(Advances in Quantum Chemistry 65) Dževad Belkić (Eds.) - Theory of Heavy Ion Collision Physics in Hadron Therapy-Academic Press, Elsevier (2013) PDF

VOLUME SIXTY FIVE
Advances in
QUANTUM CHEMISTRY
Theory of Heavy Ion Collision Physics
in Hadron Therapy
EDITORIAL BOARD
Guillermina Esti ú (University Park, PA, USA)

Frank Jensen (Aarhus, Denmark)
Mel Levy (Greensboro, NC, USA)
Jan Linderberg (Aarhus, Denmark)
William H. Miller (Berkeley, CA, USA)
JohnW. Mintmire (Stillwater, OK, USA)
Manoj Mishra (Mumbai, India)
Jens Oddershede (Odense, Denmark)
Josef Paldus (Waterloo, Canada)
Pekka Pyykko (Helsinki, Finland)
Mark Ratner (Evanston, IL, USA)
Dennis R. Salahub (Calgary, Canada)
Henry F. Schaefer III (Athens, GA, USA)
John Stanton (Austin, TX, USA)
Harel Weinstein (New York, NY, USA)
VOLUME SIXTY FIVE
Advances in
QUANTUM CHEMISTRY
Theory of Heavy Ion Collision Physics
in Hadron Therapy
Edited by
DŽEVAD BELKIĆ
Professor of Mathematical Radiation Physics
Nobel Medical Institute, Karolinska Institute
Stockholm
Sweden
Series Editors
John R. Sabin
and Erkki Brändas
Amsterdam • Boston • Heidelberg • London

New York • Oxford • Paris • San Diego
San Francisco • Singapore • Sydney • Tokyo
Academic Press is an imprint of Elsevier
Academic Press is an imprint of Elsevier
The Boulevard, Langford Lane, Kidlington, Oxford, OX51GB, UK
32, Jamestown Road, London NW1 7BY, UK
Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands
225 Wyman Street, Waltham, MA 02451, USA
525 B Street, Suite 1900, San Diego, CA 92101-4495, USA
First edition 2013
Copyright © 2013 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted

in any form or by any means electronic, mechanical, photocopying, recording or otherwise
without the prior written permission of the publisher.
Permissions may be sought directly from Elsevier’s Science & Technology Rights
Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333;
email: [email protected]. Alternatively you can submit your request online by visit-
ing the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining
permission to use Elsevier material.
Notices
No responsibility is assumed by the publisher for any injury and/or damage to persons
or property as a matter of products liability, negligence or otherwise, or from any use or
operation of any methods, products, instructions or ideas contained in the material herein.
ISBN: 978-0-12-396455-7
ISSN: 0065-3276
For information on all Academic Press publications

visit our web site at store.elsevier.com
Printed and bound in USA

13 14 15 10 9 8 7 6 5 4 3 2 1
Preface
This Special Issue is within a wider realm of physical and biological effects
of irradiation of tissue and tissue-like targets by energetic heavy ions of
high relevance to hadron therapy. The main goal is to review the lead-
ing theories describing fast collisions of ions with atoms and molecules
by emphasizing the possibilities for improving the existing data bases for
energy losses of heavy charged particles during their passage through
matter. Ion-atom collisions are included in this topic by presenting those
theoretical formalisms that are universally applicable to general targets,
including molecules from tissue. Although the main focus is on energy
losses due to electromagnetic interactions, also reviewed are the pertinent
cross sections and stopping powers for nuclear reactions. Data bases of elec-
tronic and nuclear stopping powers coupled with the associated modeling
of biological responses of cells to irradiation are essential to hadron therapy.
When determining a treatment plan for a patient with cancer, the radia-
tion oncologist must make a key assumption on the actual amount of dose
needed to eradicate all the tumor cells. It is here that the biophysical input is
required, accounting for the precise extent of the deposited physical doses,
as well as for their biological counterparts that modify the initial impact
of radiation by the cell repair processes. The overall success of radiotherapy
is contingent upon the dose planning, dose delivery and dose verification
systems.To meet with success, radiotherapy must include the most adequate
descriptions of energy losses of particle beams in tissue and the cell recov-
ery. Deep-seated tumors are usually treated with energetic hadrons because
of the optimal conformity of heavy ions to the targets by way of a very
precise local deposition of doses in the vicinity of the Bragg peak.
Versatile biophysical aspects of the topics of this Special Issue are
expounded through 14 chapters with the following specific themes:
Chapter 1 (H. Bichsel) examines the stochastic variations of energy
losses and biological effects of protons and carbon nuclei in their high-
energy collisions with water.
Chapter 2 (H. Paul) performs a comparative analysis of the accuracy of
different methods and simulation codes for stopping powers and ion ranges.
Chapter 3 (J.R. Sabin, J. Oddershede and S.P.A. Saue) reviews the theo-
retical and experimental aspects of determination of the mean excitation
energy of water.
xiii
xiv Preface
Chapter 4 (F. Ziad) studies the molecular scale Monte Carlo simulations
of ion tracks using the GEANT4-DNA code with the inclusion of the
effects of secondary electrons.
Chapter 5 (J. Beebe-Wang, P. Vaska, F.A. Dilmanian, S.G. Peggs and
D.J. Schlyer) investigates the radiation treatment verifications in proton
therapy using positron-emission tomographic imaging and Monte Carlo
simulations.
Chapter 6 (I. Abril, R. Garcia-Molina, P. de Vera, I. Kyriakou and D.
Emfietzoglou) reports on inelastic collisions of energetic protons in tissue-
like media using a combination of molecular dynamics and Monte Carlo
simulations.
Chapter 7 (C.C. Montanari and J.E. Miraglia) presents a combination
of the dielectric formalism with the shell-wise local plasma approximation
for inelastic processes in high-energy ion-matter collisions.
Chapter 8 (M.A. Bernal-Rodriguez and J.A. Liendo) assesses the useful-
ness of the available empirical cross sections relative to the continuum dis-
torted wave theories and experimental data for single ionization of liquid
water by protons, alpha particles and carbon nuclei.
Chapter 9 (R.D. Rivarola, M.E. Galassi, P.D. Fainstein and C. Champion)
reviews distorted wave methods for electron capture, ionization and excita-
tion processes in high-energy inelastic collisions of ions with water.
Chapter 10 (C. Champion, J. Hanssen and R.D. Rivarola) presents the
results of the first Born approximation for ionization and electron transfer
in energetic collisions between multiply-charged ions and water.
Chapter 11 (T. Kirchner, M. Murakami, M. Horbatsch and H.J. Lüdde)
reports on cross sections for single- and multiple-electron processes in ion-
water collisions using the time-dependent density functional theory in the
independent electron model.
Chapter 12 (Dž. Belkić, I. Mančev and N. Milojević) deals with the
four-body formalism of distorted wave second-order perturbation methods
for double electron transitions through simultaneous electron transfer and
ionization processes in ion-atom collisions at high impact energies.
Chapter 13 (V.Yu. Lazur and M.V. Khoma) reviews the theoretical con-
cept of the Dodd-Greider integral equations with Coulomb interactions
for one- and two-electron capture processes in fast ion-atom collisions.
Chapter 14 (Dž. Belkić and K. Belkić) contributes to a further
improvement of the effectiveness of the current radiation treatments of
cancer through the amended dose planning systems based on an adequate
Preface xv
description of cell survival valid at all doses as predicted by the new

mechanistic repair-based Padé linear-quadratic biophysical model.
Dževad Belkić, Guest Editor

Professor of Mathematical Radiation Physics
Nobel Medical Institute, Karolinska Institute
Stockholm
Sweden
ACKNOWLEDGMENT
The Guest Editor is highly appreciative of the support from the Swedish
Cancer Fund, Karolinska University Hospital's Radiumhemmet and the
Karolinska Institute Research Fund.
Contributors
Isabel Abril
Departament de Física Aplicada, Universitat d’Alacant, E-03080 Alacant, Spain
Joanne Beebe-Wang
Brookhaven National Laboratory, P.O. Box 5000, Upton, NY 11973, USA
Dževad Belkić
Nobel Medical Institute, Karolinska Institute, P.O. Box 260, S-171 76 Stockholm, Sweden
Karen Belkić
Mario A. Bernal-Rodríguez
Instituto de Física Gleb Wataghin, Universidade Estadual de Campinas, Brazil
Hans Bichsel
CENPA, University of Washington, Seattle, WA 98195-4290, USA
Christophe Champion
Université Bordeaux 1, CNRS/IN2P3, Centre d’ Etudes Nucléaires de Bordeaux-
Gradignan, CENBG, Gradignan, France
Pablo de Vera
Departament de Física Aplicada, Universitat d’Alacant, E-03080 Alacant, Spain
F. Avraham Dilmanian
Dimitris Emfietzoglou
Medical Physics Laboratory, University of Ioannina Medical School, GR-45110 Ioannina,
Greece
Pablo D. Fainstein
Centro Atómico Bariloche, Comisión Nacional de Energía Atómica, Avda E. Bustillo
9500, 8400 Bariloche, Argentina
Ziad Francis
Saint Joseph University, Faculty of Sciences, Department of Physics, Mkalles, Beirut,
Lebanon
Mariel E. Galassi
Instituto de Física Rosario (CONICET-UNR) and Facultad de Ciencias Exactas,
Ingeniería y Agrimensura, Universidad Nacional de Rosario, Avenida Pellegrini 250, 2000
Rosario, Argentina
Rafael Garcia-Molina
Departamento de Física, Centro de Investigación en Óptica y Nanofísica, Universidad de
Murcia, E-30100 Murcia, Spain
xvii
xviii Contributors
Jocelyn Hanssen
Laboratoire de Physique Moléculaire et des Collisions, UMR CNRS 7565, Université de
Lorraine, Metz, France
Marko Horbatsch
Department of Physics and Astronomy, York University, Toronto, Ontario, Canada M3J
1P3
Mykhaylo V. Khoma
Theoretical Physics, Uzhgorod National University, Uzhgorod, Ukraine
Tom Kirchner
Department of Physics and Astronomy, York University, Toronto, Ontario, Canada M3J
1P3
Ioanna Kyriakou
Medical Physics Laboratory, University of Ioannina Medical School, GR-45110 Ioannina,
Greece
Volodymyr Yu. Lazur
Jacinto A. Liendo
Departmento de Física, Universidad Simón Bolívar, Caracas, Venezuela
Hans Jürgen Lüdde
Institut für Theoretische Physik, Goethe-Universität, D-60438 Frankfurt, Germany
Ivan Mančev
Department of Physics, Faculty of Sciences and Mathematics, University of Niš, Niš,
Serbia
Nenad Milojević
Department of Physics, Faculty of Sciences and Mathematics, University of Niš, Niš,
Serbia
Jorge E. Miraglia
Instituto de Astronomía y Física del Espacio, CONICET and Universidad de Buenos
Aires, casilla de correo 67, sucursal 28, C1428EGA, Buenos Aires, Argentina
Claudia C. Montanari
Instituto de Astronomía y Física del Espacio, CONICET and Universidad de Buenos
Aires, casilla de correo 67, sucursal 28, C1428EGA, Buenos Aires, Argentina
Mitsuko Murakami
Department of Physics and Astronomy,York University, Toronto, Ontario, Canada M3J 1P3
Jens Oddershede
Institute for Physics, Chemistry, and Pharmacology, University of Southern Denmark,
5230 Odense M, Denmark
Helmut Paul
Atomic Physics and Surface Science, Institute for Experimental Physics, Johannes Kepler
University, Altenbergerstrasse 69, A-4040, Linz, Austria
Contributors xix
Stephen G. Peggs
Roberto D. Rivarola
Instituto de Física Rosario (CONICET-UNR) and Facultad de Ciencias Exactas,
Ingeniería y Agrimensura, Universidad Nacional de Rosario, Avenida Pellegrini 250, 2000
Rosario, Argentina
John R. Sabin
Departments of Physics and Chemistry, University of Florida, Gainesville, Florida 32611,
USA
Stephan P.A. Sauer
Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
David J. Schlyer
Paul Vaska
CHAPTER ONE
Stochastics of Energy Loss and

Biological Effects of Heavy Ions
in Radiation Therapy
Hans Bichsel
CENPA, University of Washington, Seattle, WA 98195-4290, USA
Contents
1. Introduction 2
2. Energy loss at macroscopic level 4
2.1 Continuous transport of particles through matter: Classical approach 4
2.2 Convolutions for range straggling 5
2.3 Straggling in thin segments 7
2.4 Monte Carlo methods 7
3. Bragg functions 8
3.1 Nuclear interactions 10
3.2 Practical details for protons 12
3.3 Practical details for C-ions 12
4. Energy loss and deposition at
microscopic levels 14
4.1 Energy loss 14
4.2 Energy deposition 15
4.3 Energy deposition in microscopic volumes 15
5. Stochastics of energy loss in cells 17
5.1 General features 17
5.2 MC simulation of stochastics at the Bragg peak 20
5.2.1 Simulation of random number nj of ions traversing cell j 21
5.2.2 Energy deposition z in cells 21
6. Bio-effects 21
7. Conclusions 22
Appendix A. Collision cross sections CCS 23
A.1 Rutherford cross section 23
Appendix B. Bethe–Fano collision cross section 25
B.1 Fermi-virtual-photon (FVP) cross section 28
B.2 Collision cross sections for Si, water, and DNA 29
B.3 Integrals: cumulative functions and moments for water 29
Appendix C. Multiple scattering 32
Appendix D. Nuclear reactions 32
Appendix E. Cell survival 33
Acknowledgments 33
References 33
Advances in Quantum Chemistry, Volume 65 © 2013 Elsevier Inc.
ISSN 0065-3276, http://dx.doi.org/10.1016/B978-0-12-396455-7.00001-7
All rights reserved. 1
2 Hans Bichsel
Abstract
Energy loss functions for protons and Carbon ions (not including fragmentation) are
calculated with convolutions. For C-ions a treatment dose D = 2 Gy at the Bragg peak
needs only ten ions traversing a cell of area 100 μm2. Therefore the use of a mean par-
ticle fluence per cell ND rather than D is used to assess the primary source of stochastic
effects. The stochastics of specific energy z are calculated. It shows large variations
because the particle fluence per cell as well as the energy spectrum of the ions have
large variations. A new interpretation of the relation between energy deposition and
cell survival must be found.
1. INTRODUCTION
This review is a supplemental study of the theory of energetic colli-
sions of heavy ions with matter, including tissue, by using stochastic as well
as analytic methods. The theory in this field offers a wealth of important data
bases of immediate usefulness in many research areas including health sci-
ences.1 The most frequently used data from this branch of collision physics
are differential and total cross sections, electronic stopping power, ranges, and
mean excitation energies. They constitute critical input data for modeling
energy losses of heavy ions passing through matter.The accuracy of these data
is of direct relevance to biological effects of ions, especially near the position
of the Bragg peak.2, 3 The theory used here for the calculations is based on
the Bethe–Fano method. It is outlined in the Appendix and the text. It gives
agreement of 1% or better with existing experimental and calculated data for
particle energies above 2–5 MeV/u for ions from protons to Carbon ions.
In radiation therapy a major problem is the determination of the effect
of the radiation dose on the irradiated tissues. It has been found that dose-
effect functions are steep: to achieve tumor control4 between 25% and 75%
the difference in dose is about 20%. It thus is reasonable to aim for uncer-
tainties of ±2% in physical quantities in dosimetry.5
The biological effects of radiation therapy occur in cells. Therefore we
must study energy losses and energy deposition in microscopic volumes.
In medical radiation therapy the cells of interest usually are located deep
inside the body. For heavy charged particles (protons, M0 c 2 = 938 MeV,
Carbon ions, M0 c 2 = 11178 MeV) beams of ions are used to penetrate to
these volumes. One method used for therapy is the scanning of the tumor
volume with pencil beams.6, 7
This method is studied here. Since the calculations described are made
with computers, they can readily be extended for other beam geometries
as well as cell geometries.
Stochastics of Energy Loss and Biological Effects of Heavy Ions in Radiation Therapy 3
The study consists of several parts:

• Calculation of Bragg functions with major emphasis of generating the
energy spectra of the particles for each layer of absorber along the tra-
jectories of the particles, Sections 2 and 3.
• Calculation of detailed energy deposition spectra for therapy doses at the
Bragg peak, Sections 4–6.
• For the biological effects a major problem is the stochastics of particle
fluence per cell, Section 5.2.
• Conclusions.
In order to explicitly evaluate the stochastic nature of the energy deposition
in cells the interaction of particles with matter cannot be explored on the basis
of the Bethe mean energy loss dE/dx. Instead the detailed nature of the indi-
vidual collisions of the particles with the absorber must be studied, using “col-
lision cross sections” CCS. Cross sections are described in Appendices A and B.
The CCS are needed for the convolution method which is used here,8–12
Sections 2.2 and 2.3 as well as for Monte Carlo calculations, Section 2.4.13–15
It must be understood that in general energy loss and energy deposition
must be clearly distinguished, Section 4.2. For many of the particle inter-
actions the particle speed υ = βc is the relevant parameter rather than the
energy. The parameter γ = 1/(1 − β 2 ) is also used, where M0 c 2 (γ − 1)
is the kinetic energy of the ion with mass l MeV.
To study the limitations of calculations and measurements of the physi-
cal processes in biological cells we explore the spectra of specific energy z
imparted16, 17 to cubic volumes
Vc = 10 × 10 × 10 µm3 (1)
by particles incident perpendicularly on one of the surfaces. For vol-
umes with different shapes the data derived here will have larger spreads.
Methods to determine energy loss and energy deposition are described in
Sections 2–5. Bio-effects are explored briefly in Section 6.
In this paper it is assumed that all calculations are made numerically on
computers because none of the functions can easily be written in analytic
form. It is an advantage of numerical results (i.e., tables) that it is trivial to invert
a function such as given in Eq. (3) and Figures B.3 and B.4.† Calculations made
with continuous functions will be called “analytic” here in contrast to Monte
Carlo calculations.A mean particle fluence ND per cell is used to quantify radiation.
† ne inverts the columns of numbers of the data, then uses cubic spline interpolation to get a
O
convenient scale for the inverted primary variable.
4 Hans Bichsel
2. ENERGY LOSS AT MACROSCOPIC LEVEL

The “classical” approach based on Bohr, Bethe and Fano theories of
continuous energy losses, i.e., stopping power, is discussed in Section 2.1.The
more detailed study using straggling functions is described in Sections 2.2, 2.3,
and Appendices A and B. The method of following individual particles along
their tracks with Monte Carlo simulations is discussed in Section 2.4. Multiple
elastic scattering is relegated to Appendix C, and nuclear interactions to Section
3.1 and Appendix D. The macroscopic properties of the energy deposition
field are described by Bragg functions, Section 3. Microscopic energy loss and
energy deposition are described in Section 4 and stochastic variations in
Section 5. Consequences for biological effects are outlined in Section 6.
Calculations are made for T = 200 MeV protons (βγ = 0. 687) and
T = 3600 MeV C-ions (βγ = 0. 865) traversing water (see Appendix B
about organic matter). For simplicity a cubic cell of water is used to approx-
imate the microscopic volume Vc for which energy deposition is calculated,
see Section 5. The ion beam is incident perpendicularly on the cube.
Since pencil beams are used for the scanning treatment of tumors6, 7
they are used here for the calculations.
The most important interaction of the ions with matter is the collision
with molecular electrons, see Appendix B for details. The elastic interac-
tion with atoms produces multiple scattering, Appendix C. It broadens a
pencil beam and is of some importance for protons, but can be neglected
for C-ions. Nuclear reactions are infrequent and not correlated with the
above interactions and can be dealt with independently for the continuous
transport and convolution methods, Section 3.1 and Appendix D.
2.1 Continuous transport of particles through matter:

Classical approach
In classic theory the Bethe stopping power S(T),frequently written18 as dT/dx,or
M1(β) in Eq. (B.6), is used to calculate the mean energy loss of particles with
energy T traversing absorbers of thickness ξ (units: length). If the mean energy
of the ions is reduced by less than ∼ 5% we can use the approximation
�� = ξ · S(T ). (2)
For larger energy losses, the analytic range calculation can be used to
follow the ions through the absorber with
T0 T
dT dT
y= and R(T ) = , (3)
T0 −�� S(T ) 0 S(T )
where y is the distance traversed in the absorber and R(T ) is the “mean
CSDA range” of the particles. It is practical to calculate a range table R(T )
as a function of T (Refs. 19,20).
Range approximations have been used extensively for the calculation of
“radiation doses”.21–24
In the nonrelativistic Bohr approximation25 the spectrum of energies
T of the particles emerging from the absorber of thickness y is assumed to
be a Gaussian of width σ related to the second moment M2 of the collision
cross section,26, 27 Eq. (B.6).‡
(4)
2Z

′ 2
M2 = 0. 1569z (MeVcm) /g, σ = yM2′ MeV,
A
where ze is the electric charge of the particle, Z the atomic number of the
′
absorber, A its number of nucleons per molecule. Note that M2 does not
depend on particle speed. A more sophisticated method was derived by
Tschalär.28§
The energy loss spectra (“straggling”) implicit in Eq. (2) actually are not
symmetric.9, 10, 29–31 This can be taken into account by the use of convolu-
tions for Eq. (3).
2.2 Convolutions for range straggling

For the transport through thick absorbers a convolution calculation can be
used.8, 10–12, 29, 32 Assume that we know the energy spectrum �(y, T )
(T2 < T < T1) at a plane surface at y, see Figure 1.1. To calculate the
spectrum �(y + ξ , T ) after the traversal of a thin layer ξ, we use straggling
functions V (ξ , �, T ) to determine the energy loss for ions with energy
between T and T + δT (where δT ≪ (T1 − T2 )), for each energy bin δT
between T1 and T2. These energy losses are then added together to produce
the spectrum at y + ξ at T − . The process is shown in Figure 1.1. The
straggling spectra V (ξ , �, T ) for thin absorbers are calculated with the
methods described in Section 2.3.
For one track segment ξ, Figure 1.1, a convolution integral for each
energy interval T , T + δT between T1 and T2 is calculated (s is the range
in of V (ξ , �, T + �), Figure 1.1)
‡ The
density of water is approximated by 1.00. Both g and cm3 are used as units interchangeably, as
convenient. We also assume that LET is the same as dT /dx.
easurements for 70 MeV protons at NIRS11 were analyzed with this method and that of Section
§M
2.2 for T < 20 MeV. The reproducibility of the measurements was better than 1%. The I-value for
water was derived as 79.7 ± 0.5 eV.
6 Hans Bichsel
s
�(y + ξ ; T ) = �(y, T + �) × V (ξ , �, T + �)d� or (5)
0

�(y + ξ ; T − �) = �(y, T ) × V (ξ , �, T )d�. (6)
The reader is invited to select the equation which looks more plausible.
This integral then is calculated for each energy between T1 and T2.
For present purposes (e.g., total range of 200 MeV protons11, 33 or
3600 MeV C-ions12 in water) Vavilov–Fano straggling functions V are used.
The results of the calculations are given in Section 3. This method is of the
order of 50 times faster than corresponding MC calculations.
Figure 1.1 Illustration of the convolution method. The initial energy spectrum �(y; T )
is shown at the top. For each narrow band of energies δT (shown by the vertical paral-
lel lines) the straggling spectrum V (ξ ,�,T + �) is added successively to the spectrum
�(y + ξ ; T ) shown at the bottom.
2.3 Straggling in thin segments

Analytic methods to calculate straggling for thin absorbers have been
developed for over 80 years.29–31, 34–37 One of the most comprehensive
calculations and comparison with experimental data for Si detectors is in
Ref. 9. Here the convolution method for straggling9, 30, 31 has been used.
The agreement between measurements and calculations is good.The reader
is invited to assess the data.
2.4 Monte Carlo methods

Many computer programs use Monte Carlo (MC) methods to calculate
particle spectra.13, 14, 38–42 Time consuming computations are needed to
obtain accurate data for the functions describing the depth dose functions.33
An outline of the method is given here, details are given in Section 4. In one
system the interactions occurring during the passage of the particles through
matter are simulated one at a time, collision by collision,43, 44 and include
secondary collisions by the δ-rays produced. For the calculations shown in
Figure 1.7, the following procedure was used: A particle j travels random
distances xi between successive collisions, calculated by selecting a random
number rr and determining the distance xi to the next collision from the
mean free path between collisions λ(v) = 1/Mo (v) (particle speed v is used
here rather than T )
xi = −λ(v) ln rr . (7)
The energy loss Ei is selected with a second random number from the
cumulative collision spectrum, Eq. (B.6), shown in Figures B.3 and B.4. This
process is repeated until xi exceeds the segment length x. The total energy

loss j of the particle is j = i Ei. To get Ei practically, the inverse function

E(�; βγ ) of �(E; βγ ) is calculated with e.g., cubic spline interpolation44 (see

footnote †). By binning the j the straggling function f (�) is obtained.44, 45
The Monte Carlo method can be used for all absorber thicknesses, but
with decreasing particle speed44 it is necessary to change λ(υ) (Fig. B.4)
and �(E; v) (Figs. 1.3 and 1.6) at appropriate values of υ. It may not be very
practical for very thick absorbers, e.g., for the full range of one 200 MeV
proton in water (R ∼ 25 cm) the number of collisions is of the order of
3 million. In order to get reasonable straggling functions, tracks for 2 million
protons may be needed.33**
** A study of the equivalence of the convolution and MC methods is being prepared.

8 Hans Bichsel
In several systems of MC calculations a “condensed history” approach

is used.13, 15, 40, 46 It consists of using energy loss straggling functions for
the determination of successive random electronic energy losses in short
track segments. Since the convolution method also uses such functions, any
problems†† found in their use will also appear in MC calculations, but will
be more difficult to evaluate in the MC.
3. BRAGG FUNCTIONS

A Bragg function describes the properties of the energy spectrum
�(y, T ) of the ions in the absorber as a function of penetration y in the
absorber. The methods described in Section 2 have been used to calculate
�(y, T ) and its properties at y (Eqs. (8)–(11) and B.7):
• particle fluence N (y);
• particle mean energy T (y);
• mean energy loss D(y) in a thin layer ξ;
• standard deviation σ (y) of �(y, T );
• the specific energy imparted z to cells at y;
• number of particle s(y) stopping in ξ.
The Bragg functions shown here have been calculated with the con-
volution method. In most publications about Bragg functions the only
function shown is D(y) (Figs. 1.2 and 1.5). If Bragg functions are evaluated
with the ‘‘classical” methods of Section 2.1, there will be differences from
the data given in Table 1.1.
Since the major features of the Bragg peak result from the interactions
of particles with atomic electrons, I have excluded nuclear interactions and
multiple scattering to show the features of the electron interactions clearly, 12
also see Section 3.1 and Appendices C and D.
Table 1.1 Properties (Eqs. (8)–(12)) of the fluence functions �(y,T ) shown in
Figure 1.3 for protons. Five functions are shown from a total of 1400 functions
calculated. The skewness of �(y,T ) is given by 3, for  see Sect. 3.2
N0 y cm T σ γ3 N (y) D (y) r%
0 24.52 42.68 3.822 0.299 1.00 14.0 0
2 25.532 25.68 5.827 0.576 0.9945 21.9 0.04
3 26.046 15.35 6.214 0.030 0.7598 28.7 0.55
4 26.27 12.06 5.616 −0.212 0.4624 26.6 0.95
7 26.58 9.03 4.643 −0.464 0.1058 6.2 1.6
†† Such as problems related to effective particle energy at the center of segments etc.
Figure 1.2 Bragg functions for 200 MeV protons traversing water. The thickness
of water traversed is y, the dose is D(y), the number of residual protons is N(y), the
mean energy of the spectrum �(y; T ) is T , and the standard deviation of �(y; T )
is σ (y).
The quantities defined above are calculated as follows using the method
shown in Figure 1.1
N (y) = �(y; T )dT .
(8)
In each track interval ξ the number N (y) is reduced by the number of

ions reaching T = 0, as given by s(y) in Figure 1.4. The specific dose Ds (y)
at y is given in terms of an average stopping power

Ds (y) = �(y; T )S(T )dT , (9)
where S(T ) is the stopping power for ion energy T. The mean ion energy
at y is given by

�T �(y) = �(y; T )T dT . (10)
10 Hans Bichsel
Figure 1.3 Proton spectra �(T ; y) (straggling functions for energy loss of T0 = 200 MeV
protons) at different depths y in water. Data are given in Table 1.1: T is the mean ener-
gy of �(T ; y), σ its standard deviation, γ3 its skewness and N(y) the residual number of
protons in the beam. For each function T is indicated by a vertical line near the peak.
There will be a small difference between Ds (y) and S(T (y)), Figure 1.4.
The variance σ 2 of the energy loss distribution at y is calculated with

σ (y) = �(y; T )(T − �T �)2 dT .
2
(11)
The above functions are shown in Figures 1.2–1.6.
3.1 Nuclear interactions

Three aspects of nuclear interactions can be distinguished:
• reduction of fluence �(y; T ) of primary particles. This can be taken into
account in Eqs. (5) and (6) by reducing �(y; T ) by δ� = N ξ σn (T )
where N is the number of nuclei per unit volume and σn (T ) is the cross
section for nuclear interactions. For protons it can be approximated by
Bichsel47
σn (T ) = 5. 3 · 10−26 A2/3 cm2 . (12)
Figure 1.4 Additional Bragg functions for 200 MeV protons traversing water. The thick-
ness of water traversed is y, the number of residual protons is N(y) , and the range
straggling function is s(y). Its FWHM is w = 6.4 mm. Two dose functions are given: the
dashed line is D(y) given by Eq. (9), the solid line gives the energy deposited Dd(y), in
each segment ξ. The number of ions stopping inside ξ at R0 is ∼ 0.2 %.
For water, N σn (T ) ∼ 0. 012/cm, for a mean free path = 82 cm, for

C-ions in water ∼ 25 cm (Ref. 48), in PMMA ∼ 20 cm (Ref. 49).
• electronic energy losses by secondary particles from the nuclear interac-
tions. It will be necessary to make a supplemental MC simulation for this
process. See Appendix D for other details.
• multiple scattering of the particles. Since this process produces negli-
gible energy losses it will not change the spectrum �(y; T ) of primary
particles. For broad beams of particles, the lateral displacement of the
particles will only affect the edges of the beam.11 There is a shorten-
ing of the projected track length which can be calculated analytically
with the Bichsel–Uehling method.50 For 200 MeV protons in water
the range shortening is given as 0.12% or 0.3 mm in.19 For pencil
beams the analytic form of the Moliere theory can be used51 to cal-
culate lateral scattering, see Appendix C. For 3600 MeV C-ions it is
negligible.12
12 Hans Bichsel
3.2 Practical details for protons

Calculations were made for protons with kinetic energy T = 200 MeV.
Their csda range is Ro (T ) ∼26.2 cm, Figure 1.2. The use of Vavilov strag-
gling functions V (ξ ; T + �, �) requires a choice of ξ for which ≪ T .
This means that ξ must be reduced with decreasing T. For the calculations
described here, for y < 24.5 cm (at a mean energy of 42.2 MeV, with a
spread from 28 to 68 MeV) ξ = 1 mm was used, for y > 24.5 cm ξ = 20 µ
m was used. The functions �(y; T ) given here must be considered prelimi-
nary because even with ξ = 20 µm some problems with accuracy remain.
Results for the Bragg peak are given in Figures 1.2 and 1.4. Some approxi-
mations are used in the program to simplify it. They cause errors of the
order of 0.2% in the mean range R0. Also they contribute the irregularity in
the spectra in Figure 1.3. For present purposes they are insignificant. Good
agreement between calculations and measurements for 70 MeV protons
was found,11 also with MC calculations for 200 MeV.‡‡ For y > 25. 3 cm
the energy Dd (y) deposited in each segment consists of the energy lost by
particles traversing ξ and the residual energy lost by particles stopping inside
§§
ξ, see s(y) (the range straggling function) in Figure 1.4. Energy deposited
is recorded in two parts: Dh for protons with initial energy T 1. 5 MeV and
Dl for T 1. 5 MeV, Dd (y) = Dh + Dl.
The ratio r = Dl /Dd is given in Table 1.1. The largest number of pro-
tons stopping inside of ξ = 20 µm at the peak of s(y) is 0.3% and their
contribution Dl to dose is less than 1%.Thus they are unlikely to contribute
to an increase in RBE.
3.3 Practical details for C-ions

Calculations were made for C-ions with kinetic energy T = 3600 MeV.
Their csda range is Ro (T ) ∼17.31 cm. For the calculations described here,
for y < 15.26 cm (at a mean energy of 1057 MeV, with a spread from 970 to
1175 MeV) ξ = 0. 5 mm was used, for y >15.26 cm ξ = 100 µm. Results
for the Bragg peak are given in Table 1.2 and in Figure 1.5. The spectra in
selected layers ξ = 100 µm are shown in Figure 1.6. The functions given
must be considered preliminary because ξ is too large and the Vavilov
functions were calculated with a spacing of 3 MeV in T. This causes the
structures seen for functions 3–5.
‡‡ In a comparison in 2008 with Monte Carlo functions calculated at Karolinska Institute with
SHIELD-HIT52 differences of less than 0.5 mm were seen for the functions.
§§ This is an arbitrary choice: S(1.5MeV)∼20 keV/µm =LET (∞) corresponds to the calculations in
Ref. 53.
Table 1.2 Properties (Eqs (8)–(12)) of the fluence functions of C-ions with initial
T0 = 3600 MeV shown in Figure 1.6. Data are given for six from a total of 220 functions
calculated. Depth in water is y. T and σ in MeV, D(y) in MeV/cm. For r see Sect. 3.2.
# y cm T σ γ3 N (y) D (y) r%
1 16.05 809 17.27 −1.06 0.99 346 0
2 16.55 617 21.30 −1.06 0.984 425 0.04
3 16.95 419 28.52 −0.75 0.975 575 0.55
4 17.15 285 39.24 −0.15 0.972 792 0.95
5 17.31 136 50.39 −1.19 0.794 1350 1.6
6 17.35 106 44.4 −2.0 0.526 1113 1.6
Figure 1.5 Bragg functions for 3600 MeV C-ions traversing water. The thickness of
water traversed is y, the number of residual ions is N(y). The mean energy of the ions
at y is T (y), the dose is D(y), Eq. (9), dashed line, σ is the standard deviation of the
functions shown in Figure 1.6. Nuclear fragments are not included in order to show the
nature of electronic collisions more clearly, see Section 3.1.
The range of C-ions with T = 10 MeV is ∼10 µm (Refs. 20, 54). Because
they will deposit all their energy in one or two cells, details about their inter-
actions (e.g., charge state of ion) are not important, see Sections 4.3 and 5.
The number of ions stopping in one cell at the Bragg peak is ∼0. 8%.
14 Hans Bichsel
Figure 1.6 C-ion spectra �(T ; y) = f (T ) (straggling functions for energy loss) at differ-
ent depths y in water. Data are given in Table 1.2: T is the mean energy of �(T ; y), σ its
standard deviation, γ3 its skewness and N(T ) the residual number of ions in the beam.
For function 5 the value T is indicated by a vertical line near the peak. The narrow
peaks in functions 3-5 are due to a coarse table of V (xi,�,t) of Figure 1.1.
The mean range is R0, it is given at N (y) = N (0)/2 and corresponds to

the csda range. Numerical values of T are given in Table 1.1.
Good agreement between calculations and measurements for 3600 MeV
C-ions was found.12
4. ENERGY LOSS AND DEPOSITION AT

MICROSCOPIC LEVELS
4.1 Energy loss
In order to assess the limitations of calculations and measurements we
explore the spectra of specific energy z imparted16, 17 to cubic volumes
Vc = 10 × 10 × 10 µm3. Here, z is a stochastic quantity and the determina-
tion of its local values and variations is a major purpose of this study. Since the
stochastic variation of the particle fluence is largest for C-ions we consider
primarily C-ion beams with kinetic energy T = 3600 MeV (or 300 MeV/u).

In addition we restrict the absorber to be water for most of the study (see
Appendix B). Three kinds of interactions of the ions are considered:
• inelastic collisions with atomic electrons (‘‘electronic”), Appendices A, B;
• elastic collisions with atoms, Appendix C;
• nuclear interactions, Section 3.1 and Appendix D.
Interactions are described as collisions and the effect on the particle is an
energy loss, Figure 1.7.
The cross sections are described in Appendices A and B. A schematic
description of the electronic collisions is given in Figure 1.7. Further details
are given in Appendices B, C, and D.
For particles heavier than electrons, elastic collisions cause small energy
losses (of order of eV) and angular deflections of the particles leading to
multiple scattering, described in Appendix C.
4.2 Energy deposition

For large volumes (>1 ml in solids, >1 l in gases) energy loss approximates
energy deposition closely. For smaller volumes wall effects must be taken
into account. In particular, delta rays and secondary photons can be shared
by neighboring volumes. For cells the conversion of the primary energy
into heating and chemical effects is complex (radicals, diffusion etc.55, 56)
and is of minor relevance for the present study. The collisions of particles
traversing matter occur at random. This is shown in Figure 1.7 for protons.
For C-ions with T = 150 MeV (at Bragg peak) the mean free path between
collisions is λ ∼ 0. 4 nm (Table A.1), but, presumably, the pattern is still
similar. Note that for protons λ is much larger than the diameter of DNA
(∼2. 5 nm), but for C-ions with T < 1000 MeV it is less.
4.3 Energy deposition in microscopic volumes

For fast C-ions the energy deposition is in-homogeneous in space. As an
example T = 150 MeV ions are described. Their LET is 136 keV/μm, and
the mean free path between collisions is 0.4 nm, Table A.1. The procedure
of Figure 1.7 is used, for one ion, with data from Figure B.4. It is assumed
that for energy losses E > 30 eV a secondary electron (“delta ray”) is
produced. The delta rays in turn will lose their energies in multiple colli-
sions. Frequently ‘‘radial dose distributions” are used to describe the energy
deposition and to assess radiation effects.57 Here a different method is used
which demonstrates the stochastics of energy deposition. Consider water layers
of thickness ξ = 1 µm. Numbers are listed in Table 1.3: four cylindrical
16 Hans Bichsel
Table 1.3 Data for energy deposition around the track of one C-ion (T = 150 MeV)
traversing ξ = 1 µm of water, based on Figure B.4. Eu(eV), r(nm), V(nm3), 0 number
of ion collisions per µm, 1 (keV/μm) energy lost per ion per cylinder, z (Gy) specific
energy inside each sylinder
cyl. Eu r Vu 0 1 z
A 152 3 3 104 2160 68.5 365000
B 1012 35 4 106 78 28.6 1100
C 8371 1000 4 109 11 26.6 1
D 28000 15000 7 1011 1 14.0 0.003
volumes Vu around the center of the track are studied. They are based on
the regions of energy loss shown in Figure B.4. The radii of the volumes
are defined by the mean radial range r(Eu ) in nm (Ref. 58) of the delta ray
associated with the highest energy loss Eu (eV) in each region.*** The outer
volumes are exclusive of the inner††† ones as are the values of 0 and 1,
Figure B.4. The total energy loss of the ion to the delta rays stopping in a
cylinder is 1 keV. The energy deposition by one C-ion is given. The sums
are 0 = 2250 primary collisions/μm, 1 = 137. 3 keV/μm total energy
loss. The average energy loss per collision is �E� = 61 eV.
Note: On the average, a single delta ray carries away 10% of the energy
loss into volume D.‡‡‡
The average energy deposition per ion collision in volume A is 32 eV.
One primary collective excitation fills a volume of ∼ 14 nm3 (Ref.60)
for a total volume in cylinder A of 2160 · 14 = 30240 nm3, i.e., volume
A is a solid tube of collective excitation with specific energy z ∼ 365
kGy, Table 1.3. This does not include the contributions from the collisions
by delta rays going into the outer volumes.
A reader wishing to know this contribution is invited to calculate it.§§§
In cylinder B, 0 = 78 electrons deposit energy into individual volumes
from 5 to 40 nm3, i.e., at most ∼ 1000 nm3 of the total volume VB = 4
million nm3.
*** o correction is made for the reduction of the energy of the delta ray due to the binding of the elec-
N
trons (K-shell and collective excitations). Thus for the outer three volumes the energy deposition is
equal to the energy loss by delta rays. For cylinder A the energy density is so high that this is irrelevant.
††† “Annular space between cylinders.”
‡‡‡ Delta rays lose energy at the rate of 30 to 50 eV/collision, with mean free paths from 50 nm to 1
nm along their track.59 The volume of a collective excitation along the electron tracks is estimated
to be from 15 to 1 nm3 (Ref.60). Some energy may be carried into neighboring layers, “wall effect.”
§§§ For electrons with E < 0.8 keV S∼25 eV/nm, for E < 0.8 keV S∼13.5/E0.767 eV/nm. For details
see Refs. 59, 61, 62.

In cylinder C the energy deposition is by 11 electrons filling volumes

between 40 and 1000 nm3, inside the total volume VC = 4 billion nm3.
In cylinder D one electron (with an energy between 8 and 28 keV) on
the average fills volumes between 1000 and 100000 nm3. This electron will
usually deposit its energy in more than one layer of water. Its contribution
to z on the track is between 0 and 20%.
This suggests that the LEM model57, 63 will have large stochas-
tic variations outside of cylinder A and will not be applicable outside
of cylinder C. Similar descriptions have been given in chapter 9 of
IAEA-TECDOC-799,64 but without the concept of localized collective
excitations.
Readers are encouraged to explore patterns of collective excitations in
volumes B, C, and D and explore their effects on DNA.
This energy deposition pattern by heavy ions differs greatly from that
expected for X-rays.
5. STOCHASTICS OF ENERGY LOSS IN CELLS

5.1 General features
We calculate the details of energy loss of C-ions in the “cell volume”
Vc = 1000 µm3, Eq. (1), the surface area of Ac = 100 µm2 and the thick-
ness ξ = 10 µm with the method shown in Figure 1.7. Let us consider
segment 5 of Table 1.2, y = 17. 31 cm with mean energy �T � = 136 MeV.
The mean energy loss is �� = ξ · M1 = 1443 keV (Table A.1), the mean
number of collisions is n ∼ 25000. The uncertainty of this number is less
than 1% and thus we can disregard straggling (but see Section 4.3).The aver-
age energy loss per collision is �E� = M1 /M0 ∼ 60 eV.
To explore the stochastics of the physical processes we must use the specific
energy deposited per ion in the cells rather than the dose. As a reference
concept for z, for one C-ion with kinetic energy T to a cell with volume
Vc, we use the approximation
z1 = S(T )/6250 Gy, (13)
where S(T) has units MeV/cm.**** If we assume a mean fluence ND = 20
ions cell the irradiation dose is D = 4 Gy. Next we calculate the actual
**** One ion per Ac corresponds to = 106 ions/cm2, Eq. (9). As an example, assume <T> =
136 MeV and S(T) = 1443 MeV/cm (near the mean energy at the Bragg peak). The dose D then
is D = �(T )S(T ) = 144×1015 eV/g=1.44 × 1018 eV/kg= 0.231 Gy. With S(T) = 1443 in Eq.
(13), z = 0.231 Gy.
18 Hans Bichsel
Figure 1.7 Monte Carlo simulation, Section 2.3, of the passage of T = 16 MeV

(β 2 = 0.033 at Bragg peak) protons (index j), through one layer of water of thickness
ξ = 6λ, where λ = 20 nm, Table A.2, is the mean free path between collisions. The
direction of travel is given by the arrows. Inside the absorber, the tracks are straight
lines defined by the symbols showing the location of collisions (total number nj = 56).
At each collision point a random energy loss Ei is selected from the distribution func-
tion �(E; T ), similar to Figure 1.13. Two symbols are used to represent energy losses: o
for Ei < 33 eV, × for Ei > 33 eV. Segment statistics are shown to the right: the number
of collisions for each track is given by nj, with a nominal mean value �n� = x/λ = 6.
The total energy loss is j = Ei, with the nominal mean value �� = xdT /dx = 370

eV, where dT /dx is the stopping power, M1, in Table A.2. The largest energy loss Et on
each track is also given. The mean value of the j is 325 ± 314 eV and is 15% less than
. Note that the largest possible energy loss in a single collision is EM = 33000 eV,
Eq. (A.1).
process in cells at the Bragg peak for a therapy irradiation. There are many
stochastic variations which occur during the irradiation:
• From Figure 1.8 we see that T has values between 10 and 280 MeV,
with LET between 700 and 80 keV/ μm.†††† Within these limits we
have no control over T or z for any ion in any cell.
• The ions intercept a flat layer of cells at random locations and the number
of ions traversing a cell is given by a Poisson distribution Pn (ND ), Eq. (6.1)
in Ref 9. We assume the mean value �T � = 136 MeV (Table A.1)
†††† We are not interested in details for energies below 10 MeV, see Section 3.3.
Figure 1.8 Enlarged energy spectrum f (T ) of C-ions traversing water at Bragg peak
(function 5 in Figure 1.6). The mean value of the energy of the ions is �T � ∼ 136 MeV.
The structure of f (T ) is due to an approximation of using a limited number of Vavilov
functions (V (ξ ,�,T ) in Fig. 1.1) for the convolutions, Eq. (6). The cumulative function
�(T ), Eq. (B.7), normalized to 1, is also given. It is used for the MC calculations in Section
5.2.2, also see footnote §§. The jagged structure of f (T ) is due to the coarseness of the
energy loss grid.
(S(T ) = 1443 MeV/cm) and make a calculation for ND = 20 ions.‡‡‡‡

The “mean dose” then is
D = ND · z1 = 4. 6 Gy.
(14)
The results of the MC simulation in Figure 1.9 will show us whether
this use of T is appropriate.
• In experiments to determine survival functions Sv, the effects on a limited
number of cells are measured.
• The ions traverse cells at random time intervals.
‡‡‡‡
The practice of defining a mean S(T) at the mean energy <T> of the spectrum of the ions to
calculate the dose in cell irradiations is frequently used.
20 Hans Bichsel
Figure 1.9 MC calculation of the spectrum of the number n of ions traversing cells
is given by o. It approximates a Poisson distribution with mean value ND = 20. The
spectrum of z per cell, Eq. (15), is given by ×, values of z are (n/10) Gy. The mean value
is �z� = �D� ∼ 5.4 Gy, shown by the arrow. The value D(T ) is �z� ∼ 4.6 Gy, Eq. (14),
shown by the short arrow.
• Nuclear interactions in the cell layer are infrequent, Appendix D, and

fragments from earlier layers traverse the cell and contribute to the dose,
but they are disregarded.§§§§
• Delta rays with large energies E and K-shell X-rays occur infrequently.
• Several other stochastic variations are mentioned in Appendix D.
For the calculation of energy deposition we shall simulate the stochas-
tics of the first two items only.Variations thus obtained will be smaller than
the total which would include all the above effects. Programs and results
are described in the next sub sections.
5.2 MC simulation of stochastics at the Bragg peak

A Monte Carlo calculation simulates two primary effects given above.
§§§§ ragments have smaller LET and only one or two will traverse the cell65 at ND = 10. The effect
F
will increase the variance.
5.2.1 Simulation of random number nj of ions traversing cell j

The following simple minded algorithm is used: Place a square Ac the size
of the cell (i.e., 10 × 10 μm) into the middle of a square with side length
s = 100 µm. Select two random numbers for two sides (at 90° of the large
square). If the point given by these coordinates is inside the cell use it as the
location of an incident ion. Repeat this for ND · 100 ions, Eq. (14), which
results in nj ions randomly distributed over Ac. For many cells it can be
represented analytically by a Poisson distribution, as shown in Figure 1.9.
The average value of nj is ND.
5.2.2 Energy deposition z in cells

With a second random number an energy Ti from �(T ), Figure 1.8, is
selected for each ion. Its energy deposition z1 in the cell is approximated
using Eq. (13) S(T), from Table A.1)***** and the corresponding total energy
zj imparted to cell j is given by
nj

zj = z1 (15)
1
The distribution (“spectrum”) of z for K = 10000 cells at the Bragg

peak is given in Figure 1.9. The simulated Poisson distribution from
Section 5.1 is also given. The average value of z for all cells is the dose D
of the irradiation.
6. BIO-EFFECTS
A detailed study of biological effects on cells (e.g., cell survival) is
beyond the scope of this paper. We conclude though that the broadness
of the energy spectra of the ions seen in Figures 1.8 and 1.9 requires a
more detailed study than that with the assumption of mean values shown
in Figure 1.9, used e.g., in Ref. 66. For C-ion therapy the small number
of ions passing through individual cells will cause a further spread in the
radiation dose per cell.
Studies of these effects by the author are in progress.67 A difficulty
which must be assessed is the wide spread in particle energies (and the
associated LET66) at the Bragg peak, Figure 1.8.
***** The contribution from straggling9 is small and is not included. Also see Table 1.3.
22 Hans Bichsel
7. CONCLUSIONS
I. A remarkable result of this study is the wide spread of cell doses due
to the random nature of the two independent functions in Figure 1.9.†††††
II. A major problem in determining the biological effects of irradiations
with C-ions at the Bragg peak is in the large values and the wide varia-
tion of the specific energy z per cell Vc: the variation for T is from 20 to
250 MeV, giving z1 from 1.1 to 0.14 Gy. For the most probable ion energy
at the Bragg peak, Figure 1.8, T ∼ 160 MeV, it is z1 = 0. 2 Gy, footnote.†††
III. The second problem is the random number of ions traversing
the cells given by the Poisson distribution Pn. A mean doseD = 2 Gy thus
requires only ND = 10 ions traversing a cell. As a consequence the large
spread in dose in Figure 1.9 would be even larger. Note that D and
D(T ) differ by close to 20%, Figure 1.9. If T is derived from the range
energy function by using the thickness t at which the measured Bragg peak
is located D may differ even more.
IV. For heavy charged particles the primary description of the beam should
be made with the fluence ND per cell Vc.This will remind the user of the impor-
tance to consider the Poisson distribution. The dose D is a derived quantity
which is spread out as seen in Figure 1.9.‡‡‡‡‡ In particular, D from Eq. (14)
cannot readily be used in the derivation of survival values.
V. The physical processes described in Figures 1.1–1.9 and the
Appendices are well understood and have been found to be accurate to
about to 2% (Refs. 9, 10, 12, 18, 44, 45).
The stochastic nature for the biological effects become evident when we
study small volumes and small doses: Figure 1.9. Further variations occur
due to the nature of cell survival which is expressed by the exponential
form of the relation between cell survival Sv and physical dose D (Ref. 66).
VI. At this time I have not considered the consequences for radiation
therapy with C-ions. I suggest that a more detailed analysis of the physical
parameters (or quantities) than just the “dose” should be made for cell sur-
vival experiments to establish survival functions. In particular the random-
ness due to the particle fluence, Eq. (14) should be studied. For Ne-ions the
particle fluence would be half as much as that for C-ions.
VII. It is evident that for the doses currently considered for radia-
tion therapy with C-ions (2–6 Gy) the randomness of the number of ions
††††† It must be understood that all the data points of any one of the methods are represented by one
single value on the survival curves of the biological measurements.66
‡‡‡‡‡ The agreement between calculations and measurements was good in Ref. 12.
traversing cells will cause large variations in the survival of cells as well as
on the local spectrum of the energies of the ions.These effects seem to have
been taken into account in some methods of treatment planning41, 53, 57, 68
but the procedures used are not evident to me.
VIII. To properly convert specific dose into bio-effect it may be neces-
sary to measure primary bio-effects for mono-energetic ions.
IX. The absolute accuracy of the data for water has not been deter-
mined. In view of the wide spread of the specific energy found in Figure
1.9 it may not be very important, but a comparison with experiments
similar to that for Si, Ref. 9, should be made.
X. Experimental data for 70 MeV protons and 3600 MeV C-ions
agreed to better than 1% with calculations described in Sections 2 and 3,
see Refs. 10, 11, 54.
APPENDIX A. COLLISION CROSS SECTIONS CCS

The most frequent interactions of heavy ions with matter occur
between the electric charge ze of the ion and the electrons of matter,
resulting in an energy transfer (or loss) E by the particle in inelastic collisions.
The energy is transferred to excited states of atoms, to many electrons as a
collective excitation or to single free electrons called delta rays. They can be
considered as Secondary Radiations. Details about Auger electrons, fluo-
rescence, Bremsstrahlung, Cherenkov radiation, etc. are less important for
current purposes and are not discussed.
To describe the probabilities for collisions cross sections are defined.
A.1 Rutherford cross section

For the collision process with free electrons the Rutherford or Coulomb
collision cross section27, 69–71 σR (E) is used. It serves as a first approximation
for cross sections.28, 30, 31, 35 For a heavy particle with charge ze and speed
β = v/c colliding with an electron with mass m at rest it can be written
as§§§§§
k (1 − β 2 E/EM )
σR (E, β) = ,
β2 E2 (A.1)
2π e 4 2
k= · z = 2. 54955 · 10−19 z2 eVcm2 ,
mc 2
§§§§§ Additional
factors for electrons and positrons and for particles with spin 1 and spin 1 at high speeds
2
(β ∽ 1) are given e.g. in Uehling.72 An extensive description can be found in Evans.69
24 Hans Bichsel
Table A.1 Bethe functions of C-ions traversing water with kinetic energies T (MeV), cal-
culated with the FVP method. M1(keV/µm) is the “stopping power,” also called dT /dx,
the number of collisions is M0 /µm and λ(nm) is the mean free path between collisions.
The accuracy of M1 is the order of 2%, but the uncertainty of M0 and λ may exceed 10%
T (MeV) M1 M0 λ
50.12 329.173 5723.27 .1747
63.10 274.492 4708.34 .2124
79.43 228.745 3870.74 .2583
100.00 190.147 3180.51 .3144
125.89 157.882 2612.56 .3828
158.49 131.100 2145.86 .4660
199.53 108.702 1762.86 .5673
251.19 90.190 1448.94 .6902
316.23 74.784 1191.94 .8390
398.11 62.101 981.81 1.0185
501.19 51.631 810.21 1.2343
630.96 42.977 670.25 1.4920
794.33 35.892 556.25 1.7978
1000.00 30.082 463.54 2.1573
1258.93 25.342 388.27 2.5755
1584.89 21.461 327.28 3.0554
1995.26 18.313 278.00 3.5971
2511.89 15.762 238.304 4.1964
3162.28 13.711 206.454 4.8439
3981.07 12.086 181.065 5.5230
5011.87 10.815 161.116 6.2070
6309.57 9.820 145.77 6.8603
7943.28 9.055 133.78 7.4752
10000.00 8.481 124.48 8.0331
where EM ∼ 2mc 2 β 2 γ 2 is the maximum energy loss of the particle in one

collision. The mass of the particle does not appear in the equation.69 This
cross section has been used by Bohr,25 Landau,35 Vavilov,30 and Tschalär28
for the derivation of straggling functions. The approximation is quite good
for large energy losses, Figures B.1 and B.2.
For gases and solid matter a more complex structure must be used for the
cross sections. It is my opinion that at present the Bethe–Fano cross section is
closest to reality for particles with energies above a few MeV/A. In a study of
straggling in Si9 the average difference between calculated and experimental
data was less than ±0. 2% for one set of more than 30 measurements, with
average uncertainties of ±2% for single measurements. The reader is invited
to look at the other nine data sets and to form her own opinion.
Table A.2 Bethe functions of protons traversing water with kinetic energies T (MeV),
calculated with the FVP method and Dingfelder ε (I = 81.8 eV).78 M1 (keV/ µm) is the
“stopping power,” also called dT /dx (or dE/dx ), the number of collisions is M0/µ m
and λ(nm), Eq. (7), is the mean free path between collisions. The accuracy of M1 is of
the order of 2%, but the uncertainty of M0 and λ may exceed 10%
T (MeV) M1 M0 λ
1.000 26.6878 521.76 1.92
1.259 22.6437 432.76 2.31
1.585 19.1699 358.31 2.79
1.995 16.1517 296.18 3.38
2.512 13.5691 244.49 4.09
3.162 11.3836 201.57 4.96
3.981 9.5160 166.01 6.02
5.012 7.9388 136.60 7.32
6.310 6.6114 112.31 8.90
7.943 5.5032 92.29 10.83
10.000 4.5706 75.82 13.19
12.589 3.7959 62.27 16.06
15.849 3.1477 51.15 19.55
19.953 2.6095 42.04 23.79
25.119 2.1653 34.57 28.92
31.623 1.7977 28.47 35.13
39.811 1.4930 23.48 42.59
50.119 1.2430 19.41 51.51
63.096 1.0373 16.10 62.12
79.433 .8686 13.40 74.63
100.000 .7304 11.21 89.22
125.893 .6179 9.43 106.02
158.489 .5261 8.00 125.06
199.56 .4518 6.84 146.24
251.189 .3920 5.91 169.26
316.228 .3442 5.17 193.60
398.107 .3070 4.58 218.39
501.187 .2777 4.13 242.31
630.957 .2551 3.77 264.93
APPENDIX B. BETHE–FANO COLLISION CROSS SECTION

Bethe73 derived an expression for a cross section as a function of
energy loss E and momentum transfer K using the first Born approximation
for inelastic scattering by electrons of the atomic shells. Fano74 extended
26 Hans Bichsel
the method for solids. In its nonrelativistic form it can be

written
as the Rutherford cross section modified by the ‘‘inelastic form
factor”:74–77
E2
σ (E, Q) = σR (E)|F(E, K)|2 2 ,
Q (B.1)
where Q = q2 /2m, with q = K the momentum transferred from the inci-

dent particle to the absorber, and F(E, K ) is the transition matrix element
for the atomic excitations or ionizations. For large momentum transfers,
Q ∼ E, Figure 1.2 in Ref. 74.
Usually, F(E, K) is replaced by the generalized oscillator strength (GOS)
f (E, K ) defined by
E
f (E, K ) = |F(E, K)|2 , (B.2)
Q
Figure B.1 Collision cross section for T = 50 MeV protons in Si, relative to the
Rutherford cross section for 14 electrons. Solid line: calculated with B-F approximation,
dashed line: w ith FVP approximation. The horizontal line represents the Rutherford
cross section, Eq. (A.1). The differences between B-F and FVP balance to some extent.
This indeed is the case for M1, Eq. (B.6), but for the cumulative M0 (equal to 0 in Figure
B.4, also Eq.(B.7)), the difference is 20% at E=20 eV, dropping to 8% at 10 keV.
Figure B.2 Inelastic Collision cross sections for water, calculated with FVP approxima-
tion, relative to the Rutherford cross section for 18 electrons. Solid line: T = 3600 MeV
C-ions, dashed line: T = 50 MeV C-ions. The horizontal line represents the Rutherford
cross section, Section A.
and Eq. (B.1) then is written as

E
σ (E, Q) = σR (E) f (E, K ). (B.3)
Q
The relativistic expression is given by Eq. (23) in Ref. 29 or Eq. (47)
in Ref. 74.
In the limit K → 0, f (E, K ) becomes the optical dipole oscillator strength
(DOS) f (E, 0) (Refs. 29, 45, 78).
The cross section as a function of energy loss E is obtained by integrat-
ing Eq. (B.3) over Q

dQ
σ (E; β) = σR (E; β) Ef (E, K ) . (B.4)
Qmin Q
with Qmin ∼ E 2 /2mυ 2.The dependence on particle speed β enters via Qmin.
In our current understanding, this approach to the calculation of σ (E) (solid
28 Hans Bichsel
line in Figure B.1) is closest to reality. Because of the factor 1/Q in Eqs. (B.2),
(B.3), the accuracy of f (E, 0) enters significantly78 into the calculations of
cross sections and their moments, Eqs. (B.4–B.7).******
In condensed matter energy losses of the order of 20 eV are considered
to cause collective excitations of the outer electrons of the atoms as well as
Coulomb collisions with single electrons.79 Energy losses greater than 40
to 100 eV will be mainly to single electrons, producing delta rays.††††††
We follow the suggestion of Chatterjee and Holley60 by using a collec-
tive core (with a radius of 1.5 nm for T = 150 MeV C-ions) in which 50%
of the energy loss is deposited by over 90% of the collisions, as described
in Table 1.3 and Figure B.4.
The collective excitations have a similar nature for organic matter, espe-
cially DNA and proteins.80–86 For present purposes a detailed study of the
differences is not necessary.‡‡‡‡‡‡
B.1 Fermi-virtual-photon (FVP) cross section

The FVP calculation is an approximation based on the use of photo
absorption cross sections σγ (E) (where E = ω is the photon energy) and
of the dielectric function ε(ω) = ε1 (ω) + iε2 (ω) (Refs. 29, 74, 87). §§§§§§
The collision cross section in the nonrelativistic form is given by Allison
and Cobb88
E
2 2 ′ ′
σ (E, β) = σR (E, β) Ef (E, 0)2 ln(2mc β /E) + f (E , 0)dE ,
0 (B.5)
where the integral over Q in Eq. (B.4) is approximated by the square

bracket.
This model has the advantage that the generalized oscillator strength
(GOS)29, 45 for the absorber is not needed. Data for f (E, 0) can be
extracted from a variety of optical measurements or from electron energy
†††††† etails of the ionization of the inner shells of the atoms may have to be taken into account for
D
specific biological effects.
‡‡‡‡‡‡ The collective excitations will form a full tube of energy deposition as long as the mean free path
between collisions is small: λ < 1 nm, i.e. T < 400 MeV for C-ions, Table 1.3.
****** A description of the approximations in the Bethe derivation of “stopping power” can be found
in Ref. 18.
§§§§§§ The approach is also known under the names Photo-Absorption-Ionization model (PAI) and
Weizsäcker–Williams approximation.
loss measurements.89 A detailed description of the relativistic PAI model is

found e.g., in Refs. 29, 45, 88, 90. For E > EM , σ (E) = 0.
Examples of σ (E) for Si and water are given by the lines in Figures B.1
and B.2.****** The FVP approximation has been used extensively for the
calculation of straggling functions and Monte Carlo calculations.
For convolutions and MC calculations the only parameter needed is M0
and therefore the difference seen in Figure B.1 is of importance.
B.2 Collision cross sections for Si, water, and DNA

Cross section data for water and Si are given because both††††††† have
been used for the dosimetry of heavy ion beams. For inelastic collisions of
heavy ions the electronic structure of the atoms in the absorber (especially
binding energies of the electrons) is important because probabilities for
energy transfers E depend strongly on these structures, Figures B.1 and
B.2. The collisions are also called the inelastic scattering of the particles.
For a quantitative description the Bethe–Fano (B–F) method9, 18, 29, 31, 73, 74,
77, 91 is closest to reality. For water a detailed study was made by Dingfelder,
Inokuti and Paretzke,78, 92 and should be used for accurate calculations.The

Fermi-Virtual-Photon (FVP) method34, 87, 88, 92, 93 requires less detailed
input. Since the B–F method is accurate to about 1% (Ref. 9,18) it should
be used for energy calibration of dosimeters and detectors. For current pur-
poses electrons appear only as delta rays. Usually most or all of their energy
is deposited inside the cell volumes Vc, Table 1.3, therefore details of their
interactions are not important. The interactions with the organic molecules
(DNA) in cells show a structure quite similar to that of water.80 Delta rays
with ranges larger than cell dimensions are infrequent: Table 1.3 and their
contribution to z is small.‡‡‡‡‡‡‡
B.3 Integrals: cumulative functions and moments for water

The cumulative functions �ν and the moments Mν of the collision cross
sections45 are defined by
E EM
ν
�ν (E, υ) = N Es σ (Es ; υ)dEs Mν (υ) = N Esν σ (Es ; υ)dEs ,
Em Em
(B.6)
******* Calculations have also been made for several gases.45
††††††† or simplicity it is assumed that DNA and other constituents of cells have the same dipole
F
oscillator strength as water.80
‡‡‡‡‡‡‡ The reader is encouraged to consider the geometry and stochastics in greater detail and compare
it to the description in Refs. 13, 57.

30 Hans Bichsel
Figure B.3 Cumulative energy loss functions �(E), Eq. (B.7), for single collisions in Si
are shown for several values of βγ . The excitation energy for L2 electrons is 100 eV, for
K electrons it is 1840 eV. 43 A table of the functions is given in Ref. 44.
where N is the number of electrons per unit volume, and ν = 0, 1, 2, 3 . . .

They are calculated numerically with Eq. (B.5). Examples are given in
Figures B.3 and B.4.
For water moments for C-ions are given in Table A.1, for protons in
Table A.2.§§§§§§§ The moment M0(v), which is also written as �t (υ), is the
mean number of collisions per unit track length. It is an important quantity
for the calculation of straggling functions9, 31 f (�; x, υ) because it is used
to calculate mc = ξ · M0, the mean number of collisions in a track segment
ξ, Figure 1.7. From Figures B.1 and B.2 it is evident that the exact shape of
σ (E) for small E will greatly influence values of M0 because of the factor
1
E in Eq. (B.4).
A calculation of M0 with the Rutherford spectrum, Eq. (A.1) cannot be
made because the lower limit in Eq. (B.4) is not defined.
§§§§§§§ For Si and some gases momentsM0 and M1 are given in Ref. 29, 45.
Figure B.4 The cumulative collision cross sections �(E; v) of Eq. (B.7) calculated with
FVP for C-ions in water for two values of T. The functions are normalized to 1.0. The solid
line is for T = 50 MeV, the dashed line for T = 3600 MeV. The difference between the
two functions is less than 1%. Also shown are functions for the cumulative stopping
power M1, normalized to coincide at the largest energy loss for 50 MeV ions, EM = 9 keV.
The moment M1 (υ) is usually known as the Bethe–Fano stopping

power dT /dx, i.e., the mean energy loss per unit track length.********
For Si it was found that B–F and FVP calculations differ in M1 by less
than 1%, but for M0 the differences may exceed 10% (Refs. 29, 45). M2 is
the Bohr straggling parameter.
In many publications it is customary to write the particle kinetic energy
as E, then the stopping power is dE/dx. Here the names Barkas and Bloch
are not included in the name for the stopping power because the expression
for σ (E; υ) does not contain these corrections. On the other hand, Fano74
formulated the expression for solids given in Ref. 78.
The higher moments can be used to calculate the shape of straggling
functions for large energy losses.26, 28, 71 For thin absorbers M1(v) will
result in misleading information.9 For the studies in his paper, the moments
are calculated numerically with Eq. (B.6). Only M0 and M1 are used.
******** A detailed study was made for Al and Si.9, 18

32 Hans Bichsel
The dependence of M0 (βγ ) and M1(βγ ) on particle speed is given in Tables

A.1 and A.2. The normalized zeroth cumulative moment
E ∞
�(E, υ) = σ (E ′ ; υ)dE ′ / σ (E ′ ; υ)dE ′ (B.7)
0 0
is used extensively for the MC calculations. Examples are shown in Figures

1.8, B.3, and B.4.
A quantity which is of great importance for dealing with stochastic pro-
cesses is the number of collisions per unit track length M0 of Eq. (B.6)(Refs.
78, 94, 95). It is the inverse of the mean free path λ between collisions.
APPENDIX C. MULTIPLE SCATTERING

For this study the effects of multiple scattering are small and have
been neglected. For a detailed study use the papers by Molière.96–98
Experimental measurements give good agreement with the Molière the-
ory.51 For the determination of lateral displacement in ion trajectories, cal-
culations have been made using98 with results given in Ref. 99. For broad
beams of protons and C-ions the effect was discussed in Refs. 11, 12. For a
1 mm pencil beam of 200 MeV protons after a distance of 26.2 cm in water
(residual mean energy 6 MeV) the lateral spread has a FWHM ∼ 8 mm. For
C-ion beams a description is given in Ref. 7.
For measurements with a broad beam (several cm in diameter), multiple
scattering will influence the dose distribution only at the edges of the beam.
In the central part of the beam the range of the particles is reduced.11, 12, 32, 50
Since the number of elastic collisions is of the same order of magnitude as that
of electronic collisions, the effect can be calculated analytically with Moliere
theory.96–98
For 3600 MeV C-ions the influence of multiple scattering on the
CSDA range was less than 0.1% (Ref. 12).
APPENDIX D. NUCLEAR REACTIONS

The mean free paths , Eq. (12), for nuclear interactions are large, so
that only a fraction of the ions will interact. The net effect for the convolu-
tion method in Section 2.2 is to reduce the fluence �(y, T ) at each layer
by δ� = ξ/�, Eq. (6). With ξ = 0. 01 cm for the C-ion calculations, the
reduction would be−dE/ρdℓper layer, for a total reduction of about 30% at

the Bragg peak. Further details for these estimates are given in Ref. 12, in par-
ticular it was found that the nuclear interactions in the passage across the peak
will change the shape of the peak only slightly. For a study of effects of nuclear
interactions a Monte Carlo calculation is needed. I am using results obtained
with SHIELD-HIT14, 15 to consider the influence of nuclear products for the
radiation effects discussed in Section 6. The contribution to the dose at the
Bragg peak estimated in Ref. 12 agree approximately with those shown in
Ref. 14. The spectrum of fragments calculated in Ref. 65 shows mostly low
LET particles which will have a small effect on the cell survival functions.
APPENDIX E. CELL SURVIVAL

Cell survival is a complex process, in particular the study of sub-
volumes and sub-lesions53,100–102 goes beyond the scope of this study.67 The
description of radial distribution of energy deposition, Section 4.3, Table
1.3, may be more suitable for this purpose than “radial dose distribution”.41
Some effects contributing53 to stochastic or systematic variations in addi-
tion to those described in Section 5 are listed here:
a. The randomness of the energy deposition due to straggling.
b. The randomness in cell shape and size.
c. “Wall effects” due to multiple scattering, oblique traversals or delta rays.
d. The use of “linear-quadratic” survival assumptions.
e. Assumptions about the nature of cell survival, e.g., interactions of sub-
lesions etc.53
f. The variety of responses for different cell types.66
ACKNOWLEDGMENTS
I am grateful for the help and advice by S. Hoedl, E. Swanson, J. Schwartz,
R. Seymour, I. Gudowska, D.J. Brenner and M. Dingfelder.
REFERENCES
1. Belkić, Dž. Reviews of theories on ionization in fast ion-atom collisions with pros-
pects for applications to hadron therapy. J. Math. Chem. 2010, 47, 1366.
2. Andreo, P. On the clinical spatial resolution achievable with protons and heavier
charged particle radiotherapy beams. Phys. Med. Biol. 2009, 54, N205.
3. Paganetti, H. Range uncertainties in proton therapy and the role of Monte Carlo
simulations. Phys. Med. Biol. 2012, 57, R99.
34 Hans Bichsel
4. Battermann, J. J.; Hart, G. A. M.; Breur, K. Dose-effect relations for tumor control and
complication rate after fast neutron therapy for pelvic tumors. Br. J. Radiol. 1981, 54,
899–904.
5. Wyatt, M.; Corredor, C.; Tamimi, M.; Miller, L. F. Comparison of treatment planning
dose calculations with measurements and Monte Carlo calculations in a rando phan-
tom. Rad. Prot. Dosim. 2005, 116, 461.
6. Pedroni, E.; Scheib, S.; Böhringer, T., et al. Experimental characteriztion and physical
modelling of the dose distribution of scanned proton pencil beams. Phys. Med. Biol.
2005, 50, 541–561.
7. Krämer, M.; Jäkel, O.; Haberer, T.; Kraft, G.; Schardt, D.; Weber, U. Treatment planning
for heavy-ion radiotherapy: physical beam model and dose optimization. Phys. Med.
Biol. 2000, 45, 3299–3317.
8. Bichsel, H. Unpublished calculations of range straggling functions for pions by convo-
lution of straggling functions for thin absorbers at Los Alamos, 1979.
9. Bichsel, H. Straggling in thin silicon detectors. Rev. Mod. Phys. 1988, 60, 663–699.
10. Bichsel, H. In Energylosses of electrons below 10 keV, Pfefferkorn Conference, Park City,
May 1989.
11. Bichsel, H.; Hiraoka, T. Energy loss of 70 MeV protons in elements. Nucl. Instrum.
Methods B 1992, 66, 345–351.
12. Bichsel, H.; Hiraoka,T.; Omata, K. Aspects of fast ion dosimetry. Radiat. Res. 2000, 153,
208–219.
13. Berger, M. J. Monte Carlo calculation of the penetration and diffusion of fast charged
particles. Methods Comput. Phys. 1963, 1, 135–215.
14. Gudowska, I.; Sobolevsky, N.; Andreo, P.; Belkić, Dž.; Brahme, A. Ion beam transport
using the Monte Carlo code SHIELD-HIT. Phys. Med. Biol. 2004, 49, 1933–1958.
15. Gudowska, I.; Sobolevsky, N. Calculations of particle and heavy ion interactions with
space shielding materials using the SHIELD-HIT transport code. Radiat. Meas. 2006,
41, 1091.
16. ICRU Report 33, Radiation quantities and units, International Commission on
Radiation Units, 7910 Woodmont Ave., Washington, DC, 1979.
17. ICRU Report 36, Microdosimetry, International Commission on Radiation Units,
7910 Woodmont Ave., Washington, DC, 1983.
18. Bichsel, H. Shell corrections in stopping powers. Phys. Rev. A 2002, 65, 052709.
19. ICRU Report 49, Stopping Powers and Ranges for Protons and Alpha Particles,
International Commission on Radiation Units and Measurements, 7910 Woodmont
Ave., Bethesda, MD, 1993.
20. ICRU Report 73, Stopping of ions heavier than Helium, International Commission
on Radiation Units, 7910 Woodmont Ave., Washington, DC, 2005. Note that the
tables for water are labelled incorrectly: the table for −dE/ρdℓ labelled “liquid
water” is for water vapor and vice-versa, the range table is for water vapor rather than
liquid water. Corrected Tables are at ICRU.org/images/pdf/errata-and-adenda.pdf.
21. Carlsson, A. K.; Andreo, P.; Brahme, A. Monte Carlo and analytical calculation of pro-
ton pencil beams for computerized treatment plan optimiztion. Phys. Med. Biol. 1997,
42, 1033–1053.
22. Hollmark, M.; Uhrdin, J.; Belkic, Dz.; Gudowska, I.; Brahme, A. Influence of multiple
scattering and energy loss straggling on the absorbed dose distributions of therapeutic
light ion beams: I. Analytical pencil beam model. Phys. Med. Biol. 2004, 49, 3247.
23. Kempe, J.; Gudowska, I.; Brahme, A. Depth absorbed dose and LET distributions of
therapeutic 1H, 4He, 7Li, and 12C beams. Med. Phys. 2007, 34, 183–192.
24. Kempe, J.; Brahme, A. Energy-range relation and mean energy variation in therapeutic
particle beams. Med. Phys. 2008, 35, 159.
25. Bohr, N. The penetration of atomic particles through matterDan. Mat. Fys. Medd 1948,
18.
26. Bichsel, H. Straggling of heavy charged particles: Comparison of Born hydrogenic-
wave-function approximation with free-electron approximation. Phys. Rev. B 1970, 1,
2854–2862.
27. Bichsel, H. Passage of Charged Particles Through Matter. In American Institute of Physics
Handbook, 3rd ed.; Gray, D. E., Ed.; McGraw Hill, 1972; Chapter 8d.
28. Tschalär, Ch. Straggling distributions of large energy losses. Nucl. Instrum. Methods
1968, 61, 141.
29. Bichsel, H. Interaction of Radiation with Matter. In Landolt-Boernstein, New Series,
Group I, Elementary Particles, Subvolume B, Detectors and Data Handling, Schopper,
H. Ed., Springer, 2010; Vol. 21.
30. Vavilov, P. V. Ionization losses of high-energy heavy particles. Sov. Phys. JETP 1957, 5,
749–751.
31. Bichsel, H.; Saxon, R. P. Comparison of calculational methods for straggling in thin
absorbers. Phys. Rev. A 1975, 11, 1286–1296.
32. Bichsel, H.; Hiraoka, T. Energy spectra and depth-dose curves for 70 MeV protons. Int.
J. Quantum Chemistry: Quantum Chem. Symp. 1989, 23, 565–574.
33. Gudowska, I. Private Communication, 2009.
34. Williams, E. J. The straggling of b Particles. Proc. Roy. Soc. A 1929, 125, 420.
35. Landau, L. On the energy loss of fast particles by ionization. J. Phys. 1944, VIII,
201–205.
36. Herring, J. R.; Merzbacher, E. Statistical problems in energy loss and ionization pro-
cesses. J. Elisha Mitchell Sci. Soc. 1957, 73, 267.
37. Kellerer, A. M. G. S. F. Bericht B-1, Strahlenbiologisches Institut der Universität
München, 1968.
38. Berger, M. J. Penetration of proton beams through water. I. Depth-dose distributions,
spectra and LET distributions, Report NISTIR 5226, National Institute of Standards
and Technology, Gaithersburg, MD, July 1993.
39. Nikjoo, H.; Uehara, S.; Emfietzoglou, D., et al. Track structure codes in radiation
research. Radiat. Meas. 2006, 41, 1052–1074.
40. Niita, K.; Sato, T.; Iwase, H., et al. PHITS—a particle and heavy ion transport code
system. Radiat. Meas. 2006, 41, 1080–1090.
41. Elsässer, T.; Cunrath, R.; Krämer, M.; Scholz, M. Impact of track structure calcula-
tions on biological treatment planning in ion radiotherapy. New J. Phys. 2008, 10,
075005.
42. Agostinelli, S.; Allison, J.; Amako, K., et al. Geant4—a simulation toolkit. Nucl. Instrum.
Methods A 2003, 506, 250–303.
43. Bichsel, H. Inelastic electronic collision cross sections for Monte Carlo calculations.
Nucl. Instrum. Methods B 1990, 52, 136.
44. Bichsel, H. <http://faculty.washington.edu/hbichsel>.
45. Bichsel, H. A method to improve tracking and particle identification in TPCs and
silicon detectors. Nucl. Instrum. Methods A 2006, 562, 154–197.
46. Berger, M. J.; Seltzer, S. M. Multiple Scattering Corrections for Proton Range
Measurements. p. 69–98 in Studies in Penetration of Charged Particles in Matter; Nat.
Acad. Sci.-Nat. Res. Council, Publication 1133, 1964.
47. Bichsel, H. Charged Particle Interactions. In Radiation Dosimetry, 2nd ed.; Attix, F. H.,
Roesch, Wm. C., Eds.; Academic Press, 1968; pp. 157–228.
48. Schall, I.; Schardt, D.; Geissel, H., et al. Charge changing nuclear reactions of relativistic
light-ion beams (5 < Z < 10) passaging through thick absobers. Nucl. Instrum. Methods
B 1996, 117, 221–234.
36 Hans Bichsel
49. Matsufuji, N.; Fukumura, A.; Komori, M.; Kanai, T.; Kohno, T. Influence of fragment
reaction of relativistic heavy charged particles on heavy-ion radiotherapy. Phys. Med.
Biol. 2003, 48, 1605–1623.
50. Bichsel, H.; Uehling, E. A. Multiple scattering correction for proton ranges and
the evaluation of the L-shell correction and I-value for Al. Phys. Rev. 1960, 119,
1670–1680.
51. Bichsel, H. The multiple scattering of protons. Phys. Rev. 1958, 112, 182.
52. Gudowska, I. Private communication, July 2008.
53. Brenner, D. J. Track structure, lesion development and cell survival. Radiat. Res. 1990,
124, S29–S37.
54. Hiraoka, T.; Bichsel, H. Stopping power and range tables for C, N, Ne, Si and Ar ions
in various dosimetric materials. Medical Standard Dose 2000, 5(Supplement 1), 1–145.
55. Cobut,V.; Frongillo,Y.; Patau, J. P.; Goulet,T.; Fraser, M.-J.; Jay-Gerin, J.-P. Monte Carlo
simulation of fast electron and proton tracks in liquid water-I. physical and physico-
chemical aspects. Radiat. Phys. Chem. 1998, 51, 229–243.
56. Dingfelder, M. Track structure: time evolution from physics to chemistry. Radiat. Prot.
Dosimet. 2006, 122, 16–21.
57. Elsässer, T.; Scholz, M. Cluster effects within the local effect model. Radiat. Res. 2007,
167, 319.
58. LaVerne, J. A.; Mozumder, A. Penetration of low-energy electrons in water. Radiat. Res.
1983, 96, 219–234.
59. Dingfelder, M.; Ritchie, R. H.;Turner, J. E.; Friedland,W.; Paretzke, H. G.; Hamm, R. N.
Comparisons of calculations with PARTRAC and NOREC: transport of electrons in
liquid water. Radiat. Res. 2008, 169, 584–594.
60. Chatterjee, A.; Holley, W. R. Computer simulation of initial events in the biochemical
mechanisms of DNA damage. Adv. Radiat. Biol. 1993, 17, 181.
61. Dingfelder, M.; Hantke, D.; Inokuti, M.; Paretzke, H. G. Electron inelastic-scattering
cross sections in liquid water. Radiat. Phys. Chem. 1998, 53, 1–18.
62. Turner, J. E.; Paretzke, H. G.; Hamm, R. N.; Wright, H. A.; Ritchie, R. H. Comparative
study of electron energy deposition and yields in water in the liquid and vapor phases.
Radiat. Res. 1982, 92, 47–60.
63. Beuve, M. Formalization and theoretical analysis of the local effect model. Radiat. Res.
2009, 172, 394–402.
64. Atomic and Molecular Data for Radiotherapy and Radiation Research, IAEA-
TECDOC-799, May 1995.
65. Gudowska, I.; Kempe, J.; Sobolevsky, N. Low and high LET dose components in
Carbon beam. Radiat. Prot. Dosimet. 2006, 122, 483.
66. Furusawa,Y.; Fukutsu, K.;Aoki, M.; Itsukaichi, H.; Eguchi-Kasai, K.; Ohara, H.;Yatagai, F.;
Kanai, T.; Ando, K. Inactivation of aerobic and Hypoxic cells from three different cell
lines by accelerated He-, C- and Ne-Ion beams. Radiat. Res. 2000, 154, 485–496.
67. Bichsel, H. Cell survival, Presentation at Annual Meeting of Rad. Res. Soc., September
2010, Maui.
68. Scholz, M.; Kellerer, A. M.; Kraft-Weyrather, W.; Kraft, G. Computation of cell survival
in heavy ion beams for therapy. Radiat. Environ. Biophys. 1997, 36, 59.
69. Evans, R. D. The Atomic Nucleus; McGraw Hill.
70. Leroy, C.; Rancoita, P.-G. Principles of radiation interaction in matter and detection,
World Scientific, 2004.
71. Sigmund, P. Particle Penetration and Radiation Effects; Springer.
72. Uehling, E. A. Penetration of heavy charged particles in matter. Ann. Rev. Nucl. Sci.
1954, 4, 315.
73. Bethe, H. Zur Theorie des Durchgangs schneller Korpuskularstrahlen durch Materie.
Ann. Phys. 1930, 5, 325.
74. Fano, U. Penetration of protons, alpha particles, and mesons. Ann. Rev. Nucl. Sci. 1963,
13, 1.
75. Manson, S. T. Inelastic collisions of fast charged particles with atoms: ionization of the
Al L shell. Phys. Rev. A 1972, 6, 1013.
76. Inokuti, M. Inelastic collisions of fast charged particles with atoms and molecules—the
Bethe theory revisited. Rev. Mod. Phys. 1971, 43, 297.
77. Bote, D.; Salvat, F. Calculations of inner-shell ionization by electron impact with
the distorted-wave and plane-wave Born approximation. Phys. Rev. A 2008, 77,
042701.
78. Dingfelder, M.; Inokuti, M.; Paretzke, H. Inelastic-collision cross sections of liquid
water for interactions of energetic protons. Radiat. Phys. Chem. 2000, 59, 255–275.
79. Ritchie, R. H.; Hamm, R. N.; Turner, J. E.; Wright, H. A.; Ashley, J. J.; Basbas, G. J.
Physical aspects of charged particle track structure. Int. J. Radiat. Appl. Instrum. Part D
1989, 16, 141–155.
80. Wilson,W. E.; Nikjoo, H. A Monte Carlo code for positive ion track simulation. Radiat.
Environ. Biophys. 1999, 38, 97–104.
81. Isaacson, M. Interaction of 25 keV electrons with the nucleic acid bases, adenine, thy-
mine, and Uracil, I. outer shells, II. inner shells, J. Chem. Phys.1972, 56, 1803 and 1813.
82. Johnson, D. E. The interaction of 25 keV electrons with Guanine and Cytosine. Radiat.
Res. 1972, 49, 63.
83. Inagaki, T.; Hamm, R. N.; Arakawa, E. T.; Painter, L. R. Optical and dielectric proper-
ties of DNA in the extreme ultraviolet. J. Chem. Phys. 1974, 61, 4246.
84. Killat, U. Optical properties of C6H12 ,..., C5H5N in the solid and gaseous state
derived from electron energy losses. J. Phys. C: Solid State Phys. 1974, 7, 2396.
85. Killat, U. Electron energy-loss measurements on 1,3-cyclohexadiene and 1,4-cyclo-
hexadiene. Z. Physik 1974, 270, 169.
86. Handbook of optical constants of solids: three volumes; Academic Press: New York, 1985,
1991 and 1998.
87. Fermi, E. Ueber die Theorie des Stosses zwischen Atomen und elektrisch geladenen
Teilchen. Z. Phys. 1924, 29, 315–327.Fermi, E. Sulla teoria dell’urto tra atomi e cor-
puscoli elettrici. Nuovo Cimento 1925, 2, 143–158.
88. Allison, W. W. M.; Cobb, J. H. Relativistic charged particle identification by energy loss.
Ann. Rev. Nucl. Part. Sci. 1980, 30, 253.
89. Raether, H. Excitation of plasmons and interband transitions by electrons; Springer Verlag:
New York, 1980.
90. Blum, W.; Rolandi, L. Particle Detection with Drift Chambers; Springer-Verlag
91. Fermi, E. The ionization loss of energy in gases and condensed materials. Phys. Rev.
1940, 57, 485.
92. Fernandez-Varea, J. M.; Salvat, F.; Dingfelder, M.; Liljequist, D. A relativistic optical-data
model for inelastic scattering of electrons and positrons in condensed matter. Nucl.
Instrum. Methods B 2005, 229, 187–217.
93. Liljequist, D. A simple calculation of inelastic mean free path and stopping power for
50 ev—50 keV electrons in solids. J. Phys. D 1983, 16, 1567.
94. Saxon, R. P.Total cross section for inelastic scattering of charged particles by atoms and
Molecules. VII. The Neon atom. Phys. Rev. A 1973, 8, 839–849.
95. Inokuti, M.; Saxon, R. P.; Dehmer, J. L. Total cross-sections for inelastic scattering of
charged particles by atoms and molecules-VIII. Systematics for atoms in the first and
second row. Int. J. Radiat. Phys. Chem. 1975, 7, 109.
96. Molière, G. Theorie der Streuung schneller geladener Teilchen I, Einzelstreuung am
abgeschirmten Coulomb-Feld. Z. Natf. 1947, 2A, 133.
97. Molière, G. Theorie der Streuung schneller geladener Teilchen II, Mehrfach- und
Vielfachstreuung. Z. Natf. 1948, 3A, 78.
38 Hans Bichsel
98. Molière, G.Theorie der Streuung schneller geladener Teilchen III, Die Vielfachstreuung
von Bahnspuren unter Berücksichtigung der statistischen Koppling. Z. Natf. 1955,
10A, 177.
99. Bichsel, H.; Hanson, K. M.; Schillaci, M. E. Lateral displacement in small angle multiple
scattering. Phys. Med. Biol. 1982, 27, 959–961.
100. Rossi, H. H.; Zaider, M. Compound dual radiation action. Radiat. Res. 1997, 132,
178–183.
101. Sachs, R. K.; Hahnfeld, P.; Brenner, D. J. The link between low-LET dose-response
relations and the underlying kinetics of damage production/repair/misrepair. Int. J.
Radiat. Biol. 1997, 72, 351–374.
102. Carlson, D. J.; Stewart, R. D.; Semenenko, V. A.; Sandison, G. A. Combined use of
Monte Carlo DNA damage simulations and deterministic repair models to examine
putative mechanisms of cell killing. Radiat. Res. 2008, 169, 447–459.
CHAPTER TWO
On the Accuracy of Stopping

Power Codes and Ion Ranges
Used for Hadron Therapy
Helmut Paul
Atomic Physics and Surface Science, Institute for Experimental Physics, Johannes Kepler University,
Altenbergerstrasse 69, A-4040, Linz, Austria
Contents
1. Introduction 39
2. Tables and programs 41
3. Liquid water as a target 43
3.1 Stopping power of water for hydrogen ions 43
3.2 Range measurements for water, and mean
ionization energy 45
4. Other target substances and statistical comparisons 49
4.1 Statistical comparisons for H and He ions 51
4.2 Application to therapy using H ions 55
4.3 Statistical comparisons for carbon ions 56
5. Conclusions 58
6. List of acronyms 58
References 59
Abstract
Various stopping power theories and codes are compared statistically to available
experimental data, in view of judging their reliability for hadron therapy. Both stopping
powers and ranges are taken into consideration. The mean ionization energy of water
is discussed in detail.
1. INTRODUCTION
One of the main characteristics of therapeutical hadron beams is the
range which is mainly determined by the mean ionization potential I of the
material. According to the relativistic Bethe formula (without corrections),1, 2
the mass stopping power is given by
Z12 Z2
S/ρ = (0. 307075 MeVcm2 g−1 ) L(β), (1)
β 2 A2

40 Helmut Paul
where S = −dE/dx is the linear stopping power (or linear energy transfer,
LET), x is the distance, ρ is the density of the material; Z1, E, and υ are
charge number, energy and velocity of the ion; Z2 and A2 are charge num-
ber and mass number of the target; β = υ/c, c is the speed of light; and the
stopping number L is given by
2mυ 2
L(β) = ln − ln I − β 2 , (2)
(1 − β 2 )
where m is the mass of the electron. Eq. (2) is reliable at energies high
enough (but not so high that the density correction becomes appreciable2).
To extend the validity to lower energy, one customarily adds shell, Barkas–
Andersen and Bloch corrections2 to Eq. (2) if necessary.
In place of the mass stopping power, one also uses the stopping cross
section S/n (where n is the atomic density), usually expressed in units of
10−15 eVcm2/atom. For a solid absorber, this corresponds roughly to the
energy loss per atomic layer.
For a compound, following the custom of Ziegler,18 we express
the stopping cross section in terms of 10−15 eVcm2/atom, dividing the
molecular cross section by the number of atoms in the molecule. This has
the advantage that stopping cross sections are nearly the same for substances
of similar composition but different A2 (e.g., for polymers).
Since stopping power depends essentially on velocity rather than energy,
one usually lists or plots it as a function of specific energy, i.e., energy per
nucleon, or of velocity squared† (e.g., in MeV/u).
For a mixture or, assuming Bragg’s additivity rule,3 for a compound, the
mass stopping power is obtained by a linear combination of the constituent
stopping powers:2
S S
= wj , (3)
ρ j
ρ j
where wj is the fraction by weight, and (S/ρ)j is the mass stopping power
of the jth constituent.
Deviations from Bragg’s additivity have been investigated for many
years (see, e.g., Ref. 49). In Ziegler’s SRIM program,18 a correction is built
in that can be used for many compounds (see Section 4.1). The SRIM
program also provides a correction for the negative solid-gas difference that
exists between the mass stopping powers of substances in different phases,
† Velocity squared in MeV/u is often incorrectly called “energy.”
On the Accuracy of Stopping Power Codes and Ion Ranges Used for Hadron Therapy 41
at low energy.4 The positive solid-gas difference at high energy is, however,
only described31 by CasP,8 not by SRIM.
The relation for the mean ionization energy corresponding to Eq. (3) is

Z2j Z2
ln I = wj ln Ij / , (4)
A2j A2
where

Z2 Z2j
= wj , (5)
A2 A2j
where Z2j and A2j define the jth constituent.

Lists of mean ionization energies can be found in ICRU Report 49.2
Some of these values are probably outdated, and a Report Committee of
the ICRU is working to improve the values for water and graphite.
To find the simplest approximation to the range R, it suffices to inte-
grate the reciprocal of the linear stopping power over energy:
R 0 E0
dE dE
RCSDA = dx = = , (6)
0 E0 dE/dx 0 S
where E0 is the initial energy. This is called the continuous-slowing-down
(CSDA) approximation.2 Due to scattering, actual ranges are somewhat
smaller; this fact can be described by the “detour factor”.2 Projected ranges,
i.e., the projection of the range onto the axis of the original direction of
travel, are calculated by program SRIM.18
The therapeutic value of hadron beams lies in the fact that the mean
ionization per unit length has its maximum (the “Bragg peak”) close to the
end of path.
2. TABLES AND PROGRAMS

The stopping power tables and computer programs discussed here are
listed in Table 2.1. Program PASS (on which the tables in ICRU Report 73
and in the Erratum are based) and the program by Lindhard and Sørensen5
(LS) are based on first principles only. The same is true for CasP, except that
empirical values are used here for the ionic charge. The tables by Janni, by
Hubert et al. and by Ziegler, and the program MSTAR are semi-empirical.
Program LET is not further considered here since it is not independent,
but based on Ziegler’s programs. Program BEST has been used to produce
the high energy parts of the tables in ICRU Report 49;2 it is useful in that
42
Table 2.1 Tables and computer programs for the stopping power of matter for positive ions. “u” is the unified atomic mass unit, also called
dalton
Name, reference Z1 Z2 (Specific) energy range Remarks
ATIMA6 1–92 1–92 ≥10 MeV/u (as used Baseda on Lindhard–Sørensen above 30 MeV/u
here)
BEST7 1–92 1–92; 180 compoundsb ≥0.5 MeV/u Bethe theory with corrections; bare ions
CasP v. 5.08 1–92 1–92, any compoundc 0.0001–200 MeV/u Default settings used here for target ionizationd
HISTOP9 many 6 0.01–30 MeV/u HISTOP for the valence electrons, SCA for the
K shell of carbon
Hubert et al.10 2–103 36 solid elements 2.5–500 MeV/u
ICRU Reporte 1, 2 25 elements, 48 0.001–10000 MeV (p); Programs NIST PSTAR, NIST ASTAR
492 compounds or mixtures 0.001–1000 MeV (α)
ICRU Report 3–18, 26 25 elements, 31 0.025–1000 MeV/u Based on PASS
7311, 12 compounds
Janni13 1 1–92; 63 compounds 0.001–10000 MeV
LET14, 15 1–92 19 materials 0.2–1000000 MeV/u Based on Ziegler’s TRIM/SRIM programs
(before 1999)
MSTAR16 3–18 31 elements, 48 0.00025–250 MeV/u Based on alpha stopping powers of ASTAR
compounds or mixtures
PASS17 many many Above 0.025 MeV/u Binary Theory, used for ICRU Rep. 73
SRIMf 200318 1–92 1–92, many compounds 1.1 eV–10 GeV/u SRIM stopping was not changed since 2003
Ziegler et al.19 1–92 1–92; many other targets 0.1–100000 keV/u First program to treat all ions, all targets
a Below 10 MeV/u, the values are based on an old version of SRIM (Ziegler et al. 1985). Between 10 and 30 MeV/u, the values are interpolated between SRIM and LS.
b Additional compounds may be calculated by entering a chemical formula, assuming Bragg’s additivity.
c Compounds are calculated according to chemical formula, assuming Bragg’s additivity.
d Target and projectile ionization must be calculated separately, and added.
e At high energy, the ICRU table was calculated using BEST; the low energy part is based on experimental data.
Helmut Paul
f SRIM was called TRIM in earlier times.

the value of I can be changed, but it assumes a bare nucleus and is therefore
not reliable below about 1 MeV/nucleon.
To represent stopping for heavy ions at the highest energies correctly,
it is necessary to use the non-perturbative LS theory which is fully
relativistic and, in addition, assumes projectile nuclei of finite size. For con-
venience, we have employed the program ATIMA instead which is based
on the LS p rogram above 30 MeV/u and which includes shell, Barkas and
Fermi-density effect corrections and in addition, a correction for projectile
mean charge. Results for ions heavier than Ar are shown in Ref. 31. But
the use of the LS program will hardly be necessary for radiation therapy,
since even for oxygen ions at 690 MeV/u, there is no difference between
LS theory and Bethe theory [Eq. (1)].20
3. LIQUID WATER AS A TARGET

Because of the large water content of the human body, liquid water
is one of the most important substances to be considered.
3.1 Stopping power of water for hydrogen ions

Figure 2.1, taken from our large collection of stopping graphs and data,21
shows experimental values and theoretical or semi-theoretical descriptions
of the stopping power of liquid and solid water for protons. In particular, it
shows the recent measurements by the Kyoto group26, 27 who used a liquid
water jet in vacuum, and by the Jyväskylä group28 who used a thin water
sheet (enclosed between two thin copper sheets) in transmission. Figure 2.2
shows some of the same data again, but divided by the data from the proton
table of ICRU Report 49,2 to make small differences visible.
For the curves in Figure 2.2, the I value is given where known. On the
average, above 10000 keV, the BEST (Bethe) curve (I = 78 eV) is 0.52%
below the ICRU 49 curve at unity (I = 75 eV), as one expects comparing
L(β) for the two different I-values; for the PASS curve, the difference with
respect to unity is slightly larger.
Inspection of Figure 2.2 shows a problem:30, 31 the Bethe equation (i.e.,
BEST) is generally reliable32 and depends essentially only on I and on the
shell correction in the region above 10000 keV, and the latter correction is
quite small here. The GarM09 curve and the PASS curve are very close to
BEST, about 1% below ICRU 49 (i.e., unity), and they therefore all seem
reliable. Hence it appears that the Kyoto measurements (and the Emf 09
44 Helmut Paul
Figure 2.1 Electronic stopping power of solid and liquid water (we assume that the
I-values for solid and liquid water are equal46) for protons, versus energy. The letters
show experimental data; the corresponding data references can be found in Ref. 21.
Some of the general tables and theories are listed in Table 1. In addition, there are the
following theoretical curves: Emf06,22 Emf09,23 GarM09,24 and PASS.25
theory)‡ might be too low by about 10%. But Shimizu et al.33 checked the
accuracy of their stopping data by using He ions on their water jet target,
and by measuring proton energy loss and scattering from an Al wire; in
both cases, the results agreed with the data from ICRU 49.This discrepancy
is not yet understood.
As seen in Figure 2.2, the Jyväskylä measurement (Sn11) is in good
agreement with the GarM09, PASS, and BEST curves. The average value
of the normalized Sn11 points is 0.986, i.e., 0.6% below the BEST curve.
This corresponds to an experimental I value of 81.7 eV, but with a rather
large error of 28 eV, due to the systematic error of 4.6% stated by Siiskonen
et al.28 for the stopping power values.
‡ The
low value is due to an unusually large shell correction in the Emf09 theory (priv. communication
from D. Emfietzoglou to H. Paul, 2012).
Figure 2.2 Electronic stopping power of liquid water for protons, normalized by the
data from the proton table of ICRU Report 49.2 The designations for tables and for
experimental points are as in Figure 2.1. I values are shown in parentheses. The correc-
tion to Eq. (2) is −0.68% at 10 MeV, and even smaller beyond.
For He ions in liquid water, there are four data sets in the literature
spanning the range from 300 to 7500 keV (see Ref. 21). The data points
agree with the curve of ICRU 49 within 3%, on the average; this c orresponds
roughly to the experimental accuracies stated.
For heavier ions, there is only one data set: the stopping power for Li
ions of 0.025–0.1 MeV/nucleon, measured by the inverted Doppler shift
attenuation method.34
3.2 Range measurements for water, and mean

ionization energy
Evidently, stopping power measurements are not yet sufficiently accurate for a
precise determination of the I value of water, but one might hope that range
measurements are better suited. Table 2.2 gives an overview of calculated
and measured values of the mean ionization energy of liquid water,31 based
mostly on range and ionization measurements, and on theory (using optical
data). On the basis of the data available in 1984, the value I = 75.0 eV was
46 Helmut Paul
Table 2.2 I values of liquid water. The missing references can be found in Ref. 31
I (eV) Reference Method or remark
75.4 ± 1.9 Thompson, 1952 Range, 340–200 MeV p,
assuming. ICu = 322 eV
74.6 ± 2.7 Nordin et al. 1979 Stopping power, 60 MeV pions
75 Ritchie et al. 1978 Dielectric response function
75.4 Ashley, 1982 Dielectric response function
81.77 Janni, 1982 Averaging data for H and O
79.7 ± 2 Bichsel et al. 1992 Ionization curves, 70 MeV p
81.8 Dingfelder et al. 1998 Dielectric response function
80.0 ± 1.3 Bichsela et al. 200037 Ionization curves, C ions,
290 MeV/u
77 Kramer et al. 2000 Depth dose curves for C ions
78.4 ± 1 Kumazaki et al. 200729 Depth dose curves for protons
77.8 ± 1 Emfietzoglou et al. 200923 Dielectric response function
79.4 Garcia-Molina et al. 200924 Dielectric response, MELF-
GOS model
75.0 ± 3 Chosen in ICRU 37 and 49
78.0 Chosen in Sigmund et al.12 Erratum; replaces the value
67.2 eV implicit in ICRU
Report 73
a The error is quite small due to the high precision of measurements (±0.05 mm).
chosen in ICRU Report 3735 and again in ICRU Report 49.2 But evidently,
all the more recent determinations indicate a larger I value.
Using a precision water column, Schardt et al.36 have recently mea-
sured Bragg peak positions for 1H, 7Li, 12C, and 16O beams at altogether
22 different energies, see Tables 2.3 and 2.4. To deduce mean ranges from
Bragg peak positions, they used Bichsel’s37 empirical rule that the mean
range is the depth at which the Bragg curve has dropped to p = 82% of its
maximum value. Bichsel et al.37 deduced this rule for high energy carbon
ions. The value p = 80% was originally derived by Berger38 for protons,
and it has been confirmed by Bortfeld.39 Figure 2.3 shows a comparison
of an analytical calculation by Bichsel40 with a Monte Carlo calculation by
Gudowska41 for 200 MeV protons in water.§ The values found (p = 82%
and p = 80%) are in fair agreement with Berger’s value. Schardt et al.,42
§ Allattempts to determine p by means of SRIM Monte Carlo calculations gave lower values: p = 76%
for 200 MeV protons, and even lower values for carbon ions. This probably points to some error in
the scattering cross sections used in SRIM.
Table 2.3 Bragg peak (BP) positions of Li, C and O ion beams in water measured by
Schardt et al.36 at a density of 0.997 g/cm3 (error 0.20 mm), deduced mean ranges, and
comparison to the ICRU 73 range tablea(based on I = 78 eV)12
Ion v2 Spec. energy BP pos. Mean range ICRU12 Ratio
(MeV/u) (MeV/A1) (mm) (mm) (g/cm2) (g/cm2)
7Li 100 100.201 60.65 60.89 6.0707 6.031 0.9935
200 200.410 203.7 204.5 20.389 20.206 0.9910
12C 100 100 26.16 26.26 2.6181 2.601 0.9935
150 150 53.16 53.32 5.316 5.296 0.9962
200 200 87.22 87.49 8.7228 8.705 0.9980
250 250 127.32 127.7 12.732 12.72 0.9991
300 300 172.50 172.9 17.238 17.24 1.0001
400 400 275.36 276.1 27.527 27.51 0.9994
16O 200 199.881 65.44 65.6 6.5403 6.514 0.9960
250 249.852 95.43 95.67 9.5383 9.513 0.9973
300 299.822 129.27 129.6 12.921 12.888 0.9974
400 399.763 206.24 206.7 20.608 20.572 0.9983
Average 0.9966 ±
0.0028
a Only those measured values are shown that are close to calculated range values in the ICRU table.
Table 2.4 Bragg peak (BP) positions (error 0.2 mm) and deduced ranges for protons
in water measured by Schardt et al.36 at a density of 0.997 g/cm3, compared to a
combination of the ICRU 49 table with BEST (I = 78 eV)
Ion v2 Spec. energy BP pos. Mean range ICRU49& Ratio
(MeV/u) (MeV/A1) (mm) (mm) (g/cm2) BEST(78)
p 125 125.91 115.8 117 11.665 11.664 0.9999
150 151.092 159.4 161 16.052 16.056 1.0002
175 176.274 208.4 210.3 20.967 20.986 1.0009
200 201.456 262.1 264.6 26.381 26.410 1.0011
Average 1.00054 ±
0.00055
using CR39 sheets to count particles, obtained p = 85% for 3He ions in
water at 170 MeV/u; they state that this agrees within uncertainties with
Bichsel’s p = 82%. It seems appropriate therefore to assume an average
value p = (82 ± 3)% for this conversion. Using the shape of the Bragg peak
(see Fig. 2.3), this leads to an (additional) error of 0.08% or 0.2 mm, to be
added to the error of the Bragg peak position. Assuming that both error
48 Helmut Paul
Figure 2.3 Bragg curve (ionization per unit distance) and integral range distribution for
protons of 200 MeV in water, as calculated by Gudowska41 (dashed) and by Bichsel40
(full lines), respectively.
contributions are systematic, we arrive, conservatively, at a total range error

of 0.4 mm.
Table 2.3 also lists the calculated ranges from the ICRU Erratum12 and
compares them to the measured ranges. Evidently, the calculated ranges
(for I = 78 eV) are (0.34 ± 0.28)% below measured ranges, on the average.**
To do a proper comparison between measured ranges and theoretical
stopping power it would seem necessary to use a theory that has been
tested against experiment. But unfortunately, there are no stopping power
measurements for C and O ions, and almost none for Li ions, so a compari-
son is presently impossible. It seems better therefore, to resort to Schardt’s
proton measurements (Table 2.4), since here, comparisons with stopping
power measurements are possible (see Figs. 2.1 and 2.2). For this purpose,
we use the table from ICRU Report 49, merge it with the BEST curve for
78 eV at about 500 keV and integrate the reciprocal.††
** The standard deviations given here are always based on Eq. (9).
†† Actually, the part of ICRU 49 makes an insignificant contribution to the integral.
Table 2.5 Like Table 2.4, but comparing to range values obtained from PASS
(I = 78 eV) by integration
Ion v2 Spec. energy Bragg peak Mean range PASS Ratio
(MeV/u) (MeV/A1) pos. (mm) (mm) (g/cm2) (I = 78 eV)
p 125 125.91 115.8 117 11.665 11.722 1.0049
150 151.092 159.4 161 16.052 16.134 1.0051
175 176.274 208.4 210.3 20.967 21.086 1.0057
200 201.456 262.1 264.6 26.381 26.532 1.0057
Average 1.0053 ±
0.0004
The resulting ratio (1.00054 ± 0.00055) shows complete agreement

with BEST.
Note that irrespective of the value of I, the range of the ions, with an
accuracy of 0.4 mm, may be taken from Tables 2.3 and 2.4.
An additional comparison using PASS is shown in Table 2.5, which leads
to a slightly larger ratio. A simple average of the ratios 0.9966, 1.0005, and
1.0053 from Tables 2.3–2.5 leads to an average ratio 1.0008 between calcu-
lated and measured ranges, with an estimated error of 0.008 (including the
error of the range measurements), and hence,‡‡ to a value I = (78.5 ± 5) eV,
in agreement with the unweighted average I = (79.2 ± 1.6) eV of the val-
ues§§ from the period 1992 to 2007 listed in Table 2.2.
4. OTHER TARGET SUBSTANCES AND STATISTICAL

COMPARISONS
Many other substances are of interest in hadron therapy. A list of
compositions of many non-elemental substances can be found in connec-
tion to the ESTAR program at NIST.43 For some of these, stopping powers
are given in ICRU Reports 49 and 73. Bragg functions and ranges have
been measured for several substances by Bichsel et al. 37 Average I values for
‡‡ The conversion from range error to error of I is conveniently done using program BEST in which I
can be changed. This comparison shows that the relative error of I is 8 times larger than the relative
error of range.
§§ The Report Committee of the ICRU is probably arriving at a value around I = 78 eV (pers. comm.
from Dr. Fernandez-Varea to H. Paul, February. 2012.

