Martin Stutzmann, Christoph Csoklich - The Physics of Renewable Energy

Graduate Texts in Physics
Martin Stutzmann
Christoph Csoklich
The Physics
of Renewable
Energy
Series Editors
Kurt H. Becker, NYU Polytechnic School of Engineering, Brooklyn, NY, USA
Jean-Marc Di Meglio, Matière et Systèmes Complexes, Bâtiment Condorcet, Université
Paris Diderot, Paris, France
Sadri Hassani, Department of Physics, Illinois State University, Normal, IL, USA
Morten Hjorth-Jensen, Department of Physics, Blindern, University of Oslo, Oslo,
Norway
Bill Munro, NTT Basic Research Laboratories, Atsugi, Japan
Richard Needs, Cavendish Laboratory, University of Cambridge, Cambridge, UK
William T. Rhodes, Department of Computer and Electrical Engineering and Computer
Science, Florida Atlantic University, Boca Raton, FL, USA
Susan Scott, Australian National University, Acton, Australia
H. Eugene Stanley, Center for Polymer Studies, Physics Department, Boston
University, Boston, MA, USA
Martin Stutzmann, Walter Schottky Institute, Technical University of Munich,
Garching, Germany
Andreas Wipf, Institute of Theoretical Physics, Friedrich-Schiller-University Jena,
Jena, Germany
Graduate Texts in Physics publishes core learning/teaching material for graduate-
and advanced-level undergraduate courses on topics of current and emerging fields
within physics, both pure and applied. These textbooks serve students at the MS-
or PhD-level and their instructors as comprehensive sources of principles, defi-
nitions, derivations, experiments and applications (as relevant) for their mastery
and teaching, respectively. International in scope and relevance, the textbooks cor-
respond to course syllabi sufficiently to serve as required reading. Their didactic
style, comprehensiveness and coverage of fundamental material also make them
suitable as introductions or references for scientists entering, or requiring timely
knowledge of, a research field.
Martin Stutzmann · Christoph Csoklich
The Physics of Renewable

Energy
Martin Stutzmann Christoph Csoklich
Walter Schottky Institute Electrochemistry Laboratory
Technical University Munich Paul Scherrer Institut
Garching, Bayern, Germany Villigen, Switzerland
ISSN 1868-4513 ISSN 1868-4521 (electronic)

ISBN 978-3-031-17723-1 ISBN 978-3-031-17724-8 (eBook)
https://doi.org/10.1007/978-3-031-17724-8
© Springer Nature Switzerland AG 2022

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
Cover image: The Sun’s south pole as seen by the ESA/NASA Solar Orbiter spacecraft. These images
were recorded by the Extreme Ultraviolet Imager (EUI) at a wavelength of 17 nanometers. Permission
for reproduction is gratefully acknowledged.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
We dedicate this book to a sustainable future
of our planet Earth.
May it remain an inhabitable home for our
children, grandchildren, and many future
generations to come!
Indeed there is no planet B!
Preface
This book is based on the script of a lecture about renewable energy held regularly
at the Physics Department of Technische Universität München since 2007. Over
the years, the contents of this lecture were updated and extended to the present
version. Although originally intended for master students in engineering physics,
the lecture was also frequently followed by bachelor students and students of other
faculties interested in this topic. That is why the requirements for pre-existing
knowledge in physics vary from chapter to chapter. Some topics can be appreciated
with a general scientific background, while others require the knowledge of more
advanced physical concepts. In this way, the book hopefully appeals to a larger
community of potential readers.
The main aim of the book is to make interested scientists aware of the scien-
tific backgrounds, the potentials, but also the fundamental limitations of different
renewable energy technologies in terms of achievable energy densities, maximum
physical efficiencies, and future scientific and technological challenges. When-
ever possible, we have tried to keep units, notations, and nomenclature consistent
across the different forms of renewable energies, and to refrain from unphysical
prejudices and biases.
We gratefully acknowledge the constructive feedback and questions of many
hundred students over the years and the support of colleagues in this field by
making their results available for this book. And we thank PA for just being it!
Munich, Germany Martin Stutzmann

Villigen, Switzerland Christoph Csoklich
July 2022
vii
Contents
1 Energy—A Brief Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Energy and Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Potential and Kinetic Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Noether Theorem and Energy Conservation . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Inner Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 Quantifying Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Forms of Energy and Their Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 Mechanical Potential Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Kinetic Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Wave Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.1 Mechanical Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.2 Electromagnetic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Electrostatic and Magnetostatic Energy . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 Latent Heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.6 Chemical and Electrochemical Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.7 Nuclear Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.7.1 Fission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7.2 Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3 The Sun–Earth System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1 The Sun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1.1 General Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1.2 Details of Proton Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.1.3 Shell Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 The Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.1 General Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3 A Possible Energy Scenario Until 2050 . . . . . . . . . . . . . . . . . . . . . . . . . . 41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
ix
x Contents
4 Energy from Waves, Tides and Osmosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.1 Wave Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.1.1 Deep Water Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.1.2 Shallow Water Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2 Tidal Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2.1 Solar Tides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.2 Lunar Tides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Osmosis Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5 Wind Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1 General Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2 Energy Content of Wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.3 Efficiency of Wind Turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.4 Types of Rotors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.4.1 Drag-Type Rotors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.4.2 Lift-Type Rotors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.4.3 New Types of Wind Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.5 Optimization of Wind Turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.5.1 Optimized Radial Profile of a Lift-Type Blade . . . . . . . . . . . . 76
5.5.2 Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.6 Some Practical Aspects of Wind Engines . . . . . . . . . . . . . . . . . . . . . . . . 79
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6 Thermal Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.1 Geothermal Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.1.1 Contributions to Geothermal Energy . . . . . . . . . . . . . . . . . . . . . . 83
6.1.2 Use of Geothermal Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.2 Solar Thermal Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7 Photosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.1 General Considerations of Biomass Usage . . . . . . . . . . . . . . . . . . . . . . . 99
7.2 Biophysical Principles of Photosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.3 Basic Biomolecular Processes of Photosynthesis . . . . . . . . . . . . . . . . . 105
7.4 Details of Photon Absorption and Energy Transfer
in the Light-Harvesting Complexes of Photosystems . . . . . . . . . . . . . 110
7.5 Technical Use of Biomass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.6 Artificial Photosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
8 Photovoltaics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.1 General Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.2 Basic Processes in Photovoltaics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.2.1 Photons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.2.2 Photon Density of States (DOS) . . . . . . . . . . . . . . . . . . . . . . . . . . 122
8.2.3 Absorption, Reflection, Emission . . . . . . . . . . . . . . . . . . . . . . . . . 124
Contents xi
8.2.4 Thermalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

8.2.5 Recombination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
8.2.6 Separation and Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
8.3 Types of Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
8.3.1 Crystalline Si p/n Diffusion Cell . . . . . . . . . . . . . . . . . . . . . . . . . . 144
8.3.2 Metal-Insulator-Semiconductor (MIS)-Schottky
Contact Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
8.3.3 Amorphous Si Thin Film Drift Solar Cells . . . . . . . . . . . . . . . . 148
8.3.4 CdTe and Cu(In,Ga)Se2 Compound Thin Film Solar
Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
8.3.5 Dye-Sensitized Solar Cells (DSSC) . . . . . . . . . . . . . . . . . . . . . . . 150
8.3.6 Organic Bulk Heterojunction Cell . . . . . . . . . . . . . . . . . . . . . . . . 150
8.3.7 Perovskite Thin Film Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . 152
8.4 I-U-Characteristics of Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8.4.1 Ideal Diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8.4.2 Real Diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.5 Efficiency Limits of Single Junction Solar Cells . . . . . . . . . . . . . . . . . . 157
8.6 Increasing Solar Cell Efficiencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
8.6.1 Down Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
8.6.2 Tandem Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
8.6.3 Impurity-Band Photovoltaics (Optical Up-Conversion) . . . . 164
8.6.4 Impact Ionization (Carrier Multiplication) . . . . . . . . . . . . . . . . . 164
8.7 Energy Payback Times of Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
9 Thermoelectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
9.1 Basic Physics of Thermoelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
9.2 Thermoelectric Generators (TEGs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Knowledge Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Acronyms
ADP Adenosine diphosphate

AM Air mass
APS Artificial photosynthesis
ATP Adenosine triphosphate
AU Astronomical unit
BHJ Bulk heterojunction
BSF Back surface field
CAES Compressed air energy storage
CB Conduction band
CCS Carbon capture and storage
Chl Chlorophyll
CIS Copper-indium-diselenide
CSP Concentrated solar power
Cys Cysteine
Cyt bf Cytochrome bf
DOS Density of states
DSSC Dye-sensitized solar cell
EF Fermi level
ETP Electron transfer phosphorylization
Fd Ferredoxin
FF Fill factor
HAWT Horizontal axis wind turbine
HDR Hot dry rock
HOMO Highest occupied molecular orbital
IHP Inner Helmholtz plane
ITCZ Intertropical conversion zone
ITER International thermonuclear experimental reactor
ITO Indium-tin oxide
kgCE Kilogram of coal equivalent
kgOE Kilogram of oil equivalent
LHC Light-harvesting complex
LO Longitudinal optical
LUMO Lowest unoccupied molecular orbital
MIS Metal-Insulator-Semiconductor
xiii
xiv Acronyms
MPP Maximum power point

NADP Nicotineamid adenine dinucleotid phosphate
NHE Normal hydrogen electrode
OEC Oxygen-evolving center
OHP Outer Helmholtz plane
OPV Organic photovoltaics
PCy Plastocyanin complex
PECVD Plasma-enhanced chemical vapor deposition
PEFC Polymer electrolyte membrane fuel cell
Ph a Phaeophytin a
PQ Plastoquinone
PS I Photosystem I
PS II Photosystem II
Pyr Pyrrole ring
RC Reaction center
SC Solar constant
SHE Standard hydrogen electrode
SOFC Solid oxide fuel cell
TCO Transparent conducting oxide
TEG Thermoelectric generator
VB Valence band
Energy—A Brief Introduction
1
Abstract
This chapter deals with a theoretical introduction to the physical meaning of
energy and work. The principles of energy conversion from one form to another are
explained, as well as the impossibility of energy “destruction”. After some thermo-
dynamic considerations about energy, the Noether theorem is shortly explained.
The chapter finishes with common units of energy and some numbers, to get a
first impression of the magnitudes mentioned in this book.
The physical description of our world starts from the fundamental coordinates in
space, r, and time, t. For a moving object, the space vector is a function of time,
r(t), defining the trajectory of the object with the velocity v(t) as the time derivative
of r(t) (Fig. 1.1). Closely linked to the velocity is the momentum p = m · v of an
object with mass m. The space and momentum vectors form a pair of so called con-
jugated coordinates, which in quantum mechanics gives rise to the first Heisenberg
uncertainty relationship between space and momentum. Also the time t has such a
conjugated coordinate, namely the energy E, giving rise to the time-energy uncer-
tainty relationship in quantum mechanics. As we will briefly discuss further below,
the well-known conservation laws for momentum and energy also are the result of the
invariance of our world with respect to translations of the corresponding conjugated
coordinates space and time.
Of course, the concept of energy was developed long before modern theoretical
physics and quantum mechanics. The word energy comes from the Greek expression
en-érgeia (ν-ργ ια) which can be translated as acting force. A more modern
version of this definition would be (Fig. 1.1):
Energy is the variable of state of a physical system describing the ability of

the system to perform work on other physical systems via an interaction due
to forces between the systems.
© Springer Nature Switzerland AG 2022 1

M. Stutzmann and C. Csoklich, The Physics of Renewable Energy,
Graduate Texts in Physics,
https://doi.org/10.1007/978-3-031-17724-8_1
2 1 Energy—A Brief Introduction
Fig. 1.1 A point mass m at

position r moves along the
trajectory r(t) and changes
its original momentum due
the acting force F(r, t)
1.1 Energy and Work
The change in energy E of a system is given by the work W performed by an external

force F(r , t) acting on this system (cf. Fig. 1.1):
r2
dW = F(r, t) · dr ⇒ W1,2 = F(r, t)dr (1.1.1)
r1
Depending on the sign of dW , two qualitatively different cases can occur:
• dW > 0: if the external force has a component parallel to dr, dW is positive and
the energy of the system increases by transferring energy from a second, external
system from which the force originates.
• dW < 0: in this case, the energy of the system decreases by performing work on
a second system.
1.2 Potential and Kinetic Energy
Since according to the discussion above the nature of the interaction force determines
how the energy of a system will be changed, we will have a closer look at the kind
of forces encountered in nature. Here, one distinguishes between conservative and
non-conservative or dissipative forces. A force is called conservative, if the work
performed by it along any closed trajectory is 0:
r1
W = F(r, t) · dr = 0 (1.2.1)
r1
1.3 Noether Theorem and Energy Conservation 3
In this special case, the work W1,2 performed by the conservative force on a trajectory
between the point r1 and r2 will be independent of the specific trajectory chosen.
Then we can define a potential energy E pot such that
E pot = E pot (r2 ) − E pot (r1 ) = W1,2 (1.2.2)

F = −∇ · E pot (1.2.3)
with the gradient vector operator ∇:

∂ ∂ ∂
∇ = grad = , ,
∂ x ∂ y ∂z
Note that only the change E pot can be related to the measurable physical quantity
W1,2 , so that the origin of the potential energy scale can be chosen arbitrarily (or con-
veniently). Moreover, the potential energy provides a convenient way to reconstruct
the interaction force as the negative gradient of E pot . Unfortunately, most forces
encountered in reality are non-conservative, e.g. due to friction or turbulences.
A second interesting case is a force which accelerates a mass according to Newtons
second law F = ma. Then, a simple calculation leads to the definition of a kinetic
energy E kin :
r2 r2
dv
E kin = F(r, t) · dr = m · dr = (1.2.4)
r r dt
r2 1 r2 1
dr 1 1
mdv · = mv · dv = mv 2 (r2 ) − mv 2 (r1 ) (1.2.5)
r1 dt r1 2 2
If we now choose the origin of the kinetic energy as 0 in r1 , we can define the
kinetic energy as:
1 2
v1 = 0 : ⇒ E kin = mv (1.2.6)
2
As an important result, the Noether theorem discussed below states that in sys-
tems with only dissipation-less, conservative forces the total mechanical energy E
is conserved:
E = E kin + E pot = const. (1.2.7)
1.3 Noether Theorem and Energy Conservation
The conservation of energy is one of many conservation laws in physics. According

to the general theorem formulated by the mathematician Emmy Noether in 1918,
conservation laws are a consequence of symmetries (invariant operations) in a phys-
ical system [1]. The symmetries and corresponding conserved physical quantities
are summarized in Table 1.1.
Table 1.1 Symmetries and corresponding conserved physical quantities

Invariance Conserved quantity
Translation in 3D space Momentum mv
Translation in a periodic lattice Quasi-momentum k a
Translation in time Energy E
Rotation in 3D Angular momentum mr × v
Mirror operation Parity
Phase shift of a quantum-mechanical wave Electric charge q
function
a wavevector k
Here, we just give a brief sketch of this theorem as applied to an invariance of a

system under translation in time. The argument is based on the Lagrange formalism
in classical mechanics, where a mechanical system is described by the Lagrange
operator L(qi , q̇i ) with generalised coordinates qi and q̇i = dqi /dt, i.e. the total
derivative of qi with respect to time. The Lagrange operator is obtained from the
kinetic energy T and the potential energy V as L = T − V . For the simplest case of
a one-dimensional motion of a mass m this gives L(x, ẋ) = 21 · m ẋ 2 − V (x). Using
the variational principle of Hamilton, this leads to the Euler-Lagrange equation as
the generalized equation of motion:

d ∂L ∂L
= (1.3.1)
dt ∂ q̇i ∂qi
For our simple one-dimensional example, this just gives the Newton equation of
motion, F = ma. Now consider the following coordinate transformation:
qi (t) → qi (s, t) with qi (0, t) = qi (t) (1.3.2)
Taking the partial derivative of the Euler-Lagrange equation (1.3.1) with respect to
s and rearranging, this gives:

∂ d ∂ L (qi (s, t), q̇i (s, t)) ∂qk (s, t)
L (qi (s, t), q̇i (s, t)) = (1.3.3)
∂s dt
∂ q̇k (s, t)

∂s

I II
If the transformation is a symmetry operation which leaves the system invariant,

then the left side (part I ) must be zero, meaning that expression I I on the right
side is constant in time, i.e. is preserved. For a translation in time, the coordinate
transformation to be considered is qi (s, t) = qi (s + t). From the Noether theorem
then follows:
1.4 Inner Energy 5

d ∂ L (qi (s, t), q̇i (s, t))
q̇k (t) − L (qi (t), q̇i (t)) = 0 (1.3.4)
dt ∂ q̇k (t)
Here, the expression within the brackets is the total energy of the system, which
accordingly is conserved. This again can be seen using the simple example of a free
mass m without a potential, V (x) = 0:
1 2 ∂L 1 1
L= m ẋ ; ẋ − L = m ẋ 2 − m ẋ 2 = m ẋ 2 = E kin (1.3.5)
2 ∂ ẋ 2 2
1.4 Inner Energy
In general, processes on earth are non-conservative and irreversible due to friction,

turbulences, mixing, etc. Then, on a macroscopic level, energy conservation seems
to be violated, because the mechanical energy E kin + E pot is no longer constant.
Instead, it is partially transformed into inner energy E I (or heat Q) stored in the
interacting atoms of a macroscopic system. This inner energy is the sum over the
(kinetic and potential) energy of every single atom with respect to the center of
gravity R S of the system:

EI = E i (r − R S ) (1.4.1)
i
The conversion of macroscopic potential or kinetic energy to inner energy by

dissipation can occur to a 100% (e.g. the kinetic energy of a moving car is completely
transformed to heat by the brakes), while the conversion of inner energy or heat back
to macroscopic energy is limited by the laws of thermodynamics, in particular by the
Carnot efficiency of heat engines. Thus, the second law of thermodynamics states
that it is impossible to realise a cyclic heat engine which transforms heat into work
with an efficiency of 100%. This fact motivates the distinction of two components
of energy with different properties for practical use:
Exergy is the part of the total energy of a system which can be converted into useable
macroscopic energy or work.
Anergy is the part of the total energy which cannot be converted into useable work.
For reversible processes, the relative amounts of exergy and anergy are conserved.
Irreversible processes involving friction, turbulences, etc. cause an irreversible con-
version of exergy to anergy, meaning a loss in useable energy.
Fig. 1.2 Schematic energy

flow in a cyclic heat engine.
A cyclical process takes a
heat Q W from a hot reservoir
at an absolute temperature
TW . Part of this heat is
transformed into usable work
W , whereas the rest flows as
waste heat Q C into a cold
reservoir at temperature
TC < TW
Consider a simple case: When a stone falls down onto the Earth from a height
h 0 , its original macroscopic potential energy E pot = mgh 0 is transformed con-
tinuously into macroscopic kinetic energy E kin . If we assume negligible fric-
tion with the surrounding air during the fall, there will be no increase of the
initial temperature T0 of the stone and, thus, no energy dissipation into the
inner energy of the stone. Thus, during the fall, all atoms of the stone will have
on average the same velocity v as the center of gravity of the stone. When the
stone hits the ground in an ideal inelastic way, E kin becomes abruptly 0 (due to
v = 0) and is completely transformed into inner energy E I . This corresponds
to an increase in temperature: T1 > T0 so that E = mgh 0 = E I . This increased
temperature T1 of the stone could be used by a heat engine to lift the stone back
up. However, due to the limited efficiency of the engine, the final reachable
height h 1 will be smaller than the original height h 0 from which the stone has
fallen.
The efficiency of the heat engine shown in Fig. 1.2 is limited by the Carnot effi-
ciency:
|W | |Q W | − |Q C | TW − TC
η= = ≤ (1.4.2)
|Q W | |Q W | TW
For irreversible processes the anergy is lost as an the increase of entropy (d S =

d Q/T )
QC
E an = TC S = TC = QC (1.4.3)
TC
1.5 Quantifying Energy 7
1.5 Quantifying Energy
The derived SI unit for energy is the Joule:
J = kg m2 s−2 (1.5.1)
However, depending on the scientific or practical context, a wide variety of other

energy units is still in use. This is also due to the vast range of energy magnitudes
that need to be described. Very common in physics is the unit electron volt (eV), to
quantify e.g. energy scales for atoms or fundamental particles, whereas the energy
produced by the Sun per second (around 45 magnitudes higher) can be better quan-
tified in yottajoule, (YJ). The calorie (cal), now mostly used in the description of
the energy content of food, was also popular in chemistry and is sometimes still in
use. In engineering some other energy units, such as e.g. the kilowatt hour (kW h),
are common (1 kW h corresponds to about 3.6 × 106 J). In the general public, quite
often a confusion exists between energy and power: power is the change in energy
per unit time, leading to its unit, the Watt (W).
W = J s−1 (1.5.2)
The common confusion of kilowatt (a power unit) and kilowatt hour (an energy
unit) should be carefully avoided. The Tables 1.2 and 1.3 show the relations of
the most common physical/chemical units. Other units common in energy tech-
nology are the British Thermal Unit 1 BTU = 1055 J, the (kilogram of) Oil Equiva-
lent 1 kgOE = 4.2 × 107 J = 11.63 kW h, (kilogram of) Coal Equivalent 1 kgCE =
2.93 × 107 J = 8.141 kW h, or the horse power 1 PS = 735.5 W as a publicly used
unit of power. On the global scale, the terawatt year (TWa) is another useful energy
unit.
Table 1.4 shows a brief overview of how much energy is consumed currently on
different scales by human beings compared to the energy typically available from
Table 1.2 Important physical and chemical units of energy

J eV cal
J 1 6.24 × 1018 0.239
eV 1.602 × 10−19 1 3.83 × 10−12
cal 4.1855 2.61 × 1019 1
Table 1.3 Important technical units of energy

kW h TWa kgOE kgCE
kW h 1 1.14 × 10−13 8.60 × 10−2 0.123
TWa 8.76 × 1012 1 7.53 × 1011 1.08 × 1012
kgOE 11.63 1.33 × 10−12 1 1.43
kgCE 8.141 9.28 × 10−13 0.7 1
Table 1.4 Comparison of the magnitudes of energy consumption and renewable energy supply on
Earth
Power/Energy
Basic metabolic rate of a human being 80 W = 700 kWh/a = 86 kgCE/a
Basic metabolic rate of the world population 7 × 1011 kgCE/a
Average energy consumption in 5 kW = 45 MWh/a = 5500 kgCE/a
Germany/capita
Total energy consumption per year
GER (82 M inhabitants) 5 × 1011 kgCE
US (330 M) 3 × 1012 kgCE
Chad (10 M) 9 × 108 kgCE = 90 kgCE/capita
Total solar irradiation 1.7 × 1017 W = 1.8 × 1017 kgCE/a
Including direct radiation energy 3 × 1016 kgCE/a (18%)
Wind, water, waves 5 × 1016 kgCE/a (30%)
Biomass 2 × 1014 kgCE/a (0.1%)
Earth heat 3.5 × 1013 kgCE/a (0.02 %)
Tidal wave energy 3.5 × 1012 kgCE/a (0.002 %)
Total renewable energy production in GER 6 × 1010 kgCE/a (≈12%)
different renewable energy sources. The minimum power to keep us alive is our
basic metabolic rate of about 80 W [2]. Over the time of a year, this sums up to an
energy demand of 700 kWh/a or 86 kgCE/a. Thus, we would need about our body
weight in coal per year to simply exist. For the current Earth’s population of about 8
billion people, the accumulated minimal energy demand would be 7 × 1011 kgCE/a.
However, especially in rich countries much more energy is consumed to satisfy
our current standards in mobility, health, culture or luxury. Thus, in Germany we
presently use on the average a constant power of 5 kW per capita instead of the 80 W
we really need [3]. In other countries like the US this total power consumed per
person is even larger (currently around 10 kW). Only in very poor countries like the
Chad the actual power demand is close to the basic metabolic rate.
On the other side, the Sun supplies us with an enormous power of solar irradiation
of continuously more than 2 × 1017 W. The solar energy arriving on Earth within
one hour would be enough to satisfy our current global energy demand for an entire
year. Of course, not all of this incoming solar energy is readily available for our
energy supply. Almost 50% of the incoming radiation is radiated back into space by
reflection at the atmosphere or thermal radiation of the Earth. But the other 50% could
be harvested by us using direct solar radiation via solar cells, wind engines, water
power plants, biomass etc. Compared to that, other renewable energy sources like
geothermal energy or tidal energy are almost negligible. But overall, there is basically
no physical limitation to satisfy our energy demands from renewable energy sources
if we really want to. This means that there is still a lot of room for improving the
current contribution of about 12% of renewable energy to the total energy production
in e.g. Germany, and the same holds for the entire Earth as a whole. To convey a
References 9
feeling of which physical restrictions we will have to beat or live with in this challenge
is what this little book is all about.
Further Reading
• Our world in data: research and data to make progress against the world’s largest
problems: ourworldindata.org. Accessed from 26 July 2022.
• Data on the German electricity marked: smard.de. Accessed from 26 July 2022.
• Worldwide total primary energy supply (TPES) by source:
www.iea.org/data-and-statistics/. Accessed from 26 July 2022.
• BP Statistical Review of World Energy:
https://www.bp.com/en/global/corporate/energy-economics/statistical-review-
of-world-energy.html. Accessed from 30 July 2022.
References
1. Noether, E.: Invariante Variationsprobleme, Nachrichten von der Gesellschaft der Wissenschaften
zu Göttingen. Mathematisch-Physikalische Klasse 1918, 235–257 (1918)
2. Mifflin, M.D., et al.: A new predictive equation for resting energy expenditure in healthy indi-
viduals. Am. J. Clin. Nutr. 51, 241–247 (1990)
3. BP: Statistical Review of World Energy 2021, vol. 70 (2021)
Forms of Energy and Their Density
2
Abstract
In this chapter, we recall briefly different forms of energy in view of their potential
for energy storage or transmission. Although most physics students will be familiar
with most of the formulae below for the different energies E, they will be much
less familiar as far as the related achievable volumetric energy densities E/V are
concerned. So it is worthwhile to have a closer look at this.
2.1 Mechanical Potential Energy
As already introduced in Sect. 1.2, potential energy can be defined in systems sub-
jected to a conservative force field. The gravitational field is conservative to a very
good approximation and, thus, a potential energy between different points in the
gravitational force field can be defined. In the simplest case this gives:
E pot = m · g · h (2.1.1)
Here m denotes the mass of an object in the Earth’s gravitational field, g the gravi-
tational acceleration,1 and h the height of the object above ground as the arbitrarily
chosen zero point (h should be small enough so that g still can be considered con-
stant).
1 Inthe following examples we use the approximated value for the gravitational acceleration at the
Earth’s surface of g = 9.81 m s−2 .

12 2 Forms of Energy and Their Density
Fig. 2.1 Illustration of the

volume work required to
compress gas
As an example, 1 t of water is carried up a mountain to a height of 100 m:
E pot = mgh ≈ 1 × 106 J
This is equivalent to 0.3 kWh/250 kcal/0.04 kgCE and corresponds to an

energy density (assuming ρ = 0.998 g cm−3 for the density of water) of only
E/V = 1 × 106 J m−3 (2.1.2)
As a consequence, the energy density in pumped hydro-storage is quite low.

Therefore, reservoirs require huge volumes and big enough falling heights to achieve
significant storage capabilities.
Another storage system using the potential energy difference between two states
is the compressed air energy storage (CAES). The compression and expansion of air
is used to take up excess energy e.g. from a wind park and return it to the electrical
grid when demanded. For this, a turbine operates in connection with an underground
cavern.
The compression of an ideal gas follows the ideal gas equation:
pV = N kT = ν RT (2.1.3)
with the mol number ν, the Boltzmann constant k, and the gas constant R. For
compression we have to perform volume work on the system (cf. Fig. 2.1):
dW = −Fdx = − p Adx = − pdV
If a gas is very well insulated from the environment, then the compression
occurs adiabatically: the heat Q exchanged between the gas and the sur-
rounding is ≈ 0. This is a good approximation for a CAES system, since Earth
is a bad thermal conductor.
2.1 Mechanical Potential Energy 13
For a more detailed calculation of the adiabatic compression, we start from the
first law of thermodynamics:
dU = δ Q + δW (2.1.4)
V
Here, U is the inner energy of the gas, W is the mechanical work W = V12 − pdV
upon volume change, Q is the heat, here defined by the heat capacity C: Q =
C · T . More exactly, the heat capacity depends on whether the inner energy U is
changed under the condition of constant volume or constant pressure:

dU 3 3
CV = = kN = νR and (2.1.5)
dT V 2 2

dU 5 5
CP = = k N = ν R = CV + ν R (2.1.6)
dT P 2 2
For an adiabatic compression of the gas we then have:

δ Q = 0 = dU − δW = C V dT + pdV p = ν RT
V

dV
C V dT + ν RT =0 : CV T
V

dT ν R dV νR C P − CV
+ =0
T CV V C = CV
=γ −1
V
with γ , the adiabatic coefficient:
CP
γ = (2.1.7)
CV
Integration then gives:
ln T + (γ − 1)lnV = const
ln(T V γ −1 ) = const

TV γ −1
= const T = pV
νR
This leads to the equation of adiabaticity:
pV γ = const. (2.1.8)
Now consider the adiabatic compression of a gas from {T0 , p0 } to the final con-
dition {T1 , p1 } with r = p1 / p0 as the ratio of compression:
γ γ −1
T1 p1
= (2.1.9)
T0 p0
Thus, the amount of energy stored in the compressed gas is:

T1 1− 1
E = C V (T1 − T0 ) = C V T0 − 1 = C V T0 (r γ − 1) (2.1.10)
T0
The adiabatic coefficient of dry air is γ = 1.4. We start from a temperature

of T0 = 0 ◦ C and a pressure of p0 = 1 bar. The heat capacity of air is C V =
21 J K−1 mol, where 1 mol of air at the starting conditions has a volume of
22.4 L. From this, we can calculate now the energy density:
E 1
= 0.07(r − r γ )kW h m−3 (2.1.11)
V
This expression depends only on the ratio of compression r . Assuming a mod-
erate value of r = 30 for the compression ratio, we obtain an energy density
of:
E/V = 1.3 kW h m−3 = 5 × 106 J m−3 (2.1.12)
To store 1 GW h we then need a volume of 106 m3 , which corresponds to a

relatively small cubic cavern of size 100 m × 100 m × 100 m.
2.2 Kinetic Energy
As discussed in Sect. 1.2, kinetic energy is the energy stored in the linear motion of
an object, described by the translational velocity v:
1 2
E kin = mv (2.2.1)
2
2.2 Kinetic Energy 15
As an example, consider a car with a mass of m = 1000 kg, driving with a

speed of v = 100 km h−1 = 28 m s−1 . This gives a kinetic energy of:
E kin = 4 × 105 J ≈ 0.1 kW h = 8.6 gOE
So, the entire kinetic energy of a fast-moving car is only equivalent to the energy
content a few mL of oil. Similar to our example for potential energy above, this again
shows how little energy is contained in a mechanical form compared to a chemical
form of energy.
In addition, also rotating extended objects have kinetic energy in the form of
rotational energy. Using v = ω · r for the velocity of a mass element rotating with
angular velocity ω at a distance r around a fixed rotation axis, integration over the
rotating object gives for the rotational energy:
1 2
Er ot = Iω (2.2.2)
2
Here, I is the moment of inertia of the rotating object obtained by integration over
all mass elements.
An interesting example is the rotational energy of the Earth. Approximating

the Earth by a homogeneous sphere, we have I = 2/5mr 2 , with m = 5.9 ×
1024 kg, and r = 6.4 × 106 m. The angular velocity is given by ω = 2π/24 h =
7.3 × 10−5 s−1 . Inserting this into the above equation, we find :
Er ot = 2.5 × 1029 J = 8 × 109 TWa (2.2.3)
This is a huge energy when compared to the total energy consumption of mankind
of about 15 TWa. However, it is very difficult to use this energy source for our needs.
This can only be accomplished by tidal energies as discussed in Chap. 4.
Rotational energy storage is practically used in flywheel farms e.g. for server
facilities as a quickly accessible backup energy source. Similarly, the linear kinetic
energy of a moving car could be transferred to or from a flywheel storage system
during braking or acceleration. The requirements to store the kinetic energy of our
car above in a flywheel can be easily calculated:
Assume a flywheel in form of a cylindrical torus with mass m = 50 kg and

radius r = 50 cm. To store the linear kinetic energy of the moving car discussed
above, this flywheel would have to rotate at about ω = 270 s−1 ≈ 2500 rpm, a
value which could cause serious problems for the driving behaviour of the car
because of unwanted gyroscopic effects of the flywheel.
2.3 Wave Energy
In a mechanical oscillator like a pendulum, potential and kinetic energy are periodi-
cally converted into each other. This also is the case for waves in coupled mechanical
oscillators. Such waves have the interesting property that they can transport energy
from one place to another without a corresponding transport of mass. We will have
a closer look at water and tidal waves in Chap. 4. Here we just briefly summarize
some basic aspects of their energy content.
2.3.1 Mechanical Waves
Mechanical waves like water waves have an energy density per unit wave front given
by (cf. Sect. 4.1):
1
E= ρlin ω2 A2 l (2.3.1)
2
Here, ρlin is the linear mass density of the medium (in units [kg m−1 ]), A is the wave
amplitude and l the length of the wave front.
A water wave with amplitude A = 1 m and frequency f = 0.5 Hz has a linear

power density (per m wave front) of:
dP
≈ 2000 W m−1 (2.3.2)
dx
This may appear a lot for a human being enjoying waves at a beach, but for
technical use this is again a very small density: to construct a wave power plant with
a total power of 1 GW (i.e. the power of a typical nuclear plant) we would need
500 km of coastline.
2.4 Electrostatic and Magnetostatic Energy 17
2.3.2 Electromagnetic Waves
Electromagnetic waves, especially in the form of microwaves or light, are an interest-

ing technical alternative to transport energy (and information) without a mechanical
connection or an electric transmission line. After all, the Earth is powered by the Sun
in this way, as we will discuss in detail in Sect. 3.1. From the Maxwell equations we
find for the energy density of an infinite electromagnetic plane wave with the electric
field E and magnetic field H :
E 1 1 1
E · D + H · B)
= εε0 E2 (t) + μμ0 H 2 (t) = (E (2.3.3)
V 2 2 2
To transmit a power of 1000 mW through vacuum or air with a microwave beam
at a frequency of 2.45 GHz, the electrical field strength has to be approximately
1 × 104 V m−1 .
2.4 Electrostatic and Magnetostatic Energy
Also static electric and magnetic fields can store a considerable amount of energy.
For electric fields E , this can be most easily seen by considering a plate capacitor with
capacitance C which is charged up to a voltage U , corresponding to a total stored
charge qtot = C · U . By integration from 0 to U , this gives for the stored energy:
1 1
E= CU 2 = qtot U (2.4.1)
2 2
Using the well known formula for the capacitance C of a parallel plate capacitor
with plate area A and plate distance d and replacing the voltage U by the electric
field E = U /d, we can easily calculate the energy density E/V :
A
C = εε0 (2.4.2)
d
E 1
= εε0E 2 (2.4.3)
V 2
Note that this agrees also with the energy density of the electrical part given above
for electromagnetic waves. A very popular application of this form of energy storage
are supercapacitors or ultracapacitors with a very high capacitance C. This can be
achieved by using dielectric insulators with a high relative dielectric constant ε (high-
k dielectrics), by minimizing the effective plate distance d (e.g. by making use of
the Helmholtz layer at a metal/electrolyte interface, see below), and by maximizing
the effective plate area A using dense arrays of conducting nanowires or nanoporous
metallic foams instead of flat metal plates.
Consider a commercial supercapacitor with the dimensions 10 cm × 10 cm ×

