MEASUREMENT OF PARITY-VIOLATING ELECTROWEAK

ASYMMETRIES IN −
→
e p SCATTERING AT
0.1 < Q2 < 0.4 (GeV/c)2

A Dissertation

Presented to

The Faculty of the Department of Physics

The College of William and Mary in Virginia

In Partial Fulfillment

Of the Requirements for the Degree of

Doctor of Philosophy

by

Jeffery A. Secrest

2005
 APPROVAL SHEET

This dissertation is submitted in partial fulfillment of

the requirements for the degree of

Doctor of Philosophy

Jeffery A. Secrest

Approved, November 2005

Dr. Allison Lung, Chair

Dr. David S. Armstrong

Dr. John M. Finn

Dr. Jeffrey K. Nelson

Dr. Marc Sher

Dr. Philip G. Roos, University of Maryland

ii
 CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii

LIST OF ILLUSTRATIONS . . . . . . . . . . . . . . . . . . . . . . . xii

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

CHAPTER

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Strange Quarks in the Proton . . . . . . . . . . . . . . . . . . . . 1

1.2 Parity-Violating Electron Scattering . . . . . . . . . . . . . . . . 7

1.3 Overview of the G0 Experiment . . . . . . . . . . . . . . . . . . . 10

2 Physics Formalism . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.1 Structure of the Proton . . . . . . . . . . . . . . . . . . . . . . . . 19

2.2 Electromagnetic Form Factors . . . . . . . . . . . . . . . . . . . . 21

2.3 Weak Neutral Form Factors . . . . . . . . . . . . . . . . . . . . . 27

2.4 GSM , GSE and G0 : Quark Decomposition of Form Factors . . . . . . 28

2.5 Elastic Parity-Violating Electron-Nucleon Scattering . . . . . . . . 31

2.6 Strange Form Factors and Physics

Asymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.7 Survey of Previous Experiments . . . . . . . . . . . . . . . . . . . 34

3 Experimental Apparatus . . . . . . . . . . . . . . . . . . . . . . 39

3.1 The G0 Magnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.2 The G0 Hydrogen Target . . . . . . . . . . . . . . . . . . . . . . . 41

iii
 3.3 The G0 Detector . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.4 The G0 Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.4.1 The G0 Beam Structure . . . . . . . . . . . . . . . . . . . 51

3.5 The CEBAF Injector and the Polarized Source . . . . . . . . . . . 53

3.5.1 The G0 TIGER Laser . . . . . . . . . . . . . . . . . . . . . 53

3.5.2 Optical Elements . . . . . . . . . . . . . . . . . . . . . . . 54

3.5.3 Helicity Generation . . . . . . . . . . . . . . . . . . . . . . 60

3.5.4 Photocathode . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.5.5 The CEBAF Accelerator at Jefferson Lab . . . . . . . . . 63

3.5.6 Beam Instrumentation . . . . . . . . . . . . . . . . . . . . 63

3.6 Halo Monitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.7 Lumi Monitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3.8 Moller Polarimeter . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.9 Data Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3.10 The G0 Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . 71

3.10.1 North American electronics . . . . . . . . . . . . . . . . . 72

3.10.2 French electronics . . . . . . . . . . . . . . . . . . . . . . . 76

4 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

4.1 Measured Asymmetry . . . . . . . . . . . . . . . . . . . . . . . . . 78

4.2 Statistical Widths . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

4.3 Cuts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

4.4 Raw Asymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

4.5 Passive Sign Reversal . . . . . . . . . . . . . . . . . . . . . . . . . 85

4.6 Dead Time Corrections . . . . . . . . . . . . . . . . . . . . . . . . 87

4.7 Correcting Helicity-Correlated Beam Properties . . . . . . . . . . 89

4.8 Inelastic Dilution Factors . . . . . . . . . . . . . . . . . . . . . . . 96

iv
 4.8.1 French Results . . . . . . . . . . . . . . . . . . . . . . . . . 96

4.8.2 North American Results . . . . . . . . . . . . . . . . . . . 98

4.8.3 Inelastic Dilution Factor Results . . . . . . . . . . . . . . . 106

4.9 Background Inelastic Asymmetry . . . . . . . . . . . . . . . . . . 108

4.10 Polarimetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

4.11 Radiative Corrections . . . . . . . . . . . . . . . . . . . . . . . . . 111

4.12 Q2 Determination . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

5.1 Discussion of Engineering Run . . . . . . . . . . . . . . . . . . . . 115

5.2 Measured Physics Asymmetry . . . . . . . . . . . . . . . . . . . . 117

5.3 Theoretical Predictions . . . . . . . . . . . . . . . . . . . . . . . . 120

5.4 Future Parity-Violation Experiments . . . . . . . . . . . . . . . . 123

APPENDIX A
G0 Abbreviation & Acronym Glossary (GAAG) . . . . . . . . . . 127

APPENDIX B
Detector Testing and Calibration . . . . . . . . . . . . . . . . . . . 131

B.1 Output signals of the North American Focal Plane Detectors . . . 131

B.1.1 Light Pipe Characteristics . . . . . . . . . . . . . . . . . . 132

B.1.2 PMT attributes . . . . . . . . . . . . . . . . . . . . . . . . 143

APPENDIX C
Injector Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

C.1 Overview of 2002 Results . . . . . . . . . . . . . . . . . . . . . . . 147

C.1.1 BPM Noise . . . . . . . . . . . . . . . . . . . . . . . . . . 147

C.1.2 PZT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

C.1.3 Intensity Attenuator Cell . . . . . . . . . . . . . . . . . . . 158

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

VITA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

v
 LIST OF TABLES

1.1 Expected relative contributions of statistical and systematic errors on

GsE , GsM , and GeA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.1 Lengths and areas of the NA G0 scintillators. . . . . . . . . . . . . . . 48

3.2 The measured run averaged helicity-correlations for January 18-26,

compared to the G0 specifications for 700 hours of data taking. . . . . 51

3.3 Table of the IA calibration slopes measured in Hall C. . . . . . . . . . 57

3.4 Table of the limited number of PZT calibration slopes. . . . . . . . . 60

4.1 Comparison of North American and French raw asymmetries under

reversal of the halfwave plate. . . . . . . . . . . . . . . . . . . . . . . 87

4.2 IHWP out false asymmetries. The data are from 49 asymmetry runs. 95

4.3 IHWP in false asymmetries. The data are from 58 asymmetry runs. . 95

4.4 The corrected measured asymmetries for detectors 1-14. . . . . . . . . 96

4.5 The average inelastic dilution factors for the French detectors for the
cuts on the elastic peak which were ∼ 4 ns. . . . . . . . . . . . . . . . 99

4.6 The average inelastic dilution factors and the final absolute errors for
the North American detectors. . . . . . . . . . . . . . . . . . . . . . . 105