50 Helmut Paul
various “soft tissues” have been presented in Ref. 44, showing considerable
variation.
Moyers et al.45 have recently measured the linear stopping powers for
protons at 135, 175, and 225 MeV in many compounds of interest to par-
ticle therapy, relative to a water target. They compared their results to the
Janni13 or LET15 tables, finding agreement within 1 to 3%. As examples,
Figure 2.4 shows a few results by Moyers et al., compared to the Janni,
BEST, and SRIM tables. The BEST calculation uses the I-values of ICRU
Report 49, except that I = 78 eV was taken for water.
Inspection of Figure 2.4 shows that the curves are essentially d etermined
by the I values. In particular, the Janni curves are always above the BEST
curves because of Janni’s rather high I value for water and his low I values
for the other substances. Evidently, BEST agrees best with the Al measure-
ments. For the compounds, BEST appears slightly low; this might point
to slight errors of the I values used. Even so, if such high energy stopping
powers are required and if I values are known or calculable using Eq. (4), a
calculation using BEST should be useful.
Figure 2.4 The linear stopping power of Al, clear polystyrene (CLPS), high density poly-
ethylene (HDPE), and polymethyl methacrylate (PMMA), for protons relative to that of
water, compared to the tables Janni, BEST, and SRIM. The 3-digit ID numbers from ICRU
49 are shown in parentheses. For the curves, the I values for both substances are shown
in parentheses, where available. Experimental data are from Moyers et al.45, 47
A fairly complete collection of experimental stopping powers and many

graphical presentations can be found in Ref. 21. Additional information is
given in Ref. 31.
To determine the accuracy of various stopping tables and programs by
statistics, we use our program ‘‘Judge”.46 This program calculates the nor-
malized differences
δ = (Sexp − Stab )/Sexp (7)
for every data point. Here, Sexp is the experimental value, and Stab the cor-
responding table value for the same ion, same target, and same energy. In
every range of specific energy, i.e., energy per nucleon, it then determines
the average normalized difference:
� = �δ� (8)
and its standard deviation
(9)

σ = �δ 2 � − �δ�2 .
The averages are unweighted, except that obviously discrepant data are
rejected (see “Statistical analysis” in Ref. 21). A small Δ usually signifies
good agreement between table and experimental data; in such a case, σ is
related to the mean experimental accuracy, and σ may be taken as a measure
of the accuracy of the table, as determined from experiment.
In the following, “condensed” or “gaseous” means a substance con-
densed or gaseous at normal temperature and pressure.
Note that all the experimental data analyzed in the following chapter
can be downloaded from our collection.21
4.1 Statistical comparisons for H and He ions

In Tables 2.6 and 2.7, the reliability of various stopping power tables for
H and He ions in solid elements is given in terms of Δ ± σ. Here, E is
the energy and A1 is the mass number of the ion. These tables were origi-
nally published by Paul and Schinner,32 but many new data have since
been added.31 This has not changed the results much, but it adds to the
reliability.
One can see that σ always decreases with increasing energy, due to the
higher accuracy of measurements at high energy. The numbers of experi-
mental points averaged are also shown, to give an idea of the accuracy.
52 Helmut Paul
Table 2.6 Mean normalized deviations Δ ± σ (in %) for H ions in 17 solid elements
(Ag, Al, Au, Be, C, Cu, Fe, Gd, Ge, Mo, Pb, Pt, Si, Sn, Ti, U, W) covered by the ICRU
Table, compared to various tables
E/A1 (MeV) 0.01–0.1 0.1–1 1–10 10–100 0.01–100
No. of points 1357 2492 1212 225 5286
Janni, 198213 2.1 ± 11 −1.1 ± 7.1 −0.9 ± 3.6 −0.3 ± 0.5 −0.2 ± 7.7
Ziegler et al., −1.3 ± 11 −3.1 ± 7.8 −0.4 ± 4.2 0.4 ± 2.2 −1.9 ± 8.2
198519
ICRU, 19932 0.8 ± 11 −0.7 ± 7.0 −0.3 ± 4.0 −0.1 ± 0.5 −0.2 ± 7.5
SRIM, 200318 0.6 ± 10.3 −0.9 ± 6.7 −0.6 ± 3.7 −0.2 ± 0.6 −0.4 ± 7.2
Table 2.7 Mean normalized deviations Δ ± σ (in %) for He ions in the same solid
elements as in Table 2.6 covered by the ICRU Table (except for U), compared to vari-
ous tables
E/A1 (MeV) 0.01–0.1 0.1–1 1–10 10–100 0.01–100
No. of points 1036 1913 400 11 3360
Ziegler et al., 3.2 ± 8.7 0.6 ± 5.6 −0.8 ± 3.3 0.8 ± 2.4 1.2 ± 6.7
198519
ICRU, 19932 2.6 ± 8.3 0.2 ± 5.6 0.1 ± 3.3 0.9 ± 0.9 0.9 ± 6.4
SRIM, 200318 3.5 ± 8.2 0.6 ± 5.2 −0.3 ± 3.1 0.2 ± 0.9 1.4 ± 6.3
To provide a fair comparison with the smaller number of targets in the

ICRU table, we compare only with the targets of that table, even though
we have many more targets in our files.21 We see that generally, σ has
decreased and hence, the overall agreement between the experimental
data and the tables has improved in time, with the exception of the TRIM
program by Ziegler et al. (1985); but this was the first method capable of
treating all ions and all targets.
Table 2.8 gives results for H ions in elemental gases. Here, we exclude
measurements for low energy H ions in helium (see Ref. 31). Due to
the threshold effect48 these data would produce a very large Δ and thus
obscure any other discrepancy. Except for the tables by Ziegler et al. (1985)
(due to large discrepancies for H and He targets), the gas measurements
appear here more reliable than those on solids.
Table 2.9 shows results for He ions in elemental gases. Again, the agree-
ment with the data is much better than for solids, and we can observe a
gradual improvement in time.
Table 2.8 Mean normalized difference Δ ± σ (in %) for H ions in all elemental gases
except F, Cl, Rn
E/A1 (MeV) 0.001–0.01 0.01–0.1 0.1–1.0 1–10 10–100 0.001–100
No. of 124 335 535 303 11 1308
points
Janni, −0.9 ± 9.2 −0.0 ± 4.6 0.5 ± 3.9 0.9 ± 3.2 3.2 ± 0.6 0.4 ± 4.7
198213
Ziegler 22 ± 14 22 ± 11 0.4 ± 6.8 −1.1 ± 1.7 −1.0 ± 0.5 7.7 ± 14
et al.,
198519
ICRU, −0.6 ± 6.7 −1.2 ± 5.0 −1.2 ± 3.7 −0.8 ± 1.6 −0.2 ± 0.5−1.0 ± 4.1
19932
SRIM, 2.1 ± 5.2 −0.1 ± 4.7 −0.4 ± 3.6 −0.2 ± 1.6 0.2 ± 0.3−0.1 ± 3.9
200318
Table 2.9 Mean normalized difference Δ ± σ (in %) for He ions in all elemental gases
except F, Cl, Rn
E/A1 (MeV) 0.001–0.01 0.01–0.1 0.1–1.0 1–10 0–10
No. of points 5 267 863 238 1373
Ziegler et al. 7.2 ± 13 2.5 ± 5.9 3.0 ± 4.9 −0.5 ± 2.5 2.3 ± 5.0
198519
ICRU, 1993 0.5 ± 6.8 −1.0 ± 4.2 0.1 ± 4.2 0.7 ± 2.3 0.0 ± 4.0
SRIM, 2003 −5.4 ± 6.1 0.3 ± 3.9 0.1 ± 3.8 −0.2 ± 2.2 0.1 ± 3.7
Data for compounds have been treated in Ref. 49. In our data base,21
we have data for 150 different compounds. Table 2.10 shows results for
hydrogen and helium ions in some of these compounds, compared to
SRIM. Some of the normally liquid substances have been measured also in
gaseous form at reduced pressure. Because of the different low energy limit
chosen,*** some of the results appear somewhat better than for elements.
Again, the errors σ tend to be smaller for gases than for solids.
For the SRIM calculations in Table 2.10, Ziegler’s built-in corrections
for dependence on the target phase, and his corrections for deviations
from the Bragg additivity rule Eq. (3) have been applied. These corrections
amount to a few percent and exist only below 1 MeV/nucleon.49
Table 2.11 shows a comparison between SRIM 2003 and ICRU
Report 49, for the smaller number of compounds covered by the latter
*** This is to avoid large deviations due to the threshold effect in LiF.54
54 Helmut Paul
Table 2.10 Mean normalized deviations Δ ± σ (in %) for H and He ions in condensed
or gaseous compounds, as compared to SRIM (2003)
Ions Targets E/A1 (MeV) 0.025–0.25 0.25–2.5 2.5–30 0.025–30
H cond.a No. of pts 441 947 251 1639
Δ ± σ −1.1 ± 8.8 1.4 ± 6.3 −0.1 ± 3.9 0.5 ± 6.9
gasb No. of pts 508 378 24 910
Δ ± σ −0.9 ± 4.3 0.1 ± 3.3 −0.9 ± 2.1 −0.5 ± 3.9
He cond.c No. of pts 479 1471 14 1964
Δ ± σ 0.4 ± 6.7 −0.5 ± 4.3 −2.0 ± 3.1 −0.3 ± 5.0
gasd No. of pts 997 1742 0 2739
Δ ± σ −2.6 ± 7.2 1.1 ± 2.9 −0.3 ± 5.2
a A150 tissue equiv. plastic, Al2O3, Anthracene, CaF2, CdTe, CR39 nuclear track material, D2O, Er2O3,
Formvar, GaAs, GaN, GaP, Graphite, H2O, Havar, In2O3, InP, KCl, LR115 cellulose nitrate, LiF,
LiNbO3, MuMetal, Mylar, Permalloy, Pliolite S-5A, Polyvinyltoluene, Polycarbonate, Polyethylene,
Polyimide, Polypropylene, Polystyrene, Polysulfone, Sc2O3, SiO2, Styrene, Terphenyl, TiO2, Vyns,
ZnSe, ZnSiP2, ZnTe.
b Air, Butane, Butadiene, Butene, C H , C H , C H , C H , CCl , CH , CO , Cyclopropane, Decane,
2 2 2 6 3 8 6 6 4 4 2
Ethylene, H2O vapor, H2S, Heptane, Hexane, Heptyne, Hexyne, N2O, NH3, NO, Nonane, Octane,
Pentane, Pentyne, Propelene, Tissue equivalent gas.
c Al O , BaCl , BaF , C BrF, CCl , CdTe, CH Br, CH I, Chloroform, CO solid, CR39, Cyclohexane,
2 3 2 2 2 4 5 5 2
Dichloromethane, Ethyl alcohol, Er2O3, Ethyl cellulose, Formvar, GaAs, Graphite, H2O, Havar,
Heptane, Hexane, InP, LR-115, LiF, Methyl alcohol, MuMetal, Mylar, NE-111 scintillator, Octane,
Propyl alcohol, Pentane, Pliolite S-5A, Polyvinyltoluene, Polycarbonate, Polyethylene, Polyimide,
Polypropylene, Polystyrene, Polysulfone, SiC, SiO2, Ta2O5, Teflon, TiN1.1O0.27, UO2, Vyns, WO3,
ZnSe. ZnTe.
d Acetaldehyde, Acetone, Air, Allene, Benzene, Butyraldehyde, Butane, Butadiene, Butene, 2-Butanone,
Butyne, C2F6, C2H2, C2H2F2, C2H4F2, C2H6, C3F8, C3H8, C4F8, CBrF3, CCl2F2, CCl4, CClF3,
CF4, CH3Br, CH4, CHCl2F, Chloroform, CO2, CS2, Cyclohexane, Cyclooctane, Cyclopropane,
Cyclopentane, Cyclohexadiene, Cyclohexene, Cyclohexanone, Cyclopentene, Dichloromethane,
Diethyl ether, Dioxane, Dimethylamine, Dimethyldisulfide, Dimethyl ether, Dimethyl sulfide,
Ethyl alcohol, Ethylamine, Ethynylbenzene, Ethylene oxide, Ethylene sulfide, Ethylene, H2O, H2S,
Heptane, Hexane, Hexene, Heptyne, Hexyne, Methyl alcohol, Methylamine, N2O, NH3, NO,
Octane, Propyl alcohol, Pentane, Pentene, Pentanone, Pentyne, Propelene, Propyne, Propylene oxide,
Propylene s ulfide, SF6, SO2, Tissue eq. gas, Thiophene, Trimethylamine, Trimethylene sulfide, Toluene,
Vinylmethyl ether.
Table 2.11 Mean normalized deviations Δ ± σ (in %) for H and He ions in 23 (solid or
gaseous) compounds covered by ICRU Report 49
E/A1 (MeV) 0–0.03 0.03–0.3 0.3–3.0 3–30 0–30
No. of points 116 1036 1237 135 2524
ICRU, 1993 0.2 ± 8.9 1.4 ± 5.9 1.3 ± 5.2 1.0 ± 4.4 1.3 ± 5.7
SRIM, 2003 −7.8 ± 12 −1.0 ± 6.4 0.4 ± 5.6 −0.6 ± 4.0 −0.6 ± 6.6
table, for H and He ions together.49 For this restricted number of targets,
ICRU Report 49 is clearly better than SRIM.
4.2 Application to therapy using H ions

Inspection of Tables 2.6 and 2.8 shows that, for protons in elements, in the
range 10–100 MeV, the value of Δ is negligible for the ICRU and SRIM
tables, and σ is 0.5%, on the average. Hence, in this energy range important
for therapy, the ICRU and SRIM tables can be expected to be accurate
to 0.5%. And the same accuracy may be expected up to 1000 MeV, if the
ICRU or SRIM tables are extended††† using the pure Bethe theory Eq. (2),
since the corrections to Bethe are minimal (cf. Fig. 2.2).
For protons in compounds, the highest energy range (Table 2.10) goes
only up to 30 MeV, and σ is larger (2–4%). Hence, the predictive quality
of SRIM appears worse for compounds. On the other hand, since Bragg’s
additivity holds at high energy, the stopping power of compounds at high
energy may be calculated using Eq. (3), and in this way, the accuracy could
be improved.
In therapy applications, the particle range is of greatest importance.
The range for therapeutically relevant energies depends essentially on the I
value, as has been discussed in Section 3.2 above. Range uncertainties have
recently been considered by Andreo44 and by Paganetti.50 Unfortunately,
Andreo’s paper is based on an unrealistic variation of possible I values for
water (67, 75, and 80 eV), due to the fact that his paper was published
before the ICRU value I = 67 eV was officially declared as erroneous.12
In the discussion by Paganetti,50 the low value I = 67 eV has already been
discarded.
Using our numbers from Section 3.2 (I = 79.2 eV with an error of 2.0%,
and a conversion factor of 8), we arrive at a range error in water (due to the
uncertainty of I) of 0.25%. Paganetti50 arrives at the much larger estimate
of 1.5% (for 1.5 standard deviations) for tissues because of the much larger
variability of I values for tissues (see, e.g., Figs. 2.3 and 2.4 by Andreo).44
Concerning the conversion from dose distribution to range
(see Section. 3.2), Paganetti states that ideally, p = 80% should be taken for
proton therapy, but that, for historic reasons, p = 90% is used in most proton
therapy facilities.
Monte Carlo programs like SHIELD-HIT51 are an important tool
to describe the passage of ions in all cases relevant to hadron therapy.
††† In the case of ICRU, this simply means using the ICRU table up to 1000 MeV.
56 Helmut Paul
Figure 2.5 Electronic stopping power as a function of specific energy for C-ions in C, com-
pared to various tables (cf. Table 2.1). Experimental points are marked by letters; the refer-
ences corresponding to the reference codes given in the margin can be found in Ref. 21.
Such programs require a description of the stopping power over a large

range of energies: the lowest energies are important for a proper descrip-
tion of the Bragg peak. But as we have seen above (Figs. 2.1 and 2.2),
there is no reliable theory or table for the entire range of energy required.
Hopefully, the Continuum Distorted Wave Method52 will help to bridge
this gap.
4.3 Statistical comparisons for carbon ions

Because of the importance of carbon ions for therapy, we treat these
separately. As an example, Figure 2.5 shows stopping powers for carbon ions
in carbon. Here, there is good agreement between the experimental data
and the MSTAR, SRIM, HISTOP, and ICRU 73 tables in most energy
regions, while CasP is too low at low energy.‡‡‡
Table 2.12 shows statistical results for carbon ions in solid elements
covered by MSTAR. Since ICRU Report 73 does not treat Gd and Ta
targets, the number of points is somewhat smaller for that table. The results
‡‡‡ This discrepancy has not changed much from CasP v. 3.1 to v. 5.0.
are similar to those shown in Ref. 53, except that there, ions from 3Li to
18Ar were treated together. We find again that MSTAR and SRIM describe
the data about equally well, but that the values from ICRU Report 73 are
too high,§§§ on the average, especially at low energies.
Similarly, Table 2.13 shows results for carbon ions in condensed
compounds.
Table 2.14 shows results for carbon ions in gases. This table is of limited
usefulness because of the small number of measured data points available
for carbon ions that leads to rather erratic results. Possibly, if better results
Table 2.12 Mean normalized deviations Δ ± σ (in %) for carbon ions in Ag, Al, Au, Be,
C, Cu, (Gd), Ge, Mo, Ni, Pb, Si, Sn, (Ta), and Ti
E/A1 (MeV) 0.025–0.1 0.1–1.0 1–10 10–100 0.025–100
No. of points 202 632 229 8 1071
MSTAR −1.6 ± 9.6 0.6 ± 5.8 0.9 ± 5.1 0.0 ± 2.8 0.2 ± 6.6
SRIM, 2003 0.4 ± 8.3 −0.5 ± 5.3 −0.6 ± 5.2 1.0 ± 3.0 −0.3 ± 6.0
ICRU −13 ± 12 −9.2 ± 10.9 −2.6 ± 5.8 −0.6 ± 3.8 −8.5 ± 10.8
Rep. 73
Table 2.13 Mean normalized deviations Δ ± σ (in %) for carbon ions in condensed
compounds (Al2O3, Kapton, Mylar, Polycarbonate, Polyethylene, Polypropylene (not
covered by ICRU 73), PVC, SiO2)
E/A1 (MeV) 0.025–0.1 0.1–1 1–10 10–100 0.025–100
No. of points 27 95 98 0 220
MSTAR 4.2 ± 8.8 −1.9 ± 5.3 3.8 ± 2.9 1.4 ± 5.8
SRIM, 2003 0.5 ± 5.4 −2.9 ± 4.4 −1.8 ± 3.4 −2.0 ± 4.3
ICRU Rep. 73 −6.9 ± 8.3 −8.1 ± 4.9 −2.6 ± 3.3 −5.8 ± 5.6
Table 2.14 Mean normalized deviations Δ ± σ (in %) for carbon ions in gases (Air, Ar,
H2, He, N2, Ne, O2)
E/A1 (MeV) 0.025–0.1 0.1–1 1–10 10–100 0.025–100
No. of points 13 2 38 4 57
MSTAR −3.2 ± 8.9 0.6 ± 2.4 −1.1 ± 3.4 0.1 ± 4.2 −1.4 ± 5.3
SRIM, 2003 9.7 ± 7.1 −22 ± 2.3 −1.4 ± 7.2 −1.2 ± 8.3 0.4 ± 9.5
ICRU Rep. 73 −42 ± 21 −16 ± 0.7 −2.6 ± 5.0 0.4 ± 4.2 −12 ± 20
§§§ Note that a negative D means that the table value is too high, according to the definition Eq. (7).
58 Helmut Paul
are required, the use of Table 2.3 from Ref. 53 (for ions from 3Li to 18Ar)
may be preferable.
5. CONCLUSIONS
An overview of available tables and computer programs d escribing
the stopping power of matter for positive ions is given. Experimental
and table values for the stopping power of water for hydrogen ions are
discussed in detail. An overview of measured and calculated values for the
mean ionization energy of water is given; the average of recent values is
found to be I = (79.2 ± 1.6) eV. The recent measurements of Bragg peak
positions for light ions in water by Schardt et al. are discussed in detail;
due to the uncertainty of converting these positions to range values, the I
value 78.5 eV deduced from these is found to have the rather large error
of 5 eV. Statistical comparisons of measured stopping powers with various
tables for H, He, and C ions for many targets are given. The SRIM, ICRU
49, and MSTAR tables are found to describe the data best, on the average,
but SRIM describes many more cases than the other tables. For protons of
10–100 MeV in elements, it is found that the ICRU 49 and SRIM tables
may be expected to be accurate to 0.5%.
6. LIST OF ACRONYMS
ASTAR NIST program for alpha stopping powers

ATIMA Atomic Interactions in MAtter (a program)
BEST BEthe Stopping program
BP Bragg Peak
CasP Convolution approximation for swift Particles
CR39 Nuclear track detector material
CSDA Continuous slowing down approximation
ESTAR NIST program for electron stopping powers
HISTOP Heavy Ion Stopping program
ICRU International Commission on Radiation Units and
Measurements
ICRU 49 ICRU Report 49 for protons and alphas
ICRU 73 ICRU Report 73 for ions from 3Li to 18Ar
Judge Statistical analysis program by Paul & Schinner
LET Linear energy transfer

LS Lindhard-Sørensen
MSTAR MoreSTAR (going beyond PSTAR and ASTAR)
NIST US National Institute of Standards and Technology
PASS Peter-Andreas-Sigmund-Schinner
PSTAR NIST program for proton stopping powers
SHIELD Code for the simulation of the interaction of hadrons and
arbitrary atomic nuclei with complex extended targets
SHIELD-HIT Medical version of the code SHIELD
SRIM The Stopping and Range of Ions in Matter
REFERENCES
1. Bethe, H. Bremsformel für Elektronen relativistischer Geschwindigkeit. Z. Phys. 1932,
76, 293.
2. ICRU Report 49, Stopping powers and ranges for protons and alpha particles,
International Commission on Radiation Units and Measurements, Bethesda, MD, USA,
1993.
3. Bragg, W. H.; Kleeman, R. Range of ALPHA-particles of radium. Phil. Mag. 1905, 10,
318.
4. Paul, H. Some new results on stopping power for fast ions. AIP Conf. Proc. 2009, 1099,
251.10.1063/1.3120025
5. Lindhard, J.; Sørensen, A. H. Relativistic theory of stopping for heavy ions. Phys. Rev. A
1996, 53, 2443.
6. Geissel, H.; Scheidenberger, Ch.; Malzacher, P.; Kunzendorf, J; Weick, H. ATIMA.
<http://www-linux.gsi.de/~weick/atima/>.
7. Berger, M. J.; Bichsel, H. BEST, BEthe Stopping power program, unpublished program
recieved from M.J.Berger 1993.
8. Grande, P. L.; Schiwietz, G. CasP. In Advances in Quantum Chemistry, Cabrera-Trujillo, R.,
Sabin, J. R., Eds.; Vol. 45 p. 7, Elsevier, 2004. Program. <http://www.helmholtz-berlin.
de/people/gregor-schiwietz/casp_en.html>.
9. Arista, N. R.; Lifschitz, A. F. Non-Linear Approach to the Energy Loss of Ions in Solids.
In Advances in Quantum Chemistry, Sabin, J., Ed., 45 (2004), 47, Elsevier Inc.
10. Hubert, F.; Bimbot, R.; Gauvin, H. Range and stopping-power tables for 2.5–500 MeV/
nucleon heavy ions in solids. At. Data Nucl. Data Tables 1990, 46, 1.
11. ICRU Report 73, Stopping of ions heavier than Helium, J. ICRU 2005, 5, 1.
12. Sigmund, P.; Schinner, A.; Paul, H. Errata and Addenda for ICRU Report 73, J. ICRU
2009, 5,1.
13. Janni, J. F. Proton range-energy tables. Atomic Data Nucl. Data Tables 1982, 27, 147.
14. Zajic, V.; Thieberger, P. Heavy ion linear energy transfer measurements during single
event upset testing of electronic devices. IEEE Trans. Nucl. Sci. 1999, 46, 59.
15. Zajic, V. LET: Energy-LET-Range calculator v. 1.24. New York: Brookhaven National
Laboratory, 2001. <http://tvdg10.phy.bnl.gov/let.html>.
16. Paul, H. MSTAR version 3, 2003. <http://www.exphys.uni-linz.ac.at/stopping/

MSTARWWW/MSTAR312.zip>.
17. Sigmund, P.; Schinner, A. Binary theory of electronic stopping. Nucl. Instrum. Methods
Phys. Res., Sect. B 2002, 195, 64.
18. Ziegler, J. F. SRIM 2003–SRIM 2012. <http://www.srim.org>.
60 Helmut Paul
19. Ziegler, J. F.; Biersack, J. P.; Littmark, U. The stopping and range of ions in matter Vol.1;
Pergamon: New York, 1985.0-08-021603-X
20. Scheidenberger, C., et al. Direct observation of systematic deviations from the Bethe
stopping theory for relativistic heavy ions. Phys. Rev. Lett. 1994, 73, 50.
21. Paul, H, Stopping power for light ions. <http://www.exphys.uni-linz.ac.at/stopping/>.
22. Emfietzoglou, D.; Nikjoo, H.; Pathak, A. Electronic cross sections for proton transport
in liquid water based on optical-data models. Nucl. Instrum. Methods Phys. Res., Sect. B
2006, 249, 26.
23. Emfietzoglou, D.; Garcia-Molina, R.; Kyriakou, I.; Abril, I.; Nikjoo, H. A dielectric
response study of the electronic stopping power of liquid water for energetic protons
and a new I value for water. Phys. Med. Biol. 2009, 54, 3451.
24. Garcia-Molina, R.; Abril, I.; Denton, C. D.; Heredia-Avalos, S.; Kyriakou, I.;

Emfietzoglou, D. Calculated depth-dose distributions for H+ and He+ beams in liquid
water. Nucl. Instrum. Methods Phys. Res., Sect. B 2009, 267, 2647.
25. Sigmund, P. Personal communication to H. Paul, 2011.
26. Shimizu, M.; Kaneda, M.; Hayakawa, T.; Tsuchida, H.; Itoh, A. Stopping cross sections
of liquid water for MeV energy protons. Nucl. Instrum. Methods Phys. Res., Sect. B 2009,
267, 2667.
27. Shimizu, M.; Hayakawa, T.; Kaneda, M.; Tsuchida, H.; Itoh, A. Stopping cross sections of
liquid water. Vacuum 2010, 84, 1002.
28. Siiskonen, T., et al. Energy loss measurement of protons in liquid water. Phys. Med. Biol.
2011, 56, 2367.
29. Kumazaki, Y.; Akagi, T.; Yanou, T.; Suga, D.; Hishikawa, Y.; Teshima, T. Determination of
the mean excitation energy of water from proton beam ranges. Radiat. Meas. 2007, 42,
1683.
30. Paul, H. Recent results in stopping power for positive ions, and some critical comments.
Nucl. Instrum. Methods Phys. Res., Sect. B 2010, 268, 3421.
31. Paul, H. The stopping Power of Matter for Positive Ions, in Modern Practices in Radiation
Therapy, Natanasabapathi G., Ed., InTech, ISBN 978-953-51-0427-8, 2012, Chapter 7.
<http://www.intechopen.com/books/modern-practices-in-radiation-therapy>.
32. Paul, H.; Schinner, A. Judging the reliability of stopping power tables and programs for
protons and alpha particles using statistical methods. Nucl. Instrum. Methods Phys. Res.,
Sect. B 2005, 227, 461.
33. Shimizu, M. Email to H. Paul, February, 2012.
34. Neuwirth, W.; Pietsch, W.; Richter, K.; Hauser, U. Electronic stopping cross sections of
elements and compounds for swift Li ions. Zeit. Physik A 1975, 275, 209.
35. ICRU Report 37, Stopping powers for electrons and positrons, International

Commission on Radiation Units and Measurements, Bethesda, MD, USA, 1984.
36. Schardt, D.; Steidl, P.; Krämer, M.;Weber, U.; Parodi, K.; Brons, S. Precision Bragg-Curve
measurements for light-ion beams in water, GSI Scientific Report 2007 (2008–1) 373,
and private communication from D. Schardt.
37. Bichsel, H.; Hiraoka, T.; Omata, K. Aspects of fast-ion dosimetry. Radiat. Res. 2000, 153,
208.
38. Berger, M. J. Penetration of proton beams through water. I. Depth-dose distribution,
spectra and LET distribution, Report NISTIR 5226, National Institute of standards and
technology, Gaithersburg, MD, 1993.
39. Bortfeld, T. An analytical approximation of the Bragg curve for therapeutic proton
beams. Med. Phys. 1997, 24, 2024.
40. Bichsel, H. Personal communication to H. Paul, 2008.
41. Gudowska, I. Personal communication to H. Paul, 2008.
42. Schardt, D. et al. unpublished (2004) (See Ref. 36).
43. <http://physics.nist.gov/PhysRefData/Star/Text/table1con.html>.
44. Andreo, P. On the clinical spatial resolution achievable with protons and heavier charged
particle radiotherapy beams. Phys. Med. Biol. 2009, 54, N205.
45. Moyers, M. F., et al. Ion stopping powers and CT numbers. Med. Dosimet. 2010, 35, 179.
46. Paul, H.; Schinner, A. An empirical approach to the stopping power of solids and gases
for ions from Li to Ar. Nucl. Instrum. Methods Phys. Res., Sect. B 2001, 179, 299.
47. Moyers, M. F. Personal communication to H. Paul, 2011.
48. Fermi, E.; Teller, E. The capture of negative mesotrons in matter. Phys. Rev. 1947, 72,
399.
49. Paul, H.; Schinner, A. Statistical analysis of stopping data for protons and alphas in com-
pounds. Nucl. Instrum. Methods Phys. Res., Sect. B 2006, 249, 1.
50. Paganetti, H. Range uncertainties in proton therapy and the role of Monte Carlo simu-
lations. Phys. Med. Biol. 2012, 57, R99.
51. Hansen, D. C.; Lühr, A.; Herrmann, R.; Sobolevsky, N.; Bassler, N. Recent improve-
ments in the SHIELD-HIT code. Int. J. Radiat. Biol. 2012, 88, 195.
52. Belkić, Dž. Review of theories on ionization in fast ion-atom collisions with prospects
for applications to hadron therapy. J. Math. Chem. 2010, 47, 1366.
53. Paul, H. A comparison of recent stopping power tables for light and medium-heavy
ions with experimental data, and applications to radiotherapy dosimetry. Nucl. Instrum.
Methods Phys. Res., Sect. B 2006, 247, 166.
54. Markin, S. N. et al., Phys. Rev. Lett. 2009, 103, 113201.
CHAPTER THREE
On the Determination of the

Mean Excitation Energy of Water
John R. Sabina,b, Jens Oddershedea,b and Stephan P.A. Sauerc
aDepartments of Physics and Chemistry, University of Florida, Gainesville, Florida 32611, USA
bInstitute
for Physics, Chemistry, and Pharmacology, University of Southern Denmark, 5230 Odense M,
Denmark
cDepartment of Chemistry, University of Copenhagen, Copenhagen, Denmark
Contents
1. Introduction 63
2. Some basic theory 65
3. Theoretical determination of I0 67
4. Experimental determination of I0 69
5. Conclusion 73
Acknowledgments 74
References 74
Abstract
Water is a ubiquitous substance in nature, and thus the mean excitation energy of
water is an important quantity for understanding and prediction of the details of many
fast ion/molecule collision processes such as those involved in external beam radio-
therapy of tumors. There are several methods for determining numerical values for a
mean excitation energy for water, both theoretical and experimental. Here the factors
affecting the determination of the value of the mean excitation energy of water, espe-
cially from experiment, are discussed.
1. INTRODUCTION
“The mean excitation energy of water is 75.0 eV.” What is the mean-
ing of such a statement?
As the use of fast ions for medical purposes, especially tumor therapy,
becomes more usual, the understanding of the basic physics of the interac-
tion of swift ions with biomaterial in combination with water at clinically
relevant energies (≈3–300 MeV/u) becomes more crucial. For example,
the end result of ion radiation should be destruction of tumors without,
64 John R. Sabin, Jens Oddershede and Stephan P.A. Sauer
or with only minimal, damage to surrounding tissue. Thus, understanding

of the dosimetry for the interaction of fast ions with tissue becomes of
primary importance for prediction of dose-depth curves and the spatial
resolution of such radiotherapy beams.1 However there are uncertainties2
which also must be addressed.
As biological tissue is predominantly water, and as radiation damage to
cells is predominantly due to reaction of secondary water fragments with
cellular biomolecules,3, 4 it is of primary importance to understand the
interaction of fast ions with water, in particular the details of the energy
transfer from the fast ion to water. In this process, the kinetic energy of
the fast ion is transferred to water, resulting in electronic and vibrational
excitation, ionization, and/or fragmentation. In this regard, an accurate
determination of the ratio of ion beam energy deposition per unit path-
length in water to that in air [the stopping power ratio—STPR] is the most
important quantity for external beam radiotherapy dosimetry according to
the Code of Practice of the International Atomic Energy Agency.5 This
ratio thus becomes critically important for accurate clinical dosimetry,6 as
it is this ratio that determines the mean projected range and thus the dose-
depth curves. The ratio of the mass stopping power for an ion in water and
in air, swater, 6, 7
air, is conventionally written:
0
(Se (E)/ρ)water (Z2 /A2 )water ln(2mv 2 /I0water )

0
swater, air = = , (1)
(Se (E)/ρ)air (Z2 /A2 )air ln(2mv 2 /I0air )
where Z2 and A2 are the electron number and relative mass of the target,
ρ is the scatterer density of the medium, Se (E) is the stopping power at
projectile energy E, and m and v are the projectile mass and velocity. Thus,
the quantity I0 becomes the critical quantity. I0 measures the ease with
which energy can be deposited by the ion in the target material; in other
words, I0 is a measure of the ability of a target system to absorb kinetic
energy from a projectile, and is called the mean excitation energy of the
target (sometimes referred to as the average ionization potential). I0 is thus
central to the determination of the stopping power of the target. The mean
excitation energy, which is specific to each target, may be obtained either
theoretically or experimentally, and “Thereby hangs a tale”.8 Although the
mean excitation energy is a well-defined quantity (vide infra), values of the
mean excitation energy of water obtained by experimental and theoretical
means are found to vary by as much as 20% (vide infra), with those values
deduced from experiment generally being higher. Such a large uncertainty
On the Determination of the Mean Excitation Energy of Water 65
is unacceptable for many applications, including external beam radio-

therapy dosimetry. To quote Gottschalk:9 “Unfortunately, I0 is particularly
problematic for water.”
In this report we explore one aspect of that problem; namely the deter-
mination of the mean excitation energy of water, a quantity necessary
to the understanding of the physics of the characteristics of ion energy
deposition in water. Which values should be used for understanding and
predicting external beam radiotherapy dosimetry and dose/depth curves?
The theoretical or the experimental? We maintain that the value of the
mean excitation energy obtained from theory and the value extracted from
stopping experiments can be quite different, and that their usage should
not be confused. Although there is only one mean excitation energy for
any target substance such as water in a particular phase, we will refer to the
values extracted from theory as I0theo and the values extracted from experi-
ment as I0 exp in order to be clear as to the origin of the value of the mean
excitation energy under discussion.
2. SOME BASIC THEORY

The energy loss per unit pathlength in a material, or stopping power
of the material, for a fast ion is generally written as a function of projectile
velocity (ν):
dE 4 π n e 4 Z12 Z2
− = n S (v) = L (v) , (2)
dx me v 2
where S(v) is referred to as the stopping cross section, n is the target scat-
terer density, Z1 and Z2 are the projectile charge and target electron number,
respectively, and me is the electron rest mass. L( v), known as the target
stopping number, is frequently expanded in a Born series in the projectile
charge, which, for fixed projectile charge, can be written as:

L(v) = Z1i Li (v). (3)
i=0
Each term can then be expanded, suppressing ν, as:

Li = Lij . (4)
j=0
In the following, we ignore relativistic − ln(1 − β 2 ) − β 2 and density

10
(δ/2) corrections which are small in the energy range of interest here.
Although there are several ways to express L , Li, and Li j, for the purposes
here, the most convenient and most commonly employed are the Bethe
term L0:11
2me v 2 C(v)
L0 = ln − , (5)
I0 Z2
the Lindhard12 version of the Barkas correction:13
3 π e 2 I0 2 me v 2
L1 = L10 = ln , (6)
2 me v 3 I0
and the Bloch correction:14

∞
v02 ∼ v02
L2 = L20 = − (l + 1)−3 = −1. 202 . (7)
v2 v2
l=0
Here, v0 is the Bohr velocity (2.187691 × 106 m/s).
The quantity of interest for describing energy deposition, the mean
excitation energy, was first introduced in this context by Bethe.11 His origi-
nal derivation considered that the projectile velocity was much larger than
that of the target electrons with which it collided and exchanged energy.
As this assumption is not applicable to slow ions nor to collisions with
core electrons in atoms or molecules, the shell corrections, C(v)/Z2, were
added to correct for this. However, the mean excitation energy is a well-
defined quantity, namely the first energy weighted moment of the dipole
oscillator strength distribution (DOSD) of the target system in whatever
overall state the system finds itself:
df
ln E dE
ln I0 = dE .
df dE (8)
dE
Thus, were the entire set of excitation energies and oscillator strengths
of the system (molecule, liquid, or mixture, and gas, liquid, or solid) known,
the unique mean excitation energy of the system would be obtainable,
either theoretically or experimentally. It should be noted that, in principle,
the mean excitation energy is not a parameter. It is a well-defined function
of the electronic excitation properties of the target and only of the target.
It is not a function of projectile velocity, and, in principle, does not depend
in any way on the properties of the projectile.
3. THEORETICAL DETERMINATION OF I0
The mean excitation energy was originally defined by Bethe11 in
his derivation of the stopping cross section, and is defined in Eq. (8). The
definition is unique, and given the exact quantum mechanical solution of
the target electronic structure, that is, the DOSD, or the complete set of
all electronic transitions and their dipole oscillator strengths, the integrals
could be carried out. In practice, this is not possible, as various theoretical
approximations must be made in order to carry out the calculation of the
DOSD: level of theory, molecular geometry, basis set size, target orientation
with respect to the beam, isolated molecule or condensed system, and oth-
ers. However, at least for most atoms, one can approximate13 I0 ≈ 10Z2 eV
and then one can obtain an approximation for the value of I0 for a single
molecule by application of the Bragg rule.15
We, on the other hand, have carried out a number of calculations on
the DOSD of an isolated water molecule,16–21 using the polarization
propagator scheme22 in the dipole length formulation at the random phase
approximation (RPA) level of theory. Calculations were carried out using
various extended basis sets23–25 at the water equilibrium geometry. The
calculations yielded values of the mean excitation energy of water with an
average value of 73.0 ± 0.6 eV for the various bases employed. Calculations
of the DOSD in the length, velocity, and mixed formulations gave only
slightly different results: 73.2 ± 0.1 eV, while density functional theory cal-
culations26 with the B3LYP exchange-correlation functional in the length
formulation and a somewhat larger basis set27 gave with 72.2 eV a slightly
smaller result.
These results are in reasonable agreement with the semiempirical result
of Zeiss and Meath28 (ZM) of 71.62 eV. However, we note that although
theoretical, the ZM method does not directly evaluate Eq. (8).
In a theoretical determination of the mean excitation energy of a mol-
ecule, there are other considerations as well.
As the electronic structure of a molecule changes with its geometry,
so does the DOSD and thus the mean excitation energy. Calculation
gives a ground state mean excitation energy of 72.92 eV16 for H2O while
vibrational averaging, using the Morse-oscillator-rigid-bender scheme of
Jensen,29 yields a value of 71.61 eV, nearly 2% smaller. Although isotopic
composition will not change the DOSD of the target molecule, it will
influence the vibrationally averaged values of the mean excitation energy
through the vibrational wavefunction.
Although the target mean excitation energy does not depend at all
on the projectile, it does depend on the orientation of a non-spherically
symmetric target molecule such as water with respect to the beam, as only
molecular electronic states with the correct polarization with respect to the
beam direction can be excited.30 Calculations on water in its equilibrium
geometry, fixed in the x–z plane with the z axis as the C2 symmetry axis,18
give a rotationally averaged mean excitation energy of 72.91 eV, having
directional components in the x, y, and z directions of 69.21, 76.76, and
72.87 eV, respectively.
Finally, as water seldom exists as a single molecule, the effect of
hydrogen bonding on the mean excitation energy of water is of interest.
Calculations of the mean excitation energy of H2O and of (H2O)2 give
mean excitation energies20 of 73.56 and 73.18 eV, respectively, a difference
of less than 1%.
Consideration of the effect of a few hydrogen bonds on the mean
excitation energy of a water molecule is not sufficient to determine the
mean excitation energy of bulk (liquid) water. However, the mean excita-
tion energy for a bulk system can also be determined from a study of the
dielectric response of the bulk system.31 In this approach, the mean excita-
tion energy can be obtained from:
∞
E ln E Im [−1/ε(E, 0)] dE
ln I0 = 0 ∞ . (9)
0 E Im [−1/ε(E, 0)] dE
Using a Drude-like dielectric function of the form

−1
ε(E, q) = 1 + fj q Ep2 Ej2 (q) − E 2 − iEγj (q) , (10)
j
√
where Ep = 2 nπ a0 is the nominal plasmon energy of the medium and
n is the electronic density (for water, Ep = 21. 4 eV). Here, the quantities
[Ej , γj , fj ] refer to the energy, lifetime, and oscillator strength of each of the
j electronic transitions. Using this dielectric function in Eq. (9), the mean
excitation energy for liquid water obtained from the dielectric response
model is 77.8 eV.31 We note that a detailed description of the dielectric
response of the system is given in Ref. 31.
Thus, using theory, one can investigate the properties of the DODS and
thus the behavior of the mean excitation energy under various conditions
without reference to experiment.
4. EXPERIMENTAL DETERMINATION OF I0
On the other hand, clinical application of swift ion radiation in
terms of tumor therapy is concerned with energy deposition to a specific
place, and thus with dosimetry: prediction of dose/depth properties of ion
radiation. In order to determine the relevant quantities for proper dosim-
etry determinations, the ratio of stopping powers for air (a) and water (w)
0 32–34 A good approximation to the
swater, air, for the ion beam is needed.
water/air mass stopping ratio for a particular projectile species at a particu-
lar energy (velocity), or with a delta function fluence differential in energy,
can be written35 as Eq. (1). Thus, the mean excitation energy is of primary
importance. Although practitioners, due to the difficulties discussed in
this contribution, may simply measure dose/depth curves directly rather
than attempting to calibrate beam energies closely and then construct
air from experimentally determined mean excitation energies, I0 is
0
swater,
still reported in the literature. The question arises as to how the appropriate
mean excitation energies should be obtained.
As it is not feasible to evaluate Eq. (8) directly from experiment, the
value of the mean excitation energy is frequently deduced from experi-
exp
ment. There are several ways to obtain I0 .
Perhaps the most reasonable way to extract the mean excitation energy
from experiment while invoking the least number of assumptions is to use
an experimental dielectric response function in Eq. (9).36 This procedure
gives an experimental value of the mean excitation energy of water of
81.8 eV.
Another, and in the radiological community more usual, way to obtain
the mean excitation energy of a bulk substance is to fit stopping power data
from energy deposition experiments. In principle, one chooses an ansatz for
the stopping power or range of an ion as a function of projectile kinetic
energy† and then treats the mean excitation energy as a fitting parameter
determined by adjusting it to get the best fit of the ansatz function to
the experimental stopping or range data. Programs such as the Monte
Carlo based SHIELD-HIT37, 38 or FLUKA39 are frequently used for this
purpose.1
† In fact, projectile velocity is the relevant quantity (see Eqs. (2)–(7)). However, velocities are gener-
ally expressed in terms of the corresponding kinetic energy per nucleon: eV/u, keV/u, or MeV/u.
Energy/u has units of projectile velocity squared, but must be scaled by the one half the projectile
mass in appropriate units to give projectile velocity squared [see articles in volumes 45 and 46 (1994),
Theory of the Interaction of Swift Ions with Matter, for details].
The problem then arises as to how the relevant quantities such as shell
corrections and the Barkas and Bloch corrections are to be chosen, as the
choice of the ansatz fitting function will affect the resulting value of I0.
To extract the mean excitation energy of water from experiment, one
might choose values of the shell corrections and the Barkas correction, and
then, using the mean excitation energy as a fitting parameter to Eqs. (2–6),
determine the “experimental” value of I0 yielding the best fit to the experi-
mental data over a range of projectile velocities. The result is then reported
as the experimentally determined value of the mean excitation energy,
exp
namely I0 .
However, problems and uncertainties arise. Previous work has shown
that shell corrections may vary considerably according to the method of
calculation.40 In addition, shell corrections are generally not available for
molecules, so the approximate Bragg additivity rule:41
C(v) C(v)
= , (11)
Z2 i=atoms
Z2 i
or

C(v) C(v)
= , (12)
Z2 Z2 i
i=fragments
is frequently employed,42 introducing further uncertainty into the experi-

mental value of the mean excitation energy.
To illustrate the effect of variation in the shell corrections on the mean
excitation energy extracted from experiment, we consider the simplest, or
Bethe, case where the stopping cross section is written

4 π e 4 Z12 Z2 4 π e 4 Z12 Z2 2 me v 2 C(v)
S(v) = L0 (v) = ln − .
me v 2 me v 2 I0 Z2
(13)
exp
A formula for extraction of a value of the mean excitation energy, I0 ,
from experimental values of the stopping cross section can be written
exp 2 me v 2
I0 = .
exp me v 2 S(v)
+ C(v) (14)
4 π e 4 Z12 Z2 Z2
Consider the effect of differing shell corrections when used in extract-

ing an experimental mean excitation energy for protons with v = 6 v0, and
thus a kinetic energy of 900 keV, impinging on water. We choose a proton
energy which is low from the radiological standpoint as the corrections to
the Bethe term are largest for lower ion velocities, and thus the deviations
and variations will represent maximum encountered values. Shell correc-
tions for such a situation run from 0.1543 up to 0.50,44 with various values
in between also reported.45 Table 3.1 illustrates the variation in I0 from Eq.
(14) for 900 keV protons on water using a typical value of the stopping cross
section [S(6) = 11. 52 a.u.] and various values of the shell corrections.
Similarly, the Barkas correction, L1, can be determined in several ways,
the most usual being the numerical scheme of Ashley et al.46 or the ana-
lytical form of Lindhard.12 Again, uncertainties are introduced which will
affect the value of the mean excitation energy obtained from experimental
data.47, 48 In Table 3.2 we present some values of Lindhard’s L1 as calculated
from Eq. (6).
Over this range of mean excitation energies, the Barkas correction does
not vary much, consistent with observation.49–51 Thus, the choice of Barkas
correction is less critical to the extraction of a mean excitation energy from
experiment than is the choice of shell corrections.
Table 3.1 Mean excitation energies from equation 14 for 900 keV protons on water
for various values of the shell corrections
C/Z2 I0 (eV)
0.00 72.3
0.10 65.4
0.15 62.2
0.25 56.3
0.32 52.5
0.40 48.5
0.50 43.9
Table 3.2 Barkas correction using the Lindhard form for 900 keV protons on water
using various values of the mean excitation energy
I0 (eV) L1 (a.u.)
72.30 0.191
62.23 0.172
56.31 0.160
If one measures the oscillator strength distribution for water in its dif-
ferent phases, it is seen that the distribution is different, especially near the
maximum, depending on the target phase.52 Target phase thus affects the
value of the measured mean excitation energy. As water readily exists in all
three phases, the mean excitation energy of water depends somewhat on
exp
the sample phase, with the liquid phase having a value of I0 that is some
few percent greater than the gas.10, 53 Although there is no direct evalua-
tion of Eq. (8) for liquid water, comparison of direct calculation of I0 for
an isolated water molecule and for its hydrogen bonded dimer shows the
isolated molecule to have a larger value of the mean excitation energy by
a fraction of a percent.20
The mean excitation energy is a property of the target only, by defini-
tion. It thus does not depend on any property of the projectile, in particular
the projectile speed.Thus, in the simplest, or Bethe, case where the stopping
cross section is written as in Eq. (13), the formula for extraction of I0 from
experimental values of the stopping cross section is written as Eq. (14). This
presents another problem, as values of the experimental mean excitation
energy extracted from S(v) measured at different projectile velocities54
using calculated, Bragg rule based, shell corrections,41 yield quite different
experimental values of I0, depending on to which velocity the stopping
cross section corresponds. This is clearly incorrect behavior for a quantity
that should be independent of the projectile.
The addition of Barkas and Bloch corrections leads to the expression:
2
3 π e 2 I0 2 me v 2 v me S(v) C(v) v02
1+ ln = + + 1. 202 2 .(15)
2 me v 3 I0 4 π e 4 Z12 Z2 Z2 v
exp
This equation must be solved iteratively to determine a value for a I0 .
However, there is still no independence of the experimental value of the
mean excitation energy extracted in this manner on projectile velocity.
Another problem which must be confronted when extracting a value
of I0 from measured data is that of projectile charge exchange. As a pro-
jectile travels through a target sample, charge exchanging collisions are
expected.55, 56 Thus, a measured depth-dose curve or stopping power
curve for a projectile in a condensed phase target will, in fact, contain con-
tributions for the various charge states available to the projectile under the
circumstances. For example, one would expect contributions from both
H+ and H in measuring the stopping of a beam of protons in water. The
description of the energy loss under these circumstances using Eqs. (2), (3),
(5), (6), (7) will thus involve a value of I0 which, rather than being deter-
mined directly as mean excitation energy, will again be an effective fitting
parameter which includes charge exchange and gives a best fit to the mea-
sured depth/dose or stopping curve.
The range of an ion in a condensed phase target is of greatest inter-
est in radiological terms, as it is at the end of the projectile trajectory that
the greatest amount of projectile energy is deposited in the target.57 Were
one to calculate the range in the continuous slowing down approximation,
one would integrate the stopping cross section from the initial energy, E0
to when the ion stops.
0
1
R(E0 ) = − dE. (16)
E0 nS(E)
This would return the length of the pathway, in a straight line, from
the entrance of the ion into the medium to the point where it stopped.
However, in an experimental situation, the projectile ion undergoes many
collisions, with concomitant changes in velocity, along its pathway, and
thus the actual penetration depth, d, could differ significantly from the
pathlength.58 A mean excitation energy extracted from experiment using
Eq. (2), if depth data is used, will then have this ambiguity included.
Mean excitation energies for water varying from 68 eV59 to 81.1 eV36
have been reported resulting from both theoretical evaluations of Eq. (8)
and from fitting of experimental data. This corresponds to an uncertainty
of some 18.5%, which is certainly significant, and is unacceptable when I0
is being used to determine details of hadron therapy.
5. CONCLUSION
The conclusion is that while the mean excitation energy is well-
defined and can consistently be determined theoretically, values of the
mean excitation energy extracted from experiment are not well-defined.
Rather, they depend significantly on the assumptions made concerning
the method of extraction of the mean excitation energy from, typically, a
stopping power or range measurement. Thus, if high accuracy swater,
0
air are
needed for calculation of dose-depth curves for radiation therapy as opposed
to direct measurement, both the ansatz equation and parameters used to
obtain the experimental mean excitation energy by fitting to experimental
data need to be specified when reporting the experimental mean excitation
energy. Values of I0 obtained from theory and experiment are in principle
identical, and, from a practical standpoint, would be expected to be of the

same order. However they are not the same, as the theoretical values come
from the definition of the quantity [Eq. (8)] and the experimental values
come from using the mean excitation energy as a free fitting parameter
to experimental data. Their differences can lead to clinically unacceptable
uncertainties in dosimetry prediction.The experimental values of the mean
excitation energy should thus be reported only with reference to the details
of its determination. Thus, using I0 in Monte Carlo or other simulations
for energy deposition in hadrontherapy should only be done with care and
attention to the method by which I0 was determined.
For reasons cited, we would recommend that values of the mean excita-
tion energy derived from experiment be referred to as an effective value of
the mean excitation energy rather than as an experimental mean excitation
energy.
“The mean excitation energy of water is 75.0 eV.” What is the meaning
of such a statement? Clearly the statement itself is not meaningful with-
out qualification of exactly how the reported mean excitation energy was
determined.
ACKNOWLEDGMENTS
S.P.A.S. acknowledges support by grants from the Danish Center for
Scientific Computing, the Danish Natural Science Research Council/
Danish Councils for Independent Research (Grant No. 272-08-0486) and
the Carlsberg foundation.
REFERENCES
1. Andreo, P. On the clinical special resolution achievable with protons and heavier
charged particle radiotherapy beams. Phys. Med. Biol. 2009, 54, N205.
2. Paganetti, H. Range uncertainties in proton therapy and the role of Monte Carlo simu-
3. von Sonntag, C. The chemical basis for radiation biology; Taylor and Francis: London, 1987.
4. von Sonntag, C. Free-radical DNA damage and its repair—a chemical perspective; Springer
Verlag: Heidelberg, 2005.
5. Andreo, P.; Burns, D.T.; Hohlfeld, K.; Huq, M. S.; Kanai, T.; Laitano, F.; Smythe, V. G.;
Vynckier, S. Absorbed dose determination in external beam radiography: an interna-
tional code of practice for dosimetry based on standards of absorbed dose to water,
Technical Report Series No. 398, International Atomic Energy Agency, Vienna, 2000.
6. Lühr, A.; Hansen, D. C.; Jäkel, O.; Sobolevsky, N.; Bassler, N. Analytical expressions for
water-to-air stopping-power ratios relevant for stopping power dosimetry in particle
therapy. Phys. Med. Biol. 2011, 56, 2515.
7. Paul, H. A comparison of recent stopping tables for light and medium-heavy ions with
experimental data, and applications to radiotherapy dosimetry. Nucl. Inst. Meth B 2006,
247, 166.
8. Shakespeare, W. The taming of the shrew, Grumio (1593) Act 4, Scene 1.

9. Gottschalk, B. Comments on ‘calculation of water equivalent thickness of materials of
arbitrary density, elemental composition and thickness in proton beam irradiation’. Phys.
Med. Biol. 2010, 55, L29.
10. Berger, M. J.; Seltzer, S. M. Stopping powers and ranges of electrons and positrons,
ICRU Report #37, Bethesda, 1984.
11. Bethe, H. Zur Theorie des Durchgangs schneller Korpuskularstrahlen durch Materie.
Ann. Phys. (Leipzig) 1930, 5, 325.
12. Lindhard, J. Barkas effect—or Z13, Z14—corrections to stopping of swift charged par-
ticles. Nucl. Instrum. Methods 1976, 132, 1.
13. Barkas,W. H.; Dyer, J. N.; Heckman, H. H. Resolution of the Σ–mass anomaly. Phys. Rev.
Lett. 1963, 11, 26.
14. Bloch, F. Zur Bremsing rasch bewegter Teilchen beim Durchgang durch Materie. Ann.
Phys. (Leipzig) 1933, 16, 285.
15. Bragg, W. H.; Kleeman, R. On the alpha particles of radium and their loss of range in
passing through various atoms and molecules. Philos. Mag. 1905, 10, 318.
16. Geertsen, J.; Oddershede, J.; Sabin, J. R. Calculation of molecular mean excitation ener-
gies via the polarization propagator formalism: H2 and H2O. Phys. Rev. A 1986, 34, 1104.
17. Jensen, P.; Oddershede, J.; Sabin, J. R. Geometric dependence of the mean excitation
energy and spectral moments of water. Phys. Rev. A 1991, 43, 4040.
18. Sauer, S. P. A.; Sabin, J. R.; Oddershede, J. Directional characteristics of the moments of
the dipole-oscillator-strength distribution of molecules: H2 and H2O. Phys. Rev. A 1993,
47, 1123.
19. Sauer, S. P. A.; Sabin, J. R.; Oddershede, J. Calculated molecular mean excitation energies
for some small molecules. Nucl. Instrum. and Methods B 1995, 100, 458.
20. Sauer, S. P. A.; Oddershede, J.; Sabin, J. R. Mean excitation energies and their directional
characteristics for energy deposition by swift ions on the DNA and RNA nucleobases.
J. Phys. Chem. C 2010, 114, 20335.
21. Sauer, S. P. A.; Oddershede, J.; Sabin, J. R. Mean excitation energies for biomolecules:
Glycine to DNA. Adv. Quantum Chem. 2011, 62, 215.
22. Oddershede, J.; Jørgensen, P.; Yeager, D. L. Polarization propagator methods in atomic
and molecular calculations. Compt. Phys. Rep. 1984, 2, 33.Oddershede, J. Adv. Chem.
Phys. 1987, 69, 201.Sauer, S. P. A. Molecular Electromagnetism: A Computational Chemistry
Approach; Oxford University Press
23. Lazzeretti, P.; Zanasi, R. Quantum mechanical approach to IR intensities via nuclear
electric shielding tensors. I. Water. J. Chem. Phys 1985, 83, 1218.
24. Sauer, S. P. A.; Paidarová, I.; Oddershede, J.; Correlated and gauge origin independent
calculation of magnetic properties. 2. Shielding constants of simple singly bonded mol-
ecules., Theoret. Chim. Acta 1994, 88, 351.
25. Sauer, S. P. A.; Oddershede, J.; Sabin, J. R. Directional dependence of the mean exci-
tation energy and spectral moments of the dipole oscillator strength distribution of
Glycine and its zwitterion. J. Phys. Chem. A 2006, 110, 8811.
26. Bruun-Ghalbia, S.; Sauer, S. P. A.; Oddershede, J.; Sabin, J. R. Comparison of the direc-
tional characteristics of swift ion excitation for two small biomolecules: Glycine and
Alanine. J. Phys. Chem. B 2010, 144, 633.
27. Ligabue, A.; Sauer, S. P. A.; Lazzeretti, P. Correlated and gauge invariant calculations of
nuclear magnetic shielding constants using the continuous transformation of the origin
of the current density approach. J. Chem. Phys. 2003, 118, 6830.
28. Zeiss, G. D.; Meath, W. J. Additivity of atomic and molecular dipole properties and dis-
persion energies using H, N, O, H2, N2, O2, NO, N2O, NH3 and H2O as models. Mol.
Phys. 1980, 39, 1055.
29. Jensen, P. A new Morse oscillator-rigid bender internal dynamics (morbid) hamiltonian
for triatomic-molecules. J. Mol. Spectrosc. 1988, 128, 478. Hamiltonians for the internal
dynamics of triatomic molecules. J. Chem. Soc. Faraday Trans. II 1988, 84, 1315.
30. Mikkelsen, H. H.; Oddershede, J.; Sabin, J. R.; Bonderup, E. A Bethe theory for the
directional dependence of stopping by molecules. Nucl. Instrum. Methods Phys. Res., Sect.
B 1995, 100, 451.
and a new I-value for water. Phys. Med. Biol. 2009, 54, 3451.
32. International Atomic Energy Agency (IAEA), Absorbed dose determination in external
beam radiography: an international code of practice for dosimetry based on standards
of absorbed dose to water, Report TRS-398, Vienna, 2000.
33. Paul, H.; Geithner, O.; Jäkel, O. The influence of stopping powers upon dosimetry for
radiation therapy with energetic ions. Adv. Quantum Chem. 2007, 52, 289.
34. See also the discussion in Section 2 of Ref. 3.
35. Henkner, K.; Bassler, N.; Sobolevsky, N.; Jäkel, O. Monte Carlo simulations on the
water-to-air stopping power ratio for carbon ion dosimetry. Med. Phys. 2009, 36, 1230.
cross sections in liquid water. Rad. Phys. Chem. 1998, 53, 1.
37. Dementyev, A.V.; Sobolevsky, N. M. SHIELD—universal Monte Carlo hadron transport
code: scope and applications. Radiat. Meas. 1999, 30, 553.
38. Gudowska, I.; Sobolevsky, N.; Andreo, P.; Belkić, Dž.; Brahme, A. Ion transport in tissue-
like media using the Monte Carlo code SHIELD-HIT. Phys. Med. Biol. 2004, 49, 1933.
39. FLUKA site. <http://www.FLUKA.org/FLUKA.php>.
40. Bichsel, H. Shell corrections in stopping powers. Phys. Rev. A. 2002, 65, 052709.
41. Bragg, W. H.; Elder, M. A. On the alpha particles of radium, and their loss of range in
passing through various atoms and molecules. Phil. Mag. 1905, 10, 318.
42. Zeiss, G. D.; Meath, W. J. Accurate evaluation of stopping and straggling mean excita-
tion-energies for N, O, H2, N2, O2, NO, NH3, H2O, and N2O using dipole oscillator
strength distributions - test of validity of Braggs rule. Radiat. Res. 1977, 70, 284.
43. International Commission on Radiation Units and Measurements, ICRU Report 49,
Bethesda, MD: ICRU, 1993.
44. Emfietzoglou, D.; Pathak, A.; Nikjoo, H. Electronic stopping power of liquid water for
protons down to the Bragg peak. Rad. Protect. Dosim. 2007, 126, 97.
45. Oddershede, J.; Sabin, J. R. Orbital and whole-atom proton stopping power and shell
corrections for atoms with Z ≤ 36. At. Data Nuc. Data Tables 1984, 31, 275.
3
46. Ashley, J. C.; Ritchie, R. H.; Brandt, W. Z1 effect in stopping power of matter for
3
charged-particles. Phys. Rev. B 1972, 5, 2393. Z1 -dependent stopping power and range
3
contributions. Phys. Rev. A 1973, 8, 2402. Z1 -dependent range contributions. Phys. Rev
A 1974, 10, 737.
47. Sabin, J. R.; Oddershede, J. A comment on the abstraction of mean excitation-energies
from experimental reduced stopping powers. Nucl. Instrum. Methods Phys. Res., Sect. B
1990, 44, 253.
48. Soltani-Nabipour, J.; Sardari, D.; Cata-Danil, Gh. Sensitivity of the Bragg peak curve to
the average ionization potential of the stopping medium. Rom. J. Phys. 2009, 54, 321.
49. Andersen, H. H.; Bak, J. F.; Knudsen, H.; Nielsen, B. R. Stopping power of Al, Cu, Ag,
3 4
and Au for MeV hydrogen, helium, and lithium ions—Z1 and Z1 proportional devia-
tions from Bethe formula. Phys. Rev. A 1977, 16, 1929.
3
50. Basbas, G. Inner-shell ionization and the Z1 and Barkas effects in stopping power. Nucl.
Instrum. Methods Phys. Res., Sect. B 1984, 4, 227.
51. Reiter, G.; Kniest, N.; Pfaff, E.; Clausnitzer, G. Proton and helium stopping cross-
sections in H2, He, N2, O2, Ne, Ar, Kr, Xe, CH4, and CO2. Nucl. Instrum. Methods Phys.
Res., Sect. B 1990, 44, 399.
52. Hayashi, H.; Watanabe, N.; Udagawa, Y.; Kao, C.-C. The complete optical spectrum of
water measured by inelastic X-ray scattering. Proc. Nat. Acad. Sci. 2000, 97, 6264.
53. Thwaites, D. I.; Watt, D. E. Similarity treatment of phase effects in stopping power for
low-energy heavy charged-particles. Phys. Med. Biol. 1978, 23, 426.
54. Paul, H. Phase effect for H ions in water. <http://www.exphys.jku.at/stopping/>.
55. Cabrera-Trujillo, R.; Sabin, J. R.; Deumens, E.; Öhrn,Y. Orientational effects in energy
deposition by protons in water. Adv. Quantum Chem. 2005, 48, 47.
56. Schiwietz, G.; Grande, P. L. Stopping of protons—improved accuracy of the UCA
model. Nucl. Instrum. Methods Phys. Res., Sect. A 2012, 273, 1.
57. Inokuti, M. Inelastic collisions of fast charged particles with atoms and molecules—the
58. Lindhard, J.; Scharff, M.; Schiøtt, H. E. Range concepts and heavy ion ranges. Mat. Phys.
Medd. Dan. Vid. Selsk 1963, 33/10, 1.
59. Porter, L. E.;Thwaites, D. I. Physical state effects on the mean excitation energy of water
as determined from alpha-particle stopping power measurements. Phys. Rev. A 1982, 25,
3407.
CHAPTER FOUR
Molecular Scale Simulation

of Ionizing Particles Tracks for
Radiobiology and Hadrontherapy
Studies
Ziad Francis
Saint Joseph University, Faculty of Sciences, Department of Physics, Mkalles, Beirut, Lebanon
Contents
1. Introduction 80
2. Detailed step by step track structure
codes 84
2.1 Monte-Carlo codes 84
2.2 Collision processes: cross sections 85
2.3 Sub excitation electrons and the chemical phase 88
3. Radiation microdosimetry analysis 89
3.1 Theoretical and experimental microdosimetry 89
3.2 Ions RBE estimation 94
4. DNA damage estimation 99
4.1 Track structure detailed approach 99
4.2 Stewart and Semenenko MCDS method 100
4.3 Garty statistical approach 101
4.4 DBSCAN clustering estimation 102
5. Conclusion 104
Acknowledgments 107
References 107
Abstract
This study looks into Monte-Carlo simulation methods that are used for radiotherapy
and radiobiology studies. Although simulation methodologies for the two mentioned
fields are quite different since one deals with macroscopic effects rather than the
cellular scale effects in the case of the other, the results obtained by both fields are
complementary for understanding radiation effects on cellular substructures and the
resulting reactions of irradiated biological tissues. It has been concluded through the
literature that the deoxyribonucleic acid (DNA) damage is the most important result
to look at among other radio-induced lesions in the cell. Thus, most of the research
projects on radiation damage including Monte-Carlo studies are now concentrated on
radio-induced DNA breaks and cell repair activities. Among the simulation programs

80 Ziad Francis
we can distinguish the general purpose codes that are mostly used for macroscopic
calculations and the step by step track structure codes that are dedicated for sub-
cellular scale simulations mostly dealing with particles interactions with the DNA struc-
ture. However, none of the proposed toolkits can completely cover all the simulation
cases since we are usually limited by the lack of cross sections for a certain particle type
and in a certain energy domain depending on the code content. This obstacle can be
overridden by some approximations on the macroscopic scale but when it comes to
submicron calculations the energy deposit patterns should be as precise as possible
for radiobiological analysis. On the DNA scale, chemical reactions and free radicals dif-
fusion through target media were taken into account in some studies considering also
the DNA structure geometry in the simulation. Although this type of approaches is
aiming to be as realistic as possible, it can be a complex task to reproduce and also very
time consuming requiring an important computing power. So in an effort to simplify
and accelerate the calculations, several alternative methods were developed by differ-
ent authors giving an acceptable estimation of radio-induced DNA damage quantifica-
tion and making this issue more accessible for quick studies. The work described in this
chapter reviews some of the numerical calculations of basic microdosimetric quantities
for particles that are generally used in hadrontherapy, mainly protons and carbon ions,
taking into account the secondary electrons tracks that are generated in tissue-like
media (water). Using the Geant4-DNA package of the Geant4 Monte-Carlo simulation
toolkit it is possible to track particles in a step by step mode taking also advantage of
the available geometry modules and the standard physics processes of the standard
Geant4 toolkit. Simulations are then described and the DNA damage quantification
algorithms are studied with an effort to show the link between ionizing track structure
and the resulting damage yields. The results obtained by different numerical methods
are compared together then a discussion describes the advantages of each of the
mentioned calculations.
1. INTRODUCTION
Generally and for a long period of time treatment planning in radiotherapy
was mostly based on observations obtained from experimental data of cell
survival rates after irradiations. The linear quadratic model of cell population
survival is still in use by the medical community especially for X-ray irradia-
tions. The safety of this model was studied and confirmed by Brenner et al.1
concluding that no noticeable improvement can be added by using more
sophisticated models. However, in the case of ions the energy deposit pattern
is denser than for X-rays and ions proved to be more effective for deep seated
tumors taking advantage from the so called “Bragg peak” of the energy depo-
sition profile. Recently the number of protons and ions treatment facilities
was increasing through the world. Protons are used in the United States, e.g.,
Massachusetts, Boston, and Loma Linda, California, while heavier ions ther-
apy is still missing. In Japan we may cite the Heavy Ion Medical Accelerator
Molecular Scale Simulation of Ionizing Particles Tracks for Radiobiology 81
in Chiba (HIMAC), another accelerator in Hyogo and recently the Gunma

University facility. In Germany the Heidelberg Ion-Beam Therapy (HIT)
center started ion treatments in 2009, with parallel research activities focusing
on ions interactions in tissue and improving treatment protocols for a better
tumor control probability. Other countries in Europe are also using proton
beams for therapy or therapeutic research we can cite among many the
National Institute of Nuclear Physics (INFN) accelerator in Italy (Catania),
the Uppsala Svedberg laboratory in Sweden and one treatment facility in
the United Kingdom. In France protons are used at the Proton Therapy
Center of the Curie Institute at Orsay (ICPO) and the Lacassagne Center
Biomedical Cyclotron (CLCB) in Nice along with new projects that are in
progress for heavy ions therapy and research installations, e.g., Hadrontherapy
project opening in Lyon (France) in 2016 and the Advanced Resource
Center for Hadrontherapy in Europe (ARCHADE) mainly dedicated for
research activities. As a general observation, developed countries are now
tending to invest in heavy ions irradiation facilities with several emerging
projects especially after the success of the Heidelberg treatment center. With
the spread of ions accelerators, research emphasis was put on new treatment
protocols and new cell survival models development, since these two latter
are major database sets that are needed in the medical medium. One of the
ions specific models that are presently in use in the medical medium is the
Local Effect Model (LEM). The idea was developed by Kramer and Scholz2
where the carbon ions Relative Biological Effectiveness (RBE) could be
calculated using numerical computation and based on X-rays RBE database.
The main advantage of the LEM is that it proved its reliability in RBE pre-
diction for several tissue types and over a wide range of atomic numbers and
particle energies without the need to perform cell irradiation experiments.
Treatment planning programs were further developed by Kramer et al.3 and
reviewed by Jakel et al.4 in order to provide the essential data base for ion
treatment. In comparison with X-rays, ions have a high linear energy trans-
fer (LET) in tissue and consequently a high biological effectiveness, and in
this case the dose delivered to the tumor should be as accurate as possible in
order to avoid normal tissue damage around the tumor. Also, the difference
between the treatment planning and the treatment delivery is of high impor-
tance in the case of ions. Since this difference can have a non-negligible
impact on the surroundings of the irradiated tumor, the range accuracy in
carbon ion treatment planning was investigated by Rietzel et al.5 concluding
that carbon ranges can be controlled within ∼1 mm in soft tissue for typical
treatments of head and neck tumors.
82 Ziad Francis
Although at this stage studies are able to define the basics for ions treat-
ment planning they do not provide any precise information about processes
occurring on the single cell level, and the damage yields within the sub-
cellular structures. In fact, the cell survival probability depends on many
factors like the type of the irradiated cell, the severity of the damage that
is induced to the DNA molecule, the replication phase of the cell being
irradiated, the repair activities within the cell and also on the irradiation
properties like the beam type and energy. Understanding the cell reaction
to different types of radiation requires a tight collaborative work involv-
ing biologists and physicists studying single cell irradiation experiments.
Measurements of DNA damages including strand breaks and base lesions
were carried out by some studies using different analysis techniques like
gel electrophoresis, e.g., Stypczynska et al.,6 X-ray photoelectron spectros-
copy, e.g., Ptasinska et al.7 and the Enzyme-Linked Immunosorbent Assay
(ELISA) experiment, e.g., Smialek et al.8 Moreover, the effect of the DNA
surrounding conditions were also investigated like the presence of amino
acids combined to DNA during irradiation by Stypczynska et al.,6 the pres-
ence of scavengers and repair enzymes in the medium by Milligan,9 and
Gulston et al.10 and the influence of the chromatin organization on damage
yields by Magnander et al.11
In parallel, microdosimetry studies were carried out on different par-
ticle beams using micrometric Tissue Equivalent Proportional Counters
(TEPC), e.g., Borak et al.12 With microdosimetry techniques it is possible
to measure the lineal energy distribution of a particle type revealing the
energy deposit characteristics on the micrometric and the nanometric
scale (see Refs. 13–15). However, the literature does not cover all typical
radiation cases that are encountered in the radiotherapy field. Monte-Carlo
method simulations can be a theoretical alternative providing a good esti-
mation of such measurements.
During the last 20 years numerical modeling became the most acces-
sible method to study interactions of particles crossing biological targets
and other materials enabling users to obtain the energy deposit distribu-
tions in the target volumes. General purpose Monte-Carlo codes like Fluka
(Fluctuating Kaskade, see Ref. 16), Geant4 (Geometry and Tracking, see
Ref. 17), MCNP (Monte-Carlo N-Particle transport code, see Ref. 18)
and the SHIELD-HIT code (Heavy Ion Therapy code, see Refs. 19, 20)
are able to simulate photons and ions interactions with different materials
taking into account the secondary particles that are created in the medium.
However, these codes are practical and adapted for macroscopic simulations
and cannot be used for submicron scale calculations. In fact, simulations can
be very lengthy in many cases and in order to save computing time, using
an energy cut-off is a common solution for most of these codes.This means
that when a particle’s energy drops below the chosen cut-off the particle
is killed and its energy is deposed locally at that point. Other techniques
are also used, an energy production threshold is chosen by the user for
each particle, in this way only secondary particles with energies above this
threshold are created and simulated, this approach is adopted in the Geant4
toolkit. In this way one can avoid generating a big number of low energy
secondary particles losing all their energy within few nanometers from the
primary particle’s track and cannot be distinguished from this latter from
a macroscopic point of view. On the sub-cellular level, simulations should
include all the secondary particles that are created in the target as low as
their kinetic energy can be and these particles should be followed nearly
until complete energy loss. In other words the energy cut-off should be set
to the minimum and physics processes should take into account all the pos-
sible occurring interactions. This type of detailed simulation is also called
the “step by step” particle tracking and its use is restricted to simulations
confined within volumes of few micrometers, due to the lengthy required
computing time. However, for this method the results precision is reason-
able for molecular scale analysis. Many studies are based on track structure
modeling of DNA damage after irradiation.
In general, in the simulations we can distinguish among the different
levels of the complexity of the DNA damage since the repair activities are
strongly dependent on the type of the damage and the damage yields in
the cell. Some of the detailed history Monte-Carlo codes are able to track
chemical species and free radicals creation and diffusion to estimate the
indirect radiation effect on DNA. For the sake of completeness the DNA
structure is also modeled in some studies (Ref. 21–23, and others) show-
ing the effect of the geometrical shape and distribution of the DNA on
the resulting damages. On another hand, more approximate approaches
were developed enabling DNA damage estimation after irradiations, like
the work of Semenenko et al.24 where cell repair activities are also taken
into account, and Francis et al.25, 26 where the energy deposit spatial dis-
tribution is directly linked to DNA damage using clustering algorithms on
track structures of different particles.
In the following sections, we will briefly summarize the cross sec-
tions that are of use for submicron simulations and review the micro-
dosimetry basics citing some experiments that can be interesting for the
84 Ziad Francis
radiobiology community and also for Monte-Carlo codes validation.

Then the emphasis is put on the discussion of DNA damage estima-
tion following different methodologies from the detailed event by event
modeling to the fast approximate statistical procedures that can be found
in the literature.
2. DETAILED STEP BY STEP TRACK STRUCTURE

CODES
2.1 Monte-Carlo codes
General purpose Monte-Carlo codes are now available for public use
providing also users support mostly through manuals and forums available
on the Web. Among these codes that are specific for radiation interactions
we can mention MCNPX,18 FLUKA,16 GEANT4,17 Penelope (A Code
System for Monte-Carlo Simulation of Electron and Photon Transport,
see Ref. 27), and the SHIELD-HIT code that was tested for tissue-like
media by Gudowska et al.19 These codes are well adapted for large scale
simulations and they offer good results precision for a reasonable amount
of computing time, and some benefit from integrated variance reduction
techniques (e.g., MCNPX) in order to further improve the results statis-
tics. Generally, in these codes the tracking is done in a condensed history
steps, interactions with energy loss values less than the fixed energy cut-
off and occurring between two major energy transfers are skipped and
the total energy loss is accounted for in one point situated at the end of
the mentioned step of interaction. In this case, the total cross section is
obtained by integrating the single differential cross section between the
energy cut-off Ecut and the maximum energy that can be transferred
to the medium in one single interaction noted as Emax. It follows the
expression:
Emax
dσ
σ = dE, (1)
Ecut dE
where E represents the energy transfer per single interaction.

And the energy loss per distance unit taking into account only the
skipped interactions with energy transfers below the energy cut-off value
is obtained by: Ecut
dE dσ
= E · dE.
dx 0 dE (2)
Depending on the step size the energy loss dE is calculated using the
previous expression and it is added on the energy deposit value at the end
of the step.
This method is used by several codes (e.g., MCNP and Geant4) as it
avoids loosing time on small energy transfers and generating a big number
of secondary particles (mostly electrons) that are not important on the
whole simulation scale from the macroscopic point of view. The energies
of the skipped secondary electrons in this case do not exceed the energy
cut-off values, usually set between 250 eV and 1 keV noting that their ranges
in water and tissue remains within 1 μm from the primary particle’s track.28
One may generally conclude that the adopted spatial resolution in these
simulations is about 0.5–1 μm. We should note here that the dimensions of
a biological cell can vary depending on its type, and its nucleus dimensions
are around 10–20 μm as a rough approximation. As for a chromatin fiber
the typical dimension scale is of the order of few nanometers depending
on the number of the considered nucleosomes. DNA double strand breaks
can be formed by two single strand breaks separated by less than 10 base
pairs which is equivalent to a maximum distance of ∼3.2 nm. So for DNA
damage calculations, simulations should have a nanometer scale precision
and this is not provided by the general purpose MC codes. For this reason
detailed interactions track structure codes were developed with physics pro-
cesses and cross sections that are especially dedicated for precise molecular
scale calculations, e.g., code of Gervais et al.,29 PARTRAC (Particle Tracks,
see Ref. [21, 30, 31]), PITS (Positive Ion Track Simulation32), NOREC
(National Oak Ridge Electron Transport Code, see Ref. 33), Geant4-DNA
which is an extension to the previously introduced Geant4 code enabling
simulations on the DNA scale,34 and others. Here it is important to note
that most of the cross sections are calculated for water as we consider that
the stopping powers of water and tissue are almost equivalent.
2.2 Collision processes: cross sections

Cross sections for protons inelastic collisions (excitation and ionization) can
be calculated using different methods like the First Born Approximation
(FBA) and the Rudd semi-empirical formula as it is the case in Geant4-
DNA35 and PARTRAC36 codes. The dielectric response function of water
is fitted to experimental data of Heller et al.37 and Hayashi et al.38 using
Drude equations and the double differential cross section is obtained and
integrated to calculate the single differential and the total cross sections.
This approach works well for high incident energies and fails when the
86 Ziad Francis
incident particle speed becomes comparable to the speed of the target

orbiting electrons. So for low energy protons (below 500 keV) the FBA
should be replaced by an alternative model. The Rudd semi-empirical for-
mula gives the energy differential cross section down to incident energies as
low as 1 keV. A relativistic approach of the Rudd model was also presented
by Plante and Cucinotta43 for protons and heavier ions enabling calcula-
tions of ionization cross sections for energies between ∼1 MeV/amu and
∼105 MeV/amu using no additional alternative models. The mentioned
relativistic Rudd formula was also implemented in Geant4-DNA allow-
ing Geant4 users to simulate relativistic ions in water taking into account
ionization processes.26 Theories of ionization collisions for fast incident
ions were thoroughly reviewed and discussed by Belkić 39 where ioniza-
tion cross sections are calculated according to different theoretical assump-
tions including the continuum distorted waves approximations and the
results were compared to experimental data. It was shown that while the
FBA diverges from the experimental results for intermediate and slow ion
velocities, the Modified Coulomb Born (MCB) method, which is related
to Continuum Distorted Wave-Eikonal Initial State (CDW-EIS) method
gives better cross sections in a wider energy region. Track codes usually do
not use the second-order theories, such as the MCB, Continuum Distorted
Wave (CDW)39 or CDW-EIS approaches. One of such codes described by
Gervais et al.29 employs the CDW-EIS method for ion interaction with
water and the Rudd semi-empirical formula is used to track the generated
secondary electrons. This code also takes into account the chemical reac-
tions that take place after the physical interactions in the target medium.
However, since it is configured in a way that the projectile energy is held
constant over the whole simulation length, so that the simulation cor-
responds to a well-defined energy and a constant particle LET, it is not
possible to simulate a complete ion range and to retrieve the Bragg peak
position in this case.
Charge change contribution to energy loss cannot be neglected espe-
cially in the Bragg peak area where it reaches 30% of the total energy loss
in the case of protons. Charge transfer cross sections for protons and alpha
particles are well described by Dingfelder et al.36, 40 For heavier ions when
the cross sections for each of the charge change channels are not available
the common procedure that is followed is to consider an effective charge
that takes into account all the possible charge states of the ion depending
on its stripped charge number and also on its velocity. High energy ions
tend to be stripped and charge change contribution increases when the ion
energy decreases. Modeling the effective charge of an ion can be achieved

by using semi-empirical formulas that were fitted so that the calculated
stopping power agrees with the corresponding experimental values. We
may cite the model of Barkas et al.,41 Booth and Grant model42 that is
used in the simulation code described by Plante and Cucinotta43 and also
in Geant4-DNA for ions heavier than protons by Francis et al.26 A detailed
study on the effect of the Barkas correction to the Bethe-Bloch stopping
power can be found in Porter.44
For inelastic collisions of electrons, the FBA can be used for sufficiently
high energies in water as described by Dingfelder et al.45 and Emfietzoglou
and Nikjoo.46 For relativistic energies above ∼35 keV an adapted formalism
is applied to the FBA by Bousis et al.47 And for low kinetic energies, cor-
rections can be applied to the cross sections as described by Emfietzoglou
and Nikjoo.46 The mentioned corrections may vary from a simple fitting
factor bringing the calculated cross section closer to the experimental
values to more theoretical models taking into account higher order pertur-
bations. In Dingfelder et al.45 the differential cross section was multiplied
by an empirical factor and the final values agree with the experimental
data of electrons collisions in vapor water. Alternative corrections have
been studied like the classic coulomb field correction by Vriens48 and the
second order perturbation term described by Emfietzoglou and Nikjoo.46
The latter correction considers a second order perturbation term based
on the work of Ashley et al.49 Also Kim and Rudd50 proposed a binary
dipole semi-empirical model providing reasonable values of ionization
cross sections.
Although no energy loss is induced by elastic collisions of electrons,
the effect of this process on the particle diffusion and consequently on the
track pattern is non-negligible. Elastic scattering cross sections were fitted
by Brenner and Zaider51 mainly based on values obtained by experiments
for electrons below 200 eV. For higher energies the screened Rutherford
model is applied. A recent model was published by Champion et al.52 and
implemented into Geant4-DNA. Michaud et al.53 have also published elas-
tic scattering cross sections for electrons collisions with ice, however, these
values are not expected to be the same as for liquid water. In fact, cross sec-
tions of electron collisions have been abundantly studied in the literature;
for the sake of simplicity it is not possible to state all the proposed models
in this work. The main challenge for Monte-Carlo studies today remains in
simulating the so called sub-excitation electrons. When the electron energy
drops below about 12–8 eV its wave length size approaches the target
88 Ziad Francis
molecule dimensions and reaches ∼1 nm for energies around 1.5 eV. For
these energies the dominating processes are elastic scattering, vibrational
and rotational excitations. Experiments on sub-excitation electrons and
on electrons thermalization processes in water are technically very difficult
to realize so the data published in the literature is very scarce and the few
available sources are not in full agreement. Since there are no theoretical
models that are able to reproduce or predict the inelastic cross section shape
for such low energy electrons in water a commonly adopted procedure is
to take the data presented by Michaud et al.53 for ice and scale its values
to liquid water. The scaling is usually done by enhancing the cross section
values by a factor of 2.29 This procedure is very approximate but it can be
used as an extrapolation of electron tracks till complete energy loss, con-
sidering that the cross sections for energies below 2 eV are obtained by a
simple extrapolation of the lowest energy cross section. One way to verify
the reliability of the so obtained simulations is to compare the ranges of
sub-excitation electrons with the ranges published by the mentioned work
of Michaud et al.
2.3 Sub excitation electrons and the chemical phase

Moreover, investigations of the effect of electrons with energies below
20 eV on DNA molecules are still in progress by many authors, e.g.,
Boudaiffa et al.,54 Huels et al.,55 Sanche.56 It has been proven that for
specific energies of electrons (around 8–10 eV) there is a certain resonance
leading to a peak in the yields of DNA strand breaks. An example of low
energy electrons calculations showed that ranges of such electrons can reach
values as high as ∼30 nm which is relatively high in comparison with DNA
nucleosome size according to Uehara and Nikjoo,57 Meesungnoen et al.,58
and Munoz et al.59 Thus, numerical simulations for such electrons should
be reviewed and improved through accurate cross-sections calculations or
measurements.
The interest in these low energy electrons comes not only from the fact
that they can induce direct damage in the DNA but also at the end of its
track an electron becomes aqueous and can interact with water molecules,
oxygen or hydrogen in the medium leading to a free radical creation. Free
radicals are very reactive species and can interact with the DNA causing
strand breaks. Also the local energy depositions can lead to chemical reac-
tions in the medium creating new interactive species that can modify the
DNA structure and cause strand breaks. Creation of chemical elements
depends on the initial preceding physical process at the interaction point.
For numerical simulations chemical species creation may be summarized

as following; an ionized water molecule (H2O+) decays into H3O+ and
OH, and an excited water molecule (H2O*) decays into one of seven chan-
nels depending on the excitation state leading to H, OH, H2O2, H3O+,
or aqueous electrons production. The branching ratios for the different
channels may vary slightly in the literature from one source to another.
Differences can be due to the adjustments that are made by authors in order
to agree with the data obtained by experiments. In fact, using different cross
section models in the physical phase leads to differences in the initial energy
deposit distributions that can be corrected when handling chemistry. Then
the created species are diffused in the medium during the chemical phase
which happens between 10−12 and 10−6 s after the passage of the ionizing
particle. Diffusion follows a Brownian dynamics and the simulation at this
phase is processed using time steps that are usually of the order of 10−12 of
a second.The reason behind using time steps is that the different species can
interact with each other’s and for every time step new positions are assigned
for each element and a test runs over the different radicals in the simula-
tion to check their separating distances. Each radical type has an interaction
radius and the interaction of neighboring radicals can lead to new chemical
elements creation. The reaction radius depends on both neighboring free
radicals and can vary from 2.5 × 10−5 nm to 0.94 nm according to calcula-
tions of Uehara and Nikjoo.57 This simulation approach was also adopted
by Frongillo et al.60 and the yields of H2 and aqueous electrons from 10−12
to 10−6 s after 5 MeV and 300 MeV protons irradiations were compared to
experimental measurements showing good agreement. The chemical phase
simulation is also taken into account by other codes that we will not detail
in this chapter for the sake of brevity. However, most of the mentioned
codes are very specific and somehow difficult to access since they were built
for single type of local applications in their development laboratories. For
this reason chemistry processes were included in the Geant4-DNA toolkit
and will be released in the future public versions of Geant4 after going
through the required validation phase.
3. RADIATION MICRODOSIMETRY ANALYSIS

3.1 Theoretical and experimental microdosimetry
Radiation can be characterized using concepts introduced by the micro-
dosimetry formalism and that can be obtained numerically and experi-
mentally. Microdosimetry was first introduced by Rossi in the sixties in an
90 Ziad Francis
effort to improve the classic dosimetry through more accurate concepts. A

global review about Rossi and his work with the historical development
of microdosimetry can be found in Kellerer et al.61 Numerical methods
for microdosimetry quantities calculations were reviewed and thoroughly
explained by Kellerer and Chmelevsky.62, 63 Then the mentioned formal-
ism found its most important applications in measurement techniques for
radiation protection purposes and in radiation therapy.
In order to show the separating edge between dosimetry and micro-
dosimetry, a simple simulation of 400 MeV/amu carbon ions crossing a
spherical volume of water was realized using Geant4.9.5 standard pro-
cesses. The energy deposit points were scored in the target and the total
energy per mass unit was calculated for different target dimensions. The
results were reported in Figure 4.1 showing the variation of the energy
deposit with the different mass values. We can notice that for targets
Figure 4.1 Energy deposit per mass unit for spherical targets of different dimensions
irradiated with 400 MeV/amu carbon ions. The results are presented versus the mass of
the irradiated target and were obtained using Geant4.9.5 Monte-Carlo simulation. The
lines on the plot are kept to guide the eye.
below ∼500 g the energy deposit varies quickly with the target mass while
showing a more stable behavior for targets above this mass threshold. Note
that this threshold depends on the mean free path of the particle in the
medium that will strongly affect the energy deposition distribution inside
of the target. In fact, in large volumes the total energy deposited by a par-
ticle is not a stochastic value and remains almost the same using different
calculation methods but for relatively small volumes, e.g., in the presented
example volumes with less than 500 g of water, the particle energy loss
within the target varies from one event to another so it is considered a
stochastic variable that we call the “imparted energy” in the microdosim-
etry formalism. The specific energy is the imparted energy per mass unit
and is generally noted by the letter “z,” it is the analog of the deposited
dose in classic dosimetry. The lineal energy transfer is obtained by dividing
the imparted energy in a target over the mean cord length of the tracks
crossing the target volume, and it is the analog of the linear energy trans-
fer in classic dosimetry. The specific energy and the lineal energy transfer
are stochastic quantities and their distributions can be of importance for
radiation quality assessment. Lindborg and Nikjoo64 studied the radiation
quality factors for radiotherapy applications using X-ray, protons, neutrons,
and carbon ions irradiations and Nikjoo et al.65 presented a database of
microdosimetry quantities calculated for electrons, protons, alpha par-
ticles and carbon ions. At the HIMAC facility, experiments published by
Tsuda et al.66 were carried out measuring the lineal energy distributions
for carbon ions and also for five ion fragments (proton, helium, lithium,
beryllium, and boron) in the lineal-energy range of 0.1–1000 keV/μm
at eight different depths between 7.9 and 147.9 mm in an acrylic phan-
tom.67 These kinds of measurements are mostly done using adapted tissue
proportional equivalent counters (TEPC). Tissue-equivalent gas volumes
are used to simulate the presence of a tissue volume with micrometric or
nanometric dimensions. This assumption considers that the energy loss
of an energetic particle would be the same through both targets. This is
achieved by considering that the energy loss in tissue and gas is obtained
respectively by:

dE dE
�Et = ρt lt and �Eg = ρ g lg , (3)
ρdx t ρdx g
where ρ represents the density of the tissue and l the mean chord length
of the target volume.
92 Ziad Francis
As detailed in the International Commission on Radiation Units and

Measurements (ICRU) report 16 on Microdosimetry,68 the volume of a
biological tissue can be simulated by a gas volume when the atomic com-
positions of the tissue and the gas are identical and the mass stopping power
is independent of the density. In spite of the technical difficulties, e.g.,
detector’s wall effect and others, the challenge using this type of dosimeters
was always to reach the finest resolution of energy deposit measurements
on the nanometric level. This inspired new techniques that are able to
reach this scale for different radiation types, some related details can be
found in De Nardo et al.69 and Grosswendt et al.70
These measurements are now frequently used to characterize radiation
fields and are also of interest for Monte-Carlo codes validation. Using the
code described by Bigildeev and Michalik,71 Palajova et al.72 calculated the
lineal energy distributions for protons with energies 47, 62, 76, 93, 115, 139,
and 172 MeV crossing a spherical volume of water with 3 μm diameter.
The dimensions of the target are chosen to match with the TEPC detector
that is used in the experiments of Borak et al.12 The distributions and the
mean values of the calculated lineal energies were compared to the results
obtained by the experiment. Two experimental sets of data were presented;
one was obtained using the mentioned TEPC and the other set using a
position-sensitive silicon spectrometer. A better agreement was obtained
with simulations involving only ionization processes. For the sake of
validation, the Geant4-DNA package was also used to generate frequency
distributions and mean lineal energies for the same target geometry. The
results of mean lineal energies are reported in Table 4.1 showing data of
Borak et al. and calculations of Palajova et al. compared with Geant4-DNA
Table 4.1 Frequency mean lineal energy obtained by the experiments of Borak et al.12
using TEPC counter and a silicon detector, by the calculations of Palajova et al.79 and
calculations of Francis et al.80 using the Geant4-DNA processes of the Geant4 toolkit
Protons TEPC12 Silicon Palajova et al.79 Geant4-DNA
energy (MeV) detector12 calculations calculations80
47 1.03 1.5 – 1.36
62 0.94 1.06 0.84 1.06
76 0.81 0.86 – 0.88
93 0.69 0.73 0.63 0.725
115 0.58 0.63 – –
139 0.5 0.55 0.5 –
172 0.43 0.51 0.42 –
calculations and a good agreement was obtained for protons with the previ-
ously mentioned energies.35
Another comparison between Geant4-DNA and the code by Bigildeev
and Michalik71 consisted on shooting 1 MeV protons through separate
spherical water targets of 30 nm and 2 nm of diameter. Then the total
energy deposit is computed taking into account the build-up effect by
placing the source at a certain distance from the target achieving electronic
equilibrium. The results consisted on showing the total energy deposit
per mean track length in the target versus the eccentricity of the incident
beam which is the distance between the beam and a parallel segment going
through the target center. The two codes are in agreement for both target
sizes and the results are reported in Tables 4.2 and 4.3.
Table 4.2 Energy deposition per track length unit of 1 MeV protons crossing spherical
water target with 30 nm diameter versus the incident beam eccentricity. The compari-
son between the 2 codes calculations shows a good agreement
Eccentricity (nm) Code by Bigildeev and Geant4-DNA (eV/nm)
Michalik71 (eV/nm)
0 21.65 22.6
5 20.34 21.35
10 15.77 16.6
12 12.47 12.4
15 3.29 3.46
18 0.98 0.9
20 0.63 0.7
Table 4.3 Energy deposition per track length unit of 1 MeV protons crossing spherical
water target with 2 nm diameter versus the incident beam eccentricity
Eccentricity (nm) Code by Bigildeev and Geant4-DNA (eV/nm)
Michalik71 (eV/nm)
0 9.6 9.7
0.2 9.34 9.6
0.4 8.73 9.25
0.6 7.68 8.09
0.8 6.04 6.2
1 1.71 2.28
1.2 1.32 1.15
1.4 1.07 0.916
1.8 0.69 0.799
2.5 0.35 0.33
94 Ziad Francis
3.2 Ions RBE estimation

Most of the numerical studies on energy deposition in small nanometric
volumes were influenced by the scale of the DNA fragments size. The
outcome of these studies started to be put in practice for radioprotec-
tion and radiobiology as summarized by Lindborg and Nikjoo.64 Gerlach
et al.73 have already used a derivate formalism to calculate radiation RBE
for radiotherapy use. Experiments in microdosimetry started mostly using
X-rays and gamma radiation due to the accessibility of such sources
and the data is somehow abundantly available for this type of particles.
However, with the new needs in the ion therapy domain, studies had to
be made on a wider range of particles covering configurations that are in
use today in the medical medium. Moreover, for carbon ions, due to the
fragmentation effects the radiation field becomes very complex includ-
ing more than seven secondary particles that can contribute significantly
to the total energy delivered in the irradiated medium, e.g., protons,
alpha particles, lithium ions, tritons, deuterons, beryllium, and boron ions
including also the different elements isotopes. A simulation of carbon ions
crossing a water volume was carried out using Geant4 version 9.5. The
energy of the beam was 400 MeV/amu since this is the highest energy that
is used for medical accelerators and this would show the upper limit of
secondary contributions in radiotherapy irradiations. For more precision,
the quantum molecular dynamics model was used instead of the binary
ion cascade. The obtained results showed that 64.1% of the global dose
is deposited by carbon ions and up to 35.9% by the produced fragments
including all their isotopes. Protons contribution was about 14% against
13% for alpha particles. Other fragments contributions include lithium
1.7%, boron 4.2%, beryllium 1.3%, deuteron 1.3%, and triton 0.4%,
approximately. Neutrons are produced abundantly in the fragmentation
processes and most of their interactions in water are with hydrogen atoms
producing recoil protons so their contribution is indirectly accounted for
with protons contribution.
It is to mention here that the Geant4 fragmentation models were inves-
tigated and benchmarked with the Fluka code74 in comparison with some
experimental values. The comparison included total nuclear cross sections
for carbon interaction with hydrogen and oxygen between 10 MeV/amu
and 500 MeV/amu, total and partial charge change cross sections in water
and polycarbonate as well as other results obtained from simulations like
angular yields of H, He, Li, and B and yields of fragments with respect to
the depth of the track in water. Although an agreement can be seen for the
general trend of the obtained results, in the case of single differential values
discrepancies of 10% are noticed between the two codes. However, since
data is still scarce for this kind of validations, and here we are only pre-
senting a general approximate idea about fragmentation yields and energy
deposition, we will assume that the present state of fragmentation processes
in Geant4 is reliable for our study.
The energy deposition by 400 MeV/amu carbon ions versus the beam
depth in water is shown in Figure 4.2 revealing a Bragg peak around
27.5 cm. The total fragments contribution and individual contributions of
protons, alphas, and boron ions are also illustrated. We can notice that the
energy deposition by fragments can reach distances over 40 cm from the tar-
get entry point mainly due to protons and alpha particles especially beyond
Figure 4.2 Energy deposition by 400 MeV/amu carbon ions at different depths crossing
a water volume. Simulations were done using Geant4 (version 9.5) and the quantum
molecular dynamics for fragmentation processes. The results include total contribution
of fragments to the energy deposition and individual contributions for protons, alpha
particles, and boron ions.
96 Ziad Francis
∼37 cm. As a rough approximation we can conclude from the plot that

the energy deposition per distance unit at 40 cm depth is about 20 MeV/
cm which correspond to 10 eV of deposition per 5 nm of particle distance
mainly due to protons and alphas. Theoretically, assuming that we need at
least ∼8.2 eV of energy deposition to ionize a DNA base, it is obvious that
the fragments energy deposition tail at this depth can still induce DNA
strand breaks and have an effect on biological cells. Thus, this part of the
energy profile cannot be neglected.
On another side the kinetic energy profiles of protons and He nuclei at
30 cm depth are shown in Figure 4.3. The mean kinetic energy for protons
at this point is around ∼100 MeV and for He nuclei it is ∼212 MeV/amu.
The range of particles at these energies easily exceeds 5 cm and may reach
28 cm in the case of He nuclei which leads to an overall depth of 58 cm in
the mentioned water volume. Thus, the risk of mutation and radio-induced
cancer probability should be further investigated for such type of ions.
Figure 4.3 Energy distribution of protons and alphas at 30 cm depth in water target.
Particles are issued form 400 MeV/amu carbon ions fragmentation in the medium.
Figure 4.4 Build-up of B, Be, Li, and triton ions issued from the fragmentation of carbon
ions of 400 MeV/amu crossing a water volume.
Figures 4.4 and 4.5 show the fragments build-up through the target in
number of ions per primary event for B, Be, Li, triton, He, protons, and
neutrons. Gamma photons were also abundantly produced in the medium
but are not represented on our plots. Here also it is easy to notice that pro-
tons and He yields are the highest among other charged ions in the target.
This is also reflected in the fragments energy deposition curves presented in
Figure 4.2 where most of this energy is carried by He and protons.
The previously mentioned study of Gerlach et al.73 was mainly
focused on the analysis of the carbon complex radiation field and the
related RBE calculations for medical applications. Using a TEPC detector,
measurements of microdosimetric spectra were performed near a perspex
phantom surface for carbon ion beam with energies between 89 MeV/
amu and 430 MeV/amu and at different depths including the neighbor-
hood of the Bragg peak area. The advantage of such an experiment is
that measurements are done for the whole field including the fragments
98 Ziad Francis
Figure 4.5 Build-up of He, protons, and neutrons issued from the fragmentation of
carbon ions of 400 MeV/amu crossing a water volume.
cascade effects and their contribution to the dose deposition. Then the
total RBE including contributions of all the secondary particles can be
calculated by weighting the measured lineal energies spectra d(y) with the
so called biological weighting function r(y) as described by the following
expression:

RBE = r(y)d(y)dy. (4)
The biological weighting function depends on the cell type and on the
irradiation conditions; it can be obtained only after experimental trials. In
the study of Gerlach et al.73 the RBE of neutrons were measured at nine
different neutron irradiation facilities for mice crypt cells. The absorbed
dose was the same for the nine irradiations (8 Gy) given in one single frac-
tion. Also the lineal energies of the irradiation facilities were measured for
the same conditions using a special TEPC detector developed for therapy
radiation simulating a tissue volume of 1 μm of diameter. The biological

function was then calculated for the mentioned cell types as a solution
function for the nine irradiation cases. In an attempt to calculate the carbon
ions RBE the same algorithm was used taking into account the biological
function that was obtained after neutron irradiations. Although this method
is an interesting step toward ions RBE calculation using microdosimetry,
the effect of radiation type on the biological function is not clearly known
yet, so the obtained RBE can only be considered as an approximation that
should be validated by experimental RBE measurements in carbon ion
fields and for the same cell type. So far, numerical RBE estimation often
lacks of experimental data in the case of heavy ions irradiation. Although
recent data is being published (e.g., Ref. 66), experiments and RBE mea-
surements regarding therapy ion beams are still scarce in the literature. In
fact, due to the restricted access to the few available carbon irradiation
facilities, data sets that are usually needed to further develop and validate
the related numerical algorithms for microdosimetry properties or RBE
calculations, are somehow hard to obtain and most of the studies are still
based on the results coming from X-rays or other available sources.
4. DNA DAMAGE ESTIMATION

4.1 Track structure detailed approach
Understanding the cell reaction to radiation and the resulting RBE of
different radiation types requires a detailed knowledge of the molecular
characteristics of the damage that is created within the cell. Calculations
of DNA damage might generally be differentiated according to the com-
plexity of the adopted approach and the details that are taken into account
in the DNA modeling phase. In fact the first set of simulations started by
taking particle tracks crossing small cylinders of water with nanometric
dimensions and randomly placed in a water volume. The cylinders pre-
sented an approximation of the chromatin fiber fragments and energy
deposits within such small volumes were studied mainly for X-rays and ions
like protons and alpha particles of different energies. This kind of simula-
tion played an important role in understanding the differences between low
versus high LET radiation effects.75 Then progress was carried out toward
more detailed methods like the complete atomistic DNA modeling, e.g.,
Friedland et al.30 where the PARTRAC code was used to generate tracks
of particles through the target geometries taking into account the chemical
effects in the medium. Simulations included different chromosome shapes
100 Ziad Francis
(lymphocytes and fibroblasts) and different DNA arrangements. The yields

of single and double strand breaks and the frequency distributions of DNA
fragments with different lengths after irradiations were calculated. Further
details about similar procedures can be found in the literature and by other
references that are not mentioned here for the sake of brevity; in fact hav-
ing a geometrical model for the DNA was a principal point for many
projects trying to pre-estimate DNA damage effects using Monte-Carlo
track structure codes. In spite of the many differences between the different
adopted models, the DNA configurations might have a common standard
that was followed by almost all the authors where the DNA double helix is
surrounding the histone volume of few nanometers (6–7 nm of diameter)
forming the nucleosomes. These latter are then placed together in different
configurations to form the chromatin fibers. It is more frequent that at this
level, differences will arise from one author to another since configurations
of nucleosomes and even of chromatin fragments that constitute the whole
chromosome can be completely random or can follow some imposed
criteria to reach a certain DNA density or a desired base pair number in
a fixed volume of the target. Having the DNA geometry at hand one can
distinguish between the contributions of direct versus indirect effects of
radiation and have access to data like DNA damage complexity and result-
ing DNA fragments lengths. As advanced as this procedure can be, having
the complete perfect model of the DNA geometry is a very complex task
not to mention the required computing power that is needed in this case
due to the big number of elements one has to introduce in the simulation.
4.2 Stewart and Semenenko MCDS method

In parallel to the mentioned studies, other methods were found with sim-
pler requirements and leading to a fair approximation of DNA damage cal-
culations. An interestingly fast approach is the one adopted by Semenenko
et al.,24,76 where the initial damage is generated randomly on the DNA
array taking into account the initial particle energy and LET. So the num-
ber of damages is directly related to the particle’s properties but their spatial
distribution is sampled randomly. This approximation avoids the particle
tracks simulation phase and also the chemical species diffusion phase thus
accelerating the simulation by a considerable amount of time. Then the
emphasis is put on the damage analysis, damages on opposite strands sepa-
rated by less than 10 base pairs of distance can form a double strand break
and if more than two damages are present in such a limited area they are
considered as a complex damage. This latter type is considered lethal for
the cell since it cannot be handled by the repair mechanisms, also its pres-
ence might stop the cell reproduction cycle and lead to the cell death. In
fact, in the work of Nikjoo et al.77 damages were classified into categories;
single strand breaks representing only one strand damage and are referred
to as SSB, complex single strand breaks when more than one damage
occur on the same strand within a distance of 10 base pairs and are noted
as SSB+, double strand breaks (DSB) represent two damages on opposite
strands located within the 10 base pairs distance, damages noted by DSB+
are formed by a SSB and a SSB+ located on opposite strands and DSB++
are the most complex damages as they can be formed by a combination of
DSB damages within a limited base pairs distance. This classification was
then adopted by many other studies and other authors since it summarizes
somehow most of the occurring configurations of DNA damage types.
The work of Semenenko and Stewart included also cell repair
activities through Base excision repair and Nucleotide excision repair path-
ways, however, this subject is not to be discussed here since it is out of the
scope of this paragraph. The results obtained by Semenenko and Stewart
showed good agreement with the detailed calculation methods discussed
previously and with the experimental values on DNA mutation frequency
after irradiation. Hereafter when referring to Semenenko and Stewart
approach we mean the Monte-Carlo Damage Simulation (MCDS) that
does not take into account any repair activities in the calculations. This
is seen as an appropriate choice for a comparison with other algorithms
where the repair activities are also neglected.
4.3 Garty statistical approach

Another study that is more based on microdosimetry is the one published
by Garty et al.78 Garty and co-workers assumed that there is a one-to-
one correspondence between the ionizations formed within a considered
sensitive volume and those formed within a DNA segment of equivalent
size. Distributions of ionizations occurring in the sensitive volume can be
obtained experimentally using a gas tissue equivalent counter or by numeri-
cal simulations using a track structure code. The size of the sensitive volume
is chosen to be equivalent to the size of a DNA segment of one or two
helical turns. Thus the sensitive volume is represented by a cylinder of about
∼6 nm of length and ∼4 nm of diameter. These dimensions are based on
the assumption that damages occurring within such a distance can interact
together to form double strand breaks or more complex lesions. Assuming
also that every single ionization has a fixed probability to be converted into a
102 Ziad Francis
DNA lesion and an equal chance to fall on one of the two strands, yields of
single and double strand breaks can be obtained by the following expressions:
ρ ·VSV pSB nion

GSSB = C × × f (nion ) × 2 1 − − (1 − pSB )nion ,
W n
2
ion
(5)
ρ ·VSV pSB nion

GDSB = C × × f (nion ) × 1 − 2 1 − + (1 − pSB )nion ,
W n
2
ion
(6)
where C is a conversion factor, and for C = 9.6 × 10−10 the yield G is
obtained in (Gy−1 Da−1). The sensitive volume VSV is in (nm3), pSB is the
probability that an ionization is transformed into a strand break (usually taken
between ∼9% and ∼12%) and W is the mean energy deposited (in eV) by
a single event in the sensitive volume. f (nion ) is the frequency probability of
ionization number nion induced within the sensitive volume for one event, it
can be obtained experimentally by taking a tissue equivalent detector volume
of the same size of the studied target or also by simulating tracks crossing a
similar sensitive volume. Note that this procedure is purely statistical and its
main advantage is that it requires only one input parameter to be fixed by
the user, which is the strand break induction probability. For pSB = 11. 7%
this approach leads to good agreement with experimental results of gel elec-
trophoresis analysis on plasmid DNA irradiated with electrons and X-rays.
4.4 DBSCAN clustering estimation

Another study based on clustering algorithms is described in Francis et al.26
In this case the DNA distribution inside of the cell nucleus is considered to
be overall homogenous. So the radiation track structure simulations are car-
ried out in a water volume with the same size as the nucleus of the studied
cell type. The energy deposition locations and values are scored in a separate
file that is treated by the algorithm after the simulation phase is completed.
Chemical reactions are not modeled directly only the physical track structure
is calculated, however, in our calculation we take into account that an energy
deposition that is sufficiently close to DNA can induce indirect damage to
this latter. So we consider that there is a certain sensitive area surrounding the
DNA geometry which might also be explained as a geometrical cross sec-
tion defining the percentage of the volume where an energy deposition may
lead to DNA damage. The damage sampling follows also a linear distribution
function having 0 probability of damage induction for energy deposition

below 5 eV and a probability of 1 above 35 eV depositions. At the final stage a
Density Based Clustering Algorithm with Noise (DBSCAN) runs over all the
sampled damage points and according to their separating distances clusters of
several damages are formed.The calculated damages here follow also the clas-
sification of Nikjoo et al.77 for single strand breaks and double strand breaks.
The obtained results for protons with energies between 0.5 MeV and 50 MeV
were compared with the calculations of the PARTRAC code and with
experimental data of plasmid damage after irradiation showing a good agree-
ment. This model takes three main input parameters, the already introduced
geometrical cross section (∼16%), the maximum distance within which dam-
ages belong to the same cluster and its value should remain close to a 10 base
pairs distance (∼3.2 nm) and the damage probability function equation.
Figure 4.6 shows a comparison of the predicted DSB/SSB ratio induced
by protons irradiations with energies between 0.5 MeV and 50 MeV.
Figure 4.6 Comparison of double strand breaks over single strand breaks ratio
obtained with the PARTRAC code,21 the approach described by Garty et al.,78 the clus-
tering method of Francis et al.,25 and the MCDS model described by Semenenko and
Stewart.76
104 Ziad Francis
The comparison includes values from PARTRAC code simulations com-

pared to the results obtained by using the approach of Garty combined
with ionization distributions simulated using the Geant4-DNA processes.
Also the results obtained by the DBSCAN clustering algorithm and tracks
generated with Geant4-DNA are shown with the predictions of the
MCDS of Semenenko and Stewart76 A general agreement can be noticed
between the different approaches although differences start to grow for low
energies meaning with higher LET. In fact in the PARTRAC calculations
the chemical phase was completely modeled including the scavenging
effect that is enhanced at high LET due to the high density of chemical
species created in the medium. Since the chemical species diffusion was not
taken into account in the other calculations, differences may increase with
increasing radiation LET.
In general, these models agree with each other’s and with some experi-
ments on plasmid DNA, although track structures are generated in liquid
water without taking into account the DNA interaction cross sections. It
has been shown in the literature that the differences between water and
DNA bases cross sections cannot be neglected and DNA cross sections
should be used instead of water for DNA simulations. Moreover experi-
ments with sub-excitation electrons showed that these latter can cause
DNA damage even at very low kinetic energies that are not considered
to induce direct ionizations. In fact, studies of Boudaiffa et al.54 showed
that there is a resonant formation of DNA strand breaks irradiated with
electrons of energies between 3 and 20 eV and surprisingly double strand
breaks can still be induced by electrons with energies as low as 5 eV. In most
of the track structure codes that are in use today these low energy electrons
are either killed assuming that their energy is not high enough to induce
direct DNA ionizations or simply transported with one single step of con-
densed history until their end of track location. Few specific codes studying
only sub-excitation electrons can model detailed and precise tracking in
this case. It is clear that in their present state, the approximation algorithms
that were mentioned here before cannot reproduce the results of Boudaiffa
et al. where most of the damages are induced by vibrational and rotational
excitations and by electron attachment processes.
5. CONCLUSION
Here in this chapter a brief summary was presented reviewing the
different basic approaches that can be used for modeling ionizing particles
tracks with molecular scale resolution. An emphasis on the importance of

such studies for the radiobiology and the hadrontherapy fields was built up
during the whole text showing some of the examples that were published
in the literature and presenting a clear idea of the different methodologies
that can be used to study DNA radiation damage.
In the second paragraph describing track structure codes, most of the
mentioned physics models are based on the processes that were included
in the Geant4-DNA package. Although, other existing codes are well
developed in the field, the main difficulty remains mostly in the acces-
sibility of these “homemade” programs that are often built for one single
or a restricted application type. The Geant4-DNA project aims to present
a detailed track structure code that enables all Geant4 users to simulate
particles interactions in biological tissue following a detailed step by step
tracking with nanometric scale precision. The presented particles and their
physical processes are already available to download with the public version
of Geant4 and a typical example of a microdosimetry application is also
provided in the Geant4 examples folder. The particles that can be modeled
with the present version are electrons, protons, alphas and some heavier
ions like carbon, only for a restricted energy range according to the avail-
able cross sections. On another hand, cross sections of particles interactions
in DNA material are under development and the molecules diffusion pro-
cesses that are needed for the chemical phase simulation are going through
validation tests and will be published in the upcoming releases. Also a typi-
cal atomistic model of DNA is under construction and will be included in
one of the examples in the future Geant4 releases.These efforts aim to pres-
ent a publicly available library that can be used by any researcher interested
in the mentioned applications taking benefit from a user support forum and
a publicly available manual that is under continuous development.
In spite of the multiple efforts that are put by the different communi-
ties, improvements are still required to take care of some of the commonly
adopted approximations. Here, tracking sub-excitation electrons in water
and modeling their interactions with DNA material is still an open field. So
far, the angular inelastic diffusion of sub-excitation electrons is considered
isotropic which was never precisely confirmed by theoretical studies nor by
adequate experiments. It is important here to mention the need of a reli-
able set of double differential cross sections taking into account both energy
and angular diffusions of electrons. The importance of these latter par-
ticles is underlined in the previously mentioned study of Boudaiffa et al.54
showing an unexpected sensitivity of DNA toward low incident energies.
106 Ziad Francis
Another point worth to mention is the DNA geometry models that are used
in detailed simulations.These models often represent one single specific chro-
mosome structure while a big variety of configurations are encountered in
reality which limits the applicability of the simulations to very specific cases
of cell irradiations. In fact the genome geometry depends on the cell type
and on the cell replication phase so simulations should include several geo-
metrical models to extend the applicability of their results to a wider range
of irradiation configurations. As far as the author is aware, such advanced
investigations were started by Friedland et al.30 but no other efforts were
published regarding calculations with chromosomes irradiated in different cell
replication phases.
For hadrontherapy, carbon ions fragmentation is now under the scope
of several investigations. The fragmentation gives arise to secondary ions
that are responsible for the tail of energy deposition after the “Bragg peak.”
In order to make a numerical estimation of the RBE of these ions and
their effect on the treatment area, important information can be found
through nanometric scale track structure simulations. In the Geant4-DNA
framework, using a speed scaling procedure with the previously calculated
protons ionization cross sections and an effective charge model, it is possible
to derive ionization cross sections for most of the fragmentation products.
This process is already included in Geant4-DNA and a step by step model
is used to track ions heavier than alpha particles for energies above 1 MeV/
amu taking into account ionization interactions. At the end of its track when
the ion energy drops below the mentioned threshold inaccuracies arise since
nuclear collisions are not taken into account. An adapted model of ions elas-
tic scattering is being studied in an effort to fill the gap of the lacking cross
sections, the final goal would be to follow all the fragmentation products in
a step by step mode until complete ion stop. Also here improvements are
needed regarding the interactions cross sections since most of the discrepan-
cies between nucleus collisions models arise in the low energy parts.
For validation purposes, experiments using microdosimetry techniques
for carbon ion beams analysis are of extreme importance for numerical
codes control and should be supported in order to provide a wider data
base. The required data may include lineal energies spectra in nanometric
targets of different sizes and for different beam energies. Also DNA dam-
age experiments using typical DNA samples like plasmid are of importance
for the validation of the different damage estimation algorithms. Most of
the presented approaches that were developed for damage prediction are
validated with the results of experiments using different radiation sources
sometimes of different types. However, DNA damage yields depend
strongly on the particle type even for different particles of the same LET as
demonstrated in Francis et al.26 So for hadrontherapy applications, a precise
validation of DNA damage prediction methods should mainly rely on val-
ues issued from experiments that are done using, to an extent, the same real
conditions and parameters as in the treatment cases. Finally, we can men-
tion that a comparison between the RBE calculations obtained using the
multiple different studies would be of importance for the medical purposes.
ACKNOWLEDGMENTS
The author would like to thank the Geant4 collaboration, the low energy
electromagnetic working group and the Geant4-DNA project.The support
of Professor N. Mason, Dr. A. Stypczynska and the Open University in
Milton Keynes is highly appreciated.
REFERENCES
1. Brenner, D. J.; Hlatky, L. R.; Hahnfeldt, P. J., et al. The linear-quadratic model and most
other common radiobiological models result in similar predictions of time–dose rela-
tionships. Rad. Res. 1998, 150, 83.
2. Kramer, M.; Scholz, M. Treatment planning for heavy-ion radiotherapy: calculation and
optimization of biologically effective dose. Phys. Med. Biol. 2000, 45, 3319.
3. Kramer, M.; Jakel, O.; Haberer, T., et al. Treatment planning for heavy-ion radiotherapy:
physical beam model and dose optimization. Phys. Med. Biol. 2000, 45, 3299.
4. Jakel, O.; Kramer, M.; Schulz-Ertner, D., et al. Treatment planning for carbon ion radio-
therapy in Germany: review of clinical trials and treatment planning studies. Radiother.
Oncol. 2004, 73, 86.
5. Rietzel, E.; Schardt, D.; Haberer, T. Range accuracy in carbon ion treatment planning
based on CT-calibration with real tissue samples. Radiat. Oncol. 2007, 2, 14.
6. Stypczynska, A.; Ptasinska, S.; Bahnev, B., et al. The influence of amino acids on DNA
damage induced by cold plasma radiation. Chem. Phys. Lett. 2010, 500, 313.
7. Ptasinska, S.; Stypczynska, A.; Nixon, T., et al. X-ray induced damage in DNA moni-
tored by X-ray photoelectron spectroscopy. J. Chem. Phys. 2008, 129, 065102.
8. Smialek, M. A.; Moore, S. A.; Mason, N. J., et al. Quantification of radiation-induced
single-strand breaks in plasmid DNA using a TUNEL/ELISA-based assay. Radiat. Res.
2009, 172, 529.
9. Milligan, J. R.; Aguilera, J. A.; Nguyen, T. T., et al. DNA strand-break yields after post-
irradiation incubation with base excision repair endonucleases implicate hydroxyl radi-
cal pairs in double-strand break formation. Int. J. Radiat. Biol. 2000, 76, 1475.
10. Gulston, M.; Fulford, J.; Jenner, T., et al. Clustered DNA damage induced by gamma
radiation in human fibroblasts (HF19), hamster (V79–4) cells and plasmid DNA is
revealed as Fpg and Nth sensitive sites. Nucleic Acids Res. 2002, 30, 3464.
11. Magnander, K.; Hultborn, R.; Claesson, K., et al. Clustered DNA damage in irradiated
human diploid fibroblasts: influence of chromatin organization. Radiat. Res. 2010, 173, 272.
12. Borak, T. B.; Doke, T.; Fuse, T., et al. Comparisons of LET distributions for protons with
energies between 50 and 200 MeV determined using a spherical tissue-equivalent pro-
portional counter (TEPC) and a position-sensitive silicon spectrometer (RRMD-III).
Radiat. Res. 2004, 162, 687.
13. De Nardo, L.; Colautti, P.; Baek, W. Y., et al. Track nanodosimetry of an alpha particle.
Radiat. Prot. Dosimetry 2002, 99, 355.
108 Ziad Francis
14. Grosswendt, B.; Pszona, S. The track structure of alpha-particles from the point of view
of ionization-cluster formation in “nanometric” volumes of nitrogen. Radiat. Environ.
Biophys. 2002, 41, 91.
15. Bantsar, A.; Grosswendt, B.; Pszona, S., et al. Radiation oncology Single track nanodo-
simetry of low energy electrons Nuc. Instum. Methods Phys. Res., Sect. A 2009, 599, 270.
16. Fasso, A.; Ferrari, A.; Ranft, J., et al. In New developments in FLUKA modeling hadronic and
EM interactions, SARE 3. Proceedings of the Third Workshop on Simulating Accelerator
Radiation Environments, Japan, KEK Proceedings 97–5, June, H/R/D; Hirayama, H,
Ed., 1997; p 32.
17. Agostinelli, S.; Allison, J.; Amako, K., et al. Geant4-a simulation toolkit. Nucl. Instrum.
Methods Phys. Res., Sect. A 2003, 506, 250.
18. Hughes, G.; Adams, K. J.; Chadwick, M. B., et al. In MCNPX – the LAHET/MCNP
code merger SARE 3. Proceedings of the Third Workshop on Simulating Accelerator
Radiation Environments, Japan, KEK Proceedings 97–5, June, H/R/D; Hirayama, H.,
Ed., 1997; p 44.
19. Gudowska, I.; Sobolevsky, N.; Andreo, P., et al. Ion beam transport in tissue-like media
using the Monte Carlo code SHIELD-HIT. Phys. Med. Biol. 2004, 49, 1933.
20. Henkner, K.; Sobolevsky, N.; Jakel, O., et al. Test of the nuclear interaction model in
SHIELD-HIT and comparison to energy distributions from GEANT4. Phys. Med. Biol.
2009, 54, 509.
21. Friedland, W.; Jacob, P.; Bernhardt, P., et al. Simulation of DNA damage after proton
irradiation. Radiat. Res. 2003, 159, 401.
22. Ballarini, F.; Alloni, D.; Facoetti, A., et al. Heavy-ion effects: from track structure to DNA
and chromosome damage. New J. Phys. 2008, 10, 075008.
23. Ponomarev, A. L.; Sachs, R. K. Polymer chromosome models and Monte Carlo simula-
tions of radiation breaking DNA. Bioinformatics 1999, 15, 957.
24. Semenenko, V. A.; Stewart, R. D.; Ackerman, E. J. Monte Carlo simulation of base and
nucleotide excision repair of clustered DNA damage sites. I. Model properties and pre-
dicted trends. Radiat. Res. 2005, 164, 180.
25. Francis, Z.;Villagrasa, C.; Clairand, I. Simulation of DNA damage clustering after proton
irradiation using an adapted DBSCAN algorithm. Comput. Methods Programs Biomed.
2011, 101, 265.
26. Francis, Z.; Incerti, S.; Ivanchenko, V., et al. Monte Carlo simulation of energy-deposit
clustering for ions of the same LET in liquid water. Phys. Med. Biol. 2012, 57, 209.
27. Stewart, R. D.;Wilson,W. E.; McDonald, J. C., et al. Microdosimetric properties of ionizing
electrons in water: a test of the PENELOPE code system. Phys. Med. Biol. 2002, 47, 79.
28. Francis, Z.; Incerti, S.; Karamitros, M., et al. Stopping power and ranges of electrons,
protons and alpha particles in liquid water using the Geant4-DNA package. Nucl.
29. Gervais, B.; Beuve, M.; Olivera, G. H., et al. Numerical simulation of multiple ionization
and high LET effects in liquid water radiolysis. Radiat. Phys. Chem. 2006, 75, 493.
30. Friedland, W.; Paretzke, H. G.; Ballarini, F., et al. First steps towards systems radiation
biology studies concerned with DNA and chromosome structure within living cells.
Radiat. Environ. Biophys. 2008, 47, 49.
31. Friedland, W.; Dingfelder, M.; Kundrat, P., et al. Track structures, DNA targets and radia-
tion effects in the biophysical Monte Carlo simulation code PARTRAC. Mutat. Res.
2011, 711, 28.
32. Wilson, W. E.; Nikjoo, H. A Monte Carlo code for positive ion track simulation. Radiat.
Environ. Biophys. 1999, 38, 97.
33. Semenenko,V. A.;Turner, J. E.; Borak,T. B. NOREC, a Monte Carlo code for simulating
electron tracks in liquid water. Radiat. Environ. Biophys. 2003, 42, 213.
34. Incerti, S.; Ivanchenko, A.; Karamitros, M., et al. Comparison of GEANT4 very low
energy cross section models with experimental data in water. Med. Phys. 2010, 37, 4692.
35. Francis, Z.; Incerti, S.; Capra, R., et al. Molecular scale track structure simulations in liquid
water using the Geant4-DNA Monte-Carlo processes. Appl. Radiat. Isot. 2011, 69, 220.
36. Dingfelder, M.; Inokuti, M.; Paretzke, H. G. Inelastic-collision cross-section of liquid
water for interactions of energetic protons. Radiat. Phys. Chem. 2000, 59, 255.
37. Heller, J. M.; Hamm, R. N.; Birkhoff, R. D., et al. Collective oscillation in liquid water.
J. Chem. Phys. 1974, 60, 3483.
38. Hayashi, H.; Watanabe, N.; Udagawa, Y. Optical spectra of liquid water in vacuum UV
region by means of inelastic X-ray scattering spectroscopy. J. Chem. Phys. 1998, 108, 823.
40. Dingfelder, M.; Toburen, L.H.; Paretzke; H.G. An effective charge scaling model for
ionization of partially dressed helium ions with liquid water. In The Monte Carlo Method:
Versatility Unbounded in a Dynamic Computing World, April 17–21, 2005; LaGrange Park,
Tennessee: American Nuclear Society, 2005.
41. Barkas,W. H.; Birnbaum,W.; Smith, F. M. Mass-ratio method applied to the measurement
of L-Meson masses and the energy balance in pion decay. Phys. Rev. 1956, 101, 778.
42. Booth,W.; Grant, I. S. Energy loss of oxygen and chlorine ions in solid. Nucl. Phys. 1965,
63, 481.
43. Plante, I.; Cucinotta, F. Ionization and excitation cross sections for the interaction of
HZE particles in liquid water and application to Monte Carlo simulation of radiation
tracks. New J. Phys. 2008, 10, 125020.
44. Porter, L. E.; Cabrera-Trujillo, R.; Sabin, J. R. The Barkas-Effect Correction to Bethe–Bloch
Stopping Power. Advances in Quantum Chemistry; Academic Press
45. Dingfelder, M.; Hantke, D.; Inokuti, M., et al. Electron inelastic scattering cross sections
in liquid water. Radiat. Phys. Chem. 1998, 53, 1.
46. Emfietzoglou, D.; Nikjoo, H. The effect of model approximations on single-collision
distributions of low-energy electrons in liquid water. Radiat. Res. 2005, 163, 98.
47. Bousis, C.; Emfietzoglou, D.; Hadjidoukas, P., et al. Electron ionization cross-section
calculations for liquid water at high impact energies. Nucl. Instrum. Methods Phys. Res.,
Sect. B 2008, 266, 1185.
48. Vriens, L. Binary-Encounter Electron-Atom Collision Theory. Phys. Rev. Lett. 1966, 141, 88.
49. Ashley, J. C.; Ritchie, R. H.; Brandt, W. Z3 effect in the stopping power of matter for
charged particles. Phys. Rev. B 1972, 5, 2393.
50. Kim,Y. K.; Rudd, M. E. Binary-encounter-dipole model for electron-impact ionization.
Phys. Rev. A 1994, 50, 3954.
51. Brenner, D. J.; Zaider, M. A computationally convenient parameterization of experi-
mental angular distributions of low energy electrons elastically scattered off water vapor.
Phys. Med. Biol. 1983, 29, 443.
52. Champion, C.; Incerti, S.; Aouchiche, H., et al. A free-parameter theoretical model for
describing the electron elastic scattering in water in the Geant4 toolkit. Radiat. Phys.
Chem. 2009, 78, 745.
53. Michaud, M.; Wen, A.; Sanche, L. Cross sections for low-energy (1–100 eV) electron
elastic and inelastic scattering in amorphous ice. Radiat. Res. 2003, 159, 3.
54. Boudaiffa, B.; Cloutier, P.; Hunting, D., et al. Resonant formation of DNA strand breaks
by low-energy (3 to 20 eV) electrons. Science 2000, 287, 1658.
55. Huels, M. A.; Boudaiffa, B.; Cloutier, P., et al. Single, double, and multiple double strand
breaks induced in DNA by 3–100 eV electrons. J. Am. Chem. Soc. 2003, 125, 4467.
56. Sanche, L. Low energy electron-driven damage in biomolecules. Eur. Phys. J. D 2005,
35, 367.
57. Uehara, S.; Nikjoo, H. Monte Carlo simulation of water radiolysis for low-energy
charged particles. J. Radiat. Res. 2006, 47, 69.
58. Meesungnoen, J.; Jay-Gerin, J. P.; Filali-Mouhim, A., et al. Low-energy electron penetra-
tion range in liquid water. Radiat. Res. 2002, 158, 657.
110 Ziad Francis
59. Munoz, A.; Blanco, F.; Garcia, G., et al. Single electron tracks in water vapor for energies
below 100 eV. Int. J. Mass Spectrom. 2008, 277, 175.
60. Frongillo, Y.; Goulet, T.; Fraser, M. J., et al. Monte Carlo simulation of fast electron and
proton tracks in liquid water – II. Nonhomogeneous chemistry. Radiat. Phys. Chem.
1998, 51, 245.
61. Kellerer, A. M. Microdosimetry: reflections on Harald Rossi. Radiat. Prot. Dosimetry
2002, 99, 17.
62. Kellerer, A. M.; Chmelevsky, D. Concepts of Microdosimetry. I. Quantities. Radiat.
Environ. Biophys. 1975, 12, 61.
63. Kellerer, A. M.; Chmelevsky, D. Concepts of Microdosimetry. II. Probability distribu-
tions of the microdosimetric variables. Radiat. Environ. Biophys. 1975, 12, 205.
64. Lindborg, L.; Nikjoo, H. Microdosimetry and radiation quality determinations in radia-
tion protection and radiation therapy. Radiat. Prot. Dosimetry 2011, 143, 402.
65. Nikjoo, H.; Uehara, S.; Emfietzoglou, D., et al. A database of frequency distributions
of energy depositions in small size targets by electrons and ions. Radiat. Prot. Dosimetry
2011, 143, 145.
66. Tsuda, S.; Sato, T.; Takahashi, F., et al. Measurement of microdosimetric spectra with a
wall-less tissue-equivalent proportional counter for a 290 MeV/u 12C beam. Phys. Med.
Biol. 2010, 55, 5089.
67. Endo, S.; Takada, M.; Onizuka, Y., et al. Microdosimetric evaluation of secondary par-
ticles in a phantom produced by carbon 290 MeV/nucleon ions at HIMAC. J. Radiat.
Res. 2007, 48, 397.
68. ICRU. Microdosimetry Report 36: ICRU, International Commission on Radiation
Units and Measurements, Bethesda, MD, 1983.
69. De Nardo, L.; Colautti, P.; Conte,V., et al. Ionization-cluster distributions of alpha-par-
ticles in nanometric volumes of propane: measurement and calculation. Radiat. Environ.
Biophys. 2002, 41, 235.
70. Grosswendt, B.; De Nardo, L.; Colautti, P., et al. Experimental equivalent cluster size
distributions in nanometric volumes of liquid water. Radiat. Prot. Dosimetry 2004,
110, 851.
71. Bigildeev, E. A.; Michalik,V. Charged particle tracks in water of different phases, Monte
Carlo simulation of electron tracks. Radiat. Phys. Chem. 1996, 47, 197.
72. Palajova, Z.; Spurny, F.; Davidkova, M. Microdosimetry distributions for 40–200 MeV
protons. Radiat. Prot. Dosimetry 2006, 121, 376.
73. Gerlach, R.; Roos, H.; Kellerer, A. M. Heavy ion RBE and microdosimetric spectra.
Radiat. Prot. Dosimetry 2002, 99, 413.
74. Bohlen, T. T.; Cerutti, F.; Dosanjh, M., et al. Benchmarking nuclear models of FLUKA
and GEANT4 for carbon ion therapy. Phys. Med. Biol. 2010, 55, 5833.
75. Goodhead, D. T.; Nikjoo, H. Track structure analysis of ultrasoft X-rays compared to
high- and low-LET radiations. Int. J. Radiat. Biol. 1989, 55, 513.
76. Semenenko,V. A.; Stewart, R. D. A fast Monte Carlo algorithm to simulate the spectrum
of DNA damages formed by ionizing radiation. Radiat. Res. 2004, 161, 451.
77. Nikjoo, H.; O’Neill, P.; Wilson, W. E., et al. Computational Approach for Determining
the Spectrum of DNA Damage Induced by Ionizing Radiation. Radiat. Res. 2001,
156, 577.
78. Garty, G.; Schulte, R.; Shchemelinin, S., et al. A nanodosimetric model of radiation-
induced clustered DNA damage yields. Phys. Med. Biol. 2010, 55, 761.
79. Palajova, Z.; Spurny, F.; Davıdkova, M. Microdosimetry distributions for 40–200 MeV
protons. Radiat. Prot. Dosimetry 2006, 121, 376.
80. Francis, Z.; Incerti, S.; Capra, R., et al. Molecular scale track structure simulations in
liquid water using the Geant4-DNA Monte-Carlo processes. Appl. Radiat. Isot. 2011,
69, 220.
CHAPTER FIVE
Verifying Radiation Treatment in

Proton Therapy via PET Imaging
of the Induced Positron-Emitters
Joanne Beebe-Wang, Paul Vaska, F. Avraham Dilmanian,
Stephen G. Peggs and David J. Schlyer
Contents
1. Introduction 112
2. Positron emitter production 113
3. Nuclear reaction cross sections 114
4. Monte Carlo simulations 119
5. Results 121
6. Discussion and conclusions 124
Acknowledgments 125
References 125
Abstract
Positron Emission Topography (PET) is a promising technique to verify the dose distri-
bution from proton therapy, a precise treatment modality increasingly used in radia-
tion oncology because its radiation pattern conforms more closely to the configuration
of a tumor than does that from X-ray radiation, thereby sparing normal healthy tissue.
Proton therapy produces positron-emitting isotopes along the beam’s path, allowing
PET to image the distribution of therapeutic energy, viz., a form of quality assurance
of the treatment. This ability is especially important when treating heterogeneous
organs, such as the lungs or the head-and-neck, where calculating the expected dose
distribution for treatment is complex. Here, we present the findings from our Monte
Carlo simulations of the yield of positron emitters produced by proton beams of up
to 250 MeV, followed by our statistically realistic Monte Carlo simulation of the images
expected from a clinical PET scanner. Our emphases lay in predicting accurately the
distribution of positron emitters, and in determining the quality of the PET signal near
the Bragg peak that is critical to the success of PET imaging for verifying the proton
beam’s location and dosimetry. We also demonstrate that the results depend strongly
on the accuracy of the available nuclear reaction cross section data. Accordingly, we
quantify the differences in the calculated positron-emitter yields from four different
sets of such data, comparing them to the simulated distributions of positron-emitter
production and absorbed proton energies.