0.5 mm, a mass of 15 g and a capacitance of C = 30 F. This supercapacitor
can be charged up very rapidly to a voltage of Umax = 3 V and then stores an
energy of E = 135 J = 4 mgSCE. The corresponding energy density is then:
E
= 3 × 107 J m−3 (2.4.4)
V
Note, that this is 30 times higher than the energy density of a ton of water in a
height of 100 m discussed in the beginning of this chapter.
In an analogous way we can calculate the energy stored in the static magnetic
field H of a bobbin with N windings, length l and cross-section area A conducting
an electric current I . The relevant physical quantities for this case are the inductance
L of the bobbin and the magnetic flux φ generated by the flowing current I . Then
the following relations hold:
1 2 1 A
E= L I = I , L = μμ0 N 2 , = N BA (2.4.5)
2 2 l
1 2 1
E = B l A/(μμ0 ) = μμ0 H 2 V (2.4.6)
2 2
E 1
= μμ0 H 2
(2.4.7)
V 2
Superconducting magnets have a Bmax of ca. 20 T (1 T = 1 V s m−2 ), limited

by the critical magnetic field of the superconductor used. For μ = 1 we find
−2 (for comparison: the energy density of a Li-ion battery
V ≤ 1.5 × 10 J m
E 8
9 −3
is about 10 J m ).
2.5 Latent Heat
Latent heat is a characteristic form of inner energy required for a phase change of
condensed matter: melting of a solid or evaporation of a liquid as endothermic phase
changes, and solidification of a liquid or condensation of a gas as the corresponding
exothermic processes. A practical example is the use of water freezing to prevent
the temperature in a room to fall below 0 ◦ C. The water/ice phase transition has a
latent heat of 335 J g−1 , corresponding to a volumetric energy density of E/V =
2.6 Chemical and Electrochemical Energy 19
Fig. 2.2 Melting temperatures TM and the corresponding latent heat densities HM for different
metals and compounds
3.3 × 108 J m−3 . Thus, a 10 L bucket of water can provide a latent heat of about
1 kW h or 100 gCE upon freezing.
For more modern applications of latent heat energy storage e.g. in fuel cell hous-
ings or solar thermal power plants, other materials such as metals or special “phase
change salts” (in particular metal fluorides) are used. The idea is to have the phase
change occurring at a specific temperature level in order to thermally stabilize a sys-
tem at this temperature, or to allow a high thermodynamic Carnot efficiency for the
transformation of the stored latent heat into usable exergy. Figure 2.2 shows some
examples of phase change materials for different applications.
2.6 Chemical and Electrochemical Energy
Chemical energy in the form of different fuels is the most common way in which we
store and transport energy today. The energy in the fuel is liberated by exothermic
chemical reactions, most commonly by oxidation with O2 from the air (burning).
The microscopic origin is the reduction of electronic potential energy of the reacting
atoms or molecules. The simplest example is the formation of H2 molecules from two
hydrogen atoms shown in Fig. 2.3. We start from two single H atoms whose energy
is defined as zero. For large distances r between these atoms, their interaction is
negligible. When the atomic 1s wave functions start to overlap for small r , both
electrons can lower their energy because they become less confined in space. When
the equilibrium distance r0 of the two protons in the H2 molecule is reached, the stable
bond is established with binding energy E(r0 ). The bonding orbital of the molecule is
given by a symmetric combination of the two 1s orbitals with antisymmetric spins. It
has a non-zero value of the squared electronic wave function between the two nuclei
characteristic for a covalent bond. In contrast, the high energy antibonding level
results from an antisymmetric combination of the two atomic wave functions with
a characteristic zero-crossing between the nuclei and a symmetric spin alignment.
Fig. 2.3 Schematic view of the formation of a H2 molecule from two hydrogen atoms. Left: Energy
diagram of the total energy versus the internuclear distance r . The minimum corresponds to the
binding energy stored in the bond and determines the equilibrium distance r0 between the atomic
nuclei after molecule formation. Right: Schematic sketch of the 2 wave functions and energy
levels of the electrons in the atoms and in the molecule. Arrows indicate the electronic spins
In the molecular ground state, the two electrons from the two hydrogen atoms only
occupy the bonding orbital. When one or both electrons are excited to the antibonding
level e.g. by heat or optical excitation, the bond becomes unstable and the molecule
can dissociate again.
The characteristic binding energy of a hydrogen molecule is 4.4 eV. Other

important exothermic reactions are the burning of carbon or the oxidation of
hydrogen to water:
H + H −→ H2 + 4.4 eV
C + O2 −→ CO2 + 4.2 eV
H2 + O −→ H2 O + 3 eV
In a macroscopic chemical reaction, the binding energy is dissipated in the form

of heat. This occurs through the interaction of a newly formed molecule with the
environment via its translational, rotational and internal degrees of freedom. For
the H2 molecule discussed above, these degrees of freedom are the translations in
the three orthogonal directions in space, rotation about the two orthogonal axes with
finite momentum of inertia perpendicular to the bond axis, and the internal symmetric
bond stretching vibration in the bond axis around the equilibrium distance r0 .
The unique advantage of chemical energy compared to all other forms of energy
discussed so far is that the corresponding typical energy densities are very high:
For example, a hydrogen storage tank with compressed H2 at 700 bar for the
oxidation to water in a fuel cell has an energy density of:
E/V = 5 × 109 J m−3
Another order of magnitude higher is the energy density of gasoline for com-
bustion in a car engine:
E/V = 5 × 1010 J m−3
Of great practical importance is also the direct conversion of chemical energy

into electrical energy, e.g. in batteries or fuel cells. Here, three different types of
converters are distinguished:
primary electrochemical elements: non-chargeable batteries
secondary electrochemical elements: rechargeable batteries
tertiary electrochemical elements: with input of external masses, e.g. fuel cells
All these electrochemical elements consist of combinations of metal electrodes
with electrolytes as ionic conductors. As shown in Fig. 2.4, at the metal/electrolyte
interface positive or negative metal ions are dissolved to a certain extent in the elec-
trolyte, while the electronic countercharge (electrons or holes (missing electrons))
remains in the metal. As a consequence, an electric field builds up between the elec-
trode and the electrolyte, which increases until it prevents a further dissolution of
metal ions and an equilibrium is obtained [1,2]. The interfacial electric field leads to
Fig. 2.4 The electrochemical double layer at an (metal) electrode/aqueous electrolyte interface.
Ions in the electrolyte are surrounded by a solvation shell of electrolyte molecules, lowering their
energy
Table 2.1 Galvanic series of some common elements

Element Ox. State U0 [V] Element Ox. State U0 [V]
Au 1 +1,69 Fe 3 −0,04
Cl 1 +1,35 Pb 2 +0,13
Au 3 +1,40 Sn 2 −0,14
Pt 2 +1,18 Ni 2 −0,26
Ag 1 +0,80 Fe 2 −0,45
Graphite 2 +0,75 Zn 2 −0,76
Cu 1 +0,52 Al 3 −1,66
O2 2 +0,39 Na 1 −2,71
Cu 2 +0,34 K 1 −2,93
Sn 4 +0,02 Li 1 −3,04
H2 1 +−0,00
a macroscopic potential difference between the metal electrode and the bulk of the
electrolyte with a characteristic galvanic voltage U0 . This voltage is measured under
standard conditions (room temperature, 1 M electrolyte) against a standard hydrogen
electrode (SHE) as the zero point of the voltage scale. Table 2.1 lists the galvanic
voltages for different metals and gases in different oxidation states.
Figure 2.4 provides a closer look at the structure and the potential profile in the so-
called electrochemical double layer as a function of the distance x from the electrode
surface. First, dissolved ions can adsorb directly at the electrode surface, defining the
inner Helmholtz plane (IHP). In polar solvents like water, however, most dissolved
ions are surrounded by a solvation shell of solvent molecules as indicated on the
left side of the figure. If a is the diameter of the solvation shell, the so-called outer
Helmholtz plane (OHP) due to adsorbed solvated ions occurs at x = a/2. In this
region, the double layer can be described like a plate capacitor, so that the potential
φ decreases linearly with distance x. Beyond the OHP, ions can diffuse, but still
feel a finite voltage U , which influences their local concentration c and, therefore,
the potential profile via the Nernst equation for ions with charge ±ze at an absolute
temperature T :

zeU
c = c0 exp − (2.6.1)
kT

kT c
U =− ln (2.6.2)
ze c0
According to Eq. 2.6.2, the concentration c of singly charged ions (z = 1) at room
temperature (T = 300 K) varies by one decade when U varies by 59 mV. Finally,
the potential becomes constant in the bulk electrolyte. The overall potential drop
φ corresponds the measured or applied external voltage U between the electrode
and the bulk electrolyte. The overall extent of the electrochemical double layer with
varying potential depends on the ion concentration and the voltage and usually is as
small as a few nanometers.
Fig. 2.5 Schematic view of a

lithium ion battery and the
directions of ionic and
electronic transport during
charging and discharging
By combination of two different electrodes and a suitable electrolyte, complete

electrochemical elements can be constructed. As an example, we show below a
schematic view of a rechargeable Lithium-polymer battery widely used today in
portable devices. One electrode consists of a Li-rich metallic compound such as
LiMn2 O4 , the other is made from graphite. Both electrodes are separated by a thin
polymeric electrolyte through which Li+ ions can move easily. When the battery is
charged, an external voltage is applied which causes Li ions to flow from the metal to
the graphite electrode, where they can intercalate easily between the stacked graphene
layers. To maintain charge neutrality of each electrode, a corresponding amount of
electrons is pumped from left to right through the external circuit. Upon discharge,
this process is inverted as the ions move back to the metal electrode where they have
a lower energy (Fig. 2.5).
A particular important property of batteries today is the amount of energy stored
per unit volume or per unit mass. Both parameters should be as high as possible
for practical use in portable electronics or e-mobility. Figure 2.6 shows the range of
Fig. 2.6 Volumetric and gravimetric energy densities of past, present and future primary and sec-
ondary battery technologies (Ni–MH: nickel-metal hydride, NCM–Si: nickel-cobalt-manganese
oxide-silicon)
Table 2.2 Properties of popular primary batteries

Type Zinc manganese oxide Zinc/air Li-ion (non-rechargable)
Anode Zn + 4 OH− 2 Zn + 8 OH− xLiC6
−→ [Zn(OH)4 ]2− + 2 e− −→ 2 Zn(OH)2 −4 +4 e− −→ xLi+ + xe− + xC6
Cathode MnO2 + H2 O + e− O2 + 2 H2 O + 4 e− Li1−x CoO2 + xLi+ +
xe−
−→ MnO(OH) + OH − −→ 4 OH − −→ LiCoO2
Energy 0.5 MJ kg−1 1.59 MJ kg−1 0.875 MJ kg−1
density
Cell voltage 1.5 V standard alkaline 1.4 V hearing devices 3.6 V mobile and
application battery automotive applications
Table 2.3 Comparison of two commercially available fuel cell systems [4]
PEFCa SOFCb
Operating temperature <100 ◦ C 900–1000 ◦ C
Efficiency 60% 60%
Fuel H2 , methanol, etc. H2 , CO, CH4 , etc.
Fuel reforming External Internal
Anode material Pt, Pt − X (X = Ru, Ni, Co) Ni − YSZ-Cermet
Cathode material Pt, Pt-X (X=Ru,Ni,Co) La(Sr)MnO3
Electrolyte PEM (e.g. Nafion®) ZrO2 /Y2 O3 -ceramics
Typical system power 10–100 kW Some 100 kW
Application e-mobility, domestic power Emergency power supplies
units
a Polymer electrolyte membrane fuel cell
b Solid oxide fuel cell
energy densities achieved in different battery systems today and expected densities
in future technologies. To make connection with the volumetric energy densities for
different forms of energy discussed so far in this chapter, we note that 100 W h L−1
in Fig. 2.6 are equivalent to 3.6 × 108 J m−3 , so that the volumetric energy density
of current Li ion batteries range between 0.7 × 109 to 2 × 109 J m−3 .
Without further discussion, we summarize in Tables 2.2 and 2.3 some properties
of commonly used primary (non-rechargeable) batteries and fuel cells. As a general
remark, electrochemical energy storage and conversion is probably the most dynamic
field of research and development in the context of renewable and conventional
energy, so that an up-to-date coverage is far beyond the scope of this book [3].
2.7 Nuclear Energy
The power generation by the use of nuclear energy is based on the different binding
energies of nucleons (protons (Z ) and neutrons (N )) in atomic nuclei by the strong
interaction. The different binding energies for each element and some isotopes are
2.7 Nuclear Energy 25
Fig. 2.7 Binding energy per nucleon as a function of the atomic number. The maximum occurs at
Fe (Z = 56)
shown in Fig. 2.7. The maximum is reached for 56 Fe. For lighter elements, i.e. on
the left side of this maximum, energy can be liberated by the fusion of two lighter
nuclei to a heavier one, on the right side by fission of a heavy nucleus into mostly
two lighter nuclei.
2.7.1 Fission
Nuclear fission, discovered by Otto Hahn and Lise Meitner in 1938, lead to the
fast development of the atomic bomb as well as to nuclear power plants, of which
currently 440 are in operation worldwide, producing an electric power of about
390 GW. Mainly 235 U is used as the fuel in these reactors. It is split by slow neutrons
into two different lighter daughter nuclei and sets free a vast amount of energy. The
basic process follows the reaction:
n +235 U −→ T1 + T2 + 2.43 n+ ∼ 200 MeV

During the fission of one uranium nucleus, on the average 2.43 neutrons are
produced which can be used to maintain the fission process either in a catastrophic
chain reaction or, by suitable moderation, in a constant fission rate.
Fig. 2.8 Potential well model of a 7 Li nucleus. Neutrons and protons are spin 1/2 fermions and
successively occupy discrete energy levels in the nucleus until the total numbers N and Z are
reached. The Coulomb interaction of charged protons gives rise to the Coulomb wall, which acts
as an additional energy barrier for protons approaching the nucleus from the outside, which has to
be overcome, e.g. in a fusion process
The energy density of natural uranium (99.27% 238 U and only 0.72% 235 U) is
very high:
1 gtSCE =
ˆ 53 ktU or 1 kgU =
ˆ 19 000 kgSCE
The energy density of pure 235 U is about
E/M = 7.9 × 1013 J kg−1 or E/V = 1.5 × 1018 J m−3
2.7.2 Fusion
Nuclear fusion of light nuclei in hot plasmas for power generation is still under devel-
opment. As shown by the energy diagram of Fig. 2.8, fusion reactions between light
nuclei are driven by the gain in potential energy of the strong interaction between
nucleons (neutrons and protons), but are kinetically hindered by the Coulomb repul-
sion between the positively charged protons in both nuclei approaching each other.
This gives rise to the so-called “Coulomb wall” in the energy diagram, which has to
be overcome for the fusion of charged nuclei. The outlook of cleaner nuclear energy,
less radiation and shorter half-times of the radioactive reaction products as well as
the large fuel (deuterium and tritium2 ) reserves is driving research since decades.
The enormous requirements on the reactor materials however are making the realisa-
2 Withits short half-life time it is the bottleneck, however fusion reactors try to breed T out of Li,
which is abundant.
2.7 Nuclear Energy 27
tion extremely difficult. The focus of current research are mainly hot plasma fusion
reactors with a toroidal shape. In these, the million degrees hot plasma is kept away
from the walls by a strong magnetic field, which can be only achieved efficiently by
superconducting magnets. The resulting vast temperature gradient from near 0 K to
>10 MK shows the difficulty in finding proper materials and to advance the tech-
nology. The next experimental large scale nuclear fusion reactor ITER is currently
being built in Cadarache (France) and is expected to make its first D-T fusion run
in 2035, after almost 30 years of planning and construction [5]. The probably most
likely reaction for fusion reactors will be the D-T fusion:
2
H +3 H −→4 He + n + 17.6 MeV
If we use the fusion of deuterium (2.01 u3 ) and tritium (3.02 u) as an example,

we find for the energy density for nuclear fusion:
17.6 MeV=2.82
ˆ × 10−18 MJ
M(D + T ) = 5.03 u=8.34
ˆ × 10−27 kg
E
= 3.4 × 109 MJkg−1
M
Fortunately, there is a very reliable and powerful fusion reactor available for us:
our Sun. This will be discussed in the next chapter.
Further Reading
• A comprehensive list of gravimetric and volumetric energy densities of fuels

and energy storage devices can be found on: en.wikipedia.org/energy_density.
Accessed from 26 July 2022.
• Whitman, A.M.: Work and Heat. In: Thermodynamics: Basic Principles and Engi-
neering Applications. Mechanical Engineering Series. Springer, Cham (2020).
ISBN: 978-3-030-25220-5
• For a recent example of a CAES system see: caes.ppnl.gov. Accessed from 26
July 2022.
• An overview of various fly wheel applications with further references can be
found on: en.wikipedia.org/wiki/Flywheel_energy_storage. Accessed from 26
July 2022.
• Statistics on global energy production: statista.com. Accessed from 26 July 2022.
3 Atomic mass unit: 1 u = 1.66 × 10−27 kg.

References
1. Hamann, C.H., Hamnett, A., Vielstich, W.: Electrochemistry, 2nd edn. Wiley-VCH, Weinheim
(2005)978-3-527-31069-2
2. Bard, A.J., Faulkner, L.R., White, H.S.: Electrochemical Methods: Fundamentals and Applica-
tions Wiley (2022)
3. Passerini, S., et al. (Eds.): Batteries: Present and Future Energy Storage Challenges. Wiley (2020)
4. Mench, M.M.: Fuel Cell Engines. Wiley (2008)
5. Holtkamp, Norbert: An overview of the ITER project. Fusion Eng. Des. 82, 427–434 (2007)
The Sun–Earth System
3
Abstract
The Sun is presented as the dominant energy provider for our Earth. The standard
Sun model as well as energy production in the star are explained—everything in a
view to the energy flux towards Earth. The second subchapter is about our planet,
its structure, the energy balance on Earth and other energy and climate relevant
aspects. The chapter concludes with a probable energy scenario until 2050.
3.1 The Sun
3.1.1 General Properties
Most of the energy available on Earth originates from Sun1 as the central star of
our solar system. Details of the Sun-Earth constellation are shown in Fig. 3.1. These
also determine the amount of solar energy reaching the Earth. The Sun is by far the
heaviest body in the system with a mass of M = 2 × 1030 kg and also the biggest
with a radius of R = 7 × 108 m (cf. Jupiter as the largest planet: M = 1.9 × 1027
kg at a radius of R = 7 × 107 m). However it consists mainly of light atoms: 75%
H, 23% He and only 2% heavier elements such as C, N and O. These are produced in
the core by fusion of H and He. The average density is about ρ̄ = 1.4 g/cm3 , which
is only a quarter of the average density of the Earth. This low value gives already
reason to assume an inhomogeneous structure of the Sun, which is explained further
below.
The luminosity L of the Sun is the total power radiated by the Sun into the full
solid angle = 4π :
L = 3.82 × 1026 W (3.1.1)
1 Astronomical

symbol of the Sun: .

30 3 The Sun-Earth System
Fig. 3.1 Geometry of the Sun-Earth system. The average Sun-Earth distance defines the astro-
nomical unit (AU). The half opening angle α of the Sun as seen from the Earth is important for
concentrating solar power, cf. Sect. 6.2
L is determined by measuring the solar constant (SC), which is the power imping-
ing on a unit area with normal incidence just outside the Earth’s atmosphere. Averaged
over a year this constant is:
SC = 1370 W/m2 (3.1.2)
If we take the average distance between Sun and Earth with r S E = 1 AU =

1.5 × 1011 m into account we obtain the Sun’s luminosity simply as:
L = 4πr S2 E · SC
We can further calculate the total irradiated power over one year:
E = L · 31536000 s = 1.2 × 1034 J = 3.8 × 1014 TWa =

ˆ 4 × 1026 kgCE
Using E = mc2 , this leads to a yearly mass loss of the Sun by 1.3 × 1017 kg/a (i.e.
a relative loss of 10−13 per year). By analyzing the spectrum of the light emitted
from the Sun quantitatively, we find that the Sun radiates almost like a black body.
Therefore, we can calculate the surface temperature of the Sun using the Stefan-
Boltzmann law:
Pem = e Aσ T 4 (3.1.3)
Pem stands for the total radiated power, e for the emissivity of a body (0 ≤ e ≤ 1,
with e = 1 for an ideal black body), A is the emitting area (i.e. the Sun’s surface), T
the absolute temperature of the body and σ is the Stefan-Boltzmann constant.
σ = 5.67 × 10−8 W/m2 K4
Insertion into (3.1.3) yields:
Pem = L = 4π R
2
σ Te4f f assuming e = 1
3.1 The Sun 31
Te f f = 5780 K (3.1.4)
We call this an effective surface temperature, because the real temperature varies
strongly from the core to the outside of the Sun.
3.1.2 Details of Proton Fusion
The generation of energy in the Sun via fusion of protons follows the overall reaction:
4 p + → 4 H e + 2e+ + 2νe + 26.73 MeV (3.1.5)
The number of electron neutrinos νe arriving on Earth is smaller than what is

expected from (3.1.5) and from the power arriving on Earth. The reason for this
effect are neutrino oscillations: electron, muon and tau neutrinos periodically
transform during their lifetime into each other (i.e. change their flavor). This
quantum effect indicates that neutrinos have mass and describe hereby a theory
beyond the Standard Model [1,2].
This overall reaction can be achieved via different reaction pathways. The most
direct one starts with the fusion of two protons to form a deuteron (np + ):
p+ + p+ −→ d + e+ + νe + 0.42 MeV (3.1.6)
Due to the high temperatures of several million K necessary to achieve fusion,

all atoms in the Sun are completely ionized, so that hydrogen exists in the form of a
plasma formed by free protons and free electrons. The resulting positron is needed
to conserve charge, and the electron neutrino ensures conservation of the lepton
number. After its emission, the positron annihilates immediately with a free electron
via the emission of two γ -photons.
In the next step, a deuteron fuses with a third proton to form 3 He:
d + p+ −→ 3 He + γ + 5.49 MeV (3.1.7)
After the formation of 3 He, there are three possible reactions to produce 4 He,
starting at different temperatures:
1. Dominating at T = 10 − 14 × 106 K:
3
He + 3 He −→ 4 He + 2 p+ + 12.86 MeV
2. Dominating at T = 14 − 23 × 106 K
3
He + p+ −→ 4 He + e+ + νe + 18.77 MeV
3. Dominating at T > 23 × 106 K (with 4 He as a catalyst for higher production of

4 He):
3 He + 4 He −→ 7 Be + γ + 1.59 MeV
7 Be + e− −→ 7 Li + νe
7 Li + p+ −→ 24 He + 17, 35 MeV
and:
7 Be + p+ −→ 8 B + γ + 0.14 MeV
8 B −→ 8 Be∗ + e+ + ν
e
8 Be∗ −→ 24 He
3.1.2.1 CNO-Cycle (Bethe-Weizsäcker Cycle)

In stars with more than 1.3–1.5 M (core temperatures Tcor e ≥ 20 MK) the CNO-
or Bethe-Weizsäcker-cycle dominates. This fusion chain is catalysed by the heavier
nuclei C, N, and O and also produces 4 He. In our Sun this reaction contributes about
1.6% to the overall solar energy production. In the reaction chain intermediate,
unstable nuclei are involved, i.e. 13 N and 15 O, which decay by positron emission:
• + p+ −→ 13 N + γ
12 C
−→ 13 C + e+ + νe (7 min)
13 N
• 13 C + p+ −→ 14 N + γ
• 14 N + p+ −→ 15 O + γ
15 O −→ 15 N + e+ + ν (82 sec)
e
• 15 N + p+ → 12 C + 4 He
With the overall reaction being again:
4 p+ −→ 4 He + 2 e+ + 2νe (3.1.8)
Our Sun is currently only producing elements up to carbon (C, Z = 8). Elements
heavier than C can be created in later stages by dying stars depending on their
initial mass. Stars can create elements up to iron (Fe, Z = 56) by fusion. Beyond
that, the energy balance becomes negative (cf. Sect. 2.7). The production of
heavier elements requires far higher energies which are only found in novae,
supernovae and similar events.
3.1 The Sun 33
Fig. 3.2 Equilibrium of

pressure and gravity in a star.
M(r ) is the total mass in a
sphere with radius r
3.1.3 Shell Model
A simple model of the Sun can be derived by considering the equilibrium of forces
and energy production in a stationary star2 and assuming an ideal gas for the particles
in the star. This model is called “shell model”, since the following considerations
lead to a shell structure. This model considers the interrelations between the density
ρ, the pressure p, the absolute temperature T and the energy flux as a function of
the distance r from the star center. It is based on the following three relations:
1. The Ideal Gas Equation:
k B T (r )
p(r ) = ρ(r ) (3.1.9)
m̄
Here, m̄ is the effective mass per particle (1/2(m p + m e ) ≈ 1/2m p ).
2. Equilibrium of Pressure and Gravitation:

A mass element dm in Fig. 3.2 is given by
dm = ρ dV = ρ dA dr
Each of these mass elements is subjected to the gravitational force, as well as the
pressure force caused by the core’s fusion reaction:
M(r ) M(r )
dFg = G dm = G 2 ρ(r ) d A dr (3.1.10)
r2 r
d p(r )
dF p = [ p(r + dr ) − p(r )] dA = − dr dA (3.1.11)
dr
2 Of course this model cannot explain the time evolution of a star from formation to death. The
interested reader is referred to the corresponding literature.
Here, G is the gravitational constant and M(r ) is the total mass inside r . For a
stationary star, these two forces (3.1.10 and 3.1.11) must be equal and opposite:
dp M(r )
dFg = dF p ⇒ = −G 2 ρ(r ) (3.1.12)
dr r
3. Energy Production:
The differential energy eem produced in a shell dr with radius r is:
eem (r ) = ε dM(r ) = ε4πr 2 ρ(r ) dr (3.1.13)
ε is the energy produced per unit mass. The released energy can be described as
an energy flux (r ) through the surface of a sphere with radius r :
d = (r + dr ) − (r ) = ε dM(r )
This can be rewritten with (3.1.13) as:
d
= 4πr 2 ερ(r ) (3.1.14)
dr
These coupled equations (3.1.9, 3.1.12 and 3.1.14) can be solved with the known
boundary conditions at the surface of the Sun:
M(R ) = M = 2 × 1030 kg (3.1.15)

T (R ) = Te f f = 5780 K (3.1.16)
(R ) = L = 3.82 × 10 W 26
(3.1.17)
p(R ) =0 (3.1.18)
Fig. 3.3 Shell model of the Sun. See text for more details
3.1 The Sun 35
Fig. 3.4 Convection cells in the photosphere
This leads to the following shell model of the sun (Figs. 3.3 and 3.4):
• 90% of the energy is produced in the core with radius r ≤ 0.23R . The core
temperature is 15 MK, the core density ρcor e = 100 g/cm3 ≈ 70ρaverage .
• The energy generated in the core is transported by γ -radiation up to 0.7 R . In
the outer shell energy is transported by convection cells.
• The visible radiation emitted by the Sun originates from the so-called photosphere,
a thin surface layer of about 200 km, where the temperature drops from T ≈
8000 K to T ≈ 4500 K. The photosphere also has a granular character due to
convection cells.
• Beyond the photosphere exists the chromosphere with a thickness of about 10
000 km. It consists of hot particles able to leave the photosphere according to
the barometric height formula. The emitted radiation consists of specific atomic
spectra (H, He, Mg, Fe).
• Further outside is the corona, consisting of very hot particles (T ≈1 MK), extend-
ing about 1 × 106 km. The most energetic particles can leave the gravitational
field of the Sun as the solar wind. This gives rise to an additional mass loss of
≈10−13 M per year
• In addition the Sun exhibits dynamical events of statistical nature:
Flares: short eruptions (minutes to hours)
Protuberances: fast particles remaining up to weeks in the external magnetic field

of the Sun
Sunspots: local disturbances in the magnetic field. The magnetic field lines reach
through the photosphere and hinder thereby heat transport in these regions. Thus
they are colder and appear darker (Fig. 3.5). Sunspots indicate increased solar
activity with a more disordered magnetic field pattern.3 Due to different rotation
speeds of the different shells, the number of sunspots varies with a period of
around 11 years (9–15 years, cf. Fig. 3.6).
3 First observations date back to the Antique. Among others, Galileo kept such a sunspot record for
a couple of years. Since the 19th century different records were collected and thus a long data base
was created.
Fig. 3.5 Typical structure of sunspots. Left: Umbra with surrounding penumbra. Right: Sunspots
appear mostly in pairs with common magnetic field lines and in combination with flares
Fig. 3.6 Sunspot cycles of the past 400 years. Blue dots indicate the monthly average of sunspots,
the orange line a moving average across 10 data points. Source WDC-SILSO, Royal Observatory
of Belgium, Brussels [3]
3.2 The Earth
3.2.1 General Properties
With a radius of only R♁ = 6.4 × 106 m and a mass of M♁ = 5.974 × 1024 kg, the
Earth4 is the third planet by distance from the Sun. Using a simple geometrical
ansatz, we can approximate the total incoming power from the Sun:
P♁ = (1 − R)π R 2 SC (3.2.1)
♁
π R 2 is the surface area projection towards the Sun. The additional term (1-R) with
♁
the reflectivity R takes into account that not all incoming energy is absorbed. The
Earth’s surface reflects a certain percentage, varying significantly across our planet.
Typical values of R are listed in Table 3.1.5
4 Astronomical symbol of the Earth: ♁.

5 The measure for the reflection is called reflectivity R. Together with transmittance T and absorption
A balance must hold: 1 = T + A + R.
3.2 The Earth 37
Table 3.1 Typical reflectivities of different surfaces on Earth

Surface Reflectivity
Clouds 0.2–0.7
Water (at the equator, normal incidence) 0.05
Water (northern hemisphere) 0.25
Snow 0.3–0.7
Green vegetation 0.1–0.2
Deserts 0.3
Overall, this leads to a global average reflectivity of R ≈30%. Together with the
SC we then obtain for the total incoming power6 :
P♁ ≈ 1.2 × 1017 W = 1.2 × 105 TW
From this, we can calculate the Earth’s theoretical average surface temperature
TE , if we set up the energy balance between incoming solar radiation and emitted
thermal radiation of the Earth. In a first approximation we assume that the Earth is
a black body as well with e = 1:
P♁ = (1 − R)π · R 2 · SC = 4π · R 2 · σ · TE4
♁ ♁

(1 − R)SC 1/4
⇒ TE = ≈ 254 K
4σ
Thus, the average equilibrium surface temperature of the Earth under these assump-
tions would be only TE ≈ −19 ◦ C. However, the observed average temperature today
is T ♁ = 288 K ≈ 15 ◦ C. This important difference is caused by the greenhouse
effect: without the additional reflectivity of our atmosphere also for the outgoing
radiation from the Earth’s surface, temperatures would have been most likely too
cold for life based on liquid water.
The wavelength λmax of the maximum of the emitted black body radiation can be
calculated with the Wien-law based on the experimentally observed spectra of the
incoming and emitted radiation (cf. Fig. 3.7):
hc 2898
λmax = = µm K (3.2.2)
5kB T T
⇒ λmax, = 490 nm(green-blue)

⇒ λmax,♁ = 12 µm(infra-red)
6 Inaddition to the black body radiation of the Sun, the Earth receives a weak spectrum of γ -rays
from the corona (λ ≈ 0.1nm).
Fig. 3.7 Left: Quantitative comparison of the emitted radiation of the Sun arriving on Earth before
(yellow) and after passing Earth’s atmosphere (red) [4]. Right: Earth’s infrared emission spectrum
with characteristic absorption peaks of mainly H2 O, O3 , and CO2 . Note the orders of magnitude
difference between the left and right y-axis. Source ASTM G-173 AM1.5 Reference spectra, Earth
thermal infrared emission spectrum as captured over the Sahara desert by the Nimbus 4 satellite [5]
To understand and model the very important modulation of the radiation emitted
by the Sun and the Earth when passing through our atmosphere, several relevant
physical processes have to be taken into account:
• The infrared spectrum emitted by the Earth is modulated via absorption and
reflection by strong vibrational absorption bands mainly of H2 O and CO2 in the
wavelength range between λ = 5 µm to 30 µm. The current average concentra-
tions of H2 O and CO2 are 1600 ppm and 400 ppm, respectively. This provides
the physical basis for the notorious greenhouse effect, which however is further
complicated by the existence of (partly fragile) feedback loops. Thus, the increase
in CO2 due to burning of fossil fuels leads to in increase in the surface temperature
of the Earth, resulting in an additional increase of water in the atmosphere due to
increased evaporation. The increased water concentration will further strengthen
the greenhouse effect (positive feedback), but also leads to the formation of more
clouds, increasing the reflectivity for the incoming solar radiation (negative feed-
back).
• The solar radiation arriving on the Earth surface is modulated by electronic absorp-
tion bands especially of O3 (ozone), O2 , H2 O, and CO2 . Ozone is mainly respon-
sible for absorption in the ultraviolet region of the solar spectrum (explaining the
danger related to the ozone hole), atmospheric oxygen mainly absorbs in the vis-
ible region, and water as well as carbon dioxide absorb solar radiation in the red
and near-infrared region. The integral absorption of Sun light via the atmosphere
is described by the quantity air mass (AM), which takes into account the different
effective length that sunlight travels through the atmosphere, depending on the
geographical location:
– AM = 1/ sin γ , where γ is the zenith angle
– AM 1: Sun is in zenith (at the equator)
– AM 1.5: ≈ Munich (with a maximum intensity of 1 kW/m2 = 100 mW/cm2 )
3.2 The Earth 39
Fig. 3.8 Schematic view of the elliptical orbit of the Earth around the Sun
• The geothermal energy flux through the surface of the Earth (cf. Chap. 6) is less
than 0.01% of the impinging solar energy flux and can be neglected in these
considerations.
Since the Earth orbits around the Sun on an elliptical path, the distance between
Sun and Earth exhibits a periodic modulation throughout the year. This causes also
varying values of the solar constant and, thereby, the incoming flux of energy:
• SC = 1415 W/m2 in January (Perihel, Earth-Sun distance 147 Mkm)

• SC = 1325 W/m2 in July (Aphel, Earth-Sun distance 152 Mkm)
In addition, the rotational axis of the Earth is inclined by 23.5◦ with respect to
the orbit plane of the Earth (ecliptic).7 This gives rise to the different seasons in the
northern and southern hemispheres, leading to locally even more pronounced yearly
variations of the incoming solar radiation (Fig. 3.8).
In addition to the dominant annual variation of solar irradiation due to the elliptical
Earth orbit and the rotational axis inclination, the solar luminosity L varies as well
with time. On the one hand there exist statistical variations of the luminosity by about
0.5% (causing a corresponding variation of the surface temperature of the Earth by
about T♁ ≈ 0.4 ◦ C). On the other hand, there are periodical variations by a similar
amount correlated with the number of sunspots. As discussed above, the number of
sunspots varies between 0 and about 300 spots with a period of about 11 years. There
also are indications of an additional super-period of about 100 years (see Fig. 3.6).
Since many sunspots correspond to an increased activity of the Sun, this leads to
7 The currently accepted reason for this inclination of the rotational axis of the Earth with respect
to its orbit plane around the Sun is the collision of the Earth with a Mars-like cosmic object, which
caused the shift of the axis and the simultaneous formation of our Moon from the ejected debris.
additional long-term variations of the solar luminosity and the surface temperature
of the Earth.
A still debated correlation is that between the so-called Maunder minimum, an

abnormal reduction of observed sun spots during the years 1645 and 1715, and
the “little ice age” in Europe and Northern America in about the same time
range, which lead to a severe famine catastrophe in Europe. Alternative or
additional explanations may come from enhanced volcanic activity prior and
during this time.
Figure 3.9 shows a schematic summary of the main global energy fluxes caused by
solar irradiation on the Earth. The main fluxes are linked to optical processes (absorp-
tion, reflection and thermal radiation) and thermo-mechanical processes (convection,
evaporation). Compared to those, geothermal, gravitational or biomass-related fluxes
are almost negligible. Yet, the latter have a very large influence on our daily life as
well.
Fig. 3.9 Schematic view of the energy fluxes on Earth

3.3 A Possible Energy Scenario Until 2050 41
3.3 A Possible Energy Scenario Until 2050
In the following we give a personal view of the requirements, boundary conditions

and possible solutions to provide our planet with sustainable energy by the year 2050,
based on the present situation in 2020. As a word of caution: this or similar long-term
scenarios always should be viewed with critical eyes, because political, scientific,
technological, ecological and societal aspects are likely to change significantly over
such long time spans.
We assume a continuous increase of the world population up to 10 billion people
by 2050. Moreover, we estimate an average consumption of 2 kW/human being,
including 1 kW for food production. Note that this means that most developed coun-
tries will have to save 60–90% of their current energy consumption, which already
would constitute a tremendous task. With this we require a continuous necessary
power “to run the planet” of 20 TW.
At the moment, we still have sufficient fossil fuel reserves to maintain such an
energy production for more than thousand years. Current estimates predict sufficient
oil supplies for about 50–150 years, gas for 200–600 years and coal for more than
2000 years. However, if we want to keep global warming due to the resulting emission
of CO2 below an average temperature rise of 2 ◦ C, we have to stop the use of fossil
fuels as soon as possible, reaching zero emission globally by 2050.
Possible solutions for this energy problem are very limited:
• Increased use of nuclear power (fission only, since fusion will very likely remain
on an experimental scale until 2050).
• Active removal of CO2 from the atmosphere by carbon capture and storage (CCS)
and/or artificial photosynthesis.
• Global geo-engineering to counteract the greenhouse effect by climate cooling.
• Ramping up the use of renewable energy to the required level of 20 TW by 2050.
So let us have a critical look at these four possibilities one by one.