4.7 The interpolated background inelastic asymmetries for North Amer-

ican and French detectors. . . . . . . . . . . . . . . . . . . . . . . . . 110

4.8 < Q2 > values determined by comparing Time-of-Flight differences

between pions and the elastic protons at various magnetic fields. . . . 114

vi
5.1 Extracted North American and French asymmetries. . . . . . . . . . 118

5.2 Individual contributions to the errors for Aphy . . . . . . . . . . . . . 119

5.3 Some predicted values of strangeness radius, r s2 , and strange magnetic

moment, µs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

B.1 Results of the testing with minimum ionizing particles. . . . . . . . . 142

B.2 Minimum energies left by the protons hitting the G 0 scintillators and
subsequent number of photo-electrons detected. . . . . . . . . . . . . 142

C.1 Electronic noise calculated from position differences. . . . . . . . . . . 150

C.2 Electronic noise from calculating positions. . . . . . . . . . . . . . . . 150

C.3 Electronic noise by taking a pedestal run and injecting a fake signal. . 150

C.4 The responses of the injector BPM’s position differences to the X and
Y motion of the PZT mirror on December 19, 2002. . . . . . . . . . 151

C.5 The responses of the injector BPM’s charge asymmetries to the X

and Y motion of the PZT mirror on December 19, 2002. . . . . . . . 151

C.6 These results are the angle (in degrees) between the responses of the
PZT motion in X and in Y on December 19, 2002. . . . . . . . . . . . 151

C.7 The responses of the injector BPM’s position differences to the X and
Y motion of the PZT mirror on January 14, 2003. . . . . . . . . . . 152

C.8 The responses of the injector BPM’s charge asymmetries to the X

and Y motion of the PZT mirror on January 14, 2003. . . . . . . . . 152

C.9 These results are the angle (in degrees) between the responses of the
PZT motion in X and in Y on January 14, 2003. . . . . . . . . . . . . 152

C.10 The responses of the injector BPM’s position differences to the X and
Y motion of the PZT mirror on January 24, 2003. . . . . . . . . . . . 153

C.11 The responses of the injector BPM’s charge asymmetries to the X

and Y motion of the PZT mirror on January 24, 2003. . . . . . . . . 153

vii
C.12 These results are the angle (in degrees) between the responses of the
PZT motion in X and in Y on January 24, 2003. . . . . . . . . . . . . 153

C.13 Results of the adiabatic damping as measured during G 0 commissioning.156

C.14 The IA slopes for position differences in nm/V measured in December

2002. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158

C.15 The IA slopes for charge asymmetry in ppm/V measured in December

2002. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158

viii
 LIST OF FIGURES

1.1 The tree level electromagnetic and weak contributions to the scatter-
ing cross section in electron proton scattering. . . . . . . . . . . . . . 7

1.2 Schematic of the G0 apparatus at JLab. . . . . . . . . . . . . . . . . 12

1.3 Projected errors on GsE . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.4 Projected errors on GsM . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.5 Projected errors on GeA . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.6 CEBAF accelerator at JLab. . . . . . . . . . . . . . . . . . . . . . . . 17

2.1 QCD description of the proton. . . . . . . . . . . . . . . . . . . . . . 20

2.2 Tree-level Feynman diagram for electron-proton electromagnetic scat-

tering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.3 Example of a Rosenbluth plot. . . . . . . . . . . . . . . . . . . . . . . 23

2.4 Plot of the dipole parameterizations of GpM , GpE , and GnM and the
Galster parameterization of GnE . . . . . . . . . . . . . . . . . . . . . . 24

2.5 Plot of µp GpE /GpM vs. Q2 . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.6 Tree level Feynman diagram for electron-proton weak neutral current
scattering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.7 Feynman diagrams associated with the electron-proton elastic axial

form factor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.8 SAMPLE uncertainty bands of GsM vs. GsE at Q2 =0.1 (GeV/c)2 . . . 36

2.9 Measured HAPPEX and PVA4 results. . . . . . . . . . . . . . . . . . 37

ix
3.1 A schematic of the G0 magnet collimators, target, and detectors. . . . 40

3.2 A schematic of the G0 hydrogen target. . . . . . . . . . . . . . . . . . 41

3.3 Schematic of the dummy targets. . . . . . . . . . . . . . . . . . . . . 43

3.4 Fluctuations in the statistical width (ppm) as a function of raster size

for target boiling studies. . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.5 Pictures of the G0 detector system. . . . . . . . . . . . . . . . . . . . 46

3.6 Pictures of North American and French octants. . . . . . . . . . . . . 47

3.7 The G0 beam structure. . . . . . . . . . . . . . . . . . . . . . . . . . 52

3.8 Schematic of the CEBAF polarized source laser table in the injector. 54

3.9 The IA calibration constant (ppm/V) as a function of beam position

monitor in the injector to Hall C. . . . . . . . . . . . . . . . . . . . . 56

3.10 Typical performance of the charge feedback system using the IA cell. 58

3.11 The measured intensity asymmetry as a function of the rotatable

halfwave plate angle. . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

3.12 Schematic of strained GaAs band structure and energy level diagram. 61

3.13 Plot of normalized rates as a function of run number for the 2003 G 0
engineering run as measured by the Beam Halo monitor. . . . . . . . 65

3.14 Plot of Lumi asymmetry as a function of run number for the 2003 G 0
commissioning run. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3.15 The Hall C Polarimeter. . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.16 Schematic of the North American electronics for the G 0 forward run-
ning mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

4.1 An example of a typical French Time-of-Flight spectrum. . . . . . . . 79

σmeas
4.2 Histogram of all the detectors . . . . . . . . . . . . . . . . . . . 82
σstat
4.3 North American and French IHWP in and out raw asymmetries. . . . 85

4.4 North American and French (in+out) raw asymmetries. . . . . . . . . 86

x
4.5 Plots of the helicity-correlated beam properties. . . . . . . . . . . . . 93

4.6 Detector sensitivity slopes plotted as a function of octant. . . . . . . 94

4.7 An example of a 3 Gaussian fit to French Detector 9 Octant 2. . . . . 97

4.8 An example of the stability of the inelastic dilution factor over 106
runs for detector 14 Octant 2. . . . . . . . . . . . . . . . . . . . . . . 97

4.9 French inelastic dilution factors as a function of detector number,

averaged over 4 French octants, using 3 Gaussian fit. . . . . . . . . . 98

4.10 An example of a time-of-flight spectra from a North American detector.100

4.11 An example of several white noise spectra for North American octant
3 detector 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

4.12 An example of one bin’s variation over 8 measurements over 4 months

for detector 7 octant 1. . . . . . . . . . . . . . . . . . . . . . . . . . . 102

4.13 The plots of a simulated North American ToF spectrum using the fit
parameters obtained from the French ToF and North American ToF
spectrum with DNL. . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

4.14 Histogram of the simulated extracted inelastic dilution factors for a

North American detector along with the “true” extracted inelastic
dilution factor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

4.15 Corrected and uncorrected North American ToF spectrum. . . . . . . 104

4.16 Stability of the North American inelastic dilution factors over ∼50
runs for a typical detector. . . . . . . . . . . . . . . . . . . . . . . . . 105