112 Joanne Beebe-Wang et al.
1. INTRODUCTION
The subject of the present study relates to the main topic of this book,
namely high-energy collisions of heavy ions with matter, if we extend the
subject to cover (a) biological tissues and (b) hadron therapy of cancer.1, 2
The goal is to use Monte Carlo simulations for evaluating the applicability
of positron emission tomography (PET)3 for monitoring dose distributions
produced in biological subjects in proton therapy. This chapter presents
such a simulation using a voxel-based Monte Carlo algorithm. Application
of PET to proton therapy dose monitoring rests upon detecting positron
annihilation γ rays following the decay of the small amounts of positron
emitters (typically 11C, 13N, and 15O) produced via the non-elastic nuclear
reaction of protons with the target nuclei of the irradiated tissue. The input
data for such simulations are mainly electronic and nuclear stopping powers
of protons in matter, and the attenuation of the resulting 511 keV gamma
rays in their way out of the subject towards the detector.4, 5 To arrive at real-
istic estimates, we employ several data sets from a comprehensive database of
the available cross sections and stopping powers.
The present work clearly shows that PET is potentially a very useful
tool for monitoring the distribution of the dose deposited in a patient dur-
ing proton therapy. The adequacy of the implementation of the therapy can
be verified by comparing PET images revealing the distribution of positron
activity with the predicted target dose-distribution used in planning treat-
ment. The most important feature of this technique is that the resulting
PET images are the inverse of the deposited energy distribution in the sub-
ject because the bulk of the positron-emitting isotopes are produced in the
tissues proximal to the target and in fact such isotope production is dimin-
ished at the target itself. This is because the non-elastic nuclear reaction
cross sections, where the largest amount of protons’ energies are deposited,
are the largest at the protons’ entrance to the subject and gradually dimin-
ish towards reaching the Bragg peak. However, effective dose verification
still can be made by focusing at the tissue volume at target’s edges and by
comparing the radioisotope distribution produced in that region as mea-
sured by PET with the yield of the positron emitters predicted from the
treatment planning code.
Various researchers explored the possibility of monitoring proton ther-
apy with PET through experimental measurements6–10 and by computer
simulations.11–14 So far, most of the simulations studies were performed
with general particle transport codes originally developed for high-energy
Verifying Radiation Treatment in Proton Therapy via PET Imaging 113
physics, such as FLUKA,15 GEANT4,16 and SHIELD-HIT.17 Some efforts

have been made to improve and evaluate the underlying physics in the low
energy range incorporated in the general codes, as well as to implement
them into clinical applications.18–22 However, there is a need to benchmark
them against one or more dedicated proton transport codes. In addition,
the reliability of nuclear reaction cross section data is essential to predicting
positron-emitter production. However, such data are very limited currently,
either incomplete or not compiled, because there are very few experimen-
tal measurements of these reactions, especially in the energy range below
50 MeV. Therefore, an investigation of the impact of cross section data on
the predicted positron-emitter production is also needed.
In this study, we examine the positron-emitter production through
nuclear interactions of a proton beam in a full range of proton energy
(0.1–250 MeV) with a dedicated simulation code SRNA-BNL,13 and assess
the quality of the PET signal in the region near the Bragg peak that is criti-
cal to the success of PET imaging in verifying the location and dosimetry
of the proton beam. In addition, the results are calculated with four sets of
nuclear reaction cross section data to demonstrate dependence of the result
on the reliability of nuclear reaction cross section data.
2. POSITRON EMITTER PRODUCTION

Although many isotopes are produced through different nuclear
interactions during proton therapy, only six major channels produce the
positron emitters 11C, 13N, and 15O in human tissue (Table 5.1).
Table 5.1 Major channels of nuclear interactions leading to the production of posi-
tron-emitters in human tissue during proton therapy
Nuclear reactions Threshold energy Half-life (min) Positron max.
(MeV) energy (MeV)
16O(p, pn)15O 16.79 2.037 1.72
16O(p, 2p2n)13Na 5.66c 9.965 1.19
14N(p, pn)13N 11.44 9.965 1.19
12C(p, pn)11C 20.61 20.39 0.96
14N(p, 2p2n)11Ca 3.22c 20.39 0.96
16O(p, 3p3n)11Cb 27.50c 20.39 0.96
a (p, 2p2n) includes (p, α).
b (p, 3p3n) includes (p, α pn).
c The thresholds refer to (p, α) and (p, α pn).
Table 5.2 Low-probability reactions that produce positron-emitting nuclei

Nuclear reactions Threshold energy Half-life (min) Positron max.
(MeV) energy (MeV)
12C(p, p2n)10C 34.5 0.32 1.87
12C(p, γ)13N 0 9.97 1.19
13C(p, p2n)11C 25.5 20.3 0.96
13C(p, n)13N 3.2 9.97 1.19
14N(p, nα)10C 17.2 0.32 1.87
14N(p, γ)15O 0 2.04 1.72
14N(p, n)14O 6.6 1.18 1.81
15N(p, nα)11C 14.7 20.3 0.96
15N(p, nd)13N 20.4 9.97 1.19
15N(p, t)13N 13.8 9.97 1.19
15N(p, n)15O 3.8 2.04 1.72
16O(p, γ)17F 0 1.07 1.74
16O(p, 3p4n)10C 39.1 0.32 1.87
16O(p, p2n)14O 30.7 1.18 1.81
18O(p, n)18F 2.6 109.8 0.64
Table 5.2 summarizes the other 15 more exotic nuclear reactions that
generate positron emitters. Since 13C, 15N, and 18O have very low abun-
dances in the human body, the reactions induced by protons with them
are negligible. The cross sections of the radioactive capture reactions, (p, γ),
typically are in the order of microbarns; they are three orders-of-magnitude
smaller than the six main channels listed in Table 5.1. Also, due to the very
small quantities of the 10C and 14O isotopes generated, the uncertainties
in their cross sectional data render calculations meaningless. Therefore, we
excluded from our study the reactions listed in Table 5.2.
3. NUCLEAR REACTION CROSS SECTIONS

Three factors govern the expected number of positron-emitters:
nuclear reaction cross sections; the number of incoming particles limited
by target dose; and the number of target particles. If the flux of the incom-
ing proton beam and the target’s particle-density are fixed, the production
of positron emitters is determined by the cross section data used in the
simulations. Therefore, the reliability of such data is essential to predict-
ing positron-emitter production. However, the available data are very
limited because there are very few experimental measurements of these
reactions, especially in the energy range below 50 MeV. To demonstrate the

differences in the results, in this study, we used four sets of nuclear reaction
cross section data for comparison:
1. Data extracted from the emission spectra of recoils in the ENDF
electronic file of the ICRU Report 6323 (2000) used by Beebe-Wang
et al.13
2. Data from “Experimental Nuclear Reaction Data File (EXFOR)”24
maintained by the National Nuclear Data Center (BNL) used by Parodi
et al.8
3. Data from TERA 95/19 TRA1525 (1995) used by Del Guerra et al.26
4. Data from eight different resources published during 1962-1996, and
collected by Litzenberg for his Ph.D. dissertation.18
The cross section data on the six main production channels amongst
these four sources are compared in Figures 5.1–5.6. In the higher energy
range, where some data were lacking, extrapolation was used to extend the
data to 250 MeV.
120
ICRU 63
Parodi
C(p,pn) 11 C
100 Del Guerra

Lizenberg
80
12
Cross section [mb] of
60
40
20
0
0 50 100 150 200 250
Kinetic Energy of Proton [MeV]
Figure 5.1 Nuclear reaction cross sections of 12C(p, pn)11C. The data from four different
resources are compared.
100
ICRU 63
Parodi
Del Guerra
80
Cross section [mb] of 16 O(p,pn) 15 O
Lizenberg
60
40
20
0
0 50 100 150 200 250
Figure 5.2 Nuclear reaction cross sections of 16O(p, pn)15O. The data from four differ-
ent resources are compared.
50
ICRU 63
Del Guerra
Lizenberg
N
40
13
O(p,2p2n)
30
16
20
10
0
0 50 100 150 200 250
Figure 5.3 Nuclear reaction cross section of 16O(p, 2p2n)13N. The data from three dif-
ferent resources are compared.
30
ICRU 63
Del Guerra
25 Lizenberg
O(p,3p3n) C
11
20
16
15
10
0
0 50 100 150 200 250
Figure 5.4 Nuclear reaction cross section of 16O(p, 3p3n)11C. The data from three
different resources are compared.
50
ICRU 63
Del Guerra
Lizenberg
40
N
13
N(p,pn)
30
14
20
10
0
0 50 100 150 200 250
Figure 5.5 Nuclear reaction cross section of 14N(p, pn)13N. The data from three differ-
ent resources are compared.
250
ICRU 63
Del Guerra
Lizenberg
C
200
11
N(p,2p2n)
150
14
100
50
0
0 50 100 150 200 250
Figure 5.6 Nuclear reaction cross section of 14N(p, 2p2n)11C. The data from three dif-
ferent resources are compared.
600
p + 16 O
500
p + 14 N
Total nonelastic Cross Section [mb]
p + 12 C
400
300
200
100
0
0 50 100 150 200 250
Figure 5.7 Total non-elastic nuclear reaction cross sections of protons incident on 16O,
14N, and 12C as functions of kinetic energy.23
The number of particles, protons, and produced particles through

nuclear interactions deposited at the end of the proton beam’s track is
governed by the initial flux of the proton beam and the probabilities
of proton interactions with nuclei. The total non-elastic nuclear cross
sections of protons incident on 16O, 14N, and 12C are important to this
study, as well as to treatment planning, since the accuracy of predict-
ing the dose in the target volume (tumor) is limited by their exactness.
Figure 5.7 shows the total non-elastic cross section of protons incident
on 16O, 14O, and 12C as functions of the proton’s kinetic energy used in
our simulation.23
4. MONTE CARLO SIMULATIONS

We used the SRNA-BNL software package originally developed
as SRNA-2KG by Ilić et al.;27–29 we modified the simulation code for
this work to include the production of positron-emitter nuclei. SRNA-
2KG is a Monte Carlo code employed for assessing proton transport,
radiotherapy, and dosimetry. It transports protons, within an energy
range of 100 keV–250 MeV, with pre-specified spectra in a 3D geom-
etry through material zones confined by planes and second-order sur-
faces. SRNA-2KG can treat proton transport in 279 different materials,
including elements from Z = 1 to Z = 98, and in 181 compounds and
mixtures.
The simulation of proton transport is based on the multiple-scattering
theory of charged particles, and on a model for compound-nucleus decay
after proton absorption in non-elastic nuclear interactions. For each part
of the range, the code calculates an average loss of energy30 with a fluc-
tuation from Vavilov’s distribution and with Schulek’s correction, and
then samples the deflection angle of the protons from Moliere’s distribu-
tion.27–29 Benchmarking SRNA-2KG against the well-known programs
GEANT,17 and PETRA31 demonstrated very good agreement under the
same conditions.
The positron emitters 11C, 13N, and 15O are created through the decay
processes of compound nuclei that include the emission of protons, deuter-
ons, tritons, alpha particles, or photons. The decay products are sampled using
Poisson’s distribution, with the appropriate average multiplication factors for
each particle. The energy and angle of particle emissions, and the multiplica-
tion factors, are obtained from comparing the direct cross sections available
for reaching the daughter nuclei with that from integrating the differential
cross sections for non-elastic nuclear interactions (see Figs. 5.1–5.6). Sampling
of the emission spectra yields the energy and angle of secondary neutron
emission. The transport of secondary protons follows that of primary protons
of that particular energy. The spatial location and the angle of the neutron
and the photon are recorded, but are not treated further. The emitted deu-
terons, tritons, and alpha particles are assumed to be absorbed where they
were created.
The study was for a typical absorbed dose (2 Gy) in a typical target
volume (5 cm diameter), achieved by modulating the kinetic energies
of five beam pulses so that the depths of the Bragg peaks are located
1 cm apart. The relative beam intensities were 1.0, 0.39, 0.31, 0.24, and
0.22 from the highest to the lowest energy pulse. Proton beams of 2 mm
diameter, with kinetic energies up to 250 MeV and a zero divergence
angle were transported in a human tissue using the SRNA-BNL simula-
tion code. The soft tissue in the simulation had a 0.55 ratio of the aver-
aged atomic number to atomic mass (Z/A), and a density of 1.0 g/cm.3
Its elemental composition was 10.11% hydrogen, 11.11% carbon, 2.60%
nitrogen, and 76.18% oxygen (ICRU 4-component). The estimated num-
ber of protons producing an absorbed dose of 2 Gy at the Bragg peak of
a single pulse was 2 × 107.
The positron emitter spatial distributions were simulated with the cross
section data from ICRU Report 63 (see Ref. 23) shown by the curves with
the circle markers in Figures 5.1–5.6; these data are more recent and more
complete in the low-energy range than those from other sources. To reduce
random noise, the values are obtained from averaging 45 sets of simulation data.
To obtain a realistic PET image from the resulting distribution of
positron emitters, we simulated PET data using the SimSET Monte
Carlo PET scanner simulation package.32 SimSET handles the most
important aspects of the image formation process, including photon
attenuation and scatter, the geometry and photon acceptance of the
tomograph, and binning of the coincidence data. For this study, we
introduced modifications at BNL to accommodate a block detec-
tor layout, and standard 3D sinogram binning. We simulated a clini-
cal whole-body HR+ tomograph (Siemens, Knoxville, TN) with the
proton beam direction aligned with the scanner’s axis. The attenuation
map of a typical human head (ellipse with axes of 15 cm and 18 cm in
the transaxial planes) provided more realistic statistics. We reconstructed
the output sinograms into volumetric images using the standard filtered
backprojection technique.
In order to estimate the mean and variance of the PET signal, we

produced 100 statistically independent PET data sets with SimSET
using the same input positron distribution. Each PET image was pro-
cessed by summing the activity over each transaxial plane, resulting in
a 1D depth profile of the PET image. From all 100 profiles, an aver-
age and standard deviation were calculated for each depth position.
Furthermore in order to reduce statistical fluctuations, each profile was
also fit to a polynomial and the mean and standard deviation vs. depth
similarly calculated.
5. RESULTS
The results of linear production densities of 15O, 13N, and 11C are
presented, respectively, in Figures 5.8–5.10. On the right-hand side of
these figures, we added a vertical scale of the energy absorbed by the tissue
to allow comparisons at depth. These positron emitter distributions were
4
x 10
6 30
Simulation
ICRU 63
5 Parodi 25
Positron Emitter 15 O Production [/cm]
Del Guerra
Lizenberg
Absorbed Energy [MeV/cm]
4 Absorbed Energy 20
3 15
2 10
1 5
0 0
0 5 10 15 20 25 30 35 40
Depth in Soft Tissue [cm ]
Figure 5.8 The estimated production of 15O israpy session. The distributions are
shown from the simulation and the calculations with four different cross section data
resources.
9000 30
Simulation
ICRU 63
N Production [/cm ]
Del Guerra 25
Lizenberg

Absorbed Energy
6000 20
13
15
Positron Emitter
3000 10
0 0
0 5 10 15 20 25 30 35 40
Depth in Soft Tissue [cm]
Figure 5.9 The estimated production of 13N isotopes during a single pulse-proton-
therapy session. The distributions are shown from the simulation and the calculations
with four different cross section data resources.
4
x 10
3 30
Simulation
ICRU 63
C Production [/cm ]
2.5 Parodi 25
Del Guerra
Lizenberg
2 Absorbed Energy 20
1.5 15
11
Positron Emitter
1 10
0.5 5
0 0
0 5 10 15 20 25 30 35 40
Depth in Soft Tissue [cm ]
Figure 5.10 The estimated production of 11C isotopes during a single pulse-proton-
therapy session. The distributions are shown from the simulation and the calculations
with four different cross section data resources.
Figure 5.11 A 1.7 mm thick coronal slice through the activity distribution of the 3D PET
image. The beam entered from the left. The horizontal (axial) dimension is 15.5 cm (full
scanner FOV). Pixel size is 2.4 mm horizontal by 1.7 mm vertical.
used as inputs to obtain the PET images using the SimSET Monte Carlo
tomograph simulation.
Figure 5.11 is a coronal slice from a reconstructed PET image. Despite
only about 14,000 coincidence counts in the entire image, the narrow trans-
axial distribution and lack of background activity afford sufficient contrast to
define the distribution. Figure 5.12 depicts the expected PET image’s activity
20 40
PET Image Activity
Absorbed Energy
15 30

PET Image Activity [arb]
10 20
5 10
0 0
28 30 32 34 36 38 40
Depth in Soft Tissue [cm]
Figure 5.12 Average PET activity (dark gray curves) and its standard deviation (light
gray curves), determined from 100 statistically independent PET images. The “spread-
out Bragg peak” (thick black curve), created by five proton beam pulses (thin black
curves), is superimposed with a right-side vertical scale for depth comparison. The PET
results were also processed with a simple algorithm of polynomial curve fit. The aver-
age value of the processed data (the thin curve through the middle of the error bars) is
coincident with the expected PET image activity (dark gray curve) with a much smaller
standard deviation (error bars) compared to unprocessed data (light gray curves).
signal (dark gray curve) and its standard deviation (light gray curves), deter-
mined by the data extracted from the 100 sets of PET images. The “spread-
out Bragg peak” (thick black curve), created by five proton beam pulses (thin
black curves), is superimposed with a right-side vertical scale in Figure 5.12
for depth comparison. Figure 5.12 reveals that, with the help of a simple
polynomial fit, the fitted data (the thin curve through the middle of the error
bars) retains its fidelity to the raw data (dark gray curve) but has a much
smaller standard deviation (error bars) compared to unprocessed data (light
gray curves). We assessed that the depth at half maximum of the distribution
of the PET-image activity was 6.3 mm from the end of the “spread-out Bragg
peak” in the soft tissue. Based on the projection of the error bars onto the
depth axis at that point, we estimate an error of about 1 mm.
6. DISCUSSION AND CONCLUSIONS

Our study has demonstrated that for a typical absorbed dose (2 Gy)
in a typical target volume (5 cm diameter) during a proton therapy session,
the PET image subsequently acquired has a sufficiently good signal-to-noise
ratio to determine the depth of the Bragg peak to approximately 1 mm accu-
racy. Note that this accuracy reflects the error due only to the PET imaging
process, as the error analysis was carried out with multiple statistical realiza-
tions of a fixed input positron distribution. The accuracy of this method in
a clinical setting is expected to be lower for a variety of practical reasons,
including cross section estimates as discussed below. We also demonstrated,
using our simple algorithm of polynomial curve fit, that simple PET image
analysis software could increase the prediction accuracy on the depth of the
Bragg peak. Future clinical applications require the development of a more
advanced and robust algorithm for fitting the PET data in the presence of
noise. The ultimate goal of such work is comparing the radioisotope dis-
tribution measured by PET with the yield of positron emitters predicted
from the treatment planning code. Matching these two implies that the
treatment conformed to the plan. For treatment involving multiple ports,
including some with opposing angles, the centroid of the target dose can
be also computed with that of the PET image.
We also demonstrated the differences in the predicted distributions of
the positron emitters produced by employing cross section data from dif-
ferent resources. In depths less than 30 cm, the linear density of the isotopes
is almost without structure, reflecting the fact that the cross sections remain
almost unchanged in the energy range above 100 MeV. Within this range, the
yields calculated with cross section data from four different resources agree
fairly well with each other.The exception is the level of 11C isotope produc-
tion obtained with data used in Ref. 8. Its production is only about 30% of
that obtained using other data sources. Apparently, this is mainly because the
cross section data used in Ref. 8 accounts for only one of the three major
channels producing the 11C isotope.26 In the depth range between 30 cm
and 38 cm where the Bragg peak is located, the calculated productions for
all three isotopes are significantly different (Figs. 5.8–5.10). The high yield of
13N and 15O calculated with information from ICRU 63 report23 is credited
to the cross section data on reactions 16O(p, 2p2n)13N and 16O(p, pn)15O in
the low-energy range that became available only during year 2000.
As we mentioned in Section 3 the total non-elastic nuclear-reaction
cross sections for protons incident on 16O, 14N, and 12C are also important
to this study and to the treatment planning, since their accuracy limits the
precision of dose prediction in the target volume (tumor).
This investigation points to the on-going need to develop a library of
accurate cross section data for proton- and neutron-induced reactions on the
elements in human tissue. A reliable simulation or calculation depends upon
the accuracy of such data, especially for investigations in the region near
the Bragg peak where accurate cross section information in the low energy
range (below 50 MeV) is needed. This is critical to the success of PET imag-
ing for verifying the location of the Bragg peak of the proton beam, and for
dosimetry.
ACKNOWLEDGMENTS
The authors would like to thank Dr. R.D. Ilić and the Imaging Research
Lab at the University of Washington for providing the SRNA-2KG and
SimSET software packages.
REFERENCES
1. Amaldi, U.; Kraft, G. Radiotherapy with beams of carbon ions. Rep. Prog. Phys. 2005,
68, 1861.
2. Belkić, Dž. Reviews of theories on ionization in fast ion-atom collision with prospects
3. Enghardt, W.; Parodi, K.; Crespo, P., et al. Dose quantification from in-beam positron
emission tomography. Radiother. Oncol. 2004, 73, S96.
5. Paganatti, H. Range uncertainties in proton therapy and the role of Monte Carlo simu-
6. Paans, A. M. J.; Schippers, J. M. Proton Therapy in combination with PET as monitor:
a feasibility study. IEEE Trans. Nucl. Sci. 1993, 40, 1041.
7. Oelfke, U.; Lam, G. K.; Atkins, M. S. Proton dose monitoring with PET: quantitative
studies in Lucite. Phys. Med. Biol. 1996, 41, 177.
8. Parodi, K.; Enghardt, W.; Haberer, T. In-beam PET measurements of b+ radioactivity

induced by proton beams. Phys. Med. Biol. 2002, 47, 21.
9. Knopf, A.; Parodi, K.; Paganetti, H., et al. Quantitative assessment of the physical
potential of proton beam range verification with PET/CT. Phys. Med. Biol. 2008,
53, 4137.
10. Knopf, A.; Parodi, K.; Bortfeld, T., et al. Systematic analysis of biological and physical
limitations of proton beam range verification with offline PET/CT scans. Phys. Med.
Biol. 2009, 54, 4477.
11. Ferrero, M. I.; Ragona, R.; Rolando,V., et al. Monte Carlo simulation of a Protontherapy
system for the calculation of the dose distribution in a patient, TERA 94/5 TRA 14,
1995.
12. Parodi, K.; Enghardt, W. Potential application of PET in quality assurance of proton
therapy. Phys. Med. Biol. 2000, 45, N151.
13. Beebe-Wang, J.; Dilmanian, F. A.; Peggs, S., et al. PET Imaging of dose distribution in
proton-beam cancer therapy. Nucl. Tech. Radiat. Protect. 2005, XX, 23.
14. Pshenichnov, I.; Mishustin, I.; Greiner, W. Distributions of positron-emitting nuclei in
proton and carbon-ion therapy studied with GEANT4. Phys. Med. Biol. 2006, 51, 6099.
15. Fasso, A.; Ferrari, A.; Ranft, J., et al. FLUKA: a Multi-particle Transport Code, CERN-
2005-10, INFN/TC_5/11, SLAC-R-773, 2005.
16. Agostinelli, S.; Allison, J.; Amako, K., et al. GEANT4—a simulation toolkit. Nucl.

Instrum. Methods Phys. Res. A 2003, 506, 250.
17. Gudowska, I.; Sobolevsky, N.; Andreo, P., et al. Ion beam transport in tissue-like media
using the Monte Carlo code SHIELD-HIT. Phys. Med. Biol. 2004, 49, 1933.
18. Litzenberg, D. W. On-line monitoring and PET imaging of the positron-emitting

activity created in tissue by proton radiotherapy beams. Ph.D. Thesis, University of
Michigan, 1997.
19. Sommerer, F.; Parodi, K.; Ferrari, A., et al. Investigating the accuracy of the FLUKA code
for transport of therapeutic ion beams in matter. Phys. Med. Biol. 2006, 51, 4385.
20. Parodi, K.; Ferrari, A.; Sommerer, F., et al. A MC tool for CT-based calculations of dose
delivery and activation in proton therapy. J. Phys.: Conf. Ser. 2007, 74, 021013.
21. Parodi, K.; Ferrari, A.; Sommerer, F., et al. Clinical CT-based calculations of dose and
positron emitter distributions in proton therapy using the FLUKA Monte Carlo code.
Phys. Med. Biol. 2007, 52, 3369.
22. Paganetti, H.; Jiang, H.; Parodi, K., et al. Clinical implementation of full Monte Carlo
dose calculation in proton beam therapy. Phys. Med. Biol. 2008, 53, 4825.
23. ICRU 63, Nuclear data for neutron and proton radiotherapy and for radiation, ICRU
Report 63, 2000.
24. Experimental Nuclear Reaction Data File, Energy Sciences and Technology Department,
BNL. <http://www.nndc.bnl.gov/nndc/exfor/>.
25. Mukhopadhyay, D.; Del Guerra, A.; Di Domenico, O. For PET- monitored hadron
therapy, TERA 95/19 TRA15, 1995.
26. Del Guerra, A.; Di Domenico, O. Positron Emission Tomography as an aid to in “vivo”
dosimetry for proton radiotherapy: a Monte Carlo simulation, TERA 93/10 TRA 9,
1993.
27. Ilić, R. D. SRNA-2KG—Proton Transport Simulation by Monte Carlo Techniques. Institute
of Nuclear Sciences Vinca, Phys. Lab. (010), Beograd, Yugoslavia, 2002.
28. Ilić, R. D.; Spasić-Jokić,V.; Belićev, P., et al. The Monte Carlo SRNA code as the engine
in ISTAR proton dose planning software for the TESLA accelerator installation. Nucl.
Tech. Radiat. Protect. 2004, XIX, 30.
29. Ilić, R. D.; Spasić-Jokić,V.; Belićev, P., et al. The Monte Carlo SRNA-VOX code for 3D
proton dose distribution in voxelized geometry using CT data. Phys. Med. Biol. 2005,
50, 1011.
3 0. Stopping power and ranges for protons and alpha particles, ICRU Report 49, 1993.
31. Medin, J. Studies of clinical proton dosimetry using Monte Carlo simulation and

experimental techniques. Ph.D. Thesis, Stockholm University, 1997.
32. Harrison, R. L.;Vannoy, S. D.; Kaplan, M. S., et al. Slat collimation and cylindrical detec-
tors for PET simulations using simSET. In IEEE Nuclear Science Symposium Conference.
Record 3, 2005, 20/89.
CHAPTER SIX
Inelastic Collisions of Energetic

Protons in Biological Media
⁎
Isabel Abrila, Rafael Garcia-Molinab , Pablo de Veraa,
Ioanna Kyriakouc and Dimitris Emfietzoglouc
aDepartament de Física Aplicada, Universitat d’Alacant, E-03080 Alacant, Spain
bDepartamento de Física, Centro de Investigación en Óptica y Nanofísica, Universidad de Murcia, E-30100
Murcia, Spain
cMedical Physics Laboratory, University of Ioannina Medical School, GR-45110 Ioannina, Greece
Contents
1. Introduction 130
2. Dielectric formalism for inelastic scattering 132
2.1 Projectile description: electronic charge density 134
2.2 Target description: Electronic excitation spectrum 135
2.2.1 Optical energy-loss function of biological materials 138
2.2.2 Bethe surface of biological materials 141
2.3 Mean excitation energy 142
3. Charge-exchange processes 145
4. Inelastic energy-loss magnitudes 146
5. Simulation of the depth–dose distributions 153
6. Conclusions 159
Acknowledgments 161
References 161
Abstract
We study the energy deposited by swift proton beams on materials of biological inter-
est, such as liquid water, DNA, and PMMA. An appropriate description of the target
energy-loss function, which provides its electronic excitation spectrum, is obtained
from available experimental optical data properly extended to non-vanishing momen-
tum transfers. The main magnitudes characterizing the energy-loss distribution of the
beam particles in the target are calculated analytically (in the dielectric framework)
and compared with available experimental data. The depth–dose distribution of the
energy delivered by the proton beam on the biological targets is simulated by the
SEICS (Simulation of Energetic Ions and Clusters through Solids) code, which includes
the main interaction phenomena between the projectile and the target constituents
by means of Molecular Dynamics and Monte Carlo techniques. Also, the proton-beam
energy distribution at several depths in the target for liquid water and PMMA are simu-
lated, and finally, the properties of PMMA as a water-equivalent material are discussed.
⁎
E-mail address: [email protected]

130 Isabel Abril et al.
1. INTRODUCTION
The destruction of malignant tumor cells by using energetic proton
beams was suggested in 1946 by Wilson.1 In less than a decade after this
pioneering proposal, animal studies were performed,2 and later ion therapy
was applied to human patients.3 Presently, there are over 30 ion therapy
facilities worldwide, most of which use proton beams either exclusively
or in addition to other light ions, and the number is expected to double
within the next 5–10 years.4 The energy delivered by swift ions as a func-
tion of the target depth has a nearly flat shape at the entrance of the target
and gradually grows up to having a notorious peak almost at the end of
their trajectories; this is the so-called Bragg curve. The position and inten-
sity of the Bragg peak depend on the projectile energy and type, as well as
on the target nature.The success of using ion, instead of photon or electron,
beams for cancer treatment lies in the high energy they can deposit in a
delimited region of the tumor volume with a sparing effect to the sur-
rounding healthy tissue. Another important advantage of ion therapy is that
the beam particles have larger biological effectiveness than other ionizing
radiations,5 and lower lateral scattering.
The primary beam energies used in ion therapy are typically hundreds
of MeV/u for deep-seated tumors; see Ref. 6 for a recent state-of-the-art-
review. The energy range below a few MeV/u (for light ions) becomes
relevant around the Bragg peak, where the particles exhibit their highest
RBE.7 At the energy range of interest to ion therapy, the dominant slowing
down mechanism of the swift projectiles is the electronic energy loss due
to ionization and excitation of target atoms.
For a long time the standard input data to Monte Carlo simulation
for energy losses due to ionization came from the Bethe–Bloch stopping
power formula,8 which provides energy-loss values for light ions over a
wide variety of materials.9, 10 However, this formula is only sufficiently
accurate at high energies. The usefulness of the Bethe–Bloch formula
within the Monte Carlo users’ community is due to the lack of a general
awareness on the existence of better theoretical predictions derived from
purely quantum-mechanical methodologies, such as the continuum dis-
torted wave (CDW) method11 or its simplified version, the continuum
distorted wave-eikonal initial state (CDW-EIS) approximation.12 The
interested reader is referred to Ref. 6 for a detailed discussion of the com-
putational advances of these methodologies to obtain cross sections for
energetic ion collisions aimed at hadrontherapy applications.
Inelastic Collisions of Energetic Protons in Biological Media 131
A fundamental problem in extending the Bethe–Bloch theory down

to the Bragg peak region is that its two main assumptions, namely, the
Born and dipole (or optical) approximations, turn gradually invalid.13
Specifically, as the projectile velocity decreases, the minimum momen-
tum transfer increases and non-dipole collisions practically dominate over
dipole collisions, rendering the dipole approximation invalid. The effect is
more pronounced for inner shells associated with large binding energies,
the contribution of which to the energy-loss process gradually vanishes at
low projectile energy. The so-called shell-corrections to Bethe’s theory are
meant to account (to first order) for the above effect,14 but their formal
evaluation proceeds through the, generally unknown, generalized-oscillator
strength (GOS) or, for condensed targets, the dielectric response func-
tion.15 A common approach is to employ hydrogen-like GOSs for the
inner shells and scaling laws for the outer shells ignoring any aggregation
and phase effects.14, 16
An alternative approach, which becomes particularly suitable for con-
densed-phase targets, is to use a dynamic and non-local dielectric response
function.17 This is the so-called dielectric approach, having the advantage
that shell-corrections are automatically included (as a matter of fact, they
are build into the model to all orders); in this manner, the electronic
energy loss in the Born approximation can be calculated without resorting
to Bethe’s dipole approximation. The Lindhard dielectric function of the
homogeneous electron gas within the random-phase-approximation (RPA)
is most represented in the literature due to its analytic properties which
extend over the whole energy–momentum plane.18
An analytic dielectric function from first-principles (similar to the one
derived by Lindhard) is not available for realistic materials and alterna-
tive approaches must be sought for. The extended-optical-data models
(or, simply, optical-data models) currently represent the state of the art for
constructing an analytic description of the dielectric response function of
condensed phase targets regardless of their nature (insulators, semiconduc-
tors, or conductors). Powell19 was perhaps the first to suggest that experi-
mental optical data can be used in inelastic scattering calculations within
the dielectric framework. Soon after, Ritchie and Howie20 put forward a
semi-empirical scheme whereby experimental optical data were used to
obtain the dependence of the dielectric function on the energy transfer,
with physically motivated extension algorithms providing the dependence
of the dielectric function on momentum transfer. Besides its simplicity, the
main advantage of this scheme was that the use of optical data specific to
the material being studied automatically accounts for electronic-structure

effects in a realistic manner not always possible within the electron gas
models. The first application of the Ritchie–Howie recipe to ion stopping
in solids was made by Ashley21 using a very simple extension scheme based on
the (undamped) plasmon-pole approximation with a quadratic dispersion
relation for the dependence of the plasmon energy on momentum transfer.
Several modifications over the original work of Ritchie and Howie20
have been published (see Ref. 22 for more details). One of the most fruit-
ful is based on the use of the Mermin dielectric function23 instead of the
Drude dielectric function, because the former automatically provides the
extension to nonzero momentum transfers. This scheme will be the basis of
the target description that enters in the dielectric formalism, to be presented
in Section 2. The charge-exchange processes affecting the charge-state of
the projectile will be discussed in Section 3, whereas in Section 4 we obtain
the main inelastic energy-loss magnitudes for targets of biological inter-
est (liquid water, DNA, and PMMA). These magnitudes are used as input
to simulate the depth–dose distribution and the energy delivery of the
proton beam through the target, which is discussed in detail in Section 5.
Finally, the main conclusions are summarized in Section 6.
2. DIELECTRIC FORMALISM FOR INELASTIC

SCATTERING
Based on the plane-wave Born approximation, the dielectric formal-
ism24, 25 gives simple expressions for the differential inelastic scattering cross
section in condensed media, where the charged particle interacts simulta-
neously with a large number of electrons belonging to different atoms or
molecules of the target, which is suitably described by its dielectric func-
tion ǫ(k, ω), where k and ω are, respectively, the transferred momentum
and energy from the projectile to the target electrons. When a swift pro-
jectile (with mass M1, atomic number Z1, kinetic energy T, and charge q)
moves through a target, it induces electronic excitations and ionizations in
the material, losing energy in the process. Thus, the double-differential (in
energy and momentum transfer) inverse inelastic mean free path (IIMFP)
q of the projectile, is given by
d2 �q (T , ω) e 2 M1 [Z1 − ρq (k)]2 1 −1

= 2 Im , (1)
dω dk π T k ǫ(k, ω)
where e is the absolute value of the electron charge and ρq (k) is the Fourier
transform of the projectile electronic charge density. It can clearly be
observed the contribution of two factors: one of them only depends on
the projectile, M1 [Z1 − ρq (k)]2 /T , and the other only depending on the
target properties through the energy-loss function (ELF), Im [−1/ǫ(k, ω)].
The latter accounts in an effective and compact way for the target electronic
spectrum corresponding to excitations or ionizations of individual elec-
trons or even excitations of collective modes in the condensed medium.
The macroscopic inelastic scattering cross section σ is related to the IIMFP
by the relation � = N σ , where N is the atomic (or molecular) density of
the target.
The dielectric formalism avoids the difficulty of multi-electron transi-
tions by considering either individual ionizations or collective excitations,
but the interested reader should be aware of recent advances in treating
two-electron transitions in atomic targets.26
Within the dielectric formalism the relevant magnitudes describing
the projectile energy-loss distribution can be obtained from the double-
differential inelastic cross section (Eq. (1)) by suitable integrations over the
energy ω and momentum k transfers. The single-differential (in energy
transfer) inverse inelastic mean free path is found by integrating Eq. (1) over
the momentum transfer k, which represents the probability per unit path
length Pq (T , ω) that a projectile with charge-state q and energy T produces
in the target an inelastic process of energy ω (i.e., individual and collective
electronic excitations as well as ionizations), irrespective of the momentum
transferred. This probability is
d�q (T , ω) M1 e 2 kmax dk −1

Pq (T , ω) = = [Z1 − ρq (k)]2 Im ;
dω π 2 T kmin k ǫ(k, ω)
(2)
for heavy projectiles, when M1 ≫ me (where me is the electron mass), the

√
integration limits can be approximated by kmax ≈ ∞ and kmin = ω/ 2T /M1 .
Hence, the moments of the electronic energy-loss distribution can be
found by integrating Eq. (2) over the transferred energy ω. In conse-
quence the total inverse inelastic mean free path, will be obtained by
∞
�q (T ) = d(ω)Pq (T , ω). (3)
0
The stopping power (or stopping force) represents the mean electronic
energy lost by the projectile per unit path length, and is given by
∞
Sq (T ) = d(ω)ωPq (T , ω). (4)
0
Analogously, the energy-loss straggling, which represents the mean

square deviation per unit path length of the energy-loss distribution, can
be obtained through
∞
�2q (T ) = d(ω)(ω)2 Pq (T , ω). (5)
0
In order to obtain these stopping magnitudes, Eqs. 3–5, the descrip-

tion of the projectile charge density and the target excitation spectrum by
means of its energy-loss function (ELF) are needed. In particular in this
work we will restrict to the study of energetic protons beams incident in
media of biological, such as liquid water, DNA, or PMMA.
2.1 Projectile description: electronic charge density

The charge density of the projectile is described by the statistical model
proposed by Brandt and Kitagawa (BK),27 where the effective charge of
energetic ions is calculated in the dielectric response approximation. The
projectile, with atomic number Z1 has N bound electrons and moves with
velocity v with respect to the target electrons. The BK model presupposes
that the cloud of bound electrons is screening the projectile nucleus over a
radius ϒ that depends on the projectile velocity. Therefore, target electrons
that approach the projectile with impact parameters larger than ϒ (that
is, distant collisions) perceive the projectile as a point charge with charge
q = (Z1 − N )e, irrespective of its internal structure. However, when the
impact parameter is smaller than ϒ (i.e., close collisions) the target electrons
penetrate the screening cloud of the bound electrons of the projectile, sens-
ing a partially screened potential corresponding to a projectile charge larger
than q. An average over all the impact parameters will give the charge of
the projectile, that determines their energy loss.
The Brandt and Kitagawa model27 characterizes the bound electrons of
the projectile by a generic orbital, where the screening radius ϒ is treated
as a variational parameter. In order to derive analytical results, the electronic
charge density of the projectile is described by the simple expression:
N ϒ −r/ϒ
ρq (r) = e ,
4π ϒ 3 r (6)
where r is the distance to the projectile nucleus. The internal energy of the
projectile can be written as:
E = Een + λEee + Ekin , (7)
where Een is the potential energy from electron–nucleus interaction, Eee
is the potential energy from the electron–electron interaction weighted by
the variational parameter λ, and Ekin is the electron kinetic energy. After
applying the following conditions to the energy:
dE
= 0, (8)
dϒ
dE

= 0, (9)
dN N =Z1
the following value for the screening radius ϒ is obtained:
0. 48N 2/3
ϒ= a0 , (10)
Z1 − λN /4
with λ = 4/7, and a0 is the Bohr radius (a0 = 0. 529 × 10−10 m).
A modification of the parameter ϒ was introduced when there are
N = 1 or 2 bound electrons:28
3
ϒ= a0 . (11)
2[Z1 − 0. 3(N − 1)]
From Eq. (6) we obtain the Fourier transform of the electronic charge
density of the projectile
N
ρq (k) = , (12)
1 + (kϒ)2
with ϒ given by Eq. (10) (or Eq. (11) when the projectile has only 1 or 2
bound electrons).
2.2 Target description: Electronic excitation spectrum

The ELF of the material as a function of the momentum and energy trans-
fer is known as the Bethe surface, and it contains all the information about
the electronic excitations that the target can sustain.
Experimental information of the ELF of a medium can be obtained from

the analysis of the spectra of photon or electron beams interacting with the
material. For many materials there are available optical data of the ELF (i.e.,
at zero momentum transfer, k = 0 ) over the valence excitation range up to
∼100 eV. To obtain information of the ELF at k � = 0, electron-energy-loss
spectroscopy (EELS) is typically used, where the spectra of the electrons
at different scattering angles correspond to different momentum transfers.
However, due to multiple scattering effects it is almost impracticable to
obtain information of the ELF at high momentum transfers. Therefore
to avoid this experimental difficulty it is necessary to have theoretical
models that provide the Bethe surface in a broad range of energy and
momentum transfers. Theoretical calculations from first-principles are very
hard for condensed media, then semi-empirical calculations using optical-
data models are generally employed.19, 20 They utilize the ELF at the optical
limit, k = 0, from the experimental data together with a theoretical model
to make the proper extension at k �= 0. These models can be applied to
study the inelastic scattering of swift projectiles in any material (metal,
semiconductor, or insulator), and their major advantage consists in the
incorporation of the (individual and collective) excitations of the target,
taking into account the aggregation and chemical effects inherent to con-
densed matter, since they are based on experimental ELF.29
In this work we use the MELF–GOS model (Mermin-Energy-Loss-
Function–Generalized-Oscillator Strength) to describe the energy-loss spec-
trum of the target.30, 31 Here the target electron excitations are split into two
parts, one related to excitations of the inner-shell electrons, and the other
resulting from excitations of the outer (weakly bound) electrons, namely
−1 −1 −1

Im = Im + Im . (13)
ǫ(k, ω) ǫ(k, ω) inner ǫ(k, ω) outer
The inner-shell electrons keep their atomic character due to their large
binding energies, where the excitation spectrum is insensitive to the target
phase. So a suitable and useful approximation is to describe those electrons
by their generalized oscillator strengths in the hydrogenic approach, as if
they were bound to isolated atoms. The connection between the ELF and
the GOS model for a compound Aα1 Bα2 . . . is given by15
(j)
−1 2π 2 N dfnl (k, ω)

(j)
Im = αj � ω − ωioniz,nl , (14)
ǫ(k, ω) inner ω j
dω
nl
where N is the molecular density of the target, αj is the stoichiometric

(j)
contribution of the different elements in the compound, dfnl (k, ω)/dω
(j)
and ωioniz,nl are, respectively, the GOS and the ionization energy of the (n, l)
sub-shell of the j-element of the target, finally, is the step function. The
summation in Eq. (14) must be extended over all the inner shells of all the
atoms in the compound.
The collective and single-particle excitations of the outer electrons of
the target are described by fitting the experimental optical ELF (k = 0) to
a linear combination of Mermin-type ELFs,23
−1 −1

Im = Im
ǫ(k = 0, ω) outer ǫ(k ≈ 0, ω) exp (15)
−1
Ai
= 2
Im �(ω − ωth,i ),
i
ω i ǫM (ωi , γi ; k = 0, ω)
where ǫM is the Mermin dielectric function,23 which can be expressed in

terms of the Lindhard dielectric function ǫL of complex energy as
(1 + iγ /ω)[ǫL (k, ω + iγ ) − 1]
ǫM (k, ω) = 1 + . (16)
1 + (iγ /ω)[ǫL (k, ω + iγ ) − 1]/[ǫL (k, 0) − 1]
The Mermin dielectric function goes beyond the Lindhard dielectric
function, since it includes plasmons with a finite lifetime (i.e., damped) at
any momentum transfer, as it happens in real materials. In the above expres-
sion γ is a phenomenological damping constant that represents phonon-
assisted electronic transitions via electron–ion collisions. At the limit of zero
damping, the Mermin dielectric function recovers the Lindhard dielectric
function.
The parameters ωi, γi, and Ai in Eq. (15) are related, respectively, to the
position, width, and intensity of the ith peak in the target experimental-
ELF, which are determined by a fitting to experimental optical data; ωth,i
is a threshold energy. At the optical limit (k = 0), the Mermin-ELF is iden-
tical to the Drude-ELF, therefore the following expression can be used to
obtain the fitting parameters ωi, γi, and Ai:
−1 ωi2 ωγi

Im = 2 . (17)
ǫM (ωi , γi ; k = 0, ω) (ωi − ω2 )2 + (ωγi )2
One of the main benefits of the MELF–GOS method is that once the
fitting of the ELF in the optical limit is made, the ELF is analytically and
automatically extended to k �= 0 through the properties of the Mermin

dielectric function and the GOS model.31, 32 So, dispersion relations
schemes, as in other optical-data models, are not necessary (see Refs. 22, 33
for a detailed discussion of different dispersion relation schemes).Thus, from
the experimental optical-ELF the MELF–GOS model provides the whole
Bethe surface of the target.
The ELF constructed in this manner must be consistent with physically
motivated constrains. Then, the ELF must satisfy the f-sum rule at any k
(Ref. 22)
ω
−1

me ′ ′
Neff (ω) = dω ω Im , (18)
2π 2 e 2 N 0 ǫ(k, ω′ )
which gives the effective number of target electrons per molecule that par-
ticipate in inelastic collisions with energy transfer up to a maximum value
ω. When the transferred energy ω goes to infinity, then Neff (ω → ∞)
tends to the target atomic number. As the f-sum rule of Eq. (18) is more
sensible to high-energy transfers (corresponding to inner-shell electrons),
the ELF must also fulfill further sum rules, such as the Kramers–Kronig
(KK) (or perfect screening) sum-rule, which is more sensitive to the low-
part of the excitation spectrum. The KK sum rule is given by22
2 ω ′1 −1

KK(ω) = dω ′ Im + n−2 (0), (19)
π 0 ω ǫ(k = 0, ω′ )
where n(0) is the refractive index at low-energy transfer. When the energy
transfer goes to infinity the KK-sum rule tends to unity.
2.2.1 Optical energy-loss function of biological materials

As we did mention previously, the key parameter to obtain reliable
results for the inelastic energy-loss magnitudes is the ELF of the mate-
rial, Im [−1/ǫ(k, ω)], for the whole k–ω plane. Fortunately, experimental
information about the ELF in the optical limit, k = 0, exists for important
biological materials, such as liquid water, DNA, or PMMA. We show in
Table 6.1 some characteristic data of these materials, such as their chemical
composition, atomic number, molecular mass, and mass density.
In Figure 6.1 we show by symbols the available experimental optical-
ELF for liquid water (circles),36 dry DNA (squares)37, and PMMA (tri-
angles),38 as a function of the energy transfer ω corresponding to the
excitation of outer-shell electrons. All these optical-ELFs look rather
similar, since they show only a single peak at ∼21 eV, but there are sizeable
Table 6.1 Properties of the biological targets studied in this work: liquid water, DNA,
and PMMA. We show their chemical composition, atomic number (Z2 ), molecular mass
(M2 ), and mass density.
Target Liquid water DNA34 PMMA35
Chemical formula H2 O (C20 H27 N7 O13 P2 )n (C5 H8 O2 )n
Z2 10 330 54
M2 18.0152 635.42 100.117
Density (g/cm3 ) 1.000 1.350 1.188
differences in the values of their ELF magnitude. The curves in Figure 6.1
represent the results of fitting the experimental data by the MELF–GOS
model using a linear combination of Mermin-type ELFs.
At high energy excitations usually there are not measurements of the
ELF, in particular for the biological materials here analyzed. So, in this
energy range, the optical-ELF of a compound target Aα1 Bα2 . . . will be
Figure 6.1 Energy-loss function, or ELF, for liquid water, DNA, and PMMA in the optical
limit (k = 0), as a function of the energy transfer, ω. Circles represent experimental
data of liquid water,36 squares correspond to DNA,37 and triangles are data for PMMA.38
The curves are the result of fitting those data by using the MELF–GOS model.
derived from the optical-ELFs of its elemental constituents applying the

weighted additivity rule:31, 39
−1 −1
αj
Im =N Im ,
ǫ(k = 0, ω) j
nj ǫ(k = 0, ω) j (20)
where nj and Im [−1/ǫ(k = 0, ω)]j are the atomic density and the ELF,
respectively, of the jth element of the target. The ELF of each elemental
constituent of the target can be obtained from the X-ray measurements
of the atomic scattering factors, which are available up to very high
energies.40
In Figure 6.2 we depict the optical-ELF of liquid water, dry DNA, and
PMMA at high-energy transfer as obtained with the MELF–GOS model
(solid lines). For comparison purposes we also show by dotted lines the
corresponding ELF obtained from experimental X-ray scattering factors40
for the inner-shell electrons of the elements in each compound target,
according to Eq. (20). In the case of liquid water we have added the results
(triangles) from the FFAST database of NIST for the water molecule.41
Notice that the optical ELF in Figure 6.2 contains the contribution from
the outer (Eq. (15)) as well as from the inner-shell electron excitations
(Eq. (14)). Sharp edges can be observed in the figure, corresponding to the
threshold energy of each inner shell. In our model we treated as inner shell
Figure 6.2 Energy-loss function, or ELF, in the optical limit (k = 0) of liquid water, dry
DNA, and PMMA at high-energy transfer. Solid lines correspond to the ELF resulting
from the MELF–GOS model. Dotted lines represent the results derived from X-ray atom-
ic scattering factors.40 Triangles correspond to calculations from the FFAST database of
NIST for the water molecule.41
(using hydrogenic GOS) the K-shell of O for liquid water, the K-shell of
C, N, O, and P for dry DNA, and the K-shell of C and O for PMMA.
In summary, the optical-ELF is constructed in such a wide region of
energy transfer by using the optical-ELF experimental data (in the low-
energy region) and the GOS for the elemental atoms forming the target
(in the high-energy region). These two regions are connected by a reason-
able matching by imposing the fulfillment of the sum rules, Eqs. (18) and
(19), which are satisfied to better than 99%. Proceeding in this manner, we
obtain analytical expressions for the optical-ELF that properly reproduce
the energy-loss spectra of the different targets to be discussed in this work.
2.2.2 Bethe surface of biological materials

Once we build, from the MELF–GOS methodology, the optical ELF
(k = 0) as a function of the energy transfer, the analytical properties of the
Mermin-ELF and the GOS will extend automatically the ELF to finite
values of the momentum transfer, k �= 0, without ad hoc dispersion rela-
tions,22 which allow to obtain the Bethe surface, i.e., the target electronic
excitations in the whole k–ω plane. It is worth to emphasize that chemi-
cal bonding, and aggregation effects, as well as the characteristic collective
excitations of condensed phase are included in the MELF–GOS descrip-
tion of the target excitation spectrum.
In order to check the reliability of the MELF–GOS model at finite
momentum transfers it is desirable to compare the calculated Bethe surface
with existing experimental data,32, 42 although experimental information
of the ELF at k �= 0 is scarce. Luckily for liquid water there are experi-
mental data of ELF at several finite momentum transfers (k ≈ 0–3.6 a.u.)
obtained from inelastic X-ray scattering spectroscopy measurements
(IXSS).43, 44 The experimental ELF of liquid water at k = 0.85, 1.34, and
2.79 a.u... are compared with the results obtained from the MELF–GOS
model (solid lines) in Figure 6.3. We observe a good agreement between
the experimental ELF of liquid water at k �= 0 and our model, which
reproduces the experimental broadening and shifting toward high-energy
transfers as the momentum transfer increases, and it is consistent with the
theoretical expectation that single-particle excitations should gradually
prevail over collective excitations at large momentum transfer. A detailed
discussion of different optical-data models applied to liquid water can be
found in Ref. 22.
The calculated Bethe surface from the MELF–GOS model are pre-
sented for (a) liquid water, (b) dry DNA, and (c) PMMA in Figure 6.4.
Figure 6.3 Energy-loss function, or ELF, of liquid water at the finite momentum trans-
fers k = 0.85, 1.34, and 2.79 a.u. Dotted lines are experimental data43, 44 and solid lines
represent the results of the MELF–GOS model.
As we did mention before the Bethe surface is required in the dielectric
formalism to evaluate the inelastic scattering cross section as a function of
the particle energy. In all the biological targets analyzed here we observe
that an increase in the momentum transfer k shifts the maximum of the
ELF to higher energies and decreases its amplitude. Thus, the Bethe ridge
is broadened due to chemical and/or phase effects, and its behavior is not
a delta function as it happens for two-body collision where the struck
electron is free and stationary, which implies a quadratic relation between
energy and momentum transfer and a sharply peaked Bethe ridge of zero
width. So, with the present description of the Bethe surface we have con-
sidered the many-body effects that take place in real condensed media,
which cause a smearing on the otherwise delta-like profile of the ELF over
the energy–momentum plane.
2.3 Mean excitation energy

At high projectile energies it is appropriate to use the relativistic Bethe
formula for the stopping power,8 namely:
4πe 4 Z12 Z2 N

2m e v 2
SBethe (v) = ln − (v/c) , (21)
2
v2 I (1 − (v/c)2 )
where υ and c are, respectively, the projectile and the light velocities. Here
I denotes the so-called mean excitation energy of the material, which only
Figure 6.4 Bethe surface obtained from the MELF–GOS model for (a) liquid water, (b)
DNA, and (c) PMMA.
depends on the electronic structure of the target, and is defined in terms of

the target ELF by the relation:
∞ ′ ′
dω ω ln ω′ Im[−1/ǫ(k = 0, ω′ )]
ln I = 0 ∞ ′ ′ . (22)
0dω ω Im[−1/ǫ(k = 0, ω′ )]
Note that in the high-energy limit, the Bethe stopping formula depends
on the material through the mean excitation energy I. Therefore accurate
values of I are desirable, in particular for biological materials, where a few
percent in the uncertainty of the I values might cause sizeable changes in
the range and stopping maximum (i.e., the Bragg peak) of therapeutic ion
beams,45, 46 which have energies of the order of several hundred of MeV/u.
The mean excitation energy I of a biological material can be evaluated
through Eq. (22), with the optical-ELF Im[−1/ǫ(k = 0, ω)] provided by
the MELF–GOS method. We show in Table 6.2 the I values obtained from
the MELF–GOS model corresponding to liquid water, DNA, and PMMA
targets. Our calculations predict that the outer and inner-shell electron
excitations contribute in almost the same proportion to the mean excita-
tion energy.
It is worth to notice that the I value of PMMA (70.3 eV) is consider-
ably smaller than the ones corresponding to liquid water (79.4 eV) or DNA
(81.5 eV), which implies a larger stopping of projectiles in PMMA. In Table
6.2 a comparison with other I-values from the literature is presented; for
a detailed analysis see Ref. 50. For liquid water a I-value 80 eV prevails
in most of sources, while for DNA there is more dispersion between the
I-values obtained from different sources.
Table 6.2 Mean excitation energy I (eV) obtained from the MELF–GOS model for liq-
uid water, DNA, and PMMA. A comparison with other values is also presented.
Liquid water DNA PMMA
MELF–GOS model 79.447 81.534 70.335
Bragg rule (ICRU49)14 75 ± 3 85 74.0 ± 1.5
Tan et al. 72.548 86.6449 68.3748
Paul50 80.8 ± 2.0
Kumazaki et al.51 78.4 ± 1.0
Emfietzoglou et al.52 77.8
LaVerne and Pimblott53 77.9
Akkerman and Akkerman54 68.5
3. CHARGE-EXCHANGE PROCESSES

The charge-state q of a projectile, with atomic number Z1, moving
through a condensed medium can vary due to electron-capture and -loss
processes. Thus at low and intermediate projectile energies, these processes
give a continuous dynamic charge-exchange of the projectile in its path
through the target, therefore affecting its energy loss.
When charge equilibrium is reached, usually a few femtoseconds
after the projectile penetrates into the target, the energy-loss magnitudes
(Eqs. (2)–(5)) can be obtained from a weighted sum of the corresponding
energy-loss magnitudes for each charge-state q of the projectile,31 namely
Z1

P(T , ω) = φq (T )Pq (T , ω), (23)
q=0
Z1

�(T ) = φq (T )�q (T ), (24)
q=0
Z1

S(T ) = φq (T )Sq (T ), (25)
q=0
Z1
2
φq (T )�2q (T ),

� (T ) = (26)
q=0
where φq (T ) is the probability of finding the projectile in a given charge-

state q at the energy T. The sums in Eqs. (23)–(26) extend over all pos-
sible charge-states q of the projectile. We assume the charge-state fractions
at equilibrium, which depend on the target nature, the projectile and its
energy. The values of φq are obtained from a parameterization to experi-
mental data,55 which uses Bragg’s additivity rule for compound targets. In
Figure 6.5 we show the charge-state fractions of a proton beam incident in
water, DNA, and PMMA as a function of the beam energy. No differences
appear in the charge-state fractions for these three biological targets, due to
their similar composition by light elements. It is worth to notice that the
fraction of neutral hydrogen is zero at energies 200 keV.
Figure 6.5 Charge-state fractions of H0 and H+ in water (solid line), DNA (crosses), and
PMMA (circles) as a function of the projectile energy.
4. INELASTIC ENERGY-LOSS MAGNITUDES

The inelastic collisions that a projectile experiences in its travel
through a target can be characterized by several energy-loss magnitudes, as
we discussed previously in Section 2. Now we present results of these mag-
nitudes for proton beams impinging in liquid water, dry DNA, and PMMA.
All the calculations are based in the dielectric formalism, with a realistic
description of the electronic properties of the target by the MELF–GOS
model, and taking into account charge-exchange processes of the projectile.
Figure 6.6 shows the electronic excitation probability P(T , ω), as given
by Eqs. (2) and (23), which represents the probability to produce electronic
excitations of energy ω per unit path length in liquid water (solid line),
dry DNA (dashed line), and PMMA (dotted line) by a proton with incident
energy T = 0.5, 1, 5, and 10 MeV.
We find that independently of the projectile energy T these three bio-
materials present the maximum of the electronic excitations at the same
values of the transferred energy ω, although the absolute value and shape
(a) (b)
(c) (d)
Figure 6.6 Probability per unit path length P(T , ω) that a proton beam induces
electronic excitations of energy ω in liquid water (solid line), dry DNA (dashed line),
and PMMA (dotted line). Results for several projectile energies are displayed: (a)
T = 0. 5 MeV, (b) T = 1 MeV, (c) T = 5 MeV, and (d) T = 10 MeV.
of P(T , ω) is different in each case. As the projectile energy increases the

energy for the maximum of the electronic excitations slightly shifts to
higher incident energies, going from ∼22 eV at T = 0.5 MeV up to ∼33 eV
at T = 10 MeV. Also the intensity of the maximum probability P to pro-
duce electronic excitations decreases as the proton energy increases, indi-
cating that a larger number of electronic excitations are created when the
projectile has lower energies, a circumstance that takes place at the Bragg
peak. It is remarkable to mention that the most probable electronic excita-
tions have energies in the range ∼10 − 50 eV for the analyzed targets. This
result is significant because although the damage in DNA can be produced
by direct ionization of inner shells of the target constituents, recent works
point to electrons with sub-ionizing energies as the ones responsible for
producing lethal damage in DNA.56, 57
Due to the stochastic nature of the inelastic collisions of charged
particles moving through condensed media, their electronic energy-loss
distribution can be evaluated by integrating the different moments of
the electronic excitation probability P(T , ω) over the energy transfer, ω.
In what follows we present our calculations for the main stopping mag-
nitudes: the inverse inelastic mean free path, the stopping power, and the
energy-loss straggling for proton beams traveling through liquid water, dry
DNA, and PMMA.
Figure 6.7 shows the inverse inelastic mean free path, or IIMFP, for
protons in liquid water (solid line), dry DNA (dashed line), and PMMA
(dotted line) calculated from the dielectric formalism and the MELF–GOS
model (Eqs. (3) and (24)). At proton energies lower than ∼1 MeV, sizeable
differences between the IIMFP corresponding to the three materials are
obtained, which can be ascribed to their different energy-loss function.
At the maximum value of the IIMFP, corresponding to a proton energy
∼75 keV for the three biological materials, the IIMFP in DNA is 35%
larger than in liquid water, whereas the IIMFP in PMMA is more than 50%
greater than in liquid water.
Figure 6.7 Inverse inelastic mean free path, or IIMFP, of a proton beam in liquid water
(solid line), in DNA (dashed line), and PMMA (dotted line) calculated with the dielectric
formalism and the MELF–GOS model. Squares correspond to the IIMFP in water from
Uehara et al.59 stars are the calculations in liquid water from Dingfelder et al.60 circles
correspond to the IIMFP in liquid water obtained by Emfietzoglou et al.61 and asterisks
and triangles are, respectively, theoretical calculations in liquid water and DNA from
Tan et al.62
Usually the IIMFP is required as basic input data in Monte Carlo

simulations, to account for electronic interactions and evaluate the damage
on biological tissue by the irradiation with energetic charged particles.
Although in this work we will use the IIMFP, and related magnitudes,
derived from the dielectric formalism, cross sections obtained from first
principles, like the continuum distorted wave-eikonal initial state approxi-
mation, have recently been implemented in Monte Carlo studies of the
interaction of swift ions with liquid water.58
It is important to know that all the biomaterials have not similar IIMFP,
but there are sizeable variations in their value. In Figure 6.7 we also com-
pare our calculated IIMFP with results from other models. For protons in
water we present results from Uehara et al.59 (squares), who obtain high
IIMFP values since in their model they use water vapor cross sections,
calculations from Dingfelder et al.60 (starts) and from Emfietzoglou et al.61
(circles); the latter two sets of results agree rather well with our calculations.
Results from Tan et al.62 give similar IIMFP for liquid water (asterisks) and
DNA (triangles) contrary to our findings; this coincidence of IIMFP for
liquid water and DNA may have its origin in that these calculations62 are
based in the optical data model of Ashley21 and on an evaluation approach
for the optical energy-loss function.
The calculated stopping power S (Eqs. (4) and (25)) of liquid water
(solid line), dry DNA (dashed line), and PMMA (dotted line) for proton
beams is shown as a function of the projectile energy in Figure 6.8. Our
model predicts that these biomaterials present the maximum values of S at
the same projectile energy, around 100 keV. However the stopping power
values of liquid water are always lower than those of dry DNA and PMMA;
at T = 100 keV their differences are ∼30%, but even at high proton
energies there are substantial differences between the stopping power of
liquid water and the other biomaterials. The stopping power of DNA and
PMMA are rather similar at intermediate and high proton energies, which
can be due to the balance between their ELF (larger for PMMA than
for DNA) and their different electron densities (smaller for PMMA than
for DNA: NPMMA e = 0. 385 electron/Å3 , NDNAe = 0. 421 electron/Å3.
However, when comparing the electronic properties of liquid
water and DNA or PMMA, both the ELF and the electron density
e 3
(Nliq. water = 0. 334 electron/Å ) for liquid water are smaller than for DNA
and PMMA, which explains the smaller stopping power of liquid water as
compared to the other biomaterials. We also include in Figure 6.8, experi-
mental data of the stopping power of liquid water63, 64, 65 and ice.66–68
Figure 6.8 Stopping power of liquid water (solid line), dry DNA (dashed line),
and PMMA (dotted line) for a proton beam, as a function of the incident
energy, obtained with the dielectric formalism and the MELF–GOS method.
Experimental data for protons in liquid water63-65 and ice66–68 are depicted by
symbols. The stopping power of DNA calculated by Tan et al.49 is shown by empty
triangles.
We find an excellent agreement of our stopping calculations with the new-

est experimental data in liquid water at high proton energies,65 which does
mean that the mean excitation energy I of liquid water obtained from the
MELF–GOS model, I = 79. 4 eV,47, 69 is in very good agreement with the
features of the recent experimental results.65 Note also that experimental
results for liquid water from Refs. 63, 64, covering the range from 0.3 to
2 MeV, were obtained with a thin liquid jet in vacuum, whose diameter was
treated as a fitting parameter.
The calculated stopping power by the dielectric formalism and by the
MELF–GOS model at lower energies differs from the available experi-
mental data of protons in ice. The close similarity70, 71 between the exci-
tation spectra of liquid water36 and ice72 implies that the corresponding
mean excitation energies are comparable and, therefore, the stopping
power at high projectile energies coincides for both liquid water and ice.
But, nevertheless there are small differences in the band structure of liquid
water and ice, which lead to the observed differences in the stopping power
at low projectile energies.
Finally, we also depict in Figure 6.8 by triangles the stopping power
of DNA calculated by Tan et al.49 which also were evaluated applying
the dielectric formalism to the optical-ELF of DNA, but with a different
extension algorithm to k �= 0 values.
The second moment in the energy-loss distribution corresponds to the
energy-loss straggling 2, which is related to the variance in the distribution
of the stochastic inelastic collision phenomena. We show in Figure 6.9 the
energy-loss straggling 2 of liquid water (solid line), dry DNA (dashed line),
and PMMA (dotted line) for a proton beam as a function of its energy, cal-
culated from Eqs. (5) and (26).The 2 for liquid water is appreciably smaller
than for DNA or PMMA in the whole proton energy range.The arguments
to explain this behavior are the same that we discussed in relation to the
stopping power. It is worth to notice that the contribution of the inner-shell
electrons to the energy-loss straggling is larger than in the stopping power,
representing a considerable percentage at high projectile energies.Therefore,
an accurate description of electron inner-shell excitations is desirable, which
Figure 6.9 Energy-loss straggling, calculated with the dielectric formalism and the
MELF–GOS method, of liquid water (solid line), dry DNA (dashed line), and PMMA
(dotted line) for a proton beam as a function of the incident energy.
are accounted for in this work by the GOS (in the hydrogenic approach) of
the elemental constituent of the compound target.
A proper description of the energy-loss straggling is important because
recent detailed simulations of the spatial distribution of the energy depo-
sition by proton beams in liquid water did show that the position of the
Bragg peak is determined mainly by the stopping power, whereas its width
can be attributed to the energy-loss straggling.69
In order to evaluate the possible damage induced in biological materials
due to the irradiation with energetic particles, another interesting magni-
tude is the average energy Waverage (T ) transferred by a projectile of energy
T to the target producing electronic excitations and ionizations, which is
defined by:
Z1 ∞
q=0 φq (T ) 0 dω(ω)Pq (T , ω)
Waverage (T ) = Z1 ∞ . (27)
q=0 φq (T ) 0 dω Pq (T , ω)
Figure 6.10 shows Waverage (T ) as a function of the energy of a proton

beam in liquid water (solid line), DNA (dashed line), and PMMA (dotted
line). We can observe that Waverage (T ) is rather similar for liquid water and
DNA, however the average energy is smaller for PMMA in all the range
of proton energies analyzed here. This fact is related to the behavior of the
probability to induce electronic excitations P(T , ω) in terms of the trans-
ferred energy ω, which, as has been shown in Figure 6.6, is quite smaller for
PMMA than for liquid water or DNA at high-energy transfers.Waverage (T )
increases with proton energy T, being ∼20 eV at T = 10 keV for liquid
water and DNA (∼10 eV for PMMA) up to ∼70 eV for T = 50 MeV
(∼60 eV for PMMA). We also show by dots in Figure 6.10 the values
reported by Pimblott and LaVerne73 for the average energy of the secondary
electrons generated by swift protons. These results derive from Monte Carlo
simulations of the proton tracks based on experimental optical-ELF for
liquid water, but using the quadratic extension to finite k. We can observe
that the agreement with our calculations is quite good, the small discrepan-
cies being due probably to the different extension algorithm used in each
methodology, and because we also include electronic excitations.
As the probability of electronic excitations, characterized by P(T , ω),
decreases with the projectile energy T (see Fig. 6.6), and the average energy
of the excitations increases with T, hence a maximum in the stopping
power is clearly expected, which is found around T = 100 keV for liquid
water, DNA, and PMMA (see Fig. 6.8).
Figure 6.10 Average energy Waverage transferred by the projectile producing electronic
excitations and ionizations in liquid water (solid line), dry DNA (dashed line), and PMMA
(dotted line) induced by a proton beam as a function of its energy. The calculations
were obtained with the dielectric formalism and the MELF–GOS model. Circles cor-
respond to results for ionization processes from a proton beam in liquid water from
Pimblott and LaVerne.73
An estimation of the energy carried by the secondary electrons gener-

ated by energetic protons is obtained as Waverage − Ebind, where Ebind is a
representative value characterizing the binding energy of the target elec-
trons, which is Ebind ∼ 10 eV for biomaterials.74 Therefore, the generated
secondary electrons produced in liquid water, DNA, or PMMA could be
very effective in producing DNA strand breaks, due to the low threshold
energy for radiation damage in biomolecules,75 with the average energy of
the secondary electrons produced by proton beam being similar in liquid
water and DNA, and a little bit smaller in PMMA.
5. SIMULATION OF THE DEPTH–DOSE DISTRIBUTIONS

The depth–dose profile of energetic protons in biological materials
as well as the energy distribution of the beam as a function of the depth is
simulated by the SEICS code (Simulation of Energetic Ions and Clusters
through Solids). In brief, this code uses Molecular Dynamics to follow

in detail the motion of the charged projectiles through the target, and
a Monte Carlo procedure to treat the statistical nature of the elastic and
inelastic scattering as well as the charge-exchange processes between the
projectile and the target. In what follows we present the basic features of
the simulation code; for more details the reader is referred to Refs. 69, 76.
The trajectory of an energetic particle through the target is obtained by
numerically solving its equation of motion. Given the instantaneous posi-
tion r (t), velocity v(t) acting on the projectile at a given
, and force F(t)
time t, its new position �r (t + �t) and velocity v(t
� + �t) after a time step
t, are obtained by using the velocity variant of Verlet’s algorithm77
�
F(t)
(28)
3/2
(�t)2 1 − (v(t)/c)2

�r (t + �t) = �r (t) + v(t)�t
� + ,
2M1
� + F(t
F(t) � + �t)
, (29)
3/2
(�t) 1 − (v(t)/c)2

� + �t) = v(t)
v(t � +
2M1
where c is the speed of light and the term in brackets is an ad hoc modi-
fication of the original algorithm to account for the relativistic effects on
the motion of the projectile, which must be considered when studying
the interaction of high energetic beams in biomaterials, as in the case of
hadrontherapy.
The force F(t) that slows down an energetic particle travelling with
charge-state q through the target is the electronic stopping force. But due
to the stochastic nature of the inelastic energy-loss processes, the stopping
force is obtained from a randomly sampled Gaussian distribution:
√
F = − Sq + �q / �s −2 ln ξ1 cos (2π ξ2 ) , (30)
where the mean value of the distribution is the stopping power Sq and its

variance σ is related to the energy-loss straggling 2q through σ = �2q /�s,
with �s = v(t)�t being the distance traveled in a time step t, and ξ1 and
ξ2 are two random numbers uniformly distributed between 0 and 1.
We use in the SEICS code the stopping power and the energy-loss strag-
gling for a projectile with charge q, calculated with the dielectric formalism
together with the MELF–GOS model, see Eqs. (4) and (5). Nonetheless,
at high projectile energies (E 10 MeV/u), analytical formulas from the
relativistic Bethe stopping power8 and the relativistic Bohr energy-loss

straggling are used.
Although inelastic scattering is the main responsible of the energy
loss of the projectile, the SEICS code also includes the elastic scattering
between the projectile and the target constituent atoms.These mainly result
in angular deflection of the projectile trajectory and also contributes to the
energy loss at low energies.The code uses a Monte Carlo algorithm to treat
multiple scattering based in Ref. 78.
As the electronic energy loss strongly depends on the charge state of the
projectile, which varies during its travel through the target, then electronic
capture and loss processes by the projectile are included in the SEICS code
in order to know the charge state of the projectile at each time during its
trajectory.
The simulated depth–dose distribution obtained by the SEICS code for a
100 MeV proton beam in liquid water and PMMA is shown in Figure 6.11.
Solid lines are the results when all the interactions are included in the simula-
tion, whereas dotted lines represent results obtained when energy-loss straggling
Figure 6.11 Simulated depth–dose distribution of a 100 MeV proton beam in liquid

water (black curves) and in PMMA (grey curves). Solid lines are the results obtained
from the SEICS code when all the interactions are taken into account, whereas dotted
lines represent results of the same simulation without energy-loss straggling.
was not considered. Nuclear fragmentation processes are not included in the
simulation since they do not affect significantly the depth–dose curves for
protons below 100 MeV. The plateau and peak position of the Bragg curve
are mostly determined by the electronic stopping force, but the shape of the
Bragg peak strongly depends on the fluctuations of the projectile energy loss
along its full path, accounted for by the energy-loss straggling. The Bragg
peak for liquid water (black curves) is deeper than for PMMA (grey curves),
because the PMMA stopping power (see Fig. 6.8) and its density is larger than
for liquid water, so protons in PMMA slow down faster than in liquid water,
reaching lower depths. Also, the differences in their stopping powers translate
in a higher dose in PMMA than in liquid water at all depths.
In Figure 6.12 we show the results provided by the SEICS code for
the average energy E (dashed lines) of 100 MeV protons incident on
liquid water (black dashed curve) and on PMMA (grey dashed curve), as a
function of the depth. The energy distributions of the protons at different
depths in these biomaterials are depicted by solid lines, showing the broad-
ening with depth of the initially monoenergetic beam. For comparison
purposes the depth–dose distributions corresponding to each material are
Figure 6.12 Left axis: Average energy E (dashed lines) of protons incident with
100 MeV in liquid water and in PMMA, as a function of the depth. The beam energy
distribution at each depth is depicted by solid lines. Right axis: Depth–dose distribu-
tion for liquid water and PMMA (dotted lines). Black (or grey) lines correspond to liquid
water (or PMMA).
shown by dotted lines (right axis). The average energy of protons in liquid
water and in PMMA decreases with depth mostly due to inelastic colli-
sions with target electrons, and the initially monoenergetic proton beam
becomes broader as the depth inside the target increases.
It is interesting to notice that at the Bragg peak the projectile energy
distribution is asymmetric since around that depth some projectiles start to
stop, with the consequent increase in the lower energy tail of the energy
distribution. From the previous discussion we emphasize the importance
of taking into account the energy distribution of the protons around the
Bragg peak because proton energy determines the spectrum of the gen-
erated secondary electrons, which are mainly responsible for the DNA
damage.75
The SEICS code is also a useful tool to evaluate the water-equivalent
properties of PMMA. This target is a plastic commonly used as a water-
equivalent material in dosimetry measurements, since it has relatively simi-
lar properties and since it is easier to manipulate a solid target than a liquid
one. The water-equivalence is based in the fact that a measured thickness
in PMMA can be converted to an equivalent thickness in liquid water
through79, 80
ρPMMA S PMMA
zwater = zPMMA cPMMA = zPMMA , (31)
ρwater S water
where zwater and zPMMA are the equivalent depths in water and PMMA,
respectively, and cPMMA is the depth-scaling factor (or water-equivalent
ratio, WER), which depends on the density and mean mass stopping power
ratios, ρPMMA /ρwater and S PMMA /S water, respectively.
The WER is usually calculated as the ratio of the continuous slowing
down approximation (CSDA) ranges of protons in PMMA and water. An
alternative is to use the Bragg curves calculated within the SEICS code,
defining the range as the thickness in which the dose is 80% after the maxi-
mum. We have determined the WER of PMMA in this way for 75 MeV
and 100 MeV proton beams, obtaining in both cases a value of 1.174. This
value can be compared with the ones reported by other authors, which
are summarized in Table 6.3. Note that the WER depends on the density,
according to Eq. (31), and then we have corrected all the values to match
the density ρPMMA = 1. 188 g/cm3 used by us.
In Figure 6.13 we show the Bragg curves for a 100 MeV proton beam
in water and PMMA, normalized to the entrance dose, and with the
thickness in PMMA scaled according to Eq. (31), with WER = 1.174. As

it can be seen, the two curves are totally superimposed, indicating excel-
lent water-equivalent properties of PMMA when nuclear fragmentation
processes are neglected.
Regarding the differences between the WERs shown in Table 6.3, the
maximum difference is observed between our value of 1.174 and the one
by ICRU49 CSDA ranges14 of 1.157 (1.45%). Such a difference translates,
for a 100 MeV proton beam, in an uncertainty of the order of 1.2 mm in
the Bragg peak position. Therefore, it is important to count on the most
reliable values of stopping magnitudes (and also with realistic radiation
transport codes, which allow a precise calculation of WER) if a submil-
limetric accuracy is desired.
Other useful information the SEICS code can evaluate regarding the
water equivalence of PMMA is the energy distribution of projectiles at the
Bragg peak in both targets. Although the macroscopic dose is completely
equivalent, as we have already shown, the knowledge of the equivalence
of the energy distribution is also important from a microscopic point of
Figure 6.13 Normalized (at the entrance) depth–dose curves of a 100 MeV proton beam
in liquid water (black solid line) and in PMMA at water-equivalent depth (grey solid line).
The inset shows the energy distribution of the proton beam just at the Bragg peak in
liquid water (black solid line) and PMMA (grey solid line); see the text for more details.
Table 6.3 Water-equivalent ratios, or WER, for protons in PMMA from different sources.
ρPMMA (g/cm3 ) WER WER (ρ PMMA = 1. 188 g/cm3 )

This work (75 MeV, 1.188 1.174 1.174
100 MeV)
ICRU49 (CSDA)14 1.195 1.164 1.157
(75 MeV)
Palmans et al.79 1.186 1.159 1.161
(75 MeV)
Zhang et al.80 1.185 1.157–1.167 1.160–1.170
(75–250 MeV)
view, since the energy of the particles will affect the emission of secondary
electrons. This issue is relevant if further analysis of radiation damage will
be carried out using PMMA as a water substitute. The inset of Figure 6.13
shows the energy distribution of particles, at the Bragg peak depth, for both
liquid water (black line) and PMMA (grey line), i.e., the number of projec-
tiles, N, which travel, at this depth, with a certain energy, E, divided by the
bin width, E. This division is performed in order to avoid differences in
height due to different widths of the histogram bin. As it can be seen, the
proton beam is also equivalent for water and PMMA from a microscopic
point of view. This point remarks the water equivalence of PMMA when
nuclear fragmentation reactions are neglected, and of course when the
WER is accurately known.
6. CONCLUSIONS
We have presented in this paper an account of our recent work
applied to the interaction of proton beams with materials of biological
interest, such as liquid water (∼70% constituent of living beings), DNA
(the biomaterial per excellence), and PMMA (a plastic widely used in
dosimetry). This research is linked to the recent use of proton beams in
cancer therapy, a promising and successful technique known as ion therapy
(or hadrontherapy).
The main magnitudes useful for describing the energy deposited
by a projectile moving through a biological target with charge-state q
are the stopping power Sq and the energy-loss straggling 2q, which can
conveniently be evaluated in the dielectric formalism. We use a suitable
description of the target excitation spectrum, based in optical experimental

data for the outer electronic excitations and hydrogenic generalized oscil-
lator strengths for the inner-shell electrons of the target, properly extended
to finite momentum transfers conveniently extended to finite momentum
transfers; this procedure constitutes the fundamentals of the so-called
Mermin Energy Loss Function–Generalized Oscillator Strength method,
from which we can easily obtain, besides Sq and 2q, the target mean exci-
tation energy, the projectile inverse inelastic mean free path, the average
energy transferred to electronic excitations and ionizations, or the prob-
ability that the projectile induces electronic excitations of a given energy.
The projectile electronic density is accounted for by the Brandt–Kitagawa
model.
There are sizeable differences among the calculated stopping powers of
liquid water and those of DNA and PMMA, which are rather similar. Our
stopping power of liquid water for protons compares fairly well with the
available experimental data explicitly measured for liquid water, with differ-
ences appearing for ice water stopping power (probably due to the different
phase state of the target).
The average energy transferred by the beam to the targets lies in the ∼10
and ∼70 eV range, showing the same increasing trend as the projectile energy
grows, although being slightly larger for liquid water, DNA, and PMMA (in
this order). Our results are a little bit larger than previously published results73
for liquid water, because we also consider electronic excitations.
By means of the Simulation of Energetic Ions and Clusters through
Solids code, which is based in Molecular Dynamics and Monte Carlo to
follow the trajectories of the projectiles through the target, we are able to
obtain the depth–dose curves (Bragg curves), as well as the energy distri-
bution of the proton beam as a function of the depth in liquid water and
PMMA, both materials being widely studied for the propagation of proton
beams. From these simulations we obtain the water-equivalent thickness of
PMMA as well as other interesting magnitudes in dosimetry, such as the
water-equivalent ratio, which is compared to values available in the litera-
ture, obtained by different methods. Comparing the extreme values (ours
and ICRU4914) of the water-equivalent ratio for a 100 MeV proton beam,
an uncertainty of ∼1.2 mm in the Bragg peak appears.
Finally, we have shown that the energy distributions of a 100 MeV proton
beam at the Bragg peak are practically identical in liquid water and PMMA
(when nuclear fragmentation is not taken into account), which means that
both materials are equivalent with respect to the energy distribution.
ACKNOWLEDGMENTS
This work has been financially supported by the Spanish Ministerio de
Ciencia e Innovación (Project FIS2010-17225) and the European Union
FP7 ANTICARB (HEALTH-F2-2008-201587). PdV thanks the Conselleria
d’Educació, Formació i Ocupació de la Generalitat Valenciana for its support
under the VALi+d program. This research is part of the COST Action MP
1002, Nanoscale Insights into Ion Beam Cancer Therapy.
REFERENCES
1. Wilson, R. Radiological use of fast protons. Radiology 1946, 47, 487.
2. Tobias, C. A.; Anger, O. H.; Lawrence, J. H. Radiological use of high-energy deuterons
and alpha-particles. Am. J. Roentgenol. 1952, 67, 1.
3. Tobias, C. A.; Lawrence, J. H.; Born, J. L.; McCombs, R. K.; Roberts, J. E.; Anger, H. O.;
Low-Beer, B.V. A.; Huggings, C. B. Pituitary irradiation with high-energy proton beams.
A preliminary report. Cancer Res. 1958, 18, 121.
4. Amaldi, U.; Bonomi, R.; Braccini, S.; Crescenti, M.; Degiovanni, A.; Garlasché, M.;
Garonna, A.; Magrin, G.; Mellace, C.; Pearce, P.; Pittà, G.; Puggioni, P.; Rosso, E.; Verdú
Andrés, S.; Wegner, R.; Weiss, M.; Zennaro, R. Accelerators for hadrontherapy: From
Lawrence cyclotrons to linacs. Nucl. Instrum. Methods Phys. Res., Sect. A 2010, 620, 563.
5. Nikjoo, H.; Uehara, S.; Emfietzoglou, D.; Brahme, A. Heavy charged particles in radia-
tion biology and biophysics. New J. Phys. 2008, 10, 075006.
7. Paganetti, H.; Niemerko, A.; Ancukiewicz, M.; Gerweck, L. E.; Goitein, M.; Loeffler, J.
S.; Suit, H. D. Relative biological effectiveness (RBE) values for proton beam therapy.
Int. J. Radiat. Oncol. Biol. Phys. 2002, 53, 407.
8. Inokuti, M. Inelastic collisions of fast charged particles with atoms and molecules – the
9. Sigmund, P. Stopping power in perspective. Nucl. Instrum. Methods Phys. Res., Sect. B
1998, 135, 1.
10. Sabin, J. R.; Oddershede, J. Stopping power-what next? Adv. Quantum Chem. 2005,
49, 299.
11. Belkić, Dž. A quantum theory of ionization in fast collisions between ions and atomic
systems. J. Phys. B: At. Mol. Phys. 1978, 11, 3529.
12. Crothers, D. S. F.; McCann, J. F. Ionization of atoms by ion impact. J. Phys. B: At. Mol.
Phys. 1983, 16, 3229.
13. Sigmund, P. Light-ion stopping near the maximum. Nucl. Instrum. Methods Phys. Res.,
Sect. B 1994, 85, 541.
14. ICRU, Stopping powers and ranges for protons and alpha particles, ICRU Report 49,
Bethesda, MD, 1993.
15. Fano, U. Penetration of protons, alpha particles, and mesons. Annu. Rev. Nucl. Sci. 1963,
13, 1.
16. Bichsel, H. Shell corrections in stopping power. Phys. Rev. A 2002, 65, 052709.
17. Ritchie, R. H. Energy losses by swift charged particles in the bulk and at the surface of
condensed matter. Nucl. Instrum. Methods 1982, 198, 81.
18. Echenique, P. M.; Flores, F.; Ritchie, R. H. Dynamic screening of ions in condensed
matter. Solid State Phys. 1990, 43, 229.
19. Powell, C. J. Attenuation lengths of low-energy electrons in solids. Surf. Sci. 1974, 44, 29.
20. Ritchie, R. H.; Howie, A. Electron excitation and the optical potential in electron
microscopy. Philos. Mag. 1977, 36, 463.
21. Ashley, J. C. Optical-data model for the stopping power of condensed matter to protons
and antiprotons. J. Phys.: Condens. Matter 1991, 3, 2741.
22. Garcia-Molina, R.; Abril, I.; Kyriakou, I.; Emfietzoglou, D. Energy loss of swift pro-
tons in liquid water: role of optical data input and extension algorithms. In Radiation
Damage in Biomolecular Systems; García Gómez-Tejedor, G., Fuss, M. C., Eds.; Springer:
Dordrecht, 2012; Chapter 15.
23. Mermin, N. D. Lindhard dielectric function in the relaxation-time approximation. Phys.
Rev. B 1970, 1, 2362.
24. Lindhard, J. On the properties of a gas of charged particles. K. Dan.Vidensk. Selsk. Mat.-
Fys. Medd. 1954, 28(8).
25. Ritchie, R. H. Interaction of charged particles with a degenerate Fermi–Dirac electron
gas. Phys. Rev. 1959, 114, 644.
26. Belkić, Dž.; Mančev, I.; Hanssen, J. Four-body methods for high-energy ion-atom col-
lisions. Rev. Mod. Phys. 2008, 80, 249.
27. Brandt, W.; Kitagawa, M. Effective stopping-power charges of swift ions in condensed
matter. Phys. Rev. B 1982, 25, 5631.
28. Brandt, W. Effective charges of ions and the stopping power of dense media. Nucl.
Instrum. Methods 1982, 194, 13.
29. Powel, C. J.; Jablonski, A. Evaluation of calculated and measured electron inelastic mean
free paths near solid surfaces. J. Phys. Chem. Ref. Data 1999, 28, 19.
30. Abril, I.; Garcia-Molina, R.; Denton, C. D.; Pérez-Pérez, F. J.; Arista, N. R. Dielectric
description of wakes and stopping powers in solids. Phys. Rev. A 1998, 58, 357.
31. Heredia-Avalos, S.; Garcia-Molina, R.; Fernández-Varea, J. M.; Abril, I. Calculated

energy loss of swift He, Li, B, and N ions in SiO2 , Al2 O3, and ZrO2. Phys. Rev. A 2005,
72, 052902.
32. Planes, D. J.; Garcia-Molina, R.; Abril, I.; Arista, N. R. Wavenumber dependence of the
energy loss function of graphite and aluminium. J. Electron Spectrosc. Relat. Phenom. 1996,
82, 23.
33. Emfietzoglou, D.; Abril, I.; Garcia-Molina, R.; Petsalakis, I. D.; Nikjoo, H.; Kyriakou, I.;
Pathak, A. Semi-empirical dielectric descriptions of the Bethe surface of the valence
bands of condensed water. Nucl. Instrum. Methods Phys. Res., Sect. B 2008, 266, 1154.
34. Abril, I.; Garcia-Molina, R.; Denton, C. D.; Kyriakou, I.; Emfietzoglou, D. Energy loss of
hydrogen- and helium-ion beams in DNA: calculations based on a realistic energy-loss
function of the target. Radiat. Res. 2011, 175, 247.
35. de Vera, P.; Abril, I.; Garcia-Molina, R. Inelastic scattering of electron and light ion
beams in organic polymers. J. Appl. Phys. 2011, 109, 094901.
36. Hayashi, H.; Watanabe, N.; Udagawa, Y.; Kao, C.-C. The complete optical spectrum
of liquid water measured by inelastic x-ray scattering. Proc. Natl. Acad. Sci. USA 2000,
97, 6264.
37. Inagaki, T.; Hamm, R. N.; Arakawa, E. T.; Painter, L. R. Optical and dielectric properties
of DNA in the extreme ultraviolet. J. Chem. Phys. 1974, 61, 4246.
38. Ritsko, J. J.; Brillson, L. J.; Bigelow, R. W.; Fabish, T. J. Electron energy loss spectroscopy
and the optical properties of polymethylmethacrylate from 1 to 300 eV. J. Chem. Phys.
1978, 69, 3931.
39. Moreno-Marín, J. C.; Abril, I.; Garcia-Molina, R. Stopping power calculation of rubid-
ium and strontium for protons. Nucl. Instrum. Methods Phys. Res., Sect. B 2002, 193, 30.
40. Henke, B. L.; Gullikson, E. M.; Davis, J. C. X-ray interactions: photoabsorption, scatter-
ing, transmission, and reflection at E = 50–30,000 eV, Z = 1–92. At. Data Nucl. Data Tables
1993, 54, 181.
41. The ASCII files for the f1 and f2 scattering factors of the different elements can be
obtained from Chantler, C. T.; Olsen, K.; Dragoset, R. A.; Chang, J.; Kishore, A. R.;
Kotochigova, S. A.; Zucker, D. S. X-ray form factor, attenuation and scattering tables
(version 2.1). National Institute of Standards and Technology: Gaithersburg, MD, 2005.
<http://www.nist.gov/pml/data/ffast/index.cfm>.
42. Denton, C. D.; Abril, I.; Garcia-Molina, R.; Moreno-Marín, J. C.; Heredia-Avalos, S.
Influence of the description of the target energy-loss function on the energy loss of
swift projectiles. Surf. Interface Anal. 2008, 40, 1481.
43. Watanabe, N.; Hayashi, H.; Udagawa, Y. Bethe surface of liquid water determined by
inelastic X-ray scattering spectroscopy and electron correlation effects. Bull. Chem. Soc.
Jpn. 1997, 70, 719.
44. Watanabe, N.; Hayashi, H.; Udagawa, Y. Inelastic X-ray scattering study on molecular
liquids. J. Phys. Chem. Solids 2000, 61, 407.
45. Paul, H.; Geithner, O.; Jäkel, O. The influence of stopping powers upon dosimetry for
radiation therapy with energetic ions. Adv. Quantum Chem. 2007, 52, 289.
47. Garcia-Molina, R.; Abril, I.; Denton, C. D.; Heredia-Avalos, S.; Kyriakou, I.;

Emfietzoglou, D. Calculated depth–dose distributions for H+ and He+ beams in liquid
48. Tan, Z.; Xia, Y.; Zhao, M.; Liu, X.; Li, F.; Huang, B.; Ji, Y. Electron stopping power and
mean free path in organic compounds over the energy range of 20–10,000 eV. Nucl.
49. Tan, Z.; Xia, Y.; Zhao, M.; Liu, X. Proton stopping power in a group of bioorganic
compounds over the energy range of 0.05–10 MeV. Nucl. Instrum. Methods Phys. Res.,
Sect. B 2006, 248, 1.
50. Paul, H. The stopping power of matter for positive ions. In Modern Practices in Radiation
Therapy; Natanasabapathi, G., Ed.; InTech: Rijeka, 2012; Chapter 7.
51. Kumazaki, Y.; Akagi, T.; Yanou, T.; Suga, D.; Hishikawa, Y.; Teshima, T. Determination
of the mean excitation energy of water from proton beam ranges. Radiat. Meas. 2007,
42, 1683.
and a new I-value for water. Phys. Med. Biol. 2009, 54, 3451.
53. LaVerne, J. A.; Pimblott, S. M. Electron energy-loss distributions in solid, dry DNA.
Radiat. Res. 1995, 141, 208.
54. Akkerman, A.; Akkerman, E. Characteristics of electron inelastic interactions in organic
compounds and water over the energy range 20–10000 eV. J. Appl. Phys. 1999, 86, 5809.
55. Schiwietz, G.; Grande, P. L. Improved charge-state formulas. Nucl. Instrum. Methods Phys.
Res., Sect. B 2001, 175–177, 125.
56. Boudaïffa, B.; Cloutier, P.; Hunting, D.; Huels, M. A.; Sanche, L. Resonant formation of
DNA strand breaks by low-energy (3 to 20 eV) electrons. Science 2000, 287, 1658.
57. Huels, M. A.; Boudaïffa, B.; Cloutier, P.; Hunting, D.; Sanche, L. Single, double, and
multiple double strand breaks induced in DNA by 3-100 eV electrons. J. Am. Chem. Soc.
2003, 125, 4467.
58. Gervais, B.; Beuve, M.; Olivera, G. H.; Galassi, M. E. Numerical simulation of multiple
ionization and high LET effects in liquid water radiolysis. Radiat. Phys. Chem. 2006,
75, 493.
59. Uehara, S.; Toburen, L. H.; Wilson, W. E.; Goodhead, D. T.; Nikjoo, H. Calculations of
electronic stopping cross sections for low-energy protons in water. Radiat. Phys. Chem.
2000, 59, 1.
60. Dingfelder, M.; Inokuti, M.; Paretzke, H. G. Inelastic-collision cross sections of liquid
61. Emfietzoglou, D.; Pathak, A.; Papamichael, G.; Kostarelos, K.; Dhamodaran, S.; Sathish, N.;
Moscovitch, M. A study on the electronic stopping of protons in soft biological matter.
62. Tan, Z.; Xia, Y.; Zhao, M.; Liu, X. Proton inelastic mean free path in a group of bioor-
ganic compounds and water in 0.05–10 MeV range – including higher-order correc-
tions. Nucl. Instrum. Methods Phys. Res., Sect. B 2010, 268, 2337.
63. Shimizu, M.; Hayakawa, T.; Kaneda, M.; Tsuchida, H.; Itoh, A. Stopping cross-sections
of liquid water for 0.3–2.0 MeV protons. Vacuum 2010, 84, 1002.
64. Shimizu, M.; Kaneda, M.; Hayakawa, T.; Tsuchida, H.; Itoh, A. Stopping cross sections
of liquid water for MeV energy protons. Nucl. Instrum. Methods Phys. Res., Sect. B 2009,
267, 2667.
65. Siiskonen, T.; Kettunen, H.; Perajarvi, K.; Javanainen, A.; Rossi, M.; Trzaska, W. H.;
Turunen, J.;Virtanen, A. Energy loss measurement of protons in liquid water. Phys. Med.
Biol. 2011, 56, 2367.
66. Wenzel, W. A.; Whaling, W. The stopping cross section of D2 O ice. Phys. Rev. 1952,
87, 499.
67. Andrews, D. A.; Newton, G. The stopping power of heavy ice for low-energy (10–
30 keV) deuterons. J. Phys. D: Appl. Phys. 1977, 10, 845.
68. Bauer, P.; Kaferbock,W.; Necas,V. Investigation of the electronic energy loss of hydrogen
ions in H2 O: influence of the state of aggregation. Nucl. Instrum. Methods Phys. Res., Sect.
B 1993, 93, 132.
69. Garcia-Molina, R.; Abril, I.; Heredia-Avalos, S.; Kyriakou, I.; Emfietzoglou, D. A com-
bined molecular dynamics and Monte Carlo simulation of the spatial distribution of
energy deposition by proton beams in liquid water. Phys. Med. Biol. 2011, 56, 6457.
70. Vinson, J.; Kas, J. J.; Vila, F. D.; Rehr, J. J.; Shirley, E. L. Theoretical optical and X-ray
spectra of liquid and solid H2 O. Phys. Rev. B 2012, 85, 045101.
71. Michaud, M.; Wen, A.; Sanche, L. Cross sections for low-energy electron elastic and
inelastic scattering in amorphous ice. Radiat. Res. 2003, 159, 3.
72. Kobayashi, K. Optical spectra and electronic structure of ice. J. Phys. Chem. 1983,
87, 4317.
73. Pimblott, S. M.; LaVerne, J. A. Production of low-energy electrons by ionizing radiation.
Radiat. Phys. Chem. 2007, 76, 1244.
74. Mozejko, P.; Sanche, L. Cross sections for electron scattering from selected components
of DNA and RNA. Radiat. Phys. Chem. 2005, 73, 77.
75. Sanche, L. Low energy electron-driven damage in biomolecules. Eur. Phys. J. D 2005,
35, 367.
76. Garcia-Molina, R.; Abril, I.; de Vera, P.; Kyriakou, I.; Emfietzoglou, D. In Proton beam
irradiation of liquid water: a combined molecular dynamics and Monte Carlo simula-
tion study of the Bragg peak profile; Fast Ion-Atom and Ion-Molecule Collisions, Belkić, Dž.
Ed.; World Scientific Publishing Company: Singapore, 2012, Chapter 8, pp. 271–304.
77. Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Oxford University Press:
Oxford, 1989.
78. Möller, W.; Pospiech, G.; Schrieder, G. Multiple scattering calculations on ions passing
through thin amorphous foils. Nucl. Instrum. Methods 1975, 130, 265.
79. Palmans, H.; Symons, J. E.; Denis, J.-M.; de Kock, E. A.; Jones, D. T. L.; Vynckier, S.
Fluence correction factors in plastic phantoms for clinical proton beams. Phys. Med. Biol.
2002, 47, 3055.
80. Zhang, R.; Taddei, P. J.; Fitzek, M. M.; Newhauser, W. D. Water equivalent thickness
values of materials used in beams of protons, helium, carbon and iron ions. Phys. Med.
Biol. 2010, 55, 2481.
CHAPTER SEVEN
The Dielectric Formalism for

Inelastic Processes in High-Energy
Ion–Matter Collisions
Claudia C. Montanari and Jorge E. Miraglia
Instituto de Astronomía y Física del Espacio, CONICET and Universidad de Buenos Aires, casilla de correo
67, sucursal 28, C1428EGA, Buenos Aires, Argentina
Contents
1. Introduction 166
2. The shellwise local plasma
approximation 167
2.1 Historical aspects 167
2.2 Theoretical details: The Levine–Louie dielectric function and the independent-
shell approximation 168
3. Energy loss in particle penetration
of matter 173
3.1 The SLPA in stopping power of metals for protons 173
3.2 Stopping number and Bethe limit 178
3.3 Stopping for dressed ions 180
4. Energy loss straggling 182
4.1 The SLPA for stopping and straggling of gases 186
5. Ionization probabilities 190
5.1 Total ionization cross sections 191
5.2 Multiple ionization 192
5.3 Differential cross sections 193
6. Conclusions and Future Prospects 195
Acknowledgments 196
References 196
Abstract
In this chapter we analyze the possibilities and ranges of validity of the dielectric
formalism to deal with correlated bound electrons in matter by using the shellwise
local plasma approximation. This model describes the response of the electrons of
the same binding energy as a whole (collectively), screening the interaction with
the impinging ion. It considers separately each sub-shell of target electrons, with the
corresponding dielectric response. The density of electrons and the energy gap are
included explicitly by employing the Levine and Louie dielectric function. The goal
of this chapter is to summarize and review the capability of this model to deal with
fundamental magnitudes of the atomic collisions expressed as different moments of

166 Claudia C. Montanari and Jorge E. Miraglia
the energy loss: ionization cross sections (single or multiple, differential, and total),
stopping power (and mean excitation energy), and energy loss straggling. This review
covers a wide range of the collisions of ions with gases and solids, paying special
attention to multi-electronic targets. The advantages and disadvantages of the model
in comparison with independent electron ones, ranges of validity and future prospect
will be considered.
1. INTRODUCTION
When a swift heavy particle travels through matter, different inelastic
processes take place. The atomic electrons are excited or ionized, making
the particle to lose energy. The probability of these events to happen, the
mean energy loss during the ion passage, the effects on the target atoms
(i.e., how deep are the ionized shells, what is the final charge state of the
atom) are subjects of study for the different theories on heavy particle
collisions.1, 2 This research branch finds important applications in many
fields, including medicine within hadron therapy (for recent reviews, see
Refs. 3–6).
In this chapter we discuss the possibilities and ranges of validity of
the shellwise local plasma approximation (SLPA) to deal with some of
the fundamental inelastic collisions mentioned above. This is a many-
electron model within the frame of the dielectric formalism, especially
suitable for multi-electronic targets and high energy collisions in which
target deep shells are involved.7, 8 The SLPA describes the electronic
response of each sub-shell of target electrons as a whole, including
screening among electrons.9 This is of particular interest when describ-
ing many-electron sub-shells such as 4f or 3d , for example. The main
characteristics of the SLPA are the independent-shell approximation (a
dielectric function for each sub-shell of target electrons, meaning that
only the electrons of the same binding energy respond collectively to
the ion perturbation and screen among them) and the inclusion of the
binding energy explicitly (not free-electron gas, but electron gas with
an energy threshold).
It must be stressed that the SLPA is an ab initio calculation (no param-
eters included) whose only inputs are the atomic densities of the different
sub-shells and the corresponding binding energies. It allows us to calculate
the different moments of the energy loss: ionization cross sections (single
or multiple, differential, and total), stopping power (and mean excitation
energy), and energy loss straggling. The advantages and disadvantages of the
The Dielectric Formalism for Inelastic Processes in High-Energy Ion–Matter Collisions 167
model, as well as ranges of validity and future prospects will be discussed

in the following pages.
2. THE SHELLWISE LOCAL PLASMA

APPROXIMATION
2.1 Historical aspects
When a fast heavy ion moves in a medium, it polarizes the target electron
cloud. This gives rise to an induced potential, which can be described as a
trailing wake that follows the motion of the projectile (see Ref. 9 and ref-
erences therein). The many-body consistent treatment for an ion traveling
through an homogeneous free-electron gas was developed by Lindhard10
and by Ritchie,11 within the linear response approximation (LRA).
This dielectric formalism was extended to deal with atomic bound
electrons as a free-electron gas of local density, known as the local plasma
approximation (LPA). It was applied to stopping power (energy loss per unit
length) of heavy ions in matter using the logarithmic high energy limit.12, 13
Later on developments of the LPA included the extension to isolated
atoms by Rosseau et al.14 and to intermediate energies by employing
the fully dielectric formulation.15–17 In the LPA by Lindhard and sub-
sequent works, the response of bound electrons, even local, considers
the electronic cloud as a whole by using the total density of electrons
in the atom.
In the last years we have improved and extended the use of the
dielectric formalism to deal with bound electrons of gaseous and solid
targets (insulators and metals), not only for energy loss calculations, but
also for ionization probabilities and energy loss straggling.7, 8, 18–33 The
full dielectric formalism was employed together with the theoretical
description of bound electrons through the atomic wave functions and
binding energies.
Our developments for the SLPA had two main steps. First the separate
dielectric response for each shell, by considering the independent-shell
approximation. Physically, this independent-shell approximation means that
when an electron of the nl sub-shell is ionized only the other nl-electrons
are included in the screening of the ion potential. This gave very good
results for the energy loss even at intermediate energies,18–25 as far as the
perturbative approximation is valid. A previous proposal of the independent
shells within the LPA, known as the orbital OLPA, is due to Meltzer et al.34
However, this approach uses the logarithmic high energy limit for the stop-
ping power.
The second step was the inclusion of the ionization gap using the
Levine and Louie dielectric function.35 This is a very important point to
advance in the calculation of ionization probabilities, not only for inner-
shells of metals, but also for insulators26 and gases.27 In this sense, the SLPA
provides an interesting alternative to the independent electron models, with
very good results as compared with complex formulations such as the con-
tinuum distorted wave eikonal initial state (CDW-EIS) approximation,28
and with great advantages in the computational effort and time.
The Levine and Louie dielectric function35 keeps the characteristics
of Lindhard’s,10 i.e., electron–electron correlation to all orders, collective
response, and f-sum rule (particle number conservation). It must be noted
that the SLPA is a first order approximation (as far as the dielectric functions
are calculated in LRA, like Lindhard’s 10 or variations, such as Mermin’s36
or Levin–Louie’s35). It means that it is a perturbative description valid for
ZP /v < ZT and v > ve with ve being the mean velocity of the electrons
of the nl-shell. The SLPA with this dielectric function proved to give good
results in stopping power calculations7, 29–33 in an extended energy range
that includes the maximum of the stopping, and for inner-shell ionization
of very heavy targets at high enough impact velocities.8
2.2 Theoretical details: The Levine–Louie dielectric function

and the independent-shell approximation
Consider a bare ion of charge ZP moving at velocity v in a free-electron gas
(FEG) of homogeneous density ρ, a Fermi momentum kF = [3π 2 ρ]−1/3
and a Seitz radio rS = [(4/3)πρ]1/3. Following the dielectric formalism, the
stopping power per unit length is expressed as
2ZP2 ∞ dk kv