(1) Nuclear Power
To provide the required 20 TW, about 20 000 new nuclear power plants with a typical
power of 1 GW would have to be built. This means that until 2050 every day at least
one new plant would have to go online. In reality, today less than 100 new nuclear
reactors are planned or under construction, with usually more than ten years required
for completion. Moreover, the known uranium reserves for several thousand nuclear
plants would only last for a few decades, so that the risky breeder technology would
have to be used worldwide on a large scale.
(2) Carbon Capture and Storage and Artificial Photosynthesis
The global annual CO2 emission by mankind has increased from about 6 Gt per
year (Gt/a) in 1950 to about 40 Gt/a today. Of these, about 10% alone are produced
by the metabolism of 8 billion humans and their domestic animals and, thus, are
almost unavoidable. In comparison, the net uptake of atmospheric CO2 by the oceans
only amounts to about 2 Gt/a and already has led to a noticeable and dangerous
acidification of the sea water. Storing CO2 in solid form in deep cold water at high
pressures is physically possible, but has never been tested on a large scale for a
sufficiently long time. The same is the case for CO2 storage underground on land
and usually meets great public concern even for small CCS pilot plants. Anyhow,
given the large amount of CO2 that would need to be stored annually (about ten times
the worldwide concrete and cement production of 5 Gt/a) and the related energy
consumption, avoiding CO2 emission at all would be the much better alternative.
In this respect, also artificial photosynthesis, which aims at removing CO2 from
the atmosphere and transform it with the help of sunlight into fuels like methane
or methanol is a very interesting approach (see Chaps. 7 and 8). However, even
after decades of intensive research efforts, efficient large scale and long-term stable
systems made from sustainable materials will not be available in the foreseeable
future.
(3) Climate Cooling by Geo-Engineering
Geo-engineering aims at the global manipulation of the climate by large scale removal
of CO2 from the atmosphere as discussed above, or at a global increase of the Earth’s
reflectivity for a reduction of incoming sunlight. Methods for the latter approach cur-
rently discussed are mirrors placed in an Earth orbit, distribution of small reflecting
particles like sulfur dioxide in the upper atmosphere, or increasing the reflectivity
of the Earth surface by white paint or mirrors. That such approaches indeed can
significantly lower the global temperature was last demonstrated by the eruption of
the Pinatubo volcano in 1991, which caused a temperature decrease of about 0.5◦ C.
However, it is still not possible to predict the local consequences and risks of geo-
engineering sufficiently well, and many scientists fear possible conflicts between
different countries or even the use of geo-engineering as a weapon. 8
(4) Ramping up the Use of Renewable Energy Sources
Probably the most plannable, controllable, sustainable and, therefore, most likely
solution will be the replacement of fossil fuel usage by CO2 -neutral renewable energy
sources. Before we discuss this in detail in the following chapters, we summarize
here briefly our view of the potential, limitations and pending problems of renewable
energy sources.
Energy from Rivers, Waves and Tides
Running water in streams and rivers is estimated to represent a total power of 5–
6 TW. Of these, however, only around 2 TW can be harvested in an ecologically
and economically viable manner. Especially in developed countries, most of these
viable resources are already being used. Random small scale waves in the oceans and
periodic large scale tidal waves are estimated to be able to contribute an additional
power of 2 TW.
8 As a side note, allowing CO

2 levels to further increase also has another consequence: beyond levels
of 1000 ppm the intellectual performance and capabilities of humans are noticeably decreased.
3.3 A Possible Energy Scenario Until 2050 43
Wind Energy
The global winds on-shore and off-shore are estimated to be able to realistically
contribute a total power of 4 TW to 5 TW. To fully harvest this source, about one
million wind engines with a power capacity of 5 MW each would be necessary. This
translates into 100 of such wind engines which would need to be constructed every
day until 2050, for a yearly increase of wind power by 180 GW/a. For comparison,
the currently available wind power is increasing by about 60 GW/a, so “just” a factor
of three off.
Photosynthesis and Biomass
Although the conversion of solar energy via photosynthesis by plants, algae and
bacteria in the past and at present forms the basis of our food and energy supply,
the low average efficiency of natural photosynthesis of only 0.3% limits its potential
upscaling. For the production of 20 TW, about 30% of the available continental area
would be required. In addition, a sufficient supply of water would be required in
these areas. Conservative estimates of the potential contribution of biomass to the
pool of renewable energy limit it to 5 TW.
Direct Use of Sunlight by Photovoltaics, Artificial Photosynthesis and Solar Ther-
mal Power
According to Chap. 2, the constant power provided by solar radiation on Earth is
120000 TW, with a density of about 1 kW m−2 . So, by transforming solar irradiation
into electricity with a 20% efficient solar cell, the entire power presently consumed
by the USA of about 3 TW could be produced by covering 2% of the US area with
such solar cells. This corresponds roughly to the area covered by streets, roads and
highways. To reach a power capacity of 20 TW by 2050, about one million solar
modules with a power output of 1 kW would have to be installed worldwide every
day, an annual increase by 365 GW/a. This does not seem impossible given the
current global photovoltaic annual growth rate of 60 GW/a.
To summarize, a sustainable energy supply of mankind without further detrimental
damages to our ecosphere is possible, even on the time scale of the three decades
remaining until 2050. But it will require serious and concerted efforts in (i) saving
energy especially in developed countries, (ii) replacing conventional energy sources
by an intelligent mix of renewable energy worldwide, and (iii) developing distributed
energy storage capacities on the TW scale to buffer the daily and seasonal fluctuations
of the solar energy available at a given place and time.
Further Reading
• Guenther, D.B. et al.: Standard solar model. Astr. Phys. J. Part 1, 387, 372–393
(1992).
• Currently, the NASA’s Parker Solar Probe investigates the surface of the Sun and
will be the closest satellite to the Sun ever launched. More information and images
can be found on the webpage. www.nasa.gov/content/goddard/parker-solar-probe.
Accessed from July 2022.
References
1. Eidelman, S. et al.: Review of particle physics. Phys. Lett. B592(1–4) (2004)
2. Ahn, M.H., et al.: Measurement of neutrino oscillation by the K2K experiment. Phys. Rev. D
74, 072003 (2006)
3. Sunspot data from SILSO Royal Observatory of Belgium, Brussels. www.sidc.be/silso. Accessed
from 26 July 2022
4. ASTM G-173 AM1.5 Reference spectre, available at NREL’s website: https://nrel.gov/grid/
solar-resource/spectra-am1.5.html. Accessed from 27 July 2022
5. Hanel, R.A., Conrath, B.J.: Thermal emission spectra of the Earth and atmosphere from the
nimbus 4 Michelson interferometer experiment. Nature 228, 143–145 (1970)
Energy from Waves,Tides and Osmosis
4
Abstract
This chapter deals especially with wave power and osmosis power plants. The
physics of wave generation and propagation as well as the wave’s energy content
are explained. As a special form of waves, tidal waves are discussed in more detail.
Principles of osmosis and how it can be used for power generation are also briefly
explained.
4.1 Wave Energy
Water waves are created by friction and lift forces of wind streaming over a water
surface. They are an example of the general Kelvin-Helmholtz instability at the
boundary of two media moving with different velocities. By this instability, random
fluctuations at the interface are amplified. An initial wave crest is further lifted up
according to the Bernoulli equation, because the wind velocity vwind increases at the
wave crest and, as a result, the air pressure pair is reduced there:
1
pair + ρair vwind
2
= const.
2
In addition, the wind can push a wave forward in the wind direction because of
friction and impact pressure, as sketched in Fig. 4.1.
For a non-progressing wave in deep water (meaning that the total water height h
is at least larger than a quarter of the wave length λw ), the water molecules move
on closed circles, with radii R decreasing exponentially with depth d (deep water
waves). If the wave is also progressing with mass transport along the water surface,
the speed of progression adds to the circular velocity and gives a forward spiraling
orbital motion of the molecules. The circular motion in non-progressing deep water
waves is also perturbed when the total water height h becomes less than λw /4, as
shown in Fig. 4.2:

46 4 Energy from Waves, Tides and Osmosis
Fig. 4.1 Illustration of the interaction of wind with the water surface, leading to the creation of
waves (and turbulences)
Fig. 4.2 Local orbits of molecules in deep and shallow water waves (λw is the wavelength, h the
water height)
4.1.1 Deep Water Waves
The propagation of waves in deep water (water depth h > λ/4) follows a simple
consideration: the wave does not “feel” the ground, i.e. the interaction of the water
motion with the resting bottom of the sea can be neglected. Then, the local circular
trajectories at the surface occur with a velocity vcir cle = 2π R · f w , where f w is the
frequency of the wave. Although the wave movement has a finite phase velocity vw ,
the single molecules do not propagate averaged over time (cf. Fig. 4.3).
This local circular movement causes a periodic transformation between potential
and kinetic energy of the molecules. At the surface, the molecules are falling by a
height of h = 2R, with respect to the equilibrium surface. The potential energy has
its maximum at the wave crest, whereas the velocity vtop = vw – vcir cle reaches a
minimum. At the wave trough, the potential energy is minimal, whereas the velocity
vbottom = vw + vcir cle has its maximum. The resulting changes in kinetic and poten-
tial energy are equal, from which the dispersion relation vw (λw ) can be derived:
4.1 Wave Energy 47
Fig. 4.3 Motion of water particles during a wave transition
1 2 1
E kin = m vbottom − vtop
2
= m (4vw vcir cle ) = 2m vw 2π R f w
2 2
E pot = mg · 2R
E kin = E pot ; ⇒ 2π vw f w = g and:
g
vw = (4.1.1)
2π f w
With f w = vw /λw this gives:

gλw
vw = (4.1.2)
2π
Thus, deep water waves have a pronounced dispersion: long waves move faster than
short waves. Since the dispersion relation is only determined by the gravitational
acceleration of the Earth, such waves are also called gravitational waves.
The dispersion of deep water waves can be seen by the following phase veloc-
ities for different wavelengths:
λw = 10m ⇒ vw = 4 m/s
λw = 100m ⇒ vw = 12.5 m/s
λw = 1000m ⇒ vw = 40 m/s
Fig. 4.4 Additional restoring force caused by the surface tension σ , dominating the wave velocity
for small wavelengths
In addition to gravity the velocity of water waves is also influenced by surface

tension and by a finite depth of the water [1]. The surface tension σ 1 adds a restoring
force to the wave as seen in Fig. 4.4. Every deviation from a flat surface increases the
surface area of a uniform flat water surface at rest. This increases the surface energy,
which then tries to reach a minimal value again. The influence of surface tension
modifies the dispersion relation mainly at very small wavelengths:

gλw 2π σ
vw = + (4.1.3)
2π ρλw
This gives rise to two regimes of increasing phase velocity with either increasing
or decreasing wavelength:
Capillary waves (λw < 1 cm)

vw ∝ 1/λw
Gravitational waves (λw > 10 cm)

vw ∝ λw
4.1.2 Shallow Water Waves
A finite water depth d influences the wave propagation noticeably for d ≤ λw /4.
Then the wave interacts with the ground, which leads to a more elliptical trajectory
of the molecules as shown in Fig. 4.2. The velocity of a shallow water wave is given
by:

gλw 2π d
vw = tanh (4.1.4)
2π λw
1 (with σ = 7 × 10−2 N/m, ρ = 1 × 103 kg/m3 for water).

4.1 Wave Energy 49
Fig. 4.5 Illustration of wave

parameters involved in the
energy calculation. h cg
indicates the height of the
center of gravity of the wave
√
For d << λw (tanh(x) → x), this can be approximated by vw = g · d, so that
shallow water waves have no dispersion (no dependence of vw on λw ).
This is the reason for surf or tsunami waves: long deep water waves with a high
wave velocity approach the shore, where their behaviour is influenced more and
more by the decreasing depths d < λw /4 and they “become” shallow water
waves with decreasing wave velocity. Energy conservation leads then to an
increase of the wave amplitude (i.e. potential energy) when reaching the coast.
Finally we will estimate the energy content of a wave. Let us assume a wave front
with a lateral extension of b, an amplitude A, and wavelength λw (Fig. 4.5). For a
sinusoidal wave the upper half of the wave contains a water mass M of:
λw /2
2π x λw
M = ρV = ρb A sin dx = ρ · b · A ·
0 λw π
For b = 1 km, A = 2.5 m, and a wavelength λw = 50 m, this results in a mass of

about M ≈ 40 000 t. The position of the centre of gravity h cg can be calculated as
h cg = (π/32)A
During one wave period the center of gravity drops by 2h cg and energy is trans-
formed from potential energy to kinetic energy:
π
E pot = Mg2h cg = Mg A
16
In our example this results in an energy E pot = 2 × 108 J or 78kW h or 10 kgCE

of this wave front.
The power density of a wave per unit length of the wave front is then:
d P̄ E pot Mg Aπ f w ρ A2 gλw f w
= · fw = =
dx b 16b 16
g
with λw = ⇒
2π f w2
d P̄ 1 A2
= ρg 2 (4.1.5)
dx 32π fw
For an amplitude of A = 1 m and a wave frequency of f w = 0.5 Hz the linear

power density is about
⇒ d P̄/dx ≈ 2000 W m−1
For a 1 GW wave power station this would require a coastline length of b =

500 km.
Figure 4.6 shows a map of the mean wave power per meter for the global oceans.
As one can see, the energy density per unit length is especially low in coastal regions,
where harvesting this energy would be more easy.
The first realized large-scale wave energy converter was Pelamis, a long, snake-
like semi-submerged structure with a rated power of 750 kW. Its jointed, cylindrical
sections move with the incoming waves and transfer this motion to hydraulic rams,
charging hydraulic accumulators. Electricity is then generated by induction motors
housed inside the structure [3]. A first power plant was installed off the coast of
Portugal in 2008, but in the same year taken offline again. Most systems in use today
are based on point absorber buoys. These buoys pick up the wave motion and transmit
it to a shaft connected to a station in the sea bed, where its motion is converted to
electricity. Small units are installed e.g. off the coast of Australia (CETO, rated power
240 kW per buoy) or of the United States (Azura Wave, rated power 20 kW per buoy).
Another approach to convert wave energy to electricity is based on the oscillating
water column. Here, the changing water level of incoming waves and tides pushes
air through a wind turbine and, thus, produces electricity. However, so far no system
has made it beyond a prototype phase.
4.2 Tidal Energy
Tides are created by asymmetries between gravitational and centrifugal acceleration

of water on different sides of the Earth’s surface relative to the Sun and the Moon.
4.2 Tidal Energy
Fig. 4.6 Schematic global map of the average linear power density of ocean waves. More accurate maps can be found e.g. at Cornett [2]. The color scales from
>10 kW m−1 (purple) to >120 kW m−1 (red)
51
Hence we can differentiate between solar and lunar tides. The complex overall tidal
forces are a superposition of these two contributions depending on the relative con-
stellation of Earth, Sun and Moon (e.g. spring tide).
4.2.1 Solar Tides
Solar tides result from the differences between the gravitational pull of the Sun
and the centrifugal forces due to the orbit of the Earth around the Sun on the two
sides of the Earth’s surface closest to and furthest away from the Sun as shown in
Fig. 4.7. On the side facing towards the Sun, the gravitational pull is larger than the
centrifugal force, whereas on the opposite side the centrifugal force is larger than
the gravitational pull. Only in the center of the earth both forces cancel each other
exactly. There, the average centrifugal acceleration of the Earth on its orbit around
the Sun is determined by the Sun-Earth distance (r S E = 1AU = 1.5 × 1015 m) and
the angular velocity (ω = (2π )/yr ≈ 2 × 10−7 s−1 ) of the Earth’s yearly orbit:
āc f = r S E ω2 = 6 × 10−3 m s−2
This average value varies throughout the Earth by
ac f = 2R♁ ω2 = 2.5 × 10−7 m s−2
The average gravitational acceleration caused by the mass of the Sun on the other
hand is
G M
ḡ
= (= āc f only in the center of the Earth)
r S2 E
with a variation on both sides of the Earth of g

:

dg G M
|g | =

· 2R = 2
♁ · 2R♁ = 1 × 10−6 m s−2
dr S E r S3 E
Fig. 4.7 Solar tides under the assumption of a global ocean expanding around the Earth
4.2 Tidal Energy 53
Since ac f ≈ (1/4)g

, the gradient of the gravitational pull dominates the
solar tides. Assuming a global ocean as the simplest approximation, this would give
rise to two tidal waves on the opposite sides of the Earth with a height difference
between low and high tide of only up to 20 cm. Due to the Earth’s daily rotation we
obtain a high tide period of 12 h at a given point on the Earth’s surface.
4.2.2 Lunar Tides
The calculation of the lunar tides’ height follows the same approach, with one impor-
tant difference: Due to the large mass of the Sun and the large distance between Sun
and Earth we could assume that the position of the center of mass of the Sun-Earth
system coincides with the center of the Sun. For the Earth-Moon system, however,
we have to take into account that its center of mass deviates significantly from the
center of the Earth, as illustrated in Fig. 4.8.
The gradient of gravitational acceleration caused by the Moon is:
Gm M
g M = 2 · 2R♁ = 233 × 10−6 m s−2
r E3 M
The corresponding calculation of the gradient of centrifugal force takes the posi-
tion of the center of mass and the resulting additional centrifugal acceleration on
the side facing away from the Moon (angular orbit frequency: ω = 2π/27.3 d) into
account:
ac f ≈ (R♁ + 4650 km)ω2 = 8 × 10−5 m s−2
Fig. 4.8 Illustration of the Sun-Moon system: notice the position of the center of mass about 4650
km away from the center of the Earth. Earth and Moon rotate around the axis through the center of
mass once per month
Fig. 4.9 Superposition of the solar and lunar tides for neap and spring tides
From these simple considerations we can see that the variation of forces through
the Earth and therefore, also the lunar tides for a global ocean (50 cm) are larger than
the solar tides (20 cm). Both contributions locally depend on the relative constellation
of Sun, Earth, and Moon, as shown in Fig. 4.9, giving rise to periodic phenomena
such as neap and spring tides.
The still rather small average amplitude of tidal waves for a global ocean can be
locally enhanced when a tidal wave enters into a narrow, tapered channel, and also
by local λ/4-resonances. As shown in Fig. 4.10, these occur when a tidal wave with
a wavelength λ enters a coastal bay with a length of x = λ/4 perpendicular to the
coast line. The distance λ/4 traveled by the incoming wave from the entrance to the
end of the bay corresponds to a change of the wave phase by π/2. The wave is then
reflected by the fixed end of the bay, giving rise to an additional phase change π .
Finally, the reflected wave undergoes a second phase change of π/2 until it reaches
again the entrance of the bay, where it interferes constructively with the next incoming
wave:
π π
+π + ⇒ 2π phase difference
2 2
The Bay of Fundy in Canada shows one of the highest tides in the world. The
bay has a length of L≈300 km and a depth of about d = 75 m. This corresponds
to the limit of a shallow water wave for a tidal wave with a wavelength of about
1000 km. We can use this to estimate the time delay between the wave entering
the bay and leaving it again:
4.2 Tidal Energy 55
Fig. 4.10 Schematic illustration of the λ/4 resonance for a tidal wave with amplitude A and a
period of T = 12 h
Fig. 4.11 Spectrum of ocean waves by frequency, classification and cause

vw = g · d ≈ 30 ms−1 ; λw = 4 · L ≈ 1200 km
−5
f = vw /λ ≈ 2.5 × 10 Hz ⇒ Tw = 1/ f w = 11 h 6 min ≈ 12 h
To a first approximation this matches quite well with the observed tidal period
of around 12.4 h.
Figure 4.11 provides a qualitative summary of the relative energy provided by

ocean waves of different origin as a function of the wave frequency. Tidal waves
appear as delta functions with periods of 12 and 24 h at the low frequency end of the
spectrum.
Fig. 4.12 Left: Polar water molecules decrease the energy of ions of the solute (NaCl) by forming a
hydration shell around them. Right: Working principle of an osmosis power plant using the osmotic
pressure: solvent molecules will diffuse into the second chamber with a higher solute concentration
to further dilute the solute, giving rise to an osmotic pressure difference posm
4.3 Osmosis Power
Osmosis power plants use the difference in salinity of e.g. (salty) ocean water and
(sweet) river water. Their key feature is a semi-permeable membrane, which is per-
meable for the solvent (i.e. H2 O) and in-permeable for the solute (i.e. NaCl). A
simple osmotic cell consists of two compartments separated by such a membrane
(cf. Fig. 4.12).
Across the membrane, an osmotic pressure develops, which can be estimated by
the van-’t-Hoff equation:
posm V = ν RT
Here, ν is the number of moles of solute in the volume V and R is the universal
gas constant. This formula for the osmotic pressure is completely analogous to the
ideal gas equation, where the solute ions correspond to the ideal gas particles and
the solvent corresponds to the vacuum (in which the particles move).
This osmotic pressure posm can be used e.g. to drive a water turbine for power
generation. The global power generation potential of osmotic power plants can be
estimated by assuming a salt/sweet water combination with typical salt concentration
of 3.5%. For a temperature of T = 10 ◦ C an osmotic pressure of about posm ≈ 20 bar
can be expected. Realistic power densities of semi-permeable membranes are about
3 W/m2 . Taking into account the limited sweet water resources (only 2.6–3.5% of
the Earth’s water is sweet water, with most of this in the form of snow and ice) we
can assume a realistic osmosis power potential for Germany of about 40 MW and
worldwide of about 60 GW.
The first osmosis power plant prototype worldwide was opened in 2010 close to
Oslo in Norway and was designed for an electric power of 10 kW. It was decommis-
sioned again in 2013.
References 57
Further Reading
• Drew, B., Plummer, A.R., Sahinkaya, M.N.: A review of wave energy converter
technology. J. Pow. Eng., Part A, 223,8, 887–902 (2009).
• López, I., Andreu, J., Ceballos, S., Martínez de Alegría, I., Kortabarria, I.: Review
of wave energy technologies and the necessary power-equipment. Renew. Sustain.
Energy Rev. 27, 413–434 (2013).
Some examples of wave energy converters:
• CETO: carnegiece.com/ceto-technology
• Azura wave: http://azurawave.com
• Statkraft Osmotic: www.power-technology.com/projects/statkraft-osmotic/.
References
1. Elmore, W.C., Heald, M.A., Stumpf, F.B.: Physics of waves. Jour. Acoust. Soc. Am. 81, 204–204
(1987)
2. Cornett, A.: A global wave energy resource assessment. Sea Technol. 50, 59–64 (2009)
3. Yemm, R., Pizer, D., Retzler, C., Henderson, R.: Pelamis: experience from concept to connection.
Phil. Trans. Roy. Soc. A 370, 365–380 (2012)
Wind Energy
5
Abstract
In this chapter the principles of wind, its energy content and the forms of conver-
sion are explained. The maximum efficiency of turbines is derived, followed by
a discussion of different types of real wind engines, their loss mechanisms and
possible routes for optimization.
5.1 General Considerations
For the existence of weather phenomena such as wind a planetary atmosphere is

necessary. In our solar system, not all planets have a sufficiently thick atmosphere
or do not even have one at all. Mercury for example has an “atmosphere” thinner
than today’s best achievable vacuum. It seems that all particles that could build up an
atmosphere are blown away by the solar wind. Jupiter on the other hand as a gaseous
planet consists nearly only of one thick atmosphere.1 The different colorful cloud
bands and the distinctive great red spot show the vivid motion in this atmosphere.
In general wind is a directed motion of atmospheric masses. On our Earth the
wind pattern is mainly determined by three influences:
• Convection due to different surface temperatures

• Coriolis force due to Earth rotation (F = 2m v × ω , see below)
• Local high and low pressure zones
Thus, wind energy is just a transformation of solar energy, caused by the solar
irradiation and the inhomogeneous temperature of the Earth’s surface. The global
energy in wind corresponds to about 2% of the solar irradiation energy. Apart from
local and temporal aberrations we can simplify the system to a more or less steady,
1 Of course this is an exaggeration: at the higher pressures inside the gases become liquid. Whether
Jupiter has a solid core or not, however, has not been proven yet.

60 5 Wind Energy
Fig. 5.1 Schematic view of the global wind patterns on Earth. See text for details
global wind pattern as shown in Fig. 5.1. The high solar irradiation at the equator
heats the air there, so that it rises upwards and flows towards the poles in six, large
convection rolls. The air masses coming from the equator start sinking already at
about 30◦ N/S. Cool air, coming from the poles and flowing towards the equator on
the Earth surface warm up on their way and rise at about 60◦ N/S.
The air masses moving in these convection rolls develop an increasing component
of their surface velocity v perpendicular to the rotational axis of the Earth the closer
they are to the poles. Thus, they are increasingly influenced by the Coriolis force
(FC , cf. Fig. 5.2). On the northern hemisphere, poleward winds are deflected to the
right, on the southern hemisphere to the left. At the equator, the influence of FC is
only relevant for rising air. Altogether, this leads to an east/west deflection of the
convection pattern. Note that for a given convection cell the direction of deflection
due to the Coriolis force is opposite for the air high up in the atmosphere (leading to
the formation of jet streams) and the air directly above the Earth surface which we
realize as global winds.
Apart from the rather small influence of local high and low pressure zones, which
are more important for the (short time) local weather, we can summarize these effects
with the development of three cells2 :
Tropical or Hadley Cell
The deflected air masses cause rather steady NE and SE winds3 called trade winds
or Hadley circulation. Since these two wind belts meet close to the equator, there
2 The free, educative website earth.nullschool.net displays the life movement of Earth’s wind and
ocean streams.
3 The direction of winds is named after the point of compass from where they come.
5.2 Energy Content of Wind 61
Fig. 5.2 Illustration of the Coriolis force at different positions on the Earth surface
exists a global equatorial low pressure channel, called the intertropical conversion
zone (ITCZ), which varies in its exact position during the year.
Moderate or Ferrel Cell
Here the Coriolis deflection causes a west wind belt, the westerlies. Especially in
higher altitudes, close to the tropopause, there exist very fast, narrow and meandering
west wind belts, called the jet-stream.
Polar Cell
Here the cold air from the poles flows towards the equator and warms up, until at
around 60◦ N/S it rises again and moves backwards. The so called Polar front between
Polar and Ferrel cell is very unstable due to turbulences and leads to a meandering
border line in 4–6 so called Rossby waves.
5.2 Energy Content of Wind
After discussing the global wind patterns of the Earth, we now take a look at the
energy stored in wind. Since wind is a motion of particles, we consider the kinetic
energy of streaming air of a cylindrical volume element dV (cf. Fig. 5.3):
1
dE kin = dmv 2
2
We can rewrite this expression by using the following relations for the mass
element:
dx
dm = ρdV dV = Adx v=
dt
⇒ dm = ρ Av dt
62 5 Wind Energy
Fig. 5.3 Geometry for

calculating the energy
content of wind: a volume
element dV = Adx
containing the mass dm is
moving with wind velocity v
along a cylindrical cross
section with area A
The mass flux ṁ is thereby defined as:
dm
= ṁ = ρ A v
dt
This leads to the final relations for kinetic energy and the power density per unit area
of streaming air:
1 3
dE kin = ρv A dt
2
P 1 dE kin 1
= = ρv 3 (5.2.1)
A A dt 2
The density of air at normal pressure (i.e. p = 1 bar) at sea level is ρ ≈1.2 kgm−3 .
It can be calculated from the ideal gas equation:
p·M p
ρ= =
R·T RS · T
where Rs is the specific gas constant R/M and T the absolute temperature.
• Temperature: For a temperature variation from −25 ◦ C to +35 ◦ C4 the density

varies only by ρ = 0.27 kg m−3 from 1.42 to 1.15 kg m−3
• Pressure: The air pressure varies mostly with height, as can be derived from the
barometric height formula. It ranges from normal pressure at sea level 1013.25
hPa to about 325.4 hPa at the top of Mt. Everest.
However, in practice the contributions by the variation of ρ are negligible, since a

wind turbine is most likely installed at a fixed height. Therefore, the arriving wind
power density is mainly affected by variations of the velocity. From these simple
4 At normal pressure at sea level.

5.3 Efficiency of Wind Turbines 63
considerations we can see that the energy content of wind is determined by the long-
term average wind velocity v and is proportional to v 3 . Thus, wind energy harvesting
systems should be located in areas with a high average wind speed over the year.
For a wind speed of v = 10 m s−1 and a regular air density of ρ = 1.2 kg m−3
we find a power density of P/A = 600 W m−2 . A small wind turbine with a
rotor area of A = 100 m2 receives an incoming power of P = 60 kW.
The distribution of wind speed v varies strongly as a function of time and place
(cf. Fig. 5.4), whereas its variation of v with height is weaker and dependent on the
average wind friction with the surface. An empirical relation for this dependence is:
f∗
h
v = v10m
10m
where f ∗ is a surface-dependent friction coefficient: it ranges from f ∗ ≈ 0.12 for

flat surfaces (e.g. the sea) to f ∗ ≈ 0.4 in cities with high buildings. The height
h above the surface is normalized to 10 m. Typical wind velocities in Europe lie
between 0 < v < 30 m s−1 . In storms, speeds up to 60 m s−1 (i.e. 200 km h−1 ) can
be reached. However, the commercial use of wind energy begins for a minimum
wind speed of 4 m s−1 . As shown in Fig. 5.5, the relative wind speed distribution
at a given site can be approximated quite well by the so-called Weibull or Rayleigh
distributions:

k v k−1 v k
f W eibull (v) = exp − (5.2.2)
a a a

π v2 π v2
f Rayleigh (v) = exp − (5.2.3)
2 v̄ 2 4 v̄ 2
5.3 Efficiency of Wind Turbines
Wind turbines are the most important technology to convert wind energy into electric-
ity. Though many different designs and realizations can be found, the ideal efficiency
of a wind engine can be calculated independent of the type of turbine, shown first
64
Fig. 5.4 Wind resource map showing the mean wind speed at 100 m above surface. Map obtained from the Global Wind Atlas 3.1, a free, web-based application
developed, owned and operated by the Technical University of Denmark (DTU) [1]. The permission to reprint is greatfully acknowledged
5 Wind Energy
Fig. 5.5 A typical local wind speed distribution with different mathematical approximations as
described in the text
Fig. 5.6 Generalized ansatz

for the efficiency calculation
for wind engines according
to Betz. A1 and v1 are the
area and wind velocity in
front of the rotor plane, A2
and v2 behind the rotor
plane. The pressure p and
therefore also the density ρ
remain unchanged
sufficiently far from the rotor
by Betz [2]. It limits the maximum efficiency which can be extracted by a turbine
based on the parameters shown in Fig. 5.6.
Let a wind engine interact with incoming air of speed v1 and density ρ1 in an area
A1 . The pressure in front of the wind turbine and in some distance behind it will be
the same. This means, that also the density of air is constant. Thereby the speed has
to drop after passing the turbine to v2 < v1 and flows through a larger area A2 > A1 .
A given mass element dm 1 = ρdV = ρ A1 dx = ρ A1 v1 dt is conserved when
going through the rotor plane. Assuming that ρ1 = ρ2 = ρ this gives (continuity
equation):
66 5 Wind Energy
ρ A1 v1 dt = ρ A2 v2 dt ⇒ v1 A1 = v2 A2
The wind velocity v R in the rotor plane can be calculated using the Bernoulli equation:
1
vR = (v1 + v2 )
2
According to Sect. 5.1 the incoming wind power is:
P 1
P1 = A1 = ρ A1 v13
A 2
Accordingly the outgoing wind power is:
P 1
P2 = A2 = ρ A2 v23
A 2
The power extracted by the wind engine is then:
1 1 1
PR = P1 − P2 = ρ(A1 v13 − A2 v23 ) = ρ(A1 v1 v12 − A2 v2 v22 ) = ρ A1 v1 (v12 − v22 )
2 2

2
=X =X
Alternatively, using ρ Av = dm/dt this can be written as:
ṁ 2
PR = v1 − v22 (5.3.1)
2
Thus, PR is proportional to the air mass flow and the change in kinetic energy (∝ v 2 ).
In the rotor plane we have (using A R = A1 ):
1
v = vR =
(v1 + v2 ) ṁ = ρ A R v R
2
1
⇒ ṁ = ρ A1 (v1 + v2 )
2
2
1 1 v2 v2
PR = ρ A1 (v1 + v2 )(v1 − v2 ) = ρ A1 v1 1 +
2 2 3
· 1−
4 4 v1 v1
Since the incoming power is P1 = 21 ρ A1 v13 the efficiency c P of the (ideal) wind
engine is (Fig. 5.7):
2
PR 1 v2 v2
c P := = 1+ · 1−
P1 2 v1 v1
Fig. 5.7 Betz efficiency of a

generalized wind engine
versus the speed ratio v2 /v1 .
The maximum of the
efficiency is about 60%,
occurring for a speed ratio of
1/3
Derivation of the condition v R = 21 (v1 + v2 ) for the wind velocity v R in the

rotor plane using the Bernoulli equation p + 21 ρv 2 = const. (Indices: 1: far in
front of the rotor, 2, far behind it, −R: just before, +R: right after the rotor).
1 1
p1 + ρv12 = p−R + + ρv−R2
(5.3.2)
2 2
1 2 1
p+R + ρv+R = p2 + + ρv22 (5.3.3)
2 2
Here v−R = v R = v+R can be assumed in good approximation. Subtraction
of these two equations yields with p1 = p2
1 2
ρ v1 − v22 = p−R − p+R (5.3.4)
2
The force acting on the rotor due to the pressure difference is:
F = A R ( p−R − p+R )
On the other hand the force is given by the time derivative of the momentum:
dm
F= (v1 − v2 ) = A R ρv R (v1 − v2 )
dt
This results in:
⇒ p−R − p+R = ρv R (v1 − v2 )
Inserting this into (5.3.4) above finally gives:
1
vR = (v1 + v2 ) (5.3.5)
2
68 5 Wind Energy
5.4 Types of Rotors
5.4.1 Drag-Type Rotors
Drag-type rotors use the drag force of solid bodies in streaming air with velocity v:
1
FW = cW ρ v 2 A
2
where A is the (effective) area perpendicular to the wind direction. The drag coeffi-
cient cW depends on the shape of the body as shown in Table 5.1.
A simple drag rotor is sketched in Fig. 5.8. The maximum drag force on the upper
panel is:
1
FW = cW ρ A(v1 − u)2
2
Table 5.1 Drag coefficients in air for bodies with different shapes
Shape
Stream-lined 0.03
Wing-shaped 0.03-0.1
Sphere 0.4
Semi-sphere 0.8
Plane 1.2
Hollow semi-sphere 1.4

5.4 Types of Rotors 69
Fig. 5.8 Simple example of a drag-type wind engine. Two panels with area A can rotate with a
circumferential velocity u around a common axis perpendicular to the direction of the wind velocity
v1 . A trench acting as wind shield for the lower half of the engine breaks the symmetry of the panel
arrangement
where u is the circumferential moving speed of the panel. This gives a power taken
from the wind by the rotor of:
dE d
PR = = FW dx = FW · u
dt dt
The efficiency of this drag-drag type wind engine is accordingly:
PR cW 21 ρ A(v1 − u)2 · u
cp = =
2 ρ A · v1
P1 1 3
2
u u
c P = cW 1− · (5.4.1)
v1 v1
The maximum of c P is again obtained for u/v1 = 1/3. The maximum value is
c P,max = cW (4/27), so that for cW ≤ 2 the maximum efficiency of drag type wind
engines is ≤ 30%. This is a factor of two lower than the ideal Betz efficiency, showing
that drag-type rotors are sub-optimal for wind energy harvesting.
5.4.2 Lift-Type Rotors
Better efficiencies, close to the Betz limit, can be obtained for lift-type rotors. These
use the lift force of a wing-shaped rotor:
Due to the specific wing profile, the air above the wing has to pass a longer
trajectory than the air below the wing. As shown in Fig. 5.9, this results in a larger
70 5 Wind Energy
Fig. 5.9 Schematic drawing of the air flow around a lift-type rotor and the resulting difference in
pressure above and below the wing
Fig. 5.10 Forces acting on a wing with a finite angle of attack α relative to direction of wind
air velocity v+ > v1 above the wing and thus, according to the Bernoulli equation,
in a lower pressure p+ < p1 there. This gives rise to a lift force FA according to the
following equation, where A is the area of the wing parallel to the wind and c A the
so-called lift coefficient, defined in a similar way as the drag coefficient above:
FA = ( p1 − p+ )A (5.4.2)
1
= c A ρ A v12 (5.4.3)
2
An important parameter for the operation of lift-type rotors is the angle of attack
α between the wind direction and the profile axis of the wing (cf. Fig. 5.10).
A variation of α allows to change the effective lift coefficient c A , which is accom-
panied by a smaller change of the drag coefficient cW . For a given wing profile, this
variation is documented in the so-called polar profile plot. A typical example of such
a plot is shown in Fig. 5.11.
For a given angle of attack, c A and thus the lift force vanish (stall condition). A
second stall condition may occur for large angles of attack in very strong winds,
if the angle of attack is not actively corrected. Then, with increasing wind velocity
v1 and for constant circumferential turning velocity u, the angle of attack will also
Fig. 5.11 Polar plot for an optimized profile, showing the dependence of c A and cW on the angle
of attack. In particular, c A vanishes at the stall angle (−6◦ in this case)
Fig. 5.12 Stall condition in strong winds, caused by a too large angle of attack
increase, as is shown in Fig. 5.12. This eventually causes the drag coefficient to
strongly increase and the lift coefficient to decrease. As will be discussed further
below, for lift-type rotors the wing or blade profile needs to vary along the blade
to account for the different air velocities along the blade, i.e. from the center of
the rotor to the tip of the blade. Also, in modern wind engines the angle of attack
is continuously adjusted to optimize the lift coefficient and, thus, the lift force for
different wind velocities. This is called pitching.
Another important parameter of a wing profile is the ratio of FA and FW , which
is called glide number G 5 :
FA
G :=
FW
5 Optimized profiles can reach glide numbers close to 100.