4.17 Comparison of Time-of-Flight spectra have been corrected using the

same white noise run. . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

4.18 Runs corrected by a single white noise calibration run. . . . . . . . . 107

4.19 North American inelastic dilution factors as a function of detector

number, averaged over the North American octants. . . . . . . . . . . 107

4.20 Example of the coarser asymmetry binning to find the background

inelastic asymmetry for North American detector 4. . . . . . . . . . . 108

4.21 North American and French extracted background asymmetries. . . . 109

xi
4.22 Polarization measurements made during the engineering run. . . . . . 111

5.1 Plot of the extracted North American and French asymmetries vs.
momentum transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

5.2 Kaon loop diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

5.3 Pole loop diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

5.4 Expected forward angle results from the G0 , HAPPEX II, and A4
along with result from the HAPPEX experiment. . . . . . . . . . . . 125

B.1 Positions of interest along the G0 scintillator. . . . . . . . . . . . . . . 135

B.2 Cosmic ray/Ru calibration signal for G0 scintillator . . . . . . . . . . 138

B.3 Results of the testing with minimum ionizing particles. . . . . . . . . 140

B.4 Minimum number of photo-electrons detected by the G 0 NA FPD . . 141

B.5 Ratio of the gain of the actual PMT attached to the light pipe to the
gain necessary to perfectly match the light pipe. . . . . . . . . . . . . 144

B.6 Dispersion of the ratio of the measured to the predicted cosmic am-
plitudes in the Clean Room. . . . . . . . . . . . . . . . . . . . . . . . 144

C.1 Plots that show the BPM resolution. . . . . . . . . . . . . . . . . . . 148

C.2 Plots of the instrumental noise of the BPMs. . . . . . . . . . . . . . . 149

C.3 Plots of the helicity correlated position differences in X and Y as a

function of PZT X and Y motion and as a function of beam monitor
in the injector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

C.4 Plots of the adiabatic damping measured at different BPMs in the

injector compared to BPM G0B in Hall C. . . . . . . . . . . . . . . . 157

C.5 Charge asymmetry calibration constants (ppm/V) for the IA. . . . . 159

xii
 ABSTRACT

This work is based on the first engineering run of the G 0 experiment from
October 2002 though January 2003 in Hall C at Jefferson Lab. The G 0 experiment
will be the first to measure the weak neutral form factors: G ZE (Q2 ), GZM (Q2 ) and
GeA (Q2 ) and to extract the proton’s strange form factors: GsE (Q2 ) and GsM (Q2 ) via
a Rosenbluth separation over a range of Q2 (0.1 − 1.0 (GeV/c)2 ). This will require
four sets of measurements: forward angle measurements with a proton target, and
three sets of backward angle measurements with a hydrogen and deuterium target.
The measurements are made of the parity-violating asymmetries in elastic electron
scattering.
The G0 experiment is a major installation at Jefferson Lab’s Hall C with a new
dedicated spectrometer. The superconducting magnet is made up of 8 coils with
a maximum field of 1.6 T·m. The scintillator detector array (detector solid angle
between 0.4 - 0.9 sr) detects recoiled protons in the forward angle measurement
(where θp = 70◦ ± 10◦ corresponding to scattered electron angles of a few degrees)
and to detect scattered electrons in the backward angle measurements. This detector
array is made up of a set of 16 pairs of scintillators arranged in 8 sectors around
the beam line. Custom electronics handle the high data rate (approximately 1 MHz
per detector). The target is a 20 cm long liquid hydrogen cryotarget. Besides the
check-out of the new hardware, G0 has stringent requirements on the performance
of the polarized electron beam in order to minimize false asymmetries. Further
complicating this fact, in the forward mode, was the requirement that the time
structure of the JLab beam had to be changed from 499 MHz to 31 MHz in order
to count the recoiled protons in the spectrometer. Data was collected over a 12 day
period at the end of the engineering run. These data were analyzed for a Q 2 range of
0.1 − 0.4 (GeV/c)2 corresponding to measured electroweak asymmetries that ranged
from (-4.4 ± 1.6 ± 1.6 ppm) to (-8.5 ± 2.8 ± 2.5 ppm).

xiii
CHAPTER 1

Introduction

The existence of the proton has been known since the early 20th century, yet

its structure is still not entirely understood. The current theory of the strong in-

teraction, Quantum Chromdynamics (QCD), describes the proton as being made

up of three valence quarks (two up and one down quarks) within a complicated

“sea” of quarks, antiquarks and gluons. A troubling difficulty of QCD is that while

it successfully describes the strong interaction at high momentum transfers where

the theory is perturbative, the theory is more difficult to handle at low momentum

transfers where the theory is non-perturbative.

Strange and anti-strange quarks, which are the next lightest quarks after the

up and down, are found in the quark “sea” around the proton. Since strange quarks

have a comparable mass with the proton, the question can be asked, what role do

they play in contributing to the static properties of the proton?

1.1 Strange Quarks in the Proton

Strange quarks are the lightest quarks not to contribute to the proton’s valence

distribution. The strange quarks exist as only strange quark-antiquark pairs in the
1
 2
quark sea surrounding the proton. The net strangeness of the proton is zero. This

might lead one to believe that strange quarks cannot contribute to the properties

of the proton. Experiments have indicated that strange quarks do, in fact, play a

fundamental role in the understanding of the proton [1–5, 12].

One set of strange quark observables is related to the so-called strange quark

matrix elements of the nucleon. These matrix elements have the form

hP |s̄Γs|P i (1.1)

where |P i is the proton state, s̄Γs is an operator containing strange quark, s, fields

bilinearly, Γ is a matrix in spinor space which takes the form Γ = 1 4 , γ µ , or γ µ γ5

depending on whether one is interested in the scalar, vector, or axial strangeness of

the proton.

One of the original indicators that strange quarks play a fundamental role in

the proton came from looking at the pion-nucleon sigma term. Strange quarks

contribute to the mass of the proton via the matrix element hP |s̄s|P i. This matrix

element can be inferred from the so-called “sigma term” in π-N scattering. The

pion-nucleon sigma term is defined to be

¯ i
σπN = m̂hP |ūu + dd|P (1.2)

where m̂ = 12 (mu + md ), the average of the up and down quark masses. The proton

mass, under the SU(3) flavor assumption that one can neglect cc̄, b b̄ and tt̄, can then

be written as

Mp = M0 + σs + σ (1.3)

where Mp is the physical mass of the proton, M0 is the mass of the proton as the

quark masses go to zero (in the so-called chiral limit), σ s is the mass contribution

due to strangeness (if any) and σ is the mass contribution from non-strange quarks.
 3
The pion-nucleon term σπN can be used to find the non-strangeness quark con-

tribution σ by extracting σπN at the unphysical Cheng-Daschen point (that is, where

the Mandelstam variables s = Mp2 and t = m2π ) by use of dispersion relations. The

standard value [1] for this result, after taking into account higher-order contribu-

tions, is σ = 45 ± 8 MeV.