−1
S= ω Im dω, (1)
π v2 0 k 0 ε(k, ω)
with ε(k, ω) being the quantum dielectric function of the medium.9

Different approximations are employed for the dielectric function, with
the Mermin–Lindhard dielectric function36 being the most accurate one
to account for the plasmon time decay. This dielectric function depends
on the constant density of electrons ρ and the plasmon damping γ, both
characteristics obtained from the optical properties of each material37 and
from tabulations.38
The SLPA formulation has a general expression for the different

moments of the energy loss. It describes the inelastic collision consider-
ing the interaction with each nl sub-shell of target electrons separately,
with Snl(t) being its energy moment of order t (t = 0 the ionization cross
section, t = 1 the stopping power, t = 2 the square energy loss straggling)
given by
∞ 2 kv
2 ZP dk −1
(t)
Snl = t
ω Im dω. (2)
π v2 0 k 0 εnl (k, ω)
The total
moment will be the addition of the shell contributions
S (t) = nl Snl .29 In the case of metals, electrons are treated differently
(t)
depending on whether they belong to the free-electron gas (FEG) or to

the inner-shells (bound electrons).
The dielectric function εnl (k, ω) in Eq. (2) is a mean value of a local
response that depends on the density of electrons of the shell ρnl (r) and the
ionization gap ǫnl

−1 −1
Im
εnl (k, ω)
= Im LL F (r), ǫ )
dr (3)
ε (k, ω, knl nl
F (r) = [3π 2 ρ (r)]1/3

with knl nl the local Fermi velocity. The dielectric func-
tion employed εLL, is the Levine–Louie one,35 which includes explicitly
the energy gap of each shell. It is defined as
F (r)) , ω > |ǫ |
L
Im ε (q, ωg , knl nl
(4)
LL F

Im ε (q, ω, knl (r)) =
0, ω < |ǫnl |

with ωg = ω2 + ǫnl2 and εL (q, ω, knl F (r)) being the usual Lindhard
dielectric function.10 Note that if we consider no binding energy,

ǫnl = 0, the usual expression for the probability in the dielectric for-
malism (Lindhard) is recovered. Once the imaginary part is defined,
the real part, Re εLL (q, ω, knl F (r)) is obtained in closed form through
the Kramers–Kronig relation, as shown in Ref. 35. This model for the
dielectric function, proposed originally for semiconductors and insula-
tors,35 satisfies the so-called f-sum rule, or particle number conservation,
that is the desirable feature for a dielectric function. In fact, this modified
version of the SLPA was first applied to calculation of stopping power
in insulators.26
The density of electrons of the shell ρnl (r) and the ionization gap ǫnl
are the only inputs for the SLPA. For atoms they can be obtained from the
Hartree–Fock wave functions tabulated by Clementi and Roetti39 or by
Bunge et al.40 For very heavy atoms (Z > 54) they can be calculated using
the relativistic solutions of the Schrödinger equation.7, 8, 32
In the case of an ion with N bound electrons and charge state
q = ZP − N , the interaction with target electrons can be expressed
(q)
through a screened ion-charge ZP (r) that depends on the distance to the
nucleus (with the ion electrons frozen, just screening the nucleus). It veri-
(q) (q)
fies ZP (r) → q for long distance collisions and ZP (r) → ZP for close
collisions.
In order to include this inhomogeneous screened ion-charge in Eq. (2),
(q)
instead of ZP, we need the Fourier transform of ZP (r) given by
N
(q)

ZP (k) = ZP − ϕj e ik.r ϕj . (5)

j=1
Using Flannery integrals41 and the Slater-type expansion for the projectile
wave functions (Clementi and Roetti tables39 for neutral atoms and for
positive ions), ZP(q) (k) has an analytical closed form (see the Appendix in
Ref. 8 for the details).
(q)
We calculated ZP (r) for different ions from He+q (q = 0, 1) up to
Ne+q (q = 0, . . . , 9). In order to express it in a simple way to be included
in the programs of calculus we fitted the exact ZP(q) (r) with two exponen-
tials as follows:
2
(q)

ZP (r) =q+ Zj exp(−aj r). (6)
j=1
In Tables 7.1 and 7.2 we include the parameters Zj , aj for the different ions
and charge states.
The inclusion of ZP(q) (k) in Eq. (2) is straightforward. The SLPA for
the total energy moment of order t due to the interaction with an ion of
nucleus charge ZP and charge state q is
(t)
2 ∞ [Z (q) (k)]2 dk kv
−1

S(q) = P t
ω Im dω. (7)
πv 2 0 k 0 εnl (k, ω)
n,l
In the following sections we present and discuss the current state of

the SLPA to deal with energy loss (Section 3), its straggling (Section 4),
and with ionization cross sections (Section 5). The order was chosen by
Table 7.1 Parameters to fit the screened ion-charge as function of the distance to
the nucleus for He+q to N+q. These functions were obtained using Hartree–Fock wave
functions by Clementi and Roetti.39 for neutral atoms and for positive ions
Ion q Z1 a1 Z2 a2
He+ 1 −1.289 4.722 2.289 3.529
He0 0 −0.288 7.784 2.288 2.359
Li2+ 2 −0.101 1.017 1.101 3.411

Li+ 1 −0.195 0.957 2.195 3.024
Li0 0 1.123 0.531 1.877 3.347
Be3+ 3 −0.100 1.359 1.100 4.554

Be2+ 2 −0.334 16.900 2.334 5.298
Be+ 1 1.172 0.942 1.828 4.635
Be0 0 2.412 0.824 1.588 5.161
B4+ 4 −0.106 1.730 1.106 5.674

B3+ 3 3.684 7.832 −1.684 11.561
B2+ 2 1.250 1.376 1.750 6.138
B+ 1 2.532 1.255 1.468 6.823
B0 0 3.631 1.075 1.369 7.285
C5+ 5 −0.097 2.101 1.097 6.891

C4+ 4 −0.208 2.120 2.208 6.459
C3+ 3 1.783 7.227 1.217 1.731
C2+ 2 2.591 1.661 1.409 8.473
C+ 1 3.753 1.520 1.247 9.114
C0 0 4.801 1.331 1.199 9.446
N6+ 6 −0.058 2.353 1.058 8.420

N5+ 5 −0.216 2.452 2.216 7.553
N4+ 4 1.805 8.357 1.195 2.086
N3+ 3 2.639 2.077 1.361 10.093
N2+ 2 3.846 1.965 1.154 11.027
N+ 1 4.953 1.792 1.047 11.776
N0 0 5.973 1.596 1.027 12.094
Table 7.2 Similar to Table 7.1 for O+ q to Ne+ q ions

Ion q Z1 a1 Z2 a2
O7+ 7 −0.093 2.776 1.093 9.228

O6+ 6 −0.207 2.825 2.207 8.744
O5+ 5 1.766 9.774 1.234 2.514
O4+ 4 1.299 10.976 2.701 2.591
O3+ 3 3.899 2.395 1.101 12.836
O2+ 2 5.043 2.237 0.957 13.962
O+ 1 6.113 2.052 0.887 14.782
O0 0 7.087 1.824 0.913 14.749
F8+ 8 −0.043 2.881 1.043 11.032

F7+ 7 −0.130 2.543 2.130 10.157
F6+ 6 1.754 11.088 1.246 2.912
F5+ 5 1.296 13.463 2.704 2.896
F4+ 4 3.958 2.834 1.042 15.007
F3+ 3 5.190 2.750 0.810 16.145
F2+ 2 6.277 2.554 0.723 17.556
F+ 1 7.245 2.286 0.755 18.684
F0 0 8.213 2.064 0.787 18.438
Ne9+ 9 −0.095 3.316 1.095 11.422

Ne8+ 8 −0.166 3.366 2.166 11.214
Ne7+ 7 1.787 12.100 1.213 3.247
Ne6+ 6 1.309 14.616 2.691 3.279
Ne5+ 5 1.024 16.471 3.976 3.256
Ne4+ 4 5.181 3.129 0.819 19.045
Ne3+ 3 6.299 2.953 0.701 21.086
Ne2+ 2 7.350 2.738 0.650 22.470
Ne+ 1 8.360 2.524 0.640 23.075
Ne0 0 9.326 2.306 0.674 22.586
historical reasons: the dielectric formalism, and more specifically, the local
plasma approximation, was developed to describe the energy loss of heavy
particles in solids. Instead, using a collective formalism to calculate ioniza-
tion probabilities of solids and, even more of gaseous targets, is much more
demanding and is new theoretical development due to the SLPA.
3. ENERGY LOSS IN PARTICLE PENETRATION

OF MATTER
The mean value of the energy loss of particles in matter, also
known as stopping power, is a necessary ingredient of many parts of
basic science, of medical and technological applications.42, 43 It is an
average of the ion energy loss per unit path length. At intermediate to
high impact velocities, this energy loss is related to ionization of target
electrons. The higher the ion energy, the deeper the excited electrons. In
the case of metals, it means that at intermediate and high impact ener-
gies, the description of the stopping power due to excitation of valence
electrons (the FEG of metals) is not enough, and target inner-shells must
be included.
Different experimental methods are used to determine stopping
powers44 and important collections of data, statistics, and semiem-
pirical functions are available in the web.45, 46 Many semiempirical and
theoretical models have been developed (see Refs. 47–50 and references
therein). However, the description of very heavy targets, i.e., those with
the close 4f sub-shell with 14 electrons, remains a heavy task for first
principle theoretical calculations. On the contrary, the SLPA, being a
many-electron model, describes with the same degree of complexity
Ne, Cu, or Au targets. Furthermore, the more electrons the shell has,
the more suitable the method is. On the other hand, this model works
within the dielectric formalism, so dynamical screening effects among
electrons are included. This proved to be of particular interest for the
case of the f-shells.32
The only inputs for the SLPA are the density of electrons around the
nucleus for each sub-shell and the binding energy. As far as these inputs are
known, the SLPA calculations reduce to rather simple numerical integra-
tions, given by Eqs. (2) and (7). Moreover, the extension of this formalism
to deal with complex targets (compounds, molecules of biological interest,
water) is straightforward if these inputs are known.
3.1 The SLPA in stopping power of metals for protons

The SLPA gives good results for energy loss by protons in metal targets.7, 31, 32
We consider and discuss here three cases of special interest , Cu (Z = 29),
W (Z = 74), and Au (Z = 79).This choice is founded on their experimental
and technological interest, and on the availability of data. Cu and Au are

the two targets with the largest amounts of stopping measurements. For
W, new interest is related to its properties as inner-wall of fusion reac-
tors.51 For our theoretical interest, these are targets for which outer elec-
trons can be described as FEG and we can deal separately with bound
electrons using the SLPA. This separate description of the response of
bound and valence electrons as inhomogeneous and homogeneous elec-
tron gases, respectively, is part of the good agreement obtained. On the
other hand, the separate calculations allow using different models in each
case (i.e., the SLPA for bound electrons and a non-perturbative formalism
for the FEG).
For Cu, the densities and binding energies were obtained from the
tabulated Hartree–Fock wave functions.39, 40 For Au and W, the atomic
wave functions are the solutions of the relativistic Dirac equation instead
the nonrelativistic Schrödinger equation. We employed to this end the fully
relativistic ab initio wave functions obtained by using the hullac52 code (see
Ref. 7 for the details). The binding energies were compared with tabulated
experimental binding energies in solids by Williams.53
For H ions in solids we calculate the stopping cross section just consid-
ering H+(Refs. 7, 32). At low velocities, the screening of the H+ nucleus is
strong enough to have very loose or directly no bound electron.54, 55
The total stopping of protons in Cu is displayed in Figure 7.1. This
value was obtained by adding the bound electron and the FEG contribu-
tions. The former was calculated with the SLPA formalism (Eq. (2) with
t = 1) and adding the contributions of the different sub-shells to get
total stopping of bound electrons. The latter by employing the dielec-
tric formalism (perturbative approximation) with the Mermin–Lindhard
dielectric function.36 The characteristic plasmon frequency and width
employed here for Cu FEG are ωp = 0. 703 a.u. and γ = 0. 950 a.u.,
respectively.This yields a mean value of electrons in the FEG as Ne = 3. 14,
and a Seitz radius as rS = 1. 82 a.u. These values were obtained from the
optical data of the energy loss function37 by considering only the first
important peak. These number of electrons in the FEG is similar to the
experimental value recommended by Isaacson.38 To keep the total num-
ber of electrons, we considered Cu as [Ar] 3d 7.86 and the "3. 14" electrons
as FEG.
In Figure 7.1 we compare our total stopping cross sections of Cu
for protons, with the experimental data available,45 and with srim08
results.46 The contribution from the FEG and the bound electrons are
Figure 7.1 Stopping cross sections of Cu for protons. Curves: solid line, total stopping
adding the contributions by bound electrons (blue dashed line, SLPA) and by the FEG
(red dotted line, Mermin–Lindhard dielectric calculation); orange dash dotted line,
46 45
srim08 results. Symbols: experiments as reported by Paul; different symbols marked
within the figure, experimental data since 1980; black hollow circles, experimental data
corresponding to (1935–1979).
displayed separately. About the experimental data, we indicate separately

only the data since 1980, while earlier data is plotted together with a
single type of symbol. The agreement with the experimental data and
with the srim08 curve46 is very good in the whole energy range. The
stopping maximum is correctly described in energy and value. The dif-
ferential description for the FEG and the bound electron contribution is
the basis for this agreement.56
We display in Figures 7.2 and 7.3 the stopping cross sections of Au
and W for protons.7, 32 These are very interesting targets because the 4f
sub-shell plays a major role. For solid Au, we describe the FEG with the
Mermin–Lindhard dielectric function and the following parameters (in
atomic units): Seitz radio rS = 1. 17 a.u. (17 electrons in the FEG), plasmon
energy ωP = 1. 37 a.u., and a damping γ = 1. 37 a.u. These values were
obtained as a first approximation to optical energy loss function tabulated
in the Handbook of Optical Constants by Palik and Ghosh.37 This means
that the atomic 5p6 , 5d 10 and 6s1 electrons are considered as the homoge-
neous FEG, and the first bound electrons are those of the 4f sub-shell. The
bound electron contribution was calculated using the SLPA with the fully
relativistic wave functions and binding energies.7
In Figure 7.2 we display our theoretical results for the stopping cross
section together with the large amount of experimental data available,45
and the srim 2008 values. The contributions of the FEG and the bound
electrons are displayed separately. The FEG contribution in gold, even the
main one, cannot explain the total results for energies above 100 keV. As
observed in the figure, there is an important dispersion of experimental
values, mainly around the stopping maximum, which makes any theoretical
description not conclusive. In order to have a cleaner picture of the situa-
tion, we displayed separately the stopping measurements of the last 25 years.
This arbitrary classification allowed us to note that, except for the results
by Martinez-Tamayo et al.57 the latest stopping measurements tend to be
close to a single curve. The semiempirical srim08 code 46 gives a curve in
accordance with these latest experimental data.
The total stopping cross section obtained using the SLPA for bound
electrons describes well the experimental data in the whole energy range. It
has a maximum at 130 keV, shifted and higher than srim08,46 and in rather
good agreement with theoretical predictions by Heredia-Avalos et al.58 and
Figure 7.2 Stopping cross section of Au for protons. Curves: dashed line, present SLPA
calculation for the bound electrons; dotted line, FEG contribution using the Mermin–
Lindhard dielectric function;36 solid line, total stopping as the addition of the previous
two contributions; dashed dotted line, srim08.46 Symbols: experiments as reported by
Paul;45 different symbols within the figure are data since 1987; hollow circles, previous
measurements.
with the measurements by Martinez-Tamayo et al.57 and to previous ones

by Kreussler et al.59 and Santry and Werner.60
It is worth mentioning that the relativistic results for the binding ener-
gies of Au show spin-orbit split in energy En,l,l± 1 (see Fig. 1 in Ref. 7).
2
It is an important point, because these are sub-shells that are very close in
energy. As mentioned before, the SLPA describes collectively those elec-
trons with equal binding energy, allowing screening among them. We use
Ei±Ei as criterion of equal binding energy, with Ei being the quantum
uncertainty
1 v
Ei ≈ = , (8)
ti �r�i
with Ei and ri being the energies and mean radius of the sub-shell.
Using this criterion we have found that in the cases studied, the spin-
orbit split is not resolved. Physically this implies that the 14 electrons of
the 4f sub-shell respond together shielding the projectile, and not the 8
electrons 4f7/2 and the 6 electrons 4f5/2 separately. This separated or not-
separated response of the sub-shells is very clear in the case of protons in
W, as displayed in Figure 7.3. For W, not only cannot the 4f7/2 and 4f5/2 be
resolved, but there is also inter-shell screening between 5p and 4f electrons.
As displayed in Table 7.3, these sub-shells are very close in energy and
respond collectively to the ion passage.The importance of this effect is clear
Table 7.3 Fully-relativistic binding energies of the N and O-shells of Au and W. These
results have been calculated by Mitnik and collaborators7, 32 with the HULLAC code52
for isolated atom. Also included are the experimental values, E expt, compiled by
Williams.53 The binding energies are in atomic units
expt th expt th
E Au E Au EW EW
4s 28.0 26.0 21.8 20.8

4p1/2 23.6 22.8 18.0 17.3
4p3/2 20.1 19.2 15.6 14.8
4d3/2 13.0 12.5 9.40 9.0
4d5/2 12.3 11.9 8.95 8.5
5s 3.94 4.1 2.78 2.9
4f5/2 3.22 3.2 1.23 1.3
4f7/2 3.08 3.1 1.16 1.2
5p1/2 1.66 1.9
5p3/2 1.35 1.5
Figure 7.3 Stopping cross sections of W for protons. Curves: solid lines, present theo-
retical results for the contributions by bound electrons (SLPA) and the FEG, and total
stopping as the addition of the previous two; dash lines, results obtained using the
SLPA with independent 5p and 4f responses. Symbols: different symbols within the
figure are the experimental data as reported by Paul.45
while comparing total theoretical and experimental stopping in Figure 7.3.

Note that the screening among electrons reduces the stopping cross section
in all the cases (in Au, too). This effect can only be taken into account with
a many-electron model and a collective description of bound electrons.
The results displayed in Figure 7.3 describe the stopping maximum quite
well. However an overestimation of the experimental data around 2 MeV
is found. We will return to this matter in the following section in relation
to the Bethe high energy limit.61
3.2 Stopping number and Bethe limit

The SLPA provides an interesting alternative to describe the energy loss of
different materials for energetic ions. At high energies the ion losses energy
due to the interaction with deep bound electrons, so the behavior of the
stopping power at high energies is a good test for the model.
The stopping power can be expressed in terms of the dimensionless
stopping number L(v) as
4πZP2 ZT
S(v) = L(v). (9)
v2
In the high-but-nonrelativistic velocity regime, the stopping number is

described by the Bethe asymptotic formulae61
2
2v
lim L(v) = L Bethe
(v) = ln , (10)
v→∞ I
with I being the mean excitation energy, characteristic of each target.62

Our theoretical description of the energy loss should converge to the Bethe
high (but nonrelativistic) energy limit.
In Figure 7.4 we display the stopping number comparing the SLPA,
Bethe limit, and srim08 values,46 the latter representing the behavior of the
experimental data for each target. Using the Lindhard and Scharff scaling
with v 2 /ZT (valid for elements of high atomic number63) we plot together
the SLPA results for the stopping number of W and Au. The straight-line
in the logarithmic scale of v 2 is the Bethe limit.
It is an interesting figure because different aspects can be observed:
the Lindhard and Scharff scaling is valid for W and Au; the SLPA tends to
the Bethe limit, as expected theoretically; the description of the experi-
mental data with the SLPA is good at the lower energies displayed in this
figure, but in the high energy region the SLPA is closer to Bethe than to
the experiments. This is a 10% of theoretical-experimental difference in
the energy region 1–4 MeV, the energy region observed in Figure 7.3.
Figure 7.4 Lindhard scaling for the stopping number L. Curves: SLPA results for Au
(solid line) and W (dashed line); Bethe high energy limit obtained using Eq. (10) and the
mean excitation energies I suggested by the ICRU49 Report (Ref. 44) IW = 727 ± 30 eV,
and IA u = 790 ± 30 eV. Symbols: srim08 values for Au (hollow circles) and W (stars).
This means a certain overestimation by the SLPA in the high energy

region, which is related to the theoretical tie to the Bethe limit.
3.3 Stopping for dressed ions

By dressed ions we mean those ions with bound electrons, whether they
be neutral or with a certain charge state. In this case, while the ion moves
through a medium, loss and capture processes take place until reaching
an equilibrium distribution of charge states q (q = 0, . . . , ZP) within the
foil, depending on the ion velocity v. In the equilibrium regime, the mean
energy loss will be the average of the stopping cross sections S(q) given
by Eq. (7), with the calculated screened ion-charge (Tables 7.1 and 7.2),
weighted with the data of fraction of ions with charge q at that velocity,
φq (v). Namely
ZP

S= φq (v)S(q) . (11)
q=0
The equilibrium charge fractions at each impact velocity, φq (v), are exter-
nal inputs. A fitting of a large amount of experimental values of charge
states exiting the solid, combining different ions and targets is due to
Schiwietz and Grande.64 It is available within the CasP5.0 code.65
On the other hand, for ions with nuclear charge ZP 2, the
perturbative description for the FEG contribution is found to underesti-
mate the experimental data for energies below that of the maximum of
stopping power. For this reason for He and heavier ions we combine the
SLPA for the bound electrons with a non-perturbative model for the FEG.
As the energy loss by ionization of inner-shells contributes at high energies,
the perturbative approximation used in the SLPA is still valid. In fact, for
the stopping power of Zn for different ions we found the description of
bound electrons (1s up to 3d 9 ) with the SLPA to be valid for He up to C
projectiles, as displayed in Figure 7.5. For O ions in Zn the perturbative
SLPA overestimates the total stopping cross section.33
Figure 7.5 summarizes previous results for the different ions22, 29, 30, 33 in
Zn.The non-perturbative model for the FEG employed in these calculations
is the Transport Cross Section-Extended Friedel Sum Rule (TCS-EFSR)
approach by Arista and coworkers.66, 67
To analyze in detail the stopping of dressed ions in matter, we consider
the stopping of Cu and Au targets for He. In Figure 7.6 we display the
Figure 7.5 Total stopping cross sections of different ions in Zn. Curves: theoretical
calculations adding SLPA for bound electrons and HISTOP (non-perturbative) for the
FEG.22, 29, 30, 33 Symbols: experimental data for He, Li, Be, and B ions as in Figure 6 of
Cantero et al.;30 for C ions, Cantero et al.33
eV cm /atom)
80 (a) (b) He + Cu
80
2
with non-
perturbative
FEG
60 60
-15
Total
Stopping cross section (10
Bound
40 40
Mertens (1980)
20 Stoquert (1982) 20
FEG Sakamoto (1991)
(perturbative) Bak (1994)
Sillanpaa (1998)
0 0
2 3 4 5 2 3 4 5
10 10 10 10 10 10 10 10
Energy (keV) Energy (keV)
Figure 7.6 Stopping cross sections of Cu for He ions. Curves: (a) dashed line, the bound
electron contribution obtained using the SLPA; dotted line, the FEG contribution with
Mermin dielectric function (perturbative); solid line, total stopping adding both contri-
butions (similar to Figure 7.1); (b) black solid line, total stopping adding the SLPA results
for bound electrons and a perturbative description for the FEG with Mermin dielectric
function as in (a); gray solid line, total stopping adding the SLPA results for bound elec-
trons and a non-perturbative model for the FEG by Arista for He in Zn;22 dashed dotted
line, srim08;46 dashed double-dotted line, Casp5.0.65 Symbols: experiments as reported
by Paul;45 different symbols within the figure are data since 1980; hollow circles, previ-
ous measurements.
partial and total stopping cross section for He in Cu. In Figure 7.6a the
contributions of bound electrons and FEG are displayed separately. The
former calculated with the SLPA, the latter with the dielectric formal-
ism (perturbative, LRA), using the Mermin–Lindhard dielectric function
(similar to the case of protons in Cu). The behavior of the total stopping
displayed in Figure 7.6a has already been found for Zn (Ref. 22), i.e., below
100 keV/amu (v < 2 a.u.) the perturbative model does not describe the
experimental data. The replacement of the perturbative FEG-stopping by
the non-perturbative one gives the total stopping displayed in Figure 7.6b.
For the FEG we employed, as first approximation, the results for the FEG
of Zn for He (with rs = 2. 02) by Arista using the TCS-EFSR in Ref. 22.
The theoretical-experimental disagreement at low energies may be related
to this. However, the tendency to correct the underestimation due to the
perturbative calculation is clear. We also include in Figure 7.6b Schiwietz
and Grande results with the unitary convolution approximation (UCA) for
swift particles (available in the CasP5.0 code65), and the semiempirical val-
ues by the srim08 code.46 Above the stopping maximum all the theoretical
results agree quite well. For energies below this value the combination of
different calculations, for bound electrons and FEG shows very good agree-
ment if the non-perturbative FEG model is used.
For He in Au, in Figure 7.7 we display the theoretical results obtained
in perturbative approximation, with the SLPA for bound electrons and
Mermin–Lindhard dielectric function for valence electrons. As observed in
the case of protons in Au, for He in Au the dispersion of the experimental
data is important for impact energies below the stopping maximum. The
theoretical description is quite good for impact energies above 400 keV.
The maximum of the stopping is correctly described around 1 MeV. For
energies below 400 keV we are outside the limit of validity of the pertur-
bative approximation. The combination of SLPA with a non-perturbative
description for stopping due to valence electrons of Au (as for Cu or Zn
targets) is expected to improve these results.
4. ENERGY LOSS STRAGGLING

The theoretical square energy loss straggling, 2, or the second
moment of the energy loss (t = 2 in Eqs. (1) and (7)), describes the statisti-
cal dispersion of the energy loss. It represents the energy loss variance per
unit path length of a Gaussian-type energy loss distribution.68 The condi-
tion for obtaining a Gaussian distribution is that the energies transferred in
Figure 7.7 Stopping cross sections of Au for helium. Curves: dashed line, the bound
electron contribution obtained using the SLPA; dotted line, the FEG contribution with
Mermin dielectric function (perturbative); solid line, total stopping adding both con-
tributions (similar to Figure 7.1); dashed double-dotted line, srim08 results;46 dashed
dotted line, ICRU49;44 dashed double-dotted line, Casp5.0.65 Symbols: experiments
as reported by Paul;45 different symbols within the figure are data since 1982; hollow
circles, previous measurements.
the individual collisions should be small as compared to the width of the

final distribution.69
The high energy limit for the energy loss straggling was calculated by
Bohr69 as
�2B = 4πZP2 ZT , (12)
which is proportional to ZT, the total number of target electrons active in
the collision at sufficiently high energies.
The energy loss straggling is an interesting parameter to study theoreti-
cally and experimentally. It represents a sensitive input for many calculations
(Ziegler46 or Ma et al.70) and computer simulations, like simnra by Mayer71
for material analysis, or seics by Garcia-Molina and coworkers72, 73 for ele-
ments of biological interest.
For the stopping power, there are important compilations of data avail-
able45 and wellknown and tested semiempirical programs46 and recommended
values.44 However, the situation is completely different for energy loss strag-
gling.The most widely used electronic energy loss straggling is Yang empirical
fitting74 included in modern ion beam analysis codes such as simnra, ndf,
75
corteo, and mcerd. However the accuracy of this formulae is questioned
for different reasons.25, 76 The source of the Yang expression is a compila-

tion of data prior to 1990 which presents serious problems. Measurements
of energy loss straggling set severe requirements to target preparation (well
defined thin films, uniformity, and homogeneity).77 Experimental methods,
such as transmission or Rutherford back scattering, are very sensitive to
roughness and inhomogeneity of the samples, which introduce important
additional energy loss straggling, especially at low energies.68, 77, 78 It produces
overestimation in a region around the stopping maximum68 and an important
dispersion among data. The weight of this contribution in the experimental
straggling is clear in some measurements previous to 1980 included in Yang’‘s
compilation and fitting.74 Fortunately, there is a great number of recent
measurements from different laboratories and using different techniques that
show less spread and tend to be close to a single band.79–85
In this section we present different theoretical results calculated with the
SLPA. These results improve those by Chu86 using the LPA with Hartree–
Fock densities and considering the electronic cloud as a whole. The differ-
ence between our results and those by Chu is the shell to shell description
of the dielectric response.
In Figure 7.8 we display the SLPA results for square energy loss strag-
gling normalized to Bohr value69 for H and He ions in Cu. The different
Figure 7.8 Squared straggling of Cu for H and He ions, normalized to Bohr high energy
limit. Curves: solid lines, our SLPA results. Symbols: experimental data as indicated in
the figure. For H+ in Cu (Refs. 80–82, 88); for He in Cu (Refs. 83, 87).
charge states of He in Cu were considered to obtain these values, however

no important differences with respect to He2+ were obtained for energies
above 200 keV/amu. We include in Figure 7.8 the experimental data by
Hoffman and Powers87 and by Friedland and Kotze,88 which are not cor-
rected to exclude the inhomogeneity contribution. The overshooting of
these values is clear, mainly in the data by Hoffman and Powers.87
The SLPA values for the energy loss straggling tend to the Bohr limit
from below, and do not show the overshooting (Bethe–Livingston shoul-
der89) around the energy of maximum stopping power predicted by the
binary collision formalisms.89, 90 At high energies, the square energy loss
straggling tends to a value proportional to the total number of target elec-
trons, as predicted by Bohr.69 If we calculate the square straggling as

�2 /�2B = �2nl /�2B , (13)
nl
2 2
we can say that each term verifies that �nl /�B → Nnl /ZT, with Nnl being
the number of electrons in the nl sub-shell.7, 25 This is an interesting point
because it indicates the high energy limit expected for each shell. But it also
represents a demanding requirement for the theoretical calculation, because
all shells, even deep ones, contribute to the total straggling, i.e., the L-shell
of Au with eight electrons is 10% of the Au energy loss straggling.
On the other hand, we found that the energy loss straggling normal-
ized to ZP2 (Bohr limit) is almost independent of the ion atomic number
ZP, at least for low Z ions showing a perturbative dependence with the
ion-charge.25 In order to test this in Figure 7.9 we plotted together the
experimental data for different ions in Au together with the SLPA values
for the bare ions in Au. The experimental data included are those that
explicitly take into account the roughness and inhomogeneity of the
sample. In the case of the experimental data by Andersen et al.91 they have
been corrected in 10% due to the estimation of this contribution. The data
by Møller et al. shows asymmetric error bars that correspond to Figure 5
in Ref. 92.
We can observe that the soft dependence with the ion-charge of the
experimental data normalized to Bohr limit is valid at least for H, He, and
Li ions. The data for B in Au by Hsu et al.79 could indicate a deviation for
higher Z ions. On the other hand, Figure 7.9 emphasizes the good descrip-
tion of the straggling obtained with the SLPA, even for unexpected low
energies. Note that all target electrons have been considered in the calcula-
tion, even the very deep ones.
Figure 7.9 Squared straggling of Au for H, He, Li and B ions, normalized to Bohr high
energy limit. Curves: solid line, SLPA straggling for bare ions; dashed line, Chu values.86
Symbols: indicated in the figure; for H ions;68, 84, 85, 91, 92 for He ions;68, 79, 83 for Li and
B ions.79
Based on the interest in new general expressions for the energy loss
straggling, we used Lindhard’s scaling, which proved to be valid for stop-
ping by high Z-targets. In Figure 7.10 we plotted the square energy loss
straggling normalized to Bohr, as function of v 2 /ZT, including experi-
mental data for different ions (H to B) and targets (Cu, Au, Pb, and Bi).
The theoretical SLPA results for Au to Bi are actually very close (they are
hardly distinguishable in this figure). The SLPA results for Cu (dashed line)
separates slightly from the others. This scaling for the energy loss straggling
is an interesting proposal because it introduces the possibility of a simple
universal function to describe it.
4.1 The SLPA for stopping and straggling of gases

The SLPA evolved from models developed to deal with solid targets.
However there is no reason to restrict its use just to solid targets. The
response of bound electrons employed is based on the atomic wave func-
tions and binding energies, described in full Hartree–Fock or Hartree–
Fock–Dirac methods (the latter for targets with ZT > 54) .
In the previous sections we show the performance of the SLPA, with
good results for the description of the energy loss of ions in solid matter, is
Figure 7.10 Scaling for the squared straggling normalized to Bohr as function of

Lindhard parameter υ 2 /ZT. Curves: SLPA results for Au (solid line), Bi (dotted line) and
Cu (dashed line). Symbols: similar to Figures 7.8 for H and He in Cu and to 7.9 for differ-
ent ions in Au; additional data included84, 93–95 as signed inside the figure.
the intermediate to high energy region. In Figures 7.11 and 7.12 we display
the SLPA results for stopping and straggling of protons in four rare gases,
and compare them with experimental data available.
Note that for gaseous targets we should include not only the ionization
channel (as for solid Cu or Au, for which the valence electrons as FEG
fills the outer bands). For gases, the excitation channel is allowed. For the
rare gases we use the SLPA formulation given by Eq. (2) with energy gaps
equal to that to the first excited state (i.e., in atomic units, we use for Ne,
E3s = −0. 1809; for Ar, E4s = −0. 1666; for Kr, E5s = −0. 1598; for Xe,
E6s = −0. 1517). The energy gap is a sensitive point in the SLPA. Different
values for these excitation energies will change present results.
We display in Figures 7.11 and 7.12 the SLPA stopping and straggling
of rare gases for protons. The stopping description in Figure 7.11 is good
for proton impact above 300 keV in all the rare gases. Below this energy
the theoretical values underestimate the data, especially for Ar. These results
look quite different from those in metals. This is not surprising. In general
stopping power in metals is better known and described than in gases. For
example, for stopping of Ne, Ar, and Kr the very recent calculations by
Figure 7.11 Stopping cross section of rare gases for H+. The curves correspond to the
SLPA results. The experimental data is available in Ref. 46.
Grande and Schiwietz 65 with the UCA consider that the underestimation
is due to the contribution of projectile electron capture and loss. The SLPA
results displayed in Figure 7.11 are quite similar to CasP5.065 without loss.
To this stage and related only to the SLPA calculations, we consider as first
step to review the values for the excitation energy. Changes in this value
affects more to outer-than to inner-shells, so the effect will be noted in the
low to intermediate energy region.
In Figure 7.12 we display the SLPA results and the corresponding mea-
sured values for energy loss straggling of the four rare gases for protons. The
results are good, but for Xe they seem to underestimate the experimental data.
The comparison of this collective electron model with the other independent
electron formalisms, such as the UCA, the CDW-EIS,96, 97 or the continuum
distorted wave (CDW)98 methods would be a good framework for future
developments. Note that the CDW98 and CDW-EIS96, 97 methods use the
same scattering wave function for the final state in the exit channel.They dif-
fer in entrance channel, where the CDW method employs the full Coulomb
wave function for the electronic continuum state, which is in the CDW-EIS
method approximated by the associated asymptotic form given by the eikonal
logarithmic Coulomb phase. This difference yields the corresponding differ-
ence in the perturbation potentials in the CDW and CDW-EIS methods.
Figure 7.12 Straggling of rare gases for H+. The curves correspond to the SLPA results.
Experimental data by Besenbacher et al.68
The SLPA presents no specific problem to deal with the straggling and the
stopping power, its calculation is straightforward. On the c ontrary, numerical
difficulties are found to calculate the stopping power and straggling within
the CDW-EIS approach. These magnitudes involve the elements, first and
second moments of the energy, which depend on the behavior of very
energetic electrons. The CDW-EIS approximation has serious numerical
problems in two fronts. First, a good description of energetic electrons is
required, which demands not only a calculation of a substantial grid of
energies in this region (that would not contribute to the ionization cross
section). Second, these states require a large amount of angular momenta lmax
to describe the final continuum state.
To illustrate this point, Figure 7.13 shows the straggling and the stopping
power of protons in Ne as a function of the proton velocity. Three sets of
values are displayed: the results of CDW-EIS calculations with lmax = 4, 8,
and the exact value (lmax ≈ 28). This exact value agrees quite well with the
SLPA, as can be observed in Figure 7.13. As the proton velocity increases,
the straggling evidently requires a greater amount of angular momenta
which makes the calculation more lengthy and harder. In addition, inner-
shells become very important sources for the straggling. Although these
inner-shells hardly contribute to the ionization cross sections, or even
Figure 7.13 Stopping and straggling of H+ in Ne. The curves correspond to the SLPA
and CDW-EIS results, as marked in the figures. CDW-EIS values for different Lmax are
shown (see the text for details).
to the stopping power in our range of energy, they do contribute to the

energy loss straggling. While the outer-shells are enough to calculate the
cross section, for the straggling even deep shells are important. This makes
the CDW-EIS calculation highly demanding in terms of computing time.
This is a great advantage of the SLPA over distorted wave methods.
5. IONIZATION PROBABILITIES

As mentioned in the introduction, the energy moment of zero order,
or cross section, is a much more sensitive parameter for the theoretical
model than the stopping or straggling. In what follows we show the results
of this dielectric model to deal with the description of a fundamental col-
lisional process, the ionization.
5.1 Total ionization cross sections

The total ionization probabilities are calculated from Eq. (2) with t = 0. In
Figure 7.14 we display the SLPA results for total ionization cross section of
protons in four rare gases. We also include, as in the previous section, the
comparison with the CDW-EIS values. The experimental total ionization
cross sections include not only the direct ionization but also the post-
collisional ionization also. As proved from multiple ionization calculations
including Auger-type post-collisional contributions, the theoretical total
cross section must include this contribution too.
The total cross sections displayed in Figure 7.14 are calculated adding
the ionization of each sub-shell, and are afterwards corrected to include
post-collisional ionization. To this end we used the ratios between total
cross sections with and without post-collisional ionization in Ref. 99. In
fact, this contribution is negligible for Ne and Ar targets, so the total cross
sections are just the addition over the sub-shell contributions. But post-
collisional influence increases with the target atomic number. For Kr it goes
CDW-EIS CDW-EIS
CDW-EIS CDW-EIS
Figure 7.14 Total ionization cross sections of rare gases by proton impact. Curves: solid
lines, present SLPA results; dashed lines, CDW-EIS calculations.99 Symbols: experimen-
tal data by DuBois et al.100 and Cavalcanti et al.101 and recommended values by Rudd
et al.102 For Xe, we also included the electron impact data at high energies by Schram
et al.103 and by Nagy et al.104
from 5% at 300 keV to 24% at 5 MeV, and for Xe from 15% at 300 keV to

32% at 5 MeV (see Table 1 in Ref. 99). In Figure 7.14, for Xe target, we dis-
play both, direct ionization and total ionization including post-collisional
ionization. We also include in Figure 7.14 the multiple-ionization data for
electron in Xe in order to extend the description to high energies. The
experimental values in Xe show the importance of post-collisional ioniza-
tion in the total ionization.
The SLPA total ionization cross sections displayed in Figure 7.14 are
amazing. The SLPA is good in the same energy region that the CDW-EIS.
These values for ionization cross sections together with the stopping and
straggling comparison in Figures 7.11 and 7.12 establish the SLPA as a
good framework for further calculations.
These results consider the contribution of every shell of target electrons.
We have also test the model for ionization of certain shells, such as K-shell
of low Z elements21 or, recent calculations for L- and M-shells of Au, Pb,
and Bi,8 with good agreement with the experimental data.
5.2 Multiple ionization

Multiple ionization is a quite demanding calculation for any theoretical
model. It depends on the probabilities as function of the impact parameter.
The SLPA implies a space-mean value of the dielectric response as given
by Eq. (3). In principle the capability to describe total values may not be
valid for the differential description. Moreover, the SLPA depends directly
on the density of electrons of each sub-shell. It is tight to the zeros of the
density distribution.
In Figure 7.15 we display the SLPA results for multiple ionization of Kr
by high energy proton impact. In a many-electron description like this, the
multiple ionization follows a Poisson distribution (instead of the multino-
mial distribution of the independent electron models).
The high energy region is governed by the post-collisional ionization.27
The values displayed in Figure 7.15 take into account the post-collisional
ionization following the method described in Ref. 27. This method
includes the post-collisional electron emission in a semiempirical way by
employing branching ratios of ionization distribution measured in sophis-
ticated photo ionization experiments (see for example the values tabulated
in Ref. 106).
We also display in Figure 7.15 the theoretical values for direct mul-
tiple ionization, in order to make the importance of the post-collisional
Figure 7.15 Multiple ionization of Kr by proton impact. Curves: SLPA results with (solid
line) and without (dashed line) post-collisional ionization. Symbols: stars, DuBois et
al.;100 full up-triangles, Cavalcanti et al.;101 full down-triangles, Haugen et al.;105 open
symbols for high energy electron impact data, open squares Schram et al.;103 open
triangles, Nagy et al.104
contribution clearer. The direct multiple ionization almost describes the

single-ionization, but double-ionization cannot be explained only with
direct ionization above 1 MeV. For higher orders of ionization, the impor-
tance of the post-collisional contribution increases at even lower energies
than double-ionization, as can be noted for triple and quadruple ionization
of Kr. As observed in Figure 7.15, the agreement of the SLPA with the
experimental data is good. However some differences have been found for
other targets (mainly for Ne) that require further research (analyzes of the
branching ratios employed, other post-collisional contributions not consid-
ered, possibility of a semi-localized model, not so tied to the zeros of the
density of electrons, etc.).
5.3 Differential cross sections

Another test for the SLPA is the differential cross section as function of
the electron energy. In Figure 7.16 we display the SLPA results and com-
pare them with experimental data by Rudd 107, 108 and by Toburen.109
Two kinds of differential values are displayed: the differential cross sec-
tions dσ/dω, with ω the energy transferred to the emitted electron; and
Figure 7.16 SLPA Differential ionization cross sections as function of the energy of the
emitted electron for protons in Ar and Xe targets. Platzman plots for protons in Ne and
Kr targets. Experimental data in Refs. 107–109.
the Platzman plots Y, which is a ratio between dσ/dω and the modified
Rutherford differential cross section (dσ/dω)R by impact of one electron
with the same energy,110
dσ/dω v 2 (ω + I )2
Y = = dσ/dω, (14)
(dσ/dω)R 4π
where I is the binding energy of the outermost shell. The difference

between the modified Rutherford expression and the “original” one is
the presence of I. For the rare gases we used the values of Bunge et al.40
(I = 0.850, 0.591, 0.524, and 0.457 a.u. for Ne, Ar, Kr, and Xe, respectively).
The energy distributions present a scale involving several orders of mag-
nitudes and so the finer details could go unnoticed. An alternative way to
plot this is the Platzman plot, whose physical interpretation is the effective
number of electrons. The magnitude Y is a powerful tool for analyzing and
identifying the different features in the differential cross sections in fine
detail.110
In Figure 7.16 we present the theoretical Platzman plots for Ne and

Kr, while for Ar and Xe we display the differential cross sections dσ/dω.
The SLPA results displayed in this figure are actually overwhelming.
The distribution in electron energy describes the experimental data quite
well. The comparison with a much more sophisticated and computer-time
consuming calculation as the CDW-EIS confirms this evaluation (see this
comparison in Ref. 110).
6. CONCLUSIONS AND FUTURE PROSPECTS

In this chapter the properties, possibilities, and ranges of validity
of the SLPA have been presented, analyzed, and compared with other
theoretical descriptions and with experimental data. This comparison for
stopping power, energy loss straggling, and ionization cross sections, estab-
lishes this model as a trustworthy alternative to the independent electron
model calculations. In addition, the SLPA is a simple-low-time calculation,
and allows to correctly describe the inner-shells, which proved to be very
important in energy loss straggling calculations
The possibilities for further developments may start in three different
areas:
i. The extension of the model to complex targets: As far as the space depen-
dent density of electrons in the shell and energies are available the
SLPA calculation can be performed. This seems to be the next and
most important step, with a wide spectrum of possibilities to deal with
molecular, clusters, or even bio-structures.
ii. A delocalized approximation for multiple ionization: The SLPA describes
the response of the bound electrons as function of their density. In
this way the SLPA cancels where the density of electrons does. But
the interaction with the ion should include a region around each
point. This may be taken into account through a delocalized version
that does not change the total values, which is currently under study.
iii. Antiscreening: Finally the description of the inelastic processes of the
target with a dielectric function states a difference between projec-
tile and target. Extending the SLPA to describe inelastic processes
in the projectile should not be difficult, and would allow to extend
the model to deal with antiscreening processes (inelastic processes
in both centers111).
ACKNOWLEDGMENTS
This work was partially supported by the Consejo Nacional de Investigaciones
Científicas y Técnicas, and the Agencia Nacional de Promoción Científica y
Tecnológica of Argentina, as well as by the University of Buenos Aires.
REFERENCES
2. Belkić, Dž. Quantum Theory of High-Energy Ion-Atom Collisions; Taylor and Francis:
London, 2008.
3. Schardt, D.; Elsasser, T.; Schuls-Ertner, D. Heavy-ion tumor therapy: physical radiobio-
logical benefits. Rev. Mod. Phys. 2010, 82, 383.
4. Belkić, Dž. Reviews of theories on ionization in fast ion-atom collisions with pros-
pects for applications to hadron therapy. J. Math. Chem. 2010, 47, 1366.
5. Paganatti, H. Range uncertainties in proton therapy and the role of Monte Carlo
simulations. Phys. Med. Biol. 2012, 57, R99.
6. Paul, H. The Stopping Power of Matter for Positive Ions. In Modern Practices in
Radiation Therapy; Natanasabapathi, G., Ed.; InTech, 2012; p 113. <http://www.inte-
chopen.com/books/modern-practices-in-radiation-therapy>.
7. Montanari, C. C.; Archubi, C. D.; Mitnik, D. M., et al. Energy loss of protons in Au, Pb,
and Bi using relativistic wave functions. Phys. Rev. A 2009, 79, 032903.
8. Montanari, C. C.; Mitnik, D. M.; Miraglia, J. E. A collective model for inner shell ion-
ization of very heavy targets. Radiat. Eff. Defects Solids 2011, 166, 338.
9. Echenique, P. M.; Flores, F.; Ritchie, R. H. Dynamic screening of ions in condensed
matter. Solid State Phys. 1990, 43, 229.
10. Lindhard, J. On the properties of a gas of charged particles. Mat. Fys. Medd. Dan. Vid.
Selsk 1954, 28, 1.
11. Ritchie, R. H. Interaction of charged particles with a degenerate fermi-dirac electron
gas. Phys. Rev. 1959, 114, 644.
12. Lindhard, J.; Scharff, M. Energy loss in matter by fast particles of low charge. Mat. Fys.
Medd. Dan. Vid. Selsk 1953, 27, 1.
13. Bonderup, E. Stopping of swift protons evaluated from statistical atomic model. Mat.
Fys. Medd. Dan. Vid. Selsk 1967, 35, 1.
14. Rousseau, C. C.; Chu, W. K.; Powers, D. Calculations of stopping cross sections for
0.8- to 2.0-MeV alpha particles. Phys. Rev. A 1971, 4, 1066.
15. Chu, W. K.; Powers, D. Calculations of mean excitation energy for all elements. Rev.
Lett. A 1972, 40, 23.
16. Wang, Y.-N.; Ma, T.-C. Stopping power for hydrogen molecular ions in solids: influ-
ence of the inner-shell electrons of target atoms. Phys. Rev. A 1994, 50, 3192.
17. Fuhr, J. D.; Ponce, V. H.; García de Abajo, F. J., et al. Dynamic screening of fast ions
moving in solids. Phys. Rev. B 1998, 57, 9329.
18. Montanari, C. C.; Miraglia, J. E.; Arista, N. R. Dynamics of solid inner-shell electrons
in collisions with bare and dressed swift ions. Phys. Rev. A 2002, 66, 042902.
19. Montanari, C. C.; Miraglia, J. E.; Arista, N. R. Antiscreening mode of projectile-
electron loss. Phys. Rev. A 2003, 67, 062702.
20. Kadhane, U.; Montanari, C. C.; Tribedi, L. C. K-shell processes in heavy-ion collisions
in solids and the local plasma approximation. Phys. Rev. A 2003, 67, 032703.
21. Kadhane, U.; Montanari, C. C.; Tribedi, L. C. Experimental study of K-shell ionization
of low-Z solids in collisions with intermediate-velocity carbon ions and the local
plasma approximation. J. Phys. B 2003, 36, 3043.
22. Lantschner, G. H.; Eckardt, J. C.; Lifschitz, A. F. Energy loss of helium ions in zinc. Phys.
Rev. A 2004, 69, 062903.
23. Kadhane, U.; Kumar, A.; Montanari, C. C. K-shell ionization of low Z elements in ion–
solid collisions and applicability of local plasma approximation. Radiat. Phys. Chem.
2006, 75, 1542.
24. Montanari, C. C.; Miraglia, J. E. Stopping power for swift dressed ions. Phys. Rev. A
2006, 73, 024901.
25. Montanari, C. C.; Miraglia, J. E.; Heredia-Avalos, S. Calculation of energy-loss
straggling of C, Al, Si, and Cu for fast H, He, and Li ions. Phys. Rev. A 2007, 75,
022903.
26. Garcia, A. J.; Miraglia, J. E. Influence of surface channeling in the stopping of protons
colliding with LiF surfaces. Phys. Rev. A 2006, 74, 012902.
27. Archubi, C. D.; Montanari, C. C.; Miraglia, J. E. Many-electron model for multiple
ionization in atomic collisions. J. Phys. B: At. Mol. Opt. Phys. 2007, 40, 943.
28. Miraglia, J. E.; Gravielle, M. S. Ionization of the He, Ne, Ar, Kr, and Xe isoelectronic
series by proton impact. Phys. Rev. A 2008, 78, 052705.
29. Montanari, C. C.; Miraglia, J. E.; Behar, M., et al. Theoretical and experimental study
of energy loss of Li ions in Zn. Phys. Rev. A 2008, 77, 042901.
30. Cantero, E. D.; Fadanelli, R. C.; Montanari, C. C., et al. Experimental and theoretical
study of the energy loss of Be and B ions in Zn. Phys. Rev. A 2009, 79, 042904.
31. Montanari, C. C.; Miraglia, J. E. Theoretical Stopping Power of Copper for Protons Using the
Shellwise Local Plasma Approximation; Cornell University Library, 2009. Available from:
<arXiv:0904.1386v1> [physics.atom-ph].
32. Montanari, C. C.; Mitnik, D. M.; Archubi, C. D., et al. Energy loss of protons in W
using fully relativistic calculations and mean excitation energies of W, Au, Pb, and Bi.
Phys. Rev. A 2009, 80, 012901.
33. Cantero, E. D.; Montanari, C. C.; Behar, M., et al. Experimental and theoretical study
of the energy loss of C and O in Zn. Phys. Rev. A 2011, 84, 014902.
34. Meltzer, D. E.; Sabin, J. R.; Trickey, S. B. Calculation of mean excitation energy and
stopping cross section in the orbital local plasma approximation. Phys. Rev. A 1990,
41, 220.
35. Levine, Z. H.; Louie, S. G. New model dielectric function and exchange-correlation
potential for semiconductors and insulators. Phys. Rev. B 1982, 25, 6310.
36. Mermin, N. D. Lindhard dielectric function in the relaxation-time approximation.
Phys. Rev. B 1970, 1, 2362.
37. Palik, E. D., Ghosh, G., editors The Electronic Handbook of Optical Constants of Solids; CA
Academic: San Diego, 1999.
38. Isaacson, D. Compilation of rs Values; Radiation and Solid State Laboratory, New York
University: New York, 1975; Doc. No. 02698.
39. Clementti, E.; Roetti, C. Roothaan–Hartree–Fock atomic wavefunctions: basis func-
tions and their coefficients for ground and certain excited states of neutral and ionized
atoms, Z = 54. At. Data Nucl. Data Tables 1974, 14, 177.
40. Bunge, C. F.; Barrientos, J. A.; Bunge, A.V., et al. Hartree–Fock and Roothaan–Hartree–
Fock energies for the ground states of He through Xe. Phys. Rev. A 1992, 46, 3691.
41. Flannery, M. R.; Levi, H. Simple analytic expression for general two-center Coulomb
integrals. J. Chem. Phys. 1969, 50, 2938.
42. Paul, H. New developments in stopping power for fast ions. Nucl. Instrum. Methods
Phys. Res. B 2007, 261, 1176.
43. International Commission on Radiation Units and Measurements, ICRU Report 73,
Oxford University Press, Oxford, 2005.
44. ICRU Report 49, International Commission on Radiation Units and Measurements,
Bethesda, MD, 1993.
45. Paul, H. Stopping Power for Light Ions. Graphs, Data, Comments and Programs. <http://
www.exphys.uni-linz.ac.at/stopping>.
46. Ziegler, J. F. The Stopping and Range of Ions in Matter. <http://www.srim.org>.
47. Sigmund, P. Particle Penetration and Radiation Effects Springer Series in Solid-State
Physics; Springer: Berlin, 2006.
48. Arista, N. R. Energy loss of ions in solids: non-linear calculations for slow and swift
ions. Nucl. Instrum. Methods Phys. Res. B 2002, 195, 91.
49. Grande, P. L.; Schiwietz, G. Ionization and Energy Loss Beyond Perturbation Theory.
In , ; Cabrera-Trujillo, R., Sabin, J., Eds.; Advances in Quantum Chemistry; Elsevier:
New York, 2004; 45, pp 7.
50. Abril, I.; Garcia-Molina, R.; Denton, CD., et al. Dielectric description of wakes and
stopping powers in solids. Phys Rev. A 1998, 58, 357.
51. Balluc, N.; Abe, K.; Boutard, J. L., et al. Status of R&D activities on materials for fusion
power reactors. Nucl. Fusion 2007, 47, S696.
52. Oreg, J.; Goldstein, W. H.; Klapisch, M., et al. Autoionization and radiationless electron
capture in complex spectra. Phys. Rev. A 1991, 44, 1750.
53. Williams, G. P. Electron Binding Energies of the Elements; CRC Handbook of Chemistry
and Physics, 66th ed.; Electron Binding Energies, X-Ray Data Booklet, Lawrence
Berkeley Laboratory Pub-490 Rev., 1986; Vol. F170, Chapter 2.2. <http://www.jlab.
org/gwyn/ebindene.html>
54. Arista, N. R. Charge States and Energy Loss of Ions in Solids. In Ion Beam Science,
Solved and Unsolved Problems; Sigmund, P., Ed.; Mat. Fis. Medd. 2006, 52, 595.
55. Müller, J.; Bürgdorfer, J. Dynamical thresholds for the existence of excited electronic
states of fast ions in solids. Phys. Rev. A 1991, 43, 6027.
56. Paul, H. Recent results in stopping power for positive ions, some critical comments.
Nucl. Instrum. Methods Phys. Res. B 2010, 268, 3421.
57. Martinez-Tamayo, G.; Eckardt, J. C.; Lantschner, G. H. Energy loss of H+ and He+
in Al, Zn, and Au in the very low- to intermediate-energy range. Phys. Rev. A 1996,
54, 3131.
58. Heredia-Avalos, S.; Abril, I.; Denton, C. D. J. Phys.: Condens. Matter. 2007, 19,
466205.
59. Kreussler, S.; Varelas, C.; Sizmann, R. Electronic stopping power and effective charge
of 50- to 230-keV D+ and He+ in carbon, aluminum, gold, and cesium. Phys. Rev. B
1982, 26, 6099.
60. Santry, D. C.; Werner, R. D. Stopping powers of C, Al, Si, Ti, Ni, Ag and Au for deu-
terons. Nucl. Instrum. Methods Phys. Res. B 1981, 188, 211.
61. Bethe, T. Zur theory des durchgangs schneller korpuskular-strahlen durch materie.
Ann. Phys. 1930, 5, 325.
62. Kamakura, S.; Sakamoto, N.; Ogawa, H., et al. Mean excitation energies for the stop-
ping power of atoms and molecules evaluated from oscillator-strength spectra. J. Appl.
Phys. 2006, 100, 064905.
63. Lindhard, J.; Scharff, M. Stopping power of heavier substances. Phys. Rev. 1952,
85, 1058.
64. Schiwietz, G.; Grande, P. L. Improved charge-state formulas. Nucl. Instrum. Methods
Phys. Res. B 2001, 125, 175–177.
65. Grande, P. L.; Schiwietz, G. Stopping of protons. Improved accuracy of the UCA
model. Nucl. Instrum. Methods Phys. Res. B 2012, 273, 1.
66. Lifschitz, A. F.; Arista, N. R. Electronic energy loss of helium ions in aluminum using
the extended-sum-rule method. Phys. Rev. A 1998, 58, 2168.
67. Arista, N. R.; Lifschitz, A. F. Non-Linear Approach to the Energy Loss of Ions in Solids.
In , ; Cabrera-Trujillo, R., Sabin, J., Eds.; Advances in Quantum Chemistry; Elsevier
Inc., pp 47.
68. Besenbacher, F.; Andersen, J. U.; Bonderup, E. Straggling in energy loss of energetic
hydrogen and helium ions. Nucl. Instrum. Methods Phys. Res. B 1980, 168, 1.
69. Bohr, N. Penetration of atomic particles through matter. K. Dan. Vidensk. Selsk. Mat.
Fys, Medd. 1948, 24, 19.
70. Ma, L.; Wang,Y.; Xue, J., et al. Energy loss and straggling of MeV ions through biologi-
cal samples. J. Appl. Phys. 2007, 102, 084702.
71. Mayer, M. SIMNRA simulation program for the analysis of NRA, RBS and ERDA.
AIP Conf. Proc 1999, 475, 541.
72. Heredia-Avalos, S.; Garcia-Molina, R. Energy-loss of swift Cn+ (n = 2 − 60) clusters
through thin foils. Phys. Rev. A 2007, 76, 012901.
73. Garcia-Molina, R.; Abril, I.; Heredia-Avalos, S., et al. A combined molecular dynamics
and Monte Carlo simulation of the spatial distribution of energy deposition by proton
beams in liquid water. Phys. Med. Biol. 2011, 56, 6475.
74. Yang, Q.; O’Connor, D. J.; Wang, Z. Empirical formulae for energy loss straggling of
ions in matter. Nucl. Instrum. Methods Phys. Res. B 1991, 61, 49.
75. Barradas, N. P.; Rauhala, E. Data Analysis Software for Ion Beam Analysis. In Handbook
of Modern Ion Beam Materials Analysis, ; Wang, Y., Nastasi, M., Eds.;; MRS: Warrendale,
Pennsylvania, 2009., pp 307.
76. Msimanga, M.; Pineda-Vargas, C. A. .; Comrie, C. M. Heavy ion energy loss straggling
data from time of flight stopping force measurements. Nucl. Instrum. Methods Phys. Res.
B 2012, 273, 6.
77. Moon, D. W.; Le, H. I.; Kim, K. J., et al. Estimation of the electronic straggling using
delta-doped multilayers. Nucl. Instrum. Methods Phys. Res. B 2001, 183, 10.
78. Eckardt, J. C.; Lantschner, G. H. Determination of thin foil thickness inhomogeneities
by analysis of ion beam energy loss spectra. Thin Solid Films 1994, 249, 11.
79. Hsu, J. Y.; Yu, Y. C.; Liang, J. H. et al. Energy loss of He, Li and B isotopes with MeV
energies in Au. Nucl. Instrum. Methods Phys. Res. B 2004, 219–220, 251.
80. Kido, Y. Energy straggling for fast proton beams passing through solid materials. Nucl.
Instrum. Methods Phys. Res. B 1987, 347, 24–25.
81. Kido, Y.; Koshikawa, T. Energy straggling for medium-energy H+ beams penetrating
Cu, Ag, and Pt. Phys. Rev. A 1991, 44, 1759.
82. Kawano, A.; Kido,Y. Effects of energy straggling on surface analysis with fast ion beams.
J. Appl. Phys. 1988, 63, 75.
83. Amadon, S.; Lanford, W. A. He stopping power and straggle in Al, Ti, Co, Cu, Ag, Ta,
and Au from 1.5 to 4 MeV. Nucl. Instrum. Methods Phys. Res. B 2006, 249, 34.
84. Eckardt, J. C.; Lantschner, G. H. Experimental energy straggling of protons in thin solid
foils. Nucl. Instrum. Methods Phys. Res. B 2001, 93, 175–177.
85. Alberts, H. W.; Malherbe, J. B. Energy loss and straggling of p-, d- and a-particles in Au
in the energy region 0.2 to 2.4 MeV. Radiat. Eff. 1983, 69, 231.
86. Chu, W. K. Calculation of energy straggling for protons and helium ions. Phys. Rev. A
1976, 13, 2057.
87. Hoffman, G. E.; Powers, D. Energy straggling of a particles in solid materials. Phys. Rev.
A 1976, 13, 2042.
88. Friedland, E.; Kotze, C. P. Energy-loss straggling of protons, deuterons and a-particles
in copper. Nucl. Instrum. Methods Phys. Res. 1981, 191, 490.
89. Livingston, M. S.; Bethe, H. A. Nuclear dynamics, experimental. Rev. Mod. Phys. 1937,
9, 245.
90. Sigmund, P.; Schinner, A. Barkas effect, shell correction, screening and correlation in
collisional energy-loss straggling of an ion beam. Eur. Phys. J. D 2003, 23, 201.
91. Andersen, H. H.; Csete, A.; Ichioka, T., et al. An apparatus to measure stopping pow-
ers for low-energy antiprotons and protons. Nucl. Instrum. Methods Phys. Res. B 2002,
194, 217.
92. Møller, S. P.; Csete, A.; Ichioka, T., et al. Antiproton and proton energy loss straggling
at keV energies. Eur. Phys. J. D 2008, 46, 89.
93. Andersen, H. H.; Besenbacher, F.; Goddiksen, P. Stopping power and straggling of
80–500 keV lithium ions in C, Al, Ni, Cu, Se, Ag, and Te. Nucl. Instrum. Methods Phys.
Res. 1980, 168, 75.
94. Eckardt, J. C. Energy loss and straggling of protons and helium ions traversing some
thin solid foils. Phys. Rev. A 1978, 18, 426.
95. Malherbe, J. B.; Alberts, H. W. Energy-loss straggling in lead of p, d and a-particles in
the energy region 0.2–2.5 MeV. Nucl. Instrum. Methods Phys. Res. B 1982, 196, 499.
96. Crothers, D. S. F.; McCann, J. F. J. Phys. B: At. Mol. Opt. Phys. 1983, 16, 3229.
97. Fainstein, P. D.; Ponce,V. H.; Rivarola, R. D. A theoretical model for ionisation in ion-
atom collisions. Application for the impact of multicharged projectiles on helium. J.
Phys. B: At. Mol. Opt. Phys. 1988, 21, 287.
98. Belkić, Dž. A quantum theory of ionization in fast collisions between ions and atomic
systems. J. Phys. B 1978, 20, 3529.
99. Montanari, C. C.; Miraglia, J. E. Antiproton, proton and electron impact multiple ion-
ization of rare gases. J. Phys. B: At. Mol. Opt. Phys. 2012, 45, 105201.
100. DuBois, R. D.; Toburen, L. H.; Rudd, M. E. Multiple ionization of rare gases by H+
and He+ impact. Phys. Rev. A 1984, 29, 70.
101. Cavalcanti, E. G.; Sigaud, G. M.; Montenegro, E. C. Absolute cross sections for multiple
ionization of noble gases by swift proton impact. J. Phys. B: At. Mol. Opt. Phys. 2003,
36, 3087.
102. Rudd, M. E.; Kim,Y-K.; Madison, D. H.; Gallagher, J.W. Electron production in proton
collisions: total cross sections. Rev. Mod. Phys. 1985, 57, 965.
103. Schram, B. L.; Boerboom, A. J. H.; Kistermaker, J. Partial ionization cross sections
of noble gases for electrons with energy 0.5–16 keV: I. Helium and neon. Physica
1966, 32, 185.Schram, B. L. Partial ionization cross sections of noble gases for
electrons with energy 0.5–18 keV: II. Argon, Krypton and Xenon. Physica 1966,
32, 197.
104. Nagy, P.; Skutlartz, A.; Schmidt,V., et al. Absolute ionisation cross sections for electron
impact in rare gases. J. Phys. B: At. Mol. Opt. Phys. 1980, 13, 1249.
105. Haugen, H. K.; Andersen, L. H.; Hvelplund, P., et al. Multiple ionization of noble gases
by fully stripped ions. Phys. Rev. A 1982, 26, 1962.
106. Montanari, C. C.; Montenegro, E. C.; Miraglia, J. E. CDW-EIS calculations for mul-
tiple ionization of Ne, Ar, Kr and Xe by the impact of H+ and He+, including post-
collisional electron emission. J. Phys. B: At. Mol. Opt. Phys. 2010, 43, 165201.
107. Rudd, M. E.; Kim, Y.-K.; Madison, D. H., et al. Electron production in proton
collisions with atoms and molecules: energy distributions. Rev. Mod. Phys. 1992,
64, 441.
108. Rudd, M. E.; Toburen, L. H.; Stolterfoht, N. Differential cross sections for ejection of
electrons from argon by protons. At. Data Nucl. Data Tables 1979, 23, 405.
109. Toburen, L. H. Distribution in energy and angle of electrons ejected from xenon by
0.3- to 2.0-MeV protons. Phys. Rev. A 1974, 9, 2505.
110. Miraglia, J. E. Ionization of He, Ne, Ar, Kr, and Xe by proton impact: single differential
distributions in energy and angles. Phys. Rev. A 2009, 79, 022708.
111. Montenegro, E. C.; Meyerhof, W. E. Sum rules and electron–electron interaction in
two-center scattering. Phys. Rev. A 1991, 43, 2289.
CHAPTER EIGHT
Single Ionization of Liquid Water

by Protons, Alpha Particles, and
Carbon Nuclei: Comparative
Analysis of the Continuum
Distorted Wave Methodologies
and Empirical Models
Mario A. Bernal-Rodrígueza and Jacinto A. Liendob
aInstituto de Física Gleb Wataghin, Universidade Estadual de Campinas, Brazil
bDepartmento de Física, Universidad Simón Bolívar, Caracas, Venezuela
Contents
1. Introduction 204
1.1 The liquid water in radiation physics 205
1.2 The ionization problem 206
1.2.1 Brief description 206
1.2.2 Liquid water phase effects 208
2. Theoretical approaches 208
2.1 The first Born approximation (B1) 210
2.2 The distorted wave formalism and its main variants 213
2.2.1 The continuum-distorted wave approximation (CDW) 213
2.2.2 The continuum distorted wave eikonal initial state approximation (CDW-EIS) 213
3. Experimental works 214
4. Semiempirical methods 215
4.1 The Rudd model 215
4.2 The HKS method 216
4.3 Formalism based on the dielectric response function 216
5. Comparison between experimental, theoretical, and semiempirical results 217
5.1 Ionization cross sections for water 218
6. Conclusions and perspectives 226
Acknowledgments 227
References 227
Abstract
Single ionization of liquid water by the impact of fast, but nonrelativistic heavy
charged particles is reviewed. Special attention is focused on protons, alpha particles,

ISSN 0065-3276, http://dx.doi.org/10.1016/B978-0-12-396455-7.00008-X
204 Mario A. Bernal-Rodríguez and Jacinto A. Liendo
and carbon ions. This phenomenon has been extensively studied by using theo-
retical methods during the last decades. Quantum-mechanical as well as semiclassical
approaches have been developed. Nevertheless, experimental studies in this field
are very scarce. Based upon both theoretical and experimental results, semiempirical
formalisms have been reported. At the beginning, the first Born (B1) approximation
emerged with some success in reproducing ionization cross sections corresponding
to impact energies above a few hundreds of keV/u. The introduction of the distorted-
wave formalism brought a remarkable improvement with respect to B1, mainly
because of reproduction of the well-known two-center effects. The B1 approxima-
tion is treated here in order to be used as a reference model for all the subsequent
comparisons. Also presented are the distorted-wave formalism and its variants. On
the other hand, available experimental works for measuring ionization cross sections
corresponding to water vapor targeted by hydrogen, helium, and carbon ions are
summarized. In addition, the most relevant semiempirical approaches intended to cal-
culate water ionization cross sections are addressed. Some of the experimental single
ionization cross sections mentioned above are compared to those determined by the
distorted-wave and semiempirical formalisms for liquid water. Finally, perspectives on
the studies of the ionization problem are briefly commented.
1. INTRODUCTION
This chapter addresses the single ionization of water due to the
impact of fast but not relativistic light ions (Z ∼ 1 − 6). The main interest
is focused on protons, alpha particles, and carbon ions with energies from
a few tens of keV/u to about 10 MeV/u. The physical process under con-
sideration has a great importance in areas such as medical radiation phys-
ics, space physics, plasma physics, material sciences, and others. Ionization
cross sections (CS) are of critical importance in Monte Carlo codes used
to simulate the passage of ionizing radiation through matter. In this work,
the ionization problem is first explained from a phenomenological point
of view. Later on, theoretical approaches to determine ionization CS are
briefly reviewed: the first Born approximation and those based on the
continuum-distorted wave approximation. Afterwards, a review of experi-
mental works carried out to determine ionization CS in water vapor due
to the impact of light ions is presented. In addition, a few semiempiri-
cal methods to calculate these CS are treated. Finally, the theoretical and
semiempirical methods expounded here are compared with corresponding
experimental results.
Atomic units are used throughout this chapter, except otherwise
stated. In this system, the mass is expressed in units of the electron
Single Ionization of Liquid Water by Protons, Alpha Particles, and Carbon Nuclei 205
rest mass m; the length, in units of the Bohr radius a0 = 2 /me 2; the
energy, in units of Hartrees (1 Hartree = me 4 /2 = 2 Ry = 27. 2 eV); the
momentum, in units of /a0 = me 2 /; and the speed, in units of e 2 /
(e = m = = 1).
1.1 The liquid water in radiation physics

It is well known that water is used as the main reference medium to study
interactions of ionizing radiation with tissue, not only experimentally but
also theoretically. Human body is composed of about 70% of water and,
therefore, the procedure of considering water molecules as the approximate
tissue-like targets seems to be reasonably well justified.
Ionizing radiations affect the genetic material when impacting on living
beings. The DNA molecule can be damaged through direct and indirect
processes. In the former, ionizing particles impact this macromolecule caus-
ing single, double, or multiple strand breaks. In the latter, chemical species
created after the interaction of radiation with the medium will react with
DNA structures. The nanodosimetry is the branch of physics that studies
the energy distribution at nanometric scale, which is a very important mat-
ter in investigations where the radiation induced DNA damage is simulated
numerically.
It is well known that light ions are more efficient to produce biologi-
cal damage than electrons and photons, mainly because ions have higher
linear energy transfer (LET). However, the LET does not determine the
radiobiological effectiveness (RBE) of the radiation.1 It is very important
to accurately know the RBE corresponding to each ion beam, specially for
those beams used in the radiation therapy of cancer.
Many secondary particles are produced during the passage of fast ions
through matter, most of them are electrons released by ionization of tar-
geted atoms or molecules. These very light charged particles are responsible
for the deposition of the greatest fraction of incident energy no matter
which the primary particles are. Consequently, electrons have to be taken
into account in any application related to nanodosimetry and/or radiobiol-
ogy. For these reasons, among others, the knowledge of ion-related ioniza-
tion CS are extremely important.
The water molecule has 10 bound electrons. Its mass and molecular
densities are about 1. 0 g/cm3 and 3. 34 × 1022 l/cm3, respectively, in the
liquid phase.Table 8.1 shows electron binding energy and occupation num-
ber corresponding to the five molecular orbitals of liquid and gaseous water.
Table 8.1 Electron occupation numbers and binding energies for liquid and gaseous
water subshells
Shells N Ialiquid (eV) Ibvapor (eV)
1b1 2 10.79 12.61
3a1 2 13.39 14.73
1b2 2 16.05 18.55
2a1 2 32.30 32.2
1a1 2 539.0 539.7
a Values from Dingfelder et al.2
b Values from Rudd et al.3
The four most outer shells have energies up to a few tens of eV and they
contribute to the total ionization cross section much more than the inner
shell, which corresponds to the oxygen K-shell.
Unfortunately, no experimental determination of ionization CS has
been carried out in liquid water, mainly due to the difficulty of having an
adequate liquid target and problems associated to the detection of very low
energy electrons. In a forthcoming section, all experiments, to the best of
our knowledge, carried out to determine ionization CS due to the impact
of light ions on water vapor are summarized (see Section 3).
1.2 The ionization problem

1.2.1 Brief description
When a totally stripped ion impacts a molecular or atomic target, it
interacts with bound electrons through the Coulomb field. The magnetic
component of the interaction potential is negligible at the incident veloci-
ties considered here. In this situation, a bound electron can be released
to the continuum through the so-called direct ionization phenomenon.
Otherwise, the electron can be excited to a higher energy state or captured
into a projectile bound state. These mechanisms are known as electronic
excitation and electronic capture, respectively. It should be remarked that
in the latter process, the target is also left ionized as in the direct ionization
case. Hereafter, direct ionizations will be simply referred to as ionizations.
The main difficulty in the theoretical treatment of the ionization prob-
lem is the representation of the initial and final system states. Hydrogen-
type wave functions have been a common choice used to describe the
initial electron bound state,4 even when the target is not hydrogen.5–7
The hydrogen atom has the particularity that can be solved analytically by
using quantum mechanics. In the exit channel, the electron moves under
the influence of two Coulomb potentials (long range), due to its interac-
tion with the projectile and the residual target. An exact solution for the
final electron wave function in the ion–atom ionization problem has not
been found yet, but the continuum distorted wave (CDW) function has
been derived.8 The CDW8 and the CDW-type4 approaches included in the
present work satisfy asymptotic conditions both at the entrance and exit
channels, but the first Born approximation (B1) does not.
Our efforts are focused on the single ionization mechanism described
by reaction (1), although references will be made to the electron loss and
electron capture processes. Double ionizations are relatively important
at very low energies and for highly charged projectiles, but they are not
treated in this work. Although the electron capture phenomenon is out of
the scope of our work, it is marginally mentioned due to its relation with
ionizations. In fact, these mechanisms are extremely competitive at impact
energies below about 100 keV/u. More complex processes can take place
when the incident ion carries one or more bound electrons:
PZP + + T(ZT −1)+ → PZP + + TZT + + e − . (1)
Based on the fact that the projectile mass is much greater than that of the
electron, the incident particle loses a small fraction of its original energy
during a collision with the electron. When an impact between an ion and
an electron, assumed initially at rest, is frontal, the energy transferred to
the electron is maximum and equal to 2v 2, and the maximum fractional
energy loss is given by �E/E = 4/M where E, v, and M are the projectile
incident energy, speed, and mass, respectively. In addition, the scattering of
one nucleus by the other at impact energies above a few tens of keV/u
is restricted to angles much lower than 1°. Furthermore, if the de Broglie
wave length of the relative motion of the nuclei is much less than the tar-
get characteristic dimension, a trajectory can be assigned to the projectile.
All said in this paragraph supports the idea that these energetic and rela-
tively massive particles travel along a straight path. This argument is used
by impact-parameter-based approximations (see Section 2). Under these
conditions, a straight-line trajectory can be associated to the projectile, so
that the transferred momentum during the collision q is minimum. Then,
neglecting terms of order (1/M )2 and higher,
�E
qmin = Ki − Kf = (2ME)1/2 [1 − (1 − �E/E)1/2 ] ≈ , (2)
v
where Ki (Kf ) and E are the initial(final) projectile momentum and

energy loss, respectively.
1.2.2 Liquid water phase effects

The water aggregation state influences the cross sections corresponding to
the interaction of charged particles with this medium. One of the effects is
the decrement of the electron binding energies in liquid water relative to
water vapor due to the intermolecular interaction present in the liquid water
phase (see Table 8.1). Note that outer orbitals are the most affected while
the inner ones are practically independent of the aggregation state. Another
important effect is that described by Enrico Fermi many decades ago. For a
given traversed medium mass thickness, he explained the reduction of the
expected energy loss of a primary particle in a condensed medium relative
to a rarefied one. Let us briefly describe this phenomenon by a classical
approach. The quantity qmin is related to the maximum impact parameter
ρmax (ρmax ∼ 1/qmin). In condensed media and for sufficiently high ion
velocities, ρmax is much greater than the mean inter-target distance (mol-
ecules or atoms), so that the projectile interacts with many targets simul-
taneously producing electric polarization in the medium. This polarization
screens the electric field generated by the projectile on the target, the inter-
action potential weakens and the energy lost by the projectile per unit mass
path length decreases. The Fermi density effect is important at high impact
velocities, where ρmax is large, and for low emitted electron energies, where
ρmax is also large. For energies (velocities) up to about 10 MeV/u and the
projectile charges analyzed in this work, the Fermi-density correction can
be safely disregarded. There are other phase effects such as the controversial
collective excitations occurring in liquid water but they are not related to
ionization cross sections so that they are not treated in the current work.
2. THEORETICAL APPROACHES

The purpose of this section is to give an overall description of the
ionization process that will help the reader to better understand the experi-
mental results shown below (see Section 3). Mathematical details on the
approaches presented are not included in this work as they can be found in
references given in this section.
Let us suppose the a totally stripped ion with charge, mass, velocity, and
momentum along the z-axis ZP , MP , v, and K, respectively, impacts on a 1s
hydrogen-like target with nuclear charge and mass ZT and MT, respectively.
In this case MP , MT ≫ 1 so that the electronic states can be considered

as independent of the relative motion of the involved nuclei (Born–
Oppenheimer separation, see Chapters 1 and 4 of Ref. 9). The target recoil
is neglected since the energy transferred to the electron is much greater than
that imparted to the target nucleus, even in proton-hydrogen collisions. In
addition, Ka ≫ 1, where a is the characteristic target dimension, so that a
straight-line trajectory defined by ρ · v = 0 can be assigned to the projectile,
where ρ is the 2D impact parameter vector. Under these conditions, the pro-
jectile position with respect to the target will always be R = ρ + vt.
For the reasons given just above, the theoretical approaches treated here
are based on the straight-line impact parameter model (IPM),9 which is
considered semiclassical due to the assigned projectile trajectory. The pro-
jectile–electron interaction is time-dependent and so is the perturbation
theory used. Coupling with electron capture and excitation channels is not
considered.The theoretical models addressed in this chapter describe the
process shown in reaction (1). Consequently, we define
x = re − rT , s = re − rP ,
R = rP − rT , r = 1/2(x + s),
where re , rT, and rP are the position vectors of the electron, target nucleus
and projectile, respectively measured with respect to the laboratory frame
of reference. Figure 8.1 depicts relevant vectors used for clarity purposes.
Schrödinger’s equation for reaction (1) is

∂ (3)
He − i � = 0,
∂tr
Figure 8.1 Relevant vectors used in theoretical models.

where
1 Z T Z P Z T ZP
He = − ∇r2 − − + (4)
2 x s R
is the electronic Hamiltonian, and (r, tr ) are the independent variables. It
is important to notice that ionization cross sections do not depend on the
chosen origin O if this point belongs to the line joining the target and
the projectile, and the resulting potential due to the projectile vanishes as
R → ∞ (Ref. 9).
2.1 The first Born approximation (B1)

In this approximation, the interaction potential between the projectile
and the target electron is treated as a first-order perturbation. This is jus-
tified if this potential satisfies the condition |U | ≪ v/d, where d is its
action radius.11 Taking U ∼ ZP /d, this approximation can be written as
ZP /v ≪ 1. The inter-nuclear potential is discarded since it only introduces
a phase factor in the transition amplitude, producing no change in the ion-
ization differential cross section with respect to the electron parameters.4, 8
In this approximation, the initial state is described by a stationary bound
electron wave function modified by translational phases depending on v
due to the coordinate transformation existing between frames of reference
centered at the positions of the target and point O. In the exit channel,
the movement of the electron with respect to the target is represented by a
Coulomb distorted plane wave plus translational factors. According to the
B1 approximation, the presence of the projectile causes no perturbation
on the electronic wave functions corresponding to both the entrance and
exit channels. Mathematical details on this approximation can be found
elsewhere (see Chapter 7, Section 2 of Ref. 9). After the corresponding
manipulations, for the ionization of a 1s hydrogen-like target, the B1
transition probability is given by8
7 5 2
(B1) 2
2 ZT ZP
Rif (η) = |N (ξ )|2
π 2 v 2 q4

2ZT k
(5)
−1
× exp −2ξ tan 2
q 2 + ZT − k 2
[(q2 + q · k)2 + (q · k)2 ξ 2 ]
× ,
[ZT2 + (q + k)2 ][ZT2 + (q − k)2 ][ZT2 + (q + k)2 ]4
with E
q = −η − v̂,
v (6)
where η is the transverse momentum transfer, k is the ejected electron

momentum and ξ = ZT /k.
This transition probability shows a local maximum at high electron
energies. The leading term, shown between brackets with exponent
4, has a minimum when (q + k) → 0. Based on the fact that a large
amount of moment is transferred during a head-on collision, the initial
electron momentum can be safely neglected so that q + k = 0. In this
case, the final electron momentum is equal to the momentum transfer.
Combining this equation with Eq. (6) and having into account that
η · v = 0, we have
E
k · v̂ − = 0. (7)
v
This equation defines the binary sphere,12 which describes the kinemat-
ics of a binary projectile–electron encounter. The double differential
cross sections (DDCS) show a maximum at electron momenta satisfying
Eq. (7), which represents the binary encounter (BE) peak. Furthermore,
�E = εf − εi = 21 k2 − εi, so that the solutions for the final electron
momentum in this sphere, up to first order, are
1/2
+ 2εi εi
kBE = v cos(θ ) 1 + 1 + 2 2 ≈ 2v cos θ + , (8)
v cos θ v cos θ
1/2
− 2εi εi
kBE = v cos(θ ) 1 − 1 + 2 2 ≈− ,
v cos θ v cos θ (9)
where θ is the electron ejection angle with respect to the incident particle
direction.
Firstly, the BE peak is observed in experimental DDCS (e.g., see
+
Ref. 16) for electron momenta close to kBE. When the initial elec-
+
tron energy is disregarded, the result for kBE is the one obtained for a
binary encounter of a heavy particle (MP ≫ 1) with an electron at rest
(kBE = 2v cos θ). Secondly, it can be concluded, from the analysis of Eq.
(5), that the maximum DDCS is obtained at k ≈ 0, which conforms the
−
“soft collision” (SC) region. The kBE solution is usually close to zero and
the corresponding peak is overlapped by the SC region, therefore it is
commonly unspotted.
Summarizing, the B1 approximation treats the projectile–electron inter-
action as a first-order perturbation, condition that is not fulfilled in the
following situations: (i) the projectile moves with velocities of the order of
those of the bound electrons, (ii) the projectile is very close to the electron,
for example when k ≈ v, and (iii) the projectile charge is large. According
to the transition amplitude (5), differential and total ionization cross sec-
tions scale as (ZP /v)2. Moreover, B1 cross sections depend on the projectile
velocity, instead of its energy.
Hansen and Kocbach5 used a variant of the B1 approximation to study
electron emission after K-shell ionization due to protons. In fact, they fol-
lowed an impact parameter plane-wave formulation of this approximation
(IP-PWBA). Furthermore, they described the initial and final electron
states through 1s hydrogen-type and plane wave functions, respectively. In
this case, the transition amplitude is
−i
+∞
aifHK = dz exp(iqz)��f (x)|V (s)|�i (x)�, (10)
v −∞
where q is the minimum momentum transfer (see Eq. (2)), �i (x) and �f (x) are
the initial and final electron wave functions andV (s) is the Coulomb projectile–
electron potential. If the final wave function �f (x) = (2π )−3/2 exp(−ik · x)
is substituted in Eq. (10) and the initial wave function is represented by a 1s
hydrogen-like wave function (�i (x) = π −1/2 p3/2 exp(−px)), then by using
the Bethe integral, the peaking approximation and neglecting the initial
electron momentum with respect to the final one, the integral shown in Eq.
(10) can be solved. Thus, the Hansen–Kocbach (HK) transition amplitude is
given by
−i23/2 ZP p5/2 exp(−iρ · k⊥ ) ρK1 (ρ[p2 + (q − k� )2 ]1/2 )

aifHK = , (11)
π vk2 [p2 + (q − k� )2 ]1/2
where k� = k · v̂, k⊥ is the electron momentum normal to v and K1 is the

first-order modified Bessel function of second kind.
Notice that this transition amplitude is maximum when
k · v̂ − �E/v = 0, which reproduces the binary encounter sphere
described above (see Eqs. (7) and (2)). Based on this theoretical approach,
Stolterfoht developed a semiempirical method to determine ionization
cross sections for the impact of light ions on matter.13, 14 This is the so-
called HKS model which is treated in Section 4.
2.2 The distorted wave formalism and its main variants

The CDW formalism was introduced by Cheshire10 to study the resonant
charge transfer in proton-hydrogen collisions. Subsequently, this methodol-
ogy was extended by Belkić8 to investigate the ionization problem when
ions impact on atomic systems. Afterwards, working within the CDW
theory,8 Crothers and McCann4 introduced a modification in the nor-
malization of the initial electron state and replaced the perturbation in the
entrance channel by an eikonal phase (CDW-EIS).
2.2.1 The continuum-distorted wave approximation (CDW)

In this approximation, the Hamiltonian shown in Eq. (4), for both entrance
and exit channels, is separated as follows:8
He = H0 + Ui + Wi = H0 + Uf + Wf . (12)
The perturbation potentials Wi and Wf are chosen such that the Schrödinger
equations
∂
H0 + Ui,f − i
(CDW)±
χi,f =0 (13)
∂tr
can be solved exactly.The unperturbed parts of the initial and final distorted
(CDW)±
wave functions χi,f are written in terms of the B1 wave functions in
the entrance and exit channels, respectively. In the CDW approximation,
the perturbation potentials are given by the scalar products of two gradient
operators applied to the pertinent electronic wave functions of the bound
and continuum states.
2.2.2 The continuum distorted wave eikonal initial state

approximation (CDW-EIS)
In this case, the exit channel is treated as in the pure CDW approximation,
but the wave function distortion in the entrance channel is accounted for
through an eikonal phase.4 In addition to the mentioned gradient operators
in the perturbation potential from the transition amplitude, this phase for
the eikonal initial state yields an extra term given by the electronic kinetic
energy operator. In the CDW-EIS approximation, the final CDW states
corresponding to different k are regarded as orthonormal and uncoupled,
which is not an exact assumption because they are not the eigenfunctions
of the total Hamiltonian. Moreover, the initial and final states are not
orthogonal since their corresponding Hamiltonians are different. However,
this is not an important problem for ionization as it is for charge transfer.
Note that the distorted wave Born (DWB) approximation has also been
obtained8 as an alternative hybrid model when the B1 and the CDW wave
functions are used in the entrance and exit channels, respectively.
The expression for the triple differential cross sections (TDCS) obtained
by means of the CDW and CDW-EIS approximations can be found in Refs.
8 and 4, respectively. Some important results can be extracted from that
expression: (a) the CDW-EIS cross sections depend directly on the transi-
tion amplitude obtained from the B1 approximation and the other terms
represent the effect of the distortion produced by the projectile (which is
neglected within the B1 approach). As the projectile speed increases, the
TDCS in the CDW and CDW-EIS approximations tend to the one obtained
in the B1 case, (b) the BE peak is reproduced once again, and (c) the factor
2π ζ
|N (ζ )|2 = (14)
1 − e −2π ζ
with ζ = ZP /p, is maximized as p → 0, where p is the electron momentum
in the projectile reference system. This means that when the final electron
momentum relative to that of the projectile is close to zero (k ≈ v), a peak
is produced in the TDCS as well as in the DDCS. This is the so-called
electron capture into the continuum (ECC) peak, which is one of the two-
center effects (see Section 5). This factor represents the density of electron
continuum states around the projectile.
3. EXPERIMENTAL WORKS

Experiments for the determination of ionization cross sections due to
the impact of ions on water are very scarce. All of them have been carried
out by using water vapor as a target. Many research areas, such as nanodosim-
etry and radiobiology require of the knowledge of liquid water target cross
sections but to setup experiments with this aggregation state has been very
difficult. Table 8.2 summarizes these experiments for reference purposes.
Table 8.2 Summary of experiments for the determination of ion–water interaction

cross sections. All experiments were carried out by using water vapor as a target
Reference Projectile Energy (MeV/u) Cross sections Uncertainties
Toburen et al.15 H + , H0 0.1–2.5 Single electron 10–12%
capture and loss
Toburen and H+ 0.3, 0.5, 1.0, Single ionization 20%
Wilson16 and 1.5
Toburen et al.17 He+ 0.075, 0.2, 0.3, Single ionization 20%
0.4, and 0.5
Table 8.2 (Continued)
Reference Projectile Energy (MeV/u) Cross sections Uncertainties
He2+ 0.2, 0.3, 0.4, Single ionization 20%
and 0.5
Wilson et al.18 H+ 3.0 and 4.2 Single ionization 20%
Rudd et al.19 H+ 0.007–4.0 Single ionization 8–20%
and electron capture
Rudd et al.20 He+ 0.005–0.45 Single ionization, 5–12%
electron capture
and loss
Bolorizadeh H+ 0.015–0.15 Single ionization 20%
and Rudd21
Werner et al.22 H+ 0.1–0.35 Water fragmentation ∼10–50%
He2+ 0.25 Water fragmentation ∼10–50%
Gobet et al.23 H+ 0.02–0.15 Water fragmentation N/A
Ohsawa et al.24 He2+ 6.0 and 10.0 Single ionization ∼13%
4. SEMIEMPIRICAL METHODS

Various semiempirical models have been developed to reproduce
ionization cross sections due to the impact of ions on water.3, 2, 13, 25 In this
section only three of them will be addressed.
4.1 The Rudd model

Rudd et al.3, 26 built a semiempirical model to determine single ionization
cross sections corresponding to proton impact on water as a function of
secondary electron energy. It is based on pioneer works of E. Rutherford
and J.J. Thomson on the impact of charged particles with matter (binary
encounter approximation). The single differential cross sections (SDCS)
depend only on the ejected and bound electron energies and the projectile
velocity. The formulae to determine the ionization cross section differential
in the secondary electron energy can be found in Ref. 3. This model can
also be applied to multi-electronic targets by using the electron binding
energy in each molecular or atomic shell.
Rudd and coworkers made a distinction between inner and outer sub-
shells in multielectron targets. Those with binding energy higher and lower
than twice that of the outermost subshell are regarded as inner and outer
subshells, respectively. Dingfelder et al.2 used this model to obtain liquid
water stopping powers reported by ICRU27 for proton energies below
0.5 MeV. In addition, they introduced a partition factor to weight the cross

sections corresponding to each subshell in order to reproduce B1 approxi-
mation results at impact energies above 0.5 MeV. They determined a com-
plete set of parameters and partition factors for liquid water. All parameters
used in this model, corresponding to inner and outer subshells of liquid
water (Dingfelder) and water vapor (Rudd) can be found in Table 3 of
Ref. 2. Binding energies and partition factors for both water phases are
displayed in Table 4 of the same reference.
4.2 The HKS method

Stolterfoht13 used the transition amplitude (Eq. (11)) determined previously
by Hansen and Kocbach5 to develop a semiempirical method to calculate
ionization cross sections for light ions. He found that these cross sections
diverged as k → 0 due to the neglect of the initial electron momentum. A
semiempirical term was then included in the transition amplitude to remove
that singularity. In addition, other modifications were incorporated in the
original formalism to reproduce the results determined by using the B1
approximation formula published by Landau and Lifshitz.11 This is the so-
called HKS method. However, Bernal and Liendo6 realized that previously
published HKS expressions13, 14 had some misprints so that they carried out
all calculations analytically and corrected the HKS formulae. In addition, they
compared the results obtained from the corrected HKS expressions for water
ionization due to the impact of protons and alpha particles (see Section 5).
4.3 Formalism based on the dielectric response function

As pointed out by Fano,28 macroscopic differential cross sections for energy
and momentum transfer in a condense medium can be obtained within the
B1 approximation from the expressions
d� 2 2ZP2

−1
(15)
= Im ,
dE dq π v2q ǫ(E, K )
d� 2ZP2
qmax
−1

dq
= Im , (16)
dE π v2 qmin ǫ(E, K ) q
where ǫ(E, K ) is the medium dielectric response function (DRF); E and

q represent the energy and momentum transfer, respectively; qmin is given
by Eq. (2), qmax = (2MP )1/2 [τ 1/2 + (τ − E)1/2 ] ≈ 2MP v is the maximum
momentum transfer, and τ = MP v 2 /2 is the projectile kinetic energy.
This approach accounts for the polarization of the medium, collective exci-
tations and many-body interactions in a consistent manner. The correspond-
ing cross sections scale as (ZP /v)2 because they are determined within the
B1 approximation. An important advantage of this treatment is that available
experimental data for liquid water (see Ref. 29, and references therein) can
be used to fit the corresponding DRF. Dingfelder et al.30 and Emfietzoglou
et al.25 based a series of works on this methodology to construct semiempirical
formalisms for the determination of ionization and excitation cross sections
due to the impact of electrons on liquid water. Later on, they extended these
formalisms to protons impinging on this medium.2, 25 These authors obtained
the DRF from a linear superposition of Drude-type functions. In this case,
bound electrons were supposed to move in a forced and damped oscillator.
The group of Dingfelder used experimental optical reflectance data obtained
by Heller et al.31 while that of Emfietzoglou have used more recent inelastic
X-ray scattering spectroscopy (IXSS) data reported by Hayashi et al.29 The idea
was to separate the discrete and continuum parts of the term Im [-1/ (E, K)]
in Eq. (15), also known as the energy-loss function (ELF). In addition, deriva-
tive Drude-type DRF were used for excitations, leading to more pronounce
peaks. Experimental data were fitted to obtain the optical dielectric response
function (ODRF), satisfying some physical constraints (see Ref. 32). For
instance, as IXXS data were only reported for transferred energies in the range
6–160 eV, Emfietzoglou and Moscovitch32 only included the four outermost
subshells of the liquid water molecule. Parameters for the K-shell were deter-
mined by use of the NIST photoelectric database.33
In a more recent publication,25 the oxygen K-shell was treated sepa-
rately and a hydrogen-type generalized oscillator strength (GOS)2 was
employed to determine the ELF of this shell, using the same NIST data-
base. It should be pointed out that several approximations have been used
to extend transition energies to finite momentum transfers, many of them
have been reviewed and compared in Ref. 25.
5. COMPARISON BETWEEN EXPERIMENTAL,

THEORETICAL, AND SEMIEMPIRICAL RESULTS
Experimental cross sections for the ionization of water due to the
impact of ions are very scarce, as mentioned previously (see Section 3).
Moreover, these data have been only determined for water vapor which
makes very difficult the validation of methods for the liquid water ionization
cross section determination.
The B1 approximation has become a classical method to study ion–atom
collisions. Most of the scientific publications generated in this field use this
approach as a reference to compare their experimental and/or theoretical
results. For this reason, the analysis of the B1 treatment is not carried out
independently but together with CDW formalisms.
After the fundamental works of Cheshire,10 Belkić,8 and Crothers and
McCann,4 the CDW methodologies have been used by several investiga-
tors to study ion–atom collision problems (see, e.g., Refs. 34–36). Many of
these variants have been employed to investigate processes such as transfer
ionization, direct ionization, electron interference during direct ionization,
target excitation, electron loss from negative ions due to proton impact, and
more complex phenomena like multiple ionizations and the simultaneous
projectile ionization and target excitation and/or ionization. The success
of the CDW theories is mainly based on the reproduction of two-center
effects,34 which are due to the influence of both the projectile and residual
target on the electron under consideration. Some of these effects are com-
mented in the light of CDW methodologies in the following paragraphs.
On the contrary, the B1 approximation is not capable to explain these
effects. Only the most prominent effects are analyzed just below.
5.1 Ionization cross sections for water

As far as the authors know, Olivera et al.35 published the first work on the use
of the CDW-EIS formalism to determine single ionization cross sections for
protons bombarding water.They employed the complete neglect of differen-
tial overlap (CNDO) method to obtain electron wave functions. However,
their important work was aimed to the water vapor phase. Figure 8.2 shows
total ionization cross sections reported by these authors in Ref. 37. Electron
capture cross sections calculated using the Bragg’s additivity rule instead of
molecular wave functions are also displayed. Excellent agreement between
theoretical and experimental results was obtained for projectile energies
above approximately 30 keV.
Bernal and Liendo published a work on the determination of inelastic
cross sections for the interaction of protons, alpha particles, and carbon ions
with liquid water.7 They calculated ionization cross sections by use of the
CDW-EIS and HKS formalisms (see Section 2). In both methods hydro-
gen-like wave functions were used to represent the initial electron state.
Occupation number and ionization energy values used in that research
Figure 8.2 Total ionization and electron capture cross sections for the impact of pro-
tons on water vapor. Solid lines represent CDW-EIS calculations carried out by Olivera
et al. squares are data obtained from experiments by Barnet et al. circles and triangles
represent experimental results published in Ref. 19. Reprinted from Publication,37
Copyright (1996), with permission from Elsevier.
are shown in Table 1 of Ref. 7. DDCS for 0.3 and 1.5 MeV protons, and
0.3 and 0.5 MeV alpha particles impacting on liquid water calculated by
those authors are shown in Figures 8.3 and 8.4, respectively. Each curve is
for a given secondary electron energy. Corresponding experimental values
for water vapor16, 17 are also displayed. The use of hydrogen-like wave
functions to describe the electron initial state causes an underestimation
of DDCS at backward angles.13 Furthermore, CDW-EIS accounts for
the projectile–electron interaction explicitly so that the attraction of the
positively charged projectile on the atomic electron tends to reduce the
electron ejection probability at large angles. This is one of the well-known
two-center effects34 and the reason why CDW-EIS underestimates the
DDCS at backward angles more than HKS does. The attraction mentioned
here also causes CDW-EIS to overestimate, for all projectile energies, the
DDCS at forward angles with respect to the HKS values when the elec-
tron energy is close to the ECC peak energy WECC. This effect is more
important for the strongest perturbation (ZP /v), currently represented by
the alpha particle case where DDCS corresponding to the entire electron
energy range are affected. The shift of the CDW-EIS binary encounter
+ +
0.3 MeV H + water CDW-EIS 1.5 MeV H + water CDW-EIS
WECC=163 eV HKS WECC=817 eV HKS
symbols Toburen77 -18 symbols Toburen77
-18 10
10
12 eV
-19 12 eV
10
dσ/dWdΩ (cm eV Sr )
-19
-1
-1
10
-1
-1
-20
50 eV 10 50 eV
2
2
-20
10
-21
10 100 eV
-21
10 100 eV -22
10
250 eV
-22 250 eV
10 -23
750 eV
10
750 eV
2200 eV
-23 -24
10 1 0.5 0 -0.5 -1 10 1 0 -1
0.5 -0.5
cos(θ) cos(θ)
(a) (b)
Figure 8.3 DDCS determined by use of the HKS (dashed lines) and CDW-EIS (solid
lines) methods for (a) 0.3 MeV and (b) 1.5 MeV protons. Corresponding experimental
values for water vapor extracted from Ref. 16 (Toburen77). Reprinted from Publication,7
Copyright (2007), with permission from Elsevier.
CDW-EIS 2+ CDW-EIS
2+
0.3 MeV/u He + water 0.5 MeV/u He + water
HKS HKS
-17 W ECC=163 eV WECC=272 eV
10 symbols Toburen80 -17 symbols Toburen80
10
-18
10
-18
-1
-1
15 eV 10
15 eV
-1
-1
-19
10
2
-19
10
-20 52 eV
10
-20
100 eV
10
-21
10
160 eV 10
-21
10
-22 500 eV
700 eV
750 eV 1000 eV 274 eV
-23 -22
10 1 0.5 0 -0.5 -1 10 1 0.5 0 -0.5 -1
cos(θ) cos(θ)
(a) (b)
Figure 8.4 DDCS determined by use of the HKS (dashed lines) and CDW-EIS (solid lines)
methods for (a) 0.3 MeV/u and (b) 0.5 MeV/u alpha particles. Corresponding experi-
mental values for water vapor extracted from Ref. 17 (Toburen80). Reprinted from
Publication,7 Copyright (2007), with permission from Elsevier.
peaks towards smaller angles with respect to those determined by the

HKS method is another evidence of the presence of two-center effects. As
expected, this fact is more conspicuous in the alpha particle case. Although
the CDW-EIS approximation accounts for two-center effects, it was unable
to reproduce experimental DDCS near 0°. This behavior could be caused
by the use of hydrogen-like instead of realistic bound electron wave func-

tions. Neither of the CDW-EIS and HKS formalisms provides a good
agreement with experimental water vapor DDCS.
SDCS for protons and alpha particles impacting on liquid water
computed by the CDW-EIS and HKS models7 are displayed in Figure
8.5a and b, respectively. Corresponding experimental values previously
published for water vapor are also included for comparison.17, 18 In gen-
eral, the HKS method provides a better agreement with the experimental
results than the CDW-EIS formalism, for protons as well as for alpha
particles. CDW-EIS cross sections are lower than those predicted by the
HKS approach for all projectile energies, probably due to the significant
underestimation of the DDCS at large ejection angles (see Figs. 8.3 and
8.4). At the same projectile speed (0.5 MeV/u), CDW-EIS and HKS
SDCS are consistent with experimental data corresponding to W values
above 10 eV and 30 eV for proton and alpha particle beams, respectively.
This may be due to the fact that both CDW-EIS and HKS methods are
first-order perturbative approaches. The greater the projectile charge, the
worse the expected results for the same projectile velocity. Discrepancies
below about 10 eV could be attributed to experimental difficulties, as
commented in Ref. 18.
Figure 8.5 SDCS for (a) 0.5, 1.5, 3.0, and 4.2 MeV protons and (b) 0.3 and 0.5 MeV/u
alpha particles impacting on liquid water calculated by the CDW-EIS (solid lines) and
HKS (dashed lines) formalisms. Experimental results corresponding to water vapor are
shown (symbols).18, 17 Reprinted from Publication,7 Copyright (2007), with permission
from Elsevier.
CDW-EIS
HKS
Schutten66
Rudd&Goffe85
-14 6+ Rudd85
10 C
σion (cm )
2
2+
-15 He
10
+
H
-16
10
-17
10 1 10
E0 (MeV/u)
2+
Figure 8.6 Total single ionization cross sections for H , He , and C6+ ions impacting
+
on liquid water calculated by the CDW-EIS (solid lines) and HKS (dashed lines) formal-
isms. Available experimental values are also shown (symbols).38, 19, 20 Reprinted from
Publication,7 Copyright (2007), with permission from Elsevier.
Figure 8.6 shows the total single ionization cross sections for H+ , He2+,
and C6+ ions impacting on liquid water calculated by the CDW-EIS and
HKS formalisms. Corresponding experimental data for the H+ (Ref.19) and
He2+ (Ref. 20) ions bombarding water vapor are also displayed for com-
parison purposes. In addition, analog values obtained by Schutten et al.38 for
electrons impinging on water vapor with the same speed as protons are also
included in this figure. The agreement between the theoretical and experi-
mental values for protons is better for E0 below and above about 1 MeV/u
for the CDW-EIS and HKS formalisms, respectively. The total cross section
(TCS) obtained from experiments carried out with electrons as projectiles
underestimate the theoretical values importantly for protons energies below
about 1 MeV. Unfortunately, the experimental values for He2+ are not
enough to make a good comparison but they show a good high-energy
tendency.
Bernal and Liendo calculated corresponding excitation cross sections
to determine inelastic electronic stopping cross sections (SCS) for protons,
alpha particles, and carbon ions and compared their results with those
2+
Figure 8.7 Electronic stopping cross sections for H , He , and C6+ ions impacting
+
on liquid water calculated by the CDW-EIS (solid lines) and HKS (dashed lines) formal-
isms. Corresponding reference values from ICRU reports are also shown (symbols).27, 39
Reprinted from Publication,7 Copyright (2007), with permission from Elsevier.
provided by ICRU.27, 39 Figure 8.7 shows that the best SCS agreement
with the reference values is obtained with the CDW-EIS formalism. It is
important to remark that the CDW-EIS methodology is fully theoreti-
cal and only requires the knowledge of electron occupation numbers and
binding energies of the target under study to be implemented.
Dal Cappello et al. published a work40 in which double, single, and
total ionization cross sections for water vapor bombarded by 6 MeV/u C6+
ions were determined by means of three variants of the HKS model: one
obtained directly from the Hansen and Kocbach work5(HKSorig ), another
modified by Stolterfoht13(HKSStol ), and the corrected HKS expressions
reported by Bernal and Liendo7(HKSBL ). In addition, corresponding
experimental and B1 approximation values, referred to as FBA-CW by
Dal Cappello et al., were also reported. Moreover, the previously described
model developed by Rudd (see Section 4.1) was included for SDCS
comparison purposes. Figure 8.8 depicts SDCS reported in Ref. 40. In
our opinion, the HKSBL model is the most successful, in contrast to the
conclusions of Dal Cappello et al. The reader should be aware that the
Figure 8.8 SDCS for the ionization of water vapor by the impact of 6 MeV/u C6+ ions.
Solid circles show experimental values from the Ohsawa group (see Ref. 24). Results
from the original HKS model5 are represented by the dashed line; the HKS variant
due to Stolterfoht et al.,13 by the dotted line; the corrected HKS model due to Bernal
and Liendo,7 by the dashed-dotted line; and the Rudd model,3 by the dashed-dotted-
dotted line. Reprinted from Publication,40 Copyright (2009), with permission from
Elsevier.
greatest TCS components come from low energy SDCS corresponding

to the interval W < 100 eV, where the HKSBL variant provides better
results than the other two HKS modifications. Among all the semiempiri-
cal models studied, the Rudd model produces the best agreement with
the experiment. It is worth noting that the HKS expressions reported in
Ref. 13,14 have some misprints. In other words, they cannot be obtained
from the corresponding integration of the original transition probability
reported by Hansen and Kocbach.5
Electronic stopping cross sections for protons on liquid water are
shown in Figure 8.9. In this case, results from Emfietzoglou et al.41 were
Figure 8.9 Electronic stopping power of liquid water for protons as reported by

Emfietzoglou et al.41 Solid and dotted lines show results obtained by those authors
from the fit of inelastic X-ray scattering (IXS)29 and optical reflectance data (REF),31
respectively. The other results represent calculations and experimental determinations
reported by the cited references. Printed from Publication,41 Copyright (2006), with
permission from Elsevier.
obtained by use of the semiempirical model based on the electronic

response function of the medium (see Section 4.3). Other calculations
as well as experimental determinations are shown. The reference num-
bers shown in this figure are according to Ref. 41 instead of the cur-
rent work. In general, the agreement between all the methods shown
is excellent for proton energies above 300 keV. Discrepancies arise near
the Massey peak where the stopping cross section reaches a maximum.
It should be remarked that the calculations carried out by Emfietzoglou
et al.,41 based on IXS data, represent the best fit of the ICRU reference
values.
The semiempirical models developed by Rudd et al. (see Section
4.1) have been widely used to reproduce electron capture and single dif-
ferential and total ionization cross sections for the impact of light ions
on water vapor19, 40 and liquid water.2 They have also been used to fit
cross sections corresponding to water vapor fragmentation due to proton
bombardment.22, 42 These models have the advantage of being easy to be
implemented in Monte Carlo codes, providing a very high computing
efficiency. The quality of the results obtained from them can be appreci-
ated in Figure 8.8.
6. CONCLUSIONS AND PERSPECTIVES

The CDW methodologies have proven to be very successful
to reproduce experimental single ionization cross sections for several
projectile-target combinations in a very wide projectile energy range.
Among them, the CDW-EIS approximation has thus far been the most
successful. In general, these formalisms are capable to explain, unlike the
B1 approximation, certain complex phenomena involved in ion–atom
collisions, such as the well-known two-center effects. However, realistic
bound electron wave functions must be used to obtain more accurate
DDCS that are sensible to the initial electron state. In some physical situ-
ations, such as the double ionization of a multielectron target and the
interaction between a partially ionized projectile and a hydrogen-like
target, the four-body extended CDW formalisms arise as powerful tools
to determine the corresponding cross sections. Efforts have been made in
this sense by Belkić36, 43 resulting in good agreement between theoretical
and experimental cross sections for electron detachment from negative
hydrogen ion by proton impact. This author and colleagues have recently
reviewed four-body formalisms for various processes that take place in
ion–atom collisions.44
To the best of our knowledge, all Monte Carlo codes developed
to simulate energetic ion transport through matter use semiempirical
methods to determine ion–atom interaction cross sections. Several of
these methods lack a physical basis and are limited to fitting experi-
mental cross sections. In our opinion, the CDW-EIS approximation can
be used to calculate all the electronic cross sections required by these
Monte Carlo codes, including the case of relativistic projectile energies
of heavy ions.
The determination of experimental ionization cross sections for liquid
water is a topic which still presents a considerable challenge to experimen-
talists. Although it is a very difficult task, we are optimistic in this respect.
Recently, Kaneda et al.45 have carried out the first experiment to measure
the relative ion-fragment yields due to proton impact on liquid water.
The availability of liquid water ionization cross sections will allow the needed
tests of the corresponding theoretical models. This, in turn, would improve
our understanding of the role of the liquid water phase effects that influence
the ionization process.
ACKNOWLEDGMENTS
The authors are very grateful to professors R. Rivarola, C. Champion, and
D. Emfietzoglou for kindly having provided experimental data and/or fig-
ures shown in this chapter. The authors also thanks Ms. Camilla Sampaio
for having helped us during the construction of figures. M. Bernal wishes
to thank the FAPERJ foundation in Brazil for the research Grant No.
E-26/102.524/2009 and Prof. Carlos E. de Almeida for his support during
the realization of this work.
REFERENCES
1. Friedland,W.; Jacob, P.; Paretzke, H., et al. Simulation of light ion induced DNA damage
patterns. Radiat. Prot. Dosim. 2006, 122, 116.
2. Dingfelder, M.; Inokuti, M.; Paretzke, H. Inelastic-collision cross sections of liquid water
for interactions of energetic protons. Radiat. Phys. Chem. 2000, 59, 255.
3. Rudd, M. E.; Kim, Y.; Madison, D. H., et al. Electron production in proton collisions
with atoms and molecules: energy distributions. Rev. Mod. Phys. 1992, 64, 441.
4. Crothers, D.; McCann, J. Ionisations of atoms by ion impact. J. Phys. B: At. Mol. Opt.
Phys. 1983, 16, 3229.
5. Hansen, J. P.; Kocbach, L. Ejection angles of fast delta electrons from K-shell ionisation
induced by energetic ions. J. Phys. B 1989, 22, L71.
6. Bernal, M. A.; Liendo, J. A. The HKS model for electron production in liquid water by
light ions. Nucl. Instrum. Methods Phys. Res., Sect. B 2006, 251, 171.
7. Bernal, M. A.; Liendo, J. A. Inelastic-collision cross sections for the interactions of totally
stripped H, He and C ions with liquid water. Nucl. Instrum. Methods Phys. Res., Sect. B
2007, 262, 1.
8. Belkić, Dž. A quantum theory of ionisation in fast collisions between ions and atomic
9. McDowell, M.; Coleman, J. Introduction to the Theory of Ion–Atoms Collisions; North-
Holland: Amsterdam, 1970.
10. Cheshire, I. M. Continuum distorted wave approximation; resonant charge transfer by
fast protons in atomic hydrogen. Proc. Phys. Soc. 1964, 84, 89.
11. Landau, L. D.; Lifshitz, E. M. Quantum Mechanics: Nonrelativistic Theory; Pergamon Press:
Barcelona, 1977.
12. Fainstein, P. D.; Ponce, V. H.; Rivarola, R. D. Electron emission from multielectronic
atoms by ion impact at intermediate and high energies. J. Phys. B: At. Mol. Opt. Phys.
1989, 22, 1207.
13. Stolterfoht, N.; DuBois, R.; Rivarola, R. D. Electron Emission in Heavy Ion–Atom

Collisions; Springer: Berlin, 1997.
14. Secondary electron spectra from charge particle interactions, Tech. Rep. 55, ICRU,
Bethesda, MD, 1996.
15. Toburen, L. H.; Nakai, M. Y.; Langley, R. A. Measurement of high-energy charge-
transfer cross sections for incident protons and atomic hydrogen in various gases. Phys.
Rev. 1968, 171, 114.
16. Toburen, L. H.; Wilson, W. E. Energy and angular distributions of electrons ejected from
water vapour by 0.3–1.5 MeV protons. J. Chem. Phys. 1977, 66, 5202.
17. Toburen, L. H.; Wilson, W. E.; Popovich, R. J. Secondary electron emission from ioniza-
tion of water vapor by 0.3 to 2.0 MeV He+ and He2+ ions. Radiat. Res. 1980, 82, 27.
18. Wilson, W. E.; Miller, J. H.; Toburen, L. H., et al. Differential cross sections for ioniza-
tion of methane, ammonia and water vapour by high velocity ions. J. Chem. Phys. 1984,
80, 5601.
19. Rudd, M. E.; Goffe, T. V.; Dubois, R. D., et al. Cross sections for ionization of water
vapor by 7–4000 keV protons. Phys. Rev. A 1985, 31, 492.
20. Rudd, M. E.; Itoh, A.; Goffe, T. V. Cross sections for ionization, capture, and loss for
5–450 keV He+ on water vapor. Phys. Rev. A 1985, 32, 2499.
21. Bolorizadeh, M. A.; Rudd, M. E. Angular and energy dependence of cross sections for
ejection of electrons from water vapor. II. 15–150 keV proton impact. Phys. Rev. A 1996,
33, 888.
22. Werner, U.; Beckord, K.; Becker, J., et al. 3D imaging of the collision-induced Coulomb
fragmentation of water molecules. Phys. Rev. Lett. 1995, 74, 1962.
23. Gobet, F.; Eden, S.; Coupier, B., et al. Ionization of water by (20–150) keV protons:
separation of direct-ionization and electron-capture processes. Phys. Rev. A 2004, 70,
062716.
24. Ohsawa, D.; Kawauchi, K.; Hirabayashi, M., et al. An apparatus for measuring the energy
and angular distribution of secondary electrons emitted from water vapor by fast heavy-
ion impact. Nucl. Instrum. Methods Phys. Res., Sect. B 2005, 227, 431.
25. Emfietzoglou, D.; Garcia-Molina, R.; Kyriakou, I., et al. A dielectric response study of
the electronic stopping power of liquid water for energetic protons and a new I-value
for water. Phys. Med. Biol. 2009, 54, 3451.
26. Rudd, M. E. Differential cross sections for secondary electron production by proton
impact. Phys. Rev. A 1988, 38, 6129.
27. Stopping powers and ranges for protons and alpha particles, Tech. Rep. 49, ICRU,
Bethesda, MD, 1993.
28. Fano, U. Penetration of protons, alpha particles, and mesons. Annu. Rev. Nucl. Sci. 1963,
13(1), 1–66.
29. Hayashi, H.; Watanabe, N.; Udagawa, Y., et al. The complete optical spectrum of liquid
water measured by inelastic X-ray scattering. Proc. Natl. Acad. Sci. USA 2000, 97, 6264.
30. Dingfelder, M.; Hantke, D.; Inokuti, M. Electron inelastic scattering cross sections in
liquid water. Radiat. Phys. Chem. 1998, 53, 1.
31. Heller, J. M.; Hamm, R. N.; Birchoff, R. D., et al. Collective oscillation in liquid water.
J. Chem. Phys. 1976, 60, 3483.
32. Emfietzoglou, D.; Moscovitch, M. Inelastic collision characteristics of electrons in liquid
33. <http://www.nist.gov/physlab/data/ffast/index.cfm>.
34. Fainstein, P. D.; Ponce, V. H.; Rivarola, R. D. Two-centre effects in ionization by ion
impact. J. Phys. B: At. Mol. Opt. Phys. 1991, 24, 3091.
35. Olivera, G. H.; Rivarola, R. D.; Fainstein, P. D. Ionization of the excited states of hydro-
gen by proton impact. Phys. Rev. A 1995, 51, 847.
36. Belkić, Dž. Electron detachment from the negative hydrogen ion by proton impact.
J. Phys. B: At. Mol. Opt. Phys. 1997, 30, 1731.
37. Fainstein, P. D.; Olivera, G. H.; Rivarola, R. D. Theoretical calculations of the stopping
power for protons traversing H, He and simple molecular targets. Nucl. Instrum. Methods
Phys. Res., Sect. B 1996, 107, 19.
38. Schutten, J.; DeHeer, F. J.; Moustafa, H. R., et al. Cross and partial ionization cross sec-
tions for electrons on water vapor in the energy range 0.1–20 keV. J. Chem. Phys. 1966,
44, 3924.
3 9. Stopping of ions heavier than Helium, Tech. Rep. 73, ICRU, Bethesda, MD, 2005.
40. Dal Cappello, C.; Champion, C.; Boudrioua, O., et al. Theoretical and experimental
investigations of electron emission in C6+ − H2 O collisions. Nucl. Instrum. Methods Phys.
Res., Sect. B 2009, 267, 781.
41. Emfietzoglou, D.; Nikjoo, H.; Pathak, A. Electronic cross sections for proton transport
in liquid water based on optical-data models. Nucl. Instrum. Methods Phys. Res., Sect. B
2006, 249, 26.
42. Gobet, F.; Farizon, B.; Farizon, M., et al.Total, partial, and electron-capture cross sections
for ionization of water vapor by 20–150 keV protons. Phys. Rev. Lett. 2001, 86, 3751.
43. Belkić, Dž. Quantum-mechanical methods in loss-excitation and loss-ionization in fast
ion–atom collisions. Adv. Quantum Chem. 2009, 56, 251.
44. Belkić, Dž.; Mančev, I.; Hanssen, J. Four-body methods for high-energy ion–atom col-
45. Kaneda, M.; Shimizu, M.; Hayakawa, T., et al. Mass spectrometric study of collision
interactions of fast charged particles with water and NaCl solutions. Nucl. Instrum.
Methods Phys. Res., Sect. B 2009, 267, 908.
46. Akkerman, A.; Breskin, A.; Chechik, R., et al. Calculation of proton stopping power
in the region of its maximum value for several organic materials and water. Rad. Phys.
Chem. 2001, 61, 333.
47. Xu, Y. J.; Khandelwal, G. S.; Wilson, J. W. Proton stopping cross sections of liquid water.
Phys. Rev. A 1985, 32, 629.
48. Bauer, P.; Kaferbock,W.; Necas,V. Investigation of the electronic energy loss of hydrogen
ions in H2 O: influence of the state of aggregation. Nucl. Instrum. Methods Phys. Res., Sect.
B 1994, 93, 132.
49. Wenzel, W. A.; Whaling, W. Cross section and angular distribution of the D(D, p)T
reaction. Phys. Rev. 1952, 88, 1149.
CHAPTER NINE
Computation of Distorted Wave

Cross Sections for High-Energy
Inelastic Collisions of Heavy
Ions with Water Molecules
Roberto D. Rivarolaa, Mariel E. Galassia, Pablo D. Fainsteinb and
Christophe Championc
aInstitutode Física Rosario (CONICET-UNR) and Facultad de Ciencias Exactas, Ingeniería y Agrimensura,
Universidad Nacional de Rosario, Avenida Pellegrini 250, 2000 Rosario, Argentina
bCentro Atómico Bariloche, Comisión Nacional de Energía Atómica, Avda E. Bustillo 9500, 8400 Bariloche,
Argentina
cUniversité Bordeaux 1, CNRS/IN2P3, Centre d’ Etudes Nucléaires de Bordeaux-Gradignan, CENBG,
Gradignan, France
Contents
1. Introduction 232
2. The distorted wave model for inelastic collisions 234
2.1 The theoretical description for charged projectiles 235
2.2 The theoretical description for neutral projectiles 239
3. Electronic stopping power 241
4. The case of water molecules 245
4.1 Differential and total cross sections 246
4.2 Electronic stopping power 250
5. Multiple ionization of water molecules 253
5.1 Linear energy transfer 260
5.2 Free radicals formation in water radiolysis 260
6. Concluding remarks 262
References 263
Abstract
Irradiation of water and other molecules of biological interest by heavy ion beams is
studied. Distorted wave models are employed to investigate the corresponding inelas-
tic collisions. Cross sections for electronic capture, ionization, and excitation processes
are determined as well as equilibrium charge-states and electronic stopping power.
The influence of multiple ionization in liquid water radiolysis is analyzed.