72 5 Wind Energy
Fig. 5.13 Schematic drawing of a two-bladed lift-type rotor as seen from the direction of the
incoming wind
Fig. 5.14 View (rotated by 90◦ ) from the top of the rotor plane showing the velocity and the angle
of attack
For further, more quantitative aspects we have to take a closer look at the geometry
of lift-type rotors. As shown in Fig. 5.13 for a two-blade rotor, the direction of the
incoming wind (v1 ) is perpendicular to the rotor plane. The same holds for the rotor’s
angular velocity ω. At a distance r from the rotor center, a blade section moves with
the local circumferential velocity u = ω · r within the rotor plane.
Turning the rotor plane by 90◦ and looking from the top (cf. Fig. 5.14), we see
that the velocity of attack v A , i.e. the velocity with which the air hits the blade front,
is the vector sum of the wind velocity and the blade velocity, vA = v1 + u. The
corresponding angle of attack is the angle between the profile axis and the direction
of v A .
Fig. 5.15 Geometry of the lift force F A and the drag force FW perpendicular and parallel to the
direction of the velocity of attack
The ratio
u
λ := (5.4.4)
v1
is called speed ratio.
• For drag type rotors v1 and u are co-linear, so that u ≤ v1 ⇒ λ ≤ 1. Opti-

mized efficiency is reached for λ ≈ 1/3
• For lift-type rotors u(r ) = ωr can become much larger than v1 ⇒ λ > 1.
Depending on the tip speed ratio λs = ωv1R for the tip of the rotor (r = R)
one distinguishes:
– λ S < 3: slow rotors
– λ S ≥ 3: fast rotors
The direction of the velocity of attack also defines the direction of the lift force
F A and the drag force FW , according to Fig. 5.15.
The total force on the rotor is then F R = F A + FW ≈ F A to a good approxi-
mation, because for good profiles the glide number G = FA /FW >> 1. Writing
F R = F R, + F R,⊥ ( and ⊥ to the rotor plane) we get
v1
FR, ≈ FA cos(γ ); FR,⊥ ≈ FA sin(γ ); cos(γ ) =
vA
74 5 Wind Energy
The power harvested by a blade segment dr at a distance r from the rotor axis is
then:
v1
dP(r ) = dF R, · u(r ) = ωr cos(γ )dFA = ωr dFA
vA
with
1 1
FA = c A ρ Av 2A ⇒ dFA = c A ρv 2A

dA
2 2
=b·dr
where b is the breadth of the blade at position r . Together this gives:
v1 1 2
dP(r ) = ωr c A ρv b dr |v 2A = v12 + u 2
vA 2 A

1 u2 u(r )
= c A ρωv1 1 + 2 b r dr
2
|λ(r ) =
2 v1 v1
1
⇒ dP(r ) = c A ρωv12 1 + λ2 (r )b · r · dr (5.4.5)
2
Then, a simple rotor with z blades of length R and constant breadth b with a varying
lift coefficient c A (r ) along the blade harvests a total wind power of:
R
1
P= dP(r ) = ρωv12 bz 1 + λ2 (r )c A (r )r dr
2 0
Fig.5.16 Typical variation of the profile of a rotor blade between center and tip. The blade segments
have been turned by 90◦ out of the rotor plane
5.5 Optimization of Wind Turbines 75
• P is proportional to v12 and ω

• u = ωr varies strongly from the beginning to the tip of a blade, so that the
direction and size of the velocity of attack v A change accordingly. Therefore,
as shown in Fig. 5.16, the blade profile has to be adjusted along r and as a
consequence, the lift coefficient c A becomes a function of r .
• As already discussed in connection with the profile polar plot, the angle of
attack changes c A and cW which can be used to optimize P for different v1
(“pitching”). In particular cW can be enhanced so much, that the rotor type
changes from lift to drag.
To estimate the optimum number z of blades for a wind engine, consider a trape-
zoidal blade with an average breadth b̄ at a distance r̄ = 2/3 R. Then, the total area
coverage of a rotor with z such blades is:
z b̄
B=
2π r̄
The optimal coverage Bmax depends on the tip speed ratio λ S := (ω R)/v1 : for the
maximum rotor efficiency c P = 16/27 one can then deduce:
⎡ ⎤
−1
8 1 ⎣ r̄ 2 4 2 r̄ 2 ⎦ 1
Bmax = 1 + λS · 2 (5.4.6)
9 cA R 9 R λS
Thus, the optimum number of blades z has to be reduced with increasing tip speed
ratio λ S , which in turn needs to be adjusted to remain close to the Betz efficiency
maximum.
5.4.3 New Types of Wind Engines
A very recent development and still in pre-industrial phase are bladeless wind engines
based on oscillating towers (vortexbladeless.com) or energy kites, flying several
hundreds of meters above ground. The interested readers are referred to the literature
and provided links.
5.5 Optimization of Wind Turbines
In deriving the maximum Betz efficiency of wind engines, so far all losses have
been neglected. In reality, there are three main loss mechanisms which have to be
considered:
76 5 Wind Energy
Fig. 5.17 Left: Blade breadth b(r ) and differential segment dr as a function of distance r from the
center of the rotor. Right: Velocity triangle defined by the turning velocity u, the wind velocity v R in
the rotor plane, and the resulting velocity of attack v A . v R is shown for the optimum Betz condition
v2 = v1 /3. α is the angle of attack, t(r ) the effective blade thickness determining the friction force
FW (r )
profile losses: due to the non-zero cW caused by friction.

tip losses: due to the fact that c A → 0 as r → R, leading to an effective shortening
of the blades.
wake losses: due to the fact that the streaming air is deflected radially by the rotor
blade, so that behind the rotor v2 is not parallel to v1 anymore.
5.5.1 Optimized Radial Profile of a Lift-Type Blade
In order to minimize friction losses and to maximize the extracted power, the wing
breadth b(r ) has to be optimized as a function of the distance r from the center of
the rotor. For a first approximation, we consider the schematic drawing in Fig. 5.17.
Assuming that the blade is operated close to the optimum Betz efficiency, the
following relations hold:
3 v1
2
1 1 2 2 R
v2 = v1 ; v R = (v1 + v2 ) = v1 ; tan(90◦ − γ ) = =
3 2 3 ωr 3 λS r
The resulting differential lift and friction forces acting on the blade segment dr are:
ρ 2
dFA = v c A (r )b(r )dr
2 A
ρ
dFW = v 2A cW (r )t(r )dr
2
The extracted rotor power according to the Betz approximation is:
16 ρ 3 2
PR = c P · P1 = v πR
27 2 1
5.5 Optimization of Wind Turbines 77
The corresponding differential power is:
16 ρ 3
dPR = v 2πr dr
27 2 1
To extract this power with z blades with the breadth b(r ) to be determined we need:
dP = z FA sin(90◦ − γ )ωr dr = dPR

ρ 16 ρ 3
⇒z v 2A (r )c A (r )b(r ) sin(90◦ − γ )ωr dr = v 2πr dr
2 27 2 1
1 16 2πr v13
⇒b(r ) =
z 27 c A (r ) v 2A ωr sin(90◦ − γ )
ωR
By introducing the tip speed ratio λ S = v1 , this can also be written as:

1 8 r 2 4 −1
⇒ b(r ) = 2π R λS λS
2 +
z 9c A (r ) R 9
For fast rotors (λ S > 3) and r /R ≥ 0.1, this relation can be approximated by:
1 8 R
b(r ) ≈ 2π R (5.5.1)
z 9c A (r ) r · λ2S
This means, that b(r ) should decrease hyperbolically like 1/r along the blade and
scale with the inverse square of the tip speed ratio.
5.5.2 Losses
Taking into account the friction losses of real profiles via cW (r ) > 0, the real differ-
ential power becomes:

d Pr eal = zωr FA sin(90◦ − γ ) − FW cos(90◦ − γ ) dr
The ratio d Pr eal /d Pideal is called the “efficiency” of the profile used:
FW
η pr o f ile = 1 −
FA tan(90◦ − γ )
With the glide number G = FA /FW and tan(90◦ − γ ) = 23 R/(r λ S ) this gives:
1 3 r λS
η pr o f ile = 1 − =1− (5.5.2)
G tan(90◦ − γ )
RG
2
profile losses
78 5 Wind Energy
Fig. 5.18 Close to the blade tips, air can flow around the rotor plane without contributing to the lift
force. As a consequence, the effective lift coefficient c A decreases towards the end of a blade
Profile Losses
The friction or profile losses increase towards the blade tip (r → R) and are pro-
portional to λ S . If for a given profile G is independent of r , the integration over the
entire blade gives:

16 ρ 3 R 16 ρ 3 2 λS
PR = v1 η pr o f ile 2πr dr = v1 π R 1 −
27 2 0 27 2 G
Tip Losses
These losses6 are due to the pressure gradient across the rotor plane. This causes
air to flow radially around the blade tips, as is shown in Fig. 5.18. As a result, c A (r )
decreases close to the tips. An empirical formula for tip losses of fast rotors with
λ S > 3 was provided by Betz:
1.84
ηti p = 1 −
zλ S
Wake Losses
These are caused by the forces of the blades acting on the streaming air (actio =
reactio). The rotating blades give the passing air also a radial velocity component
perpendicular to the direction of the incoming wind. Thus, behind the rotor plane,
outgoing air follows a spiraling motion. This reduces the maximum efficiency dras-
tically, especially for slow rotors (Fig. 5.19).
The derivation of wake losses as a function of the tip speed ratio was provided by
Schmitz [3]. According to his lengthy calculation, the maximum power that can be
extracted from a wind engine with wake losses is given by:
r 2 sin3 ( 2 α1 ) r
ρ 3 2 1
PSchmit z = v πR 4λ S 3
d
2 1 R sin2 (α1 ) R

0
c P,Schmit z

with α1 = arctan λRS r . The integral constitutes the Schmitz efficiency, c P,Schmit z ,
as a function of the tip speed ratio. In addition, the overall Betz efficiency limits
6 Tip losses are not due to turbulences at the tips. Turbulence losses are taken into account by the
profile losses.
5.6 Some Practical Aspects of Wind Engines 79
Fig. 5.19 Illustration of the origin of wake losses due to a radial velocity component v2,radial
behind the rotor plane
Fig. 5.20 Maximum

efficiency of wind engines
including wake losses
according to Schmitz. The
wake losses are indicated by
the hashed area
applies. Thus, as shown in Fig. 5.20, the overall efficiency of wind engines including
wake losses but neglecting friction losses approaches c P,Bet z asymptotically for
λ S → ∞ and goes to 0 for λ S → 0.
5.6 Some Practical Aspects of Wind Engines
Finally, we briefly want to discuss some other practical aspects of modern wind
engines. Table 5.2 summarizes typical dimensions of engines in the MW design
power range. Note that most modern engines have a horizontal axis (horizontal axis
wind turbine (HAWT)) and use three blades.
The reason for this becomes clear when considering real efficiencies in the frame
of the efficiency limits set by Betz and Schmitz as shown in Fig. 5.21. Slow rotors are
subject to large wake losses and were only used historically due to limited mechanical
stability of construction materials. The best efficiencies with a maximum close to
50% are currently reached with fast rotor 3-blade HAWTs operating around tip speed
80 5 Wind Energy
Table 5.2 Typical dimensions of various HAWT systems, depending on their nominal power
Type Power [MW] Height [m] Rotor Ø [m] Rotor area [m2 ]
3 bladed HAWT 0.5 50 40 1300
1 70 55 2400
2 90 90 6400
3 100 100 7900
V236- 8 138 236 43742
15.0 MWTM
[5]
Fig. 5.21 Typical efficiencies of historical and present wind turbines as a function of the tip speed
ratio. The ideal Betz and Schmitz efficiency limits are indicated for comparison
ratios of λ S = 7. The optimum number of three blades is a result of the analysis of

the optimum area coverage versus blade number and tip speed ratio discussed above.
For fewer blades, the turbines have to be operated at higher tip speed ratios, which
results in higher friction losses, since the friction force increases with the square of
the velocity of attack.
A frequent point of concern about wind turbines is the acoustic noise generated
by them. This has to be taken into account particularly when planning wind parks
close to populated areas. Figure 5.22 shows how the sound intensity (measured in
dB(A), decibel acoustic) emitted by a turbine decreases with distance from the rotor
position. For comparison, a sound level of 100 dB(A) in close proximity would be
characteristic for a very noisy industrial area. In contrast, the noise level of 40 dB(A)
in a distance of 1 km corresponds to a very quiet recreational area.
References 81
Fig. 5.22 Sound level of a typical modern wind turbine in dB(A) as a function of the distance to
the rotor
Further Reading
• Time and spatially resolved animated world climate map (temperature, winds,
currents, precipitation, particulates, etc.) with data from meteorological satellites:
https://earth.nullschool.net/
• Bladeless oscillating column wind engine: vortexbladeless.com
• On shore wind energy kite system: skysails-power.com.
References
1. Map obtained from the Global Wind Atlas 3.1, a free, web-based application developed, owned
and operated by the Technical University of Denmark (DTU). The Global Wind Atlas 3.1 is
released in partnership with the World Bank Group, utilizing data provided by Vortex, using
funding provided by the Energy Sector Management Assistance Program (ESMAP). For addi-
tional information. https://globalwindatlas.info
2. Betz, Albert: Das Maximum der theoretisch möglichen Ausnützung des Windes durch Windmo-
toren. Zeitschrift für das gesamte Turbinenwesen 26, 307–309 (1920)
3. Schmitz, G.: Theorie und Entwurf von Windrädern optimaler Leistung, Wiss. Zeitschrift der
Universität Rostock, 5. Jahrgang (1955/56)
4. Ahrens, U., Diehl, M., Schmel, R. (eds.): Airborne Wind Energy. Springer (2013)
5. Prototype 15 MW wind turbine from Vestas (DK): https://vestas.com. Accessed from 27 July
2022
Thermal Energy
6
Abstract
This chapter gives a short introduction to the basics of geothermal energy and how
this potential can be exploited. The physics of heat pumps are explained as well
in this sub-chapter. The second part deals with the principles of solar (thermal)
energy conversion, including the physical principles of solar irradiation, its usage
and some examples.
Thermal energy is generally available as waste heat in exothermal processes or can

be harvested from the Earth (geothermal energy) or the radiation of the Sun (solar
thermal energy). Depending on the temperature level of the heat source, thermal
energy can be used directly or in combination with a heat pump for heating purposes
or as process heat. Moreover, it can be transformed into e.g. electrical energy via
turbine-driven generators or thermoelectric generators (discussed in Chap. 9). It also
can be transported or stored on an industrial scale. Thus, thermal energy constitutes
a ubiquitous and very versatile form of energy.
6.1 Geothermal Energy
6.1.1 Contributions to Geothermal Energy
The Earth was formed about 4.5 billion years ago by a process called differential
accretion of solar dust: elements with a high density (e.g. Fe, Ni) are dominant in
the center, less dense minerals (SiO2 , Al2O3 , CaO, MgO) are dominant in the outer
mantle. The present structure of the earth is mainly known from the analysis of
earthquake waves and numerical modelling and is summarized in Table 6.1.
Some points to note:
• The inner core temperature is higher than the surface temperature of the Sun. Yet
the core is solid due to the high pressure. The core contains the following elements:

84 6 Thermal Energy
Table 6.1 Structure of the Earth

Depth Density Pressure Temperature Phase
[km] [tm−3 ] [kbar] [K]
Inner core 5100–6370 13.5 3700 ≈7000 Solid
Outer core 2900–5100 10–12 3340 ≈3000 Liquid
Lower mantle 700–2900 4–5 1350 ≈2000 Liquid
Transition 400–700 4 300 1300–2000 Viscous
zone
Upper mantle 35–400 3–4 260 600–1300 Viscous
Crust 0–35 2.8–3.0 9 300–600 Solid
80% Fe, 7% Si, 5% Ni and 4% O. It rotates by about 0.3...0.5◦ / year faster than
the mantle. This is called super-rotation and results in about one additional turn
in ≈ 1000 years).
• The liquid outer core is responsible for the earth magnetic field.
• The Earth’s crust consists of an oceanic crust which is rather thin (5–10 km) and a
thicker continental crust (30–60 km). Both swim on the so-called asthenosphere.
The crust consists to 50% of O.
• The deepest drill holes so far are: 12 km (Kola peninsula, Russia); 10 km (Oberp-
falz, Germany).
The average thermal energy flow through the Earth’s surface amounts to only
about 60 mW m−2 . This is very small compared to on the average several 100W
m−2 arriving on the earth surface due to solar irradiation. However, in local hot spots
the geothermal energy flow can be higher by many orders of magnitude.
The geothermal flow has two components:
1. 50% come from cooling of the inner Earth: during accretion of the earth, the
dissipated gravitational potential and kinetic energy was sufficient to reach an
average temperature of 2000 K. Due to the small thermal conductivity of the crust
(≈2–3 W m−1 K−1 ), cooling due to radiation stopped very quickly and a stable
crust developed. The geothermal flux is caused by a stable temperature gradient1 :
heat flux Q̇ = 0.06 Wm−2

thermal conductivity λ = 2 W/mK
T Q̇
temperature gradient = ≈ 0, 003 K/m = 3 K/km
x λ
Strong deviations from the average value occur close to sites where liquid magma
comes close to the surface: the resulting geothermal anomalies can have local
heat fluxes up to 0.3W m−2 or higher.
1 The so-called geothermal gradient, e.g. known from deep mines and drill holes.
6.1 Geothermal Energy 85
Table 6.2 Most frequent radioactive decays in the Earth

Element Decay T1/2 [a] Energy [MeV]
235 U α 7 ×108 ≈100
238 U α 4.5 ×109 ≈40
232 Th α 1.4 ×1010 ≈50
40 K β−, β+ 1.3 ×109 ≈1.5
2. The second 50% of geothermal heat are due to radioactive decays: For an average
mass concentration of radioactive isotopes of 10g t−1 stone (10ppm) this gives:
6 × 1023 atoms 1
Q̇ = 10 g/t · 50 MeV ≈ 10µW t−1 ≈ 3µW m−3
230 g 5 × 109 a
The total power reaching the Earth surface due to radioactive decays is about
16 TW, corresponding to an average flux of approximately 0.03W m−2 ) (Table 6.2
and Fig. 6.1).
• In the young Earth the radioactive contribution was three times higher.
• 99% of the Earth are hotter than 1000◦ C. From this, the total thermal energy
stored in the Earth can be estimated to about 3 × 1015 TW h (For compari-
son: the rotational energy of the Earth is about 6 × 13TW h).
• Strictly speaking, geothermal energy is not renewable.
6.1.2 Use of Geothermal Energy
In 2020 about 50 GW of direct geothermal energy (heating, process heat) and 15 GW

for production of electricity were used, mainly in China, Sweden, USA, Iceland,
Turkey, and Germany [1,3,5].
In Germany, the currently installed geothermal power for electricity generation is
around 0.5 GW. In addition, about half a million heat pumps are in use, mainly for
the heating of buildings.2 In addition to heat pumps, one distinguishes the following
sources of geothermal heat:
• high enthalpy sites close to geothermal anomalies

• low enthalpy sites, e.g. natural aquifers
• artificial hot dry rock (HDR) sites
2 According to German law, geothermal heat is a ”bergfreier Bodenschatz”: if you own a piece of
land, you do not own the heat underneath (§3 BergG).
86
Fig. 6.1 Schematic map of the global heat flow (from the crust/surface). The variation of the heat flow ranges from ca. 30mW m −2 in Antarctica, to above
110mW m −2 in the pacific ocean at locations of the pacific fire ring. More accurate data and representations can be found at the given resources [2,3]
6 Thermal Energy
Fig. 6.2 Schematic view of a high enthalpy site close to a magma duct. Red arrows indicate main
water flows. Typical lateral temperature profiles are also indicated
High enthalpy sites can be found close to active or extinct volcanoes or at other
special hot spots, where one can extract water/steam mixtures (2-phase mixtures)
with temperatures between 100 and 300◦ C, which can be used directly for electric-
ity production with turbines. Hot 2-phase mixtures exist at high pressures (10–100
bar), so that the boiling point of water increases to about 200–300◦ C. Some typical
properties of such high enthalpy sites are listed below:
• To extract a thermal power of 1 MW using water at 150◦ C one needs a volume

flow V̇ of about 2,5 Ls−1 .
Q̇ = T · cV · V̇
with cV ≈ 4 kJL−1 K−1 , and lowering the water temperature by T = 100 K.

• The 2-phase mixture very often is contaminated by sulphur and other minerals,
giving rise to unwanted corrosion and smell. This can be avoided by using a heat
exchanger, separating the geothermal water from the turbine circuit. Ideally, the
resulting cold water is pumped back into the site.
• Even for extinct volcanoes there can still exist magma ducts that remain hot for
several million years.
• Typical conditions at high enthalpy sites are shown schematically in Fig. 6.2.
European high enthalpy sites for electricity production are currently exploited
mainly in Turkey, Southern Italy and Iceland, with a total installed power of about
3 GW [1].
88 6 Thermal Energy
Table 6.3 Low enthalpy sites in Germany

Region Max. extractable power (GW) For a duration of (years)
Norddeutsches Becken 100 100
(Northern German basin)
Oberrheingraben 20 100
(Upper rhine valley)
Süddeutsches Molassebecken 30 100
(Southern German molassis
basin)
Fig. 6.3 Schematic layout of

a hot dry rock (HDR)
geothermal site
Low enthalpy sites typically make use of deep (1–2 km) aquifers with temperatures
around 100◦ C. Here, the extracted water can quickly be replenished because of the
high hydraulic conductivity in such aquifers.
In Germany there are three main regions for low enthalpy geothermal energy, as
listed in Table 6.3.
One example for a very successful low enthalpy site is the geothermal plant in
Erding, Bavaria, which extracts hot water with a relative low temperature of 65◦ C
from the Southern German molassis basin in a depth of about 2400 m with a flow
rate of 50 Ls−1 , generating a geothermal power of 35 MW. This is used to power the
currently largest thermal spa in the world (“Therme Erding”) and for the heating of
residential buildings.
In regions where no aquifers can be used, one can alternatively apply the hot-dry-
rock method (HDR) (Fig. 6.3). Here, one creates an artificial aquifer by fracturing
hot rock and injecting water to extract the stored heat. As an example, granite has a
heat capacity of cV = 840 J kg−1 K −1 and a density of 2800 kg m−3 . Starting from a
temperature of 200◦ C and cooling the extracted hot water to room temperature, this
would correspond to a specific energy density of 3 g CE per kg of rock.
A major problem of HDR plants is the small thermal conductivity of the unfrac-
tured surrounding rock. It is much smaller than the thermal conductivity of aquifers,
which is determined by the hydraulic conductivity according to the law of Darcy: if
(for vertical flow) v is the flow velocity, σW the hydraulic conductivity, H the height
of the water column above the flow region driving the flow and L the length of the
flow region, then the law of Darcy states:
Table 6.4 Porosities and hydraulic conductivities of different materials

Mineral Porosity σW
[%] [m/day]
Clay 45–60 < 10−2
Sand 30–40 1–500
Gravel 25–35 500–10 000
Granite 10−4 < 10−3
H
v = σW (6.1.1)
L
Together with the specific heat of the hot flowing water, this then results in
the effective thermal conductivity of the aquifer. The hydraulic conductivity σW
is mainly, but not only, determined by the porosity of the aquifer region. Some
examples are given in Table 6.4.
In the dry rock surrounding the HDR region, the thermal conductivity in contrast
is determined by the transport via atomic oscillations (phonons) and in general is
much lower. As a result, according to the estimate below, it would take very long
times for surrounding heat to flow back into the HDR region.
Consider a granite cavern with a volume of V = 100 × 100 × 100 m3 at a

starting temperature of T = 220◦ C. For a density of 3000 kg m−3 , the total
rock mass in the cavern is m = 3 × 109 kg.
• The total energy content of the HDR region in the cavern (when cooling
down to 20 C) is:
Q = T · m · cV = 200 K · 3 × 109 kg · 840 J kg−1 K−1 = 5 × 1014 J
• For extracting a heat flow of Q̇ = 10 MW, this reservoir will last for about
one year
• The heat transport from the surrounding back to the cavern is determined
by the normal thermal conductivity λ:
A
Q̇ = λT
d
With a surface area of the cavern of A = 6 × 104 m2 , assuming a tempera-

ture gradient of T /d = 1 Km−1 and the typical low thermal conductivity
of solid rock, it would take about 100 years to refill the extracted heat.
90 6 Thermal Energy
Fig. 6.4 Working principle

of a compression heat pump
If the temperature of the extracted geothermal water is not sufficiently high for
the intended use, the temperature is usually increased with the help of a heat pump.
Heat pumps are heat engines running in reverse. They convert work into heat and at
the same time pump additional heat from a cold to a warm reservoir (c.f. Fig. 1.2 in
Chap. 1). Heat Q C is extracted from a cold reservoir (at temperature TC ) and pumped
with the help of external work W into a warm reservoir (Q W at temperature TW ).
Energy conservation requires that Q W = Q C + W . Therefore, the ideal thermo-
dynamic efficiency of a heat pump is given by:
QW TW
ηhp = = >1
W TW − TC
The main technical components of a compression heat pump are shown in the
schematic diagram of Fig. 6.4.
1. The working medium is evaporated in a heat exchanger and thereby removes the
necessary heat of evaporation Q c from the cold reservoir.
2. The now gaseous medium is compressed in a compressor, thereby adding the
mechanical work W (as adiabatically as possible).
3. A second heat exchanger transfers the heat Q W to the warm reservoir, causing
liquefaction of the working medium.
4. A throttle between the two heat exchangers is used to adjust the different pressure
levels on both sides
The most common reservoirs for TC are outside air, groundwater or deeper sur-
face regions in the earth, where seasonal variations of the temperature are strongly
reduced. Some typical working media for heat pumps are listed in Table 6.5.
As a practical example, consider a floor heating system with a temperature TW =
50◦ C using groundwater with TC = 5◦ C as the cold reservoir. Transforming degrees
Table 6.5 Common working media for heat pumps

Medium Boiling point ◦ C
CCl2 F2 (no longer legal) –30
C3 H8 (propane) –42
NH3 –33
CO2 –57
Fig. 6.5 Qualitative seasonal

divergence between the heat
needed and heat pump
efficiency for building
heating with a heat pump
using outside air as the cold
reservoir: in the winter, TC is
low, so the temperature
spread T = TW − TC is
large. This causes a lower
efficiency, although the
required heat flux is large. In
the summer, TC approaches
TW , so the efficiency
increases, but much less heat
is required
Celsius into absolute temperatures T , this would be possible with an ideal efficiency
of:
50 + 273
ηhp,ideal = ≈7 (6.1.2)
50 − 5
Typical efficiencies of commercial heat pumps are between 2 and 6, depending

on the temperature spread TW − TC . This, for example, gives rise to a seasonal
divergence between the heating power of a heat pump and the amount of heat needed
in a building (Fig. 6.5).
The overall system efficiency ηsystem of a heat pump is lower than ηhp , because
the work W can only be provided with a finite efficiency ηW :
ηsystem = ηhp · ηW
Real heat pumps have a lower efficiency of ηhp,r eal ≈ 1/2 · η H p,ideal . If the exter-
nal work is provided by an electrically driven compressor, as it is commonly the case,
92 6 Thermal Energy
then also ηW ≈ 0.35, limited by the overall efficiency of commercial electrical power
plants. Then, the resulting system efficiency is ηsystem ≈ 0.35 · 3.5 ≈ 1. This means
that, viewed on a system-wide scale, the use of a heat pump only recovers the energy
lost during electric power generation in a thermal power plant. A better solution is
provided by the use of a combustion engine for W (ηW ≈ 0.25). If in addition the
waste heat of the combustion engine is used to increase TC (e.g. 5◦ C → 35◦ C), then
the overall efficiency would be increased to:
50 + 273
ηsystem = 0.25 · 1/2 ≈3 (6.1.3)
15
6.2 Solar Thermal Energy
The thermal use of solar radiation requires the efficient transformation of the broad
spectrum of solar photons impinging on the Earth into heat in the form of low
energy thermal vibrations (phonons) in the respective absorber materials. Only in
the small infrared portion of the solar spectrum, this transformation can occur directly
through the absorption of infrared photons by excitation of atomic vibrational modes
in molecules or solids. For the much larger part of the spectrum, the photons first
excite electrons into higher energy states, which subsequently decay back to the
electronic ground state by the generation of atomic vibrations. This process is called
thermalization and is governed by the details of the coupling between electrons and
phonons in the absorber material.
As the simplest example, we briefly discuss the electron-phonon coupling in
molecules.3 As shown in the sketches below, in isolated atoms photons can couple to
electrons only via transitions between the discrete energy levels of the atomic orbitals
driven by the oscillating electric field of the photons. In these processes, for photon
energies in the eV range, the atoms basically remain at rest because of the very small
momentum of the photons. In molecules, however, the situation is quite different.
Here, an optically excited electronic transition e.g. from the molecular ground state
(bonding orbital) to an excited antibonding orbital will lead to a rearrangement of
the equilibrium distance between the two atoms forming the bond. The very fast
electronic transition into an antibonding level causes a sudden repulsion of the atoms,
which excites a harmonic vibration with frequency around the new equilibrium
distance due to the inertia of the atomic nuclei (Fig. 6.6).
If many molecules can interact with each other in condensed matter (gases, liquids,
solids), the excitation of a given molecule will spread to all other molecules, until
thermal equilibrium among all possible vibrational (and rotational) modes is reached,
resulting in a corresponding increase in temperature.
There are four basic processes through which photons interact with condensed
matter:
3 (More details are provided in Chap. 8).