If the nucleon is free of strange quarks, σ should equal the SU(3)-octet scaler

quark density which can be calculated from the baryon masses

¯ − 2s̄s|P i.
σ̂ = m0 hP |ūu + dd (1.4)

This has been calculated [2, 3],to be σ̂ = 35 ± 5 MeV.

Comparing σ̂ and σ, deviations from the equality, are assumed to be coming

from strange quarks. This can be written as

σ̂ = σ(1 − y). (1.5)

where y is the strangeness content of the proton, which can then be written as

2hP |s̄s|P i
y= ¯ i. (1.6)
hP |ūu + dd|P

This leads to a value of y = 0.2 ± 0.2, indicating that strange quarks might

contribute to the mass of the proton. This implies that as much as 200 MeV of

the proton’s mass might be due to the strange quarks. This result also indicates

a violation of the OZI rule [7] which assumes that the nucleon is free of strange

quarks. This result must be accepted with some degree of skepticism since there is a

significant uncertainty due to the data and the extensive theory needed to interpret

this result.

Another piece of evidence for strange quark contributions to the nucleon come

from some p̄p annihilation channels [8, 9]. In these channels, an observed enhance-

ment of φ relative to ω production is in disagreement with the OZI prediction by a

 4
factor of 30-50. A possible interpretation of these results is that the nucleon wave

function contains some significant fraction of polarized s̄s pairs.

Another indicator that strange quarks play a role in the proton comes from po-

larized deep inelastic scattering (DIS) experiments. This method allows for access to

the hP |s̄γ 5 s|P i matrix element. The focus of these experiments is the spin structure

of the nucleon. These spin structure experiments indicate that the fraction of the

proton spin carried by the valence quarks is Σ ∼ 0.3. This naturally leads to the

question: is some of the missing spin due to strange quarks?

Each of the quark (antiquark) flavors can be described by a single quark (an-

tiquark) distribution function q(x) (q̄(x)) over a range in Bjorken x, where Bjorken

x is the fraction of four-momentum carried by a parton in the proton. The distri-

bution q(x) is the probability that a parton carries a fraction x of the momentum

of the proton. A quantity of interest is the net spin polarization, ∆q, of the quark

flavor q
Z1
∆q ≡ [q ↑ (x) − q ↓ (x) + q̄ ↑ (x) − q̄ ↓ (x)]dx. (1.7)
0

The g1 (x) structure function is the charge-weighted vector sum of the quark polar-

izations in the nucleon

1X
g1 = Qq ∆q (1.8)
2 q

The first moment of the g1 (x) structure function, Γ1 , describes the total spin carried

by the quarks

Z1
Γ1 ≡ g1 (x)dx. (1.9)
0

The Ellis-Jaffe sum rule [10] connects the structure functions to the quark spin

distributions using Equations 1.8 and 1.9. The Ellis-Jaffe sum rule for the proton
 5
can be written as
   
1 5 3F/D − 1
Γp1 = 1+ (1.10)
12 3 F/D + 1

where F and D are the universal weak decay constants. Ellis and Jaffe assumed

that the strange quarks do not contribute to the nucleon’s spin and SU(3) flavor

symmetry. This results in ∆s = 0. Ellis and Jaffe’s calculation results in the

relation,
 
1 4 1
Γp1 = ∆u + ∆d = 0.17 (1.11)
2 9 9

The EMC collaboration [11] measured Γp1 and found

Γp1 = 0.126 ± 0.010(stat) ± 0.015(syst) (1.12)

which is in disagreement with the calculated results in Equation 1.11 where ∆s was

frozen out. This was a hint that strange quarks might play a role in the properties

of the proton.

Now with these results one can isolate the individual flavor components. The

total fraction of the nucleon spin carried by the quarks is [4, 5]

∆u + ∆d + ∆s = 0.20 ± 0.10. (1.13)

The portion of the spin due to strange quarks is [4, 5]

∆s = −0.1 ± 0.1. (1.14)

The minus sign implies that the strange quarks and antiquarks are polarized nega-

tively with respect to the direction of the nucleon spin. This extracted value is to be

taken with care since the strange quark extraction is sensitive to SU(3) flavor break-

ing, and information from neutron beta decay and hyperon semi-leptonic decays had

to be incorporated into the analysis assuming exact SU(3) flavor symmetry.

 6
Another DIS technique that is employed is the scattering of neutrinos and anti-

neutrinos with nucleons to probe the ss̄ sea. The NuTeV experiment [12] at Fermilab

looked at the production of charmed particles in charged-current interactions of neu-

trinos and anti-neutrinos with nucleons in the deep inelastic region. These charmed

particles are produced in d − c and s − c transitions. The neutrinos interact with the

d and s quarks by raising their charge and producing a negative lepton. The d − c

transition is a Cabibbo suppressed one. This enhances the possibility of studying

the strange sea. By observing two muon-neutrino events, the NuTeV collaboration

extracted the total momentum fraction κ of the strange sea as [12]

R1
2 0 (s + s̄)dx
κ = R1 = 0.42 ± 0.07 ± 0.06 (1.15)
¯
(u + ū + d + d)dx
0

at Q2 = 16 (GeV/c)2 .

From experiment, there has thus been evidence that the strange quark-antiquark

sea may play a significant role in the proton. Much work has gone into extracting

the π − N sigma term to find the strange quark mass contribution, but theoretical

uncertainties still exist. The strange quark-antiquark contribution to the spin of the

proton has been another subject of intense research, but once again the result suffers

from uncertainties in the theoretical interpretation. Low energy neutrino scatter-

ing will offer the best hope in measuring ∆s. This can be technically challenging

since there can be uncertainties in knowing the neutrino flux, the detector efficien-

cies, and nuclear target effects. With all these tantalizing hints, parity-violating

electron-proton scattering is attractive, as it is a technique that has the potential

to provide a direct and clean measurement of the strange vector and axial currents

in the nucleon.
 7
1.2 Parity-Violating Electron Scattering

In order to access the strange magnetic and electric form factors, parity-violating

electron scattering has been employed. The strange magnetic and electric form fac-

tors, represented by GsE and GsM , are physical observables related to the strange

quark charge and magnetic distributions in the nucleon. In electron-proton scatter-

ing, two different kinds of interactions are involved: the electromagnetic interaction

via the exchange of a photon and the weak interaction via Z 0 exchange. This can

be seen from the cross-section in Figure 1.1.

σ∝

FIG. 1.1: Electromagnetic and weak contributions to the scattering cross section in elec-
tron proton scattering.

Typically, at low momentum transfers, the weak interaction is ignored since

it is small in comparison to the electromagnetic contribution. However, the weak

interaction violates parity, thus by using polarized electrons and forming the ratio of

the difference of polarized cross sections over the sum of the polarized cross sections,

an asymmetry can be formed, which is non-zero only due to the weak interaction.