232 Roberto D. Rivarola et al.
1. INTRODUCTION
The interaction of charged particles with matter has been a subject of
study since the pioneering works of Bethe.1 In these seminal works Bethe
applied the first-Born approximation to calculate the total cross section
for single ionization and excitation of H by swift electron and ion impact.
Using a multipole expansion of the perturbation, and keeping only the
dipole term, he was able to obtain closed formulas of the total cross sections
and stopping power. The comparison with experimental results showed
good agreement at high impact energy, where these reactions dominate,
and that it largely overestimated the cross sections at intermediate and small
impact energies. This showed that such a first-order theory could not give
a complete representation of the processes. However, the simplicity of the
calculation, the compact form of the result, the requirements from different
applications for a large quantity of data, and the difficulty to develop higher
order methods lead to the addition of semi-empirical corrections to the
Bethe formula. This produced what is now called the Bethe–Bloch theory2
which is the relativistic version of Bethe’s theory including the Barkas,
shell and polarization corrections.3 Results with this theory for atomic
and molecular targets, the later obtained employing Bragg’s rule, are readily
available for electron and proton impact in a large range of impact energies.
The corrections included in the Bethe–Bloch theory are mainly related
to the dependence of the total cross sections (or stopping power) on the
properties of the target. They depend therefore on the number of electrons
(shells) or the polarizability of the electron gas. It is well known that in
first-order theories the total cross section and stopping power scale with
the square of the projectile charge ZP. Experiments performed by Brandt
and coworkers4 employing He2+ and Li3+ projectile ions impinging on
heavy targets, showed for the first time large discrepancies as deviations
from the ZP2-scaling law. These discrepancies were explained introducing
the binding and polarization effects. The binding effect, which is important
from intermediate to low impact energy, arises from the increasing binding
of the active electron due to the combined projectile and target potential.
This effect lowers the total cross section since the increased binding results
in a lower emission probability. The polarization effect, which is important
at intermediate to high energies, arises from the distortion of the initial
state by the projectile potential. The experiments showed that this effect
was responsible for an increase of the total cross section with respect to the
prediction of the Bethe–Bloch theory. The projectile potential attracts the
Distorted Wave Methods for Ion-Water Collisions 233
electronic cloud, therefore increasing the effective impact parameter and

thus increasing the transition probability. Basbas and coworkers developed
the perturbed-stationary-state (PSS) theory where all these effects were
taken into account as higher order corrections to the first-Born approxima-
tion and found very good agreement with the experimental data.5 In this
approach the total cross section results in an expansion in powers of the
ratio between the projectile and target nuclear charge. The first-term, pro-
portional to ZP2 corresponds to the first-Born approximation. The appear-
ance of even and odd powers indicated that there are not only higher order
contributions but also that they differ on the sign of the projectile charge.
In this way Brandt and Basbas6 explained the difference in the total cross
sections and stopping power between particle and antiparticle which had
already been found experimentally by Barkas and coworkers.7
The PSS model has been found to yield good agreement with experi-
ments for asymmetric systems, when the projectile charge ZP is much
smaller than the target nuclear charge ZT. It is clear that when the projec-
tile charge increases, the parameter ZP /ZT will be large and unreasonable
for a perturbative development. Therefore a different kind of theoretical
approach is required to represent the total cross sections and stopping pow-
ers for all values of ZP and ZT.
A theoretical method which is well suited for this problem is the dis-
torted wave theory. Since all the particles involved are charged, the inter-
actions are governed by the long-range Coulomb forces. It is well known
that the Coulomb potential acts even at infinite long distances. The inter-
action between two charged particles at long distances might be small but
never zero. As a consequence the wavefunction representing this situation
cannot be written as a product of free-particle (plane-waves) wavefunc-
tions. The long-range nature of the potential results in the appearance of a
Coulomb phase or distortion. In distorted wave theory, the initial and final
state wavefunctions are chosen in such a way as to represent the physi-
cal problem and the correct asymptotic conditions at very large distances.
These wavefunctions can be chosen in different ways leading to different
distorted wave models. The Continuum Distorted Wave (CDW) model
was first proposed by Cheshire8 to investigate electron capture and after-
wards extended by Belkic 9 for electron ionization. One of the most suc-
cessful theoretical descriptions is the Continuum Distorted Wave-Eikonal
Initial State (CDW-EIS) model introduced by Crothers and McCann10 to
study electron ionization of monoelectronic targets and latter extended
by Fainstein et al.11 for multielectronic targets. The CDW-EIS model was
subsequently formulated to investigate the electron capture reaction.12 The

CDW-EIS model was applied with great success in more symmetric sys-
tems by Fainstein et al.13 for single ionization of He by H+ , He2+, and Li3+
impact. A very good agreement was obtained with the experimental data
available in whole energy range from a few keV/amu to a few MeV/amu.
Moreover the calculations explained the large difference between results
for proton and antiproton impact which had been found experimentally at
CERN by Andersen et al.14 All these results could be interpreted in terms
of the binding and polarization mechanisms of Brandt and coworkers.
For symmetric systems however, another kind of effect, not included
in the Bethe and PSS models, must be considered. As the impact energy
decreases the cross sections for electron capture increase and become even
larger that the corresponding ones for ionization and excitation. As a result,
a fraction of the particles in the projectile beam will suffer a change in
charge-state. For example, a fraction of a proton beam will be neutralized
and even, at very low impact energy, become negatively charged (H−).
In equilibrium at a given impact energy there will be several charge-
states and their contribution must be considered to calculate accurately
the stopping power. The difficulty in this approach is that a huge number
of cross sections are required for all the open channels. Such an approach
has been considered up to now with two different models. Schiwietz and
coworkers15,16 calculated the cross sections employing the coupled-channel
method using atomic orbitals while Fainstein et al.17 and Olivera et al.18
employed the distorted-wave (DW) method. Both methods showed very
good agreement with the experiments available for proton impact on H
and He targets. Furthermore, the DW method was also successfully applied
to stopping power calculations for simple molecular targets.19
2. THE DISTORTED WAVE MODEL FOR INELASTIC

COLLISIONS
When a fast bare ion beam impacts on atoms or molecules, three
main electronic reactions are produced: electron ionization or excitation of
the target and electron capture of target electrons by the projectile. This last
reaction dominates at low-intermediate collision energies, corresponding
to impact velocities v ve, where ve is the velocity of the transferred
electron in the initial unperturbed target orbital. As the energy increases
electron ionization becomes the dominant channel whereas electron
excitation gives larger contributions than electron capture.
In order to describe these one-electron processes in the penetration

of light ion beams in a gaseous target a distorted wave model was formu-
lated.17 In such theoretical description the CDW-EIS model was employed
for electron capture12 and electron ionization10, 11 while target excitation
was included employing the Symmetric Eikonal (SE) one.20 These models
were applied with success to study numerous collision systems (see Ref. 21
and references therein) not only for processes involving one active electron.
However, for e.g. two-electron capture from helium by alpha particles, the
four-body version of the CDW-EIS model gives total cross sections that
are smaller than the experimental data by a factor varying from 1000 to 10
at impact energies 0.1-3 MeV (see review22a and references therein). These
theoretical approximations introduce the effect of the distortion produced
on the promoted electron by the presence of coulombic perturbative
potentials in the entry and exit channels, paying particular attention to the
asymptotic conditions corresponding to their long-range character.
2.1 The theoretical description for charged projectiles

For simplicity we will consider a three-body system composed by a bare pro-
jectile, an active electron, and a one-center residual target. Moreover, it is con-
sidered that the passive electrons (those ones that are not promoted) remain
in their orbitals during the collision. This assumption is valid at high enough
impact velocities for which the collision time is smaller than the one cor-
responding to relaxation of the passive electrons. For molecular targets it will
correspond also to collision times smaller than the vibrational and rotational
ones. In the following, the notation is that usually employed for distorted
wave models and atomic units will be used except where otherwise stated.
The straight line version of the impact parameter approximations is
used, for which:
� = ρ� + vt,
R � (1)
where R is the internuclear vector, ρ the impact parameter, v
the collision
velocity, and t the time evolution. The system is described from a reference
frame fixed on the target nucleus.
In the α-entry channel, the initial one-active electron distorted wave-
function is chosen as:

ZP ZT
χα+ ln vR − v 2 t

= ϕα (�x) exp (−iεα t) exp [−iν ln (vs + v
� · �s)] exp i
v

ZP Z T
ln vR − v 2 t ,

= ψα exp i
v (2)
where the super-index (+) indicates that it preserves correct outgo-

ing boundary conditions. In Eq. (2), ϕα (x) represents the non-perturbed
initial electron bound orbital and εα the corresponding orbital energy,
x and s are the electron position vectors with respect to the target and
projectile nucleus, ZP and ZT are the projectile and target nuclear charges
and ν = ZP /v (also called the Sommerfeld parameter). The eikonal phase
included in the entrance channel depending of the s coordinate, takes
account of the fact that the electron bound to the target is simultaneously
traveling in a continuum state of the projectile field. In fact, the electron is
moving in a combined field of the residual target and of the projectile. In
this sense χα+ is a two-center wavefunction which contains the informa-
tion on the long-range coulomb interaction between the projectile and the
electron. The other eikonal phase depending on the internuclear distance
corresponds to the Coulomb interaction between the nuclei present in the
reaction.
The choice of the final distorting wavefunction in the β-exit channel
depends on the process investigated. Thus, for electron capture this wave-
function is chosen as
v2

−(c)
χβ = ϕβ (�s) exp −iεβ t + i v � · x� − i t
2

ZP ZT
× N ∗ (ς)1 F1 (−iς; 1; −ivx − i v ln vR + v 2 t

� · x�) exp −i
v

Z P ZT
= ψβ(c) exp −i ln vR + v 2 t .

v (3)
Now, the super-index (−) indicates that correct ingoing conditions are
satisfied, ϕβ (s) represents the non-perturbed electron bound wavefunction
in the exit channel with εβ the corresponding orbital energy. The other
terms in the first phase factor are due to the Galilean transformation of the
electron wavefunction from the projectile frame to the target nucleus one.
Also in expression (3), 1 F1 (−iς; 1; −ivx − i v � · x�) is a continuum factor
of the electron–residual target interaction, ς = ZT∗ /v, with ZT∗ the effective
target nuclear charge. Thus, the final distorted wavefunction presents again
a two-center character associated with the fact that the electron evolves in
the combined field of the projectile and residual target ones. The function
N (a) = exp(πa/2)Ŵ(1 − ia) gives the normalization of the continuum
factor and the eikonal phase depending on the internuclear distance repre-
sents again the interaction between the nuclei. According to the prescription
of Belkic et al.22b the effective charge ZT∗ is chosen as ZT∗ = −2nα2 εα ,

where nα is the principal quantum number. This choice corresponds to

consider an effective Coulomb potential to represent the interaction of the

electron with the residual target in the exit channel, even in the case of
multielectron targets.
To describe the electron ionization process the final distorted wavefunction
is taken as:
k 2
−(i) −3/2
χβ = (2π) exp i k� · x� − i t N ∗ (ξ )1 F1 −iξ ; 1; −ikx − i k� · x�
2

∗
Z P ZT 2

× N (ζ )1 F1 −iζ ; 1; −ips − i�p · �s exp −i ln vR + v t
v

(i) Z P ZT 2

= ψβ exp −i ln vR + v t ,
v (4)
where k and p� = k� − v � are the linear momenta of the electron with respect
to the target and projectile nucleus, respectively, ξ = ZT∗ /k and ζ = ZP /p.
The first part of Eq. (4), depending on the coordinate x, is a wavefunction
describing the electron in a continuum state of the residual target whereas
the continuum factor depending on the coordinate s corresponds to the
electron–projectile interaction. Thus, the final distorted wavefunction indi-
cates that the electron is traveling in a continuum state of the projectile and
residual target fields, and their actions on the emitted electron are consid-
ered on equal footing.
Finally, for the electron excitation reaction the final distorted wavefunc-
tion is chosen as:

Z P ZT
χβ−(e) = ϕβ (�x) exp −iεβ t exp [iν ln (vs − v ln vR + v 2 t

� · �s)] exp −i
v

(e) Z P ZT
ln vR + v 2 t

= ψβ exp −i
v (5)
with ϕβ (x) representing an excited state of the target with orbital energy
εβ and the eikonal phase depending on the coordinate s describes the elec-
tron in a continuum state of the projectile. The last phase factor takes into
account the internuclear interaction.
The first-order scattering amplitude for any of the cases mentioned
above (the elastic scattering channel is not considered) can be written as:
+∞
+ − ∂ † + +
� = −i
Aαβ (ρ) dt χβ Hel − i χα = (ρv)2iZP ZT /v aαβ �
(ρ)
−∞ ∂t
(6)

and +∞
− − ∂ + −
χ = (ρv)2iZP ZT /v aαβ

� = −i
Aαβ (ρ) dt χβ Hel − i �
(ρ)
−∞ ∂t α (7)
in its post- and prior-versions, respectively, with Hel the electronic

Hamiltonian which results from solving the full time-dependent Schrödinger
equation in the eikonal approximation. In Eqs. (6) and (7), the final dis-
torted wavefunction must be replaced by the corresponding ones defined
in Eqs. 3–5 if electron capture, electron ionization or electron excitation are
considered, respectively. Also,
+∞ †
+ ∂
aαβ � = −i
(ρ) dt ψβ H
el − i ψα
(8)
−∞ ∂t
and
+∞

− ∂
(9)

aαβ �
(ρ) = −i dt ψβ H
el − i ψα
−∞ ∂t
are scattering amplitudes where the internuclear potential is excluded. This

interaction is completely contained in the phase factors (ρv)2iZP ZT /v and
not included in the electronic Hamiltonian H el. These scattering ampli-
tudes (in the prior- or post-versions) can be related to the transition matrix
element:
4π 2 (10)
�) = i
Tαβ (K Rαβ (�
η)
v
by using the Fourier transform:

1
� =
aαβ (ρ) η e −iρ·�
d� �η
Rαβ (�
η), (11)
2π
where K is the total momentum transfer and η its transverse component

defined in a direction perpendicular to the incident velocity vector. It
is interesting to note that the associated matrix elements Rαβ ( η) can be
obtained in analytical forms for any of the single electron reactions consid-
ered. Thus, for example, a general analytical expression has been obtained
by Martínez et al.23 for electron capture from any target orbital to any final
projectile state. Then, for single electron capture and single electron excita-
tion of atomic targets the total cross sections can be obtained from:

2 2
σαβ = dρ� aαβ (ρ)
� = η Rαβ (�
d� η ) , (12)
where the Parseval identity has been employed. In a similar way, for elec-
tron ionization doubly differential cross sections as a function of the final
electron energy Ek and the solid angle k subtended by the emitted elec-
tron can be found from:10, 24
d2 σαk

2 2
(13)

= k dρ� aαβ (ρ)
� = k d�
η Rαβ (�
η ) .
dEk d�k
Single differential cross sections dσαk /dEk and dσαk /d�k can be obtained
by integrating Eq. (13) on k and Ek, respectively. Total cross sections σαk
can then be calculated by integration of dσαk /dEk on the final electron
energy Ek or by integration of dσαk /d�k on the solid angle k.
2.2 The theoretical description for neutral projectiles

For neutral projectiles the incoming nucleus is screened by its bound electrons.
The perturbation between the dressed projectile and the target electrons is
of short-range, having thus a different character than for the case of charged
projectiles. It has been shown25 that the DW model reduces to the first-Born
approximation for neutral aggregates.We will restrict our analysis to the impact
of hydrogen atom beams, neglecting the possibility of the double charge
exchange reaction22c associated with the formation of H−. By simplicity we
will consider an effective monoelectronic target model.
In the present case several channels are opened because both collisions
partners can be excited and/or ionized. The final state of the incoming
projectile defines the projectile-elastic and projectile-inelastic channels
if it remains in the initial ground state or is excited to a bound or con-
tinuum state, respectively. In a similar way, target-elastic and target-inelastic
channels can be defined. Two-active electron processes like excitation or
ionization of the target with simultaneous excitation or ionization of the
projectile must be also considered. Following Bates and Griffing,26 we will
investigate all these reactions by using the first-order Born approximation
(B1). Initial and final wavefunctions are given now as:
� xp −iv
φαα ′ = ϕα (�xt )ϕα ′ ( �sp )e iv·�
2 t/2−i
(εα +εα′ )t , (14)
and
� xp−iv 2 t/2−i(ε
β +εβ ′ )t
φββ ′ = ϕβ (�xt )ϕβ ′ ( �sp )e iv·� , (15)
where ϕα (�xt )(ϕα ′ (�sp )) describes the initial state of the target (projectile)
electron and ϕβ (�xt )(ϕβ ′ (�sp )) excitation or continuum states of the tar-
get (projectile) electron in the field of the parent nucleus. Also, xt (sp ) is
the target (projectile) electron coordinate referred to its parent nucleus.
Furthermore, εα (εα′ ) and εβ (εβ ′ ) are the initial and final energies of the
target (projectile) electron, respectively.
Excluding the elastic scattering between the aggregates of the collision
−
and inserting the wavefunctions φαα ′ instead of χα+ and φββ ′ instead of χβ
in Eqs. (6) and (7), we obtain the B1-scattering amplitude corresponding to
its post- and prior-versions. Moreover, it is easy to prove24 that:
(16)
∂
Hel − i φαα ′ = Vint φαα ′
∂t
with
ZT Z 1 (17)
Vint = − − P + .
R� + �sp R
� − x�t R
� + �sp − x�t
The first to third terms in expression (17) give, following this order,
the interactions between the target nucleus and the projectile electron,
between the projectile nucleus and the target electron and, between the
projectile and target electrons. Moreover, the first (second) term does
not contributes to the scattering amplitude if a change in the target
(projectile) state is produced.27 For projectile (target) elastic scattering
(monoelectronic reaction) the electron–electron interaction produces
a screening of the projectile (target) charge. However, for dielectronic
reactions where projectile and target states change simultaneously, this
interaction produces an effect known as antiscreening. In fact, the pro-
jectile nucleus and the projectile electron act incoherently on the target
electron, so that the target electron feels separately the influence of these
particles. 28 This effect is present for large momentum transfer under close
encounter collisions.
The combined use of the distorted wave approximations for ion beams
and B1 approximation for neutral projectiles is known as the distorted wave
model (DW).17, 18
3. ELECTRONIC STOPPING POWER

In order to calculate the electronic stopping power for a bare ion
beam penetrating a gaseous target we will consider the processes of excita-
tion, ionization, and electron capture involving only one active electron, so
that it is given by:
∞ (i)
(e) (e) (c) (c)

(i) dσαk (18)
Se+ = �Eαβ σαβ + �Eαβ σαβ + dEk �Eαk ,
0 dEk
β
where the super-indexes (e), (c), and (i) indicates the excitation, capture and ion-
ization processes with the corresponding energy changes �Eαβ (e)
= εβ − εα ,
(i)
�Eαβ = εβ + v 2 /2 − εα, and �Eαk = Ek − εα, respectively.
(c)
We illustrate in Figure 9.1 distorted wave (DW) calculations of the

electronic stopping power Se+ for protons impacting on an atomic hydrogen
gas target. The only reactions that will give contributions to the energy loss
are single excitation:
H+ + H(1s) −→ H+ + H(nβ lβ mβ ), (19)
Figure 9.1 Electronic stopping power for proton charge-state Se+ on H target. Solid
lines, DW from Ref. 17; dashed lines, B1 from Ref. 29.
single ionization:
H+ + H(1s) −→ H+ + H+ + e − , (20)
and single electron capture:
H+ + H(1s) −→ H(nβ lβ mβ ) + H+ . (21)

The contributions from each of these processes are taken into account (see
review 22c and references therein). Electron capture has been considered
to final states with principal quantum number nβ 2 and excitation to
final states with nβ 3. It is found that electron capture dominates the low
energy part of the spectrum whereas ionization gives the main contribu-
tion at larger collision velocities. At these velocities, the excitation process
contributes to Se+ in a percentage of the order of 30% with respect to the
ionization reaction. A comparison is given with previous B1 predictions.29
For energies lower than 100 keV, they overestimate the DW results for the
three processes considered. This can be attributed to the limitations of B1
to describe the case of charged beams. However, at high enough velocities
results using both models converge to the same values for each one of the
processes considered.
In the case of a proton beam, protons can capture electrons and become
neutralized. The neutral beams may then be transformed back into protons
by electron loss processes and an equilibrium between charge-state com-
ponents of the beam is consequently attained. At high enough collision
energies, the beam consists of a mixture of neutral (H0) and proton (H+)
charge-states. The fraction at equilibrium of each one of these charge-states
is given by the expressions:
σL
f (H+ ) =
σC + σ L (22)
and
σC
f (H0 ) = , (23)
σC + σ L
where σC is the total capture cross section and σL is the total loss cross
section.
Denoting by Se0 the electronic stopping power for the neutral H0 beam,
the total mean stopping power is given by:
Se = f (H+ )Se+ + f (H0 )Se0 . (24)

To calculate Se0 several processes are considered. They are the ones of
single excitation:
H(1s) + H(1s) −→ H(1s) + H(nβ lβ mβ )
−→H(nβ ′ lβ ′ mβ ′ ) + H(1s), (25)
single ionization:
H(1s) + H(1s) −→ H(1s) + H+ + e −
(26)
−→H+ + e − + H(1s),
double excitation:
H(1s) + H(1s) −→ H(nβ ′ lβ ′ mβ ′ ) + H(nβ lβ mβ ) (27)
with nβ , nβ ′ �= 1; double ionization:
H(1s) + H(1s) −→ H+ + e − + H+ + e − (28)

and simultaneous excitation and ionization:
H(1s) + H(1s) −→ H(nβ ′ lβ ′ mβ ′ ) + H+ + e −

(29)
−→ H+ + e − + H(nβ lβ mβ ).
The possible formation of H− ions is neglected in this chapter (for the

contribution from the electron capture process H(1s)+H(1s)→H-(1s2)+H+,
see Figures 6.7 and 6.8 in Ref. 22c). Excited states with nβ , nβ ′ 3 are
considered, assuming that contributions from higher excited states can be
neglected.26, 30 Then, Se0 is calculated using the expression:
∞ (i)
dσαα ′ ,k
(e) (e) (i) (30)

Se0 = �Eαα ′ ,ββ ′ σαα ′ ,ββ ′ + dEk �Eαα ′ ,k ,
0 dEk
β,β ′
where the first term accounts for target and projectile excitation whereas
the second one for ionization from an initial state αα ′.
Equilibrium charge-state fractions are given in Figure 9.2. DW calcula-
tions are in reasonable agreement with experiments, where Bragg’s rule has
been used to obtain the data for H from measurements for H2.31 Results
are also in qualitative agreement with B1 calculations.29 It is shown that
the neutral beam fraction is larger than the charged beam fraction at low
collision energies while the situation is reversed for energies larger that
Figure 9.2 Equilibrium charge-state fractions for protons beams traversing H as a func-
tion of projectile energy. Solid line, DW calculation; (◦), B1 from Ref. 29; (•), experiments
for H2 targets from Ref. 31.
approximately 50 keV. T he dominance at low energies can be attributed to

the resonant condition of charge exchange.
The total mean electronic stopping power Se is shown in Figure
9.3 for the case considered above. DW calculations are compared with
B1 results29 and one-center coupled-channel calculations using atomic
orbitals (AO)15, 18 corrected to include double transitions. In fact, AO
results are obtained employing Se+ computations from Refs. 15, 32 while
Se0 was obtained with the B1 approximation.29 Experimental charge-
state fractions for H2from Ref. 31 were then used to determine the
mean stopping power. We have tested that DW results are not essentially
modified if experimental charge fractions are used instead of the theo-
retical ones. Also included in Figure 9.3 are Se recommended values from
Andersen and Ziegler33 and various measurements. All calculations give
a very good description of experimental data at energies larger than the
ones corresponding to the Se maximum. However, while DW and AO
give a good estimation of the corresponding peak, B1 largely overesti-
mates it. In principle, at energies lower than this peak, the best agreement
with experiments is given by the AO calculations whereas DW results
underestimate them. In order to explain this difference and considering
that for the H + H collisional system target and projectile are identical,
Figure 9.3 Total mean electronic stopping power of proton beams on H targets as a
function of projectile energy. Theory: solid line, DW calculations; dashed line, B1 results
from Ref. 29; dash-dot line, one-center AO results from Ref. 15 corrected for double
transitions; dash-dot-dot line, DW including exchange. Experiments for H2 targets: see
Ref. 17. Stars, tabulated values from Ref. 33.
we included in DW the corresponding exchange contributions. The

agreement now is very good.
4. THE CASE OF WATER MOLECULES

In the case of water molecules we must deal with the molecular
orbitals of the target. Different possibilities are to use simply Bragg’s rule,
where the molecule is treated as the ensemble of its separated atomic
compounds, or to treat it within a molecular representation. The first

method only requires the atomic cross sections or stopping power values.
The second case is much more difficult and for that purpose we have cho-
sen to describe the molecular orbitals using a method named Complete
Neglect of Differential Overlap (CNDO). In this method the molecular
orbitals are expanded in terms of atomic orbitals and both the weights of
the expansion and the molecular energies are consequently determined.34
This method was first proposed by Senger and Rechenmann35 who
employed the B1 approximation considering that the molecular cross sec-
tion for each molecular orbital can be given by the addition of the atomic
cross sections weighted with the corresponding parentage coefficients.
4.1 Differential and total cross sections

Double differential cross sections (DDCS) for single electron ionization
from water vapor produced by 0.3 MeV-protons and 2 MeV-alpha particles
are shown in Figure 9.4. CDW-EIS results are obtained using the CNDO
representation of the initial orbitals and effective Coulomb continuum
electron–residual target factors with corresponding effective charges given
by ZT∗ = −2n2 εα , where now εα is the orbital energy and n the principal

Figure 9.4 Doubly differential cross section for ionization of water vapor: (a) 0.3 MeV H+
and (b) 2 MeV He2+ ions. Theory: solid line, CDW-EIS; experiments: (•), extracted from
Senger and Rechenmann.35
quantum number corresponding to each one of the atomic components

that constitutes the molecular orbital considered. To describe the oxygen
orbitals Roothaan–Hartree–Fock wavefunctions were used.36 A good
agreement with experimental data is obtained for fixed ejection energies
as a function of the emission angle except in the backward direction where
the theoretical results underestimate the experimental data. In a previous
work,37 where hydrogenic wavefunctions have been used to describe the
initial orbitals within the CDW-EIS formalism, some differences with
experimental data were found, showing the importance of considering the
molecular character of the target.
Single differential cross sections (SDCS) for single ionization as a func-
tion of the electron ejection energy are shown in Figure 9.5. CDW-EIS
Figure 9.5 Single differential cross section for single ionization of water vapor by 70
keV and 100 keV proton impact as a function of the electron energy. Theory: solid line,
CDW-EIS molecular method; dashed line, CDW-EIS Bragg’s rule. Experimental data: (•),
from Ref. 38.
calculations are performed employing Bragg’s rule and the molecular

method with CNDO orbital wavefunctions. Proton beams with 70 keV
and 100 keV energies are considered, obtaining a good agreement with
existing measurements,38 except at energies lower than 4 keV, for which
the perturbative models are not adequate. Differences between theoretical
singly differential cross sections calculated using the two different molecular
methods are relatively small over all the energy domain considered.
Total cross sections (TCS) for the H+ + H2 O collision system are
presented in Figure 9.6. The one corresponding to single ionization is
obtained by integrating the SDCS (calculated using the molecular CDW-
EIS model) on the electron ejection energy. Bragg’s rule has been employed
to compute the TCS for electron capture. For both reactions CDW-EIS
calculations give a very good description of experiments39, 40 for collision
energies larger than 30 keV.
In Figure 9.7, ionization TCS for H+ projectiles at energies larger
that 100 keV show a very good agreement with experimental data.41, 42
TCS for impact of He2+ and C6+ ion beams are also presented. For He2+
the agreement between theory and experiments is also very good. For
C6+ it is difficult to give any conclusion due to the fact that only one
value exists for comparison. This value has been obtained by numerical
integration of the corresponding experimental single differential cross
section.43 Experimental results for Cl6+ from Ref. 44 are also included
Figure 9.6 Total cross sections for electron capture and single ionization of water
vapor by proton impact. Solid lines, CDW-EIS calculations. Experiments: open squares,
from Ref. 39; open circles and open triangles, from Ref. 40.
Figure 9.7 Total cross section for single ionization of water vapor by H+ , He2+, and C6+
ions. Theory: solid lines, CDW-EIS. Experiments: open squares and open circles from
Refs. 40 and 41; close circles from Ref. 45; close triangles from Ref. 43; open stars from
Ref. 42; close squares from Ref. 44; stars from Ref. 46.
Figure 9.8 Cross section ratios. Theory: Solid line, C6+; dashed line, He2+; dot dashed
line, H+.
in Figure 9.7. It is interesting also to analyze the cross section ratio

σαβ (ZP )/[ZP2 σαβ (ZP = 1)] (with σαβ (ZP ) the ionization cross section for
a projectile of charge ZP) to determine deviations from the ZP2-scaling law
of B1. The results presented in Figure 9.8 show a strong deviation from B1,
which increases with ZP. This behavior can be attributed to the binding
effect, which arises at low enough impact velocities. The electron moves in
the combined fields of the attractive projectile and the attractive residual
target. Thus, the electron feels a noticeable increasing of the binding energy
and its emission to continuum states is strongly reduced.13
4.2 Electronic stopping power

When Bragg’s rule is employed, the DW mean electronic stopping power
of atomic hydrogen and atomic oxygen are simply added, so that the stop-
ping power for H2 O is given by:
Se (H2 O) = 2Se (H) + Se (O), (31)
where for H the channels considered are the ones indicated in expression
19–21 and 25–29. In the case of atomic oxygen the reaction processes taken
into account are:
H+ + O −→ H+ + O+ + e − , (32)
H+ + O −→ H + O+ , (33)
H+ + O −→ H∗ + O+ , (34)
H + O −→ H + O+ + e − , (35)
H + O −→ H∗ + O, (36)
H + O −→ H+ + e − + O+ + e −
(37)
indicating with asterisk projectile excitation and ionization channels.
Oxygen excitation channels are not included, but their influence on the
determination of Se (H2 O) can be estimated using first-Born calculations
for excitation produced by electron impact.47
Golser and Semrad48 proved that for a mixture of hydrogen and

atomic gases, the electron stopping power deviates from Bragg’s rule.
This has been attributed to the fact that the corresponding equilibrium
charge-state fractions for the mixture differ from the ones predicted
by the additivity rule. Thus, a molecular Bragg’s rule is considered so
that the electronic stopping power for the neutral and charged beams
is taken as:
Se0 (H2 O) = 2Se0 (H) + Se0 (O), (38)
Se+ (H2 O) = 2Se+ (H) + Se+ (O). (39)
The corresponding equilibrium charge-state fractions can be written as:

σL (H2 O)
fH2 O (H+ ) = (40)
σC (H2 O) + σL (H2 O)
and
σC (H2 O)
fH2 O (H0 ) = (41)
σC (H2 O) + σL (H2 O)
with
σC (H2 O) = 2σC (H) + σC (O) (42)
and
σL (H2 O) = 2σL (H) + σL (O), (43)
where σC (J) with J = H2 O, H, O indicates electron capture cross sections
from water, atomic hydrogen, and atomic oxygen, respectively. In a similar
way σL (J) indicates the electron loss cross sections for hydrogen beams
impacting on the corresponding targets. Thus, the total mean electronic
stopping power can be obtained applying the expression:
Se (H2 O) = fH2 O (H+ )Se+ (H2 O) + fH2 O (H)Se0 (H2 O). (44)
Dissociative processes are not included in the calculations. In particular, the
CNDO approximation of the molecular orbitals is applied only to describe
the ionization channel:
H+ + H2 O −→ H+ + (H2 O)+ + e − . (45)

Figure 9.9 Equilibrium charge-state fractions of hydrogen beams in water vapor as

a function of the impact energy. Theory: solid line, H+ charge-state fraction; dashed
line, H0 charge-state fraction; both DW calculations. Experimental data: (•) and (◦),
from Ref. 39.
Expression (39) must be accordingly modified to incorporate this molecu-

lar description of the target ionization reaction.
Equilibrium charge-state fractions for impact of hydrogen beams on
water vapor are displayed in Figure 9.9. DW calculations are found to be
in qualitative agreement with existing experiments.39 Total mean electronic
stopping power is compared in Figure 9.10 with tabulated49 and experi-
mental data. At the maximum of the stopping power and at larger energies
molecular DW calculations give a good description of experiments and rec-
ommended values, except at the peak region when they are compared with
the tabulated values from Ref. 49. At energies lower than around 50 keV,
molecular DW underestimates the measured and recommended values.
Figure 9.10 Electronic stopping power of water vapor as a function of the impact energy
(molecular approximation). Theory: solid line, DW mean stopping power; dash dotted line,
contribution of the charged fraction to the mean stopping power; dotted line, contribu-
tion of the neutral fraction to the mean stopping power; dashed line, atomic DW Bragg’s
rule. Tabulations: open diamonds, from ICRU.49 Experiments data obtained from Ref. 19.
Contributions to the mean stopping power from the charged and neutral
beams are discriminated, showing that protons dominates at large energies
while neutral hydrogen plays the principal role at lower energies. Also in the
figure DW calculations employing Bragg’s rule are included. They largely
overestimate the stopping power peak region.
5. MULTIPLE IONIZATION OF WATER MOLECULES

The independent electron model has been used to study theoretically
multiple transitions. In this picture each electron evolves independently in
a mean field created by the projectile and the other electrons. Then, the
calculation of cross sections for multiple processes is done by employing a

binomial distribution where the computation of single-particle probabilities
as a function of the impact parameter is required.50 The probability Pm to
remove m electrons from one particular shell that contains N-equivalent
electrons can be expressed as:
m
N −m (46)
p(ρ)m 1 − p(ρ)

Pm (ρ) = CN ,
where m = 0, 1, 2, . . . , N and p(ρ) is the probability per electron of single

ionization of the shell. In this expression,
m N!
CN = (47)
m!(N − m)!
is a binomial coefficient. The cross section corresponding to this multielec-
tronic transition is given by:
+∞ +∞
(N ) m
N −m
dρ ρp(ρ)m 1 − p(ρ)

σm = 2π dρ ρPm (ρ) = 2π CN .
0 0
(48)
(N )
This approximated expression for σm contains information both on
the dynamics of one-single electron collision and on the statistics of several
equivalent electrons in the target. The collision dynamics is included by the
choice of a particular p(ρ), which depends on the single-electron potential
used, and on the binomial statistics employed.51 Usually, σm(N ) is named
exclusive cross section. If we do not explicit the final states of the remaining
N − m electrons, we obtain the inclusive cross sections, expressed as:
+∞
(N ) m
sm = 2πCN dρ ρp(ρ)m (49)
0
N −m
with m = 0, 1, 2, . . . , N . The exclusion of the term 1 − p(ρ) is
associated with the electrons that have a “passive” role in the collision. It
must be noted that nowadays, due to the improvements in experimental
techniques involving many particle detection in coincidence, the exclusive
measured cross sections can be determined.
Exclusive cross sections can be related to the inclusive cross sections and
vice versa using the binomial inverse pair relations:52
N
(−1)j−m Cjm sj(N )

σm(N ) = (50)
j=m
and
N
Cjm σj(N ) .

sm(N ) = (51)
j=m
In the m = 1 case, Cjm = j, this equation reduces to:
N +∞
(52)

s1(N ) jσj(N )

= = 2π N dρ ρp(ρ).
j=1 0
It provides an example of an inclusive cross section that can be directly

50,53 N (N )
measured. The sum j=1 jσj is a quantity directly obtained from
experiments where the total current produced by all different ionization
(N )
events is measured, without the knowledge of any particular σj . This
inclusive cross section is also called net ionization cross section or simply
total ionization cross section.54
We focus here the interest in the calculation of exclusive cross sections
for multiple ionization of the water molecule, which as it is well known
presents five i-molecular orbitals with an occupancy number Ni = 2 for
each one of them. After some algebra, exclusive probabilities Pq, for q-fold
ionization are given by the expression:
N1
,...,N5 5
q
CNi i [pi (ρ)]qi [1 − pi (ρ)]Ni −qi , (53)

Pq (ρ) =
q1 ,...,q5 =0;q1 +···+q5 =q i=1
where pi (ρ) is the probability to ionize one electron of the orbital iand qi
its corresponding ionization degree. Then, the q-fold ionization cross sec-
tion σq is given by:
+∞
σq = 2π dρ ρPq (ρ). (54)
0
Following the previous definition of the total or net ionization cross sec-
tion, we obtain:
 
5 � +∞ � +∞ 10 10
(55)
� � �
σT = 2πNi dρ ρpi (ρ) = 2π dρ ρ  qPq (ρ) =
 qσq .
i=1 0 0 q=1 q=1
As in the case of single ionization, the different probabilities that appear
in the previous equations are calculated employing the CDW-EIS approxi-
mation. We must mention that CDW-EIS has been used with success in
the calculation of double ionization of He by heavy ion impact55 and

direct multiple ionization of atoms and molecules by proton and multiply
charged ion impact.54, 56–58 To describe the different molecular orbitals of
Figure 9.11 Charge effects in the single particle probabilities corresponding to each
molecular orbital.
liquid water the CNDO approximation previously employed for water

vapor has been used59 together with RHF initial atomic bound states
wavefunctions. In this case, experimental orbital binding energies for liquid
water were considered.60 As the biological tissue is mainly composed by
liquid water we focuss our following analysis on this phase of the target.
Probabilities pi (ρ) are obtained using the method given in Galassi et al.61
Single ionization reduced probabilities ρpi (ρ) are presented in Figure
9.11 for the case of 70 MeV/u-proton and Ar18+ impact on liquid water.
Ionization from different orbitals are discriminated, being the reduced
probabilities normalized to unity at their maxima. For the 1b1 and 3a1
external orbitals, with binding energies of −0.32 and −0.44 a.u., respec-
tively, some shift toward larger impact parameter is shown to appear for
Ar18+ with respect to the H+ case. However, in this perturbation regime
(the Sommerfeld parameter being ν ≃ 0. 34 for Ar18+ and ν = 0. 019 for
H+) this shift almost disappears for the inner shells 1b2, 2a1, and 1a1, with
binding energies of −0.62, −1.19, and −19.84 a.u., respectively. Thus, the
impact parameter at which the reduced probability presents its maximum
is not very sensitive to the projectile charge. Moreover, we have tested that
Figure 9.12 Reduced multiple ionization probabilities as a function of the impact

parameter for 70 MeV/u–Ar18+ ions impact on liquid water.
the impact parameter position corresponding to the maxima above dis-

cussed has a weak dependence with the projectile velocity.56
In Figure 9.12 we observe how the ionization degree affects the reduced
probabilities ρPq (ρ). The case of 70 MeV/u-Ar18+ is analyzed showing that,
as expected, as the q-degree increases ρPq (ρ) decreases. Moreover, smaller
impact parameters (closer collisions) give the main contribution to the reac-
tion as the ionization degree increases.Thus, postcollisional interorbital Auger
and/or intraorbital Coster–Kronig ionization processes may dominate direct
ionization (we understand by this the ionization produced by the interaction
of the ejected electron with the projectile during the collision time) at high
enough energies as q increases. As an example and just to show this effect, in
Figure 9.13 we present q-ionization total cross sections for impact of protons
on a Ne gas (which presents an isoelectronic character with water).The cases
of q = 1, 2, and 3 are shown56 and compared with existing experimental
data.62, 63 The direct interaction of the projectile produces the ionization of
the target, generating thus holes in the orbitals that are filled through Auger
Figure 9.13 Single and multiple ionization cross sections of Ne by proton impact as a
function of the projectile energy.
and Coster–Kronig transitions. In order to estimate the influence of these

mechanisms, experimental photoionization probabilities were used.64 To do
that, it is assumed that the postcollisional ionization reaction is independent
of the nature of the primary vacancy production.65 This effect was also
shown to appear for impact of protons on O2 , N2, and CO molecular tar-
gets (see Ref. 57). Moreover, it has been proven for double ionization that
when these molecules collide with multicharged ions, postcollisional mech-
anisms dominate the direct one for values ν = Zp v 0. 3 of the Sommerfeld
parameter with a small shift to larger parameters ν as q increases.58 So, as the
projectile charge increases, the effect appears at larger impact velocities.This
behavior can be understood from the fact that as ZP increases direct single
ionization probabilities pi present a ZP2 dependence at high enough ener-
gies. As a consequence, the production of qi electrons by direct ionization
increases whereas the postcollisional emission gives the same contribution
for any value of the projectile charge, independently of the collision veloc-
ity. Thus, for Ar18+ the dominance of postcollisional effects should happen
for collision energies larger than approximately 90 MeV/u.
Figure 9.14 Multiple ionization cross sections as a function of the ionization degree for
proton and argon ions at different impact energies.
In Figure 9.14 exclusive σq-direct ionization cross sections are p resented for
q-degrees going from 1 to 4.The evolution of the cross section as the collision
energy of proton beams increases is shown.The slope of the curves that can be
drawn interpolating σq for the different q-values is more pronounced for larger
impact velocities,proving that multiple ionization plays a weaker role as the veloc-
ity increases. For 70 MeV/u-Ar18+ ions, exclusive cross sections take the values
σ1 = 9. 597×10−16 cm2 , σ2 = 0. 635×10−16 cm2 , σ3 = 0. 103×10−16 cm2 ,
and σ4 = 0. 014 × 10−16 cm2. However, multiple ionization contributes at
least by 14.56% to the total ionization cross section for this system. We will
show in the following that multiple ionization may play a determinant role in
the formation of free radicals with oxygen excess in water radiolysis. It must
be also noted that for the case considered, K-shell ionization followed by
Auger emission has been estimated to contribute with a value approximately
equal to 0. 06 × 10−16 cm2 (Ref. 66).
5.1 Linear energy transfer

While mean stopping power refers to the energy lost by the particle beam
traversing the surrounding media, linear energy transfer (LET) refers to the
energy absorbed by the media per unit of distance travelled by the ionizing
radiation. LET or restricted stopping power is defined as the ratio:
dE�ε (56)
L�ε = ,
dl
where dE�ε is the local mean energy absorbed by the media by means of
collisions involving transfer energies lower than a specific value �ε (cut
value) and dl is the distance traversed by the projectile. We understand by
“local” the consideration of absorbed energies limited up by the maximum
value �ε. Thus, L�ε is related to the part of the mean stopping power
involving transfer energies smaller than the cut value �ε. It can also be
related to the maximum distance reached (range) by electrons from the
projectile track.
Different cut levels are usually selected to separate delta rays as they are
appropriate for different reactions. L∞ is defined as the energy absorbed by
the media per unit of distance traversed by the projectile when all possible
energy transfers are considered. This quantity coincides numerically with
the stopping power.
5.2 Free radicals formation in water radiolysis

The ionization of water molecules during the irradiation of biological
matter with fast heavy ions produces the formation of free radicals. This
process is known as water radiolysis. Three different consecutive stages can
be distinguished in water radiolysis:
1. the physical stage where the interaction of the projectile with the water
molecules and the subsequent electronic cascade is produced during the
first femtoseconds,
2. the physico-chemical stage which deals with the fast dissociation pro-
cesses that results from molecular excitation and ionization and that
ends at the picosecond time scale,
3. the chemical stage that accounts for the evolution of the radicals and
molecular species during their diffusion through liquid, which ends at
the microsecond scale.
Usually Monte Carlo simulations have been developed to analyze the
time evolution of water radiolysis. In the physical stage the cross sections
for all interaction processes of the projectile and of the electrons with the
medium are calculated and used as input in the Monte Carlo codes.67–69
Water radiolysis is characterized by the radiolytical yields G. This is the
number of species created per unit of deposited energy. A large amount of
experimental work has been performed to measure the yields for solvated
electrons, OH, H2 O2, H, and H2, which are the main radicals produced by
low-LET radiations. Monte Carlo simulations gave an adequate quantita-
tive description of the experimental results, explaining also the different
mechanisms involved in the process.70 The situation is not the same for
high LET, where the formation of HO2 radicals has been experimentally
observed.71, 72 Moreover, recent experiments based on direct optical mea-
surements of HO2 /O− 2 produced in deaerated water for 70 MeV/u Ar
(LET = 280 keV/μm) and 77 MeV/u S (LET = 250 keV/μm) ions were
measured.73 However, they are lower than the ones expected by extrapola-
tion of previous experiments for similar LET.72 It was suggested74,75 that
multiple ionization of water could be responsible for this effect. The main
idea is that multiple ionization generates the production of O atoms, which
reacts with OH radicals to form HO2 ones. Moreover, the reaction of
HO2 with hydroxyl radical will produce molecular oxygen. The complete
process has been recently investigated by Gervais and coworkers68,69 con-
sidering water irradiation by carbon ions. To consider multiple ionization
Figure 9.15 Monte Carlo simulation of HO2 and HO2 + O2 yields at 1 μs as a func-
tion of LET. Lines are square fit to simulation results. For C6+ ions: dot dashed
line, MC simulation considering multiple ionization (MI) channels; dashed line,
MC simulation considering only single ionization (SI); closed circles, experiments
from.72 For Ar18+ ions: solid line, MC simulation considering MI; squares, experi-
ments from Ref. 73.
the independent electron model described in the previous section was

used. Their results are summarized in Figure 9.15. The inclusion of mul-
tiple ionization of liquid water in the physical stage of water radiolysis
plays a main role to explain experiments where yields of HO2 + O− 2 and
O2 + HO2 + O− 2 formation were measured. This can be concluded by
comparison with the case where only single ionization of water is included
in the Monte Carlo simulation. Resulting O2 + HO2 + O− 2 yields con-
sidering single and multiple ionization are included in the figure, showing
the increasing influence of multiple ionization as the linear energy transfer
is larger. If only single ionization is taken into account, negligible contri-
butions to the HO2 + O− 2 formation are obtained in comparison to the
yields obtained with the inclusion of multiple ionization.
6. CONCLUDING REMARKS

A detailed description of the distorted wave model has been revisited
to study the different electronic channels resulting from charged or neutral
heavy particles on atomic and molecular targets. This model includes the
Continuum Distorted Wave-Eikonal Initial State approximation to describe
both single electron ionization and single electron capture reactions for
the case of impact of charged particle beams. This approximation takes
account properly of the large range of Coulomb interactions present in
the different electronic processes. For this type of beams the Symmetric
Eikonal approximation is proposed to determine single excitation cross
sections. However, in the cases where they were evaluated for molecular
targets, estimations have been given from electron impact predictions. The
case of impact of neutral charged beams is analyzed using the first-order
of the Born approximation. Multiple electron processes that involve elec-
tronic reactions of both the projectile and target aggregates are also studied
employing this last theoretical description. So, the combination of all these
different approximations constitutes the distorted wave model.
The adequacy of this model is tested by comparison of its theoretical
predictions with existing experimental differential and total cross sections
not only for hydrogen atomic targets but also for water vapor molecules.
The completeness of the model allows to extend this analysis to the obten-
tion of equilibrium charge fractions and the consequent calculation of
mean electronic stopping power.
Multiple electron ionization of liquid water is investigated employing the
Continuum Distorted Wave-Eikonal Initial State approximation within an
independent particle model.The influence of postcollisional Auger-type and
Coster–Kronig-type emissions on total cross sections is also estimated. The
main role played by multiple ionization of water during the physical stage of
radiolysis (until a time of the order of one picosecond) on the formation of
free radicals with oxygen excess (after a few microseconds) is also revisited.
REFERENCES
1. Bethe H. Zur theorie des durchgangs schneller korpuskularstrahlen durch materie. Ann.
Phys. (Leipzig) 1930, 5, 325.
2. Bloch F. Zur bremsung rasch bewegter teilchen beim durchgang durch materie. Ann.
Phys. 1933, 16, 285.
3. Ziegler J. F. Stopping of energetic light ions in elemental matter. J. Appl. Phys. 1999,
85, 1249.
4. Ashley J. C.; Ritchie R. H.; Brandt W. Z13-Dependent stopping power and range con-
tributions. Phys. Rev. A 1973, 8, 2402.
5. Basbas G.; Brandt W.; Laubert R. Universal cross sections for K-shell ionization by
heavy charged particles. II. Intermediate particle velocities. Phys. Rev. A 1978, 17, 1655.
6. Brandt W.; Basbas G. Antiparticle excitation of atomic inner shells. Phys. Rev. A 1983,
27, 578.
7. Barkas W. H.; Dyer N. J.; Heckmann H. H. Resolution of the −-mass anomaly. Phys.
Rev. Lett. 1963, 11, 26.
8. Cheshire I. M. Continuum distorted wave approximation; resonant charge transfer by
systems. J. Phys. B 1978, 11, 3529.
10. Crothers D. S. F.; McCann J. F. Ionisation of atoms by ion impact. J. Phys. B 1983,
16, 3229.
11. Fainstein P. D.; Ponce V. H.; Rivarola R. D. A theoretical model for ionisation in
ion–atom collisions. Application for the impact of multicharged projectiles on helium.
J. Phys. B 1988, 21, 287.
12. Martínez A. E.; Deco G. R.; Rivarola R. D.; Fainstein P. D. K-Shell vacancy production
in asymmetric collisions. Nucl. Instrum. Methods B 1988, 34, 32.
13. Fainstein P. D.; Ponce V. H.; Rivarola R. D. ZP3 effects in the ionization of helium by
ion impact. Phys. Rev. A 1987, 36, 3639.
14. Andersen L. H.; Hvelplund P.; Knudsen H.; Møller S. P.; Elsener K.; Rensfelt K. G.;
Uggerhøj E. Single and double ionization of helium by fast antiproton and proton
impact. Phys. Rev. Lett. 1986, 57, 2147.
15. Schiwietz G. Coupled-channel calculation of stopping powers for intermediate-energy
light ions penetrating atomic H and He targets. Phys. Rev. A 1990, 42, 296.
16. Grande P. L.; Schiwietz G. Nonperturbative stopping-power calculation for bare and
neutral hydrogen incident on He. Phys. Rev. A 1993, 47, 1119.
17. Fainstein P. D.; Ponce V. H.; Martínez A. E. Distorted-wave calculation of stopping pow-
ers for light ions traversing H targets. Phys. Rev. A 1993, 47, 3055.
18. Olivera G. H.; Martínez A. E.; Rivarola R. D.; Fainstein P. D. Electron-capture contri-
bution to the stopping power of low-energy hydrogen beams passing through helium.
Phys. Rev. A 1994, 49, 603.
19. Olivera G. H.; Martínez A. E.; Rivarola R. D.; Fainstein P. D. Theoretical calculation of
electronic stopping power of water vapor by proton impact. Rad. Res. 1995, 144, 241.
20. Deco G. R.; Fainstein P. D.; Rivarola R. D. Symmetric eikonal approximation for elec-
tron excitation in ion–atom collisions. J. Phys. B 1986, 19, 213.
21. Rivarola R. D.; Fainstein P. D. Electron emission in collisions of highly charged ions
with atoms and diatomic molecules. Nucl. Instrum. Methods B 2005, 205, 448.
22. (a)Belkić, Dž. Review of theories on double electron capture in fast ion-atom colli-
sions. J. Math. Chem. 2010, 47, 1420.(b)Belkić, Dž; Gayet, R.; Salin, A. Electron capture
in high-energy ion-atom collisions. Phys. Rep. 1979, 56, 279.(c)Belkić, Dž. Quantum-
mechanical methods for loss excitation and loss-ionization in fast ion-atom collisions.
Adv. Quantum Chem. 2009, 56, 251.
23. Martínez A. E.; Bullrich J. A.; Maidagan J. M.; Rivarola R. D. The continuum distorted
wave-eikonal initial state model for single electron capture in ion–atom collisions.
J. Phys. B 1992, 25, 1883.
24. Stolterfoht N.; DuBois R. D.; Rivarola R. D. Electron Emission in Heavy Ion–Atom
Collisions; Springer-Verlag: Berlin, 1997.
25. Ramírez C. A.; Rivarola R. D. Electron excitation of hydrogenic targets in collisions
with ions or atoms. J. Phys. B 1993, 26, 3835.
26. Bates D. R.; Griffing G. Inelastic collisions between heavy particles. I: Excitation and
ionization of hydrogen atoms in fast encounters with protons and with other hydrogen
atoms. Proc. Phys. Soc. London Sect. A 1953, 66, 961.
27. Montenegro E. C.; Meyerhof W. E.; McGuire J. H. Role of two-center electron–
electron interaction in projectile electron excitation and loss. Adv. Atom. Mol. Opt. Phys.
1994, 34, 249.
28. Manson S. T.; Dubois R. D. Multielectron transitions resulting from interactions between
target and projectile electrons in ionizing collisions. Phys. Rev. A 1992, 46, R6773.
29. Dalgarno A.; Griffing G. W. Energy loss of protons passing through hydrogen. Proc. Phys.
Soc. London Sect. A 1955, 232, 423.
30. Hill J.; Geddes J.; Gilbody H. B. Electron capture, loss and collisional destruction of
5–26 keV metastable and ground-state hydrogen atoms in collisions with atomic and
molecular hydrogen. J. Phys. B 1979, 12, 3341.
31. Allison S. K. Experimental results on charge-changing collisions of hydrogen and

helium atoms and ions at kinetic energies above 0.2 keV. Rev. Mod. Phys. 1958, 30, 1137.
32. Grande P.; Schiwietz G. Impact-parameter dependence of electronic energy loss and
straggling of incident bare ions on H and He atoms by using the coupled-channel
method. Phys. Rev. A 1991, 44, 2984.
33. Andersen H.; Ziegler J. F. The Stopping and Ranges of Ions in Matter; Pergamon
34. Siegbahn K.; Nordling C.; Johansson J.; Hedman J.; Hedén P. F.; Hamrin K.; Gelius U.;
Bergmark T.; Werme L. O.; Manne R.; Baer Y. ESCA Applied to Free Molecules; North
Holland
35. Senger B.; Rechenmann R. V. Angular and energy distributions of ±-rays ejected from
low-Z molecular targets by incident protons and α particles. Nucl. Instrum. Methods B
1984, 2, 204.
36. Clementi E.; Roetti C. Roothaan–Hartree–Fock atomic wavefunctions: basis functions
and their coefficients for ground and certain excited states of neutral and ionized atoms,
Z 54. At. Data Nucl. Data Tables 1974, 14, 177.
37. Bernal M. A.; Liendo J. A. Inelastic-collision cross sections for the interactions of totally
stripped H, He and C ions with liquid water. Nucl. Instrum. Methods B 2007, 262, 1.
38. Bolorizadeh M. A.; Rudd M. E. Angular and energy dependence of cross sections for
ejection of electrons from water vapor. II. 15150-keV proton impact. Phys. Rev. A 1986,
33, 888.
39. Barnett C. F.; Ray J. A.; Ricci E.; Wilker M. I.; McDaniel E. W.; Thomas E. W. Atomic
Data for Controlled Fusion; Oak Ridge National Laboratory
40. Rudd M. E.; Goffe T. V.; DuBois R. D.; Toburen L. H. Cross sections for ionization of
water vapor by 74000-keV protons. Phys. Rev. A 1985, 31, 492.
41. Rudd M. E.; Kim Y.-K.; Madison, D. H.; Gallagher, J. W. Electron production in proton
collisions: total cross sections. Rev. Mod. Phys. 1985, 57, 965.
42. Wilson W. E.; Miller, J. H.; Toburen L. H.; Manson, S. T. Differential cross sections for
ionization of methane, ammonia, and water vapor by high velocity ions. J. Chem. Phys.
1984, 80, 5631.
43. Dal Capello C.; Champion C.; Boudrioua O.; Lekadir H.; Sato, Y.; Ohsawa D.

Theoretical and experimental investigations of electron emission in C6+ + H2 O colli-
sions. Nucl. Instrum. Methods B 2009, 267, 781.
44. Cocke, C. L. Production of highly charged low-velocity recoil ions by heavy-ion bom-
bardment of rare-gas targets. Phys. Rev. A 1979, 20, 749.
45. Ohsawa, D.; Kawauchi H.; Hirebayashi M.; Okada,Y.; Homma T.; Higashi, A.; Amano S.;
Hashimoto Y.; Soga, F.; Sato Y. An apparatus for measuring the energy and angular distri-
bution of secondary electrons emitted from water vapor by fast heavy-ion impact. Nucl.
Instrum. Methods B 2005, 227, 431.
46. Toburen, L. H.; Wilson, W. E.; Popovich, R. J. Secondary electron emission from ioniza-
tion of water vapor by 0.3- to 2.0-MeV He+ and He2+ ions. Radiat. Res. 1980, 82, 27.
47. Kazaks, P. A.; Canas, P. S.; Green, A. E. S. Electron-impact excitation and ionization of
atomic oxygen. Phys. Rev. A 1972, 6, 2169.
48. Golser, R.; Semrad, D. Failure of Braggs rule for a mixture of nonreacting gases. Phys.
Rev. A 1992, 45, R4222.
49. Stopping powers of protons and alpha particles, Report 49, International Commission
on Units and Measurements, Bethesda, MD, 1992.
50. Lüdde, H. J.; Dreizler, R. Comment on inclusive cross sections. J. Phys. B 1985, 18, 107.
51. Sant’Anna, M. M.; Montenegro, E. C.; McGuire, J. H. Inversion relations for exclusive
and inclusive cross sections within the independent electron approximation. Phys. Rev.
A 1998, 58, 2148.
52. Roberts F. S. Applied Combinatorics; Prentice-Hall: Englewood Cliffs, 1984.
53. Ford, A. L.; Reading, J. F.. In Atomic, Molecular and Optical Physics Handbook, ; Drake,
G. W. F., Ed.; Atomic, Molecular and Optical Physics Handbook; AIP: New York, 1996., pp
571–577.
54. Kirchner T.; Gulyás L.; Moshammer R.; Schulz, M.; Ullrich, J. Doubly differential
electron-emission spectra in single and multiple ionization of noble-gas atoms by fast
highly-charged-ion impact. Phys. Rev. A 2002, 65, 042727.
55. Fainstein P. D.; Gulyás, L.; Dubois, A. Single and double ionization of helium by highly
charged ions in the independent electron model. J. Phys. B 1998, 31, L171.
56. Galassi, M. E.; Fainstein, P. D.; Rivarola, R. D. Multiple electron emission from noble
gases colliding with proton beams, including postcollisional effects. Phys. Rev. A 2007,
75, 052708.
57. Tachino C. A.; Galassi M. E.; Rivarola R. D. Postcollisional effects in multiple ionization
of diatomic molecules by ion impact. Phys. Rev. A 2008, 77, 032714.
58. Tachino, C. A.; Galassi, M. E.; Rivarola, R. D. Role of Auger-type emission from
diatomic molecular targets interacting with fast multicharged ions. Phys. Rev. A 2009,
80, 014701.
59. Pople, J. A.; Segal, G. A. Approximate selfconsistent molecular orbital theory. II.

Calculations with complete neglect of differential overlap. J. Chem. Phys. 1965, 43, S136.
cross sections in liquid water. Radiat. Phys. Chem. 1998, 53, 1.
61. Galassi, M. E.; Abufager, P. N.; Martínez, A. E.; Rivarola, R. D.; Fainstein, P. D. The
continuum distorted wave eikonal initial state model for transfer ionization in
H+ , He2+ + He collisions. J. Phys. B 2002, 35, 1727.
62. DuBois, R.; Toburen, L. H.; Rudd, M. E. Multiple ionization of rare gases by H+ and
He+ impact. Phys. Rev. A 1984, 29, 70.
63. Cavalcanti, E. G.; Sigaud, G. M.; Montenegro, E. C.; Sant’Anna, M. M.; Schmidt-
Böcking, H. Post-collisional effects in multiple ionization of neon by protons. J. Phys. B
2002, 35, 3937.
64. Carlson, T. A.; Hunt, W. E.; Krause, M. O. Relative abundances of ions formed as the
result of inner-shell vacancies in atoms. Phys. Rev. 1966, 151, 41.
65. Stolterfoht, N. High resolution Auger spectroscopy in energetic ion atom collisions.
Phys. Rep. 1987, 146, 315.
66. Gaigeot, M. P.; Vuilleumier, R.; Stia, C.; Galassi, M. E.; Rivarola, R.; Gervais, B.; Politis,
M. F. A multi-scale ab initio theoretical study of the production of free radicals in swift
ion tracks in liquid water. J. Phys. B 2007, 40, 1.
67. Champion, C.; L’Hoir, A.; Politis, M. F.; Fainstein, P. D.; Rivarola, R. D.; Chetioui, A. A
Monte Carlo code for the simulation of heavy-ion tracks in water. Radiat. Res. 2005,
163, 222.
68. Gervais, B.; Beuve, M.; Olivera, G. H.; Galassi, M. E.; Rivarola, R. D. Production of
HO2 and O2 by multiple ionization in water radiolysis by swift carbon ions. Chem. Phys.
Lett. 2005, 410, 330.
75, 493.
70. Frongillo,Y.; Goulet, T.; Fraser, M. J.; Cobut,V.; Patau, J. P.; Jay-Gérin, J.-P. Monte Carlo
simulation of fast electron and proton tracks in liquid water – ii. Nonhomogeneous
chemistry. Radiat. Phys. Chem. 1998, 51, 245.
71. Burns, W. G.; Sims, H. E. Effect of radiation type in water radiolysis. J. Chem. Soc. Faraday
Trans. 1981, 77, 2803.
72. LaVerne, J. A.; Schuler, R. H. Track effects in the radiolysis of water: HO2 production
by 200–800 MeV carbon ions. J. Phys. Chem. 1992, 96, 7376.
73. Baldacchino, G.; Bouffard, S.; Balanzat, E.; Gardès-Albert, M.; Abedinzadeh, Z.; Jore, D.;
Deycardand, S.; Hickel, B. Direct time-resolved measurement of radical species formed
in water by heavy ions irradiation. Nucl. Instrum. Methods B 1998, 146, 528.
74. Ferradini, C.; Jay-Gérin, J.-P. Does multiple ionisation intervene for the production of
HO2 radicals in high-let liquid water radiolysis. Radiat. Phys. Chem. 1998, 51, 263.
75. Meesungnoen, J.; Filali-Mouhim, A.; Snitwongse Na Ayudhya, N.; Mankhetkorn, S.;
−
Jay-Gérin, J.-P. Multiple ionisation effects on the yields of HO2 /O2 and H2 O2
produced in the radiolysis of liquid water with high-let 12 C6+ ions: a Monte-Carlo
simulation study. Chem. Phys. Lett. 2003, 377, 419.
CHAPTER TEN
The First Born Approximation for

Ionization and Charge Transfer in
Energetic Collisions of Multiply
Charged Ions with Water
Christophe Championa, Jocelyn Hanssenb and Roberto D. Rivarolac
aUniversité Bordeaux 1, CNRS/IN2P3, Centre d'Etudes Nucléaires de Bordeaux–Gradignan, CENBG,
Gradignan, France
bLaboratoire de Physique Moléculaire et des Collisions, UMR CNRS 7565, Université de Lorraine, Metz,
France
cInstituto de Física Rosario, CONICET and Universidad Nacional de Rosario, Rosario, Argentina
Contents
1. Introduction 270
2. Ion-induced ionization and charge transfer cross sections in water: a review
of the existing data 272
2.1 Experimental background 273
2.2 Theoretical background 275
2.2.1 Semi-empirical approaches for describing the ionization process 275
2.2.2 Semi-empirical approaches for describing the charge transfer process 279
3. Molecular description of the water target 281
4. Born approximations 283
4.1 Theoretical description of the ionization process within the 1st Born
approximation 288
4.1.1 Basic formalisms 288
4.1.2 Final state description 292
4.1.3 Angular distributions of the secondary ejected electrons 294
4.1.4 Energy spectra of the secondary ejected electrons 299
4.1.5 Total ionization cross sections 301
4.1.6 From vapour to liquid water 302
4.1.7 Influence of the target description on the ionization cross sections 304
4.2 Theoretical description of the charge transfer process within the 1st Born
approximation 305
4.2.1 Basic formalisms 305
4.2.2 Total cross sections 308
5. Conclusions 309
References 310

270 Christophe Champion, Jocelyn Hanssen and Roberto D. Rivarola
Abstract
With the more and more regular use of ionizing radiations in medicine and more par-
ticularly in hadrontherapy, it is today necessary to describe—with the highest degree
of accuracy—the biological consequences of irradiations. To model the track-structure
of charged particles in biological matter and then to quantify the full spectra of radio-
induced cellular damages, Monte Carlo simulations are the preferential methods. The
latter consist in modelling the history of the ionizing particles by means of a large set of
input data, namely, the differential and total interaction cross sections in order to finely
describe the complete kinematics of the ion-induced collisions. In these conditions,
we clearly understand the necessity for the radiobiologists and the radiotherapists to
access to accurate cross sections—in particular for collisions with water target—the
latter being commonly used as surrogate of the cellular medium.
We here report a review of the existing 1st Born predictions for describing the ioniza-
tion and the charge transfer processes in the high-impact energy-regime particularly
investigated in hadrontherapy.
1. INTRODUCTION
Interactions of light ions (Z ≤ 10) with water are of great interest in
radiotherapy where protons (and carbon ions in rare cases) are today com-
monly used in cancer treatments.1 In fact, compared to photons, ion beams
have much more favourable dose-depth distributions, the concentration of
the energy deposits at the end of their range giving access to a better bal-
listic precision. However, the different treatments must be also compared in
terms of preservation of the healthy tissues and organs at risk in the tumor
vicinity. Indeed, to improve the dose-tumor control, it is essential to con-
form the delivered dose to the tumor: in this way, hadrontherapy—which
uses hadrons i.e., collimated beams of compound particles made of quarks
(like neutrons, protons, pions, and heavier ions)—was suggested to treat
radio-resistant tumours.2 Thus, protons, neutrons, and carbon ions have
been independently investigated for their dose-depth distribution particu-
larity, namely, the significant increase of the dose profile at the end of the
particle range: the so-called Bragg peak.
Nowadays, neutron therapies have been progressively stopped in the
most countries, essentially due to their poor depth-dose distribution what
implies high effective dose in the tumor as well as in the surrounding
healthy tissues.3 On the other hand, protons—which have been first used
in treatments in 1954 at Berkeley4—are today clinically used in a large
number of centres (with more than 40000 patients irradiated by January
2005) as reported by Amaldi and Kraft1 who predict that protontherapy
The First Born Approximation for Ionization and Charge Transfer 271
will continue to spread with in particular an increasing number of treated

patients reaching the value of about 1000–2000 for every 10 million inhab-
itants. Carbon ion beams have been till now less used in tumor treatments
and we only report in 2005 two centres, namely, the GSI Laboratory at
Darmstadt (with more than 250 patients treated at the end of 2004) and
the Japanese centre HIMAC (for Heavy Ion Medical Accelerator Centre)
with more than 2000 treated patients. In fact, despite the disadvantage to
have dose deposition after the Bragg peak due to the projectile fragmen-
tation, carbon ions present numerous advantages compared to protons in
radiotherapy like, for example, minor lateral and longitudinal straggling of
about three times smaller than for protons and a relative biological effi-
ciency (RBE) which increases at the end of the particle range. Thus, as it
has been the case for the evolution of the synchrotron radiation sources,
hadron therapy is today part of the medical business landscape and numer-
ous centres are actually under construction (or in a planning stage) like for
example in Europe (HIT at Heidelberg in Germany, CNAO at Pavia in
Italy and ETOILE at Lyon in France).
Protons and carbon ions are advantageous in Intensity Modulated
Hadron Therapy (IMHT) essentially for three physical properties: (i) the
existence of a maximum energy deposit in a well localized near-Bragg
peak region, (ii) a low diffusion during their penetration in matter, and (iii)
the possibility to be formed as narrow focused and scanned pencil beams
of variable penetration depth permitting an accurate and rapid irradiation
of the tumours. In these conditions, a beam of protons or light ions allows
highly conformal treatment of deep-seated tumours with millimetre accu-
racy, giving minimal doses to the surrounding tissues. Furthermore, note
that to reach depths of more than 25 cm in soft tissues, proton, and carbon
ion beams must have an initial energy greater than 200 MeV and 4500 MeV
(i.e., 375 MeV/u), respectively.
Several dose calculation algorithms have been then proposed for treat-
ment planning in hadron therapy (especially for proton therapy), generally
based on the pencil beam algorithm method, which consists in dividing the
therapeutic beam into narrow pencil beams. Dose from the pencil beam is
then usually calculated by using depth-dose distributions combined to the
finest description of the beam profile in terms of lateral widening (due to
the multiple scattering effects).
In this context, many dedicated Monte Carlo (MC) codes have been
proposed since the pioneer works of Berger:5 see for example the recently
published approaches developed in the framework of the SHIELD-HIT,
PHITS, and GEANT4 or FLUKA codes (for more details we refer the
reader to our recent work6 and references therein). Indeed, in the field of
hadron therapy, MC track structure simulations play an important rôle in
particular for understanding the spatial pattern of energy depositions as well
as the relative biological effectiveness of radiation qualities. However, MC
simulations essentially rely on the accuracy of the cross sections used for
describing the ion-induced interactions in matter and more particularly in
water, this target being commonly used for modelling the living medium
because of its abundance in biological cells (up to 70–80% in mass). In this
context, total and multi differential cross sections for ionization and charge
transfer in water appear as input data of prime importance, these two pro-
cesses bringing both the dominant contributions to the slowing-down of
charged particles in matter.
The present chapter deals with the existing theoretical models—here
limited to the perturbative treatments—as well as the available experimen-
tal measurements for ionization and electronic capture induced by heavy
charged particles in water. Besides, let us add that more sophisticated models
going beyond the perturbative framework may be found in the literature and
for more details we refer the interested reader to the recent review of Belkić.7
2. ION-INDUCED IONIZATION AND CHARGE TRANSFER

CROSS SECTIONS IN WATER: A REVIEW OF THE
EXISTING DATA
In spite of the broad range of interest, cross section measurements
on ion-induced collisions in water are scarce and a number of details have
not yet been investigated up to now. Indeed, most of the existing studies so
far have been restricted to electron impact ionization while cross sections
of water ionization by ion impact remain extremely rare (see for example
Refs. 8, 9 for a review on the ionization process by proton and α-particle
impact, respectively). Besides, the charge transfer process has been even
more rarely investigated and despite the numerous studies on total capture
cross sections for incident protons in various gases, there are today only few
results on partial electron capture cross sections. With regard to the theory,
only few theoretical works—quantum as well as classical—have been devel-
oped for describing the ionization and charge transfer processes in water.
The existing approaches are essentially based on semi-empirical models
which use—for a major part—analytical formulae fitted on experimental
measurements. These are reported in the following and compared in terms
of differential as well as total cross sections with the 1st Born theoretical
models hereafter reported.
2.1 Experimental background

Considering the ionization process, the first extensive work was given by
Toburen and Wilson10 who measured doubly differential cross sections
(DDCS) for 300–1500 keV protons and more recently for 300 to 2000 keV
He+ and He2+ ions.11 Later on, Rudd and co-workers performed experi-
ments on the ionization of water vapour by protons and He+ ion impact12
and reported total direct cross sections (without capture). More recently,
Bolorizadeh and Rudd have extended these pioneer works to the DDCS
measurements for the ejection of electrons during the ionization process
of water vapour by 15–150 keV protons.13 Later on, Gibson and Reid14
reported energy and angular distributions of secondary electrons ejected
from various gases including water vapour by 50 keV protons. Finally, mul-
tiple ionization and fragmentation of water have been recently studied by
Werner et al.15 for fast protons and He+ ions by using a position- and time-
sensitive multi-particle detector: a good agreement was also observed with
the previous measurements of Rudd and co-workers12 for the total cross
sections (TCS). Furthermore, Gobet et al.16 have determined a complete
set of cross sections for the ionization of water molecule by proton impact
in the energy range 20–150 keV including total and partial cross sections
for H+ and H2O+ fragment production. They also found TCS in good
agreement with the measurements of Rudd et al. and those of Werner et al.
Recently, Gobet et al.16 have also presented a full set of absolute partial cross
sections corresponding to H+, H2O+, OH+, O+, and O2+ which allows
a detailed comparison between their cross sections for direct ionization of
water by proton impact and the electron impact ionization data of Straub
et al.17 Concerning helium ions, the literature remains very poor. We can
nevertheless cite the pioneer work of Rudolph and Melton18 on 2.2 MeV
He2+ ions, the extensive graphical representation given by Toburen et al.11
in terms of secondary electron yields, differential in ejected electron energy
and emission angle for 0.075 to 0.5 MeV/u helium ions, both singly and
doubly charged and finally, the very recent differential cross sections reported
by Ohsawa et al. for the energy and angular distributions of secondary
electrons produced in collisions of 6- and 10-MeV/u He2+ ions in water
vapour and more recently for 15-MeV/u α-particles (He2+ ions).19 In the
latter, DDCS were presented and compared in the low-energy region to
the Toburen’s data by means of Fano plots and very good consistency was
found. Singly differential cross sections (SDCS) were finally deduced and
compared to the semi-empirical model introduced by Rudd.20 Considering
carbon ions, the literature is much poorer since to our knowledge only few
experimental measurements have been reported. Let us cite the group of
Montenegro and co-workers21 who have published a study of great interest
for the field of hadron therapy on the water molecule fragmentation by car-
bon ions at the distal region of the Bragg peak. In this work, measurements
of the positive ion fragments resulting from the water fragmentation after
ionization and electron capture have been reported, allowing for the first
time a quantitative determination of the energy lost by carbon ions at the
end of the particle range. Finally, note that Dal Cappello et al.22 have recently
reported an extensive set of experimental doubly and singly differential cross
sections for 6 MeV/u C6+ ions provided by Ohsawa and co-workers.
Considering the capture process, many experimental measurements
have been reported since the pioneer works on protons in 1970s. Let us
cite the work of Rudd and co-workers on electron capture induced by
5–150 keV protons in many gases like H2 and O2 and later on the studies
of Rudd et al.12 on the determination of electron and positive ion produc-
tion cross sections from which electron capture cross sections for protons in
water vapour were extracted. More recently, Lindsay et al.23 reported mea-
surements of absolute differential cross sections for charge transfer scatter-
ing of 0.5-, 1.5-, and 5-keV protons by H2O at laboratory scattering angles
between 0.01° and 2.60°. In their recent letter, Gobet et al.16 have also
reported total and partial electron-capture cross sections for 20–150 keV
protons in water vapour. Finally, high-energy protons (15–3500 keV) have
been also recently investigated by Luna et al.24 who studied the water
molecule dissociation by proton and hydrogen impact. Concerning helium
ions, only few measurements have been reported. Thus, Rudd and co-
workers12 reported absolute total cross sections for 5–450 keV He+ ions
and for 5–150 keV/u α-particles whereas the low-energy regime was
only very recently investigated. Let us cite for example the study given by
Abu-Haija et al.25 where total cross sections for 0.1–1 keV He2+ ions on
triatomic molecules like H2O were reported, that of Greenwood et al.26
where single and double capture processes were studied in the energy range
0.05–5 keV/u and finally that of Seredyuk et al.27 where experimental
one-electron capture cross sections have been carried out in the range
0.025–12 keV/u. Heavy ions have been up to now only rarely studied. We
can nevertheless mention the very recent study of Mawhorter et al.28 who
reported absolute single, double, and triple charge exchange cross sections
for highly charged ions among which Cq+ ions (q = 5,6) colliding with
molecular species like CO, CO2, and H2O. However, this study remains
limited to low impact energies (7.0 qkeV), its aim being the interpretation
of the x-ray emissions from comets as they interact with the solar wind.
Similarly, Bodewits and Hoekstra29 have reported absolute total cross sec-
tions for 0.1–7.5 keV/u collisions of O6+ ions on H2O molecules.
2.2 Theoretical background

2.2.1 Semi-empirical approaches for describing the ionization process
On the theoretical side, we essentially find semi-empirical models which
are limited, for the major part, to the calculation of singly differential and
total ionization cross sections. Among them we can first cite the Rudd’s
model—initially developed for protons—which offers a parameterization of
secondary electron spectra after ion impact. This model may be used over
the entire range of primary and secondary energies and is in the form of an
analytic equation with parameters determined from experimental data. It
is based on a simple version of the binary-encounter approximation (BEA)
equation modified to yield the correct high-energy asymptotic dependence
on energy in agreement with the Bethe equation prediction. It has been
further modified by the use of the promotion model at low energies.20 In
brief, the approximation made consists in treating the collision—between
a projectile and a single target electron—as a classical one. The nucleus and
the remaining target electrons play no rôle except that of providing a bind-
ing energy for the ejected electron, the energy transfer E and the kinetic
energy Ee being related by E = Ee + I, where I denotes the binding energy
of the ionized subshell of the target. The justification for using a classical
model lies in the fact that doubly differential cross sections for Coulomb
scattering between two particles are the same when calculated using either
classical physics or quantum mechanics. Thus, the singly differential cross
sections are simply expressed as (in atomic units)
dσRudd S/I F1 + F2 w
= 3
, (1)
dEe 2R (1 + w) [1 + exp(α(w − wc )ṽ)]
where I refers to the binding energy (expressed in atomic units) whereas R
denotes the Rydberg energy.
The reduced quantities w and ṽ are given by

w = Ee /I , ṽ = Ei /I , (2)
where Ei denotes the incident projectile energy and with

wc = 4ṽ 2 − 2ṽ − 1/8I , S = π N /I 2 , (3)
N being the number of electrons of the ionized molecular subshell whereas
F1, F2, and α are seen as adjustable fitting parameters. Thus, we have
L1 = C1 ṽ D1 / 1 + E1 ṽ (D1 +4)

F1 (ṽ) = L1 + H1 , (4)
H1 = A1 ln(1 + ṽ 2 )/(ṽ 2 + B1 /ṽ 2 )
and

L2 = C2 ṽ D2
F2 (ṽ) = L2 H2 /(L2 + H2 ), (5)
H2 = A2 /ṽ 2 + B2 /ṽ 4 .
The different needed parameters are reported hereafter (see Table 10.1).
Total ionization cross sections were then simply deduced by numerical
integration of Eq. (1) over the kinetic energy transfers Ee, the latter ranging
from a minimum value equal to 0 to a maximum value (Ee )max defined
as (Ee )max = (M4m+m0 MP ∼ 4m0
2 Ei = MP Ei where m0 and MP refer to the electron
P 0)
and projectile mass, respectively. Relatively good agreements were then
observed for singly differential cross sections, in particular for ion energies
lower than 300 keV/u whereas large discrepancies were recently reported
by Uehara and Nikjoo30 for greater energies (up to about 30% at 2 MeV),
what led the authors to propose a polynomial fit of existing experimental
measurements for expressing the total ionization cross sections in their
numerical track-structure code—called LEAHIST—recently developed for
modelling the transport of α-particles in water.
The second semi-empirical and well-documented model—called
HKS model developed by Hansen, Kocbach, and Stolterfoht31—consists
in describing the ionization process in the impact parameter 1st Born
approximation. In this approach, the initial and the final electron states are
described by means of a hydrogenic function and a plane wave, respectively,
i.e., without taking into account the electron momentum in its bound state.
However, due to singularities observed when the ejected electron energy
Table 10.1 List of the semi-empirical parameters used in the Rudd’s model for fitting
the singly differential cross sections of proton-induced ionization of isolated water
molecules
A1 B1 C1 D1 E1 A2 B2 C2 D2 α
0.97 82.0 0.40 -0.30 0.38 1.04 17.3 0.76 0.04 0.64
tends to zero, further empirical fittings were employed—simultaneously to

other semi-empirical terms32, 33—to finally provide the well-known HKS
model, which can be consequently regarded as an impact parameter-plane
wave Born approximation (IP-PWBA).
In this approach, the authors have described the electron emission by
heavy ions of charge Zion in terms of impact parameter dependent prob-
abilities P(b,θe,Ee) where b refers to the impact parameter whereas θe and
Ee denote the ejection angle and the kinetic energy transfer, respectively.
The DDCS (per target electron) were then obtained by means of analytical
integrations of the corresponding probabilities over the impact parameter,
yielding a simple expression recently reported in Ref. 33 as
3
d2 σHKS Zion 2 32 αc2

= , (6)
d�e dEe vi 3π αkc3 αc2 + (K m − k̂t cos θe )2
where the function in the square brackets describes the binary-encounter

maximum that resembles a Lorentzian whose width is governed by
vi2

αc = α 1. 0 + 0. 7 2 , (7)
vi + ke2
√
where α = 2I corresponds to the mean initial momentum parameter,
I being the binding energy of the corresponding ionized molecular subshell.
In Eqs.(6) and (7), vi denotes the projectile velocity while Km = Km
αc
represents the normalized minimum momentum transfer with
Km = (α 2 + ke2 )/2vi.
Similarly, the authors defined the quantities kt (with k̂t = kt /αc) and
kc as small modifications of the momentum of the outgoing electron ke,
respectively given by

(8)

kt = k2 + 0. 2α 2 vi /α , kc = k2 + 2α 2 / ln(2v 2 /α 2 ).
e e i
Let us note that the original equations proposed by Hansen and

Kocbach31 imply kc = ke due to the fact that in the peaking approximation
one neglects the mean momentum of the bound electrons in compari-
son to that of the outgoing electron. However, as clearly emphasized by
Stolterfoht et al.32 this approximation produces a singularity in the low-
electron energy regime, which can be simply removed by adjusting the
value of kc to fit the model results to those of the Born approximation.
To do that, numerous expressions have been proposed like that reported in

Eq. (8). We can also mention that reported in Ref. 32, namely,

3α 2 /2 (9)
kc = ke2 + .
[ln(2vi2 /α 2 )]2/3
In the same state of mind, Bernal and Liendo34 have recently slightly
modified the original version of the HKS model, essentially to avoid the
“non-physical” descending jump appearing for each electron binding
energy in the SDCS due, in major part, to the use of the arctangent term
(see Eq. (11)). To do that, the authorsproposed
a modified expression for
2
the DDCS, denoted in the following d σHKS and expressed as
d�e dEe B−L
Zion 2 32
2 3
d σHKS 1
= . (10)
d�e dEe B−L vi 3π αkc3 1 + (Km + k̂t cos θe )2
However, as underlined by Uehara and Nikjoo30 and more recently by

Bernal and Liendo,34 the DDCS also obtained give only a limited agree-
ment with the experimental data, especially in the backward and forward
angle regions what could be relied to the use, in this model, of hydro-
genic wave functions instead of realistic bound electron wave functions to
describe the target electron initial state.
Furthermore, the semi-empirical SDCS provided by the different exist-
ing HKS approaches, namely, the original version dσHKS and the recently
dEe
modified version dσHKS may be respectively recast as
dEe B−L

Zion 2 16N

dσHKS 2k̂t
= arctan
dEe vi 3αkc3 k̂t 1+K m2 − k̂t2

m + k̂t ) + 3(K
5(K m + k̂t )3 5(K m − k̂t )3
m − k̂t ) + 3(K
+ −
2[1 + (Km + k̂t )2 ]2 2[1 + (K m − k̂t )2 ]2
(11)
and
2
m + k̂t )3

dσHKS Zion 8N m + k̂t ) + 3(K
5(K
=
dEe B−L vi αkc3 k̂t 3[1 + (K m + k̂t )2 ]2
m − k̂t )3
m − k̂t ) + 3(K
5(K
−
3[1 + (K m − k̂t )2 ]2

+ arctan(K
m + k̂t ) − arctan(K
m − k̂t ) , (12)
where the factor N corresponds to the number of target electrons per

molecular subshell.
Finally, note that for both the original HKS model and its more recent
version, the total ionization cross sections are—like for the Rudd’s model—
numerically obtained by integration of Eq. (11) or Eq. (12), respectively.
2.2.2 Semi-empirical approaches for describing the charge

transfer process
When the energy of the charged particles becomes sufficiently low
(Ei ≤ 100 keV/u), the ionization probability diminishes rapidly, and the cap-
ture process becomes the predominant energy-loss pathway. However, only
rare experimental measurements have been up to now reported for vapour
water. To overcome this lack, many semi-empirical simulations have been
proposed for modelling the electron capture process.
The first one refers to the approach proposed by Rudd et al.12 which
consists in expressing the charge transfer cross section σ as σ = σ+ − σ−
by means of adjustable parameters for fitting the measured cross sections
of positive (ion) and negative (electron) charge production (σ+ and σ−,
respectively). Thus, Rudd et al propose to write the (ionization + capture)
cross section σ+ and the pure ionization cross section σ− as
1
σ± = 1 1
, (13)
(σlow )± + σhigh
where the low- and high-energy components σlow and σhigh (expressed in
atomic units) were fitted as
(σlow )− = 4π Cki2D ,
(σlow )+ = 4π [Cki2D + F], (14)
σhigh = 4π [A ln(1 + ki2 ) + B]/ki2 ,
where ki refers to the incident momentum. The used parameters are

reported below (see Table 10.2).
The resulting cross sections agree well with the experimental data for
proton energies of 1-100 keV but nevertheless exhibit large discrepancies
Table 10.2 List of the semi-empirical parameters used in the Rudd’s model for fitting
the experimental cross sections of positive and negative charge production in proton
on water molecule collisions
A B C D F
2.98 4.42 1.48 0.75 4.80
with the experimental measurements of Toburen et al.35 for energies above

100 keV. Moreover, they tend to a constant value for energies below 1 keV
what is in complete contradiction with most of the existing experimental
values which increase with decreasing proton energies.
The second model was more recently suggested by Dingfelder et al.36
who expressed the charge transfer cross sections by analytical formula
(straight lines for low and high proton energies on a doubly logarithmic
scale, both connected by a power law) where the parameters were chosen
by considering the experimental data of Lindsay et al.23 and of Dagnac
et al.37 for low and medium energies, and those of Toburen et al.35 for
higher energies. In detail, the charge transfer cross section is expressed by
σ = 10Y (X ) (15)
where X = log(τ ) (with τ in eV) and
Y (X ) = [a0 X + b0 − c0 (X − x0 )d0 �(X − x0 )] �(x1 − X )

(16)
+ (a1 X + b1 ) �(X − x1 ),
where Θ(x) represents the Heaviside step function. The parameters a0, b0,
a1, and b1 determine the low-energy straight line and the high-energy one,
respectively, whereas c0 and d0 refer to the power law in between, connected
to the low-energy straight line at x0. The connection point x1 to the high-
energy line and the parameter b1 are calculated as
1/(d0 −1)
x1 = a0c0−a
d0
1
+ x0 (17)
b1 = (a0 − a1 )x1 + b0 − c0 (x1 − x0 )d0
by using the first derivative. The used parameters are reported below (see
Table 10.3).
Finally, let us mention a third model based on the analytical functions
developed by Green and co-workers38 and fitted to the experimental data
of Toburen et al.35 and Dagnac et al.37 The obtained formula is nowadays
Table 10.3 List of the semi-empirical parameters used in the Dingfelder’s model for
fitting the total charge transfer cross sections for protons in water
a0 b0 c0 d0 a1 b1 x0 x1
-0.180 -18.22 0.215 3.550 -3.600 -1.997 3.450 5.251
Table 10.4 List of the fitting parameters used in the semi-empirical expression of
total electron capture cross sections for protons in water39
λ J (keV) α (keV) Ω Λ C (keV)
H2 2.0 1.215 4084.0 0.271 4.80 75.8
O2 2.0 0.057 1038.0 0.258 3.50 125.0
commonly used in many track structure codes and is presented as follows

(with Ei in keV)
(Zα)� (Ei − I )λ
σ = σ0 (�+λ)
, (18)
J (�+λ) + Ei + (Zα)� Eiλ (Ei /C)�
where α, λ, J, Ω, Λ et C are different fitting parameters (see Table 10.4),

I being the ionization threshold (in keV) of the impacted target and
σ0 = 10−16 cm2.
For heavier charged particles like He+ ions, the electron capture cross
sections were also fitted to experimental data, namely, those of Rudd et al. 12
for impact energies between 1 and 100 keV/u and those of Sataka et al.40
for impact energies between 75 keV/u and 500 keV/u. For He2+ ions,
Rudd et al.12 have proposed—in the energy range 5 to 150 keV/u—least-
squares fitted total cross sections whereas for energies below 5 keV/u and
above 150 keV/u, these functions were extrapolated assuming a smooth
transition at the boundaries.30
Finally, let us note that in this last model the total cross sections for
water target were simply obtained by means of the additivity rule from the
relationship σH2 O = σH2 + 0. 5σO2 , what remains still today one of the most
common approach for describing the ionizing processes (capture as well as
ionization) in complex target molecules like water. However, this technique
completely conceals the molecular structure of the target whose influence
may become primordial especially at the multi differential scale as it will
be discussed hereafter.
3. MOLECULAR DESCRIPTION OF THE WATER TARGET

Describing the molecular ionization by a quantum-mechanical
approach remains a difficult task essentially due to the multi-centred nature
of the target. To overcome this difficulty, many attempts were made in par-
ticular by representing the molecular cross sections as a weighted sum of
the cross sections of the different components of the molecule, namely, the
well-known Bragg’s additivity rule whose applications to vapour water ion-
ization were first proposed by Olivera et al.41 for proton beams. A second
technique, called complete neglect of differential overlap (CNDO) consists
in writing the molecular orbitals in terms of atomic orbitals of the atomic
constituents, namely, the orbitals H1s, O1s, O2s, and O2p (see Ref. 42). Note
that this description was also used in the binary-encounter-dipole model
developed by Kim and Rudd43 for providing singly and total ionization
cross sections for a large set of molecules impacted by electrons. However,
in both these models the calculated doubly differential cross sections exhibit
some discrepancies with the experimental data at small angle regions. The
authors linked these discrepancies to the fact that the electronic popula-
tions were not correctly reproduced in these two descriptions, especially for
the calculations within the Bragg’s rule framework. Finally, a third method
consists in describing the populations of the target by means of molecular
orbitals constructed from a linear combination of atomic orbitals in a self-
consistent field approximation (MO-LCAO-SCF). In the past, Champion
and co-workers have successfully applied this description for treating the
ionization of simple molecules like CH4, NH3, and H2O by electrons44
as well as by light-ion impact, namely, H+, He2+, and C6+ ions.8, 9, 22 In
these works, the authors have used the simple molecular description pro-
vided by Moccia who reported one-centre ground state wave functions for
molecules of the type HXn (see Ref. 45). The molecular orbitals were then
expressed in terms of Slater-like functions all centred at a common origin
coinciding, in this case, with the X nucleus. In fact, for these molecules, the
electronic density is mainly governed by the central atom. Thus, provid-
ing suitable analytical wave functions is quite similar to the atomic case, in
which the basis set consists of functions all referred to a common origin
centred at the one nucleus.
Finally, note that the problem of evaluation of the above-cited multi-
centre integrals depends on the type of basis functions used. Indeed,
whereas it appears that there are no convenient and practical ways to evalu-
ate such integrals for more than two non-aligned centres when Slater-type
functions are used, it is worth noting that the use of Gaussian functions for
the radial part decreases the difficulties even if it is clear that the Gaussian
basis set needs probably 40% more such functions to achieve comparable
results. Under these conditions, the ten bound electrons of the water
molecule are distributed among five one centre molecular wave functions
corresponding to the five molecular orbitals denoted 1b1, 3a1, 1b2, 2a1,
and 1a1, respectively, and whose respective ionization potentials (or bind-
ing energies)—denoted Ij in the following (with j varying from 1 to 5,
respectively)—are equal to 0.4954 a.u., 0.5561 a.u., 0.6814 a.u., 1.3261 a.u.,
and 20.5249 a.u. (see Ref. 44 for more details). Let us note that these
molecular wave functions refer to the calculated equilibrium configura-
tions, i.e., to the geometrical configurations which, among many others
considered, give the minimum of the total energy, and agree with the
experimental data in terms of HOH angle, bound O–H length, 1st ioniza-
tion potential IP, and electric dipole moment μ (see Ref. 45).
4. BORN APPROXIMATIONS
In the framework of quantum mechanics, the treatment of atomic
(or molecular) ionization by charged particles may be divided into different
classes according to the intensity of the interaction between the target elec-
tron and the incoming and outgoing charged particle. Thus, Rudd et al.46
distinguished in their review three categories, namely, (i) a first one in
which the electron-target is a strong interaction whereas the electron-
projectile is weak, (ii) a second one which treats the electron-target inter-
action as strong whereas the electron-projectile interaction is strong with
the incoming particle and weak with the outgoing one, and finally (iii) a
third class where the electron-target and the electron-projectile are both
strong interactions.
In this scheme, it clearly appears the necessity to treat the ionization
process by means of different approaches according to the investigated
kinematical conditions. Among them, we distinguish the perturbative
approaches—called Born approximations—from the continuum distorted
wave methods whose main characteristic is that all two-particle Coulomb
interactions are explicitly contained in the initial and final state wave
functions. This last approach has been successfully employed for treating
ionization as well as charge transfer in many atomic and molecular target
systems including water and will be the subject of the next chapter of the
present review.
In the perturbative regime, which corresponds to the major part of the
cases investigated in the field of hadron therapy, the interaction between
the electron and the projectile is ignored in the formation of the electron
wave function and, as a result, appears only in the transition matrix ele-
ment. Furthermore, it is assumed that the projectile may be expressed as a
plane wave: this well-known approximation is referred as the plane wave
Born approximation and is commonly denoted PWBA. As reported by

Rudd et al.,46 this approximation will certainly be invalid if the projectile
is deflected through a large angle (>1°) and is therefore usually called the
standard straight-line trajectory semi-classical approximation. The distinc-
tion between the different calculations of this type then lies in the treat-
ment of the ejected electron. Thus, we speak of typical PW when the latter
is described by a plane wave whereas the distorted wave (DW) refers to a
description of the ejected electron by means of a wave function calculated
as the eigenfunction of the potential used for representing the residual tar-
get ion (see in the following). Therefore, when letter B is appended to PW
and DW, the resulting acronyms PWB and DWB will refer to the plane
wave Born and the distorted wave Born approximation, respectively. Finally,
if a pure Coulomb potential is preferred to a numerical potential obtained
from Hartree–Fock wave functions for the target in question, we speak of
the Coulomb Born wave function denoted by CW. In the present work,
where we are interested in the impact of bare ions, we consider initial wave
functions that include separately the long range Coulomb interactions of
the projectile with the electron to be ionized and of the projectile with the
rest of the target.Thus these wave functions verify correct boundary condi-
tions only in the initial channel, that’s why we will use for them the acronym
CB1 to distinguish with the case of B1, where asymptotic conditions are
not satisfied. Thus, throughout we will talk about CB1-PWB, CB1-CWB,
and CB1-DWB models. To avoid confusion with the existing literature on
ion-atom collisions,47 we re-emphasize that the term CB1, presently used
in the hybrid models such as CB1-PWB, CB1-CWB, and CB1-DWB,
will refer to the correct boundary condition satisfied only in the entrance
channel. However, when the letters CB1 are employed, as a stand-alone
acronym, they would refer to the first Born approximation with the cor-
rect boundary conditions fulfilled in both the initial and final states.We will
then observe in the sequel that the differences between the CB1-CWB and
the CB1-DWB models are more pronounced for DDCS than for SDCS
and more particularly at large ejected angles where the CB1-DWB results
generally underestimate the predictions given by the CB1-CWB model.
Moreover, note that the CB1-DWB-type calculations can be performed
by using different distortion potentials for the final states. In the following,
we will essentially speak about the numerical potential issued from the
above-cited wave functions taken from Moccia45 but the procedure here
followed could be easily generalized to any potential. In brief, the final
state of the active electron—described by means of wave function distorted
by the initial-state ion—is simply deduced from the initial self-consistent

field one-centre-expanded molecular wave functions taken from Moccia 45
by removing the active electron orbital without any modification of the
remaining orbitals: this is the well-known frozen-core approximation. A
spherically averaged local radial potential is then deduced from the charge
distribution of the target ion represented by the remaining orbitals. The
final state distorted wave for the ionized electron is then numerically calcu-
lated as an eigenfunction of this radial potential and expressed as distorted
waves, not necessarily orthogonal to the initial bound state wave functions.
Note that an orthogonalization procedure could be nevertheless used by
using, for example, the Schmidt-type procedure.
Finally, it is important to note that the three 1st Born approximations
type here reported, namely, the CB1-PWB, the CB1-CWB, and the CB1-
DWB, remain valid as far as the projectile velocity is fast in comparison to
the orbital velocity of the active electron. In these conditions, plane waves
are appropriate for describing the projectile whereas their use is no more
valid in large angle scattering conditions. In addition, the present choice for
the final state wave function of the active ejected electron should be satis-
factory as long as the interaction between the scattered projectile and the
ejected electron is weak and the ionization takes place fast enough so that
the inactive electron orbitals do not relax before the active electron leaves the
target, what corresponds to the major part of the cases encountered in the
radiotherapy field. On the contrary, when the ejected electron is leaving
the target in the forward direction with a speed comparable to that of the
projectile, the interaction between the ejected electron and the projectile
can be not necessarily weak.Thus, the ejected electron angular distributions
(DDCS) exhibit a big rise at low ejected angles i.e., in the forward direc-
tion. This process is called continuum electron capture to the continuum
(ECC) or charge transfer to the continuum. Using theoretical models that
do not include this effect becomes then inappropriate and results in a
substantial underestimation of the doubly differential ionization cross sec-
tions (see in the following). Indeed, describing this effect requires taking
into account of second order terms of perturbative theory, what is beyond
the scope of the present chapter. However, Salin has proposed in Ref. 48
a multiplicative correction factor—called Salin factor in the following—to
account for the electron capture to the continuum. Under these conditions,
the DDCS clearly present the ECC peak and improves the agreement with
the experimental observations at small ejected angles: see for example the
1st Born calculations provided by Madison49 for He targets impacted by
100 keV and 200 keV protons. Let us note nevertheless that the agreement
becomes less satisfactory at large angles as reported in Ref. 49 and shown
below for water target.
Furthermore, in view of the centre picture concept introduced by
Stolterfoht et al.50—which consists in describing the electron production
mechanism in terms of Coulomb centres associated to the different nuclei
partners involved in the final state of the collision—we distinguish the free-
electron Born approximation (or fully plane-wave Born approximation,
PWBA) in which a plane wave is used for describing the outgoing electron
whose interaction with the nuclei is completely neglected from the 1st
Born approximation (B1-CW) which includes the full target interaction in
the final state. Then, the free-electron Born approximation which contains
neither a projectile nor a target centre is usually referred to a zero-centre
case whereas the B1-CW is associated with the target-centre case. Finally,
note that if the interaction of the outgoing electron with the target nucleus
is neglected (as in the PWBA framework), the ionization process is referred
to as a binary-encounter electron emission, which may be described by a
classical theory. In this approach, the target electron is assumed to be free
and then interacts with the incident projectile independent of the target.
Theoretical DDCS were then successfully compared to experimental data
for 100 keV and 300 keV protons on He especially in the region of the
binary-encounter peak as ke increases whereas large discrepancies were
reported at small ke for small and large electron scattering angles (see Rudd
et al.51).
Considering now the charge transfer process, let us first mention that
most of the existing theories in fast ion-atom collisions fall naturally in one
of the two following categories: (i) plane wave theories where the exact
initial and the final states are described by means of products of bound states
and plane waves describing the relative motion of the colliding partners and
(ii) distorted wave theories, which represent the exact initial and final states
by products of bound states, plane waves, and distortion factors.52
In this context, Oppenheimer and then Brinkman and Kramers53 were
the first authors to propose a theoretical model for calculating the total
cross section in the 1st Born approximation framework via the well-known
OBK model. In this later, such as in many other theories, the calculations
are simplified by approximating the many-electron collisional system by a
one-active electron system. Thus, the remaining electrons screen the tar-
get nuclear charge seen by the projectile and the active electron and then
provide an effective potential where the active electron evolves, while the
“non-captured” electrons are considered as being passive: this is the frozen

core approximation previously reported for describing the ionization process.
A few years later that the OBK model was proposed, Jackson and Schiff 54
pointed out a strong disagreement between the experimental observations
for protons in hydrogen gas and the theoretical OBK predictions, these
latter being approximately four times greater than the experimental results
at 100 keV and even higher at lower energies. The authors attributed this
failure to the description of the perturbation Hamiltonian. Indeed, this
latter consists in two terms, namely, the Coulomb interaction between the
target active electron and the incident particle and the Coulomb interac-
tion between the target nucleus and the incident projectile. However, as
underlined by Jackson and Schiff, the OBK approach only considers the
first component of the Hamiltonian neglecting the second one, whose
contribution was assumed to be of the order of m0/MP (i.e., negligible)
by G. C. Wick (see added note in proof in Ref. 54). Thus, the agreement
between theory and experiment for protons in hydrogen became better for
all energies above 25 keV, even though the Born approximation was still
used. Similarly, Omidvar et al.55 investigated the rôle played by the inter-
nuclear interaction in the behaviour of the cross sections in the 1st Born
approximation for protons on helium and argon. Comparisons were then
reported between different options of calculation, namely, by using different
perturbative potentials, including: (i) only the projectile-electron interac-
tion, (ii) the projectile-electron and internuclear interactions, and (iii) the
projectile-electron interaction added to an effective internuclear potential
where the target nuclear charge was chosen equal to one. Strong discrepan-
cies were then observed exhibiting in particular that the inclusion of the
perturbative potential (iii) gave the best fit of the experimental observations
for helium and argon targets.
Later on, many authors among which Dollard,56 Mapleton,57 Carpenter
and Tuan58 have pointed out divergences in the Born series. More recently
Dewangan and Eichler59 showed that singularities appear in the transi-
tion matrix calculated within the Born series. These divergences come
from the contribution of the intermediate elastic channel. Higher-order
theories, like the impulse approximation (IA)60 and the strong potential
Born approximation (SPB)61 were also developed providing an improve-
ment of the agreement with experiments, essentially due to the non
negligible contribution of the high-order terms to the forward scattering.
However, divergences from elastic contributions were shown to appear in
the SPB model.59, 62 To correct the above reported divergences a proper
treatment of the Coulomb boundary conditions was proposed. Moreover,

it was shown that the IA violates correct asymptotic conditions. A detailed
description of the adequate treatment of boundary conditions was given
by Belkić and co-workers63 who have shown that the effective internuclear
additional term in the perturbative potential corresponds in fact to the
correction of the coulombic tail of the projectile-electron interaction. The
long-range Coulomb potentials are, in general, present even when the scat-
tering particles are at infinite separation from each other, what causes dis-
tortion of unperturbated channel states which, in turn, yields modifications
of the perturbation potentials. Numerous investigations were proposed to
include into both the initial- and the final-state wave functions the cor-
rect asymptotic behaviour at infinite internuclear separation, confirming
that a first Born approximation satisfying Coulomb boundary conditions
(referred in the literature as the CB1 model) was quite accurate in repro-
ducing the experimental data at no so high energies, where two-step colli-
sions (Thomas mechanism64) give an important contribution to differential
cross sections. In the last case a second order of the Born series with correct
boundary conditions is necessary. In particular, Belkić reported in Ref. 65
that the sole introduction of the Coulomb phase into the exit channel
within the Jackson–Schiff matrix element for H+-Ar charge exchange
reduced the calculated total cross sections by about 200 times, leading to
a better agreement with the experiment. In brief, these corrections refer
to: (i) the Coulomb distortion of the unperturbated channel states and
(ii) the perturbating potentials. Furthermore, it is worth noting that the
CB1 transition amplitude coincides fortuitously with the Jackson–Schiff
one for proton–hydrogen collisions, i.e., when the projectile charge equals
the target charge.
Finally, let us note that such considerations support the development
of first order continuum distorted wave (CDW) and continuum distorted
wave-eikonal initial state (CDW-EIS) approximations, the last one being
extensively detailed in chapter 9 of the present book.
4.1 T
heoretical description of the ionization process
within the 1st Born approximation
4.1.1 Basic formalisms
Let us here consider the direct ionization process schematized by
X + H2 O → X + H2 O+ + e − , (19)
where the initial state of the system is characterized by an incident projec-

tile X (of charge ZP and of mass μ) with an initial momentum ki , which
ionizes a stationary water molecule whereas the final state of the system
is characterized by a scattered projectile of momentum ks and an ejected
electron of momentum ke.
In these conditions, the four-fold differential cross sections correspond-
ing to this process, namely, differential in the energy of the ejected electron
ke2 /2, differential in the direction of the ejected electron dΩe, differential in
the energy of the scattered electron ks2 /2µ, and differential in the direction
of the scattered particle dΩs are simply given by

d4 σ µ2 2 ki2 ks2 ke2 � − k�e )2
(K
= |Tfi | δ − |I | − − − ,
ks2 dks d�s ke2 dke d�e 4π 2 ki 2µ 2µ 2µ 2Mion
(20)
where I denotes the ionization energy i.e., the binding energy of the
molecular subshell ionized.The transferred momentum—from the incident
projectile to the water target—is here denoted K � = k�i − k�s, whereas Mion
refers to the mass of the residual water ion.
Furthermore, the energy and the solid angle of the scattered projectile
being not measured in the common experiments, we usually define the
doubly differential cross sections (DDCS) as

d2 σ µ2 k e

k 2 k 2 k 2 ( �
K − �
ke )2
= |Tfi |2 δ i
− |I | − s − e − d�s .
d�e dEe 4π 2 ki 2µ 2µ 2µ 2Mion
(21)
Moreover, let us note also that since the amplitude Tfi decreases fast
when the momentum transfer K increases only small values of K will con-
tribute to the integration over dΩs in Eq.(21). Thus, we can neglect the
term (K� −k�e )2 in the delta function of Eq. (21) and then recast the DDCS as
2Mion
d2 σ
2
µ2 ke ks2 ke2

ki
= 2
|Tfi | δ − |I | − − d�s . (22)
d�e dEe 4π 2 ki 2µ 2µ 2µ
In this expression, the scattering matrix Tfi ≡ Tfi (K

� ) is given by
� ) = �f− Vi |ψi � ,

Tfi (K (23)
where f− and ψi are the exact final scattering wave function with correct
incoming boundary conditions and the initial wave function, respectively,
which both contain the information on the dynamics of all the particles
involved in the total collisional system.
In Eq. (23), the perturbative potential Vi depends on the choice of ψi.
Thus, extending the formulation given by Corchs et al.66 for single elec-
tron capture from molecules to the single ionization one (where all inter-
actions between the projectile and the target nuclei are treated into the
eikonal approximation, the target nuclei being supposed to remain in their
initial positions during the collision) and after a tedious algebra, the transi-
tion amplitude as a function of the impact parameter b may be written as
n

Afi (�b) = bc2iZP Zc /vi afi (�b), (24)
c=1
where the straight line version of the impact parameter approximation

�b = �r0 + v
� i t has been used with vi for the incident projectile velocity and
where r0 indicates the projectile position (see Figure 10.1).
In Eq. (24), bc defines the impact parameter with respect to the nucleus
c of the target (composed in total by n nuclei) whereas Zc designates the
corresponding nuclear charge. The transition amplitude afi (b) is a purely
electronic one which accounts the interaction of all target electrons with
the projectile.
Moreover, considering that all passive electrons, namely, the target
electrons which are not ionized, all remain as frozen during the collision,
Fainstein et al. demonstrated in Ref. 67 that the transition amplitude Afi (b)
reported in Eq. (24) could be recast as
∞ n

i
�
Afi (b) = exp − dtVs (�r0 ) bc2iZP Zc /vi ãfi (�b). (25)
vi −∞
c=1
ki
e-
b
r0 r1
H 2O target
Figure 10.1 Schematic representation of the water ionization process.