6.2 Solar Thermal Energy 93
Fig. 6.6 Photon absorption in atoms only increases electronic energy. In molecules or solids, part
of the electronic excitation gives rise to molecular vibrations via the electron-phonon coupling
1. Reflection, reflectivity R: This is a coherent process at an interface, depending

on the indices of refraction on both sides, the polarization of the light, and the
angle of incidence. Details are described by the general Fresnel formulae.
2. Scattering S: An incoherent process due to the interaction with atoms, molecules,
or small particles. Rayleigh scattering occurs when the wavelength of light λ is
much larger than the size of the scattering object L. It is isotropic and increases
with decreasing λ. Mie scattering occurs at small particles or interface roughness
for λ << L
3. Absorption, absorbance A: This is an incoherent process due to the transfor-
mation of photons to excited electrons and/or phonons. In homogeneous media
the Lambert-Beer law holds for the local light intensity (x) inside the medium
(α = α(ω) is the absorption coefficient):
(x) = (1 − R) (0) exp(−αx) (6.2.1)
4. Transmission T: Fraction of the incoming photon flux passing (transmitted)

through the medium.
Conservation of energy yields:
R+S+ A+T =1 (6.2.2)
For an optimum harvesting of solar thermal energy, all processes need to be opti-
mized. This is achieved by technically approaching the following idealized optical
elements:
• ideal absorbers (A = 1 for all wavelengths)

• selective absorbers (A = 1 in one spectral range, A = 0 for the rest)
• ideal mirrors (R = 1, e.g. for concentrating solar receivers)
• ideal anti-reflection coatings (R = 0, compare Chap. 8)
• scattering in the environment, on surfaces and in inhomogeneous media
94 6 Thermal Energy
Table 6.6 Global average values of atmospheric absorption, scattering and transmittance
γS AM A [%] S Rayleigh [%] S Mie [%] T [%]
90◦ 1.0 8.7 9.4 0–25 62–83
60◦ 1.15 9.2 10.5 1–30 57–81
30◦ 2.0 11.2 16.3 4–45 41–71
5◦ 11.5 19.5 42.5 25–86 6–35
Table 6.7 Seasonal variations in Berlin (ca. 52◦ N, varying γ S , 10 a average)

Date γS AM ˆ dir ect ˆ di f f use
[k W h m−2 d−1 ] [k W h m−2 d− ]
30.1. 19.5◦ 3 0.17 0.44
1.4. 42◦ 1.5 1.42 2.07
21.6. 61◦ 1.15 2.58 2.86
22.12. 14◦ 4 0.1 0.35
Table 6.8 Local variations of these values (yearly average) in Europe

Site Latitude (N) ˆ dir ect ˆ di f f use
[k W hm−2 d−1 ] [k W hm−2 d−1 ]
Bergen 60◦ 0.86 1.29
Vienna 48◦ 1.4 1.63
Lisbon 38◦ 3.06 1.66
In addition, condensed matter emits radiation also without incoming photons at finite
temperatures according to the Stefan-Boltzmann law, depending on the emissivity e
(cf. Sect. 3.1).
Important for solar thermal energy are first of all absorption and scattering of sun-
light in the atmosphere. This strongly depends on the season and angle of incidence
γ S of sunlight relative to the surface of the earth, as exemplified in Tables 6.6 to 6.8.
All values depend strongly on the so-called air mass (AM), describing the relative
length of travel of sunlight through the atmosphere for different angles of incidence:
1
AM = (6.2.3)
sin γ S
The large percentage of scattered light leads to diffuse radiation, which can not be
concentrated anymore. Average seasonal and local values for direct versus diffuse
sunlight energy flux densities per day are listed in Tables 6.7 and 6.8.
Because of the strong variation of γ S during the day, an active positioning of

solar collectors improves the received energy. In central Europe, the typical
gain in energy is:
• for two-axis solar trackers: ≈ 30% more energy

• for one-axis solar trackers: ≈ 20% more energyn
Currently, there are three main fields of applications of solar thermal energy:
1. The use of solar thermal elements in building architecture for the design of low-
energy houses, such as winter gardens, intelligent ventilation systems, active win-
dows, …
2. Non-concentrating (flat) solar collectors: these use a combination of a glass cover
(selective transmission: high transmission in the visible, low transmission in the
infrared spectral range) with a selective absorber (high absorbance in the visi-
ble spectral range, low emissivity in the infrared region). Since according to the
Kirchhoff law of radiation the absorbance A and the emissivity e at a given wave-
length λ are equal for directed radiation at thermal equilibrium, an ideal selective
absorber would have the spectral characteristics shown in Fig. 6.7. Commonly
used materials for such selective absorbers are “black chromium”, “black nickel”,
or TiNO X .
3. Concentrating solar receivers: These use mirrors to focus the direct sunlight onto
a thermal receiver. Depending on the degree of concentration, different receiver
temperatures can be achieved. This is then used to generate steam, heat up oil,
Fig.6.7 Spectral dependence of the absorbance A and the emissivity e of an ideal selective absorber.
The absorbance is high in the ultraviolet and visible region of the solar spectrum P(λ) schematically
shown in the lower panel. A and e drop to zero beyond a critical wavelength λc , depending on the
absorber temperature T . In this way the emission of thermal radiation from the absorber is minimized
96 6 Thermal Energy
drive a Stirling motor, melt phase change salts, drive thermochemical reactions,
etc.
The optical concentration or focusing of direct sunlight is limited by the finite
half opening angle of the Sun as seen from the Earth (cf. Sect. 3.1):
R
α = arcsin = 16 = 0.27◦ = 0.00465 rad (6.2.4)
rS E
For biaxial concentration e.g. with a circular parabolic mirror this results in a
maximum concentration factor of Cmax = 1/α 2 = 46211. For a uniaxial, lin-
√ concentrator (e.g. a parabolic trough) the physical limit is Cmax,lin = 1/α =
ear
Cmax = 215. The maximum working temperature TW of an insulated receiver
with emissivity e=1 is then obtained from the Stefan-Boltzman equation as:
1/4
C · SC
C · SC = σ TW4 ⇒ TW = (6.2.5)
σ
where SC is the solar constant. The maximum working temperature TW ,max is

reached for Cmax . The temperature in the focus of an ideal concentrator would
then approach the effective surface temperature of the Sun:
1/4
C
TW ,max = 5780 K = Te f f ,sun ; ⇒ TW = Te f f ,sun (6.2.6)
Cmax
Practical receiver temperatures of concentrating solar thermal systems are typi-

cally limited to < 1500 K, due to issues with long-term receiver material stability.
Some examples of concentrating solar thermal systems are listed below:
• The “Euro trough”: a standardized, linear parabolic collector (2001). Length =

150 m, width = 5.80 m, area = 825 m2 , C = 82
• The “Solar Electric Generation System” (SEGS) in the Mojave dessert, California
(since 1984). Area = 2.3 Mm2 , >1 Mio. parabolic troughs, 350 MW installed
electric power, oil at 300–400 ◦ C as the working medium, driving conventional
steam turbines. This power plant operates at an overall efficiency of η ≈ 30 −
38%.
• Solar towers in large heliostat fields are operated e.g. in Spain at Almeria and
Sevilla. C ≈ 600 − 1000, air or molten salt as the working medium, TW ≈ 700 −
1000◦ C, driving a steam turbine with 10 MW electrical power generation at an
overall efficiency of η ≈ 15%.
• An example of current research is shown in Fig. 6.8. It aims at the thermochemical
generation of bio-fuels via the production of syngas, a mixture of H2 and CO. To
this end, the reduced form of a metal oxide (MOred ) is used at a lower tempera-
ture level TC to reduce a mixture of water and CO2 , resulting in an oxygen-rich
metal oxide (MOox ) plus syngas in an overall exothermic reaction. MOox is then
Fig. 6.8 Schematic flow

diagram for solar syngas
production. See text above
for details
transported to a solar receiver with a working temperature TW , which is suffi-

ciently high to reduce MOox back to MOred by splitting off oxygen in an overall
endothermic reaction. Then, the cycle starts again.
Further Reading
• Hellisheiói Power Station, the largest geothermal power plant in Iceland (300MWe:
https://www.on.is/en/about-us/power-plants/)
• The world’s largest thermal spa in Erding is powered by the geothermal plant
Erding: https://www.steag.com/de/erding, https://www.geowaerme-erding.de/.
• https://www.statista.com/statistics/476281/global-capacity-of-geothermal-
energy/
• Tischner, T., Melchert, B., Ortiz, A., Schindler, M., Scheiber, J., Genter, A.:
Test- und Probebetrieb des HDR-Kraftwerks Soultz. Abschlussbericht zum BMU-
Vorhaben 0325159, 1–128 (2013).
• Summary of home use heat pumps: https://www.energy.gov/energysaver/heat-
pump-systems.
• A recent review on selective absorbers: Zhang, J., Wang, C., Shi, J., Wei, D., Zhao,
H., Ma, C.: Solar Selective Absorber for Emerging Sustainable Applications. Adv.
Ener. Sust. Res., 3(3), 2100195 (2022)
• Fortuin, S., Stryi-Hipp, G.: Solar Collectors, Non-concentrating. In: Richter, C.,
Lincot, D., Gueymard, C.A. (eds.) Solar Energy. Springer, New York, NY (2013).
• Solar tower Jülich: https://www.dlr.de/content/en/research-facilities/solar-tower-
juelich.html
• SEGS: https://en.wikipedia.org/wiki/Solar_Energy_Generating_Systems
• Solar towers: https://en.wikipedia.org/wiki/Ivanpah_Solar_Power_Facility
98 6 Thermal Energy
References
1. 2020 EGEC Geothermal Market Report, Garabetian, T., et al. (eds.): © EGEC - European
Geothermal Energy Council (2021)
2. Davies, J.H.: Global map of solid Earth surface heat flow. Geochem. Geophys. Geosyst. 14(10),
4608–4622 (2013)
3. Fuchs, S., Norden, B.: International Heat Flow Commission (2021): The
Global Heat Flow Database: Release 2021. GFZ Data Services (2021).
https://doi.org/10.5880/fidgeo.2021.014
4. Huttrer G.W.: Geothermal Power Generation in the World 2015-2020 Update Report, Proceed-
ings World Geothermal Congress 2020 Reykjavik (2021)
5. Analyzed data from the BP Statistical Review of World Energy on https://ourworldindata.
org/grapher/installed-geothermal-capacity?tab=chart&country=USA~ISL~TUR~OWID_
WRLourworldindata.org. Accessed 28 July 2022
6. Lund, J.W., Toth, A.N.: Direct Utilization of Geothermal Energy 2020 Worldwide Review. In:
Proceedings World Geothermal Congress 2020 Reykjavik (2021)
Photosynthesis
7
Abstract
In this chapter the different forms of biomass and their relative and absolute
energy content are discussed. The detailed biophysical mechanism of photosyn-
thesis is explained as the main energy conversion process in plants. Also, a brief
introduction to artificial photosynthesis is presented.
Photosynthesis is the natural transformation of atmospheric CO2 into biomass

through photo-catalytic water splitting and CO2 reduction powered by sunlight.
Although photosynthesis is mainly a biochemical process, it also has a significant
number of physical aspects which will be considered in this chapter. After all, the
products of photosynthesis in the form of biofuels and fossil fuels have been and
still continue to be the main source of primary energy for mankind. In addition to
N2 , CO2 was the main component of the Earth’s atmosphere during the first three
billion years of Earth history. Only about one billion years ago, the oxygen partial
pressure in the atmosphere for the first time exceeded its CO2 content because of the
increasing photosynthetic activity in the oceans and on the surface of the Earth. The
current atmospheric composition with about 20% O2 and less than 1000 ppm CO2
was only reached about 300 million years ago, enabling the development of higher
life forms with an oxygen-based metabolisms.
7.1 General Considerations of Biomass Usage
In view of renewable energy, the main outcome of photosynthesis is biomass. As

shown in Fig. 7.1, the amount of global carbon present as biomass is about 600
Gigatons (Gt). This is very small compared to carbon stored in the soil, the ocean
waters and the lithosphere. A similar amount of carbon is present in the atmosphere,
mainly in the form of CO2 . Moreover, about 10 000 Gt of carbon are stored as ancient
biomass in the Earth in the form of fossil fuels. Both, the surface of the sea and the land

100 7 Photosynthesis
Fig. 7.1 Global carbon inventories in the atmosphere, the oceans, land and the lithosphere. Black
arrows indicate annual carbon fluxes between these different reservoirs. In the center, the anthro-
pogenic contributions to the global carbon fluxes are shown
exchange about 100 Gt of carbon annually in both directions. The human footprint
in the annual carbon cycle amounts to about 8 Gt,1 mainly coming from the use of
fossil fuels and surface biomass. 4 Gt of these are absorbed back by the oceans and
the land, while the other 4 Gt are released into the atmosphere, gradually increasing
the amount of carbon there [1].
The transformation of CO2 and water via the use of sunlight into biomass by
plants, algae and bacteria is the most important basis for the energy presently used
by mankind:
• Fossil fuels (oil, gas, coal) have been created by photosynthesis and currently
provide ≈82% of the energy used [2].
• In developing countries ≈35% of all primary energy comes from surface biomass.
Its use occurs mainly via low-tech fireplaces with a very low efficiency of about
15%. In addition, a lot of atmospheric pollution is caused by the resulting smoke
[3].
• In Europe and the USA only around 3% of the primary energy come from surface
biomass [2,3].
1 Emission of pure C. This corresponds to 30 Gt of CO2 .

7.1 General Considerations of Biomass Usage 101
Table 7.1 Energy production by different photosynthesis systems. The corresponding values for
large scale Si solar cells are shown for comparison
System Biofuel production Energy content Efficiency
[kg/(ha · year)] [kWh/(ha · year)] [%]
Oil palm 4000 35000 0.17
Jatropha 2500 25000 0.12
Sugar cane 2500 16000 0.08
Micro algae 90000 1 · 106 4.5
Si solar cell – 3 · 106 14.0
As a typical example for the present use of surface biomass in central Europe
we mention the state of Bavaria in Germany, which is a highly developed
industrial region with a still considerable amount of agricultural area spanning
≈2 Mio. ha. Of these, about 10% are used for the production of biofuels in
the form of fire wood (40%), biomass for power plants (40%), liquid biofuels
(10%) and biogas (10%) [4].
A very problematic aspect of the use of surface biomass is the direct competition
between biomass used for biofuels and for food, as well as the strong usage of sweet
water.
The global production of biomass (per year) amounts to about:
• 1, 2 × 1011 t of dry biomass on land with a total area of 1 × 108 km2

• 6 × 1010 t of dry biomass in the oceans with a total area of 3, 6 × 108 km2
The average energy content of dry biomass is ≈20 MJ kg−1 . Thus, the global
energy potential of biomass amounts to about:
1.8 × 1014 kg · 2 × 107 MJ kg−1 = 3.6 × 1021 J
of total energy per year, which is actually 10 times the total energy consumption of
mankind. On the other hand, compared to the total solar irradiation of 4 × 1024 J per
year, this gives a global efficiency of photosynthesis of about 0.1%. As a matter of
fact, most of the incoming solar energy is used to guarantee the long-term survival
and progression of biomass. In comparison, as will be discussed below, the primary
efficiency of the photosynthesis reaction is ≈30%, which constitutes the long-term
aim of artificial photosynthesis discussed at the end of this Chapter.
As exemplified in Table 7.1, even for natural photosynthesis the overall efficiency
can vary a lot and is easily surpassed by technical use of sunlight e.g. in the form of
solar cells.
7.2 Biophysical Principles of Photosynthesis
In general, one distinguishes two different forms of photosynthesis, depending on

the underlying basic chemical reactions:
Oxygen-generating photosynthesis of most green plants according to the overall
reaction
ω
H2 O + CO2 −→(CH2 O) + O2
An-oxygeneous photosynthesis of bacteria using H2 S instead of H2 O, e.g. in the

vicinity of sulphur-containing vulcanic fumeroles in the oceans:
ω
H2 S + CO2 −→(CH2 O) + O + S
As an example, the overall reaction for photosynthetic production of glucose

is:
ω
12H2 O + 6CO2 −→C6 H12 O6 + 6O2 H = 2870 kJ/mol
The complete photosynthetic cycle is divided into a light reaction (primary reac-
tion) and a dark reaction (secondary reaction, Calvin cycle), as shown in Fig. 7.2:
Fig. 7.2 Illustration of the two components of the oxygen-generating photosynthetic cycle: in the
primary reaction, two water molecules are split into hydrogen and oxygen by the absorption of
eight photons in the red spectral range, resulting in the formation of molecular oxygen and the
intermediate energy carriers ATP and NADPH (see below). In the following secondary reaction in
the dark, the intermediate carriers drive the fixation of carbon from CO2 in (CH2 O), returning the
intermediate carriers into their low energy forms ADP and NADP
7.2 Biophysical Principles of Photosynthesis 103
For the fixation of one carbon atom two water molecules need to be split:
8ω
2H2 O −−→ O2 + 4H+ + 4e−
For the generation of one free electron, two photons of wavelength ≈700 nm (ω =
1, 8 eV, red) are necessary. The minimum energy to split water is ≈1,23 eV. Per
fixated carbon atom about 4 eV of chemical energy are stored. The resulting total
energy efficiency of the primary reaction is therefore:
4 eV
η= ≈30%
8 · 1.8 eV
Remarkably, this energy efficiency is close to the maximum thermodynamic

efficiency of solar cells, the so-called Shockley-Queisser limit. See Chap. 8 for
more details.
The protons and electrons produced by the primary reaction are used to generate
the intermediate energy storage molecules adenosine diphosphate (ADP)/adenosine
triphosphate (ATP) and nicotineamid adenine dinucleotid phosphate (NADP):
ADP + P ATP
NADP+ + 2e− + H+ NADPH
The necessary reaction takes place in the thylakoid membrane of chloroplasts

following the so-called Z-scheme, schematically shown in Fig. 7.3.
The photosynthetic process starts with absorption of red photons (wavelengths
680 and 700 nm) in the antenna complexes of the photosystems I and II (PS I and PS
II). These antenna complexes consist out of circular arrangements of several hundred
chlorophyll molecules. The main constituents of such planar chlorophyll molecules
are four pyrrole rings (C4 N rings with aromatic electron structure) arranged around a
central Mg atom. Optical absorption by one of the double bonds in such rings excites
the chlorophyll molecules from their ground state Chl to their excited state Chl*.
Details of this absorption process are discussed in the next section. The ground
state hole in PS II catalyzes the splitting of water into O and protons H+ . The
excited electron in Chl* is passed on rapidly to a chain of other molecules detailed
in Fig. 7.3 downward in energy to separate it spatially from the ground state hole,
avoiding direct recombination. Part of the freed electronic energy is used to generate
ATP from ADP and P in a process called electron transfer phosphorylization (ETP).
Finally, the excited electron from PS II ends up in PS I to neutralize the optically
excited hole there. The excited electron of PS I is again separated from the ground
state hole by a downward transfer process with the molecule ferredoxin (Fd) as an
important functional ingredient. Fd stores two electrons from two subsequent optical
104
7
Fig. 7.3 Optical absorption and electron transfers according to the Z-scheme of photosynthesis. See text for details
Photosynthesis
7.2 Biophysical Principles of Photosynthesis 105
excitation events of PS I, which then enable the synthesis of the intermediate energy
carrier NADPH from NADP+ and H+ . Again, further molecular details are discussed
in the next section.
The final fixation of carbon from CO2 in higher carbohydrates (the Calvin cycle)
is a very complicated sequence of chemical reactions and will not be discussed here
in any detail. It suffices to mention that for carbon fixation one distinguishes C3 –
and C4 –plants:
• C3 -plants bind ≤ 30 mg CO2 / 100 cm2 leaf area and hour

• C4 -plants bind 50–90 mg CO2 / 100 cm2 leaf area and hour
The overall efficiency of photosynthesis also depends on:
1. wavelength of light: chlorophyll a and b2 mainly absorb in the blue (400–480 nm)
and red (600–700 nm) spectral range, giving rise to the green colour of leaves.
For wavelengths λ > 690 nm the photosynthesis stops abruptly, known as the
“red drop” of photosynthesis. Using different pigments than chlorophyll also
other spectral ranges can be used (e.g. via carotinoids, bacteriorhodopsin).
2. Light intensity: photosynthesis starts at around 10 W/m2 with an optimum at
approximately 230 sW/m2 .
3. Temperature: a broad optimum is reached at 35 ◦ C
4. CO2 content: the optimum concentration is at ≈1vol% (i.e. much higher than the
400 ppm = 0.04% of CO2 currently in our atmosphere).
7.3 Basic Biomolecular Processes of Photosynthesis
In this section, we discuss some of the biomolecular processes mentioned above

in more detail. For successful photosynthesis, four distinct sub-reactions have to
be coordinated spatially, energetically and chemically: The first sub-reaction occurs
in photosystem I (PS I) and uses two photoexcited electrons to produce twofold
reduced ferredoxin (Fd2− ) by storing the electrons in the central Fe-S molecular
orbital of ferredoxin. This is shown schematically in Fig. 7.4. The two electrons are
fed into the ferredoxin via the phaeophytin a (Ph a) complex, which enables a fast
removal of the photoexcited electron from the reaction center of the photosystem:
2e− + Fdox −→ Fd2−

red
The two electrons stored in ferredoxin then enable the synthesis of NADPH:
+ +
Fd2−
red + NADP + H −→ Fdox + NADPH
2 Chl a/b: chlorophyll a/b.

Fig.7.4 Schematic molecular structure of ferredoxin. The area encircled by the dashed line indicates
the molecular Fe–S orbital capable of storing two additional electrons. Cys symbolizes cystein end
groups
This second step is catalyzed by the enzyme ferredoxin-NADP-reductase.

Figure 7.5 shows the relative energetic positions of the different molecules
involved in the synthesis of NADPH.
The energy scale used is the redox potential relative to the normal hydrogen
electrode (NHE)3 potential as the origin.
The second sub-reaction takes place at the photosystem II (PS II). After optical
excitation and transfer of the excited electron to Ph a, the remaining positively charged
reaction center P680+ acts as a very strong oxidant which, in combination with the
oxygen-evolving center (OEC) complex Mn-O, catalyses the formation of O2 via
splitting of two water molecules
Two excited electrons from PS II and the protons from the water-splitting reaction
are transferred to plastoquinon (PQ)), changing it from the oxidized form PQ into
the reduced and hydrogenated form PQH2 (Fig. 7.6):
ω, 2e−
PQ + 2H+ −−−−→ PQH2
This undergoes the following reaction:
Fig. 7.5 Energetic positions

of the different molecular
subunits related to the
NADPH synthesis in
photosystem I. Red/Ox
indicate the direction of
increasing reduction
(oxidation) potentials,
respectively
3 NHE: normal hydrogen electrode Pt/H2 : 2H+ + 2e− H2 at Evac -4,5eV.

7.3 Basic Biomolecular Processes of Photosynthesis 107
Fig. 7.6 Left: typical structure of the plastoquinon complex. Right: the quinon form of PQ is
changed to semiquinon by adding an electron and finally into the quinol form by adding a second
electron and two protons
Fig. 7.7 Time scales of the reaction pathway at P680. Following optical excitation, the excited
electron is transferred from P680* to Pha in typically 10ps. The subsequent transfer to PQ and the
modification to PQH by the reaction with a proton instead takes around 100 ps. Finally, a solvated
electron from the water-splitting reaction resets the P680 reaction center back into its ground state
As already mentioned, pheophytin4 (related to chlorophyll a, but without the Mg

center) is responsible for the fast and efficient electron transfer from the excited
state of the reaction center P680* of PS II, occuring spontaneously in typically 10 ps
(Fig. 7.7).
The third sub-reaction involves the molecular complex cytochrome bf (Cyt bf). It
produces a proton gradient through the thylakoid membrane (see below), powered
by the electron flux from PS II to PS I, liberating protons by the oxidation of PQH2 :
2e−
PQH2 + 2PCy(Cu2+ ) −−−−→ PQ + 2PCy(Cu+ ) + 2H+
A main functional component of cytochrome bf is the Plastocyanin Complex

(PCy), shown in the figure below. It contains a central copper atom, which can
change its oxidation state between 2+ and + by exchanging an electron with its
environment (Fig. 7.8).
Similar to photosystem I above, Fig. 7.9 below shows the energetic positions of
the different components active in photosystem II on a redox potential scale relative
to NHE:
The last sub-reaction produces ATP via the enzyme ATP-synthase, which catal-
yses the reaction ADP + P → ATP.
4 Short “Pheo”.
Fig. 7.8 Schematic structure of the plastocyanin complex in cytochrome bf. ‘Pyr’ symbolizes a
pyrrole ring (C4 N)
Fig. 7.9 Energy diagram of the different complexes involved in sub-reactions two and three. Mn-O
indicates the oxygen-evolving center where water splitting occurs via P680+
The enzyme is powered by a proton flux through it, caused by the proton con-
centration gradient inside vs. outside of the thylakoid membrane, which is generated
by Cyt bf. This leads to a typical pH-value of 4 inside the volume enclosed by the
thylakoid membrane (lumen) versus a pH-value of 7.5 outside (stroma). A summary
of all main components active within the thylakoid membrane, the main proton and
electron fluxes between these units and the two cyclic reactions involving plasto-
quinone and plastocyanin are summarized in Fig. 7.10.
To make this very complex photosynthetic machine run optimally, the spatial
arrangement of the different trans-membrane proteins is very important. Therefore,
the thylakoid membrane consists of (cf. Fig. 7.11):
• Stacked regions containing mainly PS II and Cyt bf. Here, a large part of the
generation of the H+ gradient takes place.
7.3 Basic Biomolecular Processes of Photosynthesis 109
Fig. 7.10 Main functional units of photosynthesis in the thylakoid membrane with the correspond-
ing electron and proton fluxes
Fig. 7.11 Schematic spatial arrangement of the four main photosynthetic proteins in stacked and
non-stacked regions of the thylakoid membrane
• Non-stacked regions with all four proteins, where the generation of NADPH and
ATP occurs.
The thylakoid membrane itself is contained in the cell membrane of the chloro-
plasts in the leaves of plants (Fig. 7.12).
Fig. 7.12 Schematic structure of chloroplast cells with the thylakoid membrane contained inside
7.4 Details of Photon Absorption and Energy Transfer

in the Light-Harvesting Complexes of Photosystems
Both photosystems contain an antenna protein (light-harvesting complex (LHC))

consisting of about 50 carotenoid molecules, 200 chlorophyll a and b molecules plus
a reaction center (RC). The chlorophyll molecules5 as the main optical absorption
sites are arranged in a torus around the reaction center, as shown schematically in
Fig. 7.13.
The optical absorption centers in chlorophyll are four pyrrole rings (Pyr, C4 N)
surrounding a central Mg atom in a planar structure. Absorption in the red spectral
region occurs in the double bonds of these rings as π π ∗ (bonding-antibonding)
transitions of the ppπ bonds (Fig. 7.14).
The absorption of visible light (red photons) occurs via π − π ∗ -(bonding-
antibonding) transitions of the ppπ bonds (Fig. 7.15):
The exact HOMO6 -LUMO7 -splitting depends on the specific ligands bound to
the pyrrole rings. The absorption coefficient of chlorophyll is roughly
1 × 105 cm−1 mol−1 . This means that at full sunlight about 10 photons are absorbed
per chlorophyll molecule per second. This excitation frequency is slow enough to
provide sufficient time for all subsequent reactions to proceed.
Normally, for these optical electric dipole transitions during the excitation of
chlorophyll molecules the electron spin S is preserved: the spin configuration (anti-
parallel) of the ground state singlet is preserved and leads to a singlet excited state,
whereas the total angular momentum L + S is increased by the angular momentum J
= 1 of the absorbed photon (singlet transitions S = 0, S = 0, L = 1). However,
as shown in Fig. 7.16, in a few cases one of the electron spins may flip due to spin-
orbit coupling (e.g. close to heavy metal atoms), causing a system crossing into the
triplet state T* (parallel spins, S = 1).
5 There are many variants of chlorophyll molecules varying by rest groups bound to the pyrrole
rings. The discovery of the chlorophyll structure led to three Nobel prizes: Wilstätter 1915, Fischer
1930, Woodward 1965.
6 Highest Occupied Molecular Orbital.
7 Lowest Unoccupied Molecular Orbital.
7.4 Details of Photon Absorption and Energy Transfer … 111
Fig. 7.13 Antenna protein arrangement of the light harvesting complex
Fig. 7.14 Left: one of the four pyrrole rings (with two C = C double bonds) linked to the central
Mg atom in a chlorophyll molecule. Different variants of chlorophyll have different rest groups
R attached to the rings. Right: electronic structure of a C = C double bond consisting of carbon
p-orbitals arranged in a strongly bound ppσ -chain and a weaker ppπ -bond perpendicular to the
chain + and − indicate the opposite phases of the two p-wavefunction lobes
Since molecular oxygen is always present close to the photosystem due to the
water splitting reaction, the triplet state of chlorophyll (S = 1, ms = +1, 0, –1) can
then couple to O2 -molecules by resonant energy and spin exchange (Dexter transfer).
This causes the formation of very reactive and dangerous singlet oxygen molecules.
Singlet oxygen is one of the most reactive oxidation agents known in chemistry and
would cause the destruction of the photosystem. To prevent this, the photosystem
also contains a number of carotenoid molecules, acting as anti-oxidants to get rid of
the dangerous T* or 1 O2 states.
Fig. 7.15 Illustration of a π − π ∗ -transition in a C = C carbon double bond. The weaker π -

bond forms the inner HOMO-LUMO level pair and the optical transition changes the π -orbital
configuration from bonding to anti-bonding. + and − indicate the different phases of the p-wave
functions
Fig. 7.16 Illustration of a system crossing from the excited singlet stateS1 into the excited triplet
state T* due to spin-orbit coupling. In the presence of O2 molecules, this enables the occurrence of
a resonant Dexter transfer (both energy and spin are exchanged) between chlorophyll and O2 via
the excitation of the oxygen molecule from its triplet ground state to an excited singlet state
To increase the absorption of sunlight in the photosystems, about 200 chlorophyll

molecules are arranged in the antenna complexes. In these complexes energy can be
transported very rapidly (10ps) via radiationless dipole-dipole interaction between
neighbouring chlorophyll molecules. This is called Förster transfer. The correspond-
ing transfer rate is related to the spontaneous fluorescence rate of the molecules via:
6
R0
ktrans f er = k f luor escence (7.4.1)
r
Here r is the distance between neighboring molecules and R06 = 8.8 × 10−5 Jκ 2 n −4
(n: index of refraction). J is the so-called energy-overlap factor, κ the orientation
factor:

J ≡ α(λ)FD (λ)λ4 dλ
7.4 Details of Photon Absorption and Energy Transfer … 113
Fig. 7.17 Schematic

illustration of the typical
Stokes-shift between
absorption and emission
spectra in molecules. The
shaded area contributes to
the energy-overlap factor J
Fig. 7.18 Orientation factor

κ, determined by the relative
angles of the two interacting
dipole moments
Fig. 7.19 Orientation-

dependent spectral shift of
molecular dimers coupled by
dipolar interaction.
Compared to the monomers,
a red shift occurs for parallel
and a blue shift for
anti-parallel dipole
orientations
where α(λ) is the absorption coefficient spectrum of the receiving molecule, while
FD (λ) is the fluorescence emission spectrum of the emitting molecule. Also, the
relative orientation of interacting dipoles influences the value of R0 via the orientation
factor (Figs. 7.17, 7.18).
Typical values for R0 in antenna complexes are 6–10 nm. The relative orientation
of the dipole moments also leads to an energy shift of the interacting molecules
(Fig. 7.19):
The efficient Förster transfer between the chlorophyll molecules in an antenna
complex allows all molecular excitations to finally be collected by a central reaction
center with a lower HOMO-LUMO-splitting (Fig. 7.20).
Fig. 7.20 Fast Förster transfer between chlorophyll molecules in an antenna complex and final
capture of the excitation in a reaction center (RC) with smaller HOMO-LUMO splitting
7.5 Technical Use of Biomass
Biomass usually comes from different sources in different forms. Therefore, the
typical energy content of the starting material (dry mass) can vary a lot (Table 7.2):
There are also different forms in which the energy can be extracted from biomass:
• physically: direct combustion in ovens or heat engines

• chemically: gasification, pyrolysis, carbonisation
• biologically: fermentation, fowling, feedstock for animals
A detailed discussion of the use of biomass would go beyond the scope of this book.
Therefore we just briefly mention some typical examples in the box below:
Table 7.2 Energy content in GJ per ton of the dry mass of various biomass sources
Component Energy content Resources
Starch, sugar ≈15 GJt −1 Corn, sugar cane, sugar beets
Cellulose ≈5 − 15 GJt −1 Grass, wood, …
Plant oils ≈40 GJt −1 Soy, sun flowers, palms
Algae ≈35 GJt −1
7.5 Technical Use of Biomass 115
Fig. 7.21 Overview of fuel production by different processing methods using different biomass
feedstocks
Typical uses of biomass:
• Direct combustion: here the efficiency of extraction of energy varies widely

between 5% (open fireplaces) and 50% (bio-power-plant with waste heat
usage)
• Gasification: this is the reaction of biomass with water steam and air/O2
at temperatures between 300–1000 ◦ C at pressure of 1–30 bar. The main
product is syngas, a mixture of CO and H2 , from which most carbohydrates
can be synthesised.
• Pyrolysis: the processing of biomass without air/C O at elevated tempera-
tures of 300–500 ◦ C until all volatile components have disappeared. The
main product is charcoal which can be burnt to CO2 to almost 100%. The
energy density of charcoal is twice that of the starting material.
• Bio-ethanol by fermentation: this produces normal alcohol C2 H5 OH, made
out of sugar or starch, followed by distillation. The achievable output is
about 200L per ton biomass.
The following Table gives an impression of current volumes of bio-ethanol
produced in different regions:
Country Production volume

USA 40 · 109 L
Brazil 30 · 109 L
EU 5 · 109 L
GER 1 · 109 L
This amounts to about 1% of the worldwide oil production of 6000 GL

(Fig. 7.21).
7.6 Artificial Photosynthesis
The term “artificial photosynthesis” (APS) describes the scientific attempts to mimic
the photosynthesis mechanisms of plants and bacteria in a technical context, using
the light of the sun directly to produce solar fuels. The simplest version of artificial
photosynthesis is solar water splitting for the production of hydrogen, first realized
50 years ago by Fujishima and Honda [5]. This mimics the water splitting occurring
at photosystem II described in Sect. 7.2. As the initial step, one uses the absorption
of sun light in a semiconductor or dye molecules to excite electron-hole pairs. The
excited holes are then used either directly or with the help of suitable catalysts for
water oxidation:
2H2 O + 4h+ −→ O2 + 4H+
This is shown on the left side of Fig. 7.22 and constitutes an anodic oxidation
reaction consuming the holes in the anode. The remaining excited electrons are
transported via an electrical connection to the nearby cathode, where they can be
used to reduce the protons generated at the anode to hydrogen:
Fig. 7.22 Electrode arrangement for water splitting and CO2 reduction for artificial photosynthesis
using solar irradiation. See text for details. Alternatively, an external electrical power source can be
used to drive the corresponding reactions electrochemically
7.6 Artificial Photosynthesis 117
4H+ + 4e− −→ 2H2
This completes the water splitting reaction. A semi-permeable membrane between

the electrodes can be used to spatially separate the evolution of O2 and H2 . Using
optimized photo-anodes with the right band gap and position of the hole energy level,
efficiencies up to 14% have been achieved under realistic solar irradiation conditions
so far.
A similar electrode set-up is also used in modern electrolyzers, where an external
voltage U drives the electron current, using the electrical power e.g. generated by
wind mills or solar cells for water electrolysis. In this case, the anode does not
need to be photo-sensitive and can just be a metal plate like the cathode. Such
electrolyzers have reached efficiencies for electrical water splitting around 70%.
Combined with a solar cell of 20% efficiency, this also gives an efficiency of 14%
for overall water splitting, with practical advantages such as better long-term stability
and larger operational flexibility for the combined solar cell/electrolyzer approach.
This is one of the reasons why cathodic reactions other than proton reduction have
gained significant interest for artificial photosynthesis in the last years. One promi-
nent example is the combination of anodic water oxidation with cathodic reduction
of CO2 , constituting a simplified version of the Calvin cycle in Fig. 7.2. For example,
CO2 can be reduced to formic acid by the reaction:
CO2 + H+ + 2e− −→ HCOO−
The formic acid then can be further processed to other molecules for long term energy
storage. Another possibility is the reduction of CO2 to CO:
2CO2 + 4H+ + 4e− −→ 2CO + 2H2 O
Together with solar hydrogen, this allows the production of syngas (H2 + CO) as
a starting point for other solar fuels such as methanol:
CO + 2H2 −→ CH3 OH
Further Reading
• CO2 -levels monitored by NOAA’s Global Monitoring Lab: www.climate.gov
• Data on worldwide use of bioenergy: www.iea.org/data-and-statistics/
• Hou, H.J.M., Najafpour, M.M., Moore, G.F., & Allakhverdiev, S.I. (eds.): Pho-
tosynthesis: Structures, Mechanisms, and Applications. Springer International
Publishing (2017).
• Kan, T., Strezov V., Evans, T.J.: Lignocellulosic biomass pyrolysis: a review of
product properties and effects of pyrolysis parameters. Renew. Sust. Energ. Rev.
57, 1126–1140 (2016).
• Sansaniwal, S.K., et al.: Recent advances in the development of biomass gasifica-

tion technology: a comprehensive review. Renew. Sust. Energ. Rev. 72, 363–384
(2017).
• Balat, M., Balat, H., Öz, C.: Progress in Bioethanol Processing. Prog. Energy
Combust. Sci. 34, 551–573 (2008).
• Zhang, B., Sun, L.: Artificial photosynthesis: opportunities and challenges of
molecular catalysts. Chem. Soc. Rev. 48, 2216–2264 (2019).
References
1. Green, C., Byrne, K. A.: Biomass: Impact on Carbon Cycle and Greenhouse Gas Emissions. In:
C. J. B. T.-E. of E. Cleveland (ed.), pp. 223–236 (2004)
2. BP.: Statistical review of world energy 2021 70 (2021)
3. IEA.: World Energy Balances. https://www.iea.org/data-and-statistics/. Accessed 30 July 2022
4. Energieatlas Bayern, Bayerisches Staatsministerium für Wirtschaft, Landesentwicklung und
Energie, https://www.energieatlas.bayern.de/thema_biomasse/daten.html. Accessed 30.07.2022
5. Fujishima, A., Honda, K.: Electrochemical photolysis of water at a semiconductor electrode.
Nature 238, 37–38 (1972)
Photovoltaics
8
Abstract
This chapter deals with the physics of solar cells. The basic photonic processes
such as adsorption, reflection, emission and the charge carrier processes separa-
tion, thermalisation, recombination and extraction are explained with the corre-
sponding theory and examples. The next subchapter presents the various types of
solar cells and their differences in generating electricity. Later, the current-voltage
characteristics are explained in a view to the layout of solar cells and their max-
imum efficiency. The chapter concludes with the main loss mechanisms in solar
cells and the different routes to improve efficiency.
8.1 General Considerations
Photovoltaics is the direct transformation of light into electricity. It is a purely opto-

electronic process in semiconductors without any chemical reactions, mechanical
movements or use of heat involved. Together with wind power, photovoltaics is
considered the major form of renewable energy production for a sustainable future.
Accordingly, the worldwide installed photovoltaic power has increased from about
30 GWp 1 in 2010 to more than 840 GWp in 2021, with currently more than 100 Wp
additional power capacity newly installed every year [1]. At the same time, due to
mass production, the cost per k Wp installed photovoltaic power has decreased to
about 1000e in 2020, enabling electricity production at a cost of currently about 4
ct/kWh.
As already discussed in Chap. 6, the amount of harvestable solar energy depends
strongly on the specific geographic location. As shown in Fig. 8.1, already in Europe
the available average annual solar energy density can vary by a factor of three between
Scandinavia and Mediterranean regions.
1 Here Wp means Watt peak, i.e. the maximum power generated under ideal standard illumination
conditions: AM 1.5, 1000 W m−2 incident irradiation intensity, cell temperature 25 ◦ C.