This asymmetry can be written for longitudinally polarized electrons scattering from

an unpolarized proton target, as

σ+ − σ− GF Q2
A = ≈ √ ≈ 10−4 Q2 (1.16)
σ+ + σ− 4 2πα
where GF is the Fermi constant, Q2 is the momentum transfer (0.1 < Q2 < 1.0

(GeV/c)2 for the G0 experiment), σ+(−) (sometimes this is denoted as σr(l) ) is the
 8
cross section for right (left) handed incident electrons scattering from a proton and

α is the fine structure constant. A right (left) handed particle is a particle whose

spin vector is parallel (anti-parallel) to its momentum vector. This is known as the

particle’s helicity. This asymmetry can then be related to the electromagnetic and

weak form factors. The form factors can then be written in terms of quark flavors.

Then by utilizing charge symmetry between the neutron and proton, the asymmetry

can be written in terms of known electromagnetic form factors for the proton and

neutron and the strange quark form factors.

Measuring asymmetries on the order of 10−6 with errors on the order of 10−7 is a

challenging feat. In order to make a measurement with this precision, a large number

of scattering events with specified helicity is required. The number of scattered

events, ns is given by

dσ dσ
ns = ×L= × I × ρ × L × ∆Ω (1.17)
dΩ dΩ
dσ
where is the differential scattering cross section, I is the beam current, ρ is the
dΩ
density of the target, L is the target length, ∆Ω is the solid angle of the detector

and L is the luminosity. It is most important to maximize the luminosity in order

to get as many events as possible. This in turn drives the characteristics common

to most electron-nucleon parity-violation experiments:

- high beam polarization

- high beam currents

- long targets

- large detector solid angles.

Another concern with parity experiments is the control of helicity-correlated sys-

tematic errors. The measured asymmetry is fairly insensitive to common systematic

 9
errors. Let Nsys be the number of counts associated with a common systematic

error such as small drifts in some experimental parameter. Forming the measured

asymmetry, Ameas , with this common systematic error

(N+ + Nsys ) − (N− + Nsys )

Ameas = (1.18)
(N+ + Nsys ) + (N− + Nsys )
N+ − N−
= (1.19)
N+ + N− + 2Nsys
 −1
N+ − N− Nsys
≈ 1+ (1.20)
2N N
 
Nsys
≈ AP V 1 − (1.21)
N

where N+ is the number of right-handed scattered particles, N − is the number of

N+ + N−
left-handed scattered particles, N ≈ , and AP V is the parity-violating
2
asymmetry without the systematic contribution.

On the other hand, parity experiments must guard against helicity-correlated

systematic errors. Using the variables defined above, but now with a helicity-

correlated systematic error, NHC sys the measured asymmetry, Ameas , can be written

as

(N+ − NHC sys ) − (N− − NHC sys )

Ameas = (1.22)
(N+ − NHC sys ) + (N− − NHC sys )
N+ − N− + 2NHC sys
≈ (1.23)
N+ − N−
NHC sys
≈ AP V + . (1.24)
N

These helicity-correlated systematic errors which come in as an additive factor to

the ‘true’ AP V must be controlled since they form a false asymmetry which adds

directly to the true asymmetry.

In order to control systematic errors and increase the number of scattered events

measured by the detector(s), a variety of techniques have been established. Some

of these techniques are:

 10
- GaAs and strained GaAs crystals used as photo-cathodes

- rapid pseudo-random reversal of the beam helicity

- precision beam monitors

- passive helicity reversal (typically by an insertable halfwave plate)

- beam intensity and position feedback.

In general, most electron-nucleon scattering parity experiments have similar set-ups.

Linearly-polarized light from a laser is transformed into circularly-polarized light by

a Pockels cell. By applying different voltages to the Pockels cell, left and right-

handed polarized light is produced. This circularly-polarized light shines on a GaAs

crystal liberating polarized electrons by the photo-electric effect. An insertable half-

wave plate can be inserted and retracted from the laser beam allowing for passive

helicity reversal to check for (and cancel) some systematic errors.

The polarized electrons are accelerated to some energy before impinging on a

target. During the transport through the accelerator and experimental end sta-

tion, beam position and current monitors detect helicity-correlated differences in

the beam. Position monitors in an accelerator arc can be used to measure the

helicity-correlated energy differences. These beam monitors feed data to feedback

systems which in turn attempt to minimize the helicity-correlated differences. De-

tector packages can then be arranged in the experimental end station to detect the

scattered particles.

1.3 Overview of the G0 Experiment

The full measurement of the G0 experiment will access the strange quark con-

tributions to the magnetic and charge distributions of the proton, G sM (Q2 ), GsE (Q2 ),
 11
and the electron-proton axial form factor, GeA (Q2 ). This will be done by measur-

ing parity-violating asymmetries in forward elastic electron-proton scattering and

backward scattering for elastic electron-proton and quasi-elastic electron-deuteron

scattering. The parity-violating asymmetries are expected to range from -3 to

-35 parts-per-million (ppm)1 for the forward angle measurements and the asymme-

tries measured in the backward angle configuration on hydrogen are expected to

be larger, ranging from -18 to -72 ppm. This will allow for a clean extraction of

GsM (Q2 ), GsE (Q2 ) and GeA (Q2 ) with few assumptions (such as charge symmetry).

The complete G0 experiment will be the first experiment to completely separate

GsM (Q2 ), GsE (Q2 ) and GeA (Q2 ) and to measure the evolution of these observables

at three different momentum transfers (Q2 ). G0 is a counting experiment in con-

trast to previous parity-violating experiments like SAMPLE [13] at MIT-Bates and

HAPPEX [14] in Hall A at Jefferson Lab that used an integrating technique.

The G0 experiment is located in Hall C at Jefferson Lab. A dedicated large ac-

ceptance spectrometer (see Figure 1.2) was built in order to perform the experimen-

tal program. The spectrometer is a toroidal magnet consisting of 8 superconducting

coils in a common cryostat that generate up to a 1.6 T field. The diameter of this

spectrometer is about 4 m, and it has an operating current of 5000 A. The total

energy stored in the magnet is 6.6 MJ. The solid angle is defined by collimators at

the inner diameter of the coils. This geometry allows a line-of-sight shield from the

detectors to the target. Elastically scattered particles of the same Q 2 are focused

onto individual focal plane detectors. Each detector is made of a pair of plastic

scintillators. There are 8 sets of 16 detector pairs called octants that are placed

symmetrically around the symmetry axis of the spectrometer.