In this expression, the static potential Vs (r0 ) contains the information

on the interaction between the projectile and the passive electrons and is
given by
N
ZP
Vs (�r0 ) = �ζi ({�rk })| − |ζi ({�rk })� , (26)
|�rk − �r0 |
k=2
where ζi ({�rk }) refers to the wave function describing the (N-1) passive elec-
trons whose positions are represented by the vectors rk , the ejected (active)
electron position being denoted r1.
In these conditions, ãfi (�b) appears as a one-active electron transition ampli-
tude, where all the interactions between the considered (active) electron
and the passive ones are accounted. Furthermore, it is important to note
that the interactions of the projectile with the passive electrons and with
the target nuclei are here (i.e., in Eq. (25)) considered at all orders and not
only at the first one, what is valid for any model used to describe the stud-
ied reaction. Note that it remains valid for any electronic transition such as
ionization, electron capture, or excitation.
Finally, the scattering matrix element is obtained using the well-known
Fourier transform

� ) = ivi d�b exp(i K
Tfi (K � . �b)Afi (�b). (27)
It is evident from Eq. (25) that the electronic transition is provoked by the
interaction of the projectile with the active electron while the interactions
of the projectile with the target nuclei and with the passive electrons will
only affect the projectile trajectory. Thus, in the case where integration
over all projectile scattering angles is considered in order to calculate the
corresponding DDCS, only the interaction of the projectile with the active
electron must be taken into account. In the following we will focus the
study on this case.
The electronic capture process being investigated in the next session
let us first consider the single ionization reaction. The asymptotic behavior
of the Coulomb projectile-active electron interaction must be included
into the initial and final wave functions in order to avoid the presence of
disconnected diagrams provoking divergences in the transition amplitudes
(and then in the scattering matrix elements) coming from the contribu-
tion of intermediate elastic channels.59, 62 Thus, if multiplicative asymptotic
eikonal phases depending on r0 are included into the initial asymptotic and
final exact wave functions, the one active-electron transition amplitude

ãfi (�b) can be written under the form
ãfi (�b) = b−2iZP /vi âfi (�b). (28)
The corresponding projectile-active electron perturbative potential appearing

in âfi (�b) will be reduced to its short range part. The long range part can be
thus also neglected from the calculations if the cross sections of interest do not
depend on the projectile scattering angle. Accordingly with this, we can calcu-
late the corresponding scattering matrix element using Eq. (27).
In a first-order approximation, we can consider in Eq. (23) �f− ∼ = ψf
where ψf is a specific asymptotic final wave function, so that
�) ∼ (1) �
Tfi (K = Tfi (K ) = �ψf | Vi |ψi � . (29)
Thus, following our analysis given above, this first-order matrix element
can be reduced to the one-active electron expression

(1) (K
T � ) ∼
= φ( �s , �r0 )ζf (�r1 ) V
k i
�
φ( k i , �
r0 )u(�
r1 ) , (30)
fi
where u(r1 ) describes the one-active electron orbital wave function of the
molecular target and φ(kd , r0 ) (with d = i,s) a projectile plane wave (for
the incident and the scattered projectile, respectively) whereas ζf (r1 ) refers
to the ejected electron wave function. The perturbative potential V i here
reported corresponds to the above-mentioned short range part of the
projectile-active electron interaction and is given by
ZP Z P
i =
V − , (31)
r01 r0
with �r01 = �r0 − �r1.
4.1.2 Final state description

In the context of water ionization, several models have been recently
tested by Champion and co-workers8, 9 for describing the final state of
the collision in order to go beyond the pure simple CB1-CWB approxi-
mation: a first one—referred as CB1-DWB in the following—where the
ejected electron is described via a distorted wave function instead of a pure
Coulomb wave function whereas the scattered projectile remains described
by means of a plane wave, and a second model where the scattered projec-
tile as well as the ejected electron are both described by Coulomb waves
(CB1-2CW). Let us add that in these two models, the effective charges seen
by the escaping electron and ion—defined as the effective ionic charges—

were chosen equal 8
to 1 by Boudrioua et al. whereas some authors defined
them as ZT = −2n2 εi (with εi = −Ii) where n denotes the princi-
pal quantum number of each atomic subshell that constitutes the initial
bound state of the active electron in the molecule. However, it has been
demonstrated that this discrepancy essentially affects the inner subshells of
the target for which the cross sections have minor contribution into the
ionization process. Finally, Champion and co-workers have tested a third
approach which consists in exhibiting a correct asymptotical Coulomb
three-body wave function by describing the final state by the product of
three Coulomb waves taking into account the interaction between the
scattered ion and the nucleus, the interaction between the ejected electron
and the nucleus, and the scattered ion-ejected electron interaction. Note
that for this model (hereafter denoted CB1–3CW for three-body first Born
with correct boundary conditions) we have the well-known asymptotically
correct Coulomb three-body wave function for the ejected electron and
the scattered proton in the field of the residual ion.
Furthermore, it is important to note that the bound wave functions
used in the present water molecular description correspond to a given
orientation of the target molecule, commonly expressed via the Euler
angles (α, β, γ). Thus, the above-cited multi-differential cross sections
refer to the ionization of a single oriented water molecule and need then
to be averaged over the Euler solid angle d�Euler = sin β dα dβ dγ to
be comparable to those experimentally measured, what may be analyti-
cally easily carried out thanks to the property of the rotation matrix.44
Note that this procedure is obviously performed for each of the N = 5
orbitals of the water molecule and the multiple differential cross sections
presented in the sequel correspond to “global” differential cross sections
i.e., obtained by summing up all the sub-shell contributions labelled j and
then weighted by the number Nelec of electrons per orbital, i.e., Nelec = 2.
Thus, we write
N =5
d2 σ
2
d σ
= Nelec . (32)
d�e dEe d�e dEe j
j=1
In further steps, singly-differential and total cross sections (SDCS

and TCS, respectively) are obtained by numerical integrations of the
DDCS over the ejected direction de and the kinetic energy transfer Ee,
respectively.
4.1.3 Angular distributions of the secondary ejected electrons

Figure 10.2 shows a comparison between the experimental DDCS taken
from Toburen and Wilson10 and the results provided by Boudrioua et al.8
within the two above-cited 1st Born models, namely, the CB1-CWB and
the CB1-DWB approaches for a proton incident energy of 0.5 MeV and for
ejected energies Ee ranging from 12 eV to 750 eV. Except for small angles
and at some fixed ejected electron energies, namely, 100 eV, 250 eV, and
750 eV, the CB1-CWB model (solid line) reproduces with a good agree-
ment the experimental observations.This disagreement may be explained by
the fact that in the present CB1-CWB calculations, the authors didn’t take
into account the above-cited ECC process which causes an increase of the
doubly differential cross sections at low ejected angles. This process is all the
more conspicuous that the velocity of the ejected electron is close to that
of the scattered proton and may be seen as a classical capture of a bound
0
10
Ee = 12 eV
-1
10
Doubly differential cross sections (a.u.)
Ee = 50 eV
-2
10
Ee = 100 eV
-3
10
-4 Ee = 250 eV
10
-5
10
Ee = 750 eV
-6
10
0 30 60 90 120 150 180
Ejection angle (deg)
Figure 10.2 Doubly differential cross sections for single ionization of water vapour by
0.5 MeV protons for different ejected electron energies as a function of the ejection
angle. Theory: CB1-CWB model (solid line), CB1-DWB model (dashed line). Experiments:
circles from Toburen and Wilson.10
electron from the target molecule into a continuum state of the proton. The
“captured” electron is then emitted in the moving frame of the proton.
Furthermore and as already reported by Rudd et al.,46 Boudrioua et al.,8
clearly underline the fact that the CB1-DWB and CB1-CWB methods yield
nearly identical cross sections except at large angles where the difference
between the two approaches is more pronounced. The authors also report
that the two approaches begin to differ from each other by a small amount
especially with increasing ejection angles and energies at fixed incident energy.
To improve the agreement between the experimental and the theoreti-
cal DDCS for protons, Boudrioua et al. have recently proposed to describe
the water ionization process within the CB1-2CW approach i.e., in
describing the scattered projectile by means of a Coulomb wave function.8
However, as shown in Figure 10.3 for incident proton energy of 0.5 MeV
-1
10
-2
10
-3
10
-4
10
0 30 60 90 120 150 180
Figure 10.3 Doubly differential cross sections for single ionization of water vapour
by 0.5 MeV protons impact with the ejection of a 250 eV electron as a function of the
ejection angle θe. The lines correspond to the present different models used: CB1-3CW
model (solid line), CB1-CWB model (dotted line), CB1-CWB model with the Salin fac-
tor (dashed line), CB1-2CW model (dash and dotted line) whereas the experiments of
Toburen and Wilson10 are represented by circles.
and an ejected electron of 250 eV, it appears that both the CB1-CWB and
CB1-2CW models are unable to explain the large enhancement of the
DDCS for small angles because none of these models treats the interaction
between the scattered proton and the ejected electron.
In a second step, Boudrioua et al.8 have also introduced a first-order
correction factor S, namely, the well-known Salin factor68 into the CB1-
CWB model, which consists in introducing the mechanism of electron
transfer to the continuum as a multiplicative factor given by
2πγ
S= , (33)
1 − e −2π γ
with γ = Z/p where Z denotes the dynamical effective charge defined as
Z = 1 - p/vi where p� = k�e − v � i is the final electron velocity measured
from a reference frame fixed at the projectile.
Thus, when the electron moves with a velocity close to that of the pro-
jectile (p @ 0), it travels in a continuum state of the projectile which lies just
above the threshold of the corresponding continuum spectrum, whereas
for an electron ejected in a soft collision (ke << ki), Z @ 0 what indicates
that the ejected electron moves only in the field of the target. An excel-
lent agreement with the experimental data was then observed in particular
at small ejection angles whereas a less satisfactory accord was reported for
larger angles as already observed by Madison49 for helium targets impacted
by 100 keV and 200 keV protons.
In addition, in their study, Boudrioua et al.8 have studied the water
ionization process in the CB1-3CW framework and reported a better
agreement in the large ejected angle region. However, it was noted that
the obtained DDCS largely overestimated the experimental observations
at small angles. As a matter of fact, the CB1-3CW model exhibits a correct
asymptotic Coulomb three-body wave function for the scattered proton
and the ejected electron in the residual ion field. Finally, let us note that the
Salin factor depends on a dynamical effective charge whereas in the CB1-3CW
model the projectile nuclear charge is considered.
Similar observations were reported in Figure 10.4 for α-particles
impinging on water molecules.
Very good agreement was then observed between the experimental and
the theoretical CB1-CWB results in all cases, except for the lower ejec-
tion energy reported i.e., for Ee = 19.2 eV for which they obtained DDCS
overestimate the experimental data essentially at small angles (θe < 60° and
θe > 120°). For θe > 60°, the observed agreement becomes better for each
-3
10
Doubly differential cross sections (10 cm /eV.sr)

Ee=19.2eV
2
Ee=38.5eV
-4
10
-16
Ee=96.2eV
-5
10
Ee=192eV
-6
10
Ee=385eV
Ee=673eV
-7
10
0 30 60 90 120 150 180
Figure 10.4 Doubly differential cross sections for water ionization by α-particles of
6.0 MeV/u. The experimental data (solid circles) of Ohsawa et al.19 are compared to the
theoretical results (CB1-CWB model) for 6 different ejected electron energies, namely,
Ee = 19.2 eV, 38.5 eV, 96.2 eV, 192 eV, 385 eV, and 673 eV.
ejected energy here considered exhibiting in particular the existence of

the well-known binary-encounter peak whose localisation is here well
reproduced, namely, from about 85–90 at Ee = 19.2 eV to about 79–80° at
Ee = 673 eV.
Finally, Figure 10.5 compares the CB1-CWB DDCS (solid line)
reported by Dal Cappello et al.22 to recent experimental measurements for
6 MeV/u C6+ ions ejecting electrons of different energies ranging from
9.6 eV to 384 eV. A good agreement is generally reported over the whole
range of ejected electron energies with nevertheless slight discrepancies
for θe @ 70°–80°. However, as previously, we observe that the position
of the binary-encounter peak is well reproduced. Furthermore, we have
reported in Figure 10.5, a comparison between the CB1-CWB DDCS of
10-1
Doubly differential cross sections (10 cm /eV.sr)

10-2
10-3
2
10-4
-16
10 -5
θe = 20°
10 -6 θe = 40° (x10 )
-1
-2
10 -7 θe = 60° (x10 )
-3
θ = 80° (x10 )
10 -8 e
10 -9
10-10 θ = 100° (x10 )

e
-4
10-11 θ = 120° (x10 )

e
-5
10-12
-6
θ = 140° (x10 )
e
-7
θ = 160° (x10 )
10-13 e
10-14 1
2 3
10 10 10
Ejected electron energy (eV)
Figure 10.5 Doubly differential cross sections for single ionization of water vapour
by 6.0 MeV/u C6+ ions for different ejection angles (20° < θe < 160°) as a function of the
ejected electron energy. The theoretical CB1-CWB results are reported by a solid line
whereas the semi-empirical HKS results are represented by a dashed line for the origi-
nal HKS model31 and by a dotted line for the expression given by Bernal and Liendo.34
The experimental measurements are represented by solid circles. Multiplicative factors
are used for a better clarity in the plot.
Dal Cappello et al. and those obtained by the two HKS models discussed
above, namely, the original version from Hansen and Kocbach31 (see Eq.
(6)) and that slightly modified and proposed by Bernal and Liendo34 (see
Eq. (10)). Minor discrepancies may be observed with in particular a better
agreement exhibited by the original HKS model version (dashed line) with
nevertheless an overestimation for the large ejection angles, which is all the
more important that the energy increases (Ee > 200 eV). In comparison, the
modified HKS model underestimates the experimental results at backward
angles, what is mainly due to the use of hydrogenic wave functions to
describe the bound electron initial state instead of an accurate molecular
wave function as it is the case for the Dal Cappello calculations. Indeed, the
agreement reported by the authors seems better even if we note a slight
overestimation in particular for the high energies (Ee > 200 eV) when the

ejection angle tends towards the binary-encounter peak (θe @ 70°–80°).
4.1.4 Energy spectra of the secondary ejected electrons

By integration of the DDCS with respect to emission angle θe, the singly
differential cross sections may be obtained. Figure 10.6 shows a comparison
between the available experimental data10, 13 and the theoretical results per-
formed by Boudrioua et al.8 in the CB1-CWB model for protons in water.
A reasonably good agreement between the experiments and the present
1st Born results may be observed especially for ejected electron energies
greater than 10 eV, the Auger electron peak being obviously not reproduced
0
10
-1
10
Singly differential cross sections (10 cm /eV)
2
-16
-2
10
-3
10
-4
10
-5
10
0 1 2 3
10 10 10 10
Figure 10.6 Singly differential cross sections for ejection of electrons by protons in
water vapour as a function of ejected electron energy. Experimental data taken from
Toburen and Wilson10 (open triangles for 0.5 MeV and open circles for 1.5 MeV) and
from Bolorizadeh and Rudd13 (open squares for 100 keV).
since not included in the present used model. For lower ejected energies
(Ee < 10 eV), the agreement is obviously less satisfactory, the kinematics
being far from the domain of applicability of the 1st Born approximation.
Similarly, Figure 10.7 shows a comparison between the experimental
SDCS measurements for C6+ ions reported in Dal Cappello et al.22 and
the CB1-CWB calculations as well as the existing semi-empirical (Rudd
and HKS) predictions.
A good agreement is generally observed for both the results with in
particular a regular underestimation in the low-energy regime and a con-
stant overestimation for high energies (Ee > 200 eV). However, we notice
that the best agreement is still obtained with the original HKS model in
10 0
Rudd's model
cm /eV)
CB1-CWB results
2
-16
Singly differential cross sections (10
10 -1
HKS model
10 -2
10 -3
10 -4 0 1 2 3
10 10 10 10
Figure 10.7 Singly differential cross sections for ejection of electrons by 6.0 MeV/u
C6+ ions in water vapour as a function of the ejected electron energy. The theoretical
CB1-CWB results are reported by a solid line whereas the semi-empirical HKS results
are represented by a dashed line for the original HKS model31 and by a dotted line
for the expression given by Bernal and Liendo.34 The experimental measurements
are represented by solid circles. Additionally, the Rudd’s results are represented by a
dash-and-dotted line.
the whole ejected electron energy range. Furthermore, we clearly observe

that the CB1-CWB theoretical results generally tend to overestimate the
experimental measurements, especially at low-energies (Ee < 200 eV). In
addition, we note that the results provided by the Rudd model (dash-dot-
dotted line) are in remarkably good agreement with the present experi-
mental measurements in the whole ejected electron energy range.
4.1.5 Total ionization cross sections

Figure 10.8 depicts an extensive comparison between the theoretical
CB1-CWB TCS (solid line) for protons, α-particles, and carbon ions and
the available experimental measurements. Additionally, we have reported
the semi-empirical Rudd’s and HKS predictions (dashed line and dotted
3
10
cm )
2
2
10
-16
Total ionization cross sections (10
1
10
6+
10
0 C
2+
-1
10 He
+
H
-2
10 2 3 4 5
10 10 10 10
Incident projectile energy (keV/u)
Figure 10.8 Total cross section as a function of the incident energy (keV/u). The theo-
retical CB1-CWB results are reported by a solid line whereas the semi-empirical HKS
and Rudd’s results are represented by a dotted and a dashed line, respectively. The
experimental measurements taken from various sources are represented by symbols
(for more details we refer the reader to Refs. 8, 9, 22).
line, respectively). We clearly observe a very good agreement between the

free-parameter calculations of Champion and co-workers8, 9, 22 and the
semi-empirical predictions of the Rudd’s model (initially developed for
protons) over the whole incident energy range investigated in the present
work (50 keV/u–100 MeV/u). On the contrary, the HKS model exhibits
large discrepancies with the theoretical results as well as the experimental
measurements.
4.1.6 From vapour to liquid water

Water ionization by charged particle impact (electrons as well as heavy
charged particles) has been a matter of active research since the 1970s par-
ticularly in the field of radiobiology for modelling the radio-induced dam-
ages, the biological matter being commonly simulated by water. Numerical
track-structure Monte Carlo simulations were then successfully developed
for describing the charged particle transport in biological medium and
then providing a detailed (macroscopic or microscopic) description of the
three-dimensional energetic deposit cartography. To that end, the model-
ling of the ion-induced ionizing processes in water and more particularly
of ionization is of prime importance. However, due to the scarcity of
experimental measurements in its liquid phase, water was essentially studied
in its vapour phase by assuming that describing the particle track-structure
in liquid matter could be done, in a first step, either by applying the well-
known “gas-phase approximation” i.e., via a simple linear extrapolation to
unit density environment of the liquid or by converting the highly excited
Rydberg states occurring in gaseous water into ionizations (see for example
Turner et al.69). Besides, note that some authors36 have also used the above-
cited Rudd’s model by adjusting the fitting parameters to the liquid case in
particular the binding energies whose values differ from gaseous to liquid
water by about 2–4 eV essentially for the three outermost subshells.36
Comparatively, the treatment of liquid water ionization by heavy
charged particles was less studied and presents still today a challenge for
radiobiologists who are interested by knowing the degree to which non-
scaling effects are important in simulating ion transport in biological mat-
ter. In this context, various semi-empirical models were developed since
the pioneer works of the Oak Ridge group.70 In this framework, DDCS
and SDCS were calculated within the 1st Born approximation by taking
advantage that in this framework the doubly differential ionization cross
−1
sections were related to the energy-loss function η(E, K ) = Im[ ε(E,K )]
where ε (E,K) denotes the dielectric-response function with E for the
energy transfer and K for the magnitude of the momentum transfer on a

single collision. Thus, similarly to their previous study on electron impact
ionization in liquid water, Dingfelder et al.36 proposed an extension of
their empirical model based on optical measurements, namely, reflectance
measurements for describing the liquid water ionization by fast protons
(Ei > 100 keV/u). In brief, the imaginary part of the dielectric-response
function is modelled in the optical limit, i.e., at momentum transfer K by
a superposition of Drude-like functions while the real part is analytically
calculated by using the Kramers–Kronig relation. For more details we
refer the reader to Dingfelder et al.36 Furthermore, the model uses exper-
imental data as well as theoretical constraints like sum rule and asymptotic
behaviour and the momentum-transfer dependence for the continuum
is introduced by an impulse approximation, in which a target electron
is seen as nearly free at the moment of energy and momentum transfer
whereas its binding through is taken equal to the instantaneous momen-
tum of a stuck electron. Total cross sections were then reported for proton
energies in the range from 500 keV to 10 MeV. Finally, to complete their
study and then to investigate the low-energy regime (Ei < 500 keV/u),
the authors proposed to use the Rudd’s model by adjusting the fitting
parameters in order to reproduce the recommended ICRU stopping
cross sections71 for liquid water, which is the only relevant experimental
information on the liquid phase. The authors also suggested taking into
account a realistic partitioning into the molecular sub-shells, by adjusting
in particular the sub-shell contributions of the dielectric model to those
provided by the Rudd’s model. The obtained values are reported below
(see Table 10.5).
They also obtained total ionization cross sections are reported in Figure 10.9.
It clearly appears that the vapour and liquid ionization cross sections differ
essentially in the region below 100–200 keV, namely, precisely where the
semi-empirical model of Rudd is used. On the contrary, when the dielectric-
model is used, namely, for Ei > 500 keV/u, only minor discrepancies are
observed.
Table 10.5 Parameters for the proton ionization cross sections in liquid water36
A1 B1 C1 D1 E1 A2 B2 C2 D2 α
1.02 82.0 0.45 −0.80 0.38 1.07 14.6 0.60 0.04 0.64
cm )
2
6
-16
Total ionization cross sections (10
5
0
-1 0 1 2 3 4
10 10 10 10 10 10
Figure 10.9 Total ionization cross sections as a function of the incident particle ener-
gies. Shown are the proton ionization cross sections for liquid and gaseous water (solid
and dashed line, respectively) taken from Ref. 36 whereas the experimental data are
similar to those reported above for total ionization cross sections of protons in water
vapour.
4.1.7 Influence of the target description on the ionization

cross sections
Describing the molecular ionization process at the multi-differential
level undoubtedly needs an accurate description of the target in terms of
electronic distribution. However, to overcome the difficult task of such
multi-centre problems, like here for studying the water molecule ionization,
we have seen that many authors used an atomic approach among which we
find the above-cited Bragg additivity rule—which consists in writing the
water cross sections as a mixture of H2 and O2 ones—and the complete
neglected differential overlap (CNDO) approach. In this context, Olivera
and co-workers41 have already pointed out huge differences in terms of
singly differential ionization cross sections for 70 keV protons in water vapour
by comparing these two descriptive approaches within the CDW-EIS frame-
work. A good agreement with experimental results taken from Bolorizadeh
and Rudd13 was then reported by the authors when the molecular description
was used—except at very low ejected electron energies—whereas the
Bragg’s additivity addition of atomic singly differential cross sections clearly
overestimated the experimental measurements at small electron energies.
Note that similar observations were also reported by Senger and co-work-
ers42 who have calculated—within the 1st Born approximation—angular
and energy distributions of δ-rays ejected from low-z molecular targets

(including water vapour) impacted by protons and α-particles.
Very recently, Champion et al.72 have also pointed out strong discrep-
ancies between a CNDO approach and a full-molecular (LCAO-SCF)
description. To do that, the authors have successively compared doubly
and singly differential ionization cross sections calculated by using both
the molecular wave function taken from Moccia45 and a CNDO approach
already used by many authors for describing the ionization process of
simple molecules. The selected description was that proposed by Senger
and co-workers42 in which the isolated water molecule was modelled by
means of oxygen and hydrogen orbitals, namely, O(1s), O(2s), O(2p), and
H(1s). Furthermore, it must be noted that to avoid any changes in the theo-
retical results due to the variations in the molecular orbital binding energies,
the authors have taken the same values for both calculations, namely, those
provided by Moccia.45
The CB1-CWB DDCS also calculated for 500 keV protons are
reported in Figure 10.10 (left panel). Evident discrepancies—within a fac-
tor 2—may be observed between the two water target descriptions, espe-
cially in the low ejection energy regime (Ee < 100 eV) whereas the cross
sections seem to be less sensitive to the molecular description for higher
ejected energies. Similar conclusions may be drawn when singly differential
cross sections are compared (see right panel of Figure 10.10). Surprisingly,
the CB1-CWB SDCS seem to be in better agreement with measurements
when the CNDO-initial state representation is used.
4.2 Theoretical description of the charge transfer process

within the 1st Born approximation
4.2.1 Basic formalisms
In this section, we investigate the electron capture process induced by
fast charged particles impinging on water target which still remains rarely
approached in the literature. In fact, except the work of Olivera et al.41 and
that of Fainstein et al.73 on the CDW-EIS method for single ionizing pro-
cesses (capture as well as ionization) induced by protons in gaseous water,
the quantum mechanical approaches are, to the best of our knowledge,
represented only by the recent 1st Born predictions given by Houamer
and co-workers74 for proton projectiles. The theoretical framework used
by the authors refers to the above-described CB1 model whose theoretical
background is briefly reported in the sequel.
-17
10
Singly differential cross sections (cm 2/eV)

0
10 Ee= 12 eV
E i = 500 keV
-1
10
E e=50 eV
-18
-2 10
10
Ee = 100 eV
-3
10
Ee =250 eV
-4 -19
10 10
-5 E e =750 eV
10 E i = 1500 keV
-6 -20
10 10
0 1 2
0 30 60 90 120 150 180 10 10 10 10
3
Ejection angle(deg) Ejected electron energy (eV)

Figure 10.10 Left panel: doubly differential cross sections for single ionization of water
vapour by 500 keV protons with different ejected electron energies as a function of the
ejection angle θe. The calculations were performed within the 1st Born approximation
(in the above-described CB1-CWB model) by including the Salin factor. The experi-
mental results are taken from Bolorizadeh and Rudd.13 Right panel: singly differential
cross sections for water ionization by protons of different incident energies (500 keV
and 1500 keV). The experimental results are taken from Toburen and Wilson.10 In both
cases, the theoretical results are obtained within the CB1-CWB model by using—for
describing the water target—the molecular wave function taken from Moccia45 and the
CNDO approach (solid and dashed line, respectively).
Thus, similarly to the previously referred CB1-CWB approach for

describing the ionization process, the multielectron collisional system can—for
the single electron capture process—also be reduced to a one-active electron
treatment. Indeed, if the long range part of the projectile-active electron inter-
action is preserved into the wave functions, the scattering matrix element may
be rewritten as

(1) Z P ZP +
∼ −
fi = χf − r + (34)
T χ ,
01 r0 i
with
χi+ ∼
= φ(k�i , �r0 )u(�r1 ) exp[−i(ZP /vi ) ln(vi r0 − ṽi . �r0 )], (35)
and
χf− = φ(k�s , �r0 )ξf (�r01 ) exp[−i(Zc /vi ) ln(vi r0 − v
� i . �r0 )], (36)
where ξf (r01 ) here represents a projectile bound state and Zc the charge of
the residual core. Thus, Eq. (34) can be written as

(1) = (bvi )−2iZc /vi � ∗ ZP ZP
Tfi d�r0 d�r1 [φ(ks , �r0 )ξf (�r01 )] − + φ(k�i , �r0 )u(�r1 )
r01 r0 (37)
× exp[i(Zc − ZP )/vi ] ln(vi r0 − v
� i . �r0 )].
Note that if Zc = ZP, Eq. (37) can be reduced to
(1) = φ(k�s , �r0 )ξf (�r01 ) − ZP + ZP φ(k�i , �r0 )u(�r1 ) ,

T fi (38)
r01 r0
where ξf ( r01 ) refers to an hydrogenic wave function, what coincides with
the expression provided by Jackson and Schiff 54 for the case of proton
beam impact on atomic hydrogen.
Furthermore, it is interesting to quote that the scattering matrix element
also obtained is in agreement with that reported for ionization in Eq. (30).
In order to calculate the singly differential cross sections as a function
of the projectile angular dispersion, we obviously need to take into account
all the particles of the collisional system and then consider the scattering
matrix element Tfi (K ). However, it is possible to show that this latter is
simply related to the matrix element T fi (K ) through the expression
∞
1 ′ � i
�) =
Tfi (K d �bd�
η η
exp[i(� − �
η ). b] exp[ dtVs (�r0 )]b−2iZc /vi
4π 2 vi −∞
N

× bc2iZP Zc /vi T � ),
fi (K
c=1 (39)
which is the Fourier transform of the Fourier antitransform of T ) (see

fi (K
Eqs. (27) and (37) and where η refers to the component of K perpendicular
to the impact velocity v i .
Then, similarly to the ionization case, the singly differential cross sec-
tions per molecular subshell dσ (where θs represents the scattering projectile
dθs
angle) are finally averaged over all the target orientations and then written as
MP2 1

dσ dσ
= (2π sin θs ) ≡ (2π sin θs ) 2 ( 2 ) |Tfi |2 d�Euler (40)
dθs d�s 4π 8π
where MP refers to the incident proton mass.
Thus, for low scattering angles (θs << 1), the SDCS take the simple form
MP2 . θs

dσ ∼ (41)
= (2π sin θs ) |Tfi |2 d�Euler .
dθs 16π 3
Finally, note that the scattering matrix element Tfi being dependent
on the impacted molecular subshell, the procedure here described is obvi-
ously performed for each of the N = 5 orbitals of the water molecule.
Thus, similarly to the ionization case, the “global” singly differential cross
sections are here also obtained by summing up all the j sub-shell contri-
butions [ ddθσ ]j and then weighted by the number Nelec (= 2) of electrons
s
per orbital. Thus, we write
N =5
dσ dσ
= Nelec . (42)
dθs dθs j
j=1
The total charge transfer cross section is finally obtained by writing that
2 2
1
d�b Afi (�b) ≡ d�b âfi (�b) ≡ � )2 , (43)

σ = η T
d� fi (K

(2πvi )2
where the Parseval identity has been employed.

In conclusion, it clearly appears from Eqs. (41) and (43) that when
cross sections differential in the projectile scattering angle are calculated,
we cannot ignore the interaction of the projectile with the target nuclei
and the passive electrons, whose influence may be not neglected. However,
when TCS are calculated, these latter play no rôle since the transition
is simply produced by the interaction of the projectile with the active
electron.
4.2.2 Total cross sections

We report in Figure 10.11 the total cross sections provided by the CB1
treatment developed by Houamer and co-workers.74 A good agreement
may be observed with the experimental available data in particular for
impact energies greater than 100 keV/u. In comparison, the predictions
given by the above-cited semi-empirical models are—for the major part
and in the overall energy range—in good agreement with the 1st Born
predictions, except those provided by the Rudd’s model which largely
overestimate both the experimental data and the other semi-empirical
predictions.
2
10
Total capture cross sections (10 cm )

2
1
10
-16
0
10
-1
10
-2
10
-3
10
-4
10
1 2 3 4
10 10 10 10
Figure 10.11 Total cross sections for charge transfer induced by protons in water.
Comparison between the CB1 predictions from Houamer et al.74 (solid line) and the
results provided by the semi-empirical model of Endo et al [39] (dashed line), Dingfelder
et al.36 (dotted line) and Rudd et al.12 (dash-and-dotted line). The experimental
measurements are taken from Dagnac et al.37 (solid down triangles), Gobet et al.16
(solid squares), Date et al.75 (solid up triangles), Toburen 76 (solid circles).
5. CONCLUSIONS
Single electron ionization and single electron capture of water mol-
ecules impacted by charged heavy particles—processes which are of interest
for medical applications—have been here theoretically studied within the
1st Born approximation framework.
From multiple differential to total cross sections, the present work
reports a detailed description of the ion-induced ionization process in
pointing out in particular the rôle played by the target representation in
the theoretical modeling. The cross sections also provided have been then
compared to a large set of experimental measurements as well as existing
semi-empirical predictions in order to clearly identify the pre-requested
kinematical conditions of the present perturbative treatment of water ion-
ization. Besides, the influence of the thermodynamical phase of water has
been also studied in order to assess its potential impact in numerical simula-
tions dealing with charged particle transport in biological matter.
Finally, proton-induced electron capture has been investigated and then
emphasized an overall good agreement in terms of total cross sections between
the present 1st Born predictions and the rare existing experimental data.
REFERENCES
1. Amaldi, U.; Kraft, G. Radiotherapy with beams of carbon ions. Rep. Prog. Phys. 2005,
68, 1861.
2. Wilson, R. R. Radiological use of fast protons. Radiology 1946, 47, 487.
3. Stone, R. S.; Lawrence, J. H.; Aebersold, P. C. A preliminary report on the use of fast
neutrons in the treatment of malignant disease. Radiology 1940, 35, 322.
4. Lawrence, J. H.; Tobias, C. A.; Linfoot, J. A., et al. Heavy-particle therapy in acromegaly
and Cushing disease. JAMA 1976, 235, 2307.
5. Berger, M. J. Penetration of proton beams through water I. Depth-dose distribu-
tion, spectra and LET distribution; National Institute of Standards and Technology:
Gaithersburg, MD, 1993; NISTIR Publication 5226.
6. Incerti, S.; Ivanchenko, A.; Karamitros, M., et al. Comparison of GEANT4 very low
energy cross section models with experimental data in water. Med. Phys. 2010, 37, 4692.
7. Belkić, Dž. Reviews of theories on ionization in fast ion-impact collisions with pros-
pects for applications to hadrontherapy. J. Math. Chem. 2010, 47, 1366.
8. Boudrioua, O.; Champion, C.; Dal Cappello, C., et al. Ab-initio calculation of differen-
tial and total cross sections for the ionization of water vapour by protons. Phys. Rev. A
2007, 75, 022720.
9. Champion, C.; Boudrioua, O.; Dal Cappello, C., et al. Theoretical and experimental
investigations of electron emission in He2++H20 collision. Phys. Rev. A 2007, 75, 032724.
10. Toburen, L. H.; Wilson, W. E. Energy and angular distributions of electrons ejected from
water vapour by 0.3–1.5 MeV protons. J. Chem. Phys. 1977, 66, 5202.
11. Toburen, L. H.; Wilson, W. E.; Popowich, R. J. Secondary electron emission from ioniza-
tion of water vapour by 0.3- to 2.0-MeV He+ and He2+ Ions. Radiat. Res. 1980, 82, 27.
12. Rudd, M. E.; DuBois, R. D.; Toburen, L. H., et al. Cross sections for ionization of gases
by 5–4000-keV protons and for electron capture by 5–150-keV protons. Phys. Rev. A
1983, 28, 3244.Rudd, M. E.; Goffe, T.V.; DuBois, R. D., et al. Cross sections for ioniza-
tion of water vapour by 7–4000-keV protons. Phys. Rev. A 1985, 31, 492.Rudd, M. E.;
Goffe, T. V.; Itoh, A. Ionization cross sections for 10–300-keV/u and electron-capture
cross sections for 5–150-keV/u He2+ ions in gases. Phys Rev A 1985, 32, 2128.Rudd,
M. E.; Itoh, A.; Goffe, T.V. Cross sections for ionization, capture, and loss for 5–450-keV
He+ on water vapour. Phys. Rev. A 1985, 32, 2499.
ejection of electrons from water vapour. Phys. Rev. A 1986, 33, 888.
14. Gibson, D. K.; Reid, I. D. Energy and angular distributions of electrons ejected from
various gases by 50 keV protons. Radiat. Res. 1987, 112, 418.
15. Werner, U.; Beckord, K.; Becker, J., et al. 3D imaging of the collision-induced coulomb
fragmentation of water molecules. Phys. Rev. Lett. 1995, 74, 1962.
16. Gobet, F.; Farizon, B.; Farizon, M., et al.Total, partial, and electron-capture cross sections
for ionization of water vapour by 20–150 keV protons. Phys. Rev. Lett. 2001, 86, 3751.
17. Straub, H. C.; Lindsay, B. G.; Smith, K. A., et al. Absolute partial cross sections for
electron-impact ionization of H2O and D2O from threshold to 100 eV. J. Chem. Phys.
1998, 108, 109.
18. Rudolph, P. S.; Melton, C. E. Experimental and theoretical considerations for ionization
of simple Hydrocarbons, H2, D2, O2, H2O, and the rare gases by 2.2-MeV a particles
and by electrons in a mass spectrometer. J. Chem. Phys. 1966, 45, 2227.
19. Ohsawa, D.; Sato, Y.; Okada, Y., et al. 6.0–10.0-MeV/u He2+-ion-induced electron
emission from water vapour. Phys. Rev. A 2005, 72, 062710.
20. Rudd, M. E. User-friendly model for the energy distribution of electrons from proton
or electron collisions. Nucl. Tracks Radiat. Meas. 1989, 16, 213.
21. Montenegro, E. C.; Shah, M. B.; Luna, H., et al. Water fragmentation and energy loss by
carbon ions at the distal region of the Bragg peak. Phys. Rev. Lett. 2007, 99, 213201.
22. Dal Cappello, C.; Champion, C.; Boudrioua, O., et al. Theoretical and experimental
investigations of electron emission in C6++H2O collisions. Nucl. Instrum. Methods Phys.
Res., Sect. B 2009, 267, 781.
23. Lindsay, B. G.; Sieglaff, D. R.; Smith, K. A., et al. Charge transfer of 0.5-, 1.5-, and 5-keV
protons with H2O: absolute differential and integral cross sections. Phys. Rev. A 1997,
55, 3945.
24. Luna, H.; de Barros, A. L. F.; Wyer, J. A., et al. Water-molecule dissociation by proton and
hydrogen impact. Phys. Rev. A 2007, 75, 042711.
25. Abu-Haija, O.; Kamber, E.Y.; Ferguson, S. M. Single-electron capture by He2+ ions from
triatomic molecules. Nucl. Instrum. Methods Phys. Res., Sect. B 2003, 205, 634.
26. Greenwood, B.; Mawhorter, R. J.; Čadež, I., et al. The contribution of charge exchange
to extreme ultra-violet and X-ray astronomy. Phys. Scr. 2004, T110, 358.
27. Seredyuk, B.; McCullough, R. W.; Tawara, H., et al. Charge exchange and dissociative
processes in collisions of slow He2+ ions with H2O. Phys. Rev. A 2005, 71, 022705.
28. Mawhorter, J.; Chutjian, A.; Cravens, T. E., et al. Absolute single and multiple charge
exchange cross sections for highly charged C, O, and Ne ions on H2O, CO, and CO2.
Phys. Rev. A 2007, 75, 032704.
29. Bodewits, D.; Hoekstra, R. Electron capture in collisions between O6+ ions and H2O
molecules. Phys. Rev. A 2007, 76, 032703.
30. Uehara, S.; Nikjoo, H. Monte Carlo track structure code for low-energy alpha-particles
in water. J. Phys. Chem. 2002, 106, 11051.
31. Hansen, J. P.; Kocbach, L. Ejection angles of fast delta electrons from K-shell ionization
induced by energetic ions. J. Phys. B 1989, 22, L71.
32. Stolterfoht, N.; DuBois, R.; Rivarola, R. D. In electron emission in heavy ion-atom collisions;
Springer: Berlin, 1997.
33. Rudd, M. E.; Kim, Y.-K.; Märk, T. D. et al, Technical Report 55, International

Commissions on Radiation Units and Measurements, Bethesda, MD, 1996.
34. Bernal, M. A.; Liendo, J. A. The HKS model for electron production in liquid water by
light ions. Nucl. Instrum. Methods Phys. Res., Sect. B 2006, 251, 171 Bernal, M.A.; Liendo,
J.A. Inelastic-collision cross sections for the interactions of totally stripped H, He and C
ions with liquid water, Nucl. Instrum. Methods Phys. Res., B,2007, 262, 1.
Rev. 1968, 171, 114.
36. Dingfelder, M.; Inokuti, M.; Paretzke, H. G. Inelastic-collision cross sections of liquid
37. Dagnac, R.; Blanc, D.; Molina, D. A study on the collision of hydrogen ions H1+, H2+
and H3+ with a water-vapour target. J. Phys. B 1970, 3, 1239.
38. Green, A. E. S.; McNeal, R. J. Analytic cross sections for inelastic collisions of protons and
hydrogen atoms with atomic and molecular gases. J. Geophys. Res. 1971, 76, 133.Miller, J.
H.; Green, A. E. S. Proton energy degradation in water vapour. Radiat. Res. 1973, 54, 343.
39. Endo, S.; Yoshida, E.; Nikjoo, H., et al. A Monte Carlo track structure code for low
energy protons. Nucl. Instrum. Methods Phys. Res., Sect. B 2002, 194, 123.
40. Sataka, M.; Yagishita, A.; Nakai, Y. Measurement of charge-changing cross sections in
collisions of He and He+ with H2, O2, CH4, CO and CO2. J. Phys. B 1990, 23, 1225.
41. Olivera, G. H.; Martinez, A. E.; Rivarola, R. D., et al. Theoretical calculation of elec-
tronic stopping power of water vapour by proton impact. Radiat. Res. 1995, 144,
241.Olivera, G. H.; Fainstein, P. D.; Rivarola, R. D. Contribution from the inner shell
of water vapour to dose profiles under proton and alpha particle irradiation. Phys. Med.
Biol. 1996, 41, 1633.
42. Senger, B.; Rechenmann, R.V. Angular and energy distributions of D-rays ejected from
low-z molecular targets by incident protons and a particles. Nucl. Instrum. Methods Phys.
Res., Sect. B 1984, 2, 204.Senger, B. Calculated molecular double-differential cross-
sections for ionization under proton impact. Z. Phys. D 1988, 9, 79.
43. Kim, Y.-K.; Rudd, M. E. Binary-encounter-dipole model for electron-impact ioniza-
tion. Phys. Rev. A 1994, 50, 3954.
44. Champion, C.; Hanssen, J.; Hervieux, P.-A. Differential and total (e,2e) cross sections of
simple polyatomic molecules. J. Chem. Phys. 2004, 121, 9423.
45. Moccia, R. One-center basis set SCF MO’s. III. H2O, H2S, and HCl. J. Chem. Phys.
1964, 40, 2186.
46. Rudd, M. E.; Kim,Y.-K.; Madison, D. H., et al. Electron production in proton collisions
with atoms and molecules: energy distributions. Rev. Mod. Phys. 1992, 64, 441.
48. Salin, A. Ionization of helium by proton impact. J. Phys. B 1972, 5, 979.
49. Madison, D. H. Angular distributions of electrons ejected from helium by proton impact.
Phys. Rev. A 1973, 8, 2449.
50. Stolterfoht, N.; Schneider, D.; Tanis, J., et al. Evidence for two-centre effects in the elec-
tron emission from 25 MeV/u Mo40++He collisions: theory and experiment. Eur. Phys.
Lett. 1987, 4, 899.
51. Rudd, M. E.; Sautter, C. A.; Bailey, C. L. Energy and angular distributions of electrons
ejected from hydrogen and helium by 100- to 300-keV protons. Phys. Rev. 1966,
151, 20.
fast protons in atomic hydrogen. Proc. Phy. Soc. 1964, 84, 89.
53. Oppenheimer, J. R. On the quantum theory of the capture of electrons. Phys. Rev. 1928,
31, 349.Brinkman, H. C.; Kramers, H. A. Proc. Acad. Sci. Amsterdam 1930, 33, 973.
54. Jackson, J. D.; Schiff, H. Electron capture by protons passing through hydrogen. Phys.
Rev. 1953, 89, 359.
55. Omidvar, K.; Golden, J. E.; McGuire, J. H. Single-electron Born approximations for
charge transfer from multielectron atoms to protons. Phys. Rev. A 1976, 13, 500.
56. Dollard, M. D. Ph.D. Thesis, University of Michigan, 1963.Dollard, J. D. Asymptotic
convergence and the coulomb interaction. J. Math. Phys. 1964, 5, 729.
57. Mapleton, R. A. Asymptotic form of the electron capture cross section in the second
Born approximation. Proc. Phys. Soc. 1967, 91, 868.
58. Carpenter, C. P.; Tuan, T. F. On the application of Faddeev equations and the Coulomb
t matrix to asymptotic electron capture in hydrogen. Phys. Rev. A 1970, 2, 1811.
59. Dewangan, D. P.; Eichler, J. Boundary conditions and the strong potential Born approxi-
mation for electron capture. J. Phys. B 1985, 18, L65.
60. Pradhan, T. Electron capture by protons passing through hydrogen. Phys. Rev. 1957, 105,
1250.
61. Macek, J.; Shakeshaft, R. Second Born approximation with the Coulomb Green’s func-
tion: electron capture from a hydrogenlike ion by a bare ion. Phys. Rev. A 1980, 22,
1441.Macek, J.; Taulbjerg, K. Correction to ZP/ZT expansions for electron capture.
Phys. Rev. Lett. 1981, 46, 170.Taulbjerg, K.; Barrachina, R.; Macek, J. H. Perturbation
theory for strongly interacting atomic systems. Phys. Rev. A 1990, 41, 207.
62. Rivarola, R. D.; Maidagan, J. M.; Hanssen, J. Coulombic potentials in atomic collisions.
63. Belkić, Dž; Gayet, R.; Salin, A. Electron capture in high-energy ion-atom collisions.
Phys. Rep. A 1979, 56, 279.Belkić, Dž; Taylor, H. S. First-order theory for charge
exchange with correct boundary conditions: General results for hydrogenlike and mul-
tielectron target atoms. Phys. Rev. A 1987, 35, 1991.Belkić, Dž. Reformulated impulse
approximation (RIA) for charge exchange in fast ion-atom collisions. Phys. Scripta 1996,
53, 414.
64. Thomas, L. H. On the capture of electrons by swiftly moving electrified particles. Proc.
R. Soc. Lond. A 1927, 114, 561.
65. Belkić, Dž. Electron capture by fast protons from helium, nitrogen, and oxygen: The
corrected first Born approximation. Phys. Rev. A 1988, 37, 55.
66. Corchs, S. E.; Rivarola, R. D.; McGuire, J. H. Impact-parameter formulation for electron
capture from molecular targets. Phys. Rev. A 1993, 47, 3937.
67. Fainstein, P. D.; Ponce, V. H.; Rivarola, R. D. A theoretical model for ionization in
ion-atom collisions. Application for the impact of multicharged projectiles on helium.
J. Phys. B 1988, 21, 287.
68. Salin, A. Ionization of atomic hydrogen by proton impact. J. Phys. B 1969, 2, 631.
69. Turner, J. E.; Paretzke, H. G.; Hamm, R. N., et al. Comparative study of electron energy
deposition and yields in water in the liquid and vapour phases. Radiat. Res. 1982, 92, 47.
70. Ritchie, R. H.; Hamm, R. N.; Turner, J. E., et al. In Sixth Symposium on Microdosimetry,
Brussels, Belgium; Booz, J.; Ebert, H. G., Eds.; London: Harwood Academic Publishers,
1978; p 345.
71. International Commission on Radiation Units and Measurements, Bethesda, Maryland,
1993.
72. Champion, C.; Galassi, M. E.; Weck, P. F. et al. Quantum-mechanical contributions to
numerical simulations of charged particle transport at the DNA scale. In: Radiation
Damage in Biomolecular Systems, Biological and Medical Physics, Biomedical Engineering;
Gómez-Tejedor, G. G.; Fuss, M. C. Eds.; Springer Science, 2012; p 263.
73. Fainstein, P. D.; Gulyás, L.; Martin, F., et al. Angular asymmetry of low-energy electron
emission in ion-atom collisions. Phys. Rev. A 1996, 53, 3243.
74. Houamer, S.; Popov, Yu. V.; Dal Cappello, C., et al. Charge transfer in collisions of
protons with water molecule at high projectile energies and small scattering angles.
Nucl. Instrum. Methods Phys. Res., Sect. B 2009, 267, 802.Houamer, S.; Popov, Yu. V.; Dal
Cappello, C. Total cross sections for charge transfer reactions of protons with atomic
and molecular targets at high projectile energies: the role of inner orbitals. Phys. Lett. A
2009, 373, 4447.
75. Date, H.; Sutherland, K. L.; Hayashi, T., et al. Inelastic collision processes of low-energy
protons in liquid water. Radiat. Phys. Chem. 2006, 75, 179.
76. Toburen, L. H. Ionization and charge-transfer: basic data for track structure calculations.
Radiat. Env. Biophys. 1998, 37, 221.
CHAPTER ELEVEN
Ion Collisions with Water

Molecules: A Time-Dependent
Density Functional Theory
Approach
Tom Kirchnera, Mitsuko Murakamia, Marko Horbatscha and
Hans Jürgen Lüddeb
aDepartment of Physics and Astronomy, York University, Toronto, Ontario, Canada M3J 1P3
bInstitut für Theoretische Physik, Goethe-Universität, D-60438 Frankfurt, Germany
Contents
1. Introduction 316
2. Theory 317
2.1 The density-functional theory approach to heavy-particle collisions 318
2.2 The basis generator method adapted for ion-molecule collisions 320
2.3 Extraction of measurable cross sections 323
2.4 Computational aspects 324
3. Results 325
3.1 Net cross section results 328
3.2 Fragmentation cross sections 332
4. Summary and Outlook 334
Acknowledgments 335
References 335
Abstract
Collisions of simple ions from water molecules in the energy range of 10–5000 keV/
amu are considered within an independent electron model. The basis generator
method applied in the past successfully to ion–atom collisions is adapted to deal with
molecular targets. Cross sections for single- and multiple-electron processes (capture
and transfer to the continuum) are obtained directly from solving time-dependent
Kohn–Sham-type orbital equations and using a Slater determinant based analysis.
Fragmentation yields are predicted on the basis of a semi-phenomenological model
which uses the calculated cross sections as input. Comparison with experiment is
made for proton and He+ impact collisions.

ISSN 0065-3276, http://dx.doi.org/10.1016/B978-0-12-396455-7.00011-X
316 Tom Kirchner et al.
1. INTRODUCTION
Electron removal from water molecules is one of the most frequent
processes taking place when biological tissue is irradiated by protons or
heavy ions. In hadron therapy,1 it contributes to the desired destruction of
a tumor and to the undesired damage of the surrounding tissue in at least
two ways: (i) the produced free electrons attack DNA molecules in ionizing
collisions or via formation of transient molecular resonance states;2 (ii) the
H2 Oq+ ions left behind are often instable and thus prone to fragmentation,
especially if q 1, i.e., if multiple-electron removal occured. In a liquid
environment, fragment ions may recombine to highly reactive oxygen spe-
cies such as HO2, which can damage the DNA in chemical reactions (see,
e.g., Ref. 3 and references therein).
It is thus important to understand ion-impact induced ionization
and the subsequent fragmentation dynamics of water molecules in detail.
Early experimental studies determined (total) cross sections for ioniza-
tion and electron transfer processes,4, 5 while more recent works were
concerned with measuring the yields of charged fragments, sometimes in
coincidence with an ionized electron or the projectile charge state.6–10
Theoretical efforts have focused on calculating cross sections for electron
removal processes at impact energies above 20 keV, for which relatively
simple self-consistent field descriptions of the water molecule are deemed
suitable. They include a molecular orbital based coupled-channel cal-
culation,11 Born- and continuum distorted-wave-type models12–14 and
classical trajectory Monte Carlo (CTMC) calculations based on one-
center15, 16 and three-center17 model potentials. Fragmentation cross
sections have been deduced from some of the results using (semi-)empiri-
cal models to relate electron removal to the production of the different
fragment ions.14, 15, 17
Our own work in this area is built on an extension of a quantum-
mechanical approach developed and tested for ion–atom collisions (see,
e.g., Refs. 18–20 and the review article: Ref. 21). The approach is based
on density functional theory (DFT) and uses the basis generator method
(BGM) to propagate effective single-particle (Kohn–Sham-type) orbitals.
The extension to molecular targets was accomplished by a spectral rep-
resentation of the target Hamiltonian and a single-center expansion of its
eigenstates.22 This enables to separate the ion-molecule multi-center prob-
lem into a set of two-center problems and to apply the BGM with only
modest modifications.
Ion Collisions with Water Molecules: A Time-Dependent Density Functional Theory Approach 317
The approach has been rather successful in describing experimental cross

sections for various electron transfer and ionization processes in proton– and
He+–H2 O collisions over wide ranges of impact energies.22, 23 Recently, we
have coupled it with a semi-phenomenological model for water fragmentation
after single and multiple electron removal, and have obtained results which
agree favorably with experimental cross sections for the production of charged
fragments.24, 25
In this chapter, we provide a somewhat condensed description of this
ion-molecule approach and the results obtained from it. We start with
a brief outline of the general framework of the theory and the DFT-
perspective on atomic collisions in Section 2.1. This is followed by sum-
maries of the ion-molecule BGM in Section 2.2 and the method used to
extract measurable information from the BGM solutions in Section 2.3.
Some computational aspects are discussed in Section 2.4, but for a more
detailed description the reader is referred to Ref. 24. We present results for
electron removal in proton– and He+–H2 O collisions in Section 3.1 and
for fragmentation of H2 O after proton impact in Section 3.2. The chapter
ends with a summary and an outline of directions of future work in Section
4. Atomic units ( = me = e = 4π ǫ0 = 1) are used throughout, unless
specified otherwise.
2. THEORY
Ion-molecule collisions are complex many-body problems. At pres-
ent, a full theoretical description of their dynamics in terms of a consistent
quantum theory for all degrees of freedom is not feasible. Fortunately, it
is not necessary either—at least not in the range of impact energies from
about 10 keV/amu up to a few MeV/amu in which we are interested. In
this range, the different time scales of the electron and nuclear dynamics
allow for a separation of the many-body state and a simplified treatment
of the different processes at play. The projectile ion is fast enough to
assume that it travels on a classical straight-line trajectory. Its interaction
time with the target molecule is short, so that the latter neither rotates
nor vibrates appreciably while electrons are being transferred or ejected
to the continuum. Thus, the Franck–Condon approximation, which
assumes that the molecule is frozen to its equilibrium geometry when the
electronic transitions occur, can be used. Such a calculation can make
(possibly state-specific) predictions for the formation of H2 Oq+ ions,
which may break up subsequently. Fragmentation cross sections can be
derived semi-phenomenologically by associating single-and multiple-

electron removal processes with probabilities for specific break-up
channels using experimental branching ratios as input.
This is the framework we have adopted in our studies of ion-H2 O col-
lisions.22-25 The main theoretical task is the solution of a time-dependent
Schrödinger equation (TDSE) for the electrons, which experience time-
dependent and stationary Coulomb forces due to the moving projectile
and the non-moving target nuclei, respectively, in addition to their mutual
repulsion. In the next subsection, we outline how this complicated many-
electron problem can be formulated within time-dependent density func-
tional theory (TDDFT). This is followed by a discussion of our specific
approach to solving the ensuing set of single-particle Schrödinger-type
equations and a brief description of the methods used to extract measurable
information from the solutions.
2.1 The density-functional theory approach to heavy-particle

collisions
The problem at hand can be stated as follows: the electronic state vector
fulfills the TDSE
(t)|�(t)�
i∂t |�(t)� = H (1)
for the initial condition
|�(t0 )� = |�0 � (2)
and the Hamiltonian
H +W
(t) = T int + V
ext (t), (3)
which consists of the kinetic energy
N

= 1 2
T − ∇j , (4)
2
j=1
the internal electron–electron interaction

N
1
int =
W (5)
i<j
|ri − rj |
with the electron position vectors ri being measured from a laboratory

(i.e., a non-moving) reference frame, and the external Coulomb interaction
with the heavy particles. For a target molecule described within the Born-
Oppenheimer and Franck–Condon approximations we have
N
M
Zα ZP
ext (t) = −
V + , (6)
rαj |rj − R(t)|
j=1 α=1
where M denotes the number of target nuclei with charges Zα (M = 3 and

Z1 = Z2 = 1, Z3 = 8 for H2 O), and rαj is the distance of the αth nucleus to
the jth electron. ZP is the charge of the (bare) projectile ion, which moves
along the straight-line trajectory R(t) = (b, 0, v0 t) characterized by the
impact parameter b and the constant projectile speed v0.
The backbone of TDDFT is the Runge–Gross theorem.26 It states that
the electronic state vector |�(t)� that solves the TDSE (1) is (up to a global
time-dependent phase) a unique functional of the time-dependent one-
particle density n, if the initial state |0 � is the (nondegenerate) ground state
of the system. The density at time t can be calculated without knowledge
of |�(t)� from a set of single-particle Schrödinger-type, so-called time-
dependent Kohn–Sham (TDKS), equations

1 2
i∂t ψi (r, t) = − ∇ + vTDKS [n](r, t) ψi (r, t), i = 1, . . . , N (7)
2
according to
N

n(r, t) = |ψi (r, t)|2 . (8)
i=1
The potential vTDKS in (7) is also a unique functional of n. This is another

consequence of the Runge–Gross theorem and suggests the following
route to solving the time-dependent many-electron problem: (i) determine
vTDKS [n](r, t); (ii) propagate the TDKS orbitals from initial conditions, that
combine to the ground-state density according to (8), up to a final time tf ;
(iii) extract all observables of interest from the asymptotic solutions using
the fact that |�(tf )� is uniquely determined by n.
In practice, these steps can only be taken after approximations have been
introduced, since the exact functional dependences of vTDKS and |� on n
are not known. Nevertheless, the idea to circumvent the actual solution of
the many-electron TDSE (1) is appealing. It provides a firm basis of the
independent particle model (IPM) and some clues of how to interpret and
improve it.
2.2 The basis generator method adapted for ion-molecule

collisions
In our previous works on proton–H2 O collisions22, 24, 25 we have used a
simplified TDKS Hamiltonian, in which the effective electron–electron
interaction potential is frozen to its (approximate) form in the initial state:
T
vTDKS [n](r, t) ≈ Vαβγ (r) + V P (r, t), (9)
ZP
V P (r, t) = − .
|r − R(t)| (10)
T
Vαβγ is a molecular ground-state potential on the Hartree–Fock (HF) level
and α, β, γ are the standard Euler angles, which we use to specifiy the
orientation of the water molecule with respect to the ion beam axis. The
occupied eigenstates of
T 1
αβγ
H = − ∇ 2 + Vαβγ
T
(11)
2
form the ground-state density and are the initial conditions for the TDKS-
type equations

Ŵ T
αβγ Ŵ
i∂t |ψαβγ (t)� = H + V P (t) |ψαβγ (t)�, (12)
Ŵ
|ψαβγ (ti )� = |Ŵαβγ �.
(13)
The capital Greek letters label the occupied MOs whose time evolu-
tion we study, i.e., Ŵαβγ ∈ {1b1 , 3a1 , 1b2 , 2a1 }αβγ . The innermost orbital is
excluded, since the 1a1 electrons are too tightly bound to undergo appre-
ciable transitions in the studied systems.
The no-response approximation (9) can be expected to work well if
the projectile is fast (and of low charge), whereas dynamical response may
play a role at lower impact energies, at which the electrons have more
time to react to changes in the effective potential. This tendency has been
confirmed in numerous studies of ion–atom collision systems by compar-
ing no-response and response cross sections,18, 20 and we also see some
evidence for it in the results presented further below.
The main difficulty with solving (12) is the structure of Vαβγ T . If one
Ŵ
expands the orbitals |ψαβγ (t)� in a basis, one encounters multi-center inte-
grals, which have plagued molecular structure calculations for many years.27
Our approach circumvents the explicit calculation of multi-center integrals

by going into an energy representation of the target Hamiltonian

H T
αβγ = ǫ� |�αβγ ��αβγ |, (14)
�
in which ǫ� denotes the energy eigenvalue of the MO labeled by quantum
number(s) . We use two different basis sets to proceed: an orthonormal
single-center basis {|s�} for representing the MOs

�
|�αβγ � = ds,αβγ |s� (15)
s
J
and a nonorthogonal, dynamical two-center basis {|χj (t)�} for propagating
(12):
J
Ŵ
|ψαβγ (t)� = ajJŴ,αβγ (t)|χj (t)�. (16)
jJ
Both basis sets will be specified further below. The benefit of using these
expansions is that the multi-center molecular matrix elements break up
into combinations of energy eigenvalues, (real) expansion coefficients, and
simpler overlap matrix elements:
KJ J
Mkj (α, β, γ , t) = �χkK (t)|H T
αβγ |χj (t)�
J (17)
� �
= ǫ� �χkK (t)|s�ds,αβγ du,αβγ �u|χj (t)�.
�
su
Instead of propagating the initial-state MOs we exploit the linearity of the

single-particle equations (12) and use the single-center basis states |s� as
initial conditions to obtain
J
s
|ψαβγ (t)� = ajJs ,αβγ (t)|χj (t)�
(18)
jJ
from solving a set of coupled-channel equations which are very similar to

those of a two-center ion–atom collision problem. The molecular solutions
(16) can be reconstructed
J
Ŵ Ŵ s Ŵ
|ψαβγ (t)� = ds,αβγ |ψαβγ (t)� = ds,αβγ ajJs ,αβγ (t)|χj (t)�
s
(19)
sjJ
to calculate the density n(r, t) of the system. In principle, this is the starting
point for the calculation of observable quantities in the TDDFT framework.
In practice, however, the density is of limited use due to the unknown depen-
dence of the many-electron state on it. A more fruitful, though approximate
basis for the final-state analysis is the one-particle density matrix constructed
from single-determinantal wave functions.
Before we discuss this method, we have to comment on how we deal
with collisions of water molecules with He+ ions, which involve an addi-
tional electron on the projectile. If one were to solve the many-electron
TDSE (1), this electron would have to be treated on the same footing as the
initial target electrons; in particular, the state vector |�(t)� would depend
on the coordinates of all electrons and be antisymmetric under interchange
of any pair of them. A pragmatic way to account for the projectile electron
in an effective single-particle framework is to add an appropriate screening
potential to the Hamiltonian and solve the TDKS-type Eqs. (12) for all
initially populated orbitals—the target MOs and a (moving) atomic orbital
(AO) on the projectile center. The pros and cons of this approach as well as
more sophisticated alternatives were discussed in Refs. 23, 28. Briefly, the
main shortcoming of using a common Hamiltonian with effective target
T
and projectile potentials Vαβγ and V P is that either the initial target elec-
trons or the initial projectile electron(s) experience an incorrect asymptotic
potential.
T
Our choice for He+−water-molecule collisions consists in leaving Vαβγ
in Eq. (9) unchanged, but replacing the bare projectile potential (10) by
He (r ′ )|2
2 |ϕ1s (20)
P
V (rP ) = − + d3 r ′ ,
rP |r′ − rP |
where ϕ1s He is the (normalized) ground-state HF AO of (neutral) helium and
rP = r − R(t). The potential (20) decays like −1/rP asymptotically, which

is correct from the viewpoint of the initial target electrons, but incorrect
for the projectile electron, which in reality experiences an unscreened
Coulomb potential for the charge number ZP = 2.
T + VP
One can argue that this choice of a common potential Vαβγ for
all electrons is reasonable for situations in which electron transfer from the
target to the continuum and to the projectile dominate, while projectile
electron loss is relatively unimportant. If one were to choose a different
Hamiltonian without the screening term in (20), one would improve the
description of the electron loss process, but at the expense of the capture
and target ionization channels.
One may be inclined to circumvent these problems by using different

potentials in the single-particle equations for the initial target and projectile
electrons. This is certainly a possibility, but it comes with a price, too: the
propagated orbitals lose their initial orthogonality during the time propaga-
tion, which renders the final-state analysis cumbersome.28, 29 In fact, our
main motivation for choosing a simple, common potential for all active
electrons was to preserve orthogonality at all times. This makes it straight-
forward to construct the one-particle density matrix of the system at the
final time and to extract observable information from it. Not only is this
elegant, it is also crucial for the correct description of channels, for which
Pauli blocking effects are important. This will be further discussed in the
next subsection and along with the results in Section 3.
2.3 Extraction of measurable cross sections

The first step in the density-matrix analysis of the solutions is the calcula-
tion of single-particle state-to-state transition amplitudes
Aif (tf ) = �f |ψi (tf )� (21)
at the final time tf . The propagated states |ψi (tf )� in (21) are those of Eq.
(19) plus the propagated projectile-electron orbital in the case of He+
impact. As final states |f � we consider bound projectile and bound target
states. All amplitudes are fed into the one-particle density matrix
N

′
1
�f |γ̂ (tf )|f � = Aif (tf )A∗if ′ (tf ), (22)
i=1
whose nondiagonal structure reflects the antisymmetry of the N-electron

state, i.e., the Pauli principle. So-called inclusive probabilities are obtained
as determinants of submatrices of (22),30 and all many-electron transition
probabilities of interest can be written and calculated as appropriate com-
binations of those inclusive probabilities.
In this work, we are concerned with cross sections which correspond
to probabilities Pk,l for finding k electrons at the projectile in coincidence
with l electrons in the continuum and with sums thereof. Details about
how these probabilities are calculated from the inclusive probabilities for
occupying bound target and bound projectile states have been provided in
Refs. 24, 29 and will not be repeated here.
Let us close this section with a few conceptual comments.The inclusive-

probability formalism is derived from the assumption that both the propa-
gated many-electron state and the final states of interest are single Slater
determinants.30 This implies that the TDKS-type orbitals, which combine
to the density according to Eq. (8), are assumed to carry the physical infor-
mation about the active electrons as in HF theory. In other words, such an
analysis corresponds to the IPM level and is a (further) departure from a per-
fect TDDFT, in which only the density is associated with a physical inter-
pretation, while the TDKS orbitals are viewed as auxiliary functions without
direct physical significance. The limitations of the IPM are well known, and
some of them will be discussed further below in Section 3. Nevertheless, the
IPM has been popular, since it can be interpreted relatively easily and, more
importantly, since there is no first-principles alternative available that goes
beyond it. This is, in fact, one of the main shortcomings of current TDDFT
applications to ion-impact or laser-field induced many-electron dynamics.
2.4 Computational aspects

It was mentioned in Section 2.2 that two basis expansions are involved in
our method: a single-center basis is used to represent the ground-state MOs
[cf. Eq. (15)] and a two-center basis to propagate the TDKS-type orbitals
[cf. Eq. (16)]. The latter is obtained from the two-center (TC) extension
of the BGM method, which was developed as a general means to describe
time-dependent quantum systems in a finite basis.31 The main idea of the
BGM is to construct a subspace of Hilbert space H that is tailored to rep-
resenting only the propagated states rather than representing H completely,
which is illusory in any case. It was shown in Ref. 31 how this can be
accomplished in terms of a hierarchy of dynamical subspaces of H starting
from one that includes the states which form the initial conditions of the
time-dependent problem.
In practice, a TC-BGM basis consists of sets of AOs placed on the pro-
jectile center and the target center, which for water molecules we choose
to coincide with the position of the oxygen nucleus. Both sets are endowed
with electron translation factors to ensure Galilean invariance. In addition,
time-dependent pseudo states constructed by the repeated application of a
regularized projectile potential on the set of target AOs
J
J 1
|χj (t)� = 1 − e−rP |ϕjT � (23)
rP
are included to represent the continuum and quasimolecular effects at low

collision energy.
The same set of target AOs is also used to represent the ground-state
MOs of the water molecule—taken from the minimal-basis-set HF cal-
T
culation of Ref. 32—according to Eq. (15), i.e., {|ϕj �} ≡ {|s�}. It is then
consistent to limit the spectral representation of the Hamiltonian (14) to
these initially occupied MOs as given in the single-center expansion (15).
This might sound drastic, but test calculations that go beyond this model
by using a closure approximation have lent credence to its reliability for
medium to high impact-energy collisions. Details about the basis sets used
are given in Ref. 24.
The measurements we wish to compare with did not record the orien-
tation of the water molecule with respect to the ion beam axis. Hence, we
have to perform calculations for a set of orientations characterized by Euler
angles α, β, γ and average them in an appropriate way. One may expect
a considerable orientation dependence of the electronic transitions at low
impact energies, and, in fact, there is ample evidence for that.11 However,
we found that considering just two orientations† is sufficient for the pur-
pose of obtaining net and charge-state correlated cross sections for electron
removal in the impact energy range from about 20 to 5000 keV/amu. All
results discussed in the next section have been obtained in this way.
3. RESULTS
The present model has been used to calculate data for comparison
with experiments at various levels. For bare-ion impact one first investi-
gates net processes, such as total electron yield in the continuum σ−, total
positive charged ion yield σ+, and total net charge transfer σcap for which
extensive experimental data were provided by Rudd and coworkers.4 For
the case of He+ projectiles σ− includes contributions from the projectile
electron, and one can look at the total electron loss from the projectile, as
well. For neutral projectiles these electron loss processes become important,
particularly if one is concerned with the problem of energy deposition.
In principle, the net (or gross) cross sections should be the most reliable
quantities provided in a density-functional theory framework, as in essence
they depend on how the electron density splits into three contributions:
† Namely those shown in Figure 1 of Ref. 24.

parts associated with the target after the collision, parts associated with the
projectile, and the remainder representing continuum electrons. In practice,
however, one should worry about the independent-electron approximation,
since it may enhance multiple-electron transitions. Even in the case of bare
projectiles one deals with the added difficulty that an N-electron density
distribution that originated at the target is spread over the three regions,
and that in low-energy collisions it is possible to transfer more electrons to
the projectile than is physically allowed. In a perfect TDDFT this should be
prevented by an appropriate response potential. In practice, however, this
can become an issue. Our philosophy will be to treat the net transfer prob-
abilities seriously even when no response is included in the calculations.
While the process of electron removal from the target leads to the total
positive ion yield σ+, more detailed information is obtained from q-fold
electron removal cross sections σq with
N

σ+ = qσq . (24)
q=1
For the projectile space the corresponding detail is given by k-fold capture
cross sections which sum to net capture according to
N
cap
σcap = kσk . (25)
k=1
From this expression it becomes clear that for a highly charge-asymmetric

system (a low-charge projectile scattering from a multi-electron target)
there is the problem that more electrons can be transferred independently
from the target to the projectile than what is allowed physically.
The cross sections to find n electrons in the continuum σnion add up to
the net continuum cross section according to
N

σ− = nσnion . (26)
n=1
The next level of sophistication is given by charge-state correlated cross

sections. These are obtained in experiment by a coincidence counting tech-
nique: by determining the charge states of projectile and target after the
collision for each event one has also determined the number of electrons
transferred to the continuum. In a purely statistical treatment of the density
operator these charge-state correlated processes are related by trinomial

statistics. In a treatment based on the one-particle density matrix Pauli cor-
relations are included, as well. Examples of charge-state correlated processes
are pure (direct) ionization of the target, pure capture (no simultaneous
transfer of electrons to the continuum), or transfer ionization—a process in
which both capture and transfer of target electrons to the continuum occurs.
All these cross sections can be obtained from microscopic calculations,
i.e., from the independent-electron approach to the N-electron TDSE.
A substantial body of data has been explained at this level in ion–atom
collisions. The practical goal of an IPM is not to describe all of the multi-
electron processes, but to concentrate on the dominant channels. How many
of these channels can be described successfully depends on the strength of
the projectile interaction: for a given projectile charge ZP it was found that
independent-electron dynamics works for more of the N target electrons in
proportion to ZP (see Ref. 19).
Compared to atomic targets, molecules present an additional challenge.
The production of recoils such as H2 Oq+ results in fragmentation even for
charge states as low as q = 1. T he fragmentation occurs on time scales longer
than the collision itself, and therefore needs to be modelled as a post-collision
phenomenon. At high collision energies, i.e., when single-electron removal
(q = 1) dominates completely, it is possible to deduce branching fractions
for H2 O+ survival vs. production of OH+ , H+ , O+ (and in principle also
H+ 2 ). The main ionic channels can be linked to exclusive single-vacancy
production in one of the MOs. At intermediate and low energies where
multiple electron removal from the target plays an increasingly bigger role
fragmentation patterns can be modelled by supplementing the q = 1 rule
with branching rules for q = 2 and higher. There is some support for such
(semi-)phenomenological assignments from coincidence measurements of
water molecule fragments following proton impact at intermediate ener-
gies.6 It led to the first description of a large body of experimental data.24
We provide a series of comparisons of our data with experiments for
proton–water molecule collisions at collision energies above 20 keV and
also discuss some selected results for He+-impact. Where applicable we
compare our results with three-center classical trajectory calculations. For
proton impact, we progress through the sequence laid out above: net (gross)
cross sections, then q-fold electron processes (recoils, transfer to the projec-
tile, transfer to the continuum), and finally, fragmentation. The latter can be
observed also specifically for electron removal by capture, or by transfer to
the continuum.
3.1 Net cross section results

In Figure 11.1 we show the net capture cross sections in p − H2 O colli-
sions in comparison with experiment and with the three-center CTMC
calculations of Illescas et al.17 The results from the BGM-IPM follow
the experimental data very well in the low-to-intermediate energy
regime. They overestimate the data at energies above 200 keV displaying
a systematic deviation by about a factor of two. The CTMC calcula-
tions show a somewhat different energy dependence and lie above our
calculations at both low and very high energies. We cannot offer any
explanations for the discrepancy between the data and our results at
high energies.
In Figure 11.2 a similar comparison is given for transfer of electrons
to the continuum. Note that in contrast to Figure 11.1 the vertical axis is
linear. The experimental data display visible statistical errors on this scale,
and possibly a systematic discrepancy. Our calculations agree with the data
at the 30% level. The CTMC calculation is in better agreement with the
data over the entire energy range. The classical calculation employs a more
complete three-center geometry for the collision calculation, which is
likely to be the main reason for the more favorable comparison. In the limit
of very high energies it should not yield the correct energy dependence,
net
net [17]
Figure 11.1 Net capture σcap in p − H2 O collisions vs. projectile energy. Experiments:
solid circles—Rudd et al. (1985);4 open circles: Toburen et al. (1968).33 Solid line: present
theory; dotted line: three-center CTMC calculation of Ref. 17.
net
net [17]
Figure 11.2 Net ionization σ− in p − H2 O collisions vs. projectile energy. Experiments:

solid circles—Rudd et al. (1985);4 open circles: Bolorizadeh and Rudd (1986).34 Solid
line: present theory; dotted line: three-center CTMC calculation of Ref. 17.
however, since a classical-trajectory calculation based upon a microcanoni-

cal distribution does not agree with the Bethe–Born limit.
The net recoil production cross section σ+ is equal to the sum of net
capture and net transfer to the continuum. We present it in Figure 11.3 to
demonstrate the fall-off of the net ionization cross section at high energies.
The figure also shows how more recent measurements of the individual
recoil yields (H2 O+ , OH+ , H+ , O+ , O2+ ) from three different experiments
add up to σ+. They mostly agree with the original data of Rudd et al.,4 but
also show that there is a small uncertainty about the absolute height at the
highest energies. The three-center CTMC calculation does well at high
energies, but appears to overestimate the experimental data below 50 keV.
The figure also displays how within the BGM-IPM the net recoil pro-
duction is dominated at high energies by single-electron removal (q = 1),
but that multiple electron processes become important at intermediate and
low energies.
Figure 11.4 shows neutral He production in He+–H2 O collisions. In
the case where projectiles carry active electrons into the collision one
cannot define a net capture cross section in a simple way. As shown in
net
S(q=1)
D (q=2)
net [17]
Figure 11.3 Total recoil ion production σ+ in p − H2 O collisions vs. projectile energy.
Experiments: solid circles—Rudd et al. (1985);4 other symbols: sum of fragment yields
for Refs. 6, 8, 9 respectively. Solid line: present results; dotted line: three-center CTMC
calculation of Ref. 18. Dashed and dash-dotted lines: present results for σq with q = 1, 2.
T electrons only
inclusive
Figure 11.4 Neutral He production in He+ − H2 O collisions vs. projectile energy.

Experiments: solid circles—Rudd et al. (1985);5 triangles—Garcia et al. (2008).10 Solid
line: all electrons, including the projectile electron are used in the Slater determinant
analysis; dashed line: net capture based on target electrons only.
Ref. 23 the net recoil production cross section σ+ can be reduced consid-
erably when projectile electron transfer to the target is allowed. In lieu of
the net capture cross section (25) it is necessary to define a cross section
for neutral helium production, which takes into account both projectile
electron loss, as well as single and double electron transfer to the projec-
tile from the target. In addition, one has to consider Pauli blocking, since
in collision events where the projectile electron is not lost He formation
(in the ground state) requires a spin-singlet as the final state. This is taken
care of in the inclusive analysis.
We demonstrate the effect in Figure 11.4 to show that the experi-
mental data are described very well by the inclusive analysis. A calculation
that ignores projectile electrons overestimates neutral He production by
approximately a factor of two, consistently over the entire energy range.
The fact that it does not run perfectly parallel to the full analysis can be
associated with the fact that electron loss from the projectile varies with
impact energy. Note that the theoretical results shown are inclusive in elec-
tron transfer to the continuum, i.e., they do not represent pure capture, but
include simultaneous target ionization events.
net
T electrons only
Figure 11.5 Net free-electron production in He+ − H2 O collisions vs. projectile energy.
Experiments: solid circles—Rudd et al. (1985);5 open circles—σ− minus projectile elec-
tron loss cross section; triangles—Garcia et al. (2008).10 Solid line: σ− calculation includ-
ing projectile electron; dashed line: based on propagation of target electrons only.
Figure 11.5 displays σ−, as well as σ− − σloss (where σloss is the projectile
electron loss cross section) in order to highlight the role of the projectile
electron. From the experimental data of Rudd et al. one concludes that
projectile electron removal becomes an appreciable contributor to σ− at
energies above 40 keV/amu. Note that σ− is large to begin with, since
the water molecule is easily ionized. A theoretical calculation that ignores
projectile electron contributions to σ− is seen to follow the experimen-
tal σ− σloss quite well; it displays a broader maximum than in the related
p − H2 O case shown in Figure 11.2 (this calculation is very close to the
p − H2 O case for large impact parameters, but different for close colli-
sions). The full calculation for He+–H2 O collisions agrees well with σ− at
high energies, but overestimates the data below EP = 70 keV/amu. The
projectile electron is bound too loosely in the present model, it experi-
ences a long-range Coulomb potential from the water molecule even on
the incoming part of the trajectory, and therefore is removed too easily at
lower energies.
3.2 Fragmentation cross sections

In Figure 11.6 we compare our calculated recoil production cross sec-
tions σq ; q = 1 . . . 3 for proton–water collisions with experimental data
of Werner et al. For the q = 3 case it is safe to assume that the measured
coincidences for H+ + H+ + O+ provide the dominant contribution. For
q = 1, 2 a model-dependent reconstruction was performed on the basis of
measured fragment yields. It can be seen that the BGM-IPM calculations
agree very well for q = 1, 2 and overestimate for q = 3 by about a factor
of two. It was then argued25 that fragment yields can be obtained from the
cross sections for single (q = 1), double (q = 2), and triple (q = 3) electron
removal σS,D,T according to:
σH2 O+ = 0. 68σS , (27)
σOH+ = 0. 16σS + 0. 6σD , (28)
σH+ = 0. 13σS + 1. 2σD + 1. 0σT , (29)
σO+ = 0. 03σS + 0. 2σD + 0. 5σT . (30)

This model as motivated by collision data at 100–350 keV energies takes
into account known branching ratios for the q = 1 case (high-energy
S (q=1)
D (q=2)
T (q=3)
Figure 11.6 Recoil ion production σq in p − H2 O collisions vs. projectile energy.

Experiments: q = 1: squares (model-dependent reconstruction from the data of
Werner et al. (1995);6 q = 2: circles. Present theory: q = 1 – dashed; q = 2 dash-
dotted; q = 3 dotted. Also shown: coincidence measurements from Ref. 6: triangles:
H+ + H+ + O+ (q = 3); crosses: H+ + H+ + O .
0,−
proton collision, electron collisions, photoionization), and blends them

with an assumption about which fragments are created when more elec-
trons are removed. At lower energies these multiple electron removal
events are due to processes such as transfer ionization and pure multiple
ionization.
In Figure 11.7 it is shown that the microscopic BGM-IPM calculations
when supplemented with the post-collision fragmentation model are quite
capable of describing the details of the fragmentation pattern over a wide
range of energies. The effect of multiple-electron removal is to produce a
reversal in strength of the H+ versus OH+ yields, and a dramatic increase of
O+ yields at low energies. The H2 O+ channel (which is exclusive to q = 1)
is seen to be predicted too low by the theory below 100 keV impact energy.
There are more detailed experimental data at low energies, namely
fragmentation following electron capture vs. pure ionization. An analysis
can be found in Ref. 25.
+
H2 O
+
OH
+
H
+
O
Figure 11.7 Fragmentation cross sections in p − H2 O collisions vs. projectile energy.

Experiments: Squares: H2 O+; triangles: H+, circles OH+; diamonds: O+. Symbols with
crosses: Werner et al.;6 solid symbols: Gobet et al.;8 open symbols: Luna et al.9 Present
theory: H2 O+—solid line; H+—dash-dotted; OH+—dashed; O+—dotted.
4. SUMMARY AND OUTLOOK

In the past 15 years the density functional theory based calculations
using the basis generator method as a means to propagate Kohn–Sham-type
orbitals have explained a wealth of data in ion–atom collisions.While much
emphasis was placed on neon and argon target atoms, successful applica-
tions to an open-shell system such as atomic oxygen were also developed.
In recent years we extended this work to simple molecules and applied
the formalism to the case of H2 O targets. In the regime of fast collisions
(above 10 keV/amu) we can safely ignore molecular rotational and vibra-
tional degrees of freedom due to their longer time scales. Fragmentation
of the ionized molecule happens also on much longer time scales than the
collision itself, and is therefore not treated as part of the time-dependent
dynamics.
Nevertheless, the extension of the TDDFT approach to molecular
targets faces the following challenges: (i) in principle, multi-center matrix
elements of the electrostatic interactions are necessary; (ii) an orientation
average has to be carried out when total cross sections are calculated. In
the present work multi-center matrix elements are avoided by working in
an energy representation of the target Hamiltonian. The calculations are

carried out very efficiently by representing the molecular orbitals (from
a restricted HF calculation) in terms of an atomic oxygen basis. A short
coming of this representation may be that the orbital densities are not
accurate as far as electric multipole moments are concerned. On the other
hand, it may actually help by requiring a reduced effort in the orientation
averaging procedure. For most of the results shown this does not appear to
be a significant weakness. Some work in the future should be dedicated,
however, to test these assumptions in more detail.
The independent particle model (using a determinantal analysis) does
respect the Pauli principle. This becomes important even for basic total cross
sections when the projectile carries electrons into the collision. Some open
questions remain in this subfield, and should be addressed by using a more
sophisticated effective potential which is common to both target and projec-
tile electrons, but which does not suffer from incorrect asymptotic behavior.
The dynamical TDDFT calculations (which so far are carried out only at
the no-response level) are then supplemented with a semi-phenomenological
fragmentation model.This model takes into account multi-electron processes
as predicted by the independent particle model. Fragmentation patterns
change as one goes from the high-energy limit towards lower energies.Their
main features have been explained by the present work for the first time.
Future work should concentrate on some improvements. In particular,
we can imagine that dynamic response will play a role at energies below
100 keV/amu—it ought to suppress some of the multi-electron processes.
More realistic molecular orbital representation is also on our wish list. A
short-term goal, however, should be the application of the methodology to
other small molecules.
ACKNOWLEDGMENTS
This work was supported by NSERC Canada and by SHARCNET.
REFERENCES
1. Belkić, Dž. Review of theories on ionization in fast ion–atom collisions with prospects
2. Boudaïffa, B.; Cloutier, P.; Hunting, D.; Huels, M. A.; Sanche, L. Resonant formation of
DNA strand breaks by low-energy (3 to 20 eV) electrons. Science 2000, 287, 1658.
3. Tavernelli, I.; Gaigeot, M.-P.;Vuilleumier, R.; Stia, C.; Hervé du Penhoat, M.-A.; Politis,
M.-F. Time-dependent density functional theory molecular dynamics simulations of
liquid water radiolysis. Chem.Phys.Chem. 2008, 9, 2099.
4. Rudd, M. E.; Goffe, T.V.; DuBois, R. D.; Toburen, L. H. Cross sections for ionization of
water vapor by 7–4000 keV protons. Phys. Rev. A 1985, 31, 492.
5. Rudd, M. E.; Itoh, A.; Goffe, T. V. Cross sections for ionization, capture, and loss for
5–450 keV He+ on water vapor. Phys. Rev. A 1985, 32, 2499.
6. Werner, U.; Beckord, K.; Becker, J.; Lutz, H. O. 3D imaging of the collision-induced
Coulomb fragmentation of water molecules. Phys. Rev. Lett. 1995, 74, 1962.
7. Gobet, F.; Farizon, B.; Farizon, M.; Gaillard, M. J.; Carré, M.; Lezius, M.; Scheier, P.;
Märk,T. D.Total, partial and electron-capture cross sections for ionization of water vapor
by 20–150 keV protons. Phys. Rev. Lett. 2001, 86, 3751.
8. Gobet, F.; Eden, S.; Coupier, B.;Tabet, J.; Farizon, B.; Farizon, M.; Gaillard, M. J.; Carré, M.;
Ouaskit, S.; Märk,T. D.; Scheier, P. Ionization of water by (20–150) keV protons: separation
of direct-ionization and electron-capture process. Phys. Rev. A 2004, 70, 062716.
9. Luna, H.; de Barros, A. L. F.; Wyer, J. A.; Scully, S. W. J.; Lecointre, J.; Garcia, P. M. Y.;
Sigaud, G. M.; Santos, A. C. F.; Senthil, V.; Shah, M. B.; Latimer, C. J.; Montenegro, E.
C. Water-molecule dissociation by proton and hydrogen impact. Phys. Rev. A 2007, 75,
042711.
10. Garcia, P. M.Y.; Sigaud, G. M.; Luna, H.; Santos, A. C. F.; Montenegro, E. C.; Shah, M. B.
Water-molecule dissociation by impact of He+ ions. Phys. Rev. A 2008, 77, 052708.
11. Gabás, P. M. M.; Errea, L. F.; Méndez, L.; Rabadán, I. Ab initio treatment of charge
transfer in ion-molecule collisions based on one-electron wave functions. Phys. Rev. A
2012, 85, 012702.
12. Boudrioua, O.; Champion, C.; Dal Cappello, C.; Popov,Y.V. Ab initio calculation of dif-
ferential and total cross sections for the ionization of water vapor by protons. Phys. Rev.
A 2007, 75, 022720.
13. Dubois, A.; Carniato, S.; Fainstein, P. D.; Hansen, J. P. Ionization of water molecules by
fast charged projectiles. Phys. Rev. A 2011, 84, 012708.
75, 493.
15. Errea, L. F.; Illescas, Clara; Méndez, L.; Pons, B.; Rabadán, I.; Riera, A. Classical calcula-
tion of ionization and electron-capture total cross sections in H+ − H2 O collisions.
Phys. Rev. A 2007, 76, 040701R.
16. Abbas, I.; Champion, C.; Zarour, B.; Lasri, B.; Hanssen, J. Single and multiple cross sec-
tions for ionizing processes of biological molecules by protons and ß-particle impact: a
classical Monte Carlo approach. Phys. Med. Biol. 2008, 53, N41.
17. Illescas, Clara; Errea, L. F.; Méndez, L.; Pons, B.; Rabadán, I.; Riera, A. Classical treat-
ment of ion-H2 O collisions with a three-center model potential. Phys. Rev. A 2011, 83,
052704.
18. Kirchner, T.; Horbatsch, M.; Lüdde, H. J.; Dreizler, R. M. Time-dependent screen-
ing effects in ion–atom collisions with many active electrons. Phys. Rev. A 2000, 62,
042704.
19. Kirchner, T.; Horbatsch, M.; Lüdde, H. J. Nonperturbative calculation of charge-chang-
ing processes in C4+ scattering from neon atoms. Phys. Rev. A 2001, 64, 012711.
20. Kirchner, T.; Horbatsch, M.; Lüdde, H. J. Time-dependent independent-particle model
−
calculation of multiple capture and ionization processes in p–Ar, p –Ar, and He2+–Ar
collisions. Phys. Rev. A 2002, 66, 052719.
21. Kirchner, T.; Lüdde, H. J.; Horbatsch, M. A time-dependent quantal approach to elec-
tronic transitions in atomic collisions. Recent Res. Dev. Phys. 2004, 5, 433.
22. Lüdde, H. J.; Spranger, T.; Horbatsch, M.; Kirchner, T. Nonperturbative, quantum-

mechanical approach to ion collisions from molecular targets. Phys. Rev. A 2009, 80,
060702R.
23. Kirchner, T.; Murakami, M.; Horbatsch, M.; Lüdde, H. J. Calculations for charge transfer
and ionization in heavy-particle collisions from water molecules. J. Phys.: Conf. Ser., 2012,
388, 012038 (Part 1); Murakami, M.; Kirchner, T.; Horbatsch, M.; Lüdde, H.J. Quantum
-mechanical calculation of multiple electron removal and fragmentation cross sections in
He+ − H2 O collisions. Phys. Rev. A 2012, 86, 022719.
24. Murakami, M.; Kirchner, T.; Horbatsch, M.; Lüdde, H. J. Single and multiple electron
removal processes in proton–water-molecule collisions. Phys. Rev. A 2012, 85, 052704.
25. Murakami, M.; Kirchner, T.; Horbatsch, M.; Lüdde, H. J. Fragmentation of water mol-
ecules by proton impact: the role of multiple electron processes. Phys. Rev. A 2012, 85,
052713.
26. Runge, E.; Gross, E. K. U. Density-functional theory for time-dependent systems. Phys.
Rev. Lett. 1984, 52, 997.
27. Pople, J. A. Nobel lecture: quantum chemical models. Rev. Mod. Phys. 1999, 71, 1267.
28. Kirchner, T.; Horbatsch, M.; Lüdde, H. J. Coupled mean-field description of electron
removal processes in He+ + Ne and He+ + Ar collisions. J. Phys. B 2004, 37, 2379.
29. Kirchner, T.; Santos, A. C. F.; Luna, H.; Sant’Anna, M. M.; Melo, W. S.; Sigaud, G. M.;
Montenegro, E. C. Charge-state-correlated cross sections for electron loss, capture, and
ionization in C3+−Ne collisions. Phys. Rev. A 2005, 72, 012707.
30. Lüdde, H. J.; Dreizler, R. M. Comments on inclusive cross sections. J. Phys. B 1985,
18, 107.
31. Kroneisen, O. J.; Lüdde, H. J.; Kirchner, T.; Dreizler, R. M. The basis generator method:
optimized dynamical representation of the solution of time-dependent quantum prob-
lems. J. Phys. A 1999, 32, 2141.
32. Aung, S.; Pitzer, R. M.; Chan, S. I. Approximate Hartree–Fock wavefunctions, one-
electron properties, and electronic structure of the water molecule. J. Chem. Phys. 1968,
49, 2071.
Rev. 1968, 171, 171.
ejection of electrons from water vapor. II. 15–150 keV proton impact. Phys. Rev. A 1986,
33, 888.
CHAPTER TWELVE
Four-Body Theories for Transfer

Ionization in Fast Ion-Atom
Collisions
Dževad Belkića, Ivan Mančevb and Nenad Milojevićb
aNobel Medical Institute, Karolinska Institute, Stockholm, Sweden
bDepartment of Physics, Faculty of Sciences and Mathematics, University of Niš, Niš, Serbia
Contents
1. Introduction 339
2. The independent particle/event models 340
3. The four-body continuum distorted wave method 347
4. The four-body Born distorted wave method 354
5. Conclusions 358
Acknowledgments 360
References 360
Abstract
Total cross sections for high-energy transfer ionization in collisions between bare
projectiles with helium are investigated using several three- and four-body methods
in the semi-classical impact-parameter and quantum-mechanical formalisms. The
results are compared with experimental data for transfer ionization in the H+ — He,
He2+ — He, and Li3+—He collisions. It is found that the cross sections from the four-
body continuum distorted wave and the four-body Born distorted wave methods
compare favorably with the available measured findings, especially at higher impact
energies that are within the expected region of validity of these theories.
1. INTRODUCTION
Two-electron transitions in four-body fast ion-atom collisions have
been extensively studied in the last two decades both theoretically and
experimentally. This includes transfer ionization (TI) as a collision between
a nucleus as a projectile and a two-electron target from which one electron
is captured and the other ionized. Transfer ionization, which is the subject

340 Dževad Belkić, Ivan Mančev and Nenad Milojević
of the present study, remains one of the most challenging collisional pro-
cesses with two active electrons, even in the simplest case which involves
helium-like targets. This collisional phenomenon has opened interesting
and fruitful discussions about different scattering mechanisms within the
first- and second-order theories, including the role of static as well as
dynamic electron correlations.
The majority of the theoretical investigations1–5 that have considered TI
collisions employed the independent particle model (IPM) and the inde-
pendent event (IEV) model. The IPM and IEV models ignore the inter-
electron correlations from the outset and compute the probability for TI as a
product of the individual probabilities for capture of one electron and inde-
pendent ionization of the other electron.The basic feature of these previous
investigations within the IPM and its variants is the preservation of a pure
three-body formalism, despite the fact that the studied four-body problems
include two active electrons. The IPM has been applied to TI within the
continuum distorted wave (CDW-IPM) and the continuum distorted wave
eikonal initial state (CDW-EIS-IPM) approximations. Although the IPM
or IEV versions of other competitive models were outperformed by the
CDW-IPM and CDW-EIS-IPM approximations, these two latter methods
still gave cross sections that were larger than the experimental values, espe-
cially for the case of the H+ impact. Such drawbacks can be partially over-
come by employing the four-body formalism of scattering theory within
e.g., the continuum distorted wave (CDW-4B)6, 7 and Born distorted wave
(BDW-4B)8 approximations that are capable of yielding satisfactory agree-
ment with the existing experimental data. This will be demonstrated in the
present review.
2. THE INDEPENDENT PARTICLE/EVENT MODELS

The simplest way to carry out a theoretical examination of TI in fast
ion-atom (two- or multi-electron) collisions is the IPM, in which the inter-
action between the electrons is considered primarily in the so-called mean
field approximation. In this picture each electron evolves independently in
an effective potential created by the other electrons. For evaluations of cross
sections for multiple processes, this method requires the computation of
single-particle probabilities as a function of the impact parameter.
In general, the IPM has been frequently used for describing collisions of
multiply-charged nuclei with helium-like targets, but many computations
Four-Body Theories for Transfer Ionization in Fast Ion-Atom Collisions 341
have shown that the IPM model fails to reproduce experimental data
in a systematic manner, thus casting doubt about its practical usefulness.
Therefore, it can be anticipated that multiple processes induced by impact
of light bare ions should require a theoretical description that goes beyond
the IPM when single-particle probabilities are obtained from the existing
perturbation models. One method that achieves this purpose, without actu-
ally solving the fully correlated problem, is the IEV model of Dunseath and
Crothers,1 who studied the TI process using the three-body CDW method
for computations of single-particle probabilities. In the IEV model the TI
process is a result of different steps (events) involving one active electron.
The net probability is computed as the product of individual probabili-
ties for each event. The main feature is that the second electron makes a
transition to the final state, not from its initial unperturbed state, but from
an intermediate state where it was left after the first electron made the
transition to its final state. The case where the electrons follow different
reaction channels is taken into account in the IEV model through differ-
ent sequences of events. It is clear that a full description of the processes
depends on the time sequence in which the events are produced. The IEV
model gives a reasonable framework to deal with multiple-electron pro-
cesses. Computations of total cross sections for TI in He2+ + He collisions
by Dunseath and Crothers,1 using the CDW model to obtain the single-
particle probabilities, show an improvement with respect to the IPM in
comparison with the available experimental data.
The three-body CDW method for single ionization was formulated
by Belkic.9 In this method, the distortions due to electronic continuum
intermediate states are properly included in the entrance and exit chan-
nels. Nevertheless, the corresponding full Coulomb wave function within
the initial scattering state was found in applications to lead to overestima-
tion of experimentally measured total cross sections near and below the
Massey peak. Typically, all total cross sections computed by the CDW
method keep on rising as the impact energy decreases, whereas the corre-
sponding experimental data generally decline in the same region exhibit-
ing the Massey peak. This bending of the curve for the total cross sections
can be obtained within the CDW methodologies, if the full Coulomb
wave function for the distortion of the initial state in the entrance channel
is approximated by its long-range asymptotic eikonal form through the
well-known logarithmic factor. The product of this latter factor and the
unperturbed initial state in the entrance channel represents the eikonal
initial state (EIS). The resulting simplification of the CDW method is
abbreviated as CDW-EIS.10 In this way, the Massey peak is systematically

reproduced by the CDW-EIS method, leading to quantitative agreement
with measurements. The CDW and CDW-EIS methods are computation-
ally attractive, since the fully analytical expressions are available for the
transition amplitudes for ionizing collisions involving one active electron.
Recently Belkic12 reviewed theories on ionization in fast ion-atom col-
lisions with prospects for applications to hadron therapy for patients with
deep-seated cancer for which high-energy ions (∼400 MeV/amu) are
employed.
We shall first present the CDW-EIS approximation for TI within the
IPM (CDW-EIS-IPM) and IEV (CDW-EIS-EIV) models. The CDW-EIS
approach has been introduced by Crothers and McCann10 for single ion-
ization of hydrogen-like targets and later extended by Fainstein et al.11
to multi-electronic targets with a remarkable success in many applica-
tions. In the CDW-EIS-IPM and CDW-EIS-EIV both probabilities for
electron capture and ionization are computed by means of the CDW-EIS
approximation.
Using the straight-line version of the IPM, the initial and final undis-
torted wave functions are distorted by multiplicative factors. These dis-
torted waves satisfy the asymptotic conditions for the Coulomb potentials
in both channels. In the entrance channel the target bound-state wave
function is multiplied by an eikonal phase which, at large distances,
describes the state of the active electron in the projectile field. In the
outgoing channel the distortion scheme varies according to different
reactions. For electron capture (or ionization) the projectile bound-state
(or target continuum) wave function is multiplied by a Coulomb factor
which accounts for the fact that the active electron is simultaneously in
the residual target (or projectile) continuum. Importantly, such a model-
ing acknowledges the following two features. First, in the final state the
emitted or captured electron evolves in a two-center potential. Second,
the perturbation represents the residual potential obtained from the
application of the full Hamiltonian to the ansatz for the initial or final
distorted waves. The differences between the CDW and CDW-EIS meth-
ods are in the initial bound-state distortion and the residual potential.
Several possibilities are available for the construction of the target wave
functions. Thus, for example, Galassi et al.2 followed Fainstein et al.11
and used analytical wave-functions that permit obtaining the transition
amplitude in a closed form. This is very useful since the computations
of the single-particle impact-parameter-dependent probabilities could be

very time consuming.
For computations of the total cross sections for a TI process, we will
employ the IPM and IEV. To this end, it is necessary to obtain the transition
amplitude as a function of the impact parameter ρ. The differential cross
section for the TI process is given by:
∞
dQ
= 2π dρρPTI (ρ, κ� ). (1)
κ
d� 0
The quantity PTI (ρ, κ ) in the case of IPM is expressed by:
(IPM)
PTI (ρ, κ� ) = 2PI (ρ, κ� )PC (ρ). (2)
Here, PI (ρ, κ ) is the ionization probability per electron as a function of

impact parameter ρ and the emitted electron momentum κ in the final
state. Further, PC (ρ) is the impact-parameter-dependent single-particle
probability for electron capture which is computed using the three-body
CDW-EIS model, as in the work of Busnengo et al.13
In the IEV model the collisional process involving helium can occur
by following two different pathways. One electron can become ionized
first, and then the other electron may be captured (ionization-capture). Or,
capture of one electron occurs first, and then the other electron is emitted
(capture-ionization). If we label the two electrons of helium as 1 (the first
active electron) and 2 (the second active electron) the probability for TI
can be written as:
(IEV)
PTI (ρ, κ� ) = PI1 (ρ, κ� )PC2 (ρ) + PI2 (ρ, κ� )PC1 (ρ). (3)
Quantity PI1 (or PI2 ) is the probability that electron 1 (or 2) is emitted,
whereas PC1 (or PC2 ) is the probability that electron 1 (or 2) is captured.
In the IPM model, we have PI1 = PI2 and PC1 = PC2. Therefore, taking
these features into account within Eq. (3), one can retrieve Eq. (2). In
contrast, in the IEV model these probabilities are different and they are
computed as follows. If the electron 1 is ionized, then PI1 corresponds to
single ionization of helium, as computed using e.g., the CDW-EIS model
according to the study by Fainstein et al.11 Further, PC2 is the probability
for one-electron capture from He+ (1s) which may also be computed using
the CDW-EIS model, as done by Martínez et al.14 In the second pathway,
the first electron is captured from helium and PC1 is computed using the
CDW-EIS model with the Roothaan–Hartree–Fock wave functions of
Clementi and Roetti15 to describe the initial active electron orbital. In the
computation from Ref. 2, capture is included to the ground state as well
as to the excited states with the principal quantum numbers n = 2 and 3.
The non-captured electron is supposed to remain frozen during this first
step. The second step corresponds to ionization of He+ (1s) with the prob-
ability PI2. In the case of the H+ impact, PI2 corresponds to ionization of
He+ (1s) by a neutral hydrogen atom and is computed considering that the
projectile electron remains frozen during this second step. Thus, we take
into account a screening effect due to the projectile electron, but neglect
the anti-screening effects (excitation of the projectile electron). In the
experiment of Bernardi et al.16 the emitted electron is recorded in coinci-
dence with the neutral projectile and, therefore, a process of simultaneous
emission from the projectile and target cannot occur. If the first step cor-
responds to capture to the projectile ground state, two-center effects will be
of minor importance for the H(1s) impact and thus PI2 can be computed
using e.g., the first Born approximation of Bates and Griffing.17, 18 In the
case of capture to the excited states H(nlm) (with n 2), we consider that
capture has been produced at small impact parameters, so that TI is domi-
nated by this region.
Since the excited projectile electron remains, on the average, outside
the interaction region, it is assumed that it does not contribute to the
ionization reaction. In this way, the process is considered to be produced
by the projectile nucleus and, therefore, the CDW-EIS model can be
employed. The same assumptions are used to study the He+ impact case.
The only difference with the hydrogen atom impact is in the fact that a
Coulomb asymptotic perturbation potential of the effective charge +1
appears for He+. Thus, for the He+ (1s) impact, the CDW-EIS compu-
tation is performed corresponding to an effective projectile Coulomb
potential with the net charge +1, whereas for the He+ (nlm) (n 2), the
CDW-EIS is applied using the effective projectile Coulomb potential of
charge +2. Hence, the IEV model requires not only the single capture and
ionization probabilities for a neutral helium atom, but also for the He+ ion.
Although multiple differential cross sections provide detailed infor-
mation about collision processes, the total cross sections also yield good
insights into global trends. Therefore, in the present review, we shall
focus on total cross sections. Total cross sections for TI are obtained
from differential cross sections after integration over the emitted
Figure 12.1 Total cross sections Q(cm2) as a function of the laboratory incident energy
E(keV/amu) for transfer ionization: He2+ + He −→ He+ + He2+ + e. The full curve:
CDW-EIS-IEV model;2 the dot-dashed curve: CDW-EIS-IPM model;2 the dashed curve:
CDW-IEV model;1 the symbol ◦ represents the theoretical result of the CDW-Born-IPM.3
Experimental data: n Shah and Gilbody.20
electron momenta (κ). The results of total cross sections for TI in the
He2+ + He and H+ + He collisions are plotted in Figures 12.1 and
12.2, respectively. It is seen that with the CDW-EIS-EIV model,2 there
is an improvement with respect to the findings from the CDW-IEV
approach of Dunseath and Crothers,1 most notably at the intermediate
impact energies. This could be a confirmation of the conjecture of these
authors that the normalization of the CDW wave function might be the
reason for which the CDW method overestimates the experimental data
for TI. The computations in the CDW-EIS-IPM2 are in good qualitative
agreement with the experimental data. Nevertheless, with the CDW-
EIS, the IEV model yields better results than the IPM. This supports the
picture of a process in two successive steps. Notice that the difference
between these two models is larger at intermediate impact energies and
Figure 12.2 Total cross sections Q(cm2) as a function of the laboratory incident

energy E(keV/amu) for transfer ionization: H+ + He −→ H + He2+ + e. The full curve:
CDW-EIS-IEV model;2 the dot-dashed curve: CDW-EIS-IPM model;2 the dashed curve:
CDW-IEV model;1 the symbol ◦ represents the theoretical result of the CDW-Born-IPM.3
Experimental data: n Shah and Gilbody.20
that they converge to each other at high energies. At high energies the
ionization-capture channel dominates, but capture-ionization still gives
a significant contribution of the order of 20%. Therefore, confluence of
the IEV and IPM at high energies can be attributed to a subtle addition
from different channels and not necessarily to the dominance of one of
the channels.
Gayet and Salin3 used the IPM and the Hartree–Fock wave function
for the initial two-electron bound state of helium and assumed that the
second electron is not affected when the first electron is ionized or cap-
tured. In their computations,3 the capture amplitude was evaluated within
the CDW-3B approximation by including a contribution from capture into
low-lying excited states of the projectile. The ionization amplitude was
computed using the first Born approximation. The authors from Ref. 3
carried out their computations only at a single energy (400 keV/amu) and
the results of such a CDW-Born model are displayed by the open circle in
Figures 12.1 and 12.2.
We can conclude that the total cross sections in the CDW-EIS-IEV

model are in better agreement with the experimental data than those due
to the CDW-EIV model. However, the results of the CDW-EIS-IEV are
still larger than the experimental values, especially for the case of proton
impact. In contrast, the cross sections from the IPM overestimate the
experimental data, even if we use the single-particle probabilities from the
CDW-EIS approximation. This supports a description of the investigated
process as a combination of successive events in different paths. It should
be re-emphasized that the IEV and IPM completely ignore the dynamic
correlations and compute the net probability for transfer ionization as a
product of the individual probabilities for transfer of one electron and
independent ionization of the other electron.
Focusing our interest on the case of a bare swift ion impacting on a
two-electron atomic target, a four-body theoretical representation appears
as necessary in order to include the static and dynamic correlations.
Therefore, we propose the formalism of a four-body reaction under study
as a pure four-body problem, thus avoiding the said limitations of a three-
body representation. The four-body formalism of scattering theory within
e.g., the CDW-4B and BDW-4B approximations could be anticipated as
being more capable of yielding satisfactory agreement with the existing
experimental data.
3. THE FOUR-BODY CONTINUUM DISTORTED

WAVE METHOD
The CDW-4B method for TI has been formulated and implemented
by Belkic et al.21 for the He2+ + He collisions and good agreement was
found with measurements. Later, the CDW-4B method employed by
Mančev22 has also been reported to be in satisfactory agreement with the
experimental data for the Li3+ + He collisions.
Let s1, 2 and x1, 2 be the position vectors of the electrons e1,2 relative
to the nuclear charges ZP and ZT, respectively. Let electron e1 be captured
and electron e2 simultaneously ionized. We denote by R the position vec-
tor of the target nucleus (T) with respect to the projectile nucleus (P). The
distance between the two electrons will be denoted by r12 . The transi-
tion amplitudes in the “prior” (Tif− ) and “post” (Tif+ ) forms for TI in the
CDW-4B theory can be written as:21

�
Tif− = N � d�s1 d�s2 (ρv)2iν e iα� ·�s1 +iβ·�x1 −i�κ ·�x2 ϕf∗ ( �s1 )1 F1 (iνT , 1, ivx1 + i v
dR � · x�1 )
×1 F1 (iζ , 1, ipx2 + i�p · x�2 ) [VP (R, s2 )1 F1 (iνP , 1, ivs1 + i v

� · �s1 )ϕi (�x1 , x�2 )
� x1 ϕi (�x1 , x�2 ) · ∇
−∇ � s1 1 F1 (iνP , 1, ivs1 + i v
� · �s1 )
(4)
−1 F1 (iνP , 1, ivs1 + i v
� · �s1 )(Ei − HT )ϕi (�x1 , x�2 )] ,

�
Tif+ =N � d�x1 d�x2 (ρv)2iν e iα� ·�s1 +iβ·�x1 −i�κ ·�x2 ϕi (�x1 , x�2 )
dR
×1 F1 (iνP , 1, ivs1 + i v � · �s1 )1 F1 (iζ , 1, ipx2 + i�p · x�2 )

× [VP (R, s2 ) + V (r12 , x1 )] 1 F1 (iνT , 1, ivx1 + i v � · x�1 )ϕf∗ ( �s1 )

−∇ � s1 ϕf∗ ( �s1 ) · ∇
� x1 1 F1 (iνT , 1, ivx1 + i v � · x�1 ) , (5)

1 1 1 1
VP (R, s2 ) = ZP − , V (r12 , x1 ) = − , (6)
R s2 r12 x1
where ν = ZP ZT /v, whereas the factor (ρv)2iν originates from inter-

nuclear potential ZP ZT /R and it disappears from differential Qif± (� κ ) as well
as total cross section Qif± after integration over η (see Eqs. (7) and (8)). The
symbol 1 F1 (a, b, x) stands for the usual Kummer hypergeometric function.
The momentum vector of the ejected electron e2 with respect to its parent
nucleus T is denoted by κ. The wave function of the initial bound state is
labeled by ϕi (x1 , x2 ), whereas ϕf ( s1 ) is the single electron hydrogen-like
wave function of the (ZP ; e1 ) system in the exit channel. The remaining
quantities in transition amplitudes Tif± are defined as follows:
∗ ∗
N = (2π)−3/2 N − (ζ )N ′ , N ′ = N + (νP )N − (νT ),
N − (ζ ) = Ŵ(1 + iζ )eπ ζ /2 ,
N − (νT ) = Ŵ(1 + iνT )eπ νT /2 , N + (νP ) = Ŵ(1 − iνP )eπνP /2 ,
ZP ZT − 1 ZT
νP = , νT = , ζ = , �p = v
� + κ� .
v v p
The momentum transfers α and β are given by:

v Q v Q
α� = η� − − �
v, β� = −�η− + �
v,
2 v 2 v
whereas the vector η is the transverse momentum transfer: α� + β� = −v �

and η� · v � = 0, where the impact velocity vector v is directed along
the Z -axis. The Q factor or the inelasticity factor is defined as:
Q = Ei − (Ef + Eκ ), where Ei and Ef = −ZP 2 /(2nf2 ) are the initial
and final binding energies, and Eκ = κ 2 /2. The “prior” transition
amplitude contains the term with the factor (Ei − HT )ϕi (�x1 , x�2 ) where
HT = −∇x21 /(2b) − ∇x22 /(2b) − ZT /x1 − ZT /x2 + 1/r12, with b = mT /
(mT + 1), and mT is the mass of the target nucleus. If the bound-
state wave function for helium were known exactly, the function
ϕi′ (�x1 , x�2 ) ≡ (Ei − HT )ϕi (�x1 , x�2 ) would vanish identically. Since the
helium exact wave function is unavailable, the contribution from ϕi′ (�x1 , x�2 )
is not equal to zero. This implies that this term, in principle, should be kept
throughout in the “prior” transition amplitude. Such an addendum to the
T -matrix element has been suggested by Belkic23 within the four-body
first Born approximation with the correct boundary conditions (CB1-4B)
for double electron capture. The numerical computations for double elec-
tron capture23 and for TI21 in the α–He collisions show that this correction
does not give a significant contribution at high impact energies. For this
reason, we shall not consider this term as a perturbation in the transition
amplitude.
The triple differential cross sections for the TI process is given by:
± 2
dQif± T
if (7)
Qif± (�
κ) ≡ = η
d� ,
κ
d� 2π v
whereas the total cross section reads as:

±
Qif = d� κ Qif± (�
κ ). (8)
As it has been shown in Ref. 21, after analytical calculations carried out by
means of the standard Nordsieck technique,24 the expressions for the total
cross sections for the TI process in the CDW-4B model can be reduced to
a seven-dimensional numerical quadrature.
In the “prior” and “post” forms of the transition amplitudes, there
is a common perturbation: VP (R, s2 ) = ZP (1/R − 1/s2 ). Of course,
considered outside the T-matrix element, the potential VP2 = −ZP /s2

represents the direct Coulomb interaction between e2 and ZP. Its

asymptotic value V ∞ (R) at large distances s2 is given by −ZP /R, since
P2
s2 −→ R as R −→ ∞. Hence, the term VP (R, s2 ) is precisely the differ-

ence between the finite and asymptotic value of the same overall short-range
potential VP (R, s2 ) = VP2 (s2 ) − VP∞2
(R), in accordance with the correct
boundary condition.7, 25–27 However, when placed in the T-matrices, the
potential VP2 plays the role of a perturbation which can cause capture of
electron e1. This could only occur through some kind of underlying cor-
relations between e2 and e1. For example, a part of the energy received by
the electron e2 in its collision with ZP could be sufficient to facilitate the
transfer of e1 to the projectile.
The “post” formTif+ contains an additional term:V (r12 , x1 ) = 1/r12 −1/x1,
which is completely absent from the Tif−. Here the dielectronic interac-
tion 1/r12 appears explicitly and when combined with the initial and final
distorted wave functions on both centers ZP and ZT, it describes the well-
known Thomas P-e-e scattering. Due to the perturbation V (r12 , x1 ), even
the total cross section in the “post” form should be more adequate than
its “prior” counterpart. The potential 1/x1 in V (r12 , x1 ) = 1/r12 − 1/x1
is the asymptotic tail of the 1/r12, since r12 → x1 as x1 → ∞. Hence,
the term V (r12 , x1 ) is the difference between the finite and asymptotic
values of the same potential. As such, V (r12 , x1 ) is a short-range interac-
tion, as required by the correct boundary conditions. Using the relation
r12 = |�x2 − x�1 |, we can develop 1/x1 = 1/|�r12 − x�2 | in a power series
3
around x2 according to 1/x1 = 1/|�r12 − x�2 | = 1/r12 − �r12 · x�2 /r12 + · · ·,
so that V (r12 , x1 ) = 1/r12 − 1/x1 = �r12 · x�2 /r12 + · · ·. This is justified by
3
the small value of the x2 coordinate (of the order of the Bohr radius a0),
since electron e2 always remains bound to the projectile. From here we can
see that the potential V (r12 , x1 ) contains information on the dielectronic
correlation e1 − e2. When the potential V (r12 , x1 ) is placed in the T -
matrix element, it plays the role of a perturbation which causes capture of
electron e1.
As an illustration of the validity of the CDW-4B method for the
TI process, we shall consider the total cross sections for the following
reactions:
He2+ + He(1s2 ) −→ He+ (1s) + He2+ + e, (9)
Li3+ + He(1s2 ) −→ Li2+ (1s) + He2+ + e. (10)
The total cross sections for process (9) at energies 40–1000 keV/amu

are shown in Figures 12.3 and 12.4. In Figure 12.3, a comparison is made
E(keV/amu) for transfer ionization: He2+ + He(1s2 ) −→ He+ (1s) + He2+ + e. The full
curve represents the “post” total cross section of the CDW-4B method21. The dashed
curve represents the corresponding “post” cross sections of the CDW-IEV model.1
Experimental data: ◦, Shah et al.;19 and •, Shah and Gilbody.20
between the CDW-4B method21 and the CDW-IEV model.1 From the
numerical point of view, the CDW-IEV model also encounters seven-
dimensional scattering integrals when dealing with the total cross sections
for a TI process. The main difference between the CDW-IEV and the
CDW-4B method is in the electron correlation treatment. Namely, the
CDW-IEV model includes the static electron correlation (SEC) in the tar-
get by using the bound-state wave function of Pluvinage:28
Z3T −ZT (x1 +x2 ) −ikr12 ′ (11)