120
8
Fig. 8.1 Average annual sum of solar power (in [kW h m−2 ]) available in Europe and Northern Africa. Solargis is gratefully acknowledged for the permission
Photovoltaics
for reproduction (©2020 The World Bank, Source Global Solar Atlas 2.0, Solar resource data: Solargis)
8.2 Basic Processes in Photovoltaics 121
Fig. 8.2 Feynman diagram of photovoltaic energy conversion. Dots indicate interaction vortices
where the indicated physical processes need to be optimized
To make photovoltaics profitable also in regions with solar energy densities around
1000 kW h m−2 , high conversion efficiencies of solar cells on an industrial scale are
necessary. The complexity of this task can be understood by looking at the Feynman
diagram of photovoltaic energy conversion in Fig. 8.2.
The optimisation of photovoltaic devices requires the optimisation of many mate-
rials and device parameters. Absorption of solar photons competes with reflection and
transmission, which have to be minimized. The energy stored in photoexcited elec-
trons and holes is limited by thermalization losses (emission of phonons) depending
on the mismatch between photon energies and the band gap of the chosen semicon-
ductor absorber layers. This stored energy can be lost again by recombination of
the excited carriers via emission of phonons and photons. This has to be prevented
by an efficient spatial separation of electrons and holes, similar to the first step in
photosynthesis. Finally, electrons and holes have to be extracted via suitable con-
tacts, minimizing electrical losses in a solar cell. In the following, we will discuss
challenges and optimization strategies for efficient solar cells in more detail.
8.2 Basic Processes in Photovoltaics
8.2.1 Photons
Photons are characterized by two properties:
energy h f = ω
where f and ω are the frequency and angular frequency.
momentum p = k
ω 2π f
λ = c = c (⇒ λ = c/ f ) and c, the
with k the wavevector (k = 2π
velocity of light.
122 8 Photovoltaics
Fig. 8.3 Circular polarisation of the EM-wave corresponding to a photon. Classically, the circular
polarization can be related to the circulating electrical field E between an electron on its orbit around
the positively charged nucleus. Non-classically, the orbital momentum of the circling electron is
quantized in terms of
In addition, photons are bosons with angular momentum J = 1, realized by circular

polarization of the corresponding electromagnetic wave (Fig. 8.3).
This leads to three polarisation states:
⎧
⎪
⎨|m j = +1 (↑)
⎪ right circular polarisation
J = 1 |m j = 0 √ ((↑) + (↓)) linear polarisation
1
(8.2.1)
⎪
⎪ 2
⎩|m = −1 (↓) left circular polarisation
j
Summary of basic properties of photons as planar electromagnetic waves:
E = E0 exp [i(kr − ωt)] ; B = B0 exp [i(kr − ωt)]

E0 ⊥ B0 ; E0 , B0 ⊥ k; E and B in phase; E0 = cB0
1 1 c0 √
c0 = √ ; c= √ = in matter (n = ε for μ = 1)
ε0 μ0 εε0 μμ0 n
with n the refractive index.
8.2.2 Photon Density of States (DOS)
Photons confined to a volume V = L x L y L z have a quantized momentum p:

i j k
p = ( p x , p y , pz ) = , , ; i, j, k = 0, ±1, ±2, . . .
Lx L y Lz
This follows from the Heisenberg uncertainty principle for position and momentum:
x px ≥ h analogous for y and z (8.2.2)
Therefore, the minimal phase space volume of one quantized state is:
xyz px p y pz ≥ h 3
Due to the linear photon dispersion relation ω = cp, this leads to the following
relations:
• The number Nγ (ω) of photons with energy ≤ ω (with a factor of two originating
from the polarisation) is given by:
3 |p|
4π 3
8π (ω)3
Nγ (ω) = 2V = V (8.2.3)
h3 3 (hc)3
• This results in a photon density of states (DOS):
1 dNγ (ω) (ω)2

Dγ (ω) ≡ = 2 3 3 (8.2.4)
V dω π c
• Therefore, in thermal equilibrium the spectral distribution of photons (i.e. the
spectral density dn γ (ω) of photons with energy between ω and ω + dω) is:
dn γ (ω) = Dγ (ω) f γ (ω)dω (8.2.5)
with

−1
ω
f γ (ω) = exp −1 (8.2.6)
kT
where (8.2.6) is the equilibrium occupation function for bosons.

For an isotropic distribution of propagation directions (e.g. photons in a hollow
metal sphere as a black body), the density of states into the solid angle d is:
1
Dγ , (ω) = Dγ (ω) (8.2.7)
4π
This finally gives for the photon energy per unit volume and energy interval dω
into the solid angle d (≡photon spectrum):
deγ (ω) = Dγ (ω) f γ (ω)ω dω d (8.2.8)

−1
(ω)3 ω
= exp − 1 dω d (8.2.9)
4π 3 3 (c0 /n)3 kT
124 8 Photovoltaics
Note that the photon density of states in matter is proportional to n 3 . When

going from a medium n 1 to a medium with n 2 = n 1 and keeping the distribution
function f γ (ω) constant, this requires that at the interface a part of the photons
has to be reflected
8.2.3 Absorption, Reflection, Emission
Solar cells are generally built with diamagnetic2 semiconductors as the absorbers.
This means that the interaction with photons mainly occurs via their electric field
E. The principal process is acceleration of electrons by the electric field. Thereby
an electron will take up an energy E e = |E|e · x, where x is the spatial length of
undisturbed acceleration before a scattering event takes place. Thus, the interaction
strength is given by the dipole moment e · x of electrons in the absorbing medium.
Similarly, the quantum mechanical matrix element for optical absorption contains
the dipole operator, and the transition probability W for transitions from the initial
ground state |i to the excited state | f is proportional to the light intensity I :
W ∝ |Ei|e · x| f |2 = E2 e2 |i|x| f |2 ∝ I
In general, the intensity I of an electromagnetic wave is given by:

1 1 1 2 ε0 2
I = |E × B| = E0 B0 = E = E (8.2.10)
μ0 μ0 μ0 c 0 μ0 0
or, in the photon particle model
I = ω (8.2.11)
with the photon flux density (number of photons per area and time) .
Absorption of a photon is subject to three selection rules:
1. energy conservation: E f − E i = ω
2. momentum conservation: p f − pi = k ≈ 0 (due to the very small momentum
of massless photons compared to electrons)
3. angular momentum conservation: L f − L i =
2 Relative magnetic permeability μ = 1.

• In atoms, angular momentum conservation leads to the dipole selection rule:

n = ±1 for the main quantum number n.
• In solids (esp. semiconductors) angular momentum is conserved preferably
by generation of excitons (bound electron-hole pairs).
The optical properties of solids are parametrized

√ by the complex dielectric func-
tion ε̃(ω) or the complex index of refraction ñ = ε̃. Traditionally, these are written
in the form:
ε̃ = ε1 + iε2 and ñ = n + iκ (8.2.12)
where n is the index of refraction and κ the absorption constant. For electromagnetic
waves in matter the complex index of refraction then leads to the following changes
compared to vacuum (k = wavenumber):
c0 ω ω
c0 → ⇒k= → k = ñ
ñ c0 c0
This gives for the electric field of a plane wave:

ω
E = E0 exp i (n + iκ)x − ωt (8.2.13)
c0

ω ω
= E0 exp − κ x exp i nx − ωt (8.2.14)
c0 c0
so that the intensity I ∝ |E|2 follows the law of Lambert–Beer:

ω
I (x) = I0 exp[−2 κ x] ≡ I0 exp[−αx] (8.2.15)
c0
with α ≡ 2 cω0 κ the absorption coefficient.
Note that according to this law 95% of the incoming intensity are absorbed
within a distance of x = 3/α.
Typical values for the 95% absorption depth of photons with an energy of ω =
2 eV (red-yellow) in some semiconductor absorber materials commonly used for
solar cells are listed in Table 8.1.
The relevant absorption coefficients depend strongly on the nature of possible
optical transitions close to the optical band gap of the different absorber materials.
As shown in Fig. 8.4, one distinguishes between direct and indirect band gaps for
126 8 Photovoltaics
Table 8.1 95% absorption depths of photons with ω = 2 eV for various semiconductor absorber
materials
Material α (2 eV) x (95%)
[cm-1 ] [µm]
GaAs 3 · 104 1
Crystalline Si 3 × 103 10
Amorphous Si 2 × 104 1.5
Perovskite (PbI) and organic (PCBM/P3HT) 5 × 104 0.6
Fig. 8.4 Possible optical transitions in the E(k) band diagrams close to the optical band gap of
direct and indirect crystalline semiconductors as well as disordered “quasi direct” semiconductors.
CB and VB denote the lowest conduction and the highest valence band, respectively. ω stands for
a photon transition, for a phonon transition. See text for further details
crystalline semiconductors and “quasi-direct” band gaps in disordered (amorphous)

semiconductors.
In a direct semiconductor, the maximum of the upper valence band and the mini-
mum of the lowest conduction band occur at the same wavevector, typically at k = 0
in the center of the Brillouin zone. Important examples of direct semiconductors are
GaAlAs, GaInN, ZnO and perovskites. In these materials, direct optical transitions
can occur between discrete states of the valence and conduction band (VB, CB) with
the same k-value, as indicated by black arrows in Fig. 8.4. Note that compared to
electron and hole wavevectors, the wavevector of the massless photons is negligible,
so that optical transitions occur vertically up in the E(k) band diagram. Also the
electronic states only exist within the discrete bands, so that transitions such as the
one indicated by the left-most arrow in Fig. 8.4 are not allowed.
In indirect semiconductors, the two band extrema occur at different k-values, so
that close to the band gap no direct optical transitions are possible. Instead, for k-
conservation, at least one phonon needs to be involved in the transition to provide the
difference in momentum between the initial electron at the top of the valence band
and the excited electron at the bottom of the conduction band. The phonon partici-
pation can either occur as phonon emission starting from a higher photon energy or
as phonon absorption starting from a lower photon energy. Whereas phonon emis-
sion can occur at all temperatures, phonon absorption only contributes at higher
temperatures, when sufficient phonons are thermally excited. In either case the
need of additional phonons lowers the probability of optical transitions, so that the
absorption coefficients of indirect semiconductors close to the band gap are gen-
erally smaller than those of direct semiconductors. Important examples of indirect
semiconductors are Si, Ge, and SiC.
In non-crystalline, disordered semiconductors such as amorphous silicon or most
organic semiconductors, the wavevector k is no longer a valid quantum number due
to the lacking crystalline periodicity of the atom positions. Instead, the electronic
bands E(k) are projected onto the energy axis, giving rise to the electronic density of
states D(E), indicated by the hashed regions in Fig. 8.4. Since k-conservation is no
longer required, optical transitions occur again vertically up in energy. This is why
these transitions are called “quasi-direct”.
Reflection of incident photons at the absorber surface occurs due to the change of
the complex index of refraction, ñ = n + ik. In the following, we summarize the
typical behavior of the reflectivity R, the absorbance A and the transmission T for
the case of a solar cell with normal incidence of light from air (n 1 = 1, κ1 = 0) into
an absorber (n 2 , κ2 ) without multiple internal reflection due to a sufficiently strong
absorption in the absorber layer with thickness d. We then have:
(n 2 − 1)2 + κ22
R= (8.2.16)
(n 2 + 1)2 + κ22
A = (1 − R) (1 − exp [−αd]) (8.2.17)
T = (1 − R) exp [−αd] (8.2.18)
Thus, an incoming photon flux 0 generates a volume density of excited states per
unit time of:
1
G = (1 − R) (1 − exp[−αd]) 0 (8.2.19)
d
This is called the average generation rate with units cm−3 s−1 . Whereas the
absorber thickness d has to be optimized with respect to the absorption coefficient
α, the negative influence of the reflectivity R on the generation rate G has to be
minimized by dedicated anti-reflection layers on top of the absorber. To this end,
different strategies can be used, as shown for the specific example of crystalline Si
solar cells in Figs. 8.5, 8.6 and 8.7.
1. Macroscopic surface texturing with feature sizes larger than the wavelength of
light by anisotropic etching (e.g. for c-Si, ZnO, cf. Fig. 8.5)
2. Nano-texturing with feature sizes << λ (cf. Fig. 8.6)
3. λ/4-anti-reflection coatings (e.g. Si3 N4 on Si, cf. Fig. 8.7)
Some experimental results for Si wafers textured in these different ways to sup-
press their reflectivity are shown in Fig. 8.8.
128 8 Photovoltaics
Fig. 8.5 Macroscopic anisotropic etching of a smooth Si surface. In this specific example, a crys-
talline Si (100) surface is etched by KOH, which preferentially leads to exposure of pyramids with
(111) orientation of the crystal surface. The wavelength λ of incoming light is much smaller than
the obtained feature sizes, so that the anti-reflection properties result from multiple non-normal
reflections within the surface texture
Fig. 8.6 Nano-textured surfaces with feature sizes much smaller than the optical wavelength λ.
The resulting intermediate layer forms an effective medium exhibiting a gradual transition from the
outside (index n 1 ) to the inside (n 2 ) with no jump of the refractive index, thus avoiding reflection
The thermal emission of an absorber is obtained by integrating over the photon

energy density deγ (ω) in a given volume (cf. (8.2.8)) and substituting the integrand:
ω
x=
kB T
where T is the absolute temperature and k B the Boltzmann constant. This results in:
∞
(k B T )4 x3 π 2 k 4B
eγ = c0 3 dx ·4π = c0 3 T
4
(8.2.20)
4π n 0
3 3 e x − 1 3
15 n

π 4 /15
Inserting this into the energy flux density je = 1/4 ceγ yields the Stefan-Boltzmann-
law for the emission of a black body already used in Chaps. 3 and 6. In addition,
optically excited semiconductors emit so-called luminescence by radiative recom-
bination of excited carriers in the form of characteristic luminescence spectra (cf.
Fig. 8.12).
Fig. 8.7 λ/4-anti-reflection coating formed by a thin film with refractive index n and optical thick-
ness d = λ/4n between the outer layer (n 1 ) and the inner layer (n 2 ). Due to their optical path length
difference x = λ/2, the two waves reflected from the two interfaces interfere destructively, thus
canceling the reflectivity for the specific wavelength λ
Fig. 8.8 Anti-reflection coatings with different morphologies (upper row, feature size δ << λ),
resulting in different profiles of the refractive index (middle row). Experimental results for the
reflectivity in the wavelength range between 300 and 800 nm (lower row). Left column: creation of
a λ/4 layer by partial removal of the silicon via nanopore etching. Right column: nano-texturing of
a surface layer with smooth transition of the refractive index (“black etching”) [6]
130 8 Photovoltaics
8.2.4 Thermalization
After absorption of a photon with an energy above the band gap (for semiconductors)
or the HOMO-LUMO gap (for molecules), the absorbing material is in a short-lived
highly excited state, from which it relaxes very quickly (≈10−12 s) to a state of zero
kinetic energy, with just the band gap energy remaining as potential energy. The
excess kinetic energy is lost in the form of molecular vibrational energy or lattice
phonons. This energy loss cannot be recovered and, therefore, decreases drastically
the efficiency of solar cells due to the unavoidable “thermalization losses”. In the
following, we will briefly discuss some salient features of the thermalization process.
1. Thermalization in molecules
Thermalization of excess kinetic energy in molecules can be described in the so-
called energy-configuration diagram, E(Q), as in Fig. 8.9. Here, the molecular
energy E is plotted as a function of a multidimensional configuration coordinate Q
describing the equilibrium atomic positions. For a bi-atomic molecule, Q would
just be the distance between the two atoms. Close to the initial state |i, the energy
has a minimum, so that E(Q) can be approximated by a parabola. Molecular
vibrations around the stable initial position can be accounted for by an equidistant
ladder of a harmonic oscillator added to the electronic ground state. In a similar
way, the metastable excited state | f after an optical transition can be visualized
by a second parabola, albeit shifted up in energy and with a different equilibrium
configuration coordinate and a different spacing of harmonic vibrational sub-
levels. According to the Franck–Condon principle, optically induced electronic
transitions between the states |i and | f occur on a timescale much faster than
possible changes of the molecular configuration Q, so they can be described by
vertical arrows up or down in the E(Q) diagram:
Therefore, starting from the ground |i, an optical excitation will result in the
occupation of an electronically and vibrationally excited level of state | f . If
the excited molecule is in contact with a phonon bath (e.g. other molecules or a
substrate), a first thermalization chain will occur by phonon transitions down the
harmonic oscillator ladder of state | f into the equilibrium excited state. From
there, for example a second optical recombination event may occur by emission of
a luminescence photon,3 ending up in a vibrationally excited level of the ground
state |i. From there, a second thermalization chain of phonon emissions will
finally bring the molecule back to its ground state.
Both thermalization chains give rise to the so-called Stokes-Shift between the exci-
tation energy and the observed luminescence spectra, usually containing several
vibrational side bands or phonon replica:
ωex − ωlum = N (8.2.21)
3 Also called fluorescence or phosphorescence, depending on the lifetime of the excited state.
Fig. 8.9 Energy-configuration diagram of optical transitions and thermalization in a molecule. (|i:
initial state, ex: excitation, | f : final state, lum: luminescence). See text for further details
2. Thermalization in semiconductors
In the case of semiconductors as absorber materials, thermalization is best dis-
cussed in the context of the usual E(k) band diagram, where k is the wavevector
of electrons and holes in the conduction band and valence bands, respectively
(Fig. 8.10). For most inorganic semiconductors used in photovoltaics, the elec-
tronic band structure is as qualitatively shown in Fig. 8.10. The conduction band
(CB) consists of linear combinations of atomic s-orbitals, forming strong ssσ -
bonds.4 In the valence band (VB), two possible combinations of atomic p-orbitals
are present: weak ppπ -bonds between p-orbitals perpendicular to the bond axis
and strong ppσ -bonds between p-orbitals parallel to the bond axis, as already dis-
cussed for C=C double bonds in Chap. 7. Electronic bands resulting from stronger
bonds have a larger bonding-antibonding splitting and, therefore, a stronger cur-
vature of E(k) at the band extrema. This leads to a corresponding difference in
the effective masses m ∗ of electrons and holes, defined by:
−1
∂ E2
m ∗ = 2 (8.2.22)
∂k 2
As a consequence, holes in the top-most valence of direct semiconductors have
a much higher effective mass than electrons in the lowest conduction band. As
can be seen in Fig. 8.10, this gives rise to the fact that, following a direct optical
excitation, electrons in the conduction band have much more kinetic energy than
holes in the valence band. As a consequence, thermalization occurs mainly by
the relaxation of “hot” electrons back to k = 0 via a phonon emission cascade
along the E(k) band. In comparison, much less kinetic energy is lost by the
thermalization of “warm” holes in the heavy hole valence band.
The most efficient phonon emission from hot electrons in compound semicon-
ductors such as GaAs occurs via longitudinal optical (LO) phonons close to the
4 xxσ - and xxπ - bonds denote whether the participating x-orbitals of the two bonding partners have
rotational symmetry around or perpendicular to the bond axis, respectively.

132 8 Photovoltaics
Fig. 8.10 Thermalization of optically excited electrons and holes in the lowest conduction band
(CB) and the highest valence bands (VB), respectively. See text for details
Fig. 8.11 Atomic positions and resulting dipole moments pi per unit cell in a partially ionic
compound semiconductor: at rest (top row) and with the displacement caused by a long wavelength
(k ≈ 0) longitudinal optical phonon (bottom row)
center of the Brillouin zone at k ≈ 0, i.e. very long phonon wavelengths (the
“Fröhlich” interaction). As shown in Fig. 8.11, the alternating positively and
negatively charged atoms along a lattice direction give rise to nearest-neighbor
dipole moments pi with alternating directions. For an unperturbed atomic lattice
with all atoms at rest, the macroscopic sum over these dipole moments is zero. In
the presence of a long wavelength LO-phonon, however, the relative displacement
of positively and negatively charged atoms will add up to a finite macroscopic
polarisation, P = (1/V )i pi . This macroscopic polarization P is equivalent to a
macroscopic electric field E, which changes direction according to the vibrational
frequency of the LO-phonon and, thus, scatters charge carriers very effectively.
8.2.5 Recombination
After generation with rate G and before extraction, electrons and holes can recombine
with the recombination rate R according to the overall rate equation:
dn dp
= =G−R (8.2.23)
dt dt
Here, n(t) = n 0 + n(t) is the is the total electron density, consisting of the equi-
librium density n 0 due to doping or thermal excitation, and the time-varying opti-
cally excited carrier density n(t). Similarly, the total hole density is given by
p(t) = p0 + p(t).
Note that for the absorber layers of many solar cells one has a stronger p-doping
with p0 > n 0 , so that holes are the majority carriers, whereas electrons are the
minority carriers. This is chosen, because electrons in most cases have a smaller
effective mass and, thus, a higher mobility μ and a larger diffusion constant D than
holes:
vdrift eτ kT
μ= = ∗; D= μ (8.2.24)
E m e
Here, τ is the scattering time limiting the mobility. With a higher mobility, electrons
can, thus, drift or diffuse a longer distance, so that the probability for extraction
before recombination is increased.
Recombination can occur radiatively or non-radiatively:
Radiative recombination is the inverse process to optical excitation, however occur-
ring after thermalization. Typical photoluminescence spectra of GaN layers are
shown in Fig. 8.12. The radiative recombination probability is governed by the opti-
cal dipole matrix element i|ex| f . Classically, the power emitted by an oscillating
dipole is:
1 ω4 2 2
P= e x (8.2.25)
6π ε0 c3
where x is the electron-hole distance. The corresponding radiative recombination

lifetime τrad is
ω c3
τrad = = 6π ε0 3 2 2 (8.2.26)
P ω e x
• τrad is particularly short for semiconductors with large band gaps, since
ω ≈ E gap .
• Due to momentum conservation, τrad in indirect semiconductors is much
larger than in direct semiconductors.
• A typical value for ω = 1 eV and an electron-hole distance x ≈ 1 nm (exci-
ton radius) is τrad ≈ 5 × 10−9 s (compared to ≈1 µs for indirect semicon-
ductors).
134 8 Photovoltaics
Fig. 8.12 Radiative recombination (photo-luminescence) at T = 5 K in GaN grown on Al2 O3 -

substrates by two different methods: Metal-Organic Chemical Vapor Deposition (MOCVD) and
Molecular Beam Epitaxy (MBE). Optical excitation occurred with a laser at an above-band gap
energy of 3.72 eV with an intensity of 1 mW cm−2 . The spectra consist of a sharp excitonic peak
close to the GaN band gap, recombination between shallow donor-acceptor pairs followed by a
series of phonon replica, and radiative recombination at deep defects
Non-radiative recombination occurs via the emission of phonons. This can happen
via:
• phonon cascades analogous to thermalization, if a high density of states is present

in the band gap. However, this is normally not relevant for good solar cells.
• multi-phonon emission, i.e. the simultaneous emission of:
E gap
N=

phonons in a process of order N . The corresponding probability W scales like:
W ∝ e−
N
phonon
where e− phonon << 1 is a measure for the electron-phonon coupling.
In an ideal semiconductor with no defect levels in the band gap E gap , therefore N >>
1 holds. As a consequence, non-radiative recombination decreases with increasing
band gap. Defect levels in the middle of the band gap generally act as very efficient
recombination centers, as shown in Fig. 8.13.
Fig. 8.13 Illustration of non-radiative recombination without and with midgap defects. In the
right picture the recombination happens in two subsequent steps with lower order. Therefore, the
overall probability W for recombination here is much larger compared to the left picture: W ∝
1/2 N /2 >> N
In the following, we will discuss the rate equations for some special recombination
processes in more detail.
Monomolecular recombination: Here we have
R = cn n N D R = c p pN D (8.2.27)
N D is the volume density of defects acting as recombination centers, cn the capture

coefficient of the defects for electrons with the unit [cn ] = cm−3 s−1 .5 Alternatively,
one can also write for the capture coefficient cn = vn · σ D with vn , the electron
velocity and σ D , the capture cross section of defects. The resulting rate equation for
monomolecular electron recombination (i.e. single electrons combine with a constant
density of defects) is then:
dn
= G − R = G − cn n N D
dt
In steady state (dn/dt = 0) this results in an excess density of electrons propor-
tional to the generation rate and inversely proportional to the defect density:
G
n = (8.2.28)
cn N D
By introducing a carrier recombination lifetime τr ec via τr ec = n/R, we obtain

in steady state:
n n 1
R= ⇒ G− =0 ⇒ τr ec = (8.2.29)
τr ec τr ec cn · N D
5 An analogous discussion holds for the capture of holes (index p).

136 8 Photovoltaics
Fig. 8.14 Possible recombination rates R and thermal emission rates E between the valence and
conduction band and a deep defect located at an energy E D in the band gap. The emission rates are
proportional to a Boltzmann term as shown on the right
A more detailed analysis also taking into accounts the thermal re-excitation of
electrons trapped by defects into the conduction band as well as of trapped holes into
the valence band was developed by Shockley, Read and Hall. Their model explicitly
uses the principle of detailed balance between all recombination and excitation rates
as shown in Fig. 8.14. In steady state (R = E), this gives:
N D (np − n 0 p0 )
RS R H = (8.2.30)
E D −E i −E i
c−1
p n + n i exp kT + cn−1 p + n i exp − E DkT
√
E gap
Here, n i = pi = N V NC exp − 2kT are the thermally excited intrinsic carrier
densities without doping. N V , NC stand for the effective density of states of the
valence and conduction band, for example:
3/2
NC = 2/h 3 2π m ∗e kT (8.2.31)
E i is the position of the Fermi-level E F in the intrinsic, undoped semiconductor,

defined by:

EC − Ei
n i = NC exp − (8.2.32)
kT
Typical recombination centers in the most common semiconductors are:
• intrinsic defects such as vacancies, dangling bonds or dislocations

• charged impurities with a big Coulomb-like capture cross-section (esp.
Fe, Ni, Au, …in Si).
Bimolecular recombination: Here, an optically excited electron in the conduction

band recombines directly with a hole in the valence band. In the limit of strong
excitation (i.e. n >> n 0 , p >> p0 ) this gives:
R = c∗ · n p
with n = p, so that in steady state:
0 = G − R = G − c∗ n 2
This leads to a different dependence n on G as for monomolecular recombination:

G
n = (8.2.33)
c∗
Here c∗ is an effective capture coefficient for electrons by holes, and the electron
recombination lifetime is:
τr ec = 1/(c∗ p) (8.2.34)
Characteristic for bimolecular recombination is the sublinear (square root) depen-

dence of the excited carrier density on the generation rate.
Auger recombination: Here, the energy and momentum (wavevector) of a recom-
bining electron-hole-pair is transferred through direct Coulomb interaction to a third
electron or hole, which then rapidly returns by thermalisation to the ground state
(Fig. 8.15).
Fig. 8.15 Illustration of Auger recombination. The right picture shows a highly doped semicon-
ductor, where the Fermi-level lies already within in the CB: this guarantees that there are enough
electrons for this recombination process and momentum can be conserved
138 8 Photovoltaics
Auger recombination is a three-particle process via Coulomb interaction requiring

either n 0 or p0 to be sufficiently large. Depending on whether equilibrium holes or
electrons are dominant, the Auger recombination rate is given by:
R = c A np02 for strong p-doping, p0 >> p

R= c A pn 20 for strong n-doping, n 0 >> n
For steady state in heavily p-type semiconductors this leads to:
0 = G − R = G − c A npo2
This gives for the steady-state excess electron density:
G
n = (8.2.35)
c A p02
and
1
τr ec = (8.2.36)
c A p02
with a characteristic Auger coefficient c A . In solar cells, Auger recombination may

occur especially in highly doped contact regions with n 0 , p0 1017 cm−3 .
Surface recombination: For very high-quality absorber layers with little recombi-
nation in the bulk, recombination at surfaces or interfaces may become dominant.
Most semiconductors have intrinsic surface or interface defects with energy levels in
the band gap, which then can act as efficient recombination centers for photo-excited
charge carriers coming sufficiently close to the surface or interface. For example, at
the Si surface each Si atom is missing a fourth bonding partner, resulting in one
so-called dangling bond orbital per atom (cf. Fig. 8.16 left). Neutral dangling bonds
are occupied by a single electron carrying a spin (arrows in the left picture), which
allows their detection via electron spin resonance. The dangling bonds form a defect
band with an energetic position close to the middle of the bulk silicon band gap
(center). Therefore, they can act as very efficient recombination centers for photo-
excited carriers coming sufficiently close to the surface (Fig. 8.16 right). The neutral
Si dangling bonds are amphoteric recombination centers: they can either capture a
photo-excited electron from the conduction band (becoming negatively charged) or
a hole from the valence band (becoming positively charged). The surface recombi-
nation is then completed by the effective Coulomb-attraction of an opposite charge
carrier.
Fig. 8.16 Schematic illustration of surface recombination at a free unreconstructed surface of

silicon. See text for details
The influence of the surface on the recombination lifetime τr ec is described in

terms of a surface recombination velocity S:
Rsur f ace
S := (8.2.37)
n bulk
The unit of the surface recombination rate is [Rsur f ace ] = cm−2 s−1 . The unit of the
bulk photo-excited carrier density is [n bulk ] = cm−3 . Thus, the unit of S is indeed
that of a velocity, cm s−1 . Typical values for S can vary largely between S = 1cm s−1
for passivated surfaces (e.g. Si, where all surface dangling bonds are passivated by
forming Si-H bonds with atomic hydrogen) to S = 106 cm s−1 (for bad, unpassivated
surfaces). The upper limit of S is the thermal velocity vth of minority carriers moving
in the single direction towards the surface (S < vth ):
1 ∗ 2 1
m vth = kT ⇒ Smax ≈ 2 · 107 cm s−1
2 2
As an example, Fig. 8.17 shows the relative contributions of different recombination
processes to the overall carrier lifetimes in p-doped high quality crystalline silicon.
8.2.6 Separation and Extraction
Directly after optical excitation an electron and the corresponding hole first remain
close together in a Coulomb-bound excitonic state. The excitonic binding energy is
obtained analogous to that of the hydrogen atom (1 Rydberg):
140 8 Photovoltaics
Fig. 8.17 Room temperature recombination lifetimes of electron minority carriers in p-doped solar-
grade silicon with different equilibrium hole concentrations p0 . The experimentally observed life-
time (solid line) varies between 1ms and 1µs [2]. At low p-doping levels, it is dominated by
monomolecular Shockley–Read–Hall recombination (dotted line), whereas Auger recombination
dominates at high doping levels above 1016 cm−3 (dashed line). Due to the indirect nature and the
low value of the silicon band gap, radiative recombination (fine dashed line) is always negligible
e4 m r∗ed 1 1 1 1
E bind = ; n = 1, 2, . . . ; with = + ∗
8h 2 ε2 ε02 n 2 m r∗ed ∗
me mh
m r∗ed 1
E bind (n = 1) = 1Ryd (8.2.38)
m e ε2
With 1 Ryd = 13.959 eV this typically gives E bind ≈ 10 meV due to the reduced
effective mass and high dielectric constant (ε ≈ 10) in most semiconductors. There-
fore, at room temperature phonons are already sufficient to dissociate the exciton and
separate the electron and the hole through electron-phonon interaction (Fig. 8.18).
A more efficient separation occurs in internal electric fields e.g. in p/n-junctions,
p/i/n-junctions, metal/semiconductor junctions or with an applied external bias volt-
age. In the E(x)-diagram, such an internal field leads to an inclination of both band
edges (Fig. 8.19). The required field can be estimated from the potential energy drop
across the radius R1 of the excitonic ground state:
Fig. 8.18 Thermal separation of an optically excited electron-hole pair. The excitonic binding
energy is indicated as a small dip in the conduction band when electron and hole are close together.
The dashed arrow indicates excitation of the electron by absorption of a phonon followed by exciton
dissociation
Fig. 8.19 Linear spatial

inclination of the valence
and conduction band edges
due to a constant electric
field, causing the drift of
electrons and holes in
opposite directions
eER1 ≥ E bind (n = 1)
E bind
E=
e R1
2
with R1 = 4π εε0
e2 m r∗ed
For a typical value of R1 ≈ 5 nm and a binding energy of 10 meV, this gives an

electric field of E ≥ 20000 V cm−1 .
In organic solar cells, excitonic binding energies are much larger (≈0.5 eV)
because of the lower dielectric constant. Therefore, thermal exciton dissociation is
no longer sufficient. An efficient separation of optically excited electrons and holes
is then achieved at a suitable internal hetero-junction, as shown in Fig. 8.20 below.
The transport of separated electrons and holes to the contacts, where they finally
are extracted, either occurs by diffusion (in a concentration gradient) or drift (in
electric fields). This gives e.g. for the current density of electrons in one dimension:
142 8 Photovoltaics
Fig. 8.20 Hetero-junction between two organic semiconductors with shifted HOMO and LUMO
energies. After optical excitation, electrons will preferentially be trapped in the material with the
lower LUMO energy, while holes will collect preferentially in the material with the higher HOMO
energy
Table 8.2 μn , Dn and L n for electrons at room temperature in different materials. For L n , a
recombination lifetime of 1µs has been assumed
Typical values μn (300 K) Dn (300 K) L n (τr ec ≈ 10−6 s)
[cm2 /(V s)] [cm2 /m] [µm]
crystalline Si 1000 25 50
amorphous Si 0.1 2.5 × 10−3 0.5
polymers 10−3 2.5 × 10−5 0.05
∂n
jn = env = enμn E + eDn (8.2.39)
∂
dri f t
x
di f f usion
e kB T
with μn = τ and Dn = μn
m ∗e e
Thus, large mobilities μn , μ p and the corresponding large diffusion coefficients

Dn , D p are required to allow for a good carrier extraction. During their recom-
bination lifetime τr ec charge carriers can diffuse over a distance L n or L p before
recombining:

L n, p = Dn, p τr ec (8.2.40)
Typical values of μn , Dn and L n for electrons at room temperature in different

material classes are listed in the Table 8.2. For L >> 1/α (1/e absorption depth of
light), solar cells can rely on diffusion for carrier extraction, otherwise drift in an
internal field is required.
8.3 Types of Solar Cells 143
Fig. 8.21 Most common materials used for solar cell production. The work horse of photovoltaics
with a market share above 95% is silicon in the form of monocrystalline or polycrystalline wafers.
In thin film technology, compound semiconductors with a direct band gap such as CdTe or copper-
indium-diselenide (CIS) are dominant. Multijunction cells with up to six stacked junctions predom-
inantly use high quality III-V semiconductor alloys. Due to their high cost, multijunction cells are
mainly used in space applications or in concentrated photovoltaics, where the cell sizes can be much
smaller. The second class of solar cells is based on purely organic materials or combines organic
and inorganic components in hybrid technologies
8.3 Types of Solar Cells
The history of modern solar cells for energy conversion basically started around 1955
with a brief note by Chapin, Fuller and Pearson of Bell Telephone Laboratories in
the Journal of Applied Physics about a p/n-junction device made from crystalline
silicon with an overall solar conversion efficiency of 6%. This was much more than
what had been observed for at that time commercially available semiconductor-based
photodetectors or thermoelectric devices (see Chap. 9) of below 1%. The authors also
estimated that an optimized Si solar cell could have efficiencies around 22%, which
is very close to the currently reported record efficiency of 26%. Since these early
days, solar cells have experienced a very dynamic and successful development in
terms of conversion efficiencies, industrial production and deployment as well as a
cost reduction. At the same time, many other material systems have been investigated
and developed for their use in photovoltaic converters (Fig. 8.21).
Today, many different materials and technological approaches are actively inves-
tigated for different purposes. However, only few of them currently contribute sig-
nificantly to large scale solar energy conversion, notably crystalline silicon, CdTe
and copper-indium-gallium-diselenide . One important physical parameter guiding
the choice of materials is the band gap of the absorber layer, which has to be matched
to the solar spectrum. If the band gap is too large, too few solar photons are absorbed
and the electrical current density provided by the solar cell is small. If the band
gap is too small, most energy is lost in unwanted thermalization losses and also the
voltage provided by the cell is small. For single junction cells, there is a broad band
144 8 Photovoltaics
Fig. 8.22 Absorption

coefficient spectra close to
the band gap of some
common absorber materials
in inorganic solar cells
gap range between 1 and 1.6 eV where a maximum theoretical conversion efficiency
around 30% could be reached. For multijunction cells, different absorber layers with
carefully matched band gaps and thicknesses are used. Absorption coefficients of a
small collection of materials with suitable band gaps are shown in Fig. 8.22.
A chart summarizing certified solar cell efficiencies for the different photovoltaic
technologies is published and updated annually by the National Renewable Energy
Laboratory. The 2022 version of this chart is reproduced below. Note that this chart
only refers to best results obtained for small scale research cells. Mass-produced
solar modules typically have efficiencies which are a third or more lower than these
record efficiencies.
As can be seen, mature solar cell technologies such as monocrystalline silicon
wafer cells or CIGS thin film cells have reached a plateau at around 25% efficiency.
The highest efficiencies approaching 50% have been reached for multijunction cells
using concentrated sunlight. More recent technologies with still significant improve-
ments and fast learning curves have been organic solar cells since about 2000 and
perovskite cells since about 2010. In the following we briefly discuss a few types
of solar cells in more detail to introduce some important aspects of the underlying
technology (Fig. 8.23).
8.3.1 Crystalline Si p/n Diffusion Cell
The basic band diagram of a crystalline silicon solar cell is shown in Fig. 8.24. It
is based on mono- or poly-crystalline Si wafers with a typical size of 15×15 cm2
and a thickness of 200–300 µm. This thickness is much larger than the value of 3/a
required by optical absorption. Instead it is determined by the mechanical strength
of the wafers in order to guarantee a sufficiently high yield in large scale automated
production. Usually, the major part of the wafer forms the base, which is weakly
p-type doped with an acceptor concentration N A around 1015 cm−3 . Then optically
excited electrons are the minority carriers and can diffuse freely through the entire
base due to their higher mobility compared to holes as the majority carriers. In
8.3 Types of Solar Cells
Fig. 8.23 Research solar cell efficiencies up to 2022. Permission for reproduction by NREL is gratefully acknowledged. The original graph can be found at:
https://www.nrel.gov/pv/cell-efficiency.html
145
146 8 Photovoltaics
Fig. 8.24 Schematic band edge level diagram of a c-Si p/n diffusion solar cell. p++ , p and n+ denote
different doping levels with acceptors (density N A ) and donors (N D ). The corresponding dopant
ionization energies E A and E D determine the position of the Fermi level in thermal equilibrium
shown by the dashed horizontal line (without illumination). W p and Wn are the widths of the space
charge layers at the p/n-junction, across which the built-in potential eUbi drops. AR stands for a
suitable anti-reflection coating
high quality monocrystalline Si wafers diffusion lengths of more than 1 mm can be

obtained.
The back contact is made from Al which is thermally diffused into the wafer.
Since Al is a substitutional acceptor in Si, this creates a thin very highly doped (p++ )
layer between the base and the Al contact, with two functions. On one hand, it insures
a good ohmic contact to the base, on the other hand it causes a small upward jump
in the band diagram caused by the degenerate p++ -doping. This creates the so-called
back surface field (BSF) which repels diffusing electrons away from the back contact
and attracts diffusing holes towards the p-type back contact.
On the opposite side of the absorber, a thin n-type layer called emitter with a
high doping level (n+ ) is created e. g. by in-diffusion of phosphorus donors. This
results in a p/n-diode junction with corresponding space charge regions W p and
Wn . Because of the different doping levels, the extent of the space charge region
is much larger in the weakly doped base, where also the generation rate of photo-
carriers due to the incoming light (from the right) is highest. In the electric field
of this space charge region, separation of photo-excited carriers occurs with a high
efficiency, and electrons can drift to the front contact. Because of their high doping
levels promoting fast Auger recombination, both the BSF region and the emitter do
not contribute significantly to the solar cell efficiency. The top side is completed by
an anti-reflection coating (e.g. a λ/4-Si3 N4 layer) and a grid of screen-printed Ag
contacts.
These basic ingredients of crystalline Si p/n solar cells have been varied and opti-
mized over now almost 70 years. As an example, Fig. 8.25 shows a three-dimensional
view of the so-called PERL cell (Passivated Emitter and Rear Locally diffused solar
cell) developed at the University of New South Wales in Australia in 1998. This was
Fig. 8.25 Three-dimensional

layout of a PERL cell
(Passivated Emitter and Rear
Locally diffused solar cell).
See text for more details.
Here, an n-type base instead
of a p-type base has been
used to avoid stability
problems due to boron
acceptors
the first Si p/n solar cell with a certified efficiency of 25% and made use of optical
lithography to produce local p and n contacts within an otherwise passivated Si wafer
surface covered by SiO2 . Lithography can also be used to create a regular inverted
pyramid structure at the front surface by selective etching and was combined with an
additional λ/4 layer as an efficient anti-reflection coating. The design of local back-
side contacts in combination with backside passivation is also referred to as PERC
cell (Passivated Emitter and Rear Cell). Finally, the latest improvements in Si solar
cells are based on the introduction of passivation layers using very thin hydrogenated
amorphous silicon films on backside and front surfaces. These are called HIT cells
(Heterojunction with Intrinsic Thin layer) and were introduced by Sanyo Electric
Company around the year 2000. This type of crystalline Si cells currently holds the
record in conversion efficiency of almost 27%.
To calculate the spatial profile of the conduction and valence band edges E C and
E V one starts with the Poisson equation relating the local space charge density ρ
with the divergence of the electrical field and the potential :
ρ ρ
divE = ; E = −gradφ ⇒ div(grad)φ = φ = −
εε0 εεo
In doped semiconductors the local space charge density is given by the difference
of ionized donor and acceptor densities: ρ = N D − N A .
• The built-in potential can be approximated by: eUbi ≈ E gap − E D − E A