For the measured asymmetries, which are on the order of parts-per-million, the
1
Due to the large number of abbreviations and acronyms, Appendix A is glossary of many terms
used in this experiment.
 12

FIG. 1.2: Schematic of the G0 apparatus at JLab. This is a dedicated spectrometer

in Hall C. In the forward configuration, polarized electrons hit an unpolarized hydrogen
target and the recoil protons are detected while the electrons scatter at a forward angle.
The apparatus can then be turned around so that polarized electrons hit an unpolarized
hydrogen (deuterium) target and the back-scattered electrons are detected.

level of statistical uncertainty will be about 5% and the systematic uncertainties

related to helicity-correlated effects should be on the order of 10 −7 . The G0 col-

laboration will extract GsE , GsM , and GeA at the values Q2 = 0.30, 0.50, and 0.80

(GeV/c)2 from the measured asymmetries. The projected errors from this extrac-

tion are shown in Table 1.1 and in Figures 1.3, 1.4, and 1.5. The calculation assumes

700 hours of data taking for the forward angle measurement on the hydrogen target

and for each of the backward angle measurements on the hydrogen and deuterium

targets. It is the statistical uncertainties on these measured asymmetries that dom-

inate the overall errors on GsE , GsM , and GeA . The polarization of the electron beam

is assumed to be 70 ± 2% in the calculation of these proposed errors. It is the polar-

ization measurement that is expected to dominate the systematic errors on G sE , GsM ,

and GeA . Errors coming from the uncertainties in the electromagnetic form factors

are: 20% for GnE , 3% for GnM , 2% for GpE , and 2% for GpM . A theoretical uncertainty
 13

FIG. 1.3: Projected errors on GsE . Result from the SAMPLE experiment is shown along
with the expected errors from the HAPPEX II experiments. Also shown are theoretical
predictions from chiral perturbation theory, lattice QCD, and various pole (dispersion
analysis) models.

had to be included on the isoscaler part of the electron-proton axial form factor,

GeA (T = 0) since this contribution is not accessible in the asymmetry measurements.

The momentum transfer Q2 is expected to be measured to within 1%.

In the first phase of the G0 experiment, forward angle asymmetries are mea-

sured. This is done by detecting elastically scattered protons between 62 ◦ < θp < 78◦

The 20 cm liquid hydrogen target is based on the SAMPLE [15] design. The tar-

get’s main requirement is that the density remain constant as the beam deposits

up to 500 W of power. The target provides a high longitudinal flow of about 5-10

m/s in order to provide enough mixing by turbulent flow. The spectrometer has an

acceptance of ∼ 0.9 sr defined by collimators at the inner radius of the coils of the

magnet. The measured Q2 range is from 0.1 < Q2 < 1.0 (GeV/c)2 with an incident

beam energy of 3.0 GeV. The scattered particles are detected by pairs of plastic
 14

FIG. 1.4: Projected errors on GsM . Result from the SAMPLE experiment is shown along
with the expected errors from the HAPPEX II experiments. Also shown are theoretical
predictions from chiral perturbation theory, lattice QCD, and various pole (dispersion
analysis) models.

scintillators known as the Focal Plane Detectors (FPDs). For each of the eight

sectors, there are 16 detector pairs corresponding to increasing Q 2 with increasing

detector number. The detectors are shaped into arcs of a circle in order to collect

events of approximately the same Q2 . In the forward mode, time-of-flight is used to

reject some of the backgrounds. Since it takes about 20 ns for a proton to reach a

detector from the target, the CEBAF machine producing polarized electrons must

reduce the micro-structure of the beam bursts from 499 MHz to 31 MHz in order to

to provide electrons every 32 ns for this experiment. Custom electronics were built

to accumulate the time-of-flight spectra in this configuration. These time-of-flight

data are recorded by shift registers that feed scalers. This time-of-flight measure-

ment, made over a 32 ns window, is used to supplement the momentum selection

by the spectrometer and separate elastic from inelastic contributions, allowing for
 15

GsE GsM GeA (T = 1)

2 2
Q (GeV/c) 0.30 0.50 0.80 0.30 0.50 0.80 0.30 0.50 0.80
Error (ppm) 0.032 0.037 0.052 0.090 0.059 0.041 0.188 0.159 0.137
Af (%) 20.3 11.4 12.5 0.20 0.1 0.1 1.5 0.7 0.7
Ab (%) 31.0 34.3 37.8 50.3 47.4 47.9 0.9 0.8 0.9
Adeut (%) 14.3 17.1 22.0 23.3 23.7 27.9 61.8 62.6 72.6
GpE (%) 2.1 1.9 1.0 0.0 0.0 0.0 1.1 0.4 0.1
GpM (%) 1.5 1.3 0.6 0.9 1.2 1.0 0.4 0.8 0.8
GnE (%) 12.2 10.9 5.7 0.3 0.2 0.1 0.9 0.6 0.3
GnM (%) 0.9 0.9 0.6 1.4 2.0 1.9 3.8 3.1 1.9
Q2 (%) 4.4 4.9 3.7 5.0 4.8 3.3 7.2 6.9 4.8
Pe (%) 11.4 14.3 11.8 15.6 16.5 12.5 22.3 23.8 17.8
GeA (T = 0) (%) 1.8 3.1 4.2 3.0 4.2 5.3 0.1 0.2 0.2

TABLE 1.1: Expected relative contributions of statistical and systematic errors on the
proposed full G0 experiment for GsE , GsM , and GeA . The total absolute error is noted
in the second line of the table. The statistical errors on the forward, backward, and
deuterium asymmetry measurements are denoted by Af , Ab , and Adeut respectively.

background corrections to the elastic asymmetries. The maximum rate of elastically

scattered protons for a given detector pair is about 1 MHz.

The longitudinally-polarized electrons are provided by the Continuous Electron

Beam Facility (CEBAF). Linearly-polarized light is provided by a laser in the injec-

λ
tor and is turned into circularly-polarized light by a waveplate (a Pockels cell).
4
This circularly-polarized light interacts with a “strained” GaAs photocathode and

liberates longitudinally-polarized electrons. The helicity of the electrons can be re-

versed by changing polarity of the voltage applied to the Pockels cell. As mentioned

above, G0 requires a beam energy of 3.0 GeV, with high intensity (40 µA) and a

pulsed structure (31 MHz instead of the typical 499 MHz) in the forward angle phase

of the experiment. This mode produces a charge per bunch that is 16 times larger

than normal. This requires changes to be made to the beam optics. It is important

for G0 to control any helicity-correlated beam differences. This is because these

beam differences may manifest themselves as false asymmetries.

 16

FIG. 1.5: Projected errors on GeA . Result from the SAMPLE experiment is shown. Also
shown are theoretical predictions.

In the second phase of the G0 experiment, backward angle asymmetries will

be measured. This will be done by reversing the apparatus relative to the beam

line. Elastically scattered electrons will be detected at 110 ◦ from the same 20 cm

liquid hydrogen target. In this phase of the experiment, the background is expected

to be composed of electrons and pions from inelastic processes. In this case, the

time-of-flight cannot discriminate between the different reactions and thus the se-

lection between particles will be obtained from their different trajectories. To aid in

discrimination of elastic and inelastic electrons, another set of 9 detectors, the cryo-

stat exit detectors (CEDs) will be added near the exit window of the cryostat. The

Focal Plane Detector arrays will be reduced to a single layer of 16 scintillators. This

reduction of the FPDs is due to the fact that the CEDs act as the second scintillator.