ϕi (�x1 , x�2 ) = c(k)
e e 1 F1 (1 − ik , 2, 2ikr12 ),
π
where k′ = 1/(2k) and c(k) is the normalization constant, with k being
a non-linear variational parameter. The corresponding lowest binding
energy Ei,Pluv = −2. 878 for the ground state 1 S of helium is obtained
for k = 0. 41, in which case c(k) = 0. 603366. The wave function (11)
E(keV/amu) for transfer ionization: He2+ + He(1s2 ) −→ He+ (1s) + He2+ + e. The full
and the dashed curves correspond, respectively, to the “post” and “prior” cross sections
of the CDW-4B method.21 Experimental data: ◦, Shah et al.;19 and •, Shah and Gilbody.20
contains two entirely uncorrelated hydrogen-like wave functions with the

unscreened charge ZT multiplied with a corrective r12-dependent term of
the form exp (−ikr12 )1 F1 (1 − ik′ , 2, 2ikr12 ).
In the CDW-IEV model,1 the dynamic electron correlations (DEC) are
completely neglected. In sharp contrast, the CDW-4B method explicitly
includes the DEC through the dielectronic interaction 1/r12 in the transi-
tion T-operator. It should be noted that the SEC can also be fully included
in the CDW-4B method by using the corresponding wave function, but it is
ignored in the present illustrations that provide an unambiguous assessment
of the DEC alone. The relative role of the SEC and the DEC is otherwise
apparent from the two curves associated with the CDW-IEM and CDW-4B
methods, as depicted in Figure 12.3. A comparison of these two theories
with the experimental data in Figure 12.3, clearly shows that the CDW-4B
method represents a substantial improvement over the CDW-IEV model. It

can be concluded that the DEC plays a more important role than the SEC.
Further, we evaluate the “post-prior” discrepancy which arises from
the unequal perturbation potentials used in the T-matrices (4) and (5). In
Figure 12.4, we give the “prior” (Qif− ) and the “post” (Qif+ ) cross sections
of the CDW-4B method.21 A comparison in Figure 12.4 reveals that the
“post-prior” discrepancy is very significant throughout the energy range
40–1000 keV/amu. The “post” cross sections are larger by nearly 50%
than the “prior” results at lower energies, with precisely the opposite pat-
tern at higher energies. Such a considerable difference can be attributed
to the role of the dielectronic repulsion 1/r12. We recall that the difference
between Qif− and Qif+ is due solely to the potentialV (r12 , x1 ) = 1/r12 − 1/x1,
which is present in the “post” and absent from the ‘prior” cross sections.
In the same Figure 12.4, the theoretical cross sections are compared with
the experimental data for reaction (9). It can be seen from Figure 12.4
that, in contrast to the “prior” variant Qif−, the “post” version Qif+ of the
CDW-4B method is in good agreement with the measurements at impact
energies E 80 keV/amu. At lower energies, the results for Qif+ are larger
than the experimental values, as expected, since the CDW-4B method
is a high-energy approximation. The superiority of the “post” over the
“prior” version can be attributed primarily to the electron–electron
interaction 1/r12 in the perturbation of the former variant.
The “post” and the “prior” total cross sections for TI in reaction (10),
derived with the full three-term perturbation are plotted in Figure 12.5,
where the experimental findings20, 32, 33 are also shown. The CDW-4B
method is seen to be in good agreement with the experimental data.
The “post” cross sections lie below the “prior” ones at impact ener-
gies between 100 and 3000 keV/amu, with the reverse behavior above
3000 keV/amu. These computations have been performed for electron
transfer to the ground state alone. Agreement between the CDW-4B
method and the measurements at lower impact energies could be
improved by including a contribution from all the excited states.
The theoretical results from Ref. 34 are also depicted in Figure 12.5.
These cross sections have been obtained using a relativistically covariant field
approach via the second-order Feynman diagrams. As can be observed from
Figure 12.5, the results from34 largely overestimate the experimental data.
As shown by Mančev,22 the “prior” total cross sections for reaction (10)
computed with and without the term VP (R, s2 ) differ from each other
by a considerable amount, which can reach 67% at 50–5000 keV/amu.
E(keV/amu) for reaction: Li3+ + He(1s2) −→ Li2+ (1s) + He2+ + e. The full and the
dashed curves represent, respectively, the “prior” Qif− and “post” Qif+ cross sections of
the CDW-4B method22 with the complete three-term perturbation potentials. The dot-
dashed curve is due to a second-order method34 from the field theory. Experimental
data: Shah and Gilbody;20 •, Woitke et al.;32 ◦ Sant’Anna et al.33
The potential −ZP /s2 from VP (R, s2 ) can cause capture of the electron e1
through the e1 − e2 correlation. This points to a possible role of the DEC
in the process (10). As we have seen above, the CDW-4B method is in very
good accord with the experimental data on TI for the He2+ + He and
the Li3+ + He collisions. Regarding the total cross sections for TI in the
H+ + He collisions at higher impact energies, the CDW-4B method over-
estimates some of the measured data,28–31 whereas at intermediate energies
satisfactory agreement is found with the experimental findings from Ref. 20.
4. THE FOUR-BODY BORN DISTORTED WAVE METHOD

The BDW-4B method has been formulated by Belkic and
Mančev.8 This is a fully quantum-mechanical four-body model which
explicitly takes into account all the particles and their interactions in the
collision, allowing for a systematic study of the correlated dynamics of

the process. Further, the BDW-4B method preserves the correct bound-
ary conditions in both scattering channels according to the principles of
scattering theory.25 Moreover, the BDW-4B method takes full account
of the Coulomb continuum intermediate states of the captured electron
in the exit channel. This makes the BDW-4B method coincide with the
CDW-4B21 method in the exit channel and with the CB1-4B method
in the entrance channel.
The transition amplitude in the BDW-4B method can be written as:8
Tif+ = �χf− |Uf |�+

i �. (12)
The initial wave function +

i with asymptotically correct boundary condi-
tion is given by:
� � (13)
�+ x1 , x�2 )e iki ·�ri +iνi ln(vR−v·
i = ϕi (�
� R)
,
where ri is the relative vector of ZP with respect to the center of mass of
(ZT ; e1 , e2 )i. Function ϕi (x1 , x2 ) represents the two-electron bound-state
wave function of the atomic system (ZT ; e1 , e2 )i, whereas ki is the initial
wave-vector, v is the incident velocity, and νi = ZP (ZT − 2)/v. The ini-
tial state is distorted even at infinity, due to the presence of the asymptotic
Coulomb repulsive potential, Vi∞ = ZP (ZT − 2)/R, between the projec-
tile and the screened target nucleus.
In the BDW-4B theory the distorted potential Uf and distorted wave
χf− are chosen according to:21
� s1 ln ϕf · ∇
Uf = VP (R, s2 ) + V (r12 , x1 ) − ∇ � x1 , (14)
χf− = N − (ζ )N − (νT )φf ϕf ( �s1 )1 F1 (−iζ , 1, −ipx2 − i�p · x�2 )

(15)
� �
� · x�1 )e −ikf ·�rf −iν ln(vR+v·
×1 F1 (−iνT , 1, −ivx1 − i v � R)
,
�
where the function φf is defined by φf = (2π)−3/2 e −ikf ·�rf +i�κ ·�x2, whereas
rf is the position vector of T with respect to the center of mass of the
system (ZP , e1 )f + e2 in the exit channel, kf is the final wave-vector, and
ν = ZP (ZT − 1)/v. The quantities ϕf (�s1 ), N − (ζ ), and N − (νT ) have the
−
same meaning as in CDW-4B model. The wave function χf obeys the
correct boundary conditions.
Using Eqs. (13)–(15), the expression for the transition amplitude Tif+ in
the BDW-4B method becomes:

+ � x1 −i�
Tif = M dR � d�x1 d�x2 e iα� ·�s1 +iβ·� κ ·�x2
L(R)ϕi (�x1 , x�2 )
×1 F1 (iζ , 1, ipx2 + i�p · x�2 ) {[VP (R, s2 ) + V (r12 , x1 )]
×1 F1 (iνT , 1, ivx1 + i v � · x�1 )ϕf∗ ( �s1 ) (16)

−∇ � s1 ϕf∗ ( �s1 ) · ∇
� x1 1 F1 (iνT , 1, ivx1 + i v � · x�1 ) ,
where
∗ ∗
M = (2π)−3/2 N − (νT )N − (ζ ).
The product of logarithmic Coulomb factors is denoted by the auxiliary

function L(R) which can be reduced to the single term:
�
� R)+iν �
� R) � iξ , (17)
L(R) = e iνi ln(vR−v· ln(vR+v·
= (ρv)2iνi (vR + v
� · R)
where ξ = ZP /v. The multiplying term (ρv)2iνi does not contribute to
the total cross section and can be dropped from the transition amplitudes.
Hence, such a formulated BDW-4B method is exactly equal to the
CDW-4B method21 in the exit channel and to the CB1-4B method in
the entrance channel. Therefore, the BDW-4B method satisfies the correct
boundary conditions in both scattering channels.
As an illustration of the validity of the BDW-4B model for the TI pro-
cess, we shall show the total cross sections for the following reaction:
H+ + He −→ H+ + He2+ + e. (18)
The results of computations of total cross sections at impact energies
300–10000 keV are depicted in Figure 12.6. As can be seen from this figure,
the BDW-4B method is found to be in good agreement with the available
experimental data.20, 29, 31 The displayed theoretical results in Figure 12.6
have been obtained by using only the leading term ∇ � ·∇� in Eq. (16). In
such computations the total cross section is reduced to a five-dimensional
numerical quadrature. The number of integration points is gradually and
systematically increased until convergence to two decimal places has been
reached. The computations of the total cross sections with all the terms in
perturbation potential (14) require a nine-dimensional quadrature which
is numerically extensive and shall be pursued in full details in the near
future. In the present work, for an illustration, the Table 12.1 reports on
the results obtained by means of the Gauss–Legendre quadrature with
Figure 12.6 Total cross sections (in cm2) as a function of the laboratory incident energy
for reaction H+ + He −→ H + He2+ + e. The full line represents the results of the
BDW-4B theory.8 Experimental data: •, Schmidt et al.,31 , Shah and Gilbody20 and ◦
Mergel et al.29
Table 12.1 Total cross sections (in cm2) as a function of incident energy E for electron
capture in reaction H+ + He −→ H+ + He2+ + e. The presented theoretical results
are obtained by means of the BDW-4B model using the one-parameter Hylleraas-type
wave function for the initial helium bound state. The symbols Q represent the cross
sections obtained with the complete perturbation potentials according to
Eq. (16), whereas Q∇ refer to the cross sections obtained without terms V (r12 , x1 )
and VP (R, s2 ) in Eq. (16). The results are obtained by the Gauss–Legendre quadrature
with NG L = 8, 16, and 24 points. The numbers in the square brackets denote the
powers of 10
E(keV) NGL Q∇(cm2) Q(cm2)
300 8 7.82[−21] 6.64[−21]
300 16 7.86[−21] 5.94[−21]
300 24 7.87[−21] 5.93[−21]
1000 8 5.92[−23] 5.28[−23]
1000 16 5.88[−23] 6.73[−23]
1000 24 5.89[−23] 5.73[−23]
NGL = 8, 16, and 24 points at 300 and 1000 keV/amu. A comparison is

made between the cross section obtained by using only the leading term
� ·∇
∇ � (as labeled with Q∇ in Table 12.1) and results obtained with all the
terms in perturbation potential (as denoted by Q). The latter results Q
include the dynamic correlations. Only capture into the ground state is
taken into account.
As mentioned, the BDW-4B method describes the four-body nature
of the original problem and satisfies the correct boundary conditions
in the exit and exit channel. Double charge exchange has already been
investigated by means of the BDW-4B method.7, 35–37 This approximation
has also been applied, after the appropriate adaptation, to single electron
capture.38–40 Recently, Monti et al.41, 42 proposed a four-body distorted
wave-eikonal initial state (CDW-EIS-4B) model for single ionization of
dielectronic atomic targets by impact of protons. It would be interest-
ing in the future to adapt their model41, 42 to TI and compare it with the
CDW-4B and BDW-4B theories.
5. CONCLUSIONS
We have investigated the problem of transfer ionization in colli-
sions between bare ions and two-electron atomic systems at high impact
energies. Our objective is twofold. One is to compare the formalisms of
the independent particles/events with the four-body correlated electronic
dynamics. The other is to contrast the continuum distorted wave theory,
or CDW, with its two hybrid approximations known as the continuum
distorted wave eikonal initial state, or CDW-EIS, and the Born distorted
wave, or BDW, models. These three methods coincide with each other only
in one channel regarding both the total scattering wave function and the
associated perturbation potential in the T -matrix element. In the other
channel (entrance or exit depending whether the “prior” or “post” form
of the transition amplitude is employed), both the total scattering wave
functions and the corresponding perturbation interaction potentials are dif-
ferent. The CDW method always uses the full Coulomb wave functions in
the entrance and exit channels for the independent as well as the correlated
motions of the electrons. These electronic dynamics in the continuum are
simplified in the CDW-EIS and BDW models through the replacements
of the full Coulomb waves by their asymptotic forms (eikonal logarithmic
Coulomb phases) valid at infinitely large inter-particle separations. The
� = x� − �s in
two phases differ only in the spatial variable, which is s or R
the CDW-EIS or BDW method, where x and s are the position vectors of
the electron to be captured relative to the projectile and the target nucleus,
respectively. This difference yields two different perturbation potentials in
the transition amplitudes in the CDW-EIS and BDW methods. Thus, any
difference in the performance of these two latter formalisms can be traced
back to the fact that the dynamic variables s and x are coupled through R
in the eikonal Coulomb state from the BDW method, whereas only the
decoupled position vector s appears in the eikonal logarithmic distorting
phase in the CDW-EIS approximation.
First, we compared the IPM and IEV models with the CDW and
CDW-EIS approximations. We made use of the IEV model to simulate the
successive events of ionization-capture and capture-ionization. The single-
particle probabilities as a function of impact parameters were computed
using the three-body CDW and CDW-EIS approximations. It was found
that the total cross sections obtained by means of the CDW-EIS-IEV model
are in slightly better agreement with the experimental data than the results
from the CDW-IEV model. However, the results from the CDW-EIS-IEV
model are still significantly larger than the corresponding experimental val-
ues, especially for the case of proton impact. This points to a failure of the
CDW-EIS method within the independent particle formalism for transfer
ionization. Such a finding adds to a much more pronounced breakdown of
the corresponding correlated version in the CDW-EIS-4B model, by some
formidable 2–3 orders of magnitudes for double charge exchange in fast
collisions between alpha particles and helium.
Second, the four-body theories, such as the CDW-4B and BDW-4B
methods are presently applied to transfer ionization in the H+ + He,
He2+ + He and, Li3+ + He collisions at intermediate and high impact
energies. Agreement of the measurements with the theoretical total cross
sections obtained by means of these two methods is favorable, especially at
sufficiently higher impact energies.
Overall, similarly to the situation of double charge exchange, it can be
concluded that at high energies, the CDW-4B and BDW-4B methods for
transfer ionization are capable of reproducing the available experimental
data. This represents an extrapolation of the well-known success of the
CDW method for three-body collisions. Such a stable performance builds
confidence into this theory when passing from simpler to complicated
collisional problems with more than one active electron. This is in sharp
contrast with the CDW-EIS approximation, which is utterly inadequate
for double charge exchange and also fails for transfer ionization for which,
however, the CDW and BDW methods are successful. The same situation
is likely to be encountered with molecular targets, such as water, which is
the main tissue-like medium. Two or more electrons are expected to play
a larger role in making transitions between ions and water molecules in
the close vicinity of the Bragg peak. This is because at the end of their
path through the traversed medium, the initially energetic ions abruptly
slow down, thus enhancing the probability for single and multiple elec-
tron capture, as well as for simultaneous capture and ionization/excitation.
At the last couple of centimeters before they eventually stop, partially or
fully neutralized ions readily lose one or all their electrons only to quickly
capture them again. Such direct and inverse processes occur literally thou-
sands of times for each ion near the Bragg peak prior to the establishment
of a kind of equilibrium. These higher order collisional phenomena are
anticipated to make noticeable contributions to the overall stopping power,
calling for the corresponding improvements of the existing data bases for
energy losses of ions in the traversed matter. Thus far, these data bases have
exclusively been restricted to single electron transitions consisting mainly
of the ionization or excitation scattering channels, and ignoring simultane-
ous transitions of two or more electrons.
ACKNOWLEDGMENTS
Dž. Belkic appreciates research funds from the Karolinska Institute, the
Karolinska University Hospital’s Radiumhemmet and the Swedish Cancer
Society (Cancerfonden). I. Mančev and N. Milojevic thank the Ministry of
Science of the Republic of Serbia (Project No. 171020) for support.
REFERENCES
1. Dunseath, K. M.; Crothers, D. S. F. Transfer and ionization processes during the collision
of fast H+, He2+ nuclei with helium. J. Phys. B: At. Mol. Opt. Phys. 1991, 24, 5003.
2. Galassi, M. E.; Abufager, P. N.; Martinez, A. E.; Rivarola, R. D.; Fainstein, P. D. The
continuum distorted wave eikonal initial state model for transfer ionization in
H+, He2+ + He collisions. J. Phys. B: At. Mol. Opt. Phys. 2002, 35, 1727.
3. Gayet, R.; Salin, A. Simultaneous capture and ionisation by fast ion impact on helium.
J. Phys. B: At. Mol. Phys. 1987, 20, L571.
4. Sidorovich, V. A.; Nikolaev, V. S.; McGuire, J. H. Calculation of charge-changing cross
sections in collisions of H+, He2+, and Li3+ with He atoms. Phys. Rev. A 1985, 31, 2193.
5. Shingal, R.; Lin, C. D. Calculations of two-electron transition cross sections between
fully stripped ions and helium atoms. J. Phys. B: At. Mol. Opt. Phys. 1991, 24, 251.
7. Belkić, Dž. Quantum Theory of High-Energy Ion-Atom Collisions; Taylor & Francis:
London, 2009.
8. Belkić, Dž.; Mančev, I. Transfer ionization in fast ion-atom collisions: four-body Born
distorted-wave theory. Phys. Rev. A 2011, 83, 012703.
10. Crothers, D. S. F.; McCann, J. F. Ionisation of atoms by ion impact. J. Phys. B: At. Mol.
Phys. 1983, 16, 3229.
11. Fainstein, P. D.; Ponce, V. H.; Rivarola, R. D. A theoretical model for ionisation in
ion-atom collisions. Application for the impact of multicharged projectiles on helium.
13. Busnengo, H. F.; Martínez, A. E.; Rivarola, R. D.; Dubé, J. L. Distorted-wave models for
electron capture in asymmetric collisions. II. Differential cross sections and the double
scattering region. J. Phys. B: At. Mol. Opt. Phys. 1995, 28, 3283.
14. Martínez, A. E.; Deco, G. R.; Rivarola, R. D.; Fainstein, P. D. K-Shell vacancy production
in asymmetric collisions. Nucl. Instrum. Methods Phys. Res., Sect. B 1988, 34, 32.
15. Clementi, E. C.; Roetti, C. Roothaan–Hartree–Fock atomic wavefunctions. At. Data
Nucl. Data Tables 1974, 14, 177.
16. Bernardi, G.; Focke, P.; González, A. D.; Suarez, S.; Fregenal, D. Electron emission for
transfer ionization in 70 keV H+ on He. J. Phys. B: At. Mol. Opt. Phys. 1999, 32, L451.
17. Bates, D. R.; Griffing, G. W. Inelastic collisions between heavy particles I: excitation and
ionization of hydrogen atoms in fast encounters with protons and with other hydrogen
atoms. Proc. R. Soc. A 1953, 66, 961.
18. Bates, D. R.; Griffing, G. W. Inelastic collisions between heavy particles IV: contribution
of double transitions to certain cross sections including that associated with the ioniza-
tion of hydrogen atoms in fast encounters with other hydrogen atoms. Proc. R. Soc. A
1955, 68, 90.
19. Shah, M. B.; McCallion, P.; Gilbody, H. B. Electron capture and ionisation in collisions
of slow H+ and He2+ ions with helium. J. Phys. B: At. Mol. Phys. 1989, 22, 3037.
20. Shah, M. B.; Gilbody, H. B. Single and double ionisation of helium by H+, He2+ and
Li3+ ions. J. Phys. B: At. Mol. Phys. 1985, 18, 899.
21. Belkić, Dž.; Mančev, I.; Mergel, V. Four-body model for transfer ionization in fast ion-
atom collisions. Phys. Rev. A 1997, 55, 378.
22. Mančev, I. Single-electron capture and transfer ionization in collisions of Li3+ ions with
helium. Phys. Rev. A 2001, 64, 012708.
23. Belkić, Dž. Symmetric double charge exchange in fast collisions of bare nuclei with
heliumlike atomic systems. Phys. Rev. A 1993, 47, 189.
24. Nordsieck, A. Reduction of an integral in the theory of Bremsstrahlung. Phys. Rev. 1954,
93, 785.
25. Belkić, Dž.; Principles of quantum scattering theory, Institute of Physics, Bristol, 2004.
26. Belkić, Dž.; Gayet, R.; Salin, A. Electron capture in high-energy ion-atom collisions.
Phys. Rep. 1979, 56, 279.
27. Dollard, J. D. Asymptotic convergence and the Coulomb interaction. J. Math. Phys. 1964,
5, 729.
28. Pluvinage, P. Fonction d’onde approchée a un parametre pour l’état fondamental de
l’hélium. Ann. Phys. (NY) 1950, 5, 145.
29. Mergel, V.; Dörner, R.; Achler, M.; Khayyat, Kh.; Lencinas, S.; Euler, J.; Jagutzki, O.;
Nüttgens, S.; Unverzagt, M.; Spielberger, L.; Wu, W.; Ali, R.; Ullrich, J.; Cederquist, H.;
Salin, A.;Wood, C.; Olson, R.; Belkić, Dž.; Cocke, C.; Schmidt-Böcking, H. Intra-atomic
electron-electron scattering in p-He collisions (Thomas process) investigated by Cold

Target Recoil Ion Momentum Spectroscopy. Phys. Rev. Lett. 1997, 79, 387.
30. Schmidt, H.T.; Fardi, A.; Schuch, R.; Schwartz, S.; Zettergren, H.; Cederquist, H.; Bagge,
L.; Danared, H.; Källberg, A.; Jensen, J.; Rensfelt, K.; Mergel, V.; Schmidt, L.; Schmidt-
Böcking, H. Double-to-single target ionization ratio for electron capture in fast p-He
collisions. Phys. Rev. Lett. 2002, 89, 163201.
31. Schmidt, H. T.; Jensen, J.; Reinhed, P.; Schuch, R.; Støchkel, K.; Zettergren, H.;

Cederquist, H.; Bagge, L.; Danared, H.; Källberg, A.; Schmidt-Böcking, H.; Cocke, L.
Recoil-ion momentum distributions for transfer ionization in fast proton-He collisions.
Phys. Rev. A 2005, 72, 012713.
32. Woitke, O.; Závodszky, P. A.; Ferguson, S. M.; Houck, J. H.; Tanis, J. A. Target ionization
and projectile charge changing in 0.5 – 8 MeV/q Liq++He(q=1, 2, 3) collisions. Phys.
Rev. A 1998, 57, 2692.
33. Sant’Anna, M. M.; Santos, A. C.; Coelho, L. F.; Jalbert, G.; de Castro Faria, N.V.; Zappa,
F.; Focke, P.; Belkić, Dž. Double-electron-capture cross sections for Li3+ collisions with
He at intermediate-to-high velocities. Phys. Rev. A 2009, 80, 042707.
34. Bhattacharyya, S.; Rinn, K.; Salzborn, E.; Chatterjee, L. High-energy electron capture
by fully stripped ions from He atoms—a QED approach. J. Phys. B: At. Mol. Opt. Phys.
1988, 21, 111.
35. Belkić, Dž. Double charge exchange at high impact energies. Nucl. Instrum. Methods
Phys. Res., Sect. B 1994, 86, 62.
36. Belkić, Dž.; Mančev, I.; Mudrinić, M. Two-electron capture from helium by fast α par-
ticles. Phys. Rev. A 1994, 49, 3646.
37. Belkić, Dž. Review of theories on double electron capture in fast ion-atom collisions.
J. Math. Chem. 2010, 47, 1420.
38. Mančev, I. Single charge exchange in fast collisions of alpha particles with helium.
39. Mančev, I. Single-electron capture from helium-like atomic systems by bare projectiles.
Euro. Phys. Lett. 2005, 69, 200.
40. Mančev, I. Continuum distorted wave—Born initial state (CDW-BIS) model for single
charge exchange. J. Comput. Methods Sci. Eng. 2005, 5, 73.
41. Monti, M.; Fojón, O. A.; Hanssen, J.; Rivarola, R. D. Ionization of helium targets by pro-
ton impact: a four-body distorted wave-eikonal initial state model and electron dynamic
correlation. J. Phys. B: At. Mol. Opt. Phys. 2009, 42, 195201.
42. Monti, M.; Fojón, O. A.; Hanssen, J.; Rivarola, R. D. A four-body fully distorted wave—
eikonal initial state model for ionization of He targets by ion impact. J. Phys.: Conf. Ser.
2009, 163, 012068.
CHAPTER THIRTEEN
Distorted Wave Theories for

One- and Two-Electron Capture
in Fast Atomic Collisions
Volodymyr Yu. Lazur and Mykhaylo V. Khoma
Contents
1. Introduction 364
2. Basic kinematics and dynamics 367
3. The first Born method with correct boundary conditions 374
4. The continuum-intermediate state method with the correct boundary conditions 381
5. Angular and energy dependencies of charge-transfer cross sections 387
6. The Dodd–Greider integral equation in the theory of two-electron processes 390
7. Conclusion 401
References 402
Abstract
On the basis of the Dodd–Greider integral equations with Coulomb interactions, we
have reviewed the theory of one-electron capture in collisions of hydrogen-like atoms
and positively charged ions with account taken for electron re-scattering effects. This
theory can be applied to study the influence of Coulomb interactions in the final state
on angular and energy distributions of the products of a charge-transfer reactions at
intermediate collision energies.
Using the four-body Dodd–Greider integral equations, we have reviewed a four-
particle formalism of the continuum distorted wave (CDW) method, applicable
for describing a two-electron capture reaction in energetic ion-atomic collisions.
An attractive feature of the CDW method is consistent and accurate consideration
of asymptotic wave functions, which take into account the long-range nature of
Coulomb interactions in both the entrance and exit reaction channels. In the past, this
method been applied to double electron transfer in energetic H++ He and He2+ + He
collisions, yielding good agreement with measurements.

364 Volodymyr Yu. Lazur and Mykhaylo V. Khoma
1. INTRODUCTION
The one-electron charge-transfer process:
AZα + + B → A(Zα −1)+ + B + (1)

for swift ion-atomic collisions has attracted steady attention both from
theoretical and experimental groups for many decades.1 The necessary
information about process (1) is not limited to the integral characteristics
(total cross section, reaction constant), but also includes details about the
energy and angular distributions of the reaction products. This places high
demands both for completeness and precision of the theoretical calculation.
In the context of this work, swift collisions signify that the relative velocity
of the colliding particles is greater than the effective orbital velocity υ0 of
the captured electron.
The theoretical analysis of experimental data on the processes of type
1/2
(1) may be summarized as follows. At collision velocities υ ∼ υ0 Zα the
cross sections of charge exchange and ionization are of the same order of
1/2
magnitude. In the range υ > υ0 Zα , the probability of charge transfer
decreases, such that ionization and excitation of atoms by energetic pro-
jectiles become the dominant processes. However, the latter is not always
the case. A high probability of target excitation (or ionization) initiates new
mechanisms of electron transfer that are the main subject of the present
study. It is commonly accepted that in the range of intermediate and high
collision velocities, the dominant processes are a one-step mechanism of
electron transfer (1) and a two-step Thomas mechanism of electron capture
through the intermediate discrete or continuum state. The latter becomes
especially important with increasing collision energy.
Charge exchange in low- and intermediate-energy ion-atomic colli-
sions is commonly described by close coupling methods (see for instance
Ref. 2, and references therein), whereas energetic collisions are usually
treated by analytical perturbative approaches. The latter include differ-
ent variations of the distorted wave method3 which are used for both
symmetrical and asymmetrical collisions. The main drawback of the
mentioned approaches is the assumptions about the short-range charac-
ter of the considered interaction potential. It was pointed out, however
(see Refs. 4–6) that such assumptions are not justified for applications
to atomic collisions. As an illustration, we refer to Ref. 4 where it was
demonstrated that the Strong Potential Born (SPB) approximation7 can-
not serve as a basis for constructing adequate charge-exchange theories,
Distorted Wave Theories for Charge Exchange 365
since the scattering amplitude includes a contribution from the diver-

gent terms. The reason for this drawback is the inaccurate account
taken for the Coulomb effects in the initial and final states.4–6, 8–10 This
circumstance confirms the general result of Merkuriev11 as well as of
Merkuriev and Faddeev,12 who demonstrated how theories elaborated
for short-range potentials could be adapted to long-range potentials.
In approximations of a single collision, the mentioned theoretical
approaches reproduce the standard formulae of the Oppenheimer Brinkman
Kramer (OBK) approximation.3, 13 However, one should keep in mind that
the OBK-approximation is not, strictly speaking, a first order perturbation
theory.The numerical calculations3 show that, compared to the experiment,
the OBK-approximation systematically overestimates the cross sections
and gives the wrong position of the maximum. The first order perturba-
tion theory (the first Born approximation) uses perturbation as the “full”
interaction (i.e., the interaction of the electron with the projectile and the
internuclear repulsion), which gives rise to the well known result of Jackson
and Schiff (JS).14 Generally speaking, the validity of the Born approxima-
tion (and perturbation theory in general) for describing charge-exchange
reactions is not at all obvious.3 In the case of excitation or ionization, it
is possible, physically and mathematically, to determine the conditions for
applicability of the perturbation theory. However, such conditions are not as
clear in the case of charge-exchange reactions. In Refs. 15, 16 it was shown
that at high collision velocities the second Born cross section for electron
capture from a hydrogen-like atom with charge Zβ by a bare ion of charge
Zα increases rapidly as Zβ increases. These results indicate that second- and
higher-order Born terms must be considered in calculating the cross section
for electron capture. The question of the asymptotic behavior of charge-
exchange amplitudes has also been investigated in Ref. 17 on the basis of
Faddeev equations.18
Apart from the general problem of the Born approximation conver-
gence, there is also the long-standing difficulty regarding the choice of
the perturbing potential. Due to the non-orthogonality of the initial and
final states, different forms of the perturbing potential yield different results
for the cross sections. The formal derivation (based on general scattering
theory) of the first Born scattering amplitude for charge exchange gives
rise to interaction potentials that include the internuclear potential. The
simplest of these in the one-electron model reads:
JS Zα Zα Zβ
Uα = − + ,
s R (2)
where s is the radius-vector of the electron relative to the center of ion

AZα + and R is the internuclear separation. The first term in (2) describes
the interaction of the active electron with ion AZα +. The second term
describes the internuclear interaction in the entrance channel. Due to
the orthogonality of the initial and final wave functions, the second term
of (2) makes no contribution to the amplitude of the direct ionization.
However, this term gives a non-zero contribution to the amplitude of the
charge-exchange reaction. Potential (2) has been used in Refs. 14, 19 for
calculating the Born cross sections of charge exchange of hydrogen atom
by proton. This reaction does not contain the Coulomb interaction in the
entrance and exit channels at large interatomic separations, and hence good
agreement with experiment has been obtained. However, application of
potential (2) for calculating the cross sections of electron capture by nucleus
H+ and He2+ from the inner shell of multi-electron atoms (carbon, neon,
and argon) gives rise to non-physical results that differ from the experiment
by several orders of magnitude.20, 21
By investigating the non-orthogonality problem, Bates22 has shown
that orthogonalization of the wave functions of the initial and final states
leads to a modification of the perturbing potential. More precisely, this
orthogonalization makes the contribution from the internuclear potential
in (2) vanish. Instead, a new term appears which is similar to the inter-
nuclear interaction.23 At large interatomic separations, this new term has
the asymptotic form Zα /R. The resulting perturbing potential which now
approximately accounts for orthogonalization of the wave functions reads:
Zα Zα
Uα = − + . (3)
s R
The authors Omidvar et al.24 and Halpern25 explained the usage of poten-
tial (3) by full electron shielding of the nuclei of atom B. Despite the criti-
cism of this argumentation23, 26 (especially for electron capture from inner
shells), the results of cross section calculations within the Born approxima-
tion with potential (3) are in good agreement with experiment, even for
electron capture from K-shell of the multi-electron atom.24
In this connection, it is important to be accurate and systematic in
choosing the perturbing potential. At the very least, one should realize
that the results of the obtained cross sections will differ according to the
selection of Uα. One such approach is considered in the present work and
consists of using few-body problem integral equations that are properly
modified for Coulomb interaction.
On the basis of these introductory considerations, it is clear that the cor-

rect account of the Coulomb interaction in the charge-exchange reaction
channels has not had a compelling solution based upon Refs. 20, 21, 23–25.
On the other hand, the existing methods from few-body physics allow us to
take account of Coulomb effects and to test these effects in the framework
of the widely used continuum distorted wave (CDW) approximation.3
In the present work on the basis of modified Coulomb interaction three-
body Dodd–Greider integral equations,27, 28 the method of one-electron
capture cross section calculation is presented. The amplitude of reaction (1)
is obtained as the first term of the iteration expansion of the solution of the
corresponding Dodd–Greider equations for the three-body scattering opera-
tor. This is the boundary-corrected continuum intermediate state (BCIS)5
method. It is shown that without accounting for the effects of the Coulomb
re-scattering of the captured electron, it is not possible to reproduce the
Thomas peak in the angular distribution of the reaction products. In the
approximation of the “single-step” collision, the BCIS method reduces to
the version of the boundary-corrected first Born (CB1) method,8–10 where
the asymptotic behavior of the particles in the entrance and exit channels of
the reactions is described by the two-body Coulomb wave functions. The
short-range interaction due to incomplete screening of the nucleus by atomic
electrons is taken into account in the perturbing potential.
2. BASIC KINEMATICS AND DYNAMICS

The complex problem of an ion-atomic collision reaction (1) con-
sidered here is an idealized problem of nonrelativistic interaction of three
spinless particles α (projectile A+Zα), γ (active electron e −), and β (target
ion B +) with masses mα , mγ , mβ, respectively. The center of mass motion
is assumed to be separated. Corresponding to possible separation of the
three-body problem on fragments (β, γ ) + α; (α, γ ) + β; (α + β) + γ , we
introduce, together with the full Hamiltonian H = H0 + V , the arrange-
ment channel Hamiltonians Hj = H0 + Vj ( j = α, β, γ ), where H0 is the
three-body kinetic energy
operator in the center of the mass system of
coordinates and V = α,β,γ Vj is the full interaction. The subscript index j
in Vj defines the missing particle in this interaction (i.e., the Vα is the opera-
tor of β and γ interaction). Finally, υj = V − Vj is the arrangement channel
interaction. Let us introduce the Jacobi-type coordinates:
�s = (a/mγ )�x − �rα , x� = (b/mγ )�s − �rβ , � = x� − �s,

R (4)
where a = mγ mβ /(mγ + mβ ) and b = mγ mα /(mγ + mα ) are the reduced

masses. Here x is the radius-vector of the electron relative to the center of
the residual target ion B+ in reaction (1). In terms of these coordinates, the
operator H0 can be represented in two equivalent forms:
1 1 � 1 1 �
H0 = − ��rα − ∇ x� = − ��rβ − ∇�s , (5)
2µα 2a 2µβ 2β
where the standard definition for the Laplace operator is used, and:
mα (mβ + mγ ) mβ (mα + mγ )
µα = , µβ = . (6)
mα + mβ + mγ mα + mβ + mγ
We shall split the potentials υj, j = 1, 2 into two parts:

υj = V − Vj = Uj + Wj , (7)
where Wj defines the so-called “disturbing” potential which represents the
weak long-range Coulomb forces partly responsible for the asymptotic
behavior of the wave functions and Uj is the pure short-range perturbation
responsible for electronic transitions.
From the definition of the Hamiltonian Hα, (Hβ ) immediately follows
its eigenfunctions |�αi � (|�βf �):
β
|�αi � = |ϕi (�x)� exp(i k�α · �rα ), |�f � = |ϕf (�s)� exp(i k�β · �rβ ), (8)
where ϕi (ϕf ) is the wave function of the bound state of the pair
(β, γ ) ((α, γ )) and exp(i k�α · �rα ) (exp(i k�β · �rβ )) is the plane wave, describ-
ing the relative motion of the free particles α (β) and (β, γ ) ((α, γ )) in
the initial (final) state with momentum kα (kβ ). Rigorously speaking, in
the case of charged particles, the wave functions in (8) should be distorted
by a phase factor which logarithmically depends on the distance between
the particles.9 This is because the asymptotic motion of the particles in the
Coulomb potential is not free, due to its long-range character.
Hence, following Ref. 3 we introduce modified arrangement chan-
β− β
nel asymptotic states |�α+ i � (|�f �) which, compared to |�i � (|�f �)
α
correctly describe the effects of the long-range Coulomb fields on the

course of the collision process. Let ξα = rα − k�ˆ α · �rα (ξβ = rβ − k�ˆ β · �rβ )
be the parabolic coordinates of the α (β) particles before (after) the col-
lision, then k�ˆ j (j = α, β) are the identity vectors: k�ˆ j = k�j /kj. The functions
β−
�α+i (�f ) are represented by the product of the bound state wave
function of the pair (β, γ ) ((α, γ )) and distorted plane wave fα+ (fβ− )
with the unity amplitude:
�α+
i = ϕi (�x)fα+ (�rα ) ≡ ϕi (�x) exp(i k�α · �rα + iσα ), (9)
β−
�f = ϕf (�s)fβ− (�rβ ) ≡ ϕf (�s) exp(i k�β · �rβ − iσβ ), (10)
with Coulomb phases σα and σβ given by:
σα = να ln(kα σα ), να = nα /υ, υ� = k�α /µα ,

(11)
σβ = νβ ln(kβ σβ ), νβ = nβ /υ ′ , υ� ′ = k�β /µβ ,
where nα = Zα (Zβ + Zγ ), nβ = Zβ (Zα + Zγ ), and Zj is the charge of the

jth particle (j = α, β, γ ); Zγ = −1.
We shall split the perturbing potential Wα and Wβ into two parts:
Wα = wα + Wαd , Wβ = wβ + Wβd , (12)

where wα and wβ are the arbitrary short-range potentials that depend
only on the relative coordinates rα and rβ, respectively. The distorting
potentials Wαd and Wβd in the asymptotic limits have a pure Coulomb
behavior:
nα nβ
Wαd −→ , Wβd −→ . (13)
rα →∞ rα rβ →∞ rβ
We shall introduce the arrangement channel Hamiltonian Hαd (Hβd ):

Hαd = Hα + Wαd , Hβd = Hβ + Wβd , (14)
and construct potentials Wαd and Wβd in such a way that the Schrödinger
equations:3
(Hαd − E)�α+
i = 0, E = Ei + kα2 /2µα , (15)
β−
(Hβd − E)�f = 0, E = Ef + kβ2 /2µβ , (16)
shall be satisfied. Here Ei (Ef ) is the energy of the bound state of the pair
(β, γ ) ((α, γ )) and E is the total energy of the three-body system.
The use of the arrangement channel Hamiltonian Hαd (Hβd ) in the

form of Eq. (14) has a clear physical motivation. Since electrons are influ-
enced by the Coulomb field in any point of configurational space, the
perturbations Wαd and Wβd (which approximate the potential of the distant
Coulomb center) shall be included in the arrangement channel (i.e., zero-
order) Hamiltonian.
Let us obtain Wαd in an explicit form. From definition (14),
Wαd = Hαd − Hα. Acting with operator (Hαd − Hα ) on the state |�α+ i �
and taking into account relations (5), (9), and (15), one obtains:
Wαd |�α+ x)(kα2 /2µα + ��rα /2µα )|fα+ (�rα )�

i � = ϕi (�
(17)
= (nα /rα )[1 − (να /(kα ξα ))]|�α+
i �.
Using a similar technique, one obtains:

β+ β−
Wβd |�f � = (nβ /rβ )[1 − (νβ /(kβ ξβ ))]|�f �. (18)
Let us note that representations (9) and (10) are not valid along the so-
called forward scattering direction (�rˆα = k�ˆ α , ξα = 0) ((�rˆβ = k�ˆ β , ξα = 0)),
where the phases σα (σβ ) become infinite. On this special direction, the
asymptotic of the functions �α+i and �β− f can be described in terms of
special functions.12
Consider the full Green’s function (resolvent) for the system of three
particles:
G ± (E) = (E − H ± iε)−1 , (19)

+ −
where ε is an infinitely small positive number. Let us define Gαd
as and Gβd
the Green’s functions of the model arrangement channel Hamiltonian Hαd
and Hβd, respectively:
+ −
Gαd = (E − Hαd + iε)−1 , Gβd = (E − Hβd − iε)−1 . (20)
Let us introduce the Møller wave operators ωα+ (ωβ− ):
+
ωα+ = 1 + gαd wα ≡ 1 + (E − Hαd − wα + iε)−1 wα , (21)
ωβ− = 1 + gβd
−
wβ ≡ 1 + (E − Hβd − wβ − iε)−1 wβ , (22)
β−
that transform the eigenfunction �α+ i (�f ) into the distorted wave
β−
χiα+ (χf ) in the entrance (exit) channel of reaction (1):
|χiα+ � = ωα+ |�α+

i �, (23)
β− β−
|χf � = ωβ− |�f �. (24)
± as in Refs. 27, 28:

We shall define now operators Uαβ
+ †
Uαβ = ωβ− (υβ − Wβ )[1 + G + (υα − Wα )]ωα+ , (25)
− †
Uαβ = ωβ− [1 + (υβ − Wβ )G + ](υα − Wα )ωα+ , (26)
where the symbol † denotes Hermitian conjugation. The matrix elements

± between the Coulomb asymptotic states |�α+ � and
of the operators Uαβ i
β− ±
|�f � are the transition amplitudes Tαβ from channel α to channel β cor-
respondingly in the post- and prior-formalisms:
± ± β−
Tαβ = ��f |Uαβ |�α+
i �. (27)
The integral equations for transition operators Uαβ± , obtained by Greider
and Dodd,27 read (for illustration consider the operator Uαβ

− only):
− †
Uαβ = ωβ− (υα − Wα )ωα+ + ωβ− (υβ − Wβ )Gβd
− −
Uαβ . (28)
In the prior formalism of the considered theory, potential Wβ is arbitrary

and potential Wα should not lead to rearrangement in the channel β, i.e.,
β−
lim iε��f |ωα+ |�α+
i � = 0.
ε→0
The first term on the right hand of Eq. (28) leads to the following transi-
tion amplitude in the Born approximation with distorted waves:
− β− † β−
Tαβ (DWB) = ��f |ωβ− (υα − Wα )ωα+ |�α+
i � ≡ �χf |υα − Wα |χiα+ �. (29)
Although Eq. (28) is formally exact, its solution cannot be obtained within
the standard integral equations methods. The reasoning is the follow-
ing. The kernel of integral equation (28) contains disconnected diagrams
responsible for processes where one particle does not interact with the
other two. As a consequence, integral equation (28) contains singulari-
ties due to the presence of delta-functions representing the momentum
conservation for the particles that do not interact with the selected pair of
other particles. These arguments (for detailed discussion see Ref. 27) raise
questions about the convergence of the Born series within the distorted
waves method, i.e., the iteration expansion of Eq. (28). One should note
that the arguments used in Ref. 27, although
+
applicable in the case of the
U
iteration series for transition operator αβ in post-formalism, lead to a con-
clusion about its divergence. Essentially, these findings generalize the results
of Ref. 29, concerning the divergence of the Born iteration series within
the plane wave representation of the three-body rearrangement scattering
problem.
All these circumstances dictate the need for some restructuring (or
rearrangement) of Eq. (28) similar to that carried out for the multiple scat-
tering theory30 and Faddeev equations.18 Such restructuring must be done
with the condition that the kernel of the integral equation (28) contains
no disconnected diagrams. The details of such a cumbersome restructur-
ing can be found in Ref. 28 and are omitted here. We shall present only
the final result. With this aim, we shall introduce the auxiliary potential υx,
that corresponds to the virtual intermediate channel “x” and the Green’s
operator gx+ = (E − H + υx + iε)−1 corresponding to this potential. In
these designations the modified (with account for the long-range nature
of Coulomb interaction) Dodd–Greider equation for quantum mechanical
−
operator Uαβ of the three-particle rearrangement scattering has the final
form:
− −
Uαβ = I + KUαβ , (30)
where
†
I = ωβ− [1 + (υβ − Wβ )gx+ ](υα − Wα )ωα+ ,
†
(31)
K = ωβ− (υβ − Wβ )gx+ vx Gβd
+
.
The main advantage of (30) compared to (28) is that different choices of υx
and Wβ can be made, all satisfying the prescribed condition of eliminating
disconnected diagrams. Thus, in the case when the mass of one particle is
much less (or much greater) than the mass of the other two, the auxiliary
potentials υx and Wβ can be chosen in such a way that kernel K of the
rearranged integral equation (30) is continuous in a suitable Banach space,
i.e., it does not include disconnected diagrams and all the Green’s functions
in this equation are the solutions of equations with separable variables, and
hence, can be evaluated in the explicit form. Inhomogeneous term I in
Eq. (30) differs from the inhomogeneous term of Eq. (28) and represents a
1st order approximation for convergent iteration series of the solution of
−
Eq. (30). Now, using (30), the amplitude Tαβ from Eq. (27) reads:
− β− − β− β−
Tαβ = ��f |I |�α+ α+ α+
i � + ��f |KUαβ |�i � = ��f |I |�i � + RS, (32)
where RS designate all the terms that take into account multiple re-scatter-
ing, where the rest ��β− − α+
f |KUαβ |�i �, according to (31), contains the terms
with at least three sequential re-scatterings. If we assume that the processes
with multiple re-scatterings do not significantly affect the shape of angular
distribution, the second term of (32) can be omitted. Hence, the amplitude
of reaction (1) in the prior-formalism reads:
†
−
Tαβ −
= Tαβ
β−
(DWB) + ��f |ωβ− [ gx+ (υβ − Wβ )](υα − Wα )ωα+ |�α+
i �. (33)
Comparison of (29) and (33) demonstrates that the first term Tαβ −
(DWB)
of (33) allocates the amplitude of the direct one-step mechanism for charge
transfer within the distorted wave Born approximation method. The sec-
ond term of (33) describes the two-step mechanism of electron capture
through the intermediate discrete or continuous state. A similar result holds
+
for the transition amplitude Tαβ in the post-formalism:
+ β− †
Tαβ = ��f |ωβ− (υβ − Wβ )[1 + gx+ (υα − Wα )]ωα+ |�α+
i �. (34)
To conclude this section we should discuss once more the fundamental

property of Eq. (30). From a formal point of view, it is as complicated for
a direct solution as the Faddeev-type equations.18 However, there is no
need for an exact solution of Eq. (30). The essence of the method consid-
ered here consists of the following. It is enough to make only the iterative
approximation for the operator that describes the rearrangement of the
system.
Transformation of Eq. (28) into Eq. (30) in the framework of distorted
wave method allows one to obtain the iteration series (the so-called
Coulomb boundary-corrected Born series) for the transition operator
which, as shown by many calculations (see Refs. 3, 31, 32) rapidly converges.
3. THE FIRST BORN METHOD WITH CORRECT

BOUNDARY CONDITIONS
Let us calculate the contribution of the direct one-step mechanism
to the cross section of reaction (1) in the framework of the first Born
approximation with distorted waves, when the amplitude of the process is
given by:
− β− † β−
Tαβ = ��f |ωβ− (υα − Wα )ωα+ |�α+
i � = �χf |Uα |χiα+ �, (35)
+ β− † β−
Tαβ = ��f |ωβ− (υβ − Wβ )ωα+ |�α+
i � = �χf |Uβ |χiα+ �. (36)
By acting with operators (E − Hα − Wα + iε) and (E − Hβ − Wβ + iε)

on both parts of Eqs. (23) and (24), respectively, and taking the limit ε → 0
with account of Eqs. (15) and (16), we come to the following equations:
(E − H0 − Vα − Wα )|χiα+ � = 0, (37)
β+
(E − H0 − Vβ − Wβ )|χf � = 0. (38)
The boundary conditions have the forms:

β− β−
χiα+ −→ �α+
i , χf −→ �f , (39)
rα →∞ rα →∞
where the Coulomb asymptotic waves are defined by relations (9) and (10).
Let us define the explicit form of the potentials of arrangement channel
interactions υα and υβ:
Zα Zα Z β Zβ Zα Z β
υα = Vβ + Vγ = − + , υ β = V α + Vγ = − + . (40)
s R x R
Here for R → ∞, it follows: υα ≈ Zα (Zβ − 1)/R≡Wαd c and

c
υβ ≈ Zβ (Zα − 1)/R≡Wβd. The specific form of the short-range poten-
tials wα and wβ will be chosen in the symmetric form:
c c
wα = Wαd − Wαd , wβ = Wβd − Wβd , (41)
which ensures the identical interpretation of the distortion in the

entrance and exit channels. Using the above formulae for wα and wβ, we
rewrite partitioning (7) for channel interactions υα, υβ in the following

form:
Zα (Zβ − 1) Zα Zα
Wα = , Uα = υα − Wα = − , (42)
R R s
Zβ (Zα − 1) Zβ Zβ
Wβ = , U β = v β − Wβ = − . (43)
R R x
With these partitioning Eqs. (37) and (38) read:

��rα �x� Zβ nα
E+ + + − χiα+ = 0, (44)
2µα 2a x R

��rβ ��s Zα nβ β−
E+ + + − χf = 0. (45)
2µβ 2b s R
The main technical problems with solving Eqs. (44) and (45) occur when
distortion potential Wα (Wβ ) and Laplace operator �rα (�rβ ) depend on
different relative variables. The connection between those variables follows
from Eq. (4) as:
� − (a/mβ )�x,
�rα = R � + (b/mα )�s.
�rβ = R (46)
Since the electronic mass is much smaller than the mass of the particles
α and β, one can simplify Eqs. (44) and (45). Neglecting the terms pro-
�.
portional to a/mβ and b/mα, one obtains for the coordinates: �rα ≃ �rβ ≃ R
Hence, the variables in Eqs. (44) and (45) can be separated and the wave
α+ β−
functions χi , χf can be represented in the factorized forms:
β−
χiα+ = ϕi (�x)Fα+ (�rα ), χf = ϕf (�s)Fβ− (�rβ ), (47)
where Fα+ (Fβ− ) is the continuum Coulomb wave function which describes
the motion of the particle α (β) relative to the center of mass of the pair
(β, γ ) ((α, γ )) in the initial (final) state with the relative kinetic energy
kα2 /2µα (kβ2 /2µβ ):
Fα+ (�rα ) = N (+) (να ) exp(i k�α · �rα )1 F1 (−iνα , 1, ikα ξα ), (48)
Fβ− (�rβ ) = N (−) (νβ ) exp(i k�β · �rβ )1 F1 (iνβ , 1, −ikβ ξβ ), (49)
N (±) (νj ) = Ŵ(1 ± iνj ) exp(−π νj /2), j = α, β. (50)
Here 1 F1 (a, b, x) is the confluent Kummer hypergeometric function. By

α+ β−
substitution of the explicit expressions for χi , χf and Uα,β into (35)
and (36), one obtains:

− ∗ Zα Zα � �
Tαβ = d�rα d�x ϕf (�s) − ϕi (�x)e ikα ·�rα −ikβ ·�rβ L(�rα , �rβ ), (51)
R s

+ Zβ Zβ � �
Tαβ = d�rβ d�s ϕf∗ (�s) − ϕi (�x)e ikα ·�rα −ikβ ·�rβ L(�rα , �rβ ), (52)
R x
where
L(�rα , �rβ ) = N (+) N (−) F(−iνα , 1, ikα ξα )1 F1 (iνβ , 1, −ikβ ξβ ). (53)
Taking into account that mγ ≪ mα ∼ mβ, the following kinematic relations

are straightforward:
k�α · �rα − k�β · �rβ = −av � = β� · x� + α� · �s,

� · x� − α� · R (54)
where we use the following definitions:
α� = (b/mγ )k�β − k�α = αz υ�ˆ + η�, αz = −mγ υ/2 − �E/υ, (55)
β� = (a/mγ )k�α − k�β = βz υ�ˆ − η�, βz = −mγ υ/2 + �E/υ, (56)
α� + β� + υ� = 0, �E = Ef − Ei , υ� ′ ≃ υ,
� (57)
η� = (η cos �η� , η sin �η� , 0), η2 = 4µα µβ υ 2 sin2 (θ/2), (58)
and η is the orthogonal (relative to vector υ) component of the transmitted

momentum (η� · υ� = 0) and θ is the scattering angle (the angle between
the vectors kα and kβ). Now, taking into account assumption �rα ≈�rβ ≈R
� , the
±
reaction amplitude Tαβ reads:

+ 1 1 � �
Tαβ = N1 Zβ �
dR d�x − ϕf∗ (�s)ϕi (�x)e −iaυ·�
� x−i α� ·R
F (R), (59)
R x
where

− 1 1 �R � �
Tαβ = N1 Zα �
dR d�s − ϕf∗ (�s)ϕi (�x)e ibυ·�
� s+i β·
F (R), (60)
R s
� = 1 F1 (−iνα , 1, ikα R − i k�α · R)

F (R) � 1 F1 (−iνβ , 1, ikβ R + i k�β · R),
�

π(να + νβ )
N1 = N (+) N (−) = Ŵ(1 + iνα )Ŵ(1 + iνβ ) exp − .
2
In further calculations we assume that the atoms in the entrance

and exit channels have zero orbital momentum, hence the wave functions
read:
3/2
ϕi (�x) = Zβ π −1/2 e −Zβ x , ϕf (�s) = Zα3/2 π −1/2 e −Zα s , (61)
and the energies are defined as Ei,f = −Zα,β

2 /2. We shall demonstrate the
−
calculation of the scattering amplitude on example of Tαβ . Consider the
integral:

� R −1 exp(i β� · R)
� � αβ− �
(−)
K (Zα , Zβ ) = dR F (R)I (R), (62)
where

− �
Iαβ (R) = d�s(xs)−1 exp(−Zα s − Zβ x − ibυ� · �s). (63)
Using the integral Fourier representation for exp(−Zβ x)/x:

exp(−Zβ x) 1 exp(i�q · x�)
= d�q, (64)
x 2π 2 q2 + Zβ2
� , and equality:
where x� = �s + R
1
1
= dt(ut + (1 − t)v)−2 , (65)
uv 0
the integration in (63) over s leads to expression:

1 �
− � 2 d�q exp(−i�q · R)
Iαβ (R) = dt . (66)
π b2 0 � 2 + �2− 2
|�q − Q|
Here, the following definitions have been used:
� = (t − 1)υ,
Q � �2− = υ 2 t(1 − t) + tZβ2 + (1 − t)(Zα /b)2 . (67)
Computing the integral in (66) by employing the theorem residues, one

obtains:
1
− � (68)
Iαβ (R) = 2π b−2 dt[�− ]−1 exp(−i Q
� ·R
� − R�− ).
0
Calculation of the integral in (62) can be made in the framework of the

method proposed by Nordsieck.33 We shall omit −
the algebra and give
here only the final result for reaction amplitude Tαβ in the form of a one-
dimensional integral:
1
− (−) ∂ 1 ∂ ∂ 1
Tαβ = N1 N 2 Z α dt − + �(−) (�− ), (69)
0 ∂Z β � �
− − ∂Z �
α −
iνα iνβ
4π 2µα υσ+ 2µβ υσ−
2 F1 (−iνβ , −iνα , 1, τ− ), (70)
(−)
� (�− ) = 2 1 − 2
1− 2
ω− ω− ω−
where, 2 F1 (a, b, c, x) is the Gauss hypergeometric function, and
N2(−) = (2/b2 )(Zα Zβ )3/2 , 2

ω− = p2 + �2− , (71)
2 v2
4µα µβ p⊥
τ− = 2 − 2µ υσ )(ω2 − 2µ υσ )
, σ± = i�− ± pz . (72)
(ω− α + − β −
Here, vector p� = β� − Q � is decomposed into the parallel pz and perpen-

dicular p⊥ components relative to the vector υ:

ˆ 1 �E
p� = (pz , p⊥ ), pz = �p · υ� = υ −t + , �p⊥ = η�.
2 υ (73)
In order to compute the integral (69), one should first differentiate over
the Zα , Zβ, and − variables and, at the final stage, make the following
substitutions:
να → Zα (Zβ − 1)/υ, νβ → Zβ (Zα − 1)/υ.
+
Using the same approach, one can compute the reaction amplitude Tαβ
in the post-formalism. Omitting the algebra, we represent here only the
final result:
1
+ (+) ∂ 1 ∂ ∂ 1
Tαβ = Zβ N1 N2 dt − + �(+) (�+ ), (74)
0 ∂Z α � �
+ + ∂Z �
β +
iνα iνβ
4π 2µα υσ− 2µβ υσ+
�(+) (�+ ) = 2 1− 2
1− 2 2 F1 (−iνβ , −iνα , 1, τ+ ), (75)
ω+ ω+ ω+
4µα µβ q⊥2 υ2
(+)
N2 = (2/a2 )(Zα Zβ )3/2 , τ+ = 2 2 − 2µ υσ )
, (76)
(ω+ − 2µα υσ− )(ω+ β +
and the following designations have been used:
�q = (qz , �q⊥ ), �q⊥ ·υ� = 0, qz = (t − 1/2)υ + �Eυ −1 , �q⊥ = −�

η.
Employing the same differentiation technique over the charges Zα and Zβ,
one can take into account the polynomial structure of the wave functions
of the excited states of hydrogen-like atom or ion.
Let us compare the described method for calculating charge-transfer
amplitude with the traditional approximations of the first order perturba-
tion theory. First, we discuss the asymptotic properties of the perturbing
potentials Uα and Uβ as a function of R. Using (4), we shall represent
vector s(x) in terms of R and x(s) and expand the Coulomb potential
Vβ = −Zα /s (Vα = −Zβ /x) into the series over the small parameter
x/R (s/R). As a result, we come to the following asymptotic (at R → ∞)
representations:
Zα Zα (
� · x�)
R Zβ Zβ (
� · �s)
R
Vβ ≃ − − + ··· , Vα ≃ − − + · · · (77)
R R2 R R2
which implies that when R tends to infinity, the potential Uα (Uβ ) tends
to zero as R −2:
Zα Zα Zβ Zβ
Uα = − − → −Zα O(R −2 ), Uβ = − − → −Zβ O(R −2 ). (78)
R s R x
This gives us grounds to assume that potentials Uα (Uβ ) are short-range.

Hence, one can observe a correlation between the behavior of the perturb-
ing potentials Uα (Uβ ) at large distances R and the asymptotic properties of
α+ β−
distorted waves χi and χf . This correlation leads to the short-range
character of the perturbation Uα (Uβ ) at the correct behavior (39) of the
β−
χiα+ (χf ) functions at infinity. Substitution in Eq. (51) of the plane waves
±
exp(i k�α,β · �rα,β ) instead of functions Fα,β and the channel interaction υα
instead of the transition operator Uα leads to the final result for transition
amplitude:

(B1) Zα Zβ Zα � �
Tαβ (�
η) = d�rα d�x ϕf∗ (�s) − ϕi (�x)e ikα ·�rα −ikβ ·�rβ . (79)
R s
This determines the direct one-step mechanism of reaction (1) in the

framework of the standard first Born (B1) approximation (the so-called
Jackson–Schiff approximation14). In the case of resonant charge transfer of
atomic hydrogen on the proton (Zα = Zβ = 1), the Coulomb wave func-
±
tions Fα,β reduce to the plane waves and formulae (51) and (79) coincide.
The discrepancy of these formulae for Zα,β �= 1 is due to the fact that in
the standard Born approximation, the Coulomb interaction in the entrance
and exit channels are not taken into account. Hence, the channel wave
function does not satisfy the Coulomb boundary conditions of the scatter-
ing problem. Thus, expression (51) may be considered as a correct general-
ization of the Born approximation which accounts for Coulomb distortion
of the asymptotic behavior of the wave functions in both the entrance
and exit channels of the charge-exchange reaction (1). This generalization,
which is the boundary-corrected first Born approximation, or CB1, will be
used in the next sections of this work. When the collision energies of the
projectiles are sufficiently high, the Coulomb wave functions do not differ
substantially from the plane waves, and hence the CB1 and the usual first
Born approximations almost coincide.24, 25 However, already at intermedi-
ate collision energies, it is necessary to take into account the long-range
Coulomb effects that substantially influence the behavior of the charge-
exchange cross sections.
We now reemphasize that when using the CB1 approximation, we neglect

the re-scattering effects, i.e., possible multiple-step reaction mechanisms. The
role of two-step transitions through the intermediate (bound or continuum)
state increases with increasing projectile energy.32 The quantitative descrip-
tion of such transitions becomes possible only with inclusion of the post-
collision interaction into the final state wave function. That is equivalent to
accounting for the effects of multiple-step re-scattering of the electron on
the rest of the target ion. In the next section, on the basis of the first iteration
of the Dodd–Greider equations (30) we shall disuss the one- and two-step
electron capture mechanisms and clarify their connection with the angular
distribution of the reaction products. Alternatively, re-scattering effects can be
described by the boundary-corrected second Born (CB2) approximation.32
4. THE CONTINUUM-INTERMEDIATE STATE METHOD

WITH THE CORRECT BOUNDARY CONDITIONS
−
Let us transform the expression for the reaction amplitude Tαβ in
β−
the following way. We introduce the scattering vector |�f � by definition:
β− β−
|�f � = [1 + gx+† (υβ − Wβ )]χf �. (80)
Substitution of (23) and (80) into (33) leads to the following representation
−
of the reaction amplitude Tαβ , which takes into account both the direct and
the two-step mechanisms:
− β−
Tαβ = ��f |υα − Wα |χiα+ �. (81)
In order to obtain the differential equation for the wave function �fβ−, we
†
shall multiply both parts of Eq. (80) with (E − H + υx − iε) from the left
and compute the limit ε → 0+ taking into account Eq. (38). As a result we
obtain:
β− β−
(E − H + υx† )|�f � = υx† |χf �. (82)
Since the solution of Eq. (82) with realistic local potential υx involves con-
siderable mathematical difficulties, one should attempt to replace potential
υx with an operator, which should be chosen so the condition:
β−
υx† |χf � = 0, (83)
is satisfied. Also, the solution of the corresponding homogeneous equation

should be represented as:
β−
|�f � = |ϕf (�s)hβ− � (84)
with the following boundary conditions:

β− β−
�f −→ �f = ϕf (�s) exp[i k�β · �rβ − iνβ ln(kβ ξβ )]. (85)
rβ →∞
By substituting Eq. (85) into (82) and taking into account (83), we come
to the following equation relative to the hβ−:
1�
ϕf (�s)(E − Ef − H0 − υβ )hβ− + ∇ �s ϕf (�
� �s h− + υx† (ϕf (�s)h− ) = 0. (86)
s)·∇ β β
b
We now return to the question concerning the choice of operator υx.
Closely related to this choice is the need to improve the convergence of
−
the iteration series of integral equation (30) for operator Uαβ . As described
in the previous section, the proper choice of the operator υx eliminates
the disconnected diagrams from its kernel (31). Indeed, by definition, (43)
of the perturbation Uβ = υβ − Wβ as a part of (31), contains terms that
depend solely on the relative vectorial variables x and R . Hence, in order to
eliminate the disconnected diagrams from the K operator, υx should only
act over the variable s , which is related to pair (α, γ ). Such an operator can
be constructed by the following definition:
1�
υx � = − ∇ � �s �/ϕf (�s) ,
s)·∇
�s ϕf (� (87)
b
on the set of elements ∈ H, where H is the subspace of continuum states
of the Hamiltonian H. The domain of differential operator υx includes
only those elements ∈ H whose image ′ obtained by the action υx
belongs to the same Hilbert space H (i.e., υx � = � ′ under the condi-
tion , ′ ∈ H). The latter is equivalent to the requirement that elements
admit a representation similar to (84). As a consequence, although the
operator contains singularities at the points of configurational space where
function ϕf (s) is equal to zero, the result of action υx � has no such singu-
larities. That the operator υx satisfies condition (83) is easily seen by using
the obvious relation ∇ � �s (χ β− /ϕf (�s)) = ∇
� �s F − (�rβ ) = 0.
f β
Thus, when choosing the operator υx in a form (87), the kernel K
of the integral equation (30) is defined only by those variables which
completely correspond to the connected diagrams. This means that the

iteration series of Eq. (30) should, in principle, converge faster and in a
wider energetic range than the original Born series of the three-body
problem.
Taking into account the explicit form (87) of operator υx and using
expression (5) for kinetic energy operator H0, Eq. (86) becomes:

1 1 Zβ Z α Zβ
E − Ef + ��r + �x� + − hβ− = 0. (88)
2µα α 2a x R
According to Eqs. (84) and (85), the asymptotic limit of a function hβ− at
rβ → ∞ has the form of a distorted plane wave with a unit amplitude:

iZβ (Zα − 1)
hβ− −→ fβ− (�rβ ) �
= exp i kβ · �rβ − �
ln(kβ tβ + kβ · �rβ ) . (89)
r →∞
β υ′
With the approximation R ≈ rα, the variables in Eq. (88) are separable and
the corresponding wave functions become explicitly representable in the
following form:
hβ− = C (−) F (−) (�x)ℑ(−) (�rα ). (90)
The two-body scattering Coulomb wave functions F (−) (�x) and ℑ(−) (�rα )
are defined through the confluent hypergeometric function as follows:3

iaZβ π aZβ iaZβ
F (−) (�x) = Ŵ 1 + exp + i�q · x� 1 F1 − , 1, iqx − i�q · x� , (91)
q 2q q

(−) iµα Zα Zβ π µα Zα Zβ
ℑ (�rα ) = Ŵ 1 − exp − + i�qα · �rα
qα 2q (92)
×1 F1 (−iµα Zα Zβ /q, 1, iqα rα − i�qα · �rα ).
The coefficient C (−) and vectors q, qα, that appear due to variable separa-
tion, can be defined by matching hβ− with the eikonal asymptotic limit
(89) at rβ → ∞. Using the asymptotic form of the confluent hypergeo-
metric function F(a, b, t) at t → ∞, we find from the matching condi-
tions that:
iZ /υ ′
�q = aυ� ′ ≃ υ� ′ , �qα = (b/mγ )k�β ≃ k�β , C (−) = µβ β . (93)
Summing up, let us represent the wave function of the final state �fβ−,
which describes the scattering of the charged particle β on the hydrogen-
like system (α, γ ) as:
iν ′
β−
�f = µβ t ϕf (�s) exp(i k�β · �rβ )F (−) (�x)ℑ(−) (�rα ), (94)
where
F (−) (�x) = N (+) F(−iνt′ , 1, −iυ ′ x − i υ� ′ · x�), νt′ = Zβ /υ ′ , υ� ′ = k�β /µβ , (95)
ℑ(−) (�rα ) = N (−) 1 F1 (iν ′ , 1, −ikβ rα − k�β · �rα ), ν ′ = Zα Zβ /υ ′ , (96)
N (+) = Ŵ(1 + iνt′ ) exp(π νt′ /2), N (−) = Ŵ(1 − iν ′ ) exp(−π ν ′ /2). (97)
β−
From the form of the function �f one can readily appreciate its con-
venience for studying the influence of the post-collision interaction on
differential cross sections of charge transfer, because it equally takes into
account the interaction of bound electron γ with the residual target ion β
and the interaction of heavy particles α and β. Recall that the initial wave
α+
function χi is defined by expressions (47) and (48).
Using now the expressions for the wave functions (94), (47), and (48),
the transition operator (42), and relation (81), we come to the following
charge-transfer reaction amplitude with account for Coulomb interaction
in the final state:

� � Zα Zα
−
Tαβ (BCIS) = N (υ, να ) d�rα d�x e i kα ·�rα −i kβ ·�rβ ϕf∗ (�s) − ϕi (�x)
rα s (98)
× 1 F1 (iνt′ , 1, iυ ′ x + i υ� ′ · x�)1 F1 (−iνα , 1, ikα rα − i k�α · �rα )1 F1 (−iν ′ , 1, ikβ rα + i k�β · �rα ),
where
π
−iνt′
N (υ, να ) = µβ Ŵ(1 + iνα )Ŵ(1 − iνt′ )Ŵ(1 + iν ′ ) exp να + ν ′ − νt′ . (99)
2
This is the boundary-corrected conntinuum intermediate state approxi-

mation,5 or BCIS. Without the eikonalization of the rα- and rβ-dependent
Coulomb waves, the calculation of amplitude (98) in the general case when
Zβ �= 1 is complicated by the presence of three confluent hypergeometric
functions under the integral sign. However, there is one important case,
when integration in (98) can be reduced to a one-dimensional numerical

integration.This is the case of charge exchange of a proton (or other charged
particle) with atomic hydrogen (Zβ = 1). In this case the Coulomb parameter
να = 0 and the hypergeometric function 1 F1 (−iνα , 1, ikα rα − i k�α · �rα ) in
formula (98) is equal to 1.
For further calculations we shall use expression (61) for bound state
wave functions and integral representation for confluent hypergeometric
function:
1
1
1 F1 (a, c, z) = t a−1 (1 − t)c−a−1 exp(zt)dt, (100)
B(a, c − a) 0
where B(a, c − a) is the Euler beta-function. By changing the order of integra-

tion for integrals in expression (98), we come to the following representation
−
for Tαβ :
−
Tαβ = N (v, 0)N2 [B(iνt′ , 1 − iνt′ )B(−iν ′ , 1 + iν ′ )]−1
1 1 (101)
iνt′ −1 −iνt′ ′ ′
× t1 (1 − t1 ) dt1 t2−iν −1 (1 − t2 )iν I (t1 , t2 )dt2 ,
0 0

∂2 ∂2
I (t1 , t2 ) = lim Zα − J, N2 = (Zα Zβ )3/2 π −1 . (102)
ε→0 ∂λα ∂λβ ∂λβ ∂ε
Here J is defined by the expression:

� e −λα s i(k�α +k�β t2 )·�rα e −δ1 rα e −δ2 x it1 υ� ′ ·�x
J= d�rα d�x e −ikβ ·�rβ e e , (103)
s rα x
where
δ1 = ε − ikβ t2 , δ2 = λβ − iυ ′ t1 , λα = Zα , λβ = Zβ .
Using the Fourier transformation:

exp(−λa s) 1
= exp(i k� · �s)(k2 + λ2α )−1 dk,
� (104)
s 2π 2
and the following relations:
�rβ = (b/mγ )�rα + (a/µβ )�x, �s = (a/mγ )�x − �rα ,

one can perform integration in (103) over the x and rα variables. As a result
one obtains the following expression for the matrix element J:

8mγ2 dk�
J= , (105)
a2 (k2 + λ2α ) |k� − �q1 |2 + ρ12 |k� − �q2 |2 + ρ22
where

mγ mγ ′ b �
�q1 = k�β + � t1 ,
v �
�q2 = kα − kβ + k�β t2 ,
µβ a mγ
and ρ1 = δ2 mγ /a, ρ2 = δ1. Using the results obtained in Ref. 34 for a rear-
rangement of the expression for J, and omitting the lengthy derivation, we
−
give the final result for charge-transfer amplitude Tαβ:
∞
16π 2 ∂2 ∂2
−
Tαβ = N (v, 0)N Z
2 α lim − K(x)dx, (106)
a2 ε→0 ∂λα ∂λβ ∂λβ ∂ε 0
where
1
iν ′ −1 ′
K(x) = [B(iνt′ , 1 − iνt′ )B(−iν ′ , 1 + iν ′ )]−1 t1 t (1 − t1 )−iνt dt1
0
1 (107)
′
−iν −1 iν ′ −1
× t2 (1 − t2 ) (A + Bt1 + Dt2 + Ct1 t2 ) dt2 .
0
The expressions for coefficients A, B, C, and D are quite involved and we

do not show them here. The integrals over t1 and t2 can be evaluated by the
formula from Ref. 33. The final result for K(x) reads:
′ ′
K(x) = A−1 (1 + D/A)iν (1 + B/A)−iνt
(108)
×2 F1 iνt′ , −iν, 1; (BD − AC)(A + D)−1 (A + B)−1 .
The one-dimensional integral over x in expression (106) should be com-

puted numerically.
Let us compare the proposed approach for investigating the charge-
exchange reactions with the traditional continuum distorted wave approxi-
mation,3 or CDW. The transition amplitude of the CDW approximation in
the prior-formalism is given by formulae:

−
Tαβ (CDW) = NCDW d�rα d�x exp(i k�α · �rα − i k�β · �rβ )Z − rα , �rβ ), (109)
αβ J (�
where
J (�rα , �rβ ) = N (+) N (−) 1 F1 (−iν, 1, ikα ξα )1 F1 (iν, 1, −ikβ ξβ ),

(110)
Z − = ϕf∗ (�s)1 F1 (iνt′ , 1, iυ ′ x + i υ� ′ · x�′ )∇
αβ
� x� ·ϕi (�x)·∇
� �s 1 F1 (iνp , 1, iυs + i υ� · �s),
NCDW = Ŵ(1 − iνt′ )Ŵ(1 − iνp ) exp(π(νt′ + νp )/2), (111)

with ν = Zα Zβ /υ, νp = Zα /υ, νt′ = Zβ /υ ′. The CDW approximation
was first used by Cheshire35 for calculation of the cross section of resonant
charge transfer of fast protons on atomic hydrogen. Subsequently, Gayet36
showed that amplitude (109) in the prior-formalism can be obtained as a
first term of the Dodd–Greider perturbation series28 for operators of three-
body rearrangement scattering. Note that due to small scattering angles of
heavy projectiles, the eikonal approximation can be used with the replace-
ment of all the rα - and rβ -dependent Kummer functions 1 F1 (a1 , b1 , ξα )
and 1 F1 (a1 , b2 , ξβ ) in the CB1, BCIS and CDW methods from Eqs. (53),
(98) and (110), respectively, by their logarithmic Coulomb phase factors.3
This significantly simplifies the calculations.3, 8–10, 46
5. ANGULAR AND ENERGY DEPENDENCIES OF

CHARGE-TRANSFER CROSS SECTIONS
As seen from the previous section, the quantum scattering theory
equations for a few-body system can serve as a universal mathematical
background for constructing various approximations for charge-transfer.
The iterations of such equations represent scattering amplitudes in the
form of series (32). The number of terms taken into account defines the
order of re-scattering. The obtained representations (33)–(36) for scattering
amplitude are based only on accounting for the first terms of the iteration
series of the Dodd–Greider equations. Therefore, such an approach is appli-
cable if the specific collision energy conditions are satisfied. These energies
must significantly exceed the binding energies of the compound particles
in the initial and final channels. However, they should not be extremely
high, because with increasing collision energy the role of multiple scatter-
ing of electrons is also augmented. The influence of the higher terms of the
iteration series for three-body rearrangement scattering operator has been

studied in Refs. 15–17, 32. The qualitative estimates of the contribution of
such mechanisms in the nonrelativistic collision energy range showed that
this contribution is small compared to the cross sections of one- and two-
step mechanisms.
The first applications of the reviewed formalism were for studying the
charge-transfer collisions: 3, 35
H+ + H(1s) → H(1s) + H+ . (112)

This reaction is of interest for many applications and also plays an impor-
tant role in testing a variety of theoretical approaches for rearrangement
processes. Before proceeding to analysis of reaction (112), we notice the fol-
−
lowing. For process (1) the scattering amplitude Tαβ has a sharp maximum
in the region of small angles: 3, 37
θ mγ /µ ≪ 1, µ = mα mβ /(mα + mβ ).
In the case of reaction (112), when particles α and β are the protons, the
exchange part of scattering amplitude is negligible.38 Exhaustive com-
parisons of the CB1 and CDW approximations with all the experimental
data for process (112) and for other similar charge-transfer collision, have
systematically shown good agreement at intermediate and high energies
(80 keV/amu).3, 9
The need for improvements in the theoretical approaches as well as the
development of ion-atomic collision physics led to a shift of emphasis from
studies of the integral characteristics toward the differential nature of col-
lision processes. The latter includes more detailed information concerning
the role of various reaction mechanisms and are a much more stringent test
of theoretical approaches.
The question concerning charge-transfer mechanisms at intermediate
and high velocities has remained open for a long time. The lack of reli-
able experimental data on differential cross sections hampered the choice
among various theoretical calculations. For instance, the first Born approxi-
mation14 gives a forward scattering peak in the cross sections, correspond-
ing to a direct
√one-step mechanism of electron capture with a minimum at
angle θLS ∼ 3/2mp (mp is the ratio of proton mass to the electronic mass,
θLS is the scattering angle of the formed hydrogen in the laboratory system
of coordinates). However, the more precise asymptotic analysis of √the scat-
17
tering amplitude indicated that the maximum at angle θLS ∼ 3/2mp is
related to a classical Thomas mechanism of electron capture.39 The quantum
counterpart of the classical Thomas mechanism is the electronic transition

through the continuum spectra from the target into the states bound to the
fast projectile. In the BCIS method, the effects of multiple Coulomb re-
scattering of the captured electron are taken into account by approximate
summation in the distortion factor F (−) (�x) from (95).
In the differential cross sections of charge transfer a two-step mecha-
nism of electron capture is reflected by a maximum (Thomas peak) at spe-
cial scattering angle θT . For the first time, the Thomas peak was observed
in p + He and p+H collisions at energies of a few MeV at scattering angle
0. 47 mrad. 40, 41
In Refs 1, 3 the results of the calculations of the differential cross section
for electron capture in the p+H collisions using amplitudes (69) and (109)
are compared with experimental data.41 Similar comparisons have also been
carried out between the CB2 method32 and the measurement of angular
distributions41, 42 for the same process (112).These comparisons have shown
that the CB2 method is in good agreement with experiments. The most
interesting from a methodological point of view is the comparison of calcu-
lations of angular distribution with amplitude (69) of a one-step mechanism
and calculations by formula (109), which also takes into account the two-
step (Thomas) mechanism of electron capture. Here, one can observe that
account of the two-step effects gives rise to a maximum (Thomas peak) at
the site of the dip in the JS method, obtained in a simple one-step mecha-
nism. At large scattering angles the interaction of heavy particles is taken
into account in the BCIS and CDW methods through the factors ℑ(−) (�rα )
and J (rα , rβ ) in Eqs. (96) and (100), respectively.
Regarding the discussed theoretical investigations on the Thomas peak
in differential cross sections, the most significant influence on the shape
of angular distributions is from Coulomb re-scattering of the electron on
the remaining target ion in the final state. In contradistinction, by neglect-
ing the Coulomb interaction in the final state or by calculations using the
single-step collision approximation, the Thomas peak is not reproduced.
The obtained agreement between the theoretical and experimental data
on differential and total cross sections in a wide range of collision energies
demonstrates the reliability of the described approach based on the first
iteration of the Dodd–Greider equations.
The main restriction of the method described herein is related to the
nature of the three-body problem which assumes only one-electronic
degree of freedom. This means that the theory is applicable, rigourously
speaking, only to one-electronic systems. However, in some cases
multi-electronic systems can be considered as a quasi three-body system

with a corresponding non-local or effective local interaction potential
between the active electron and the electronic core.
The multi-electronic atom is described by a one-electronic approxima-
tion, i.e., it is assumed that the active electron is moving in the frozen atomic
core. In that case, all the arguments presented in this section remain valid,
except that instead of hydrogen-like wave functions ϕi (x) and ϕf (s) from
Eq. (61) one should use the eigenfunctions of the Hamiltonians Hα , Hβ
with effective one-center potentials Vαeff and Vβeff . For calculation of the
charge-transfer amplitude, the CB1-approximation (see Section 3) can be
+
used. The electron capture amplitude Tαβ from (74) with the perturbing
post-collision potential Uβ from (43) is correct for Zα > Zβ (Ref. 3). In the
case of electron capture by protons from multi-electron atoms for Zα < Zβ,
−
one should use the transition amplitude Tαβ from (69) with the pre-collision
perturbation Uα from (42).
When describing the electronic bound states in multi-electronic target
atom, the simple model for one-particle wave functions is usually employed
in the Roothaan–Hartree–Fock (RHF)43 model with the effective charge
ZβRHF = (−2ERHF )−1/2 for the initial wave function. For testing the sensi-
tivity of the results of the chosen model on the wave functions, also utilized
is the hydrogen-like (HL) model with the effective charge ZβRHF.The values
of the cross sections computed within the CB1, CDW and BCIS approxi-
mations by means of the RHF and HL models are in good agreement with
experimental data.3, 8, 10, 44
6. THE DODD–GREIDER INTEGRAL EQUATION IN

THE THEORY OF TWO-ELECTRON PROCESSES
This section is devoted to a theoretical description of two-electron
capture:45–51
AZα + + B → A(Zα −2)+ + B 2+ , (113)

which will be reviewed following Refs. 6, 48 at intermediate and large
velocities of colliding particles. The cross sections of a two-electron process
(113) are quite large (∼10−18 − 10−16 cm2 ) and for certain collisions their
contribution to charge transfer could reach the same order of magnitude
as one-electron processes.
The success of the CDW theory regarding one-electron processes

stimulated the development of the reviewed distorted wave approach
for describing two-electron processes.48 This latter formalism is similar
to the CDW method3 for one-electron capture, but more complicated.
The amplitude of reaction (113) can be calculated by approximating the
mechanism of simultaneous capture of two electrons.The application of the
general theory has been demonstrated in the past by the example of two-
electron capture in collisions of helium atoms with nuclei.48–5
In the framework of nonrelativistic quantum mechanics, we shall con-
sider collisions in the system of four particles α, β, γ1 , γ2 in which three
particles are bound (i.e., they form a “compound” particle) both in the
initial and in the final reaction channel:
α + (β; γ1 , γ2 ) → (α; γ1 , γ2 ) + β, (114)
where symbol (λ; γ1 , γ2 ) designates the corresponding compound particle
(λ = α, β are the atomic nuclei, and γ1 , γ2 are the electrons). We did not
take into account the spin of particles (the Coulomb effects studied here do
not depend on spin). Consider the full Hamiltonian of the system H0 + V ,
where H0 is the kinetic energy operator of the four-particle system in the
center of mass system.The full interaction potential V is given by expression:
2

V = (Vα,γk + Vβ,γk ) + Vγ1 ,γ2 + Vα,β , (115)
k=1
where Vα,γ1 is the operator of the pair interaction of α and γ1, etc. We
shall denote by Vα (Vβ ) the effective interaction potential which forms
the compound particle in the initial (final) channel of reaction (114);
Hα = H0 + Vα (Hβ = H0 + Vβ ) the Hamiltonian of initial (final) chan-
nel of reaction (114); G(W ) = [W − H ]−1 is the Greens’ function of the
Hamiltonian H. Let us also define operator υλ = V − Vλ (λ = α, β).
−
In the prior-formalism the transition amplitude Tαβ from channel “α” to
channel “β” is given by the standard form:
− −
Tαβ = lim ��β |υα + υβ G(W )vα |�α � ≡ ��β |Uαβ |�α �. (116)
W →E+i0
−
Here, Uαβ is an operator responsible for the α → β transition; |�α � and
��β | are the initial and final asymptotic states of the system that are the
eigenfunctions of the operators Hα , Hβ with eigenvalues Eα ,
Eβ; E is the
total energy of the four-particle system.
In order to write the Dodd–Greider integral equation for a system of

four particles, we shall represent the channel interaction υλ(λ = α, β) as
the following sum υλ = (υλ − ωλ ) + ωλ. Here, ωα , ωβ are the disturbing
potentials in the initial and final channels of reaction (114). Corresponding
to these potentials, we shall introduce the Møller wave operators:
ωα+ = 1 + (E − Hα − wα + iε)−1 wα = 1 + gα+ wα , (117)
ωα− = 1 + (E − Hβ − wβ − iε)−1 wβ = 1 + gβ− wβ , (118)
where ε is an infinitely small positive number. By analogy with the

three-particle case, we introduce the auxiliary potential υx, which cor-
responds to an intermediate virtual channel “x” and the Greens operator
−
gx+ = (E − H + υx + iε)−1. Using these definitions the operator Uαβ reads:

− −†
Uαβ = ωβ [1 + (υβ − wβ )gx+ ](υα − wα )ωα+ + (υβ − wβ )gx+ υx Gβ+ Uαβ
−
. (119)
Up to now, Eq. (119) has been exact. Consider the approximation for the
−
transition operator Uαβ , namely on the right-hand side of (119), we shall
keep only the first iteration. As a result, we come to the following repre-
−
sentation of transition amplitude Tαβ :
− −†
Tαβ = ��β |ωβ [1 + gx+ (υβ − wβ )](υα − wα )ωα+ |�α �
−†
(120)
−
= Tαβ (DWB) + ��β |ωβ [gx+ (υβ − wβ )](υα − wα )ωα+ |�α �,
− −†
where Tαβ (DWB) = ��β |ωβ (υα − wα )ωα+ |�α � is the amplitude of reac-
tion (114) in the distorted wave Born approximation (DWB). The first
term on the right-hand side of (120) corresponds to the direct electronic
transition without any re-scattering, whereas the second term describes the
two-step mechanism of electron capture.
Consider the system which consists of four particles with coordinates ri
and masses mi (i = 1, 2, 3, 4). The numbers 1, 2, 3, 4 denote the particles γ1,
γ2, α, β, correspondingly. We shall introduce two sets of coordinates �rα , x�′k
′
and �rβ , �sk (k = 1, 2) defined by equations:
k−1
�r1 + �r2 + mβ �r4 ′ mβ �r4 + i=1 �ri
�rα = �r3 − , x�k = �rk − , (121)
mβ + 2 mβ + k − 1
k−1
�r1 + �r2 + mα�r3 mα�r3 + i=1 �ri
�rβ = �r4 − , �sk′ = �rk − , (122)
mα + 2 mα + k − 1
where mα,β = m3,4 /m and m1 = m2 = m. Introducing the radius-vectors

xk and sk which determine the position of the kth electron (γk ) relative to
� is the distance
the nuclei β and α, respectively, the difference x�k − �sk = R
between the nuclei α and β. In these designations the channel interactions
vα and vβ read:
Zα Zα Zβ Zα Zβ Zα Z β Zβ
υα = − + − , υβ = − + − , (123)
s1 R s2 x1 R x2
where Zα and Zβ are the charges of nuclei α and β, respectively.

The eigenstates |�α � (|�β �) of the Hamiltonian Hα (Hβ ) are the
product of a wave function ϕα (�x′1 , x�′2 ) (ϕβ (�s1′ , �s2′ )) of the bound state of
system (β; γ1 , γ2 ) ((α; γ1 , γ2 )) and a plane wave of the relative motion of
particles in the initial (final) state:
�α = ϕα (�x′1 , x�′2 ) exp(i k�α · �rα ), �β = ϕβ (�s1′ , �s2′ ) exp(−i k�β · �rβ ), (124)
where kα (kβ ) are the momentum of the incident (scattered) particle in the
system of a center of masses before (after) the collision.
From the formal point of view, the operator ωα+ (ωα− ) may be con-
sidered as a transition operator from the initial (final) asymptotic state
+ −
|�α � (|�β �) into the distorted wave |χα � (|χβ �) in entrance (exit) reaction
channel:
|χα+ � = ωα+ |�α �, (125)
|χβ− � = ωβ− |�β �. (126)

Note that one can consider Uα = υα − wα as the operator responsible for
the transition from state (α) into state (β).
We introduce the state vector |�β− � by the definition:

|�β− � = 1 + gx+† (υβ − wβ ) |χβ− . (127)
In designations (125)-(127), transition amplitude (120) can be represented

as follows:
−
Tαβ = ��β− |Uα |χα+ �. (128)
In order to obtain the explicit differential equations for calculation of

the distortion in the initial and final channels, we shall apply operator
(E − Hα − wα ) to both parts of Eq. (125). Taking the limit ε → 0+, we

come to the following expression for the distorted wave in the exit channel:
(E − Hα − wα )|χα+ � = (E − Hα )|�α � = 0, E = Eα + kα2 /2µα . (129)
On the basis of Eq. (126), one obtains a similar expression for the distorted
wave in the exit channel:
(E − Hβ − wβ )|χβ− � = (E − Hβ )|�β � = 0, E = Eβ + kβ2 /2µβ . (130)
Here, Eα and Eβ are the bound energies of compound particles (β; γ1 , γ2 )

and (α; γ1 , γ2 ), where µα = mα (2 + mβ )/M and µβ = mβ (2 + mα )/M are
the reduced masses of the corresponding compound and M = mα + mβ + 2
is the total mass of the system.
Further calculations are based on the special requirements concerning
−
the choice of function χβ which, essentially, make the problem considered
−
here explicitly solvable. Assume that solution χβ is represented in the fol-
lowing factorized form:
|χβ− � = |ϕβ (�s1′ , �s2′ )f (�rβ )�, (131)
where f (rβ ) is the function which describes the asymptotic motion of

a compound (α; γ1 , γ2 ) in a Coulomb field of particle β. We note that
assumption (131) is only applicable if the relative motion of heavy particles
is larger than the effective orbital velocity of a bound electron.
The differential equation (130) must be supplemented with the b oundary
condition, which in this case has the following form:

iZβ (Zα − 2)
χβ− −→ ϕβ (�s1′ , �s2′ ) exp �
−i kβ · �rβ − ln(kβ rβ − kβ · �rβ ) , (132)
�
r →∞
β υ′
−
where υ� ′ = k�β /kβ. The fulfilment of the condition on the function χβ
can be achieved by the corresponding choice of distortion potential wβ in
Eq. (130). For instance, wβ may be defined as wβ = Zβ (Zα − 2)/rβ, and
then the function f (rβ ) can be represented through the confluent hyper-
geometric function. However, we omit explicit representation of f (rβ )
since it is not necessary for our purpose.
Let us now construct the differential form of Eq. (127). Applying the
†
operator (E − H + υx ) to both sides of (127) and taking into account
(130) we obtain (in the limit ε → 0+):
(E − H + υx† )|�β− � = υx† |χβ− �. (133)
The solution of Eq. (133) for realistic potential υx is an extremely compli-

cated task. Hence, it is reasonable to substitute this potential by an operator
chosen in such a way that the following condition is satisfied:
υx† |χβ− � = 0. (134)
Similarly to expression (131)) the solution to (133) is represented as:
|�β− � = |ϕβ (�s1′ , �s2′ )Y − �, (135)
where the unknown function Y − describes the distortion of the wave

function ϕβ (�s1′ , �s2′ ) of the bound state of system (α; γ1 , γ2 ) induced by the
interaction with nucleus β in the exit channel.
By virtue of condition (134), Eq. (133) reduces to a homogeneous one:
2
Zβ Zα Zα Zβ 1
E − H0 + + − − + υx |�β− � = 0.
†
(136)
xk sk R |�s1 − �s2 |
k=1
Using the above-defined two sets of relative variables, one can represent the
Hamiltonian H0 in two equivalent forms:
2 △ ′
2 △′

△�rα x�k △�rβ �sk
H0 = − − =− − , (137)
2µα 2µβk 2µβ 2µαk
k=1 k=1
where
µβk = (mβ + k − 1)/(mβ + k), µαk = (mα + k − 1)/(mα + k).
The interaction potentials in H0 depend on different combinations of

relative variables. This circumstance leads to severe complications when
solving Eq. (136). In order to avoid this difficulty, consider the approximate
variable separation in Eq. (136) based on the natural approximation that
the masses of particles γ1 , γ2 (electrons) are much smaller than the masses
of the two other particles α and β (atomic nuclei), i.e., m1 = m2 ≪ m3,4.
Therefore, one can neglect the terms that contain the ratio mk /m3,4
(k = 1, 2) in expressions (121) and (122). As a result we come to the fol-

lowing approximate relations:
x�′k ≃ x�k , �sk′ ≃ �sk , �

�rα ≃ R, �
�rβ ≃ −R. (138)
−
By substituting expression (135) for the wave function |�β � into Eq.
(136) and taking into account (130) and (138), one obtains the equation
for Y −:

ϕβ (E − Eβ − H0 − υβ )Y − + µ−1 � � − † −
αk ∇�sk′ ϕβ ·∇�sk′ Y + υx (ϕβ Y ) = 0. (139)
k=1,2
As an operator υx we chose a potential whose action on the arbitrary func-

tion �(�rβ , �s1′ , �s2′ ) is defined as follows:

υx � = − µ−1 � �
αk ∇�sk′ ϕβ ·∇�sk′ (�/ϕβ ). (140)
k=1,2
By substituting (140) into Eq. (139) and taking into account (138), we
obtain the equation for Y −:
 � � 
△�rα � △x�′ Z β Z Z
α β −
E − E β + + k
+ ′ − Y = 0. (141)
2µα 2µβk xk rα
k=1,2
The solution to Eq. (141) should behave at infinity as a distorted plane

wave:

− � iZβ (Zα − 2) �
Y −→ f (�rβ ) −→ exp −i kβ · �rβ − ln(kβ rβ − kβ · �rβ ) . (142)
rβ →∞ rβ →∞ υ′
Solving Eq. (141) by the method of variable separations, we obtain the fol-
lowing solution, as in Ref. 48:
2

− (−)
Y =C (−)
F (−)
(�rα ) Fk (�x′k ), C (−) = const, (143)
k=1
where the two-particle Coulomb distorting functions and

(−)
Fk (�x′k ) (F (−) (�rα )) are given by:
(−)
F k (�x′k ) = N (+) (νβk
′
) exp(i�qk · x�′k )1 F1 (−iνβk
′
, 1, −iqk ξk ), (144)
F (−) (�rα ) = N (−) (να′ ) exp(i�qα · �rα )1 F1 (iνα′ , 1, −iqα ξα ), (145)
where N (±) (ν) = Ŵ(1 ± iν) exp(±πν/2) are the normalization coef-
ficients, ξk = x′k + �qˆ k · x�′k , ξα = rα + �qˆ α · �rα are the two-particle parabolic
variables, �qˆ k and �qˆ α are the unit vectors directed along the vectors qk and qα,
′ = Z µ /q ′
νβk β βk k and να = Zα Zβ µα /qα are the Sommerfeld parameters.
Energy conservation implies the following relation between qα and qk:
kβ2 qα2 q2
k
E − Eβ = = + . (146)
2µβ 2µα 2µβk
k=1,2
Matching the asymptotic forms of function (143) and their eikonal expres-
sions (142) leads to:

�qk · x�′k + �qα · �rα = −k�β · �rβ . (147)
k=1,2
The coordinates rα , rβ, x�′1, and x�′2 are related by:
mα mβ
�rβ = −a2�rα − µ−1
β bk x �′k , ak = , bk = .
(mα + k) (mβ + k) (148)
k=1,2
Substituting now (148) into the right-hand side of (147) and taking into
account the asymptotic limits:
�qα = a2 k�β −→ k�β , �qk = bk µ−1 � �′

β kβ −→ υ (k = 1, 2), (149)
mα →∞ mβ →∞

iZβ kβ rβ − k�β · �rβ −iZ /υ ′
− ′ ln −→ ln µβ β (k = 1, 2), (150)
υ υ ′ x′k + υ� ′ · x�′k rβ →∞
β −2iZ /υ ′
one obtains that C (−) = µβ . From Eqs. (143)–(145) and (149), the
distorted wave distortion function Y (−) in the exit channel can be repre-
sented as:48
2iν ′
Y − = µβ β N (−) (ν ′ )[N (+) (νβ′ )]2 exp(−i k�β · �rβ )
2
(151)
′ ′ ′ ′
×1 F1 (iν , 1, −ikβ rα − i k�β · �rα ) 1 F1 (−iνβ , 1, −iυ xk − i υ� · x�′k ),
′
k=1
where νβ′ = Zβ /υ ′ , ν ′ = Zα Zβ /υ ′. Hence, we constructed the wave func-

tion �β− (defined by Eqs. (135) and (151)) that describes the scattering of
charged particle β on a bound compound of three particles α, γ1 , γ2.
The wave function of the initial state χα+ can be constructed in a similar
manner:
χα+ = ϕα (�x′1 , x�′2 )Y + , (152)

where Y + satisfies the following differential equation:
 � � 
△�rβ � △�sk′ Zα Z α Zβ
E − E α + + + −  Y + = 0. (153)
2µβ 2µαk sk′ rβ
k=1,2
The solution of Eq. (153) can be constructed by comparison of its asymp-

totic with the corresponding eikonal approximation. We omit the cum-
bersome derivation (which was explained for the Y − construction) and
represent only the final result (in the limit m3,4 ≫ m1 = m2):
Y + = µ−2iν
α
�α · �rα )
α N (−)∗ (ν)[N (+)∗ (ν )]2 exp(i k
α
2
(154)
′
×1 F1 (−iν, 1, ikα rβ + i k�α · �rβ ) 1 F1 (iνα , 1, iυsk + i υ� · �sk′ ),
k=1
where να,β = Zα,β /υ, ν = Zα Zβ /υ.

Substituting now expressions (135), (151), (152), and (154) into (128),
−
one can obtain the reaction amplitude Tαβ which describes the simultaneous
capture of two electrons by fast ions in collisions with atoms:48
2
− (+)∗ ′
Tαβ =− N (+)∗
(να )N (νβ ) d�x′1 d�x′2 d�rα exp(i k�α · �rα + i k�β · �rβ )
2

×Z (�rα , �rβ )ϕβ∗ (�s1′ , �s2′ ) ′ ′ ′
1 F1 (iνβ , 1, iυ xj + i υ� ′ · x�′j ) � x�′ ϕα (�x′1 , x�′2 ) (155)
∇ k
j=1 k=1,2
� �s′ [1 F1 (iνα , 1, iυs1′ + i υ� · �s1′ )1 F1 (iνα , 1, iυs2′ + i υ� · �s2′ )],
·∇ k
where
−2iνβ
Z (�rα , �rβ ) = µ−2iν
α
αµ N (−)∗ (ν)N (−)∗ (ν ′ )
β (156)
×1 F1 (−iν, 1, ikα rβ + i k�α · �rβ )1 F1 (−iν ′ , 1, ikβ rα + i k�β · �rα ).
In the case of fast collisions (kα2 /2µα > |Eβ − Eα |) when forward scat-
tering is dominant, the expression (156) can be simplified. At small scatter-
ing angles (i.e., k�ˆ α ≃ k�ˆ β ) we can assume that nuclear motion occurs with
a constant velocity and the trajectories are close to linear. Hence, the vec-
tor R is represented as the orthogonal sum R � = ρ� + z� , ρ� · z� = 0. Taking
′
into account that υ� ≃ υ� and m1 = m2 = m ≪ m3 ∼ m4, one obtains
immediately:
k�α · �rα + k�β · �rβ ≃ p� · (�x1 + x�2 ) + �q · (�s1 + �s2 ), (157)
where

E β − Eα ˆ E β − Eα ˆ υ�
−2�p = υ� + υ − �
υ, 2�q = υ� − υ − �
υ, υ�ˆ = , (158)
υ υ υ
where η is orthogonal to the υ component of the transmitted momen-
tum (� η·υ� = η�·�z = ρ·�� z = 0). Using the asymptotic form3 for the function
−
Z (rα , rβ ), the amplitude Tαβ for scattering at small angles reads:

− (k)
Tαβ = −[N (+)∗ (να )N (+)∗ (νβ )]2 Iαβ , (159)
k=1,2
−
where Tαβ is given by the following expression48:

− (∗)
Tαβ = d�x1 d�x2 d�s1 exp [i�p · (�x1 + x�2 ) + i�q · (�s1 + �s2 )] ϕβ (�s1 , �s2 )
2

× � x� ϕα (�x1 , x�2 )
+ i υ� · x�j )∇ (160)
1 F1 (iνβ , 1, iυxj k
j=1
� �s [1 F1 (iνα , 1, iυs1 + i υ� · �s1 )1 F1 (iνα , 1, iυs2 + i υ� · �s2 )].
·∇ k
As an example, consider charge transfer in collision of helium atom

with nuclei. The initial and final states are the s2-states:
ϕα (�x1 , x�2 ) = ϕα (�x1 )ϕα (�x2 ), ϕβ (�s1 , �s2 ) = ϕβ (�s1 )ϕβ (�s2 ), (161)
where ϕα (xk ) and ϕβ (sk ) are the hydrogen-like wave functions with effec-
tive charges Zα and Zβ:
ϕα (�xk ) = (α 3 /π )1/2 exp(−αxk ), ϕβ (�sk ) = (β 3 /π )1/2 exp(−βsk ).