• The extent of the space charge regions W p and Wn is obtained from:
1/2
2εε0 NA + ND kT
W = Ubi − 2
e NA ND e
8.3.2 Metal-Insulator-Semiconductor (MIS)-Schottky Contact Cell
This is a second type of wafer-based diffusion cells using the electrical field generated
by a top Schottky contact instead of a p/n-diode for carrier separation and extraction.
148 8 Photovoltaics
Fig. 8.26 Band level diagram of a silicon wafer-based Metal-Insulator-Semiconductor (MIS) solar
cell. The n+ -doped emitter is replaced by a metallic Schottky contact, separated from the base by a
thin tunnel oxide for surface passivation
Because of the missing n+ -emitter, it is easier to fabricate and requires a smaller

thermal budget, however until now also has only achieved lower efficiencies. In such
a MIS cell, a thin tunnel oxide or oxynitride is placed between the base and the top
Schottky contact metal. The main purpose of the tunnel oxide is to prevent surface
recombination. Otherwise, the band diagram and layout of a MIS cell resembles that
of p/n-junction cells without emitter (Fig. 8.26).
8.3.3 Amorphous Si Thin Film Drift Solar Cells
Starting around 1980, thin films of hydrogenated amorphous silicon (a- Si:H) were
investigated as an alternative to crystalline Si cells. This development was enabled
by the then unexpected discovery that also this disordered form of silicon could be
doped n- and p-type by the addition of phosphorus and boron, similar to its crys-
talline counterpart. A-Si:H can be deposited by Plasma-Enhanced Chemical Vapor
Deposition (PECVD) from silane (SiH4 ) at much lower temperatures around 250 ◦ C
on low-cost glass substrates coated with a contact layer using transparent conduc-
tive oxides such as ZnO, SnO2 , or indium-tin oxides (ITOs). In addition, the optical
absorption coefficients of a-Si:H in the visible spectral range are much higher than
those of crystalline Si (cf. Fig. 8.22), because the lack of crystalline periodicity in the
amorphous network, removing the requirement of wavevector-conservation during
an optical excitation and resulting in a “quasi-direct” optical band gap already men-
tioned above. This enabled an a-Si:H-based thin-film technology with absorber layer
thicknesses below 1µm, featuring much less material use and a much lower ther-
mal budget. The major disadvantage of amorphous a-Si:H compared to crystalline
Si on the other hand is the much lower electron and hole mobility (typically below
1 cm2 V−1 s−1 at room temperature) due to strong carrier scattering by the disordered
atomic network. This requires the presence of a sufficiently strong electrical field
throughout the absorption layer to move photo-excited carriers by electrical drift to
the contacts. Diffusion would be too slow to extract carriers before they recombine.
Fig. 8.27 Band level diagram of an amorphous silicon thin film drift cell
This motivates the layout of an amorphous Si solar cell as a p-i-n drift cell with a
high electrical field in an undoped (intrinsic, “i”) absorber layer. This field is created
by strongly doped n+ and p+ extraction layers on opposite sides, followed by suitable
metallic and transparent conducting oxide (TCO) contacts, as shown in Fig. 8.27.
For a typical absorber layer thickness of dabs = 0.5 µm and a difference of the
equilibrium Fermi-level positions in the n- and p-contact layers of 1 eV, the electrical
field in the absorber can be estimated to about:
Ubi 1V
E≈ ≈ = 2 × 104 Vcm−1
dabs 0.5 µm
Illumination of the a-Si:H cell occurs through the glass substrate and the TCO layer
at the p-contact side. This is required by the fact that at typical operating temperatures
the mobility of holes in a-Si:H is much (about a factor of 100) lower than that of
electrons, so that optical generation of excess holes should occur as close to the
p-contact as possible. Although a-Si:H solar cells very quickly achieved conversion
efficiencies above 10% at much lower costs than crystalline Si cells, their commercial
acceptance was severely affected by the phenomenon of light-induced degradation
(the so-called Staebler–Wronski effect). During long-term (several 1000 h operation
of a-Si:H solar cells in normal sunlight, their conversion efficiency decreases by
several percent (absolute) due to the optical generation of additional recombination
centers. Similar degradation effects also plague other thin film technologies such as
the organic or perovskite solar cells described below.
8.3.4 CdTe and Cu(In,Ga)Se2 Compound Thin Film Solar Cells
Parallel to the development of thin film solar cells based on amorphous Si, also thin
film cells using polycrystalline CdTe or CuInGaSe2 (CIGS) chalcopyrite compound
absorber layers were developed since 1980. Both absorber materials are direct semi-
conductors, allowing a reduced absorber thickness around 1 µm. Due to the higher
carrier mobility in the polycrystalline absorber layers, they make use of both, carrier
drift and diffusion, for charge extraction. Their thermal budget is between that of
150 8 Photovoltaics
Fig. 8.28 Layer sequence of

a CdTe compound thin film
solar cell
a-Si:H and crystalline Si cells, depending on the specific deposition and processing
methods used. Today, both compound thin film technologies have reached conversion
efficiencies above 22% in the laboratory. Being a thin film technology, a glass sheet
with a transparent conductive oxide is used as the substrate or superstrate of the cells
(Fig. 8.28). In CdTe cells, a thin n-type CdS layer with a wider band gap of 2.4 eV is
deposited as the n-contact and as a structural buffer layer. The p-type CdTe absorber
(E gap = 1.45 eV) is deposited next. The solar cell is then completed by an ohmic
metal contact, e.g. Mo. In CIGS cells, basically only the CdTe absorber is replaced
by Cu(In, Ga)Se2 . An alloy of In with Ga instead of pure In is used to increase the
band gap of the absorber layer from around 1 eV to an optimum value around 1.4 eV.
Because of the polycrystalline nature of the absorber, the conversion efficiencies of
both types of cells are increased significantly by special grain boundary passivation
treatments.
8.3.5 Dye-Sensitized Solar Cells (DSSC)
This thin film cell type was developed since 1990 first at EPFL and is also known as
Grätzel cell after one of its inventors. It is inspired by early steps in photosynthesis
for light absorption and carrier separation. Light absorption occurs in a molecular
organic dye, which is dispersed on the large effective surface of nanostructured
TiO2 . As shown in Fig. 8.29, a photo-excited electron is quickly transferred from
the molecular LUMO to the conduction band of n-type TiO2 and from there to the
negative contact. The remaining hole in the HOMO of the dye molecule is then
neutralized by an electron coming from a redox shuttle (e.g. iodine) in a liquid
electrolyte and is transported to the TCO contact in form of the oxidized redox
ion. There, it is reduced again by taking up an electron from the TCO. Today, dye-
sensitized solar cells have achieved conversion efficiencies of up to 13%.
8.3.6 Organic Bulk Heterojunction Cell
Organic photovoltaics (OPV) is based on purely organic materials (molecular or

polymeric) for light absorption and charge separation. As already discussed in the
Fig. 8.29 Schematic structure and working principle of a dye-sensitized solar cell (DSSC)
Fig. 8.30 Working principle of an organic bulk heterojunction cell. See text for details
beginning of this chapter, organic semiconductors are characterized by large exciton

binding energies, so that abrupt heterojunctions between an electron donor material
(A) and an electron acceptor material (B) are required for efficient carrier separation.
After photo-excitation in A, the excited electron is transferred to B. Similarly, after
photo-excitation in B, the hole is transferred to A. Thus, electrons are collected in B
and holes in A. A second complication arises from the fact that excitons in organic
materials also have a very low mobility for diffusion, which limits their diffusion
lengths to a few 10 nm. To achieve a good compromise between a small average
distance between A and B for efficient carrier separation and, at the same time,
a sufficiently large combined thickness A + B for efficient photon absorption, the
concept of a bulk heterojunction (BHJ) is used. Both components are intermixed
on a nanoscale, forming a large effective interface as shown in Fig. 8.30, but still
remain singly connected to allow a good transport of the separated carriers to the
respective contacts. The development of organic solar cells started around the year
2000, leading to present record efficiencies of 18%.
A popular example of two organic components forming good bulk heterojunctions
is shown in the Fig. 8.31.
152 8 Photovoltaics
Fig. 8.31 Two components A and B for bulk heterojunctions: A = P3HT: Poly(3-hexylthiophene)
(left). B = PCBM: [6, 6]-phenyl-C61-butyric acid methyl ester (right)
8.3.7 Perovskite Thin Film Solar Cells
The latest newcomer (2013) and the fastest rising star among the so-called emerging
photovoltaic technologies with current record efficiencies above 25% is based on the
material class of perovskites. Perovskites are ionic compounds with the stoichiom-
etry ABX3 , where A and B are cations and X is an anion. The lattice structure of
perovskites consists of AX6 octahedra, eight of which surround a B cation as a cubic
cage. Perovskites used for photovoltaics can be purely inorganic (such as CsPbI3 )
or also contain organic cations. The best-known example for the latter is methy-
lammonium lead iodide (CH3 NH3 PbI3 ), which contributed a lot to the success of
perovskite solar cells. Optical band gaps of perovskites are mostly direct and can
be adapted easily to solar cell requirements by variation of the constituents. They
also can be deposited with a high electronic quality by low-temperature deposition
methods such as spin-coating from solution.
Figure 8.32 shows the basic layer sequence for solar cells using perovskite
absorbers. As for almost all thin film solar cells, glass substrates covered by a trans-
parent conductive oxide are used as the mechanical support and the optical window
for the absorber layer. The perovskite absorber layer itself is sandwiched between
other inorganic or organic layers to optimize carrier separation and collection by the
final contact layers. Many of these interlayers have been copied from their earlier
use in organic or dye-sensitized solar cells.
Fig. 8.32 Layer sequence of normal and inverted perovskite solar cells. FTO stands for the trans-
parent conductive fluorine-doped tin oxide, TiO2 , PEDOT:PSS, PCBM and Spiro-OMeTAD are
special electron or hole conductors helping to transport charge carriers to the final metallic contacts
8.4 I-U-Characteristics of Solar Cells 153
8.4 I-U-Characteristics of Solar Cells
8.4.1 Ideal Diodes
The current-voltage-characteristics of an ideal pn-diode is obtained by solving the

drift-diffusion equation for electrons and holes:
jn = −enμn E − eDn grad(n) (8.4.1)

j p = epμ p E + eD p grad( p) (8.4.2)
This gives (in the dark) (Fig. 8.33):

eU
j = js exp −1 (8.4.3)
n̂k B T
with the following quantities:
• the saturation current density for U → −∞:

Dn Dp
js = e n 0, p + p0,n (8.4.4)
Ln Lp
√
• the diffusion length L = Dτr ec
• n 0, p : equilibrium electron concentration on the p-side
• p0,n : equilibrium hole concentration on the n-side

E gap
n 0, p , p0,n ∝ exp − (8.4.5)
kB T
• n̂: the ideality factor with 1 ≤ n̂ ≤ 2

– n̂ = 1: no recombination in the space charge region W
– n̂ = 2: complete recombination in the space charge region W (e.g. in a LED)
Fig. 8.33 Ideal circuit of a

pn-junction in a
semiconductor
154 8 Photovoltaics
Fig. 8.34 Schematic current-voltage characteristics of an ideal pn-diode in the dark and under
illumination. MPP is the maxiumum power point. See text for details
With illumination the photo-generated current density j ph (e.g. j ph = e(nμn +

pμ p )E for a drift cell) is subtracted from the dark I-U characteristics:

eU
j = js exp − 1 − j ph (8.4.6)
n̂k B T
From this, the following important solar cell parameters can be deduced (Fig. 8.34):
• Uoc : the open circuit voltage

nk B T j ph
j =0 → Uoc = ln −1 (8.4.7)
e js
• jsc : the short circuit current density (U = 0)
U =0 → jsc = − j ph
• the fill factor (FF):
Um jm
FF = (8.4.8)
Uoc jsc
Here Um and jm are the voltage and current density in the maximum power point
(MPP), the point of operation where the maximum electrical power density can be
drawn from the cell. This is determined via a suitable ohmic load resistor Rload
8.4 I-U-Characteristics of Solar Cells 155
connected to the illuminated diode. With these parameters, the solar cell efficiency
η can be written as:
Pout Um jm F F · Uoc · jsc

η= = = (8.4.9)
Pin Pin Pin
where Pin is the incoming light power density ([Pin ] = W m−2 )
The open circuit voltage Uoc can also be described in terms of so-called quasi-
Fermi-levels of electrons and holes. In the dark, the equilibrium carrier concen-
trations n 0 and p0 are determined by the distance of the common Fermi-energy
E F from the band edges E C and E V :

EC − E F E F − EV
n 0 = NC exp − , p0 = N V exp −
kT kT
Under illumination, both carrier concentrations increase to their non-

equilibrium values: n 0 → n 0 + n = n and p0 → p0 + p = p, with
n, p = Gτr ec . This allows the definition of the quasi-Fermi-levels E F,n
and E F, p of electrons and holes according to:

E C − E F,n
n = n 0 + n =: NC exp −
kT

E F, p − E V
p = p0 + p =: N V exp −
kT
The open circuit potential eUoc is then given by the splitting of the two quasi-
Fermi-levels as shown in the Fig. 8.35.
8.4.2 Real Diodes
The equivalent circuit of real pn-diodes (cf. Fig. 8.36) includes an ideal diode together
with a parallel resistance R P and a series resistance R S : Both resistances modify
the I-U characteristics of real diodes (I = j · A) in the following way:

e(U − I Rs ) U − RS I
I = Is exp −1 + − I ph (8.4.10)
n̂k B T Rp
This equation takes into account that the voltage U across the ideal diode is reduced
by the voltage drop I · R S across the series resistor, and the same voltage drives an
156 8 Photovoltaics
Fig. 8.35 Splitting of the quasi-Fermi-levels and resulting open circuit potential
Fig. 8.36 Equivalent circuit

of real pn-diodes. Series
resistances usually are
caused by contacts and
external leads, while parallel
or shunt resistances are
caused by conductive defects
in the absorber layer (pin
holes, impurities, etc.)
additional leakage current through R P . R S and R P can be deduced from the slopes
of the I-U curve at U = 0 and U = Uoc :
−1 −1
dI dI
Rs = Rs + R p =
dU dU
Uoc U =0
The series and parallel resistance of real diodes directly affect the solar cell
efficiency in the following ways:
• Too large values of Rs decrease Isc and FF

• Too small values of R p decrease Uoc and FF
• in real devices with an area of 1cm−2 , Rs of a few noticeably decreases the
cell efficiency η, whereas only for R p ≤ 100 the efficiency is significantly
affected.
8.5 Efficiency Limits of Single Junction Solar Cells 157
8.5 Efficiency Limits of Single Junction Solar Cells
For the estimation of the maximum efficiency of an ideal pn-solar cell we make the
following simplifying assumptions:
1. All photons with ω ≥ E gap are absorbed, all photons with ω ≤ E gap are trans-
mitted. This means that there is no reflection.
2. No non-radiative recombination at defects or interfaces occurs.
3. No radiative recombination occurs or there is radiative equilibrium between the
Sun and the solar cell. The latter is referred to as the detailed balance or Shockley–
Queisser limit [5].
4. The solar cell is an ideal diode: R p = ∞; Rs = 0; n̂ = 1
As discussed above, the solar cell efficiency per unit area is η = (F F · Uoc ·
jsc )/Pin with Pin given by the solar spectrum:
∞
Pin = deγ (ω) with (8.5.1)
0

−1
(ω)3 ω
deγ = exp −1 dωd (8.5.2)
4π 3 3 c03 k B Tsun
Here, Tsun = 5800 K and integration occurs over according to the solid angle
extended by the solar cell, or taking into account a possible concentration system.
The short circuit current density jsc is only produced by photons with ω ≥ E gap ,
and the maximum energy provided by these photons is eUoc , due to thermalisation
losses. Using the photon density of states and the thermal equilibrium occupation
factor as derived at the beginning of this chapter, the maximum output power for a
fill factor of F F = 1 would be:
∞
Pout = eUoc · Dγ , f γ (ω)dωd = Uoc jsc
E gap
This results in a maximum efficiency of a solar cell without hot carrier extraction
(see below) of:
∞ −1
Pout E gap eUoc (ω)2 exp k Bω Tsun − 1 dω
ηmax = = ∞ −1
Pin ω
0 (ω) k B Tsun − 1
3 exp dω
Thus, ηmax depends strongly on the energy gap E gap of the absorber because:
• ηmax → 0 for E gap → 0 (Uoc < E gap → 0)

• ηmax → 0 for E gap → ∞ (no adsorption, so that jsc → 0)
158 8 Photovoltaics
Under the simplifying assumptions made so far, there is a band gap optimum for
the maximum efficiency with a value of ηmax = 42% for E gap = 1.1 eV, assuming
that E oc ≈ E gap . Thus, Si with a room temperature band gap of E gap = 1.1 eV
would be the ideal absorber.
For a more accurate estimate of the efficiency limits, however, we also have to take
into consideration the influence of a realistic Uoc < E gap and a realistic F F < 1.
As discussed above, Uoc is determined by the splitting of the quasi-Fermi-levels
E F,n and E F, p under illumination and depends on the generation rate G, but also
on recombination processes and e.g. donor and acceptor levels of the pn-diode in
the junction. This lowers eUoc to typically around 70% of E gap . A more accurate
quantitative discussion of the dependence of the open-circuit-voltage on the band
gap of the absorber is very complicated and would go beyond the scope of this book.
A second challenge is to estimate the fill factor of a solar cell in the maximum
power point assuming an ideal diode current-voltage characteristics with known
values of Uoc and Isc . A starting point is the maximum power condition, leading to
the following relations for the related voltage and current, Um and Im :

dI I
d(I U ) = 0 = U dI + I dU ⇒ =−
m dU m U m
Inserting this into the ideal diode (8.4.3) gives:

dI e eUm I
= Is exp =− (8.5.3)
dU m kT kT U m
Moreover, for ideal diodes we have (Fig. 8.37)

eUm
Im = Is exp − 1 − Isc
kT

Isc eUoc
= 1 − exp
Is kT
Fig. 8.37 Relation between

the ratio Im /Um of current
and voltage in the maximum
power point of a solar cell
and their derivate dIm /dUm
8.6 Increasing Solar Cell Efficiencies 159
Inserting this into the above equation for the maximum power point results in:

Im kT eUm
Um = − exp −
Is e kT

kT e(Uoc − Um )
= exp −1
e kT
This equation can be solved numerically to obtain Um and Im for a given Uoc and
Isc and, thus, the corresponding fill factor F F. In a good approximation one gets for
ideal diodes:

eUoc eUoc
kT − ln 1 + kT
FF ≈ (8.5.4)
1 + eU
kT
oc
This leads to values of F F ≈ 80% for optimized solar cells.
Some additional remarks concerning conversion efficiencies of real solar cells:
• Emission due to radiative recombination can usually be neglected if exciton

separation is fast enough
• Non-radiative recombination causes the largest deviation from the maxi-
mum efficiency ηmax
• Concentration of light leads to higher efficiencies due to an increase of jsc ,
resulting in a corresponding increase of Uoc due to a larger splitting of
the electron and hole quasi-Fermi-levels. For a concentration factor C this
ideally gives:

kT jsc kT C · jsc
Uoc ∝ ln → Uoc = ln
e js e js
Thus, the efficiency of concentrated solar power (CSP) cells is larger.

• For strong concentration the temperature of the cell increases, which has
to be prevented by active cooling. Otherwise, the dark saturation current js
increases exponentially, decreasing Uoc correspondingly.
8.6 Increasing Solar Cell Efficiencies
The ideal conversion of energy from the Sun to energy on Earth would be subject to
the thermodynamic Carnot efficiency limit defined by the absolute temperatures on
the surfaces involved:
160 8 Photovoltaics
Table 8.3 Different approaches to increase the expected maximum conversion efficiencies ηmax
of solar cells
Expected ηmax (%) Approach
31 Single p/n junction without reflection and recombination
39 Including optical down-conversion
44 Tandem solar cell
46 Impact ionization (carrier multiplication)
49 Optical up-conversion
51 Triple solar cell
54 Thermal conversion, thermo-photovoltaics, thermionics
68 “hot carrier” cells, N→ ∞ multijunctions
5800 K − 300 K
ηCar not = = 95%
5800 K
Of course, this is not practical and the more realistic efficiency limit for single
absorber solar cells is the Shockley–Queisser limit of about η = 31%. Higher effi-
ciencies would require circumventing the main loss mechanisms such as thermali-
sation and recombination. Indeed, several approaches exist or have been proposed
to overcome the Shockley–Queisser limit by using multi-absorber solar cells or so-
called “Third generation” solar cell concepts as listed in Table 8.3.
In the following, we will briefly sketch some examples for these approaches. A
more detailed treatment of the corresponding research activities can be found in the
related specialized literature.
8.6.1 Down Conversion
Optical down-conversion makes use of specific dye molecules, color centers or quan-
tum dots in a transparent layer in front of a solar cell, to convert high energy photons
with ω > 2E gap into two photons with ω ≈ E gap , where E gap is the band gap
energy of the subsequent solar cell absorber layer. This reduces the thermalization
losses for high energy photons compared to their direct absorption in the absorber
layer, thus increasing the overall efficiency (Fig. 8.38).
8.6.2 Tandem Cells
Tandem solar cells have been actively investigated since the advent of thin-film
solar cells around 1980. They combine two single junction solar cells with different
Fig. 8.38 Schematic electronic level sequence involved in optical down-conversion of high energy
photons
optical band gaps E gap placed optically and electrically in series. A top cell with
larger band gap absorbes higher energy photons, while lower energy photons are
subsequentially absorbed by the bottom cell with lower band gap. This theoretically
allows a significant reduction of thermalization losses compared to a single junc-
tion cell, but requires a careful optimization of thickness and band gap of the two
absorber layers for maximum efficiency under varying illumination conditions. Since
tandem solar cells require the same illuminated area but reach a higher efficiency
than single junction cells, they potentially provide a significantly lower system cost
in real applications. As shown in Fig. 8.39, there are two possibilities two combine
top and bottom cells electrically. In a three-terminal combination, a transparent con-
ductive intermediate layer provides a third contact, which allows to operate both
cells separately at their maximum power points. Most current approaches, however,
are based on a two-terminal combination, so that the same current I = I1 = I2 flows
through both cells, while the two cell voltages add up to a single external voltage
U = U 1 + U 2. In order to realize this series interconnection, a dedicated tunnel
junction has to be added as an additional element (see below).
For two-terminal tandem solar cells operated under AM0 conditions, theoretical
efficiencies of up to 44% have been calculated for the combination of a top cell
absorber with a band gap of 1.9 eV and a bottom cell absorber with a band gap of
1 eV (Fig. 8.40). These strict band gap limitations can be relaxed significantly when
accepting reduced efficiency limits of about 40%.
Fig. 8.39 Three-terminal and two-terminal electrical connections for tandem solar cells
162 8 Photovoltaics
Fig. 8.40 Theoretical maximum efficiencies of two-terminal tandem solar cells under AM0 illu-
mination for different combinations of top and bottom absorber band gaps
Fig. 8.41 Band level diagram of a thin film tandem solar cell combining an amorphous a-Si:H
top cell with a microcrystalline µc-Si:H bottom cell. A p-type amorphous Si-C alloy layer acts
as a transparent top contact and optical window. The electrical interconnect between both cells is
achieved by a highly doped p/n tunnel junction
As a recent example, Fig. 8.41 shows the band level scheme of a so-called 2nd
generation “micromorph” tandem thin film Si solar cell combining an amorphous
Si top cell and a microcrystalline Si bottom cell. Both materials can be produced
with a low thermal budget by plasma-enhanced chemical vapor deposition on low-
cost glass substrates, are close to the ideal band gap combination of tandem solar
cells and have achieved efficiencies of about 15%. Other current examples explore
perovskite-based top cells combined with bulk wafer silicon solar cells or CIGS thin
film cells.
For the electrical interconnection between two cells in a multi-junction solar
cell, an efficient loss-less recombination between electrons (holes) of the top cell
with holes (electrons) of the bottom cell needs to be realized. This is achieved by
a dedicated tunnel junction involving heavily doped n+ and p+ layers. The doping
Fig. 8.42 Details of the band level alignment at a n++ /p++ tunnel junction between two sub-cells
of a multi-junction solar cell. Due to the high doping levels, the Fermi energy E F lies within the
conduction band on the n-side and the valence band on the p-side (degenerate doping), allowing
access to a large density of empty states on both sides. In addition, the extents of the space charge
regions W are small enough to allow efficient carrier tunnelling through the junction barrier
Fig. 8.43 Schematic layer sequence of a triple-junction solar cell. Mono-crystalline Ge is used
for the bottom cell, followed by a GaAs middle cell and a GaInP2 top cell. Each cell includes a
p/n-junction. The three cells are internally electrically connected by two tunnel junctions
levels have to be high enough to reduce the corresponding space charge layer widths
sufficiently to allow effective carrier tunnelling through the junction (Fig. 8.42).
The increased efficiency of a tandem solar cell compared to a single-junction cell
can be further improved by combining more sub-cells in series, thus further reducing
thermalization losses for a wider range of the solar spectrum. So far, multijunction
cells with up to six sub-cells have been realized. As an example, we show in Fig. 8.43
the schematic layer sequence of a triple-junction solar cell. Such cells have reached
efficiencies of up to 40% in the laboratory, but require expensive deposition methods
such as MOCVD or MBE for the realization of defect-free epitaxial growth of the
different sublayers.
164 8 Photovoltaics
8.6.3 Impurity-Band Photovoltaics (Optical Up-Conversion)
Similar to optical down-conversion, it is in principle possible to achieve an optical

excitation in the absorber by combining two low energy photons with energies smaller
than E gap . At the relatively low intensity levels of unconcentrated sunlight, this
usually requires the presence of a defect band in the band gap of the absorber to serve
as an intermediate electronic state. However, optical up-conversion competes directly
with the very efficient recombination of excited carriers via the same intermediate
defect band. Thus, no convincing examples of up-conversion in real solar cells have
been reported so far.
8.6.4 Impact Ionization (Carrier Multiplication)
This process corresponds to the inverse process of Auger recombination: an opti-

cally excited hot electron with E kin > E gap creates an additional electron-hole pair
by Coulomb interaction with other carriers. For photons with energy >2E gap , this
would result in a quantum efficiency for the generation of electron-hole pairs larger
than one. Indeed, carrier multiplication has been reported for optical absorption in
semiconductor quantum dots, where due to the small volume Coulomb interaction
between carriers is very strong. To make use of carrier multiplication in photo-
voltaics, however, the extraction of multiply generated carriers has to be faster than
the competing Auger recombination. Again, no convincing evidence for a significant
improvement of solar cell efficiencies by this process has been demonstrated so far
(Fig. 8.44).
Fig. 8.44 Multiple

generation of optically
excited carriers via the
absorption of high-energy
photons followed by impact
ionization
References 165
8.7 Energy Payback Times of Solar Cells
An important aspect that has been and needs to be discussed and considered in view
of a large scale global use of solar cells (as well as for all other renewable energy
technologies) for a sustainable future energy supply is their energy payback time and
the related lifetime energy harvesting factor. This conceptually very simple question
turns out to be very complicated and controversial when applied to real products. For
solar cells, one has to take into account on the one side the energy costs for mining raw
materials such as Cu, Si, Ga or In, the energy spent for the construction of production
facilities, the energy required for the solar cell production itself, and the energetic
deployment and recycling costs. On the other side, the energy harvested by a solar cell
during its lifetime needs to be estimated, which depends on the solar cell efficiency,
their placement in the environment, their lifetime, maintenance conditions, etc.
As of today, most solar cell technologies are estimated to have energy payback
times between one and a few years and energy harvesting factors around 10. How-
ever, the current use of photovoltaics is less driven by physical considerations than
much more by economic boundary conditions, involving the costs and long-term
environmental aspects of competing energy sources and the resources needed for
solar cell production. Therefore, we refrain from a further discussion of this delicate
and indeed very complicated aspect.
Further Reading
• Partially commercial solar irradiance and weather data service by Solargis: solar-
gis.com. Accessed 29 Aug 2022
• Development of solar cell efficiency: www.nrel.gov/pv/cell-efficiency.html.
Accessed 28 Aug 2022
• Sugathan, V., John, E., Sudhakar, K.: Recent improvements in dye sensitized solar
cells: a review. Renew. Sustain. Energy Rev. 52, 54–64 (2015)
• Wang, R., et al.: A review of perovskites solar cell stability. Adv. Funct. Mater.
29, 1808843 (2019)
References
1. BP: Statistical Review of World Energy, 70 (2021)
2. Poortmans, J., Arkhipow, V. (eds.): Thin Film Solar Cells. Wiley, Chichester (2006)
3. Chapin, D.M., Fuller, C.S., Pearson, G.L.: A new silicon p-n junction photocell for converting
solar radiation into electrical power. J. Appl. Phys. 25, 676 (1954)
4. Schröder, D.K.: Carrier lifetimes in silicon. IEEE Trans. Electron Devices 44(1), 160–170
(1997)
5. Shockley, W., Queisser, H.J.: Detailed balance limit of efficiency of p-n junction solar cells. J.
Appl. Phys. 32, 510 (1961)
6. Koynov, S., et al.: Appl. Phys. Lett. 88, 203107 (2006)
Thermoelectrics
9
Abstract
The last chapter deals with the field of thermoelectrics. As the physical basis of
these devices the Seebeck and Peltier effect are explained as well as the implemen-
tation in thermoelements and thermoelectric generators (TEGs). The requirements
of efficient thermoelectric conversion are presented together with state-of-the-art
materials for these. The chapter concludes with the possible fields of application.
9.1 Basic Physics of Thermoelectricity
Thermoelectricity is a physical process allowing the direct transformation of heat

into electricity without any moving parts. Therefore, it is highly reliable and basically
maintenance-free, so that thermoelectric generators were used early on for deep space
missions such as the Voyager spacecrafts launched in 1977 [1]. Their supply with
electrical energy is based on thermoelectricity using radioactive heat sources. Since
waste heat is a by-product of many energy-conversion processes, thermoelectric
converters could have a wide range of applications also in the context of renewable
energy. The driving force behind thermoelectricity is a heat flow caused by a thermal
gradient in a conductive material. The physical basis for this is the Seebeck effect
(also referred to as thermopower) in electrical conductors such as metals and doped
semiconductors. Assume a conductive bar of length d extending in x-direction, with
two different temperatures T1 and T2 at both ends (cf. Fig. 9.1). If a volt-meter is
attached to both ends, it will measure a thermo-voltage Uth caused by different
electrical potentials φ1 and φ2 at the cold and hot end. The Seebeck effect now states
that the gradient of the electrical potential for zero electrical current through the
bar is proportional to the temperature gradient across the bar, with a proportionality
factor α, the so-called Seebeck coefficient.
E = −grad φ = α grad T

168 9 Thermoelectrics
Fig. 9.1 The Seebeck effect

for a linear conducting bar.
T2 > T1 are the absolute
temperatures of the hot and
cold ends, giving rise to the
thermo-voltage Uth for zero
current flow, j = 0
Fig. 9.2 Schematic sketch of

the equilibrium
Fermi-function f 0 for two
different temperatures T1 and
T2 > T1 . The negative
derivative of f 0 with respect
to the energy E visualizes the
thermal excitation of
electrons around the Fermi
energy E F compared to
T = 0. Below E F , electrons
are missing and form a
distribution of thermally
excited hole-like states,
whereas excited electron-like
states appear above E F with
increasing temperature
In the simple 1D-case shown in Fig. 9.1 above and for homogeneous materials,
this can also be approximated by the macroscopic relations:
T2 − T1
grad T = ; Uth = φ2 − φ1
d
Uth = α(T2 − T1 ) for j=0; [α]=V/K (9.1.1)
The emergence of a thermopower can be derived from the differences of the

electronic Fermi-function at the hot and cold end, as sketched in Fig. 9.2.
−1
E − EF
f 0 (E) = exp +1
kT
To understand how differences in the distribution of hole- and electron-like states

lead to the Seebeck effect, we note that
9.1 Basic Physics of Thermoelectricity 169
Fig. 9.3 Schematic drawing of symmetric and asymmetric distributions of electron- and hole-like
states in the vicinity of the Fermi energy of a metal. The energy (E − E F ) · l(E) transported per
electron or hole is indicated by the dashed black lines, the red curves indicate a symmetric (α = 0,
left), an asymmetric, electron-like (α < 0, middle), and an asymmetric, hole-like character (α > 0,
right)
• Thermally excited electrons around E F have an excess kinetic energy of ≈ E −

EF
• For a finite Seebeck coefficient α = 0, an asymmetry between electron-like and
hole-like states around E F is necessary.
– In semiconductors, this can be easily established by doping, giving rise to a
large difference in the equilibrium densities of electrons and holes for a given
temperature.
– In metals, more subtle differences in the densities of state, the effective masses
or the mean free paths l(E) for electron- and hole-like states are responsible
for non-zero Seebeck coefficients. For example, the characteristic transported
thermal energy per electron or hole is proportional to (E − E F ) · l(E) (cf.
Fig. 9.3).
A simplified qualitative argument for the appearance of a thermo-voltage under

the presence of a thermal gradient is that in electron-like metals electrons at the hot
end have a higher kinetic energy due to a higher thermal velocity than electrons at
the cold end. Thus, there is an overall net movement of electrons from the hot to the
cold end. As a consequence the cold end is charged negatively, while the hot end
is charged positively. The resulting electric field counteracts the further diffusion of
electrons from hot to cold, until a steady state with a constant thermo-voltage Uth is
established.
The situation in an n-type semiconductor is described in Fig. 9.4. Due to n-type
doping, the Fermi level is closer to the conduction than to the valence band edge,
making electrons the majority carriers. At the cold end, the density n of thermally
excited electrons in the conduction band is smaller than at the hot end. This difference
in electron density again causes a diffusion current of electrons from hot to cold. The
resulting electrical field drives a counteracting drift current of electrons from cold to
hot. Steady state and a constant thermo-voltage are reached, when drift and diffusion
current cancel each other. An analogous picture holds for p-type semiconductors
with thermally excited holes.
Fig. 9.4 N-type

semiconductor with a
thermal gradient. Sketched
are the different Fermi
distributions and the
resulting electron densities in
the conduction band. See text
for further details
For both, metals and semiconductors, the dominant type of carriers (electrons or
holes) present is indicated by the sign of the thermo-voltage at the cold end. This can
be used to quickly determine the doping type of a semiconductor. Furthermore, the
full steady-state thermo-voltage is obtained when drift and diffusion currents cancel
each other exactly, i.e. for zero internal current density, j = 0.
For j = 0 the drift current generated by the internal electric field is opposite and
equal to the diffusion current generated by T .
Thermocouples: To measure the Seebeck coefficient α of a sample of material

B, one needs electrical contacts usually consisting of a different material A.
As indicated below this also generates a thermo-voltage. To avoid further
complications due to additional materials in the voltmeter, the latter should be
maintained at a constant temperature T0 .
For the thermo-couple arrangement shown above, the expected thermopower can
be calculated using the Seebeck relation by integration over the different segments
of the closed conductor loop:
9.1 Basic Physics of Thermoelectricity 171
Fig. 9.5 The Peltier effect.