The decision to keep the front FPD scintillator and not the back FPD scintillator is
 17

FIG. 1.6: The CEBAF accelerator at JLab.

because the front detector could become a source of background/multiple scatter-

ing for the back scintillator. The coincidence between the CED-FPD combination

allows for a rough measurement of the electron momentum and scattering angle. To

reject pion background from the entrance and exit foils of the targets, a Cerenkov

detector will be placed between the CEDs and the FPDs. In the backward mode,

programmable logic chips in the electronics will be employed to identify elastic events

from CED/FPD coincidences. The target will also be filled with liquid deuterium

in order to perform a third set of measurements for the extraction of the axial form

factor.

This thesis is based on the work of the First G0 Engineering Run that occurred

in October 2002 through the January 2003. The First G 0 Engineering Run was

a “proof of principle” run. The layout of this thesis is as follows: Chapter 2 de-

scribes the formalism of parity-violating electron-nucleon scattering and previous

experiments, Chapter 3 describes the G0 experimental apparatus with an emphasis

on the polarized electron beam (discussed in further detail in Appendix B) and on

 18
the North American detector system (further discussed in Appendix C), Chapter 4

describes the analysis of the data from the raw asymmetries to the extracted physics

asymmetries with a particularly detailed discussion on the extraction of the inelas-

tic dilution factors, and in Chapter 5, the physics asymmetries are compared to the

standard model predictions and the performance of the individual components of

the G0 experiment are discussed.

CHAPTER 2

Physics Formalism

Parity-violating electron scattering is a practical and clean method for measur-

ing the strange quark vector matrix elements. This is done by measuring the parity-

violating amplitudes arising from the electroweak interference in elastic scattering

of polarized electrons from an unpolarized proton. In this chapter, the formalism

required to interpret the G0 experiment is presented.

2.1 Structure of the Proton

The proton is a composite particle made up of quarks and gluons (collectively

known as partons). This cluster of quarks and gluons that form the proton can be

approximately written as

p = uud
|{z} + uū
| + dd̄ {z
+ ss̄ + . .}. + g + g + . . .; . (2.1)
| {z }
valence quarks sea quarks gluons

19
 20

FIG. 2.1: The proton is made up of three valence quarks, two up and one down quark.
Gluons, the mediator of the strong interaction, are exchanged between quarks to hold the
proton together. These gluons, by virtue of the Heisenberg uncertainty principle, bubble
into quark-antiquark pairs as illustrated by the strange quarks in this figure.

The first term in Equation 2.1 refers to the so-called “valence quarks”, where the

current quark masses are

mu ≈ 5 MeV (2.2)

md ≈ 9 MeV (2.3)

ms ≈ 175 MeV. (2.4)

The second term is related to the large number of relatively low-momentum quark-

antiquark pairs, known collectively as “sea quarks”. The probability of these q q̄

pairs fall off inversely with the mass of the quark species being produced (this is

why the heavy cc̄, bb̄, and tt̄ are expected to play a relatively small role). The

quarks are all held together by the mediator of the strong interaction, gluons. In

fact, it is from these gluons that the q q̄ sea quarks are generated via the Heisenberg

uncertainty relation much like in e+ e− vacuum polarization in QED. As has been

demonstrated in Section 1.1, strange quarks may play a major role in the structure

of the proton and accessing the strange quarks says something about the pure sea

quarks. Sadly, perturbative QCD cannot light the way, since the mass of the strange
λQCD
quark, ms is comparable to the QCD scale, λQCD , where ∼ 1.
ms
 21

FIG. 2.2: Tree level Feynman diagram for electron-proton electromagnetic scattering.

2.2 Electromagnetic Form Factors

Electromagnetic form factors describe the complicated interactions of photons

with the complex structure of the nucleon in elastic scattering (see the tree-level

Feynman diagram in Figure 2.2). This tree-level scattering diagram is described by

the following scattering amplitude for the single photon exchange

4πα µ p
Mγ = − J J . (2.5)
q2 e µ

The electron is a point-like spin- 21 particle; the associated current for the electron is

Jeµ = ūγ µ u (2.6)

where u is the Dirac 4-component spinor that describes the initial electron, u is

the Dirac 4-component adjoint spinor of the final electron, and γ µ is the Dirac

gamma matrix. In contrast, the proton, being an extended spin- 21 particle, yields a

more complicated transition current due to its complex structure. Since the electro-

magnetic interaction respects parity conservation, Lorentz invariance and current

conservation, the transition current can be written in the general form as

   
κ
Jµp = ū γµ F1 + i ν
F2 σµν q u (2.7)
2Mp
 22
where F1 (Q2 ) and F2 (Q2 ) are known as the Dirac and Pauli form factors respectively,

κ = 1.79 nuclear magnetons is the anomalous part of the magnetic moment of the

proton (if the proton were structureless then κ = 0). The momentum transfer, Q 2 ,

is the only variable needed to describe the electromagnetic interaction vertex and

hence the form factors are only functions of momentum transfer.

The Sachs form factors are more physically insightful combinations of the Dirac

and Pauli form factors, and allow for a more convenient formalism to be written

without form factor cross terms in the cross section for electron-nucleon scattering.

In the Breit frame, where the exchanged boson (in this case a virtual photon or Z 0 )

is purely space-like (Q2 = 0), this implies that the initial and final momentum of the

scattered particle is equal in magnitude but opposite in sign (~p 0 = p~). These form

factors are closely related to the proton charge and magnetic moment distributions.

The Sachs form factors are related to the Pauli and Dirac form factors by

GpE (Q2 ) ≡ F1 − τ F2 GpM (Q2 ) ≡ F1 + F2 (2.8)

Q2
where τ = 4MN2 is a measure of the relativistic recoil effects. At Q2 = 0 these form

factors are normalized to

q0 µp
GpE (Q2 = 0) = GpM (Q2 = 0) = (2.9)
e e/2Mp
where q0 is the proton’s electric charge and µp is the proton’s magnetic moment.

GpE and GpM can be determined from elastic electron-proton scattering experiments.
dσ
In these experiments, the differential cross sections, , are measured at different
dΩ
values of Q2 and lab angle, θ. The Rosenbluth formula [16] which describes the

elastic electron-proton scattering can then be applied

   p 2  
dσ dσ (GE ) + τ (GpM )2 p 2 2 θ
= + 2τ (GM ) tan (2.10)
dΩ dΩ M ott 1+τ 2
dσ


where dΩ M ott
is the Mott differential cross section [17] which describes the scat-
1
tering of spin 2
Dirac particles and takes into account the recoil of the proton. A
 23
Rosenbluth plot (as shown in Figure 2.3) can be made showing the cross section as

a function of tan2 ( θ2 ) for constant Q2 . The data should lie along a straight line with

a slope of 2τ (GpM )2 and the extrapolation to τ = 0 will determine the electric form

factor GpE for that value of Q2 .

2
τ GM
e =2
Slop
d Ω d ΩMott
dσ/ dσ

G2E + τ G 2M
Y intercept =
1+τ

tan2 θ
2

FIG. 2.3: Rosenbluth plot showing the linear relationship between (dσ/dΩ)/(dσ/dΩ) M ott
and tan2 (θ/2) at a fixed value of Q2 .