We shall proceed to calculation of a matrix element (160). Consider the

(1)
first term Iαβ which we shall write in the momentum representation:

(1) � β(1) (�q − τ�)R
� α(1) (�p + τ�)Rβ(2) (�q + τ�)Rα(2) (�p − τ�),(162)
Iαβ = (2π)−3 d� τR
where the following designations are introduced:

(j) �
Rβ (k) = d�sj exp(i k� · �sj )ϕβ∗ (�sj )∇
� � �s 1 F1 (iνα , 1, iυsj + i υ� · �sj ), (163)
j

� α(j) (k)
R � = d�xj exp(i k� · x�j )1 F1 (iνβ , 1, iυxj + i v
� · x�j ) ∇� x� ϕα (�xj ) , (164)
j

(j) �
Rβ (k) = d�sj exp(i k� · �sj )ϕβ∗ (�sj )1 F1 (iνα , 1, iυsj + i υ� · �sj ), (165)

(j) �
Rα (k) = d�xj exp(i k� · x�j )1 F1 (iνβ , 1, iυxj + i υ� · x�j ). (166)
(k)
The exact numerical integration over τ in Iαβ (k = 1, 2) needed in (159)
has been carried out in Refs. 5, 6, 48–51. For a rough estimate at high
energies, a peaking approximation of the quadrature from (162) might be
made as follows. The expression under the integral (162) is localized in the
following four regions of momentum τ subspace:
|�q − τ�| 1/a, |�p + τ�| 1/a, |�q + τ�| 1/a, |�p − τ�| 1/a, (167)
where a is the effective radius of a pairwise interaction. Since R α(1) (

τ ) and
(1) (2) (2)
Rβ ( τ ) decrease faster than Rα ( τ ) and Rβ ( τ ), the contribution to the
integral (162) from the third and fourth region from (167) is negligible. In
addition, the slowly varying function R � β(1) (�q − τ�)R � α(1) (�p + τ�) ≡ Rαβ (� τ)
can be taken outside the integral sign in (162). Applying the convolution
theorem and using the Nordsieck integral technique33, we obtain:

(1) � β(1) (�q − �p)R
� α(1) (2�p) + R � β(1) (2�q)R � α(1) (�p − �q)
Iαβ = −Nα Nβ (1/2) R
∂
× J (α + β, �p + �q, να , νβ , υ,
� υ),
� (168)
∂α
where the following designations have been used:

�
� ν1 , ν2 , υ�1 , υ�2 ) =
J (λ, k, d�x x−1 e −λx+ik·�x 1 F1 (iν1 , 1, iυ1 x + i υ�1 · x�)
−iν1
4π 2(k� · υ�1 − iλυ1 )
×1 F1 (iν2 , 1, iυ2 x + i υ�2 · x�) = 1 +
(k2 + λ2 ) (k2 + λ2 )
×[1 + 2(k� · υ�2 − iλυ2 )/(k2 + λ2 )]−iν2 2 F1 (iν1 , iν2 ; 1, ω), (169)
4(k� · υ�1 − iλυ1 )(k� · υ�2 − iλυ2 ) − 2(k2 + λ2 )(υ�1 · υ�2 − υ1 υ2 )

ω= ,
[k2 + λ2 + 2(k� · υ�1 − iλυ1 )][k2 + λ2 + 2(k� · υ�2 − iλυ2 )]
with Nγ = (γ 3 /π )1/2 and γ = α, β. The interpretation of Eqs. (168) and

(1) (1)
(169) is as follows. The matrix element Rα andRβ describe the two-step
(Thomas) mechanism of capture of electron r1 through the continuum
from target atom (β; γ1 , γ2 ) in the states bounded with the projectile (α).
(1)
The factor ∂J /∂α, which corresponds to the integration in Iαβ over the
coordinates of second electron (γ2 ), is reduced to the overlap integral. In a
(2)
similar way one can construct the expression for Iαβ which is omitted here
for the sake of brevity.
In detailed applications to double electron capture5, 6, 48-51 in colli-
sions of fast ions (protons, alpha particles, lithium nuclei) with helium, the
CDW method was found to be in good agreement with measurements.
Reasonably successful description of double charge exchange has also been
accomplished by the CB1,45 BCIS46 and BDW (Born distorted wave)47
approximations, albeit in certain restricted intervals of impact energies.
In the present exposition only one mechanism of reaction (113) for
the simultaneous capture of two electrons is taken into account. The other
mechanisms, such as sequential one-electron capture may also contribute
to the cross section of (113). The electronic correlation has been neglected
in the present formalism. Obviously, simultaneous account of different
mechanisms and effects should improve the theoretical picture of this rather
complicated process of two-electron capture in atomic collisions.
7. CONCLUSION
In the present work, the three-body Dodd–Greider integral equations
have been reviewed by focusing on the correct Coulomb asymptotic forms of
the wave functions for the problem of rearrangement scattering. Specifically,
distorted wave theories for one-electron capture reactions in energetic
collisions of hydrogen-like atoms with positive ions are recapitulated.
In particular, a reaction amplitude is highlighted as a first iteration term of

the solution of the Dodd–Greider equations for the three-body scatter-
ing operator. It is emphasized that to reproduce the Thomas peak in the
angular distribution of the reaction products, it is necessary to account for
the effects of Coulomb re-scattering of the captured electron. In approxi-
mation of the “single-step” collision, this method reduces to the version of
the well-known boundary-corrected first Born approximation, where the
asymptotic behavior of the particles in the entrance and exit reaction chan-
nels is described by two-body Coulomb wave functions. The short-range
interaction due to incomplete screening of the nucleus by atomic electrons
is taken into account in the perturbing potential.
On the basis of the Dodd–Greider integral equations for a quantum
mechanical operator of a four-particle rearrangement scattering, we have
reviewed a version of the continuum distorted wave method to describe
two-electron capture in energetic ion-atomic collisions. An important
feature of this method is a consistent preservation of the proper asymp-
totic limits of the wave functions of a colliding system in the entrance
and exit reaction channels that takes into account the long-range nature
of the Coulomb interactions. The reaction amplitude can be calculated in
approximation of simultaneous two-electron transitions. The correspond-
ing results from the literature have been discussed for two-electron capture
in energetic H+ + He and He2+ + He collisions. The related theoretical
cross sections are in good agreement with the experimental data in the
region of high impact energies. Further development of this method is
foreseen by accounting for sequential mechanisms of two-electron transi-
tions and including electronic correlations in the wave functions of the
initial and final states. Such corrections are expected to improve the agree-
ment between theory and experiment at lower collision energies.
REFERENCES
1. Belkić, Dž. Principles of Quantum Scattering Theory; Taylor & Francis: London, 2003.
2. Fritsch, W.; Lin, C. D. The semiclassical close-coupling description of atomic collisions:
Recent developments and results. Phys. Rep. 1991, 202, 1.
3. Belkić, Dž.; Gayet, R.; Salin, A. Electron capture in high-energy ion-atom collisions.
Phys. Rep. 1979, 56, 279.Belkić, Dž. Leading distorted wave theories and computational
methods for fast ion-atom collisions. J. Comput. Meth. Sci. Eng. 2001, 1, 1.
4. Salin, A. Comments on ‘Strong potential Born expansions for ion-atom collisions’.
Comments At. Mol. Phys. 1991, 26, 1.Salin, A. Some remarks on theory of high
energy electron capture in ion-atom collision. J. Phys. B 1992, 25, L137.Mapleton, R.
Asymptotic form of the electron capture cross section in the second Born approxi-
mation. Proc. Phys. Soc. 1968, 91, 868.Dewangan, D. P.; Eichler, J. Boundary condi-
tions and the strong potential Born approximation for electron capture. J. Phys. B
1985, 91, L65.Macek, J. H. Treatment of divergent terms in the strong potential Born
approximation. J. Phys. B 1985, 18, L71.McGuire, J. H. Non-orthogonality in the strong
potential Born approximation. J. Phys. B 1985, 18, L75.
5. Belkič, Dž. Quantum Theory of High-Energy Ion-Atom Collisions; Taylor & Francis:
London, 2008.
7. Macek, J.; Alston, S. Theory of electron capture from a hydrogenlike ion by a bare ion.
Phys. Rev. A 1982, 26, 250.
8. Belkić, Dž.; Saini, S.; Taylor, H. S. Electron capture by protons from the K-shell of H
and Ar. Z. Phys. D 1986, 3, 59.
9. Belkić, Dž.; Saini, S.; Taylor, H. S. A critical test of first-order theories for electron trans-
fer in collisions between multi-charged ions and atomic hydrogen. Phys. Rev. A 1987,
36, 1601.
10. Belkić, Dž.; Taylor, H. S. First-order theory for charge exchange with correct bound-
ary conditions: general results for hydrogen-like and multi-electron targets. Phys. Rev. A
1987, 35, 1991.Belkić, Dž. Electron capture by fast protons from helium, nitrogen and
oxygen: the corrected first Born approximation. Phys. Rev. A 1988, 37, 55.
11. Merkuriev, S. P. On the three-body Coulomb scattering problem. Ann. Phys. 1980,
130, 395.
12. Mekuriev, S. P.; Faddeev, L. D. Quantum Scattering Theory for Several Particle Problems;
Kluwer Academic Publishing: Dortrecht, 1993.
13. Oppenheimer, J. R. On the quantum theory of the capture of electrons. Phys. Rev. 1928,
31, 349.Brinkman, H. C.; Kramers, H. A. Zur Theorie der Eifangung von Elektronen
durch α–Teilchen. Proc. Acad. Sci. Amsterdam 1930, 33, 973.
14. Jackson, J. D.; Schiff, H. Electron capture by protons passing through hydrogen. Phys.
Rev. 1953, 89, 359.Schiff, H. Electron capture cross sections. Canad. J. Phys. 1954, 32,
393.Jackson, J. D. On the use of the complete interaction Hamiltonian in atomic rear-
rangement collisions. Proc. Phys. Soc. A 1957, 70, 26.
15. Macek, J. H.; Shakeshaft, R. Second Born approximation with the Coulomb Green’s
function: electron capture from hydrogen-like ion by a bare ion. Phys. Rev. A 1980,
22, 1441.
16. Shakeshaft, R. Inner shell electron capture by a swift bare ion: second Born effects. Phys.
Rev. Lett. 1980, 44, 442.
17. Brodskii, A. M.; Potapov, V. S.; Tolmachev, V. V. Asymptotic behaviour of the cross sec-
tion for charge exchange. Sov. Phys. JETP 1970, 31, 144. [Zh. Eksper. Teor. Fiz. 1970,
58 264].
18. Faddeev, L. D. Scattering theory for a system of three particles. Sov. Phys. JETP 1961,
12, 1011 [Eh. Eksper. Teor. Fiz. 1960, 39 1459]; Faddeev, L. D. Mathematical Aspects of
the Three-Body Problem in the Quantum Scattering Theory; Israel Program of Scientific
Translations: Jerusalem, 1965.
19. Bates, D. R.; Dalgarno, A. Electron capture – I: resonance capture from hydrogen atoms
by fast protons. Proc. Phys. Soc. A 1952, 65, 919.Bates, D. R.; Dalgarno, A. Electron cap-
ture – III: capture into excited states in encounters between hydrogen atoms and fast
protons. Proc. Phys. Soc. A 1953, 66, 972.
20. Band,Y. Differential cross sections for electron capture in high energy proton-atom col-
lisions. Phys. Rev. Lett. 1976, 37, 634.
21. Halpern, A.; Low, J. Full first Born approximation for inner-shell pickup in heavy-ion
collisions. Phys. Rev. A 1975, 12, 776.
22. Bates, D. Electron capture in fast collisions. Proc. Roy. Soc. A 1958, 274, 294.
23. Lin, L.; Soong, S.; Tunnel, L. Two-state atomic expansion methods for electron capture
from multielectron atoms by fast protons. Phys. Rev. A 1978, 17, 1616.
24. Omidvar, K.; Golden, J.; McGuire, J.; Weaver, L. Single-electron Born approximation for
charge transfer from multielectron atom to protons. Phys. Rev. A 1976, 13, 500 [Erratum:
Phys. Rev A 1976, 13, 1604].
25. Halpern, A. Inner-shell charge transfer in a distorted-wave eikonal approximation using
the asymptotic interaction. Phys. Rev. A 1977, 15, 619.
26. Lin, C.; Soong, S. Differential cross section for electron capture in fast proton–multielectron
atom collisions. Phys. Rev. A 1978, 17, 499.
27. Greider, K. R.; Dodd, L. R. Divergence of the distorted-wave Born series for rearrange-
ment scattering. Phys. Rev. 1966, 146, 671.
28. Dodd, L. R.; Greider, K. R. Rigorous solution of three-body scattering process in the
distorted-waves formalism. Phys. Rev. 1966, 146, 675.
29. Aaron, R.; Amado, R. D.; Lee, B. W. Divergency of the Green’s function series for rear-
rangements collisions. Phys. Rev. 1961, 121, 319.
30. Goldberger, M. L.; Watson, K. M. Collision Theory; John Wiley & Sons: New York, 1964.
31. Lendyel,V. I.; Lazur,V.Yu.; Karbovanets, M. I.; Janev, R. K. Introduction to Atomic Collision
Theory; Lviv University, 1989 (in Russian).
32. Belkić, Dž. Second Born approximation for charge exchange with correct boundary
conditions. Europhys. Lett. 1988, 7, 323.Toshima, N.; Igarashi, A. Second Born approxi
mation differential cross sections for p+H and p+He charge-exchange collisions. Phys.
Rev. A 1992, 45, 6313.Toshima, N. Convergence of the perturbation series for high-
energy electron capture. J. Phys. B 1993, 26, L281.Shakeshaft, R.; Spruch, L. Mechanisms
for charge transfer (or capture any of any light particle) at asymptotically high impact
velocities. Rev. Mod. Phys. 1979, 51, 369.
33. Nordsieck, A. Reduction of an integral in the theory of bremsstrahlung. Phys. Rev. 1954,
93, 785.
34. Lewis, R. R. Potential scattering of high-energy electrons in second Born approxima-
tion. Phys. Rev. 1956, 102, 537.
36. Gayet, R. Charge exchange scattering amplitude to first order of a three body expan-
sion. J. Phys. B 1972, 5, 483.
37. Belkić, Dž.; Salin, A. Differential cross section for charge exchange at high energies.
J. Phys. B 1978, 11, 3905.Ho, T. S.; Eichler, J.; Lieber, M.; Chan, F. T. Calculation of the
differential cross section for electron capture in fast ion-atom collisions. Phys. Rev. A
1982, 25, 1456.
38. Karnakov, B. M. Asymptotic behavior of charge-transfer cross sections. Sov. Phys. JETP
1982, 55, 816 [Zh. Eksper. Teor. Fiz. 1982, 82 1407].
39. Thomas, L. H. On the capture of electrons by swiftly moving electrified particles. Proc.
Roy. Soc. A 1927, 114, 561.
40. Horsdal-Pedersen, E.; Cocke, C. L.; Stockli, M. Experimental observation of the

Thomas peak in high-velocity electron capture by protons from He. Phys. Rev. Lett.
1983, 50, 1910.
41. Vogt, H.; Schuch, R.; Justiniano, E.; Schulz, M.; Schwab, W. Experimental test of higher-
order electron-capture processes in collisions of fast protons with atomic hydrogen. Phys.
Rev. Lett. 1986, 57, 2256.
42. Martin, P. J.; Blankenship, D. M.; Kvale, T. J.; Redd, E.; Peacher, J. L.; Park, J. T. Electron
capture at very small scattering angles from atomic hydrogen by 25-125-keV protons.
Phys. Rev. A 1981, 23, 3357.
43. Clementi, E.; Roetti, C. Roothaan–Hartree–Fock atomic wave functions. Atom. Data
Nucl. Data Tables 1974, 14, 177.
44. Ghosh, M.; Mandal, C. R.; Mukherjee, S. C. Charge-transfer cross sections for asym-
metric collisions of protons with carbon, nitrogen, oxygen, neon and argon. Phys. Rev. A
1987, 35, 2815.Mandal, C. R.; Mandal, M.; Mukherjee, S. C. K-shell capture by He2+
and Li3+ on carbon and neon. Phys. Rev. A 1990, 42, 1803.
45. Belkić, Dž. Symmetric double charge exchange in fast collisions of bare nuclei with
heliumlike atomic systems. Phys. Rev. A 1993, 47, 189.Belkić, Dž. Two-electron capture
from helium-like atomic systems by completely stripped projectiles. J. Phys. B 1993,
26, 497.
46. Belkić, Dž. Importance of intermediate ionization continua for double charge exchange
at high energies. Phys. Rev. A. 1993, 47, 3824.
47. Belkić, Dž. Double charge exchange at high impact energies. Nucl. Inst. Meth. Phys. Res.
B 1994, 86, 62.Belkić, Dž.; Manćev, I.; Mudrinić, M. Two-electron capture from helium
by fast alpha particles. Phys. Rev. A 1994, 49, 3646.
48. Belkić, Dž.; Mančev, I. Formation of H− by double charge exchange in fast proton-
helium collisions. Phys. Scr. 1992, 45, 35.Belkić, Dž.; Mančev, I. Four-body CDW
approximation: dependence of prior and post total cross sections for double charge
exchange upon bound-state wave-functions. Phys. Scr. 1993, 46, 18.
49. Gayet, R.; Hanssen, J.; Martínez, A.; Rivarola, R. Double electron capture in ion-atom
collisions at high impact velocities. Nucl. Instr. Meth. Phys. Res. B 1994, 86, 158.Gayet, R.;
Hanssen, J.; Jacqui, L.; Martínez, A.; Rivarola, R. Double electron capture by fast bare
ions in helium atoms: production of singly and doubly excited states. Phys. Scr. 1996, 53,
549.
50. Sant’Anna, M. M.; Santos, A. C. F.; Coelho, L. F. S.; Jalbert, G.; de Castro Faria, N. V.;
Zappa, F.; Focke, P.; Belkić, Dž. Double-electron capture cross sections for Li3+ colli-
sions with He at intermediate-to-high velocities. Phys. Rev. A 2009, 80, 042707.
51. Belkić, Dž. Review of theories on double-electron capture in high-energy ion-atom
collisions. J. Math. Chem. 2010, 47, 1420.
CHAPTER FOURTEEN
Mechanistic Repair-Based Padé

Linear-Quadratic Model for Cell
Response to Radiation Damage
Dževad Belkić, and Karen Belkić
Contents
1. Introduction 408
2. Dose–effect curve (response curve or cell surviving curve) 410
2.1 Poisson distribution of radiation events, mean lethal dose 410
2.2 Extrapolation number and quasi-threshold dose 413
3. The linear-quadratic model 415
3.1 Biological effect, relative effectiveness, and biologically effective dose 415
3.2 The Barendsen formula 416
3.3 Low- and high-dose asymptotes of biological effect and surviving fraction 417
4. The Padé linear-quadratic model 419
4.1 Differentiation between physical and biological doses 419
4.2 Repair-mediated non-linear damping of linear direct cell kill mechanism 425
4.3 Initial slope, final slope, and extrapolation number 429
4.4 The Padé linear-quadratic model and the Michaelis–Menten kinetics 432
5. Results: comparison of radiobiological models with measurements 440
6. Discussion and conclusion 445
6.1 Biologically expressed response of the cell to irradiation 445
6.2 Dose–effect relationships at low, intermediate (shoulder), and high doses 446
6.3 Beyond the linear-quadratic model of cell inactivation 446
6.4 Mixed-order chemical kinetics for enzymatic cell repair systems 447
Acknowledgments 448
References 448
Abstract
This work is on a novel radiobiologal theory of cell survival after radiation of unspecified
modality or quality. The analyzed biophysical model, the Padé linear-quadratic model,
for cell surviving fractions and the related observable of clinical usefulness is equally
applicable to photon beams, as well as to light (electrons) and heavy (atomic nuclei)
charged particles of low, intermediate, or high energies. The presented formalism is
valid for both the single cell and entire cell populations. The analyzed description can

408 Dževad Belkić and Karen Belkić
be incorporated into any type of radiation delivery, including all fractionated treat-
ments ranging from the conventional (2 Gy per fraction, one fraction per day), through
hyperfractionation (more than one smaller fraction per day) to hypofractionation
(larger doses with shorter overall exposure time, as in stereotactic radiosurgery). The
major clinical application of the proposed theory is envisaged to be in providing a
better input to dose planning systems for radiotherapy, as expected from the clear
biological meaning of the derived parameters. The main strength of the Padé linear-
quadratic model is in its foundation on the mechanistic description of radiation dam-
age through enzymatic repair systems governed by the Michaelis–Menten catalysis.
It is from this latter origin, which passed the test of time in mainstream biochemistry,
that the present theory derives its biomedical adequacy. This, in turn, yields a power-
ful outperformance of the standard linear-quadratic model, the current workhorse of
radiobiological modeling for the purpose of radiotherapy, as amply illustrated in the
present study.
1. INTRODUCTION
Radiotherapy is multifaceted, since it relies upon interdisciplinary
research in order to meet with success, which is the patient’s cure. Different
radiation qualities or modalities (photons, charged particles, etc.), deposit
their energies in the traversed tissue according to different depth-dose
profiles. A typical Bragg peak is a remarkable example of such profiles for
heavy ions with most energy delivered to the encountered targets, mainly
near the very end the beam’s traversal. Electrons and photons deposit most
of their energy close to the entrance to the tissue. This makes these radia-
tions unsuitable for treating deep-seated tumors. In sharp contrast, high-
energy heavy ions can be optimally conformed to the target location deep
inside the patient’s body. Such a key feature is associated with negligibly
small multiple scattering effects of heavy ions due to large masses of atomic
nuclei relative to light electrons.
Nevertheless, irrespective of the existence of diametrically opposite
dose-depth profiles, all these different radiation modalities produce cell sur-
vival curves of a similar kind, characterized by typical sigmoid shapes, as a
function of the absorbed dose. Ionization density and linear energy transfer
are larger for heavy ions than for X-rays. This is expected to be biologically
expressed by the targeted cell in two distinct ways. Indeed, the so-called
relative biological effectiveness is typically 2–3 times larger for heavy ions
than that of X-rays. Yet, there are some other aspects of a less appreciable
variation for different radiation qualities. For example, the numbers of
double strand breaks of deoxyribo-nucleic acid (DNA) molecules could
Mechanistic Repair-Based Padé Linear-Quadratic Model 409
be close to each other for heavy ions and X-rays under the comparable
irradiation conditions.
It is of great importance for radiotherapy to have reliable predictions
on tumor control and healthy tissue complications. This is where radio-
biological modeling comes into its full function to assist tailoring dose
planning systems for individual treatments of patients in a manner which
is as comprehensive as possible.1, 2 Biophysical models can, in principle,
design dose distributions for each treated patient by taking into account
different biological factors ranging from anatomical to physiological. As to
the patient, the ultimate goal is to enhance survival and diminish toxicity
to the normal tissues. This aim could be attained through several strategies.
One of them is an improved understanding of the radiation-tissue inter-
actions on molecular, cellular, and tissue levels. Among other things, this
would yield the ever needed amendments of the current dose planning sys-
tems. Another strategy is a better comprehension of the individual patient
feedback from the administered dose, given that the same amount of the
identical irradiation could have markedly different outcomes for different
patients with the same type of cancer. These two strategies among others
should be considered in concert to achieve the best outcome.
There are two major variabilities in radiation-tissue interactions. One
is variability of dose in the irradiated volume. The other is variability of
cell response. Both variabilities are multifaceted, ranging from some self-
evident to more intricate, hidden aspects. Dose varies through the irradi-
ated tissue due to the stochastic nature of collisions between the beam
species and the targeted particles. This does not imply that dose variation
is completely random. Certain non-stochastic factors can also influence
dose variability, e.g., organ motion, some external settings, etc. Radiation
imparts damage to both normal and tumorous cells. Tumor topology
is highly complex due to intertwining of healthy with diseased tissue.
Critical to the variability of cell responses is the key difference in the way
normal and tumor cells cope with the same radiation insult. This variabil-
ity implies the existence of different interaction mechanisms of radiation
with these two kinds of cells. Radically different proliferation rates repre-
sent the main cause of unequal mechanisms for healthy and tumor cells.
The former have a controllable cell cycle, whereas the latter proliferate
uncontrollably with time changing rates.
A key to the overall success of radiotherapy is cell repair of the
imparted damage. Therefore, it is of critical importance to investigate vari-
ous repair mechanisms within the context of the mentioned variabilities
of interactions between the applied radiation and the targeted tissue. This
problem is addressed in the present chapter by reference to some of the
existing radiobiological models, such as the linear-quadratic (LQ) model3
as well as from the viewpoint of a recently introduced Padé linear-qua-
dratic model (PLQ).4–6
The most salient feature of the three constant parameters {α, β, γ }
from the PLQ model is their clear biological interpretation based
upon chemical kinetics of the Michaelis–Menten7 mechanism for cell
repair through enzyme-lesion catalysis. This automatically provides a cell
surviving fraction SF (D) of universal validity at all absorbed doses D.
Moreover, this biophysical model possesses the built-in correct asymp-
totic exponential inactivation modes at low and high doses, separated by
a shoulder. Further, the passage from the intermediate shoulder region to
both small and large doses occurs in a smooth manner through a typical
rectangular hyperbola for the dose-modifying factor—the relative effective-
ness RE(D) = {1 + (β/α)D}/(1 + γ D). The main significance of this is
to indicate that Barendsen’s concept8 of biologically effective dose (BED)
of radiation is not connected to the total absorbed dose D by a simple
relation BED = λD, with λ being a proportionality dimensionless constant
(dose-independent). Rather, the cell response, mediated by the enzymatic
repair of radiation damage, profoundly alters the physical dose D deposited
to the tissue through a modifying factor M (D), in the name of the relative
effectiveness M (D) = RE(D). This changes the said simple proportional-
ity relation to a more structured function BED(D) = D · M(D), which
becomes linear in D at both low and high doses, as indeed is typical for
most mammalian cells. Comparison with several representative sets of
experimental data for cell surviving fractions is presented to assess the rela-
tive performance of the PLQ model and to challenge the LQ model, which
is currently the workhorse of radiobiological modeling in radiotherapy.
2. DOSE–EFFECT CURVE (RESPONSE CURVE OR CELL

SURVIVING CURVE)
2.1 Poisson distribution of radiation events, mean lethal dose
Belonging to statistical phenomena, the distribution of events involving
cell-radiation interactions fluctuates following the Poisson probability. This
can be understood from the arguments which run as follows. On the one
hand, particle tracks traversing a tissue are certain to cross at least some of the
cell structures. On the other hand, randomness of radiation-cell interactions
implies that they are intrinsically uncertain, i.e., probabilistic. For example,
single and double strand breaks (SSB, DSB) or any other type of lesions
from interactions between the cell and radiation, can take place only with
some probability. It is precisely this lack of certainty, which classifies such
events as non-deterministic. Upon irradiation, the cell finds itself in a kind
of “all-or-nothing” state, as being either alive or dead with respect to its
clonogenic ability for proliferation. This naturally leads to the Bernoulli
statistics of binary (dichotomous) events. When the number of these events
is large, the Bernoulli distribution takes the form of the Poisson distribu-
tion. A p articular case of this passage from the Bernoulli to the Poisson
distribution is interesting to illustrate. In order to highlight this aspect, let
us consider some m independent hits that are delivered to the same target.
We could inquire about the probability π(m) that this target receives no hit.
Being independent, each hit has the same probability 1/m to arrive at the
same target. Conversely, the probability of missing this target is 1 − 1/m.
Furthermore, the probability that all the m hits will miss the target is
(1 − 1/m)m. This is precisely the sought probability π(m):
1 m

π(m) = 1 −
m
. (1)
In the limit of large values of m, it follows:
1 m

1
lim π(m) = lim 1 − = ≡ P(0). (2)
m→∞ m→∞ m e
This is a special case P(0) = e −1 of the more general Poisson law for the
distribution of a large number m of specific events†:
xm −x (3)
P(m) = e .
m!
Here, x is the average number of specific events. For m = 0 and x = 1, this

expression is reduced to:
1
P(0) = ≈ 0. 367879, or,
e (4)
P(0) reduced by 36. 7879% (e ≈ 2. 71828).
† The
word specific is used to refer to a particular kind of the cell-radiation interaction, e.g., single or
double ionization, excitation, etc.
The classical hit-target model for cell-radiation interaction assumes the

Poisson distribution of m events, such as creation of lesions whose aver-
age number x is supposed to be directly proportional to the absorbed
dose D:
D
x= , (5)
D0
and therefore,
m
1 D
P(m) = e −D/D0 . (6)
m! D0
In this model, the number of hits is equal to the number of events, m, and
moreover each hit is assumed to lead to a cell inactivation by producing a
lethal lesion (cell death). Thus, the probability of survival of a cell as a target
is the chance P(0) of not being hit, i.e., when no hit takes place at all, m = 0.
In the latter case, by setting x = 1:
D
x=1= , (7)
D0
it is possible to give the definition of the proportionality constant D0 in
(5) by reference to (3). Namely, since x = 1 corresponds to D = D0, as
per (7),
P(0) = e −D/D0 x=1,D=D = e −1 , (8)

0
we can say that D0 is a particular dose D, which yields, on the average,

one lethal event per target (x = 1). Due to this circumstance, D0 is usually
termed the mean lethal dose. Moreover, according to (4) and (8), quantity D0
is the dose at which the surviving fraction is reduced by 1/e ≈ 0. 367879,
or equivalently, by 36.7879% ≈ 37%. Due to this circumstance, D0 is often
called the “37% dose” and accordingly denoted by D37 (= D0 ). Reciprocal
1/D0 ≡ k0 is the measure of cell sensitivity to radiation and it is called either
the inactivation constant or the radiosensitivity constant (or radiosensitivity,
for short). By reference to the special name for dose D0, it is convenient to
refer to k0 as the mean lethal radiosensitivity, because it is also associated with
the 1/e reduction of the cell surviving fraction. Mathematically, 1/D0 is the
final slope of the terminal (exponential) part of the cell survival curve at
high doses, e −D/D0. Note that formally the same decay law or cell surviving
fraction applies to a single cell and to a cell population consisting initially

of some N0 cells. Here, the surviving fraction SF (D) would be defined by
N /N0, with the specification N /N0 = e −D/D0, where N is the number of
the surviving cells after irradiation by dose D. In this case, D0 would represent
the dose needed to deliver an average of one lethal event per cell in a total
population of N0 cells.
The curve for the cell surviving fraction SF (D) plotted as a function of
dose D is called the dose–effect curve or response curve. The latter name is
used to indicate that SF (D) describes the response of the cell to irradiation.
This response is the information about the number of cells that survived
by absorbing a dose D. Such curves have different characteristics at low,
intermediate, and high doses. They seem to decay exponentially at both
low and high doses. However, their initial and final slopes are different for
these two asymptotes when D → 0 and D → ∞, respectively. At relatively
lower doses there is a so-called shoulder. There are three other quantifying
characteristics of the dose–effect curves. These are the mean lethal dose D0,
quasi-threshold dose Dq and the extrapolation number n.
2.2 Extrapolation number and quasi-threshold dose

A shoulder in the response curve SF (D) is situated in the low-dose
region, where cell inactivation per unit dose is noticeably smaller than
that at high doses. The passage from these two regions of low and high
doses is anything but abrupt. Therefore, a transition dose for delinea-
tion of the precise extent of the shoulder cannot be sharply determined.
Nevertheless, an approximate procedure could still be designed to obtain
a reasonable estimate of a dose located near the transition point (thresh-
old) between the terminal part of the dose–effect curve and the shoulder
region. For this reason, such a dose is called a threshold dose, or more
appropriately, a quasi-threshold dose because of the said uncertainty.
An alternative name for the same dose is the shoulder width or the
quasi-width. Both names are associated with the symbol Dq. In order to
determine Dq, we introduce the high-dose exponential tail or asymptote
SF∞ (D) of SF (D) as:
SF∞ (D) ≡ e λ−D/D0 , (9)
or equivalently,
SF∞ (D) ≡ ne −D/D0 , (10)

where λ is a positive constant connected to the extrapolation number n by

the relation:
λ = ln n > 0, n > 1. (11)
When this definition is linked to a cell survival curve versus dose, the value
∞
Dq is seen as the dose located at a point at which the tangent to SF (D)
crosses the horizontal line parallel to the abscissa D at the height of the
ordinate reaching the maximal cell survival, SF (0) = 1:

Maximal surviving fraction D=0
(12)
= High dose asymptote of surviving fraction D=D .
q
In other words, Dq is the dose to which the terminating exponential region

of SF (D) is back-extrapolated to the 100% survival level. Relation (12)
represents the condition for isosurvival at which the same value of the cell
surviving fraction is obtained for two different doses D = 0 and D = Dq:
1 = e λ−Dq /D0

(Isosurviving fractions). (13)
or 1 = ne −Dq /D0
In this case, isosurvival, as the identical surviving fraction at both D = 0

and D = Dq is obtained if λ − Dq /D0 = 0, so that e λ−Dq /D0 = 1. Here, 1
(unity) is the maximal survival at D = 0, signifying the 100% survival, as
it ought to be for any form of SF (D). Therefore, the isosurvival condition
λ − Dq /D0 = 0 from (13), or equivalently:
D

ln 1 = 0 = λ − Dq0
D ,
ln 1 = 0 = ln n − Dq0 (14)
gives the shoulder quasi-width, or the quasi-threshold dose, Dq as:
Dq = D0 ln n. (15)
If a shoulder is viewed as an indication of the existence of accumulation of
sublethal damage, then, e.g., a wide shoulder width Dq would mean that the
amount of repaired sublethal damage is large.The name extrapolation num-
ber stems from “extrapolating” the terminal, high-dose asymptote ne −D/D0
of the survival curve SF (D) back to the zero dose (D = 0). The intercept
of the asymptote ne −D/D0 and the ordinate gives the extrapolation number,
{ne −D/D0 }D=0 = n.
3. THE LINEAR-QUADRATIC MODEL

3.1 Biological effect, relative effectiveness, and
biologically effective dose
The cell surviving fraction in the linear-quadratic model (LQ) is intro-
duced by:
(LQ) 2
SF (D) = e −αD−βD , (16)
or equivalently,
(LQ)
SF
(LQ)
(D) = e −EB (D)
, (17)
(LQ)
where EB is the biological effect (BE):
(LQ) (LQ)
EB ≡ − ln SF (D) = αD + βD 2 . (18)
This expression can be written in the following two alternative forms:
(LQ)
EB = αBED(LQ) , (19)
(LQ)
EB = ξ RE(LQ) , (20)
where ξ is the approximate expected number of lethal lesions:
ξ ≡ αD (Expected number of lethal lesions). (21)
The quantity BED(LQ) is the LQ-based biologically effective dose:

β 2
BED(LQ) = D + D , (22)
α
whereas the quantity RE(LQ) is the relative effectiveness (RE) from the LQ
model:
β
RE(LQ) ≡ 1 + D. (23)
α
The BED(LQ) and RE(LQ) are interconnected through the relationship:
BED(LQ) = D · RE(LQ) . (24)

Thus, the effect E(LQ)

B can also be understood as being given by the product
of the expected number of lesions αD and the relative effectiveness RE(LQ).
3.2 The Barendsen formula

The simple form (22) of BED(LQ) is attractive especially in fractionated
radiotherapy, because only the ratio β/α is employed, but the separate
values of α and β are not required. Based upon the LQ model, the ratios
α/β are estimated to be about 3 Gy and 10 Gy for healthy and tumorous
tissues, respectively. The specific numerical values for α/β can be fully
meaningful only if they are provided by a mechanistically-based radio-
biological model which is universally valid at all doses of interest. However,
the LQ model is a low-dose approximation. As a consequence, in order
to cover all the needed doses, one might be required to perform several
(say J ) separate fits to the given experimental data resulting in different
sets of values {αj /βj }( j J ) for the selected J dose intervals. This has the
drawback of yielding a dose-range dependence of the BED(LQ), which
severely limits inter-comparisons of different patterns of radiation deliv-
ery in fractionated radiotherapy. Therefore, it would be desirable to have
a model which would give the biologically effective dose applicable to all
doses for the same quantifying parameters estimated by using all the avail-
able experimental data points. Such a feature would enable the extraction
of the biologically effective dose from the reconstructed parameters that do
not change when passing from one to another dose range. The occurrence
that the mentioned ratios α/β are so different for tumorous and normal
tissue is an indication of the existence of substantially different mechanisms
by which these two types of tissues respond to irradiation. As stated in the
Introduction, the main reason for this difference is in the cell proliferation
which is uncontrolled (chaotic) in tumor, and well regulated by the cell
cycle growth in normal tissue.
What made the LQ model clinically useful was precisely Barendsen’s8
idea about linking the radiation dose with the ensuing biological effect.
In general, irrespective of any particular model, this concept states that the
biologically effective dose is equal to the product of the total dose D and a
dose-modifying factor, which is the relative effectiveness:

Biologically effective dose = {Total dose} · Modifying factor , (25)
BED = D · RE. (26)

It is through the modifying factor, the RE, that a particular pattern

of radiation delivery can be taken into account. This is especially impor-
tant for fractionated radiotherapy, where the RE can help improve the
effectiveness of the specific dose-time schedule by means of which the
total dose is administered. In searching for some plausible ways that could
make the LQ model clinically applicable, Barendsen8 entered only the
ratio β/α into the dose-dependent relative effectiveness RE(LQ), but not
the individual parameters α and β. Because the quotients of the form β/α
are tissue-specific parameters, quantity RE(LQ) = 1 + (β/α)D from (23)
can express the differences in radiation effects for various tissues. In other
words, RE(LQ) can be employed to investigate the differences in biologi-
cal effectiveness among various tissues as a consequence of alterations in
the dose delivery patterns. Thus, since different tissue effects are associated
with different values of the quotient β/α, it is possible to use Eq. (26) to
evaluate alterations in the therapeutic ratio due to changes in fractionation.
The therapeutic ratio can be improved by e.g., reducing dose per fraction,
in which case there should be proportionally more sparing of the healthy
(late reacting) than tumorous (early reacting) tissues. These considerations
indicate that the LQ model can be used as a predictive model, when
considering certain alternative radiation treatments aimed at maximiz-
ing tumorous cell kill effects, while simultaneously minimizing some of
the adverse healthy-tissue effects (e.g., normal tissue complication rates).
The Barendsen relation (26) is general, as it is not limited to fractionated
radiotherapy. Rather, it can be used for an arbitrary kind of radiation
treatment. Note that Barendsen8 originally coined a term “extrapolated
response dose” (ERD) in (26). This was subsequently renamed by Fowler9
to the “biologically effective dose,” or the BED, as a better terminology
for Barendsen’s idea.
3.3 Low- and high-dose asymptotes of biological

effect and surviving fraction
The effect (18) in the LQ model has the following asymptotic behavior at
infinitesimally small and infinitely large values of D:
(LQ)
EB −→ αD, (27)
D→0
(LQ)
EB −→ βD 2 , (28)
D→∞
respectively. This gives the corresponding asymptotes for BED(LQ) as:
BED(LQ) −→ D, (29)
D→0
β 2
BED(LQ) −→ D . (30)
D→∞ α
Using the asymptotes (29) and (30) for E(LQ)

B , it follows that the survival
(LQ)
curve given by SF (D) from (16) is mainly exponential and Gaussian at
small and large values of D:
SF
(LQ)
(D) −→ e −αD , (31)
D→0
(LQ) 2
SF (D) −→ e −βD , (32)
D→∞
respectively. A stronger repair, with a significant ratio β/α, is seen in the

plot of cell surviving fraction as a more pronounced shoulder due to the
Gaussian with its quadratic term βD 2, which yields a more curved func-
tion SF(LQ) (D). For small β/α, cell kill prevails and SF(LQ) (D) is less curvy
because of the dominance of the purely exponential function. As such, the
ratio β/α appears as a measure of the curvature of SF(LQ) (D) and this influ-
ences the cell response to radiation. Moreover, the smaller β/α implies that
the dose-response relationship will be less sensitive to fractionation when
fractionated radiotherapy is applied. Conversely, the larger β/α means that
radiation damage was accumulated to a sufficient level to produce various
lesions in DNA molecules. In such a case of an elevated β/α, the repairing
molecules are triggered more proactively, so that repair of repairable lesions
can become a key factor in determining the overall biological response of
the cell to radiation.
The number of lesions could be assumed to be proportional to D as
(LQ)
indicated in (21).Therefore, the LQ-based effect EB = (αD)(1 + βD/α)
from Eq. (18), interpreted as the yield of elementary lesions, is proportional
to the product of the average number of primary lesions (∼D) and the
average energy deposited around the lesions (∼ {1 + βD/α}). This can also
be written as:
(LQ)
EB = βD(ζ + D) = βD (zD + D) , (33)
where,
α
ζ ≡ zD = . (34)
β
Parameter ζ (or zD ≡ z̄D) is a microdosimetric quantity called the “dose-
averaged” specific energy, which is given in terms of a sequence of incre-
ments of the specific energy z associated with single events:
1 ∞
∞
zD ≡ ζ = dz z2 f1 (z), zF = dz zf1 (z), (35)
zF 0 0
where z̄F ≡ zF is the “frequency-averaged” specific energy.‡ The term “fre-
quency” refers to the frequency of occurrence of single events in a given
volume. Quantity f1 (z)dz is the probability distribution in z, where the
subscript “1” refers to single events alone. Event distributions appear in the
analysis because ionizing collisions are random and, therefore, the energy
deposited in the tissue by such collisions represents a stochastic quantity or
variable. In the microdosimetric formalism, the average number of events
N̄ at a fixed dose D is given by N̄ = ⌈D/zF ⌉, where ⌈r⌉ is the ceiling symbol
which denotes the largest integer obtained by rounding up the number r,
which can be a rational or any other real number or a real value of any
function.
4. THE PADÉ LINEAR-QUADRATIC MODEL

4.1 Differentiation between physical and biological doses
As mentioned, the two most salient aspects of the cell surviving fraction,
SF (D), are the direct cell kill and cell repair. They can be simultaneously
taken into account by introducing a biological dose DB to be determined
for the given physical, single radiation dose D. The sought dose DB can be
found from the Poisson statistics. In the context of radiation damage, the
targeted cell is certain to survive if it receives no dose when the dose DB
is expected to be absorbed. The chance for such an event to occur is given
by the Poisson probability:
P(0) = e −µDB , (36)
‡ I n microdosimetry, a single absorbed dose D is defined as the expected value of the so-called specific
energy z. On the other hand, specific energy z is the energy per unit mass per unit volume deposited
per event per cell nucleus.
where µ is the repair constant in units of Gy−1. The constant µ 0 is

connected to the repair time τ during which the cell becomes effectively
insensitive to any two consecutive hits (events, particle or ray traversals
through the treated tissue) whenever they follow each other within the
time interval t, which is smaller than the recovery time (�t < τ ). When
the repair mechanism is activated, the delivered physical dose D is reduced
and becomes only an apparent dose Dappar. On the other hand, the true
dose Dtrue, which is actually received by the cell, represents the biological
dose DB. In other words, repair effectively diminishes the values of D and
transforms it to DB. The difference between D and DB is that the latter
accounts for a correction due to the missed/wasted hits during the time
lag �t < τ whenever τ > 0. Such a discrepancy between D and DB can be
modeled by the said Poisson probability P(0). This settles the issue of the
definition of DB = DB (D) for a fixed D as:
DB P(0) = D, (37)
or explicitly,
DB e −µDB = D, (38)
where,
D DB . (39)
Employing the alternative notation:
Dappar ≡ D, Dtrue ≡ DB , (40)
we can rewrite (38) as:
Dtrue e −µDtrue = Dappar , (41)
Dappar Dtrue . (42)
It will also prove useful to introduce a repair degree by the following quo-
tient of doses:
Dappar D
ν≡ = , 0ν1 (Cell repair measure or degree). (43)
Dtrue DB
The mechanism driving the pattern (38), or equivalently (41), which

describes the cell recovery during the repair time τ consists of the following
twofold pattern:
(a) What is usually considered to be a single absorbed dose D is modified

by repair to become merely an apparent dose Dappar, which is smaller
than the true dose Dtrue, which is expected to be deposited to the sensi-
tive part of the cell.
(b) A measure or degree of the cell repair is the quotient ν of Dappar and
Dtrue from (43) as given by the Poisson probability (38) that the cell
receives no dose when the dose Dtrue = DB is anticipated. This degree
varies from 0 to 1 due to its coincidence with the Poisson probability
(36) via ν = P(0) = e −µDtrue = e −µDB.
The inequality in (39) is evident from e.g., (41) where Dtrue has to
be exponentially damped via Dtrue e −µDtrue to become equal to Dappar as
Dtrue e −µDtrue = Dappar. Stated equivalently, the apparent dose Dappar = D
is smaller than its true counterpart Dtrue = DB because D ought to be
enhanced by a positive factor 1/P(0) = e µDB in order to match the value
DB through:
De µDB = DB . (44)
Parameter µ could be related to the microdosimetric dose-averaged specific
energy zD = ζ from (34). In microdosimetric formalism, the probability of
receiving no dose when the dose DB is expected is given by P(0) = e −DB /zD.
This is the same Poisson formula which is in the formalism of this Section
denoted by P(0) = e −µDB in (36). Therefore, the parameter µ from (36) or
(38) could have a microdosimetric meaning, in which case it would repre-
sent the reciprocal of the dose average specific energy:
1 1
µ= ≡ . (45)
zD ζ
With this relation, a link to the parameters α and β from the LQ model can
readily be made. Parameter zD, which appears in the second moment of the
dose-averaged specific energy has already been identified in dosimetry as
zD = α/β and, therefore:
β
µ= . (46)
α
It is important to formulate the following inverse dose problem for radio-
biological modeling:

Given the physical dose D applied to the treated tissue, 
what would be the biological dose DB received by . (47)
the irradiated cells when the cell repair system is active?

Within the Poisson statistics (36), one of the possible answers to this ques-
tion is given by the exact real-valued solution DB of the transcendental
equation (38).
If DB were known, the dose which was applied to the tissue could
be retrieved by a direct computation of the so-called Ricker function,
DB e −µDB. Then the inverse Ricker function would give D. In reality,
however, the biological dose DB is unknown, but could nevertheless be
determined by finding the inverse Ricker function. Although the exact
inverse Ricker function is known and can be used through the Lambert
function,4, 10 in the present work we shall deal with a simpler and more
instructive exposition. To this end, we shall derive an approximate solu-
tion for DB from the non-linear, transcendental equation (38) through
the process of linearization, by using only the known simplest elementary
functions. With this goal, we start from (44) where we use the series for the
exponential e µDB as:
µDB (µDB )2 (48)

e µDB = 1 + + + ··· ,
1! 2!
where the rhs converges for every value of µDB. A further simplification of
the rhs of Eq. (45) can be deduced by assuming that the recovery time is
short (τ ≪ 1 or µ ≪ 1) and that the dose D permits the relation µDB ≪ 1.
In such a case, it is justified to retain e.g., only the first 2 terms of the series
from (48), so that (44) becomes:
DB = De µDB ≈ D(1 + µDB ) (µDB small). (49)
This implicit, linearized version of the non-linear Eq. (38) can be written
more explicitly by collecting the unknown DB on the same side of the
equation to yield:
D
DB ≈ (µDB small). (50)
1 − µD
The expression D/(1 − µD) from (50) is the diagonal Padé approximant
(PA)11 to DB (D), as a quotient of two polynomials of the same first degree
in variable D. This PA possesses an equivalent representation obtained by
employing the binomial series:
D
DB ≈ = D{(1 + µD) + (µD)2 + · · · }, (51)
1 − µD
which converges for µD < 1. If we keep only the first two terms within the
square brackets from the rhs of Eq. (51), we will finally arrive at:
DB ≈ D(1 + µD) (µD small), (52)
or equivalently, by reference to (39):
Dtrue = Dappar (1 + µDappar ) (µDappar small). (53)
As discussed for the exact relation (38), we see that the corresponding
approximation (52) also obeys the inequality D DB from (39), since
a non-negative term (µD 2 0) must be added to D 0 to obtain an
approximate DB. Division of (52) by D or (53) by Dappar yields:
DB
≈ 1 + µD, (54)
D
so that,
1
ν= . (55)
1 + µD
Using (46), the rhs of Eq. (54) can be identified as the relative effectiveness
RE(LQ) from Eq. (23). Thus, the quantity RE(LQ) can equivalently be con-
ceived as the quotient DB /D of an approximate expression for the biologi-
cal dose Dtrue (or DB) and the apparent dose Dappar (or D):
DB
RE(LQ) = . (56)
D
The approximate formula for the effect EB related to (52) is now given by:
EB ≈ αDB ≈ αD + βD 2 . (57)
Therefore, the definition (46) permits a connection of Eqs. (52) and (57)
with the LQ model via:
(LQ) (LQ)
DB ≈ DB , EB ≈ EB , (58)
(LQ)
where EB is defined in (18), and:

(LQ) β
DB =D 1+ D . (59)
α
The binomial series (51) for DB can alternatively be obtained by solving

the implicit, linearized equation (49) through iterations. The first iter-
ate is generated via the replacement of DB from the rhs of Eq. (49) by
D(1 + µDB ) so that DB ≈ D {1 + µ [D(1 + µDB )]}. If in the rhs of the
latter equation, DB is written as D(1 + µDB ), the second iterate follows,
DB ≈ D(1 + µD + µ2 D 2 + µ3 DB ). The higher iterates within this pro-
cedure of self-substitutions shall yield the binomial series, which sums up
to closed expression DB ≈ D/(1 − µD) for µD < 1 in agreement with
Eq. (51).
The approximate solution (52) for DB is the sum of the linear (∼D) and
quadratic (∼D 2 ) terms. Without repair, all the impinging radiation quanta
are absorbed by the targeted cell, so that:
DB = D, µ = 0 (τ = 0 : no repair), (60)
and this goes for both (38) and (59). However, with repair (τ > 0, µ > 0),
there would be some wasted radiation quanta, as if they were removed from
the beam or annihilated in the traversed tissue and this gives the quadratic
term ∼D 2. Our derivation shows that the quadratic term ∼D 2 is directly
rooted in the cell repair mechanism and brought about by reconstruc-
tion of the unknown biological dose DB ≈ DB(LQ) from the given physical
dose D. This gives the approximate answer (52) to the stated inverse dose
problem (47). By reference to (59), it follows that the quantity BED(LQ)
coincides with DB(LQ):
(LQ)
BED(LQ) = DB . (61)
In the present formulation of the LQ model, parameters α and β are
associated in a distinct manner with the cell kill and cell repair per Gy−1
and Gy−2, respectively. However, in the present derivation of the approxi-
mate effect E(LQ)
B from (18), these two parameters are correlated through
the relationship (46).Thus, if µ were known, only α would be a free param-
eter when reconstructing E(LQ)
B
from e.g., its least-square adjustment to the
corresponding experimental data. Parameter µ = β/α is the measure of
the deviation of the parabola αD + βD 2 from the straight line αD plotted
versus D as the abscissa. The higher the µ, the more parabolic the effect
(LQ)
EB and the more significant the repair βD 2. Conversely, the lower the
µ, the more straight line behavior of E(LQ)
B and the more pronounced the
cell kill αD.
This derivation clearly demonstrates that the LQ model represents
a low-dose approximation of a more general model,4 which solves the
transcendental equation (38) exactly, rather than using the approximate

solution (54), which is valid for small µD. More specifically, we employed
the convergence radius µD < 1 in the result DB /D ≈ 1 + µD from (54).
This latter low-dose restriction is due to the use of the binomial series
for 1/(1 − µD), which is meaningful only for µD < 1. Retaining the first
two terms in this series via 1/(1 − µD) ≈ 1 + µD leads straight to the
LQ model. The binomial 1/(1 − µD) itself is the Padé approximant as a
ratio of the simplest two polynomials of degree 0 and 1 in the numerator
and denominator, respectively, according to 1/(1 − µD) = P0 (D)/Q1 (D),
where P0 (D) = 1 and Q1 (D) = 1 − µD.
4.2 Repair-mediated non-linear damping of linear direct

cell kill mechanism
At larger values of D, experimental data for SF (D) usually exhibit an expo-
nential fall-off, SF (D) ∼ e −D/D0 = P(0), where D0 is the dose at which
the surviving fraction is reduced by 1/e ≈ 0. 367879 or by 36. 7879%,
as per (4). In other words, the final slope of most measured curves
for cell surviving fractions is given by 1/D0. This is opposed to the
LQ-type high-dose dominance of the Gaussian SF(LQ) (D) ∼ e −βD
2
from (32), which continues to bend and thus has no final slope. In
the LQ model, the initial slope, determined by the low-dose asymp-
tote SF(LQ) (D) ∼ e −αD from (31), is given by α which is associated with
single radiation events (single hits) to a sensitive part of the cell. Even in
the low-dose limit, the LQ model was seen to deviate from experimen-
tal data12 thus pointing to unreliable numerical values of the parameters
α and β in the LQ-based cell surviving fraction SF(LQ) (D). According
to the above derivation, α is also present in β. The said drawbacks of
the LQ model at both low- and high-dose asymptotic regions could
partially be attributed to the assumption that the linear part ξ = αD
is associated exclusively with lethal events as in the classical hit-target
model. We shall relax this limitation and modify the term αD so as to
allow for cell repair. In other words, as opposed to the hit-target model,
where the direct hits (∼αD) describe irreparable lesions that cause cell
death, we shall permit that even direct hits could be repaired. As dis-
cussed, this can be done by damping αD when the dose is progressively
augmented. To this end, we shall modify the LQ model by introducing
the repairable lesion ξB in lieu of ξ as:
ξB e ωξB ≡ ξ ∴ ξB � ξ , (62)
where ω 0 is a dimensionless repair-related constant. Here,

ξB = Expected number of repairable lesions from direct hits
, (63)
ξ = Expected number of irreparable lesions from direct hits
The inequality ξB ξ from (62) is evident from the defining relation

ξB e ωξB = ξ, since ξB needs to be multiplied by a non-negative number
e ωξB 0 to be equalized to ξ. Moreover, biologically, the plus sign of ξB
in the argument of the exponential in ξB e ωξB from the lhs of Eq. (62)
coheres with the fact that repair diminishes the number of expected lethal
lesions from direct hits (single interaction of radiation with the cell). At
this point of the analysis, it suffices to find an approximate solution of the
transcendental equation (62). This can be done if in Eq. (62), rewritten as
ξB = ξ e −ωξB, we replace the exponential e −ωξB by its first-order diagonal
Padé approximant11 in variable ωξB /2 as:
1 − ωξB /2
e −ωξB ≈ (Padé approximant for exponential). (64)
1 + ωξB /2
This transforms the transcendental equation (62) into a quadratic equation
for the unknown ξB:
1 2
ωξ + QξB − ξ = 0, (65)
2 B
where,
1
Q = 1 + ωξ = 1 + γ D, (66)
2
1
γ = ωα. (67)
2
The roots of the quadratic equation (66) are:

Q 2ωξ
ξB± = −1 ± 1 + 2 , (68)
ω Q
where ξB+ > 0 (physical) and ξB− < 0 (unphysical). We retain only the
+ (P)
positive-definite root ξB , which is re-labeled as ξB :
(P)
ξB+ ≡ ξB . (69)
Next, the term (1 + 2ωξ/Q 2 )1/2 is expanded in a series with powers of

2ωξ/Q 2:
2ωξ 1/2 3 (ωξ )2

ωξ
1+ 2 = 1+ 2 + + ··· . (70)
Q Q 8 Q4
By keeping solely the first two terms, i.e., the terms from the curly brackets,
and using (66), it follows:
2ωξ 1/2

ωξ
1+ 2 ≈ 1 + 2. (71)
Q Q
(P)
In this way, Eqs. (68) and (69) yield the final result ξB ≈ ξB ≈ ξ/Q or:
(P) ξ αD (72)
ξB ≈ ξB ≈ = .
1 + γD 1 + γD
(P)
The replacement of the term ξ by its Padé-equivalent ξB yields the Padé
Linear-Quadratic model, as denoted by PLQ, for the biological effect of
radiation:
(PLQ) (P)
EB ≡ ξB (1 + µD), (73)
or equivalently,
(PLQ) αD
EB = (1 + µD). (74)
1 + γD
The corresponding cell surviving fraction in the PLQ model reads as:
(PLQ)
SF
(PLQ)
(D) ≡ e −EB . (75)
If parameter µ is chosen according to (46), we can cast Eqs. (74) and (75)
into the forms:
(PLQ) αD + βD 2
EB = , (76)
1 + γD
2
(PLQ) − αD+βD
SF (D) = e 1+γ D . (77)
The numerator of the quotient from the rhs of Eq. (76) represents the effect
in the LQ model, so that we can also write:
(LQ)
(PLQ) EB (78)
EB = .
1 + γD
As per derivation of the PLQ model, there are two “repair ratios” Ŵβα and
Ŵγ α that can be introduced by:
β γ
Ŵβα = , Ŵγ α = , (79)
α α
in terms of which, the biological effect (76) can be rewritten as:
D + Ŵβα D 2 1
EB
(PLQ)
= , δ= . (80)
δ + Ŵγ α D α
Quotient Ŵβα is a repair degree β/α, which gives the relative impor-
tance of the linear (D) and quadratic (D 2 ), as also encountered in the LQ
model, according to (59) and (61). The additional repair ratio Ŵγ α in the
PLQ model is a repair measure γ /α of the strength of some higher-order
mechanisms appearing through all the powers of dose D that are implicitly
present in (80) and could be made explicit by expanding the binomial
1/(δ + Ŵγ α D) into its Macularin series. The quotient of Ŵβα and Ŵγ α is
useful, since it gives the degree of the departure of the PLQ from the LQ
model:
Ŵβα β
Ŵβγ = = . (81)
Ŵγ α γ
This quantity is also the final slope of the dose–effect curve SF(PLQ) (D),
as will be discussed later on. The expression (73) for the effect E(PLQ) B
in the PLQ model was obtained in the two main steps: (i) derivation of
the approximate biological dose DB = D(1 + µD) from (52), which by
way of the definition µ = β/α from (46) coincides with the biologically
effective dose BED(LQ) in the LQ model, DB = BED(LQ), as per (61), and
(ii) replacement of the lethal (irreparable) lesions ξ = αD in the effect
(P)
EB ≡ αDB = ξ(1 + µD) from (61) by the sublethal (repairable) lesion ξB
via ξ =⇒ ξB(P) = ξ/(1 + γ D) = (αD)/(1 + γ D).The net result of the steps
(P)
(i) and (ii), through the product of ξB and the relative effectiveness 1 + µD
(P)
as ξB (1 + µD) represents the effect E(PLQ)B = (αD + βD 2 )/(1 + γ D)
from (73) in the PLQ model. This recapitulation through the said two steps
(i) and (ii) illustrates the origin of the name “Padé + Linear-Quadratic” and
the associated acronym PLQ for this new radio-biological model which has
originally been introduced in our recent works.4–6
4.3 Initial slope, final slope, and extrapolation number

For the choice µ = β/α from (46), the low- and high-dose asymptotes of
(PLQ)
EB read as:
(PLQ)
EB −→ αD, (82)
D→0
(PLQ) β
EB −→ D, (83)
D →∞ γ
respectively. For brevity, the high-dose asymptote (83) is written to exhibit

only the leading term (∼D), whereas the constant (∼D 0 ) is ignored. As it
stands, Eq. (76) is the para-diagonal Padé approximant with the numerator
and denominator polynomial of the second- and first-degree, respectively
in variable D.
The expression for E(PLQ)
B from (76) leads to the corresponding biologi-
cally effective dose BED (PLQ) in the PLQ model:
(PLQ)
(PLQ) E D + βD 2 /α (84)
BED ≡ B = .
α 1 + γD
This can also be written in analogy with (24) as:
BED(PLQ) = D · RE(PLQ) , (85)

where RE(PLQ) is the relative effectiveness in the PLQ model:
1 + (β/α)D RE(LQ)
RE(PLQ) = = . (86)
1 + γD 1 + γD
Insertion of the asymptotes (82) and (83) for E(PLQ)

B into Eq. (84) yields:
BED(PLQ) −→ D, (87)
D →0
β
BED(PLQ) −→ D.
D→∞ αγ (88)
As expected, the PLQ and LQ models exhibit the same low-dose behav-
iors in (29) and (87), but differ substantially at high doses according to (30)
and (88).
The behaviors (82) and (83) of E(PLQ)
B yield the following two asymp-
totes of SF(PLQ) (D) at small and large values of D:
SF
(PLQ)
(D) −→ e −αD , (89)
D →0
(PLQ)
SF (D) −→ e −βD/γ , (90)
D→∞
respectively. This gives the initial and final slopes si and sf , respectively, in
the dose–effect curve from the PLQ model as:
β
PLQ : Initial slope : si = α, Final slope : sf = . (91)
γ
In the high-dose asymptotes (90), only the leading term βD 2 is retained in
the numerator of the biological effect EB(PLQ) = (αD + βD 2 )/(1 + γ D).
However, it is also useful to extrapolate the high-dose limit of the cell surviv-
ing curve back to the ordinate axis (D = 0). This would give the so-called
extrapolation number n. Thus, alongside the same high-dose approximation
for the denominator 1 + γ D ≈ γ D, which has already been made in (90), we
shall now retain the full numerator αD + βD 2 in (αD + βD 2 )/(1 + γ D)
to arrive at:
αD + βD 2

(PLQ) β β
− ln SF (D) −→ − + D
D→∞ 1 + γD γ γ

(αγ − β)D β
= + D
γ (1 + γ D) γ

(αγ − β)D β
−→ 2
+ D
D→∞ γ D γ

αγ − β β
= 2
+ D
γ γ
so that,
β−αγ
(PLQ) − γβ D
SF (D) −→ e γ2 .
D→∞ (92)
This can conveniently be rewritten as:

− γβ D

(PLQ)
SF (D) −→ ne 
(PLQ)
D→∞ , (93)
ln SF (D) −→ ln n − γβ D 
D→∞
where the extrapolation number n is given by:
β − αγ
ln n =
γ2 (94)
�sfi
= , �sfi = sf − si .
γ
Thus, the extrapolation number is proportional to the difference sfi

between the final and initial slopes, ln n ∼ sf − si = sfi. The extrapolation
number n must be positive and this imposes the following condition:
ln n > 0 if β > αγ . (95)
At high doses, it might be useful to constrain the free parameter γ to the
relationship:
γ = βD0 , (96)
in which case (90) can alternatively be written as:
(PLQ)
SB (D) −→ e −D/D0 at γ = βD0 . (97)
D→∞
With the selection (96), the extrapolation number n from (94) becomes:
1 − αD0
ln n = at γ = βD0 , (98)
βD02
where the restriction condition (95) is now specified as:
1
ln n > 0 if α < at γ = βD0 . (99)
D0
This positivity requirement (99) for the extrapolation number n > 0
reflects the proper relationship between the initial (α) and final (1/D0 )
slopes when the third parameter γ in the PLQ model is not adjustable, but
rather fixed via γ = βD0:
Initial slope (α) < Final slope (1/D0 ) at γ = βD0 . (100)

We re-emphasize that when γ is constrained to the relation γ = βD0,

the high-dose limit of the PLQ model becomes SF(PLQ) (D) ∼ e −D/D0,
as required by the experimental data. Of course, any constraint imposed
on one parameter inevitably introduces a bias into the estimates of the
remaining parameters. However, irrespective of whether γ is pre-assigned
to be of the form γ = βD0 or kept free, the other two parameters α and
β are always mutually dependent, since by definition β = µα. Moreover,
in view of (67) instead of γ , we could use ωα/2, where ω takes the role
of an adjustable parameter. This shows that in the general case without
resorting to (96), the third parameter ωα/2 in the PLQ model is connected
to the direct cell kill component α. Overall, the unconstrained version of
the PLQ model is negligibly more involved than the LQ model from the
computational viewpoint due to the presence of merely one additional
parameter γ or equivalently, ωα/2. The constrained variant of the PLQ
model, with γ fixed by the prescription γ = βD0, has only two parameters
α and β, since D0 can be considered as the input data to be read off from
the final slope of the experimentally measured cell surviving fraction at
larger values of D. In either case, the advantage of the PLQ over the LQ
model is at least twofold:
• (A) a richer mathematical function with the underlying mechanisms

and,
• (B) a smooth switch from the incorrect quadratic (Gaussian) to the cor-
rect (exponential) asymptote at high doses, as required by the measure-
ments.
4.4 The Padé linear-quadratic model and the Michaelis–

Menten kinetics
Inserting (18) for E(LQ)
B into the numerator of E(PLQ)
B from the Padé quo-
tient (78), we can alternatively write the biological effect in the PLQ model
as:
EB
(PLQ)
= v0 (ζ + D) = v0 (zD + D), (101)
where,
βD
v0 = . (102)
1 + γD
In (101), use is made of the quantity ζ, or equivalently, zD from (34), where

zD is the dose-averaged specific energy from microdosimetry. T he quantity
v0 is velocity or rate, which is equivalent to the initial reaction velocity
from the enzyme kinetics of Michaelis–Menten (MM)7 for the follow-
ing irreversible chemical reaction (enzyme catalysis) with formation and
destruction of the enzyme-lesion complex:
[E] + [L] −→[EL] −→[E] + [R].
k1 k2 (103)
Here [E], [L], [EL], and [R] are the concentrations of the free enzyme
molecules, lesions (primarily DNA), enzyme-lesion complex and repaired
lesions, respectively. Quantities k1 and k2 are the rate coefficients for forma-
tion and destruction/dissociation of the intermediate complex molecule
[EL]. In this chemical reaction, the free enzyme molecule [E] binds the
radiation damaged DNA molecule (a lesion [L]) into an intermediate
and temporarily living unstable complex molecule [EL]. This compound
facilitates the enzymatic synthesis of DNA. After completion of this inter-
mediate stage of the reaction, the complex [EL] decays, thus producing the
repaired lesions [R] and enzymes [E] that are again free for further bindings
with other lesions. As mentioned earlier in (21), the number of lesions is
usually assumed to be proportional to dose D:
[L] ∼ D = κD ≡ D, (104)
where κ could be taken as κ = k0 ≡ 1/D0 or as a constant of unit abso-
lute value (κ ≡ 1). When writing [L] = D in (104) and afterwards, it is
understood that [L] = D ≡ 1 · D, where “1” takes care of the proper units
in the passage from a dose to a molar concentration. This convention is
done to avoid introducing a superfluous parameter only for the purpose of
conversion of the units. In this way, Eq. (102) can equivalently be written as:
β[L]
v0 = . (105)
1 + γ [L]
Using Eq. (104) and the definition:
1
KM ≡ , (106)
γ
we can cast expression (105) into its form used in the MM kinetics:
vmax [L]
v0 = ,
KM + [L] (107)
where vmax is the maximal velocity given by:

β
vmax ≡ . (108)
γ
Velocities v0, and consequently, vmax are given in units Gy−1. On the other
hand, in the original Michaelis–Menten velocity for irreversible reaction
(103), we have:
k2
KM = , (109)
k1
vmax = k2 [E]0 , (110)
where [E]0 is the initial concentration of enzymes (the number of enzymes
at the onset of the reaction), which is assumed to be constant throughout
and, therefore, equal to the total enzyme concentration [E] ≡ [E]tot at any
subsequent time. Quantity KM is the Michaelis–Menten7 constant with the
dimension of concentration. More specifically, this constant for irreversible
variant (103) of the general MM enzyme catalysis, represents the concen-
tration of lesions ([L] ≈ KM ) at which velocity v0 attains one half of vmax,
as follows {v0 }[L]≈KM ≈ vmax [L]/([L] + [L]) = vmax /2:
vmax
v0 ≈ at [L] ≈ KM . (111)
2
In the same approximation, the biological effect E(PLQ)
B becomes:
(PLQ) vmax + α
EB ≈ KM at [L] ≈ KM . (112)
2
At small concentration of lesions, the reaction velocity v0 is reduced to:
vmax [L]
v0 = ≈ vmin , (113)
KM (1 + [L]/KM ) [L]/KM ≪1
where,
vmax [L]
vmin = . (114)
KM
The use of Eqs. 106 and (108)–(110) permits connecting e.g., the final
slope (β/γ ) in the PLQ model with the enzyme kinetic parameters as:
β
Final slope : Ŵβγ = = vmax = k2 [E]0 . (115)
γ
A similar connection can also be deduced for the extrapolation number using
Eq.(94) as follows: ln n = (β − αγ )/γ 2 = (β/γ − α)/γ = (vmax − α)KM ,
so that:
Logarithm of the extrapolation number : ln n = (vmax − α)KM . (116)
Hence, the larger (the smaller) the product of the enzyme concentration
[E]0 and the dissociation rate constant k2, the steeper (the shallower) the
final slope of the dose–effect curve. This is correct, since more significant
values of k2 [E]0 would enhance the chance for a greater enzymatic activ-
ity with the ensuing larger concentration [R] of repaired lesions. Such
an outcome should mitigate the influence of the direct cell kill mecha-
nism (αD) and, therefore, would increase cell survival, as manifested by
a departure from the pure exponential bending e−αD in the dose–effect
curve. Consequently, a shouldered cell surviving curve appears, as a sig-
nature of the enzymatic repair of radiation-induced lesions. This is also
reflected in the repair ratio Ŵβα = β/α from (79), which can be rewritten
in the form:
β
Ŵβα = = Ŵγ α {k2 [E]0 },
α (117)
where the second repair ratio Ŵγ α is equal to γ /α, according to (79). The
larger values of β/α imply a more noticeable influence of repair. Expression
(117) confirms this expectation through a direct proportionality between
the quotient β/α and the enzyme concentration [E]0 available for repair.
Moreover, the same ratio β/α is also directly proportional to the catalysis
rate k2 and, thus, to the efficiency of the enzymatic repair system in con-
verting the radiation damage [L] to the repaired lesions [R] in the final
reaction path [EL] −→[E] + [R], so that:
k2
Ŵβα ∼ k2 [E]0 . (118)
In the equivalence relation (118), the proportionality constant is Ŵγ α, where

Ŵγ α = γ /α from (117). This constant is associated with linear and non-
linear contributions from all the powers that are inherent in the binomial
(δ + Ŵγ α D)−1, as mentioned earlier.
Rewriting (104) as E(PLQ)
B = (αD + βD 2 )/(1 + γ D) = (β/γ )(α/β + D)
× (1/γ + D) and using (106) and (108) for 1/γ = KM and β/γ = vmax,
respectively, we can equivalently express the biological effect (76) and the
cell surviving fraction (77) in the Michaelis–Menten representation as:
(PLQ)
EB = v0 (zD + [L])
(119)

vmax [L]
= (zD + [L]) ,
KM + [L]
and,
(PLQ)
SF (D) = e −v0 (zD +[L])

vmax [L]
(120)
− KM +[L] (zD +[L])
=e .
The connection among the three parameters {α, β, γ } with the equivalent
triple {zD , vmax , KM } from the PLQ model is summarized as:
zD = βα ; Direct/Indirect mechanisms , (Cell kill)/(Cell repair) 



vmax = γβ ; Experimentally measurable maximal enzyme velocity .
KM = γ1 ; Experimentally measurable Michaelis–Menten constant

(121)
Overall, the MM rate constant KM and the maximal enzyme velocity vmax
can be measured in standard enzyme experiments. Moreover, the initial (α)
and final (vmax ) slopes can be extracted from the experimental data for the
given cell surviving curve and so could the extrapolation number n. Also,
as soon as the triple {α, vmax , KM } becomes available, the extrapolation
number can be obtained from Eq. (116). Thus, the basic elements of the
dose–effect curve in the PLQ model are recapitulated via:

si = α; Initial slope 
sf = vmax ; Final slope .
ln n = (vmax − α)KM ; logarithm of the extrapolation number

(122)
When parameter is pre-assigned as γ = βD0, by reference to (96), then
γ
according to (100) the final slope β/γ becomes 1/D0, so that:
1 1
D0 = = at γ = βD0 . (123)
vmax k2 [E]0
This result (D0 ∼ 1/k2 ) shows that the mean lethal dose D0 is proportional
to the reciprocal of rate constant k2 for enzyme-mediated creation of a
repaired lesion [R], which is the product substance in reaction (103) for
enzyme catalysis.
At high doses, the enzyme-lesion reaction velocity v0 tends to the
constant value vmax, which is the fastest rate possible for the given
enzyme concentration [E]0. This means that the rectangular hyperbola
v0 = αD/(1 + γ D) from (102) as a function of D has reached a plateau at
larger doses. Stated equivalently, the curve v0 = vmax [L]/(KM + [L]) from
(107), as a function of [L] is leveled off for higher lesion concentrations,
[L]. Such a high-dose or a high concentration of lesions described by the
reaction velocity v0 in terms of the independent variable D or [L] is due
to the limited amount of enzymes (∼ 100 enzyme molecules per lesion)
that are available for repair of radiation-damaged cells. At high doses, the
average number of lesions is sufficiently large to overwhelm and thus inac-
tivate the enzyme repair system, after which point every radiation damage
is essentially lethal. This saturation of enzymes by lesions is the signature for
a switch from the cell repair to the cell kill mechanism corresponding to
the passage from the second- to the first-order Michaelis–Menten kinetics.
Expressions (119) and (120) are written in a way which separates
the two parts of the PLQ model, by exhibiting the contributions from:
the enzyme velocity v0 = vmax [L]/(KM + [L]) and the precipitation of
the dose-averaged specific energy around the lesion zD + [L].This is merely
a formal separation, since the dose-averaged specific energy zD is not a
quantity which is independent of the enzymatic repair. Quite the contrary,
combining the definition zD = β/α with β = vmax /KM, we have:

vmax
zD = α . (124)
KM
Thus, the enzyme repair system effectively modifies the cell radiosensitivity α
by the multiplying factor vmax /KM due to the Michaelis–Menten chemical
kinetics. Because of this inter-connection between zD and {vmax , KM }, the
product v0 (zD ) + [L], or equivalently, v0 (zD + D) appearing in the effect
(PLQ)
EB from the PLQ model should not be taken too literally to mean a true
separation of the two independent mechanisms, the one being enzymatic
repair (v0 ) and the other being of microdosimetric origin (zD + D). It is
merely for the reason of drawing an analogy rather than making a one-to-
one correspondence that we used the notation zD from microdosimetry for
the defining quotient α/β of the two parameters in the PLQ model for the
cell kill (α) and cell repair (β) mechanisms of the cell response to radiation.
The microdosimetric parameter zD was also employed earlier in Eq. (33)
for the effect E(LQ)

B in the LQ model.There, following Kellerer and Rossi,13
(LQ)
we expressed the defining relation EB = αD + βD 2 = βD(α/β + D) as
(LQ)
EB = βD(zD + D), where zD = α/β. Using the assumed direct propor-
tionality between the lesion number (concentration) [L] and dose D, via
[L] = κD ≡ D, as per (106) and (108), we can further write:
(LQ)
EB = βD (zD + D)
= {β[L]} (zD + [L])
(LQ)
≡ v0 (zD + [L]) ,
where,
(LQ)
v0 = β[L]
vmax [L] (125)
= .
KM
In this way, the effect (33) and the surviving fraction (16) from the
LQ model can be cast in the following form of the Michaelis–Menten
terminology:
(LQ) (LQ)
EB = v0 (zD + [L])
(126)

vmax [L]
= (zD + [L]) ,
KM
and,
SF
(LQ)
(D) = e −v0
(LQ)
(zD +[L]) (127)

[L]
− vmax
K (zD +[L])
=e M .
Notice that both surviving fractions (120) and (127) in the PLQ and LQ
models, respectively, are expressed through three parameters {zD , vmax , KM }.
(LQ)
However, there is a special circumstance within v0 in the LQ model
permitting a reduction from this apparent three to only two degrees of
freedom. This is possible because the two parameters vmax and KM do not
appear individually in SF(LQ) (D) at different places, but rather they enter Eq.
(LQ)
(127) through v0 exclusively as the ratio vmax /KM.This leads to a reduc-
tion from {zD , vmax , KM } to {zD , β}, where β = vmax /KM and zD = α/β.
As such, the apparent three parameter set {zD , vmax , KM } is, in fact, a col-
lection of the usual two parameters α and β from the LQ model. In the
general version of the PLQ model§, none of the three parameters could
be eliminated so as to have only two remaining degrees of freedom. The
reason is that, instead of the velocity v0(LQ) = vmax /KM from the effect
(LQ) (LQ)
EB = v0 (zD + [L]) in the LQ model, the initial enzyme velocity
v0 = vmax [L]/(KM + [L]) appears in the corresponding effect in the LQ
model via:

(PLQ) vmax [L]
EB = v0 (zD + [L]) = (zD + [L])
KM + [L]
[L] −1

vmax [L]
= (zD + [L]) 1 +
KM KM
−1
(LQ) [L]
= EB 1+ ,
KM
so that,
(LQ)
(PLQ) EB (128)
EB = ,
1 + [L]/KM
as in (78). Thus, the general PLQ model possesses three parameters because
any attempt to express E(PLQ)
B through the two parameters {α, β} in E(LQ)
B
invariably leads to the emergence of the third parameter via the isolated
(LQ)
term 1/KM = γ . The above juxtaposition of enzyme velocities v0 and
v0 from the LQ and PLQ models, respectively, is instructive, since it facili-
tates one of the mechanistic levels of comparison between these two for-
malisms. This is best seen by observing that:
v0
(LQ)
= vmin , (129)
where vmin is the asymptote of the reaction velocity v0 at small concen-

tration of lesions, as per (113). Hence, the PLQ model with its Michaelis–
Menten chemical kinetics of enzyme catalysis for lesion repair can help
§ The general PLQ model is the one which excludes the special case in which the final slope vmax is
constrained to satisfy the relation vmax = 1/D0 from (123), provided that the mean lethal dose D0 is
viewed as known by e.g., reading off the ending, exponential part of the curve for the cell surviving
fraction.
understand one of the limitations of the LQ model, such as the restric-

tion to (129). A small concentration of lesions is associated with low-dose
(LQ)
exposure of cells to radiation. Therefore, restriction of v0 to only small
repair velocity vmin of enzyme molecules means that the validity of the LQ
model is limited to low doses. This conclusion from the Michaelis–Menten
formalism is in accordance with the well-known fact that the LQ model is
a low-dose approximation to cell surviving fraction.
5. RESULTS: COMPARISON OF RADIOBIOLOGICAL

MODELS WITH MEASUREMENTS
The relative performance of the PLQ and LQ models is illustrated by their
comparisons with experimental data. This is done on the level of cell sur-
viving fractions SF (D) and also by plotting the so-named full effect graph.4
Such twofold comparisons are deemed necessary for the reasons that run
as follows.
At low-to-intermediates doses, quite different radiobiological models
can still be in reasonably close agreement with experimental data when
plotted as cell surviving fractions SF (D) versus D. This is also evident
from each panel (i) on Figures 14.1–14.3 when comparing the PLQ
and LQ models with measurements. Of course, it is also clear from the
same panel (i) on these figures that this type of relatively good agreement
between these two formalisms ceases to exist at larger doses because of
the prevailing Gaussian and exponential shapes of cell surviving fractions
in the LQ and PLQ model, respectively. The displayed experimental data
for the corresponding cell surviving fractions favor the predictions by the
PLQ model at all doses. This confirms the theoretical expectation that the
PLQ model is universally valid at any dose D. By contrast, at high doses
the LQ model is seen to break down, as it largely underestimates the sur-
viving fractions from the measurements.
Overall, at small and intermediate doses, survival curves do not appear to
be the most suitable for differentiating among various models while evaluat-
ing their clinical usefulness in radiotherapy. Moreover, dose–effect functions
SF (D) are rarely of direct use in dose planning systems that, instead, most
frequently employ the biological effect EB (D) and the biologically effec-
tive dose BED(D). There is yet another useful relationship, which offers the
possibility for a more stringent assessment of clinical adequacy of different
biophysical models. This is the so-called full-effect plot, or Fe-plot,4 which
is associated with the ratio of the biological effect and the absorbed dose,
Fe(D) = (1/D)EB (D), or equivalently, Fe(D) = −(1/D) ln SF (D). This

quantity is also known by the alternative name “reactivity”13 and denoted
by R(D), which is also used in panel (ii) on Figures 14.1–14.3:
Fe(D) ≡ R(D)
1
= EB (D) (130)
D
1
= − ln SF (D).
D
Such a biological effect per unit dose represents the full effectiveness of
radiation on cell survival for each given level of dose exposure. It is this
Fe-plot, depicting Fe(D) versus D, or equivalently, R(D) as a function of
dose, which can distinguish one model from another in the most dramatic
way, as is clear from panel (ii) on Figures 14.1–14.3. In the Fe-plot, the LQ
model yields a linear radiation response, as displayed by a straight line of a
slope β and the intercept α on the ordinate:
Fe(LQ) (D) = R (LQ) (D) = α + βD. (131)
This means that the effectiveness of radiation at every dose level would
have no bound, as it would be indefinitely increased with augmentation
of D. Such a pattern is at variance with most experimental data Fe(exp) (D)
that are seen on plot (ii) of Figures 14.1–14.3 to saturate to some constant
values at high doses. This behavior is also predicted by the PLQ model
whose Fe-plot levels off to the constant final slope β/γ , as D becomes very
large:
α + βD β
Fe(PLQ) (D) = R (PLQ) (D) = −→ . (132)
1 + γ D D→∞ γ
Here, the rectangular hyperbola (α + βD)/(1 + γ D) from the PLQ model
implies the existence of repair of radiation damage to the cell through a
mechanism of the Michaelis–Menten type for enzyme-lesion catalysis. As
such, panel (ii) on Figures 14.1–14.3 for the Fe-plot shows excellent agree-
ment of the PLQ model with the corresponding experimental data.
Overall, we can conclude that the universal applicability of the PLQ
model to all doses is demonstrated in both panels (i) and (ii) of Figures
14.1–14.3 for cell surviving fractions and the Fe-plot. Simultaneously,
these comparisons prove the marked superiority of the PLQ over the LQ
model. This outperformance of the latter by the former radiobiological
model testifies to the adequacy of the mechanistic underpinning of the
Figure 14.1 Cell surviving fractions SF(D) as a function of radiation dose D in Gy (top
panel (i)). Bottom panel (ii), as the Fe-plot, shows the so-called reactivity R(D) given
by the product of the reciprocal dose D−1 and the negative natural logarithm of
SF(D), as the ordinate versus D as the abscissa. Any departure of experimental data
from a straight line indicates inadequacy of the LQ model for the Fe-plot. Experimental
data (full circles):30 the mean clonogenic surviving fractions SF(D) (panel(i)) and
R(D) ≡ −(1/D) ln(SF) (panel (ii)) for the human small cell lung cancer line (U1690)
irradiated by 190 kVp X-rays. Theories: solid curve for the PLQ (Padé linear-quadratic)
model and dashed curve for the LQ (linear-quadratic) model (the straight line α + βD
on panel (ii)).
Figure 14.2 Cell surviving fractions SF (D) as a function of radiation dose D in Gy (top
SF (D), as the ordinate versus D as the abscissa. Any departure of experimental data
data (full circles):17 the mean clonogenic surviving fractions SF (D) (panel(i)) and
R(D) ≡ −(1/D) ln(SF) (panel (ii)) for the Chinese hamster cells grown in culture and
irradiated by 50 kVp X-rays. Theories: solid curve for the PLQ (Padé linear-quadratic)
model and dashed curve for the LQ (linear-quadratic) model (the straight line α + βD
on panel (ii)).
Figure 14.3 Cell surviving fractions SF(D) as a function of radiation dose D in Gy (top
SF(D), as the ordinate versus D as the abscissa. Any departure of experimental data
data (full circles):31 the mean clonogenic surviving fractions SF(D) (panel(i)) and
R(D) ≡ −(1/D) ln(SF) (panel (ii)) for the asynchronous V79 Chinese hamster cells irradi-
ated hypoxically by 250 kVp X-rays with a concurrent 30 min exposure to the sulfhydryl-
binding agent, N-ethylmaleimide, of low concentration 0.75 µM. Theories: solid curve
for the PLQ (Padé linear-quadratic) model and dashed curve for the LQ model (the
straight line α + βD on panel (ii)).
Padé linear-quadratic formalism, which is rooted in a firm theoretical and

practical basis of chemical kinetics for repair of radiation damage by means
of enzyme-lesion catalytic reaction.
It should be noted that the concept of the Fe-plot is critically important
for both the conventional fractionation and hypofractionation. It impacts one
of the most delicate decisions by radiotherapists regarding the question: given
that radiation indiscriminately damages both tumorous and healthy cells,
how should the total dose vary as a function of the overall irradiation time,
as well as the number of fractions, in order to maintain a constant b iological
end effect and also minimize complications to the normal tissues at risk?
6. DISCUSSION AND CONCLUSION

6.1 Biologically expressed response of the cell
to irradiation
It has been argued that the ultimate success of radiotherapy rests upon
the possibility to properly understand cell repair after irradiation.14, 15 The
main focus of this chapter is on enzymatic repair mechanisms encoun-
tered in radiobiological descriptions of dose–effect relationships. With this
goal, we further elaborate the Padé linear-quadratic model, or the PLQ
model4–6 for cell surviving fraction, SF(PLQ) (D), as a function of a single
absorbed dose D. In this novel biophysical model, the biological effect
of radiation, E(PLQ)
B (D) = − ln SF
(PLQ)
(D), is given by the Padé approxi-
mant, E(PLQ)
B (D) = (αD + βD 2 )/(1 + γ D). By a smooth transition, this
rational function becomes automatically linear at both low and high doses,
(PLQ) (PLQ)
EB (D) −→ αD and EB (D) −→ (β/γ )D. Precisely such types of
D→0 D→∞
exponentials have been observed by numerous measurements in the said
two dose limits. The PLQ model has three parameters {α, β, γ }. Here, as
usual, radiosensitivity α is a single event inactivation constant in units of
Gy−1. However, parameter β, which is in units of Gy−2, is derived from the
introduction of a delay time in the cell response to radiation insult. This
delay is associated with the existence of a repair or recovery time τ . Any
two consecutive radiation events or hits would be wasted, i.e., not regis-
tered at all by the cell, if they were separated by a time interval t such that
�t < τ. The cell becomes effectively insensitive to such consecutive hits.
Parameter γ is the reciprocal of the Michaelis–Menten constant, KM, from
the theory of chemical kinetics for enzyme catalysis. This latter quantity
is the concentration of lesions at which the enzyme velocity of repair, v0,
attains one half of its maximum,7 i.e. vmax.
6.2 Dose–effect relationships at low, intermediate

(shoulder), and high doses
One of the most important advantages of the PLQ model relative to the
linear-quadratic model is of particular relevance to radiotherapy by high
doses per fraction used especially in stereotactic radiosurgery. For this treat-
ment modality, called hypofractionation, the LQ model is inadequate, since
its biological effect, E(LQ)
B (D) = αD + βD , has a high-dose asymptote
2
(LQ)
EB (D) −→ βD , which is at variance with the corresponding experi-
2
D→∞ (exp)
mental data exhibiting the exponential shape, EB (D)∼D, as D → ∞.
This severely hampers the proper use of one of the key quantities in dose
planning systems, the so-named biologically effective dose, which is a scaled
biological effect, BED(D) = (1/α)EB (D). Since the LQ model is not
universally valid at all doses, the entire set of the given experimental data
for BED(exp) (D) cannot be used for extracting the biological parameters.
Therefore, the usual practice is to carry out a segmentation of the given set
of experimental data BED(exp) (D) into different dose ranges to estimate the
ratio β/α from the postulated relation BED(exp) (D) ≈ BED(LQ) (D), where
(LQ)
BED(LQ) (D) = (1/α)EB (D) = 1 + (β/α)D. A serious disadvantage
of such a procedure is that the quotient β/α and, therefore, BED(exp) (D)
become dose-range dependent. This introduces complications in employ-
ing the BED concept to compare the conventional fractionation (2 Gy per
fraction)16–20 with hypofractionation.21, 22 Such comparisons are critical
for extrapolating the abundant experience with conventional fractionation
to hypofractionated treatments. This is vital given that larger doses per frac-
tion have a tendency of causing more severe late side effects relative to the
conventional small size fractions. Additionally, the LQ model has difficulties
in coping with cell survival curves with broad shoulders.23, 24 Moreover, on
top of continuously bending down as dose D is enlarged, pointing to the
non-existence of the final slope and the extrapolation number n, the LQ
model can break down at very low doses, as well.12
6.3 Beyond the linear-quadratic model of cell inactivation

In order to partially overcome the mentioned drawbacks of the LQ model,
Paganetti and Goitein25 introduced in 2001, within the amorphous track
partition (ATP) model, a modification containing a Heaviside step func-

tion with a transition dose DT. Their surviving fraction coincides with
2
the linear-quadratic response e−αD−βD from the LQ model at D DT
and, conversely, becomes a linear function of dose D at D > DT via

e−αDT −βDT −γ (D−DT ). Here, γ is either a third independent fitting parameter
2
or the final slope fixed by the continuity constraint of the derivative of the sur-
viving fraction at D = DT, which leads to γ = α + 2βDT. More recently in
2008, the modified LQ model from Ref. 25 has been renamed as the universal
survival curve (USC) model by Park et al.,26 and the linear-quadratic-linear
(LDL) model by Astrahan.27 However, the common feature of Refs. 25–27 is
an ad hoc switch from the incorrect D 2 high-dose component in the LQ model
to the corresponding term with a linear dose dependence (∼ D) in the cell
surviving fraction. The transition dose D = DT at this switch has no justifi-
able biological significance, as it represents just another free parameter. Typical
measurements of surviving fractions for most mammalian cell lines can be
trustworthy only down to the 10−3 survival level. For this reason, extraction of
parameter DT from such experimental data could hardly be reliable. Astrahan27
tried to attribute a clinical meaning to DT by claiming that it delineates the
region of the passage from the shoulder region to the linear component of the
LQ model. Evidently, this is merely rewording the mathematical meaning of
the mentioned Heaviside step function from Refs. 25–27 and, as such, cannot
constitute a clinical nor biological interpretation of the transition dose DT.
Moreover, it has been found in applications27, 28 that DT can be anywhere in
a quite wide dose range 15Gy–30 Gy. Such locations of DT are incompatible
with Astrahan’s27 interpretation of the transition dose, since shoulders do not
typically extend to even the lowest limit (15 Gy) of the mentioned interval.
6.4 Mixed-order chemical kinetics for enzymatic

cell repair systems
The mentioned problems with the LQ model have also been addressed
within the PLQ model.4–6 In this mechanistic description, as opposed to an
empirical transition dose DT, different passages from intermediate to high
doses are governed by natural switches from various orders (zero, first, sec-
ond) of chemical kinetics that underlies interactions of radiation with the
cell. A key role in these different switches from one to another dose depen-
dence of cell surviving fraction is the overall activity of enzyme molecules
in the process of repair of radiation damage of the cell. This mixed-order
enzyme catalysis, which is at the center of the cell repair system, guarantees
the emergence of the correct asymptotes of the biological effect at both
small and large doses. It also secures the existence of a shoulder of the
proper width at intermediate doses in typical cell surviving fractions. Such
a clear mechanism is backed by the accompanying mathematical formalism
in the PLQ model through the Padé approximant, which is known to pro-
vide optimal interpolations and extrapolations between different regions
of a given function.11 This is achieved smoothly without ever resorting to
unnecessary artifices, such as sewing two different regions by a transition
dose DT placed at an empirically found point through the Heaviside step
function as in Refs. 25–27. Our initial testings,4–6 as reviewed here, and our
more recent thorough comparisons of nine different models with six cell
lines28 resulted in the common conclusion that the PLQ model systemati-
cally provides the most satisfactory description of cell survival after irradia-
tion. This is most prominently evidenced at high doses in the reconstructed
dose–effect curves as well as in the associated Fe-plots.29
ACKNOWLEDGMENTS
This work is supported by research funds from the Karolinska Institute, the
Karolinska University Hospital’s Radiumehemmet and the Swedish Cancer
Society(Cancerfonden).
REFERENCES
1. Dale, R. G. The potential for radiobiological modelling in radiotherapy treatment
design. Radioth. Oncol. 1990, 19, 245.
2. Jones, B.; Dale, R. G. Mathematical models of tumour and normal tissue response. Acta
Oncol. 1999, 38, 883.
3. Sinclair, W. K. Biophysical aspects of radiation quality, Technical Report Series, 58,
IAEA, Vienna, Austria, 1966.
4. Belkić, Dž. Parametric analysis of time signals and spectra from perspective of quantum
physics and chemistry. Adv. Quantum Chem. 2011, 61, 145.
5. Belkić, Dž.; Belkić, K. Padé-Froissart exact signal-noise separation in nuclear magnetic
resonance spectroscopy. J. Phys. B: At. Mol. Opt. Phys. 2011, 44, 125003.
6. Belkić, Dž.; Belkić, K. High-resolution signal processing in magnetic resonance spec-
troscopy for early cancer diagnostics. Adv. Quantum Chem. 2011, 62, 243.
7. Michaelis, L.; Menten, M. L. Die kinetik der invertinwirkung. Biochem. Z. 1913, 49, 333.
8. Barendsen, G. W. Dose fractionation, dose rate, and isoeffect relationships for normal
tissue response. Int. J. Radiat. Oncol. Biol. Phys. 1982, 8, 1981.
9. Fowler, J. F. The linear-quadratic formula and progress in fractionated radiotherapy. Br. J.
Radiol. 1989, 62, 679.
10. Corless, R. M.; Gonnet, G. H.; Hare, D. E. G.; Jeffrey, D. J.; Knuth, D. E. On the Lambert
W function. Adv. Comput. Math. 1996, 5, 329.
11. Belkić, Dž. Quantum-Mechanical Signal Processing and Spectral Analysis; Taylor & Francis:
London, 2005.
12. Thames, H. D.; Ang, K. K.; Stewart, F. A.; van der Schueren, E. Does incomplete repair
explain the apparent failure of the basic LQ model to predict spinal cord and kidney
responses to low doses per fraction?. Int. J. Radiat. Biol. 1988, 54, 13.
13. Kellerer, A. M.; Rossi, H. H. RBE and the primary mechanism of radiation action. Rad.
Res. 1971, 47, 15.
14. Steel, G. G.; Duchesne, J. M.; Horwich, G. M.; Kelland, L. R.; Peacock, J. H. The dose-
rate effect in human tumour cells. Radioth. Oncol. 1987, 9, 299.
15. Stephens, T. C.; Eady, J. J.; Peacock, J. H.; Steel, G. G. Split-dose and low-dose rate
recovery in four experimental tumour systems. Int. J. Rad. Biol. 1987, 52, 157.
16. Jacobson, B. Evidence of recovery from X-ray damge in Chlamydominas. Rad. Res.
1957, 1, 394.
17. Elkind, M. M.; Sutton, H. X-ray damage and recovery in mammilian cells in culture.
Nature 1959, 184, 1293.
18. Elkind, M. M.; Sutton, H. Radiation response to mammilian cell growth in culture: I.
Repair of X-ray damage in surviving Chinese Hamster cells. Rad. Res. 1960, 13, 556.
19. Elkind, M. M.; Sutton-Gilbert, H.; Moses, W. B.; Kamper, C. Sublethal and lethal radia-
tion damage. Nature 1967, 214, 1088.
20. Fowler, J. F. What next in fractionated radiotherapy. Br. J. Cancer 1984, 49, 285.
21. Muriel, V. P. Hypofractionation for radiotherapy. Clin. Transl. Obncol. 2007, 9, 21.
22. Hopewell, J. W.; Millar, W. T.; Ang, K. K. Toward improving the therapeutic ratio in ste-
reotactic radiosurgery: selective modulation of the radiation responses of both normal
tissue and tumor. J. Neurosurg. 2007, 107, 84.
23. Hutchinson, F.; Easter, S. S., Jr; A difference between biological effect of gamma rays
and heavy ions. Science 1960, 132, 1311.
24. Utsumi, H.; Elkind, M. M. Potentially lethal damage versus sublethal damage:

Independent repair processes in actively growing Chinese Hamster cells. Rad. Res. 1979,
77, 346.
25. Paganetti, H.; Goitein, M. Biophysical modelling of proton radiation effects based on
amorphous track models. Int. J. Radiat. Biol. 2001, 77, 911.
26. Park, C.; Papiez, L.; Zhang, S.; Story, M.; Timmerman, R. D. Universal survival curve
and single fraction equivalent dose: Useful tools in understanding potency of ablative
radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 2008, 70, 847.
27. Astrahan, M. Some implications of linear-quadratic radiation dose-response with regard
to hypofractionation. Med. Phys. 2008, 35, 4161.
28. Abdisheh, B.; Edgren, M.; Belkić, Dž.; Mavroidis, P.; Lind, B.; Brahme, A. A comparative
analysis of radio-biological models for cell-surviving fractions at high doses. Tech. Cancer
Res. Treatm. Ahead of print 2012 (DOI:10.7785/tcrt.2012.500306).
29. Douglas, B. G.; Fowler, J. F.The effect of multiple small doses of X-rays on skin reactions
in the mouse and basic interpretation. Rad. Res. 1976, 66, 401.
30. Persson, L. Cell survival at low and high ionization densities investigated with a new
model. PhD Thesis (unpublished), Stockholm University, Stockholm, 2002.
31. Kimmler, B. F.; Sinclair, W. K.; Elkind, M. M. N-Ethylmaleimide sensitization of

X-irradiated hypoxic Chinese Hamster cells. Rad. Res. 1977, 71, 204.
INDEX
A energy spectra of secondary ejected

Advanced Resource Center for electrons, 299–301
Hadrontherapy in Europe final state description, 292–294
(ARCHADE), 80–81 influence of target description on
Amorphous track partition (ATP) model, ionization cross sections, 304–305
446–447 total ionization cross sections,
301–302
B from vapour to liquid water, 302–304
Barkas and Bloch corrections, mean Born cross sections, 365–366
excitation energy of water, 70–72 Born distorted wave (BDW-4B), 340,
Basis generator method, 316 354–358
adapted for ion-molecule collisions, Bragg additivity rule, mean excitation
320–323 energy of water, 70
BCIS method, 381–387 Bragg functions
BDW-4B theory, 355 C-ions, calculations for, 12–14, 30
Barendsen formula, 416–417 description, 8
Bethe–Bloch formula, 232 fluence functions properties, 8, 10
Bethe–Fano collision cross section mean ion energy, 9
C-ions, 24 3600 MeV C-ions traversing water, 13
cumulative functions, 29–31 200 MeV protons traversing water, 9, 11
inelastic form factor, 25–26 nuclear interactions, 10–12
optical dipole oscillator strength, 27 protons, calculations for, 12
Bethe formula, 232 stopping power, 9
Bethe term, mean excitation energy of Bragg peak, 270, 408
water, 66 Brandt and Kitagawa (BK) model
Bichsel–Uehling method, 11 charge density of projectile
Binary encounter (BE) peak, 211 description, 134–135
Binary projectile–electron encounter, 211 dielectric response approximation,
Biologically effective dose, 416–417, 134
429–430 Fourier transform, 135
BK model. See Brandt and Kitagawa (BK) internal energy, 134–135
model modification of parameter radius, 135
Born approximations, 283–309 screening radius, 134–135
convergence, 365–366
theoretical description of charge transfer C
process within the 1st Born Carbon ions, Hadron therapy, stopping
basic formalisms, 305–308 power codes and ion ranges
total cross sections, 308–309 condensed compounds, 57
theoretical description of ionization gases, 57–58
process within the 1st Born solid elements covered by MSTAR,
angular distributions of secondary 56–57
electrons, 294–299 stopping powers for, 56
basic formalisms, 288–292 CDW-4B theory, 347–349
451
452 Index
CDW-EIS- EIV model, 344–346 double-differential inverse inelastic mean

CDW formalism, 213 free path (IIMFP), 132–133
CDW-IEV approach, 344–346 electronic charge density of swift
Cell damage by radiation, 409–410 projectile, 134–135
Cell repair mechanism, 424 electronic excitation spectrum
Cell surviving fraction, 410, 412 Bethe surface of biological materials,
Charged projectiles, 235–239 141–142
Charge transfer cross section, 280 ELF, 135–136
Charge transfer process, 272–273 MELF–GOS model. See Mermin-
within the 1st Born approximations energy-loss-function-generalized-
basic formalisms, 305–308 oscillator strength (MELF-GOS)
total cross sections, 308–309 model
Close coupling methods, 364–365 optical-ELF of biological materials,
Collision cross sections (CCS) 138–141
inelastic collision cross sections for energy loss in particle penetration of
water, 27 matter, 173–182
Rutherford cross section, 23–26 energy loss straggling, 134, 182–190
secondary radiations, 23 in high-energy ion–matter collisions,
Complete neglected differential overlap 166–167
(CNDO) approach, 304–305 ionization probabilities, 190
Continuum distorted wave (CDW) Levine–Louie dielectric function and
method, 213, 233–234, independent-shell approximation,
386–388, 391 168–173
four-body, 347–354 mean excitation energy
Continuum Distorted Wave-Eikonal Initial Bethe formula for stopping power,
State (CDW-EIS), 213–214 142–144
method, cross sections calculation, biological material, 144
85–86 PMMA, liquid water and DNA
model, 233–234 comparison, 144
Corrected–Born approximation, 367 stopping power, Bethe formula for,
Corrected –Born (CB1) approximation, 142–144
379–380 multiple ionization, 192–193
Coulomb interaction, 367–368 projectile description, 134–135
Shellwise local plasma approximation,
D 167–173
Density Based Clustering Algorithm with single-differential inverse inelastic mean
Noise (DBSCAN) free path, 133
clustering estimation, 102–103 SLPA for stopping and straggling of
DNA clustering estimation, 102–103 gases, 186–190
DSB and SSB ratio comparison with SLPA in stopping power of metals for
PARTRAC code, 103–104 protons, 173–178
sub-excitation electrons, 104 stopping for dressed ions, 180–182
Density-functional theory approach to stopping magnitudes, 134
heavy-particle collisions, 318–320 stopping number and Bethe limit,
Dielectric formalism for inelastic 178–180
processes differential cross sections, stopping power, 134
193–195 target description, 135–142
Index 453
total inverse inelastic mean free path, extrapolation number and quasi-
133 threshold dose, 413–415
total ionization cross sections, 191–192 Poisson distribution of radiation events,
two-electron transitions in atomic mean lethal dose, 410–413
targets, 133 Dose–effect relationships at low,
Dielectric function, 169 intermediate (shoulder), and high
Dielectric response function, formalism doses, 446
based on, 216–217 Dosimetry, 419–422
Dipole oscillator strength distribution Doubly differential cross sections (DDCS),
(DOSD) 273–274, 278, 289
mean excitation energy of water, 66 Drude-like dielectric function, mean
Disconnected diagrams, 382 excitation energy of water, 68
Distorted wave Born approximation Dynamic electron correlations, 352–353
(DWB), 392–393
Distorted wave formalism and its main E
variants, 213–214 Electronic stopping power, 241–245
Distorted wave model for inelastic Energy deposition, radiation therapy
collisions, 234–241 data for, one C-ion, 16
charged projectiles, 235–239 LEM model, 17
for neutral projectiles, 239–241 microscopic volumes, 15–17
Distorted wave models, 233–234 Monte Carlo simulation, 15, 18
Distorting potential, 367–368 wall effects, 15
DNA damage, 79 Energy loss, radiation therapy
Density Based Clustering Algorithm cells, stochastics of energy loss in, 17–21
with Noise (DBSCAN) clustering description, 14–15
estimation macroscopic level, 4–8
description, 102–103 Energy loss in particle penetration of
DSB and SSB ratio comparison with matter, 173–182
PARTRAC code, 103–104 Energy loss straggling, 182–190
sub-excitation electrons, 104 Energy spectra of secondary electrons,
Garty statistical approach, 101–102 299–301
Stewart and Semenenko MCDS method Enzymatic repair, 432–435
categories of damages, 100–101 Enzyme-Linked Immunosorbent Assay
cell repair activities, 101 (ELISA)
track structure detailed approach, DNA damages measurement, 80
99–100 Extrapolation number, 413–416, 431, 433,
Dodd–Greider equation, for quantum 437–438, 448
mechanical operator, 372–373
Dodd–Greider integral equation F
Moller wave operators, 392–393 Faddeev equations, 365, 372–373, 391
in theory of two-electron processes, Fermi-virtual-photon (FVP) cross section,
390–401 28–29
DOSD. See Dipole oscillator strength First Born Approximation (FBA)
distribution (DOSD) cross sections for protons inelastic
Dose-averaged specific energy, 421, 423, collisions calculation, 85–86
437 for ionization and charge transfer,
Dose–effect curve, 413 270–272
454 Index
Born approximations, 283–309 SRIM, 53, 54

energy spectra of secondary electrons, H ions, application to therapy using,
299–301 55–56
ion-induced ionization and charge ionization energy, 41
transfer cross sections in water, liquid water
272–281 Bragg peak (BP) positions, 46–48
experimental background, 273–275 hydrogen ions, stopping power of
theoretical background, 275–281 water for, 43–45
molecular description of water target, I values of, 45–46
281–283 mean ionization energy, 45–49
Formalism based on dielectric response measured ranges and theoretical
function, 216–217 stopping power comparison, 48
Forward scattering direction, 370 range measurements, 45–49
Four-body theories for transfer ionization range value comparison from
in fast ion-atom collisions, 339–340 PASS, 49
four-body Born distorted wave method, mass stopping power, 39
354–358 SRIM program, 40–41
four-body continuum distorted wave tables and programs, 41–43
method, 347–354 target substances and statistical
independent particle/event models, comparisons
340–347 carbon ions, 56–58
two-electron transitions, 339–340 H ions, application to therapy, 55–56
Fragmentation cross sections, 316, 332–334 H versus He ions, 51–55
Free-electron gas, 167, 169 Judge program, 51
Free radicals formation in water radiolysis, linear stopping powers for protons, 50
261–263 Hartree–Fock wave function, 346
F-sum rule, 169 Heavy Ion Medical Accelerator in Chiba
(HIMAC), 80–81
G Heavy-particle collisions, density-
Garty statistical approach, 101–102 functional theory approach to,
Gel electrophoresis 318–320
DNA damages measurement, 80 Heidelberg Ion-Beam Therapy (HIT),
80–81
H Helium ions, H and He ions statistical
Hadron therapy, stopping power codes and comparisons
ion ranges elemental gases, 52–53
carbon ions, statistical comparisons ICRU Report 49 versus SRIM 2003,
condensed compounds, 57 53–55
gases, 57–58 mean normalized deviations, 51–52
solid elements covered by MSTAR, SRIM, 53–54
56–57 High-energy heavy ion collisions with,
stopping powers for, 56 232–234
H and He ions statistical comparisons case of water molecules, 245–253
elemental gases, 52, 53 distorted wave model for inelastic
ICRU Report 49 versus SRIM 2003, collisions, 234–241
53–55 theoretical description for charged
mean normalized deviations, 51–52 projectiles, 235–239
Index 455
theoretical description for neutral projectile description, 134–135

projectiles, 239–241 single-differential inverse inelastic
electronic stopping power, 241–245 mean free path, 133
free radicals formation in water stopping magnitudes, 134
radiolysis, 261–263 stopping power, 134
linear energy transfer, 260–261 target description, 135–142
multiple ionization of water molecules, total inverse inelastic mean free
253–263 path, 133
HKS model, 216–217 two-electron transitions in atomic
Hydrogen ions targets, 133
H and He ions statistical comparisons dielectric function, 131–132
elemental gases, 52–53 distorted wave model for, 234–241
ICRU Report 49 versus SRIM 2003, charged projectiles, 235–239
53–55 for neutral projectiles, 239–241
mean normalized deviations, 51–52 of energetic protons in biological media
SRIM, 53–54 inelastic energy-loss magnitudes
electronic excitation probability,
I 146–147
Inclusive probabilities, 323 energy-loss distribution, 147–148
Independent particle/event models, energy-loss straggling, 151–152
340–347 IIMFP, 148–149
Inelastic collisions stopping power, 149–151
Bethe–Bloch theory, 131 Mermin dielectric function, 132
charge-exchange processes shell-corrections to Bethe’s theory, 131
charge-state fractions, 145–146 Inelastic form factor, 25–26
energy-loss magnitudes, 145 Inelastic processes in high-energy
cross sections for water, 27 ion–matter collisions, dielectric
depth-dose distributions, simulation of formalism for, 166–167
average energy, SEICS code, 156–157 differential cross sections, 193–195
Bragg peak, 157 energy loss in particle penetration of
PMMA, SEICS code in, 157 matter, 173–182
proton beam in water and PMMA, energy loss straggling, 182–190
157–158 ionization probabilities, 190
SEICS code, 153–156, 158–159 Levine–Louie dielectric function and
trajectory of energetic particle independent-shell approximation,
through target, 154 168–173
water-equivalent ratios (WER), multiple ionization, 192–193
157–159 Shellwise local plasma approximation,
dielectric formalism for inelastic 167–173
scattering SLPA for stopping and straggling of
double-differential inverse inelastic gases, 186–190
mean free path (IIMFP), 132–133 SLPA in stopping power of metals for
electronic charge density, 134–135 protons, 173–178
electronic excitation spectrum, stopping for dressed ions, 180–182
135–142 stopping number and Bethe limit,
energy-loss straggling, 134 178–180
mean excitation energy, 142–145 total ionization cross sections, 191–192
456 Index
Intensity Modulated Hadron Therapy Ion-molecule collisions, 317–318, 322

(IMHT), 271 basis generator method adapted for,
Ion–atom collisions, 218 320–323
Ion-atomic collision reaction, 367–368 Irradiation, biologically expressed response
Ionization, 257–258 of cell to, 445–446
cross sections Isosurvival dose, 413–414
influence of target description on,
304–305 J
total, 301–302 Jacobi-type coordinates, 367–368
energy, Hadron therapy, stopping power Judge program, 51
codes and ion ranges, 41
probabilities, 190 K
Ionization process within the 1st Born Kummer hypergeometric function,
approximations 347–349
angular distributions of secondary
electrons, 294–299 L
basic formalisms, 288–292 Lacassagne Center Biomedical Cyclotron
energy spectra of secondary electrons, (CLCB), 80–81
299–301 Laplace operator, 367–368
final state description, 292–294 LEAHIST, 276
influence of target description on Levine–Louie dielectric function and
ionization cross sections, 304–305 independent-shell approximation,
total ionization cross sections, 301–302 168–173
from vapour to liquid water, 302–304 Linear energy transfer (LET), 260–261
Ionizing radiations, 206–208 Linear-quadratic model (LQ)
Ion–matter collisions, dielectric formalism Barendsen formula, 416–417
for inelastic processes in biological effect, relative effectiveness,
high-energy, 166–167 and biologically effective dose,
differential cross sections, 193–195 415–416
energy loss in particle penetration of Linear-quadratic mode of cell inactivation,
matter, 173–182 446–447
energy loss straggling, 182–190 Local Effect Model (LEM), 80–81
ionization probabilities, 190 Local plasma approximation (LPA), 167
Levine–Louie dielectric function and
independent-shell approximation, M
168–173 Mean excitation energy of water
multiple ionization, 192–193 Barkas correction, 66
Shellwise local plasma approximation, Bethe term, 66
167–173 Bloch correction, 66
SLPA for stopping and straggling of dipole oscillator strength distribution
gases, 186–190 (DOSD), 66
SLPA in stopping power of metals for experimental determination, 69–73
protons, 173–178 experimental determination of I0
stopping for dressed ions, 180–182 ansatz function for stopping
stopping number and Bethe limit, power, 69
178–180 Barkas and Bloch corrections, 70, 72
total ionization cross sections, 191–192 Barkas correction, 71
Index 457
Bragg additivity rule, 70 differentiation between physical and

dielectric response function, 69 biological doses, 419–425
ion in condensed phase target, 73 initial slope, final slope, and
projectile charge exchange, 72–73 extrapolation number, 429–432
projectile ion collisions, 73 Padé linear-quadratic model and
shell corrections, 71 Michaelis–Menten kinetics,
stopping cross section, 65, 67, 70 432–440
target phase, 72 repair-mediated non-linear damping
ratio of mass stopping power for ion in of linear direct cell kill mechanism,
water and air, 64–65 425–429
stopping power of material for fast radiobiological models with
ion, 65 measurements, comparison,
target stopping number, 65 440–445
theoretical determination, 67–69 MELF-GOS model. See Mermin-energy-
theoretical determination of I0 loss-function-generalized-oscillator
DOSD, 67 strength (MELF-GOS) model
Drude-like dielectric function, 68 Mermin-energy-loss-function-generalized-
hydrogen bonding, effect of, 68 oscillator strength (MELF-GOS)
target mean excitation energy, 68 model
vibrational averaging, 67 Bethe surface of biological materials,
ZM method, 67 141–142
Mean lethal dose, 410–413 electronic excitation spectrum
Mechanistic Repair-Based Padé Linear- benefits of, 137–138
Quadratic Model, 408–410 electron excitations, 136
beyond linear-quadratic mode of cell ELF, 137–138
inactivation, 446–447 inner-shell electrons, 136–137
biologically expressed response of cell to Lindhard dielectric function, 137
irradiation, 445–446 outer electrons, collective and
dose–effect curve single-particle excitations of, 137
extrapolation number and quasi- mean excitation energy
threshold dose, 413–415 Bethe stopping formula, 144
Poisson distribution of radiation Bethe surface obtained, 143
events, mean lethal dose, biological material, 144
410–413 stopping power, 142–144
dose–effect relationships at low, optical energy-loss function of
intermediate (shoulder), and high biological materials
doses, 446 energy-loss function (ELF), 138–140
linear-quadratic model liquid water, dry DNA, and
Barendsen formula, 416–417 PMMA,140–141
biological effect, relative effectiveness, Mermin–Lindhard dielectric function, 168,
and biologically effective dose, 180–182
415–416 Michaelis–Menten kinetics, 432–440
low- and high-dose asymptotes of Microdosimetry
biological effect and surviving biological tissue simulated by gas
fraction, 417–419 volume, 90–93
Padé linear-quadratic (PLQ) model code by Bigildeev and Michalik versus
Geant4-DNA, 93
458 Index
energy deposit points, 90–91 geometry of nuclear system, 375

Geant4-DNA process, 92–93 K -shell of carbon atom, 392
ions RBE estimation scattering vector, 381–382
biological weighting function, 97–99 two-body scattering Coulomb wave
energy deposition by carbon ions functions, 383
versus beam depth in water, 95–96 Molecular scale simulation of ionizing
fragments build-up, 97–98 particles
Geant4 fragmentation models and Bragg peak, 80–81
Fluka code, 94–95 DNA damage estimation, 80
kinetic energy profiles of protons and DBSCAN clustering estimation,
He nuclei, 96 102–104
radiation Garty statistical approach, 101–102
characterization, 89–90 Stewart and Semenenko MCDS
quality factors for radiotherapy method, 100–101
applications, 90–91 track structure detailed approach,
TEPC detector, 92–93 99–100
Modified Coulomb Born (MCB) method microdosimetry techniques, 82
cross sections calculation, 85–86 Monte-Carlo codes, 84–85
Modified distorted wave methods in processes cross sections
problem of one-electron, charge transfer, 86–87
364–367 elastic scattering cross sections, 87–88
amplitude of reaction, prior -formalism inelastic collisions of electrons, FBA
and post -formalism, 372–373 for, 87
basic kinematics and dynamics, protons inelastic collisions, 85–86
367–374 radiation microdosimetry analysis
Born approximation with distorted ions RBE estimation, 94–99
waves, amplitude of process, theory and experiment, 89–94
374–377 sub excitation electrons and chemical
boundary condition, 374–377 phase, 88–89
calculation of angular and energetic track structure codes
dependencies of charge-transfer Monte-Carlo codes, 84–85
cross sections, 387 processes cross sections, 85–88
CDW-approximations, 387–388 sub excitation electrons and chemical
CDW theory regarding one-electron phase, 88–89
processes, 391 Moliere theory, 11
confluent hypergeometric function, Moller wave operators, 370–372
374–377 Monte-Carlo codes
continuum-intermediate state method molecular scale simulation of ionizing
with correct boundary conditions particles, 84–85
(BCIS), 381–387 Monte Carlo simulations
convergence of Born series, 370–372 Positron emission topography (PET) in
Dodd–Greider integral equation in proton therapy
theory of two-electron processes, compound nuclei, decay processes of,
390–401 119–120
first Born method with correct ICRU Report 63, 120
boundary conditions (CB1), proton transport, 119
374–381 SimSET, 120–121
Index 459
SRNA-2KG, 119 coronal slice from 3D PET, 123–124

radiation therapy human tissue, nuclear interactions
description, 7–8 in, 113
simulation of stochastics at Bragg linear production densities of
peak, 21 11C isotopes, 122
Morse-oscillator-rigid-bender scheme of 13N isotopes, 122
Jensen, 67 15O isotopes, 121–123
Multiple ionization of water molecules, Monte Carlo simulations
253–263 compound nuclei, decay processes
Multiple scattering, 32 of, 119–120
ICRU Report 63, 120
N proton transport, 119
Net cross section results, 328–332 SimSET, 120–121
Net ionization cross section, 255 SRNA-2KG, 119
Neutral projectiles, 239–241 nuclear reactions, 114
Nuclear reactions, 32–33 cross sections, data for comparison,
114–118
O cross sections, total non-elastic cross
One-electron charge-transfer process, section, 118–119
364–367 results, 121–124
Oppenheimer Brinkman Kramer (OBK) Post-collision fragmentation
approximation, 365 model, 333
Orbital OLPA, 167–168 Proton therapy, PET in proton therapy.
See Positron emission topography
P (PET) in proton therapy
Padé approximant, 422–424 Proton Therapy Center of the Curie
Padé linear-quadratic (PLQ) model Institute at Orsay (ICPO),
differentiation between physical and 80–81
biological doses, 419–425
initial slope, final slope, and Q
extrapolation number, 429–432 Quasi-threshold dose, 413–414
Padé linear-quadratic model and extrapolation number and, 413–415
Michaelis–Menten kinetics,
432–440 R
repair-mediated non-linear damping of Radiation
linear direct cell kill mechanism, biologically effective dose (BED),
425–429 416–417
PCCDW (Pluvinage Correlated CDW) dose, 408–409
approximation, 404 mechanism driving pattern, 420–421
PLQ. See Padé linear-quadratic model Radiation therapy
(PLQ) Bethe–Fano collision cross section
Poisson distribution of radiation events, C-ions, water moments, 24, 30
mean lethal dose, 410–413 cumulative functions, 29–31
Positron emission topography (PET) in FVP cross section, 28–29
proton therapy inelastic form factor, 25–26
activity and standard deviation, PET, optical dipole oscillator strength, 27
123–124 biological effects of, 2, 21–22
460 Index
Bragg functions Bragg peak, MC simulation at, 20–21

C-ions, calculations for, 2, 4, 6, calculation in cell volume, 17
12–14 Enlarged energy spectrum of C-ions
description, 8 traversing water at Bragg peak, 19
fluence functions properties, 8, 10 irradiation, Bragg peak for, 17–19
mean ion energy, 9 MC simulation, 19–20
3600 MeV C-ions traversing physical processes, 17
water, 13 Poisson distribution, 18–19
200 MeV protons traversing water, straggling in thin segments, 7
9, 11 Radiation-tissue interactions, variabilities
nuclear interactions, 10–12 in, 409
protons, calculations for, 12 Radiobiological models, 419–422
stopping power, 9 cell surviving fractions, 442– 444
CCS with measurements, comparison,
inelastic collision cross sections for 440–445
water, 27 Relative biological efficiency (RBE),
Rutherford cross section, 23–26 270–271
secondary radiations, 23 Repairable lesions, 417–418, 425–429
cell survival, 33 Ricker function, 422–424
continuous transport of particles Rudd model, 215–216
through matter
mean energy loss, 4 S
range calculation, 4 Semi-empirical models, 272–273, 275,
convolutions for range straggling, 5–7 302–303
dose effect functions, 2 Shellwise local plasma approximation,
energy deposition 167–173
data for, one C-ion, 16 for stopping and straggling of gases,
LEM model, 17 186–190
microscopic volumes, 15–17 in stopping power of metals for protons,
Monte Carlo simulation, 15, 18 173–178
wall effects, 15 Single ionization of liquid water by
energy loss protons, alpha particles, and carbon
cells, stochastics of energy loss in, nuclei, 204–208
17–21 comparison between experimental,
description, 14–15 theoretical, and semiempirical
macroscopic level, 4–8 results, 217–226
key to overall success of, 409–410 ionization cross sections for water,
medical, 2 218–226
microscopic levels electronic stopping cross sections for
energy deposition, 15 protons on liquid water, 222
energy loss, 14–15 electron occupation numbers and
microscopic volumes, energy binding energies for liquid and
deposition, 15–17 gaseous water subshells, 206
multifaceted, 408 experimental works, 214–215
multiple scattering, 32 ionization problem, 206–208
nuclear reactions, 32–33 liquid water in radiation physics,
stochastics of energy loss in cells 205–206
Index 461
liquid water phase effects, 208 fragmentation cross sections, 332–334

semiempirical methods, 215–217 net cross section results, 328–332
formalism based on dielectric response theory, 317–325
function, 216–217 basis generator method adapted for
HKS model, 216–217 ion-molecule collisions, 320–323
Rudd model, 215–216 density-functional theory approach to
theoretical approaches, 208–214 heavy-particle collisions, 318–320
continuum-distorted wave extraction of measurable cross
approximation (CDW), 213 sections, 323–324
distorted wave formalism and its main Time-dependent Kohn–Sham (TDKS),
variants, 213–214 319
first Born approximation (B1), Time-dependent Schrödinger equation
210–213 (TDSE), 318
relevant vectors used in theoretical Tissue Equivalent Proportional Counters
models, 209 (TEPC), 82
transition amplitude, 212 Total ionization cross sections, 275
SLPA for stopping and straggling of gases, Transfer ionization (TI), 339–340
186–190
SLPA in stopping power of metals for W
protons, 173–178 Water
“Soft collision” (SC) region, 211–212 Hadron therapy, stopping power codes
Spectral representation, 316 and ion ranges
Stewart and Semenenko MCDS method Bragg peak (BP) positions, 46–48
categories of DNA damages, 100–101 hydrogen ions, stopping power of
cell repair activities, 101 water for, 43–45
Stochastics of energy loss in cells I values of, 45–46
Bragg peak, MC simulation at, 20–21 mean ionization energy, 45–49
calculation in cell volume, 17 measured ranges and theoretical
Enlarged energy spectrum of C-ions stopping power comparison, 48
traversing water at Bragg peak, 19 range measurements, 45–49
irradiation, Bragg peak for, 17–19 range value comparison from
MC simulation, 19–20 PASS, 49
physical processes, 17 molecules, multiple ionization of,
Poisson distribution, 18–19 253–263
Straight-line impact parameter model radiolysis, free radicals formation in,
(IPM), 209 261–263
Strong Potential Born (SPB) Water-equivalent ratios (WER), 157– 159
approximation, 364–365
X
T X-ray photoelectron spectroscopy, DNA
Threshold dose, 413–414 damages measurement, 82
Time-dependent density functional theory
(TDDFT), 316–318 Z
approach, 316–317 Zeiss and Meath (ZM) method, 67
computational aspects, 324–325 ZM method. See Zeiss and Meath (ZM)
results, 325–334 method

(Advances in Quantum Chemistry 65) Dževad Belkić (Eds.) - Theory of Heavy Ion Collision Physics in Hadron Therapy-Academic Press, Elsevier (2013) PDF

Uploaded by

Copyright:

Available Formats

(Advances in Quantum Chemistry 65) Dževad Belkić (Eds.) - Theory of Heavy Ion Collision Physics in Hadron Therapy-Academic Press, Elsevier (2013) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Advances in Quantum Chemistry 65) Dževad Belkić (Eds.) - Theory of Heavy Ion Collision Physics in Hadron Therapy-Academic Press, Elsevier (2013) PDF

Uploaded by

Copyright:

Available Formats

VOLUME SIXTY FIVE

Guillermina Esti ú (University Park, PA, USA)

Amsterdam • Boston • Heidelberg • London

First edition 2013

Copyright © 2013 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system or transmitted

For information on all Academic Press publications

Printed and bound in USA

description of cell survival valid at all doses as predicted by the new

Dževad Belkić, Guest Editor

Stochastics of Energy Loss and

The study consists of several parts:

2. ENERGY LOSS AT MACROSCOPIC LEVEL

2.1 Continuous transport of particles through matter:

2.2 Convolutions for range straggling

2.3 Straggling in thin segments

2.4 Monte Carlo methods

loss j of the particle is j = i Ei. To get Ei practically, the inverse function

E(�; βγ ) of �(E; βγ ) is calculated with e.g., cubic spline interpolation44 (see

** A study of the equivalence of the convolution and MC methods is being prepared.

In several systems of MC calculations a “condensed history” approach

3. BRAGG FUNCTIONS

In each track interval ξ the number N (y) is reduced by the number of

The above functions are shown in Figures 1.2–1.6.

3.1 Nuclear interactions

For water, N σn (T ) ∼ 0. 012/cm, for a mean free path  = 82 cm, for

3.2 Practical details for protons

3.3 Practical details for C-ions

The mean range is R0, it is given at N (y) = N (0)/2 and corresponds to

4. ENERGY LOSS AND DEPOSITION AT

primarily C-ion beams with kinetic energy T = 3600 MeV (or 300 MeV/u).

4.2 Energy deposition

4.3 Energy deposition in microscopic volumes

see Refs. 59, 61, 62.

In cylinder C the energy deposition is by 11 electrons filling volumes

5. STOCHASTICS OF ENERGY LOSS IN CELLS

Figure 1.7 Monte Carlo simulation, Section 2.3, of the passage of T = 16 MeV

(S(T ) = 1443 MeV/cm) and make a calculation for ND = 20 ions.‡‡‡‡

• Nuclear interactions in the cell layer are infrequent, Appendix D, and

5.2 MC simulation of stochastics at the Bragg peak

5.2.1 Simulation of random number nj of ions traversing cell j

5.2.2 Energy deposition z in cells

The distribution (“spectrum”) of z for K = 10000 cells at the Bragg

APPENDIX A. COLLISION CROSS SECTIONS CCS

A.1 Rutherford cross section

where EM ∼ 2mc 2 β 2 γ 2 is the maximum energy loss of the particle in one

APPENDIX B. BETHE–FANO COLLISION CROSS SECTION

the method for solids. In its nonrelativistic form it can be ­

where Q = q2 /2m, with q = K the momentum transferred from the inci-

and Eq. (B.1) then is written as

B.1 Fermi-virtual-photon (FVP) cross section

where the integral over Q in Eq. (B.4) is approximated by the square

loss measurements.89 A detailed description of the relativistic PAI model is

B.2 Collision cross sections for Si, water, and DNA

Inokuti and Paretzke,78, 92 and should be used for accurate calculations.The

B.3 Integrals: cumulative functions and moments for water

it to the description in Refs. 13, 57.

where N is the number of electrons per unit volume, and ν = 0, 1, 2, 3 . . .

The moment M1 (υ) is usually known as the Bethe–Fano stopping

2.1 Continuous transport of particles through matter:

loss j of the particle is j = i Ei. To get Ei practically, the inverse function

For water, N σn (T ) ∼ 0. 012/cm, for a mean free path = 82 cm, for

4. ENERGY LOSS AND DEPOSITION AT

the method for solids. In its nonrelativistic form it can be

where Q = q2 /2m, with q = K the momentum transferred from the inci-

3.2 Range measurements for water, and mean

4. OTHER TARGET SUBSTANCES AND STATISTICAL

2. DETAILED STEP BY STEP TRACK STRUCTURE

ρ ·VSV pSB nion

ρ ·VSV pSB nion