See text for details
T1 T2 T0
Uth = − E · ds = − α gradT · ds = − α A dT − α B dT − α A dT
T0 T1 T2
T2
= (α A − α B ) dT
T1
≈ (α A − α B )(T2 − T1 ) for small T
Therefore, with such a thermo-couple the Seebeck coefficient of test material B can
be calculated from the applied temperatures relative to the Seebeck coefficient of a
reference material A. Here, one has to further consider the following aspects:
• In general α = α(T ) is strongly temperature dependent.

• For the reference material, a highly conductive metal with a small Seebeck coef-
ficient such as Pt should be used.
• For non-zero current density j = 0, power can be extracted from a thermo-couple
(cf. (9.2.1)). This also reduces the thermo-voltage Uth .
Also the inverse of the thermopower can be observed and is known as the Peltier
effect: here a current density j due to an externally applied voltage U creates a heat
flow density q̇ which in turn generates a temperature gradient between the electrical
contacts (Fig. 9.5). The direction of the temperature gradient depends on whether the
conductivity in the Peltier element is electron- or hole-like. The Peltier coefficient
π̃ = αT is directly related to the Seebeck coefficient α and the absolute temperature
T.
A quantitative treatment of the Seebeck and Peltier effect can be performed based
on the Boltzmann equation for the Fermi-distribution f (r, k, t) in phase space:

∂ f (r, k) df ∂f
= = + k̇ gradk f + ṙ gradr f
∂t scattering dt ∂t

∂ f ∂k ∂ f ∂r
0 in stationary state ∂k ∂t ∂r ∂t
Here the last term on the right side allows the inclusion of spatial gradients of
temperature and their effect on the Fermi distribution f . Solving this equation assum-
ing different scattering mechanisms gives the following formulae for the Seebeck-
coefficient and its T -dependence in metals and semiconductors (with typical orders
of magnitude indicated at the right side):
π2 kB kB T
αmet = − for electron-like metals (≈ µV/K)
2 e EF

k B EC − E F 5
αn = − + +r for n-type semiconductors (≈ m V/K)
e kB T 2

kB E F − EV 5
αn = + + +r for p-type semiconductors (≈ m V/K
e kB T 2
Here, r is a scattering factor which depends on the dominant scattering mechanism of

the majority carriers (e.g. r = −1/2 for deformation potential scattering by acoustic
phonons or r = 3/2 for scattering at ionized impurities).
9.2 Thermoelectric Generators (TEGs)
Thermoelectric generators are devices to harvest electrical energy with the help of
suitably assembled thermoelectric elements. As shown in Fig. 9.6, they are based on
pairs of n- and p-type semiconductors (called n- and p-“legs”) connected parallel
and/or in series to larger assemblies, depending on the desired output voltage or
current.
Obvious requirements for suitable semiconductor materials for the legs are:
Fig. 9.6 Schematic layout of a thermoelectric generator. “Legs” of alternating n- and p-type semi-
conducting bars are connected in series by ohmic metal contacts. The electrical structure is encased
by a suitable ceramic to make contact to the hot and cold heat reservoirs, setting up a constant
thermal gradient through the legs. Arrows indicate the diffusion direction of the majority carriers
in the legs
9.2 Thermoelectric Generators (TEGs) 173
• a large Seebeck-coefficient α ([α]=V K−1 )

• a high electrical conductivity σ = enμ ([σ ]=A V−1 m−1 )
• a low thermal conductivity κ ([κ]= W K−1 m−1 )
• sufficient structural and electrical stability in the temperature range applied.
To compare the suitability of different thermoelectric materials one uses the thermo-
electric figure of merit Z T :
α2 σ
Z ·T ≡ ·T ([Z ] = K−1 ; [Z · T ] = 1) (9.2.1)
κ
This figure of merit determines the maximum theoretical efficiency of a TEG:

√
TH − TC 1 + Z Tave − 1
ηmax = ·√ TC
(9.2.2)

1 + Z Tave + TH
TH
ηCar not
with the average temperature Tave = 21 (TH + TC ). TH and TC are the absolute tem-
peratures of the hot and cold reservoirs. Note that:
ηmax → ηCar not for Z ·T →∞
The formula above is only valid for TH − TC << TC . Otherwise, since Z = Z (T )

is strongly dependent on temperature, one has to obtain ηmax by integration:
√
TH − TC 1 TH 1 + Z (T ) · T − 1
ηmax = √ dT
TH TH − TC TC 1 + Z (T ) · T + TTCH
For energy conversion in TEGs, only materials with Z T ≥ 1 are commercially viable.
This can be seen from the maximum theoretical efficiencies plotted in Fig. 9.7 and
listed in Table 9.1.
Fig. 9.7 Dependence of the

theoretical TEG efficiency
on the figure of merit and the
temperature difference
between hot and cold heat
reservoirs. The cold reservoir
was assumed to be at room
temperature, 300 K
Table 9.1 Theoretical efficiencies of TEGs for different TH and TC = 300 K in comparison to the
Carnot efficiency
TH ηCar not ηmax ηmax ηmax
(Z T = 0.1) (Z T = 1) (Z T = 10)
500 K 0.4 0.01 0.08 0.24
1000 K 0.7 0.03 0.17 0.45
In addition to T , Z = (α 2 σ )/κ also depends strongly on the carrier densities n or

p. In particular, we can approximate for n-type materials:
kB T 2
α∝ with EF = (3π 2 n)2/3 in metals
EF 2m ∗

EC − E F NC
α∝ = ln in semiconductors (n-type)
kB T n
σ = enμn
κ = κ phonon + κelectr on with κelectr on = Lσ T
The last equation is the Wiedemann-Franz law in metals, with the Lorentz factor L
for free electrons:

π2 kB 2
L= = 2.4 × 10−8 J2 K−2 C−2
3 e
Finally, the phonon contribution to the thermal conductivity is given by:
κ Phonon ∝ T 3 for T <

D ; κ Phonon ∝ 1/T for T ≥
D (9.2.3)
with
D the Debye-temperature for acoustic phonons.
A qualitative view of the dependence of the different contributions to the quantity
Z = (α 2 σ )/κ on charge density n in n-type semiconductors is shown in Fig. 9.8.
The Seebeck coefficient α decreases logarithmically with the carrier concentration,
as the Fermi-level gradually approaches the conduction band edge. The electrical
conductivity σ is negligible on a linear scale for small carrier concentrations. Above
about n = 1020 cm−3 , σ approaches the minimum metallic conductivity of about
104 −1 cm−1 in the metallic regime. The thermal conductivity κ is independent of
the carrier concentration and dominated by the phonon-contribution κ phonon for car-
rier concentrations below 1020 cm−3 and then increases according to the Wiedemann-
Franz law proportional to σ by the electronic contribution κelectr on . As a result, Z
typically shows a broad maximum for carrier concentrations around n = 1020 cm−3 ,
corresponding to degenerate doping levels in semiconductors.
So far, we have only discussed TEGs under open-circuit conditions. To extract
electrical power from a TEG, however, a finite current has to flow through a load
resistor. This current also flows through the TEG which, due to the highly doped legs
and the ohmic metal contacts, can be treated like an internal resistor, RT E G . This is
Fig. 9.8 Qualitative semi-logarithmic view of the different contributions to Z = (α 2 σ )/κ as a

function of the carrier density n in an n-type semiconductor. See text for details. Since the thermal
conductivity is basically constant, a similar broad maximum around n = 1020 cm−3 as for Z is also
obtained for its numerator, α 2 σ , which is called the power factor
shown in Fig. 9.10. For a TEG with N p/n-legs in series and an effective Seebeck
coefficient α operated with a temperature difference T , the open-circuit voltage is
(Fig. 9.9):
Uoc = N · α · T
The voltage U dropping across the load resistor is then:
U = Rload · I = Uoc − RT E G · I
The maximum output power is reached for RT E G = Rload :
2
Uoc
Pmax = Um · Im =
4RT E G
To quantify the performance of TEGs, in addition to the current-voltage characteris-

tics a direct measurement of the figure of merit Z T can be achieved by the so-called
Harman method:
1. A current I is passed through the TEG and produces via the Peltier effect a heat
flow Q̇ = π̃ · I = α · T · I
2. In steady state this causes a temperature difference T , given by Q̇ = αT I =
κT .
3. The voltage Ustat across the TEG in steady state is measured: (RT E G ∝ 1/σ ):
Ustat = Uoc − RT E G I
Fig. 9.9 Basic structure of a thermoelectric generator with N p/n-legs in series and a load resistor
Rload . The corresponding equivalent circuit is shown on the right
Fig. 9.10 Current-voltage

characteristics of a TEG.
Since the TEG acts like a
resistor, this leads to a linear
characteristics with slope
RT−1E G . The corresponding
fill factor is much smaller
compared to the fill factor of
solar cells with diode
characteristics
Fig. 9.11 Illustration of the

Harman method
4. The current I is stopped abruptly and immediately after Uoc = αT is measured.
5. Z T can then be calculated from Uoc and Ustat (Fig. 9.11).
α2 σ Uoc
ZT = T = (9.2.4)
κ Ustat − Uoc
Table 9.2 Applications for TEGs

With fuel heating With waste heat
Deep space missions-RTG Waste burning facilities
Pace makers Waste heat in cars
Light houses, transmitter stations Geothermal plants
Corrosion protection Miniature converters (watches, sensors,..)
Mobile power supplies Solar climate regulation (venting)
Electrical backup systems for solar plants Space missions near the sun
Fig. 9.12 Historical

development of the figure of
merit Z T for different
materials and nanostructures
Some additional remarks concerning TEGs:
• For TH ≥ 1200K, thermo-photovoltaics becomes more efficient than thermo-

electrics [2]. For TH ≥ 2500K, photovoltaic energy conversion is preferable.
• The efficiencies of current commercial TEGs are in the range of 1–10% [3].
• Some examples for applications of TEGs are listed in Table 9.2.
Current research to develop systems with high Z T values is focussed on complex

compound materials and the specific use of nanostructures. The aim is to decrease
the thermal conductivity of thermoelectric systems, while keeping the Seebeck coef-
ficient and the electrical conductivity high or even increasing them. A detailed treat-
ment of different approaches would go beyond the scope of this book, so we just
mention a few headlines (Fig. 9.12).
Early TEGs were based on crystalline SiGe alloys, since these could be produced
as large crystals with good quality and could be equally well doped n- and p-type for
the different TEG legs. SiGe reached Z T = 1 for a temperature of 1200 K and was
used primarily in early deep space missions. Most other thermoelectric materials also
used compound semiconductors with heavy atoms such as PbTe or Bi2Te3, because
Fig. 9.13 Energy band diagram of a semiconductor superlattice. The energetic filtering of electrons
or holes by the periodic band gap modulation plus the multiple scattering of phonons by the acoustic
impedance mismatch between the two constituent materials help to increase ZT for such structures
they combined a low thermal conductivity due to alloy scattering and the heavy atoms
involved with reasonable Seebeck coefficients and electrical conductivities. Further
advances were made around 1990–2000 with novel materials beyond binary alloys,
especially chlathrites and skudderudites featuring even lower thermal conductivities
due to their large crystal unit cells and so-called “rattling atoms”, i.e. heavy atoms
vibrating with relatively low energies yet large amplitudes, thereby providing effi-
cient scattering centers for other lattice phonons (“phonon glass-electron crystal”
Behaviour). With this approach Z T -values exceeding 2 were realized, e.g. for the
compound AgPb19SbTe20. Even higher Z T -values exceeding 2.5 were reached for
Bi2Te3/Sb2Te3 superlattices. As sketched in Fig. 9.13, such superlattices combined
additional multiple phonon scattering at the layer interfaces due to a sudden change
in the acoustic impedance ρvsound (density times sound velocity) of the different
materials, and the energy filtering of electrons passing through the superlattice con-
duction band. A negative aspect of the use of such compound superlattices are their
high production costs.
Even higher Z T -values above 3 were reported for thermoelectric materials based
on compound semiconductor nanostructures such as quantum dots or quantum wires.
Here, improvements of Z T have been related to an increased surface or interface
scattering of phonons with bulk mean free paths exceeding the nanostructure size.
In addition, positive effects of the altered electronic density of states in 1D- or 0D-
confined systems on the Seebeck coefficient have been postulated.
Further Reading
• Rowe, D.M. (ed.): Thermoelectrics Handbook: Macro to Nano. CRC Press, Boca
Raton (2017).
• Radioisotope TEG with plutonium-238 on the Cassini and other space probes:
solarsystem.nasa.gov/missions/cassini. Accessed from 29 July 2022.
References 179
• Wei, Jiangtao, et al.: Review of current high-ZT thermoelectric materials. J. Mat.

Sc. 55, 12642–12704 (2020).
References
1. National Aeronautics and Space Administration, Voyager spacecraft website: https://voyager.
jpl.nasa.gov/mission/spacecraft/. Accessed from 31 July 2022
2. LaPotin, A., et al.: Thermophotovoltaic efficiency of 40%. Nature 604, 288 (2022)
3. Champier, Daniel: Thermoelectric generators: a review of applications. Energy Convers. Manag.
140, 167–181 (2017)
Knowledge Check
Problems of Chap. 1
1.1 What is the unit of energy expressed in terms of the SI units kg, m, and s?
1.2 What is the meaning of the abbreviation “BTU”?
1.3 Show that the kinetic energy of a moving mass m is 1/2 · mv 2 in classical
mechanics.
1.4 According to the Noether theorem, which symmetry leads to energy conserva-
tion?
1.5 What is the basic metabolic rate of an average human being in rest?
(A) 30 W
(B) 80 W
(C) 150 W
(D) 280 W
Problems of Chap. 2
2.1 Provide the formula for the energy density E/V of a plate capacitor with capac-
itance C charged up to a voltage U .
2.2 Which of the following forms of energy storage has the lowest volumetric energy
density E/V ?
(A) a commercially available supercapacitor

(B) a Li-ion battery

182 Knowledge Check
(C) water with a height difference of 100 m

(D) a high temperature phase change salt
2.3 According to the Nernst equation, how does the voltage U developing at a solid
electrode immersed in a liquid electrolyte depend on the concentration c of dissolved
electrode ions?
2.4 Give the overall reaction equation for the nuclear fission of 235 U.
Problems of Chap. 3
3.1 What is the total power emitted by an ideal black body with a temperature
1000 K and a surface area of 1 m2 ? (Stefan–Boltzmann constant: σ = 5.67 × 10−8
in SI units)
3.2 What is the overall fusion reaction in our sun, considering all conservation laws?
3.3 What is the effective surface temperature of the Sun? Describe two ways how
this temperature can be determined here on Earth.
3.4 What would be the average surface temperature of the Earth without the green-
house effect?
1. Give the equation based on the balance between absorbed and emitted energy
using an average reflectivity of 30% for the incoming radiation.
2. Give the resulting value for the average surface temperature.
3.5 What is the current average value of the Solar Constant (SC)? How large (in %)
are its seasonal changes?
(A) ±1%
(B) ±5%
(C) ±10%
3.6 If all of the radiation of the Sun impinging on the Earth could be captured and
used, how much time of sunshine would be necessary to satisfy the current energy
need of mankind?
(A) 1 min
(B) 1h
(C) 1 day
(D) 1 week
Knowledge Check 183
Problems of Chap. 4
4.1 What is the dispersion relation v(λ) of deep water waves? Is there a difference
to the relation for shallow water waves?
4.2 Give the formula for the gradient of the gravitational acceleration caused by the
sun on the two sides of the earth pointing towards and away from the sun.
4.3 What is the main qualitative difference between solar and lunar tides when
calculating gradients of gravitational and centrifugal forces?
4.4 Why can local tidal waves be much higher than estimated under the assumption
of a global ocean?
Problems of Chap. 5
5.1 Why is there a low pressure region at the intertropical conversion zone (ITCZ)?
5.2 What is the areal power density P/A of wind with velocity v?
5.3 Draw a diagram of the variation of pressure p(x) when air passes through the
rotor plane of a wind engine at x = 0 (the pressure far away from the wind engine
is p0 ).
5.4 Draw the Betz efficiency of a wind engine as a suitable function of the wind
speeds before (v1 ) and after the engine (v2 ) and give the limiting values (maximum
efficiency, efficiency for v2 = 0 and point of zero efficiency).
5.5 Draw the vector diagram defining the velocity of attack v A of a lift type rotor.
5.6 What is the tip speed ratio λs ?
5.7 A turbine wing has the glide number G = 100 and experiences a drag force of
FW = 1000 N. What is the corresponding lift force FA ?
5.8 What are the main loss mechanisms of a lift type 3-blade rotor at
(a) very low tip speed ratio λs ?

(b) very high tip speed ratio λs ?
5.9 Wind with a velocity of 50 km h−1 flows through an area A. What is the cor-
responding power density P/A, using a density of ρ = 1.2 kg m−3 for the flowing
air?
184 Knowledge Check
5.10 Wind with the velocity v1 = 50 km h−1 enters an ideal wind engine with an
area of A = 10000 m2 and exits with a velocity of 25 km h−1 .
1. What is the Betz efficiency of this wind engine?

2. How does this efficiency change when the area of the wind engine is reduced to
1000 m2 ?
5.11 A blade segment has drag coefficient cW = 0.3, a perpendicular area A⊥ =

1 m2 , a lift coefficient c A = 0.9 and a parallel area A|| = 10 m2 . The velocity of
attack is v A = 10m s−1 . What is the glide number G of this segment?
5.12 What is the realistic maximum efficiency of a modern 3-blade wind engine?
At approximately which tip speed ratio λ S is this efficiency achieved?
5.13 Give the empirical formula for the average wind velocity as a function of height
h above ground.
5.14 How does the breadth b(r ) parallel to the velocity of attack of an optimized
rotor blade scale with the tip speed ratio λ S and the distance r from the center of the
wind engine?
Problems of Chap. 6
6.1 How large is the average geothermal gradient in the Earth’s crust?
6.2 At a given high enthalpy site the local heat flux to the surface is 300 mW m−2 and
the local thermal conductivity is 3 W/(Km). How large is then the thermal gradient
driving this heat flux?
6.3 What is the definition of the efficiency η H P of a heat pump and what is its
thermodynamic limit?
6.4 A heat pump operates between the temperature levels 100 ◦ C and 25 ◦ C. What
is the maximum efficiency of this heat pump?
6.5 Sketch the optimized wavelength-dependence of the absorbance for a non-

concentrating solar-thermal panel.
6.6 What is the maximum achievable concentration factor Cmax of direct sunlight
using a linear concentrator?
6.7 A solar tower is illuminated by a heliostat field with a concentration factor of

C = 1000. What is the maximum achievable absorber temperature?
Knowledge Check 185
Problems of Chap. 7
7.1 What is the typical carbon fixation capacity of so-called C4 plants in mg CO2
per 100 cm2 leave area and hour?
(A) 10 mg
(B) 30 mg
(C) 70 mg
(D) 120 mg
7.2 What is the current overall efficiency of global photosynthesis (energy stored in
annually produced biomass vs. integrated incoming solar irradiation)?
(A) 0.01%
(B) 0.1 %
(C) 1%
(D) >1%
7.3 What are the two main intermediate energy storage systems in photosynthesis
and to which photosystem are they linked?
7.4 Sketch the primary and secondary reactions of photosynthesis with the input and
output molecules and the intermediate energy storage compounds.
7.5 Describe the role of C-C double bonds in chlorophyll for the absorption of light.
7.6 How many chlorophyll molecules are typically present in the antenna complexes
of the photosystems?
7.7 1. What is the distance dependence of the Förster transfer rate between neigh-
boring chlorophyll molecules in an antenna complex?
2. What does this distance dependence say about the type of interaction between
the molecules?
7.8 What is synthesis gas or “Syn-gas”?
Problems of Chap. 8
8.1 Draw the “Feynman diagram” of photovoltaics.
8.2 The absorber layer of a solar cell has an absorption coefficient of α = 104 cm−1
at a wavelength of 600 nm. What is the required absorber thickness d to absorb 95%
of the incoming photons of that wavelength?
186 Knowledge Check
8.3 Describe the time derivative of the minority carrier density n in a p-type absorber
by the recombination and generation rates R and G, respectively.
8.4 For the case of bimolecular recombination, what is the dependence of the photo-
generated excess carrier density n on the generation rate G?
8.5 Sketch the energy band diagram of a Si diffusion cell including the BSF and the
built-in voltage Ubi .
8.6 Explain why the efficiency of a solar cell goes to zero in the limit of small and
of large band gaps of the absorber layer.
8.7 Which combination of band gaps in tandem solar cells is optimum for AM = 1.5
illumination conditions?
8.8 What is the time and spatial dependence E(x, t) of the electric field of a photon
with frequency ν in a matter with complex index of refraction (n + iκ)?
8.9 What is the average generation rate G in a semiconductor of thickness d with

an absorption coefficient α and a reflectivity R under illumination with an incoming
photon flux
0 ?
8.10 What is the closest value for the “Shockley–Queisser” limit for the efficiency
of an ideal solar cell under illumination with unconcentrated sun light?
(A) 12%
(B) 19%
(C) 30%
(D) 37%
(E) 42%
8.11 Sketch the equivalent electrical circuit of a real solar cell.
8.12 What is the formula for the open-circuit voltage Uoc of an ideal solar cell?
8.13 What is formula for the I -U characteristics of a real solar cell under illumina-
tion, including series and shunt resistances Rs and R p ?
Problems of Chap. 9
9.1 What is the figure of merit for thermoelectric materials?
9.2 According to the Boltzmann approximation, what is the Seebeck coefficient of

an electron-like semiconductor?
Knowledge Check 187
9.3 Sketch and describe the current-voltage characteristics of a thermoelectric power

converter.
9.4 What is the figure of merit ZT of a thermoelectric material with a Seebeck

coefficient α = 0.1 V/K, an electrical conductivity of σ = 1 A/(Vm) and a thermal
conductivity κ = 10 W/(km) operated at 100 ◦ C?
9.5 What is the Seebeck coefficient of an electron-like metal? Include at least all
terms important for the correct unit of αmet and the correct sign.
Index
A production, 101
Absorbance, 127 pyrolysis, 115
Absorber, 124 Black body, 30, 37, 123, 128
Absorption, 124 Boltzmann constant, 128
coefficient, 125 Boson, 122
depth, 125 Brillouin zone, 126
Acceptor, 144
Adenosine diphosphate/Adenosine triphos- C
phate (ADP/ATP), 103 C3 /C4 -plants, 105
Adiabatic coefficient, 13 Calvin cycle, 102
Adiabatic compression, 13 Capacitance, 17
Adiabaticity, 13 Capture coefficient, 135
Air mass, 38, 94 Carbon capture and storage (CCS), 42
Anergy, 5 Carrier density, 133
Angle of attack, 70 Centrifugal force, 52
Antenna complex, 103 Chlorophyll, 103, 105, 110
Antenna protein, 110 absorption coefficient, 110
Anti-reflection layer, 127 Chloroplast, 103
Aquifer, 88 Compressed air energy storage (CAES), 12
Asthenosphere, 84 Conduction band, 126, 131
Astronomical unit, 30 Conjugated coordinates, 1
Auger coefficient, 138 Convection cells, 35
Convection roll, 60
B Coriolis force, 59
Back surface field, 146 Coulomb interaction, 138
Band gap, 126 Covalent bond, 19
Barometric height formula, 62 Cytochrome bf, 107
Battery, 21, 24
Bernoulli equation, 45, 66 D
Binding energy, 20, 24 Dangling bonds, 138
Biomass, 99, 114 Darcy’s law, 88
energy content, 101 Dark current, 153
fermentation, 115 Deep water waves, 45, 46
gasification, 115 Detailed balance, 136

190 Index
Deuteron, 31 Energy overlap factor, 112

Dexter transfer, 110 Energy payback time, 165
Differential accretion, 83 Equilibrium occupation, 123
Diffuse radiation, 94 Exciton, 125
Diffusion length, 153 Exergy, 5
Dipole moment, 124 Extraction, 139
Dipole selection rule, 125
Dispersion relation, 47 F
Down-conversion, 160 Fermi energy, 155
Drag coefficient, 68 Fermi-function, 168
Drag force, 68 Fermi-level, 136
D-T-fusion, 27 Ferrel cell, 61
Ferrodoxin, 103
E Fill factor, 157
Earth, 36 Fluorescence rate, 112
core, 84 Flywheel, 15
crust, 84 Force
emission spectrum, 38 conservative, 2
energy fluxes, 40 Förster transfer, 112
mantle, 83 Frank-Condon principle, 130
orbit, 39 Fröhlich interaction, 132
reflectivity, 36 Fuel cell, 24
structure, 83
surface temperature, 37 G
Ecliptic, 39 Galvanic voltage, 22
Effective medium, 127 Gamma radiation, 35
Efficiency Geo-engineering, 42
APS, 117 Geothermal flux, 84
Betz, 65 Geothermal gradient, 84
Carnot, 5, 159 Glide number, 71
heat engine, 6 Gravitation, 11, 33
heat pump, 91 Gravitational constant, 34
photosynthesis, 101
Schmitz, 79
solar cell, 143, 144, 157 H
TEG, 173 Hadley cell, 60
Electric field, 17 Harman method, 175
Electrochemical double layer, 22 Heat capacity, 13
Electrolyzer, 117 Heat pump, 90
Electromagnetic waves, 122 Helmholtz planes, 22
Electron-phonon coupling, 92, 134 Hetero-junction, 141
Emission, 128 Highest occupied molecular orbit (HOMO),
Emissivity, 30, 94 110, 130, 150
Energy, 11 Hot dry rock, 88
chemical, 19 Hot enthalpy site, 87
electrochemical, 21 Hydraulic conductivity, 88
electromagnetic, 17
kinetic, 14, 61 I
mechanical, 11 Ideal gas, 12
nuclear, 24 equation, 33
potential, 11 Impact ionization, 164
tidal, 50 Incidence angle, 94
wave, 16, 45 Intensity, 124
wind, 59 Intertropical conversion zone (ITCZ), 61
Index 191
Invariant operations, 3 PERL cell, 146

Perovskites, 152
K Phase change, 18
Kelvin-Helmholtz instability, 45 Phase velocity, 46
Kinetic energy, 3 Phonon, 89, 92, 126, 130, 140
Kirchhoff law, 95 replica, 130
Photo current, 154
Photon
L
absorption, 93
Lambert-Beer law, 93, 125
angular momentum, 122
λ/4 coating, 127
density of states, 123
λ/4-resonance, 54
dispersion relation, 123
Latent heat, 18
flux, 124
Lift coefficient, 70
generation rate, 127
Lift force, 70
momentum, 122
Light harvesting complex, 110
polarization, 122
Li-ion battery, 23
reflection, 93, 127
Lorentz factor, 174
scattering, 93
Low enthalpy site, 88
spectral distribution, 123
Lowest unoccupied molecular orbit (LUMO),
spectrum, 123
110, 130, 150
transmission, 93
Lumen, 108
Photosynthesis, 99
Luminescence, 128
an-oxygeneous, 102
artificial, 116
M oxygen-generating, 102
Magnetic field, 17 red drop, 105
Maunder minimum, 40 Z-scheme, 103
Maximum power point (MPP), 154 Photosystem I/II, 103, 105, 106
Mie scattering, 93 Photovoltaics, 119
Mobility, 133 Feynman diagram, 121
Moment of inertia, 15 installed power, 119
Momentum, 1 Plasma, 26
Plate capacitor, 17
N p/n-junction, 140
Nernst equation, 22 Polar cell, 61
Neutrino, 31 Polar front, 61
Neutron, 24 Potential energy, 3
Nicotineamid adenine dinucleotid phosphate pp-transition, 110
(NADP), 103 Profile polar plot, 70
Noether theorem, 3 Proton, 24, 31
Nuclear power plant, 25 Proton fusion, 31
Nucleon, 24 Pumped hydro-storage, 12
Pyrrole ring, 103
O
Open circuit voltage, 154, 175 Q
Orbital, 19 Quasi-Fermi-level, 155
Osmosis power, 56
power density, 56 R
Osmotic pressure, 56 Rayleigh scattering, 93
Ozone, 38 Reaction center, 105
Recombination, 132
P Auger, 137
Peltier effect, 171 bimolecular, 137
PERC cell, 147 centers, 134
192 Index
monomolecular, 135 Surface texturing, 127

non-radiative, 134 Syngas, 117
radiative, 128, 133
surface, 138 T
Reflectivity, 36 Tandem cell, 160
Refractive index, 122, 125 Thermalization, 130
Rossby waves, 61 in molecules, 130
Rotor plane, 66, 72 in semiconductors, 131
Thermo-couple, 170
S Thermoelectric figure or merit, 173
Salinity, 56 Thermoelectric generator (TEG), 167
Saturation current, 153 Thermopower, 167
Scattering time, 133 Thermo-voltage, 167
Schottky contact, 147 Thin film, 150
Seebeck coefficient, 167 Thylakoid membrane, 103, 107, 108
Seebeck effect, 167 Tides, 50
Semi-permeable membrane, 56, 117 lunar, 53
Separation, 139 solar, 52
Shallow water waves, 48 Trade winds, 60
Shockley–Queisser limit, 157 Transition probability, 124
Short circuit curr. dens., 154 Transmission, 127
Solar collector Transparent conducting oxide (TCO), 150
concentrating, 96 Tropopause, 61
non-concentrating, 95
Solar constant, 30 U
Solar wind, 35 Uncertainty principle, 1, 123
Solvation shell, 22
Solvent, 22
Space charge region, 146, 153 V
Spectrum, 30 Valence band, 126, 131
Speed ratio, 73 Van-’t-Hoff equation, 56
Spin, 19 Velocity of attack, 72
Staebler–Wronski effect, 149
Standard Model, 31 W
Stefan-Boltzmann constant, 30 Wave, 45
Stefan-Boltzmann law, 94 capillary, 48
Stokes shift, 130 dispersion, 47
Stroma, 108 energy content, 49
Sun, 29 frequency, 46
chromosphere, 35 gravitational, 48
CNO-cycle, 32 power density, 50
core, 35 reflection, 54
corona, 35 tsunami, 49
emission spectrum, 38 velocity, 47, 48
flares, 35 Wave energy converter, 50
luminosity, 29 Wave function, 19
photosphere, 35 Wavevector, 126
protuberances, 35 Westerlies, 61
shell model, 33 Wiedemann-Franz law, 174
spots, 35 Wien law, 37
Supercapacitor, 17 Wind
Super-rotation, 84 circumferential velocity, 71
Surface recombination velocity, 139 deflection, 60
Surface tension, 48 energy, 59
Index 193
energy content, 61 efficiency, 65, 66, 69

friction coefficient, 63 HAWT, 79
jetstream, 60, 61 lift-type, 69
pattern, 59, 60 pitching, 71
power, 66 power, 74
speed, 63 profile losses, 76
speed map, 63 rotors, 68
stall, 70 tip losses, 76
turbine, 63 tip speed ratio, 77
velocity, 62 wake losses, 76
Wind turbine wing profile, 70
acoustic noise, 80 Work, 2, 90
drag-type, 68 World population, 41

Martin Stutzmann, Christoph Csoklich - The Physics of Renewable Energy

Uploaded by

Copyright:

Available Formats

Martin Stutzmann, Christoph Csoklich - The Physics of Renewable Energy

Uploaded by

Document Information

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

Martin Stutzmann, Christoph Csoklich - The Physics of Renewable Energy

Uploaded by

Copyright:

Available Formats

What is the purpose of the Graduate Texts in Physics series?

What is the purpose of the Graduate Texts in Physics series?

What topics does the Graduate Texts in Physics series cover?

What topics does the Graduate Texts in Physics series cover?

Graduate Texts in Physics

The Physics of Renewable

ISSN 1868-4513 ISSN 1868-4521 (electronic)

© Springer Nature Switzerland AG 2022

Munich, Germany Martin Stutzmann

1 Energy—A Brief Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

4 Energy from Waves, Tides and Osmosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

8.2.4 Thermalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

Knowledge Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

ADP Adenosine diphosphate

MPP Maximum power point

Energy is the variable of state of a physical system describing the ability of

© Springer Nature Switzerland AG 2022 1

Fig. 1.1 A point mass m at

1.1 Energy and Work

The change in energy E of a system is given by the work W performed by an external

Depending on the sign of dW , two qualitatively different cases can occur:

1.2 Potential and Kinetic Energy

E pot = E pot (r2 ) − E pot (r1 ) = W1,2 (1.2.2)

with the gradient vector operator ∇:

E = E kin + E pot = const. (1.2.7)

1.3 Noether Theorem and Energy Conservation

The conservation of energy is one of many conservation laws in physics. According

Table 1.1 Symmetries and corresponding conserved physical quantities

Here, we just give a brief sketch of this theorem as applied to an invariance of a

qi (t) → qi (s, t) with qi (0, t) = qi (t) (1.3.2)

If the transformation is a symmetry operation which leaves the system invariant,

1.4 Inner Energy

In general, processes on earth are non-conservative and irreversible due to friction,

The conversion of macroscopic potential or kinetic energy to inner energy by

Fig. 1.2 Schematic energy

For irreversible processes the anergy is lost as an the increase of entropy (d S =

1.5 Quantifying Energy

The derived SI unit for energy is the Joule:

However, depending on the scientific or practical context, a wide variety of other

Table 1.2 Important physical and chemical units of energy

Table 1.3 Important technical units of energy

2.1 Mechanical Potential Energy

© Springer Nature Switzerland AG 2022 11

Fig. 2.1 Illustration of the

As an example, 1 t of water is carried up a mountain to a height of 100 m:

E pot = mgh ≈ 1 × 106 J

This is equivalent to 0.3 kWh/250 kcal/0.04 kgCE and corresponds to an

E/V = 1 × 106 J m−3 (2.1.2)

As a consequence, the energy density in pumped hydro-storage is quite low.

dW = −Fdx = − p Adx = − pdV

For an adiabatic compression of the gas we then have:

with γ , the adiabatic coefficient:

Integration then gives:

This leads to the equation of adiabaticity:

Thus, the amount of energy stored in the compressed gas is:

The adiabatic coefficient of dry air is γ = 1.4. We start from a temperature

E/V = 1.3 kW h m−3 = 5 × 106 J m−3 (2.1.12)

To store 1 GW h we then need a volume of 106 m3 , which corresponds to a

2.2 Kinetic Energy

As an example, consider a car with a mass of m = 1000 kg, driving with a

E pot = E pot (r2 ) − E pot (r1 ) = W1,2 (1.2.2)