The proton form factors GpE and GpM have been determined by the Rosenbluth

separation up to Q2 ∼ 9 (GeV/c)2 [18–21]. Empirically both GpE and GpM approxi-

mately follow the so-called empirical dipole form (see Figure 2.4)
 −2
Q2
GpE ' GD = 1+ (2.11)
0.71 (GeV/c)2
GpM ' µp GD (2.12)

Besides the Rosenbluth separation, another experimental technique used to

 24

2.5

2
GpM
Form Factor Value (n.m.)

1.5

1
GpE
0.5
GnE
0

-0.5

-1
GnM

-1.5

-2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
2
Momentum Transfer (GeV/c)

FIG. 2.4: Plot of the dipole parameterizations of G pM , GpE , and GnM and the Galster
parameterization of GnE .

extract GpE is to measure the recoil polarization transfer in polarized electron-

unpolarized proton scattering. This technique was originally suggested in the 1970’s

[22]. This polarization transfer method is less prone to some systematic uncertainties

compared to the Rosenbluth separation. Considering only one-photon exchange, a

polarized electron beam transfers its polarization to the recoil proton with two non-

zero components, Pt , perpendicular to, and Pl , parallel to, the proton momentum

in the scattering plane. These components are related to the electric and magnetic

form factors

Pt ∝ GpE GpM and Pl ∝ Gp2

M. (2.13)

Putting in the kinematical variables, the ratio G pE /GpM can be expressed as

 
GpE Pt (Ee + Ee0 ) θe
p = − tan (2.14)
GM Pl 2Mp 2
 25
where Ee is the beam energy, Ee0 is the scattered electron energy, Mp is the proton

mass and θe is the angle of the scattered electron.

FIG. 2.5: Plot of µp GpE /GpM vs. Q2 for Rosenbluth separation technique (triangle sym-
bols) [23–28], and polarization transfer measurements (blue circles and red squares)
[30, 31] results. The systematic uncertainties of the JLab data are shown at the bottom
of the plot.

Surprisingly, the results from these polarization transfer experiments [29–31]

disagree with the Rosenbluth separation [23–28] as shown in Figure 2.5, especially
2 2 2 GpE
for Q larger than 1.0 (GeV/c) . At low Q , p ∼ 1 as measured with both
GM
 26
techniques, but the ratio falls with increasing Q 2 using the polarization transfer data.

Theoretical work is ongoing to understand the source of this discrepancy [32, 33].

Since G0 operates at momentum transfers less than 1.0 (GeV/c)2 this discrepancy

should have little effect.

The neutron form factors are defined in an analogous way. They are more

difficult to measure since there are no free neutron targets. Typically deuterium

and helium targets in quasi-elastic scattering are used and the theoretical and/or

measured proton contribution is subtracted off. GnM can be parameterized as (see

Figure 2.4)

GnM ' µn GD (2.15)

where µn is the neutron’s magnetic moment. GnE is a more difficult situation since the

neutron’s net electric charge is zero. Therefore, in the static limit, G nE (Q2 → 0) = 0.

The small value of GnE at Q2 = 0 compared to GnM makes the Rosenbluth separation

[34] challenging for the neutron. It has been found that G nE is approximated by the

so-called Galster parameterization [35] (see Figure 2.4)

−µn τ GD
GnE = . (2.16)
1 + 5.6τ

Recent experiments using recoil polarimetry [36–38] and polarization transfer [39–41]

with light nuclei in quasi-elastic scattering have been employed to measure G nE

without using the Rosenbluth separation. Interference between the magnetic and

electric scattering amplitudes produces an asymmetry that can be measured and

related to a ratio of GnE and GnM . These techniques have several advantages, one of

the most important being that many of the systematic errors cancel in these ratios.

Comparing the Rosenbluth results and these recent experiments, they seem to be

in agreement.
 27

FIG. 2.6: Tree level Feynman diagram for electron-proton weak neutral current scattering.

2.3 Weak Neutral Form Factors

Besides electromagnetic scattering via the photon exchange, electron-nucleon

scattering has a mixture component that can occur via the weak neutral current

interaction (see Figure 2.6). The scattering amplitude is

GF
MZ = √ j Zµ JµZ . (2.17)
2

Since the weak interaction is a V-A theory [42], the proton current is more compli-

cated than the electromagnetic case. It is given by:

 
Z Z σµν q ν Z e
Jµ = ū γµ F1 + i F + γµ γ5 GA u. (2.18)
2Mp 2

The weak neutral form factors F1Z and F2Z are analogous to the electromagnetic

form factors F1γ and F2γ . Besides the vector form factors, in weak scattering there

is also an axial contribution, represented by the electron-nucleon axial vector form

factor, (GZA ). This form factor can be written in general as (see Figure 2.7)

GeA = GZA + ηFA + RA

e
(2.19)

where

GZA = −τ3 GA + GsA (2.20)

 28


 







 




 γ


 



 

PV
Z0 γ 








 







 


0
Z

electron proton electron proton electron proton

FIG. 2.7: Feynman diagrams for contributions to the electron-nucleon axial coupling.
The first diagram describes a single Z 0 exchange. The second diagram describes the
parity-violating photon exchange which contributes to the nucleon anapole moment. The
third diagram is an example of radiative corrections that must be taken into account.

corresponding to the weak axial form factor associated with Z 0 exchange. It is

measured at Q2 = 0 in neutron beta decay with a value of GA (0) = 1.2601 ± 0.0028

[43], and its Q2 dependence is measured in neutrino-proton scattering [44]. The

strangeness portion, GsA (Q2 ) is measured at Q2 = 0 in deep inelastic scattering [45].

FA is the nucleon anapole moment which describes the parity-violating coupling of

the photon to the nucleon. This term is enhanced by a factor

√
8π 2α
η= (2.21)
1 − 4sin2 θW
where θW [43] is the weak mixing angle. This coupling can occur when two quarks in

the proton interact weakly while interacting with the scattered photon. This term

is unique to parity-violating interactions with charged particles such as electrons.

e
RA represents the radiative corrections that have been folded into G eA .

2.4 GSM , GSE and G0: Quark Decomposition of Form

Factors

It is possible to decompose the electromagnetic and neutral weak form factors

in terms of individual quark flavors. In this way the strange quark content of the

nucleon can be isolated.

 29
The SU(3) flavor decomposition (that is working with the u, d, and s quark

basis) of the proton electromagnetic form factors can be written for each form fac-

tor. This is the sum of the individual quark flavor form factors weighted by the

electromagnetic charge of that flavor quark

2 u 1 1
Gγ,p
E = GE − GdE − GsE (2.22)
3 3 3
2 u 1 d 1
Gγ,p
M = GM − GM − GsM . (2.23)
3 3 3

By analogy the same can be done for the weak form factors, but this time the quark

flavor form factors are weighted by the weak charge of that flavor quark:
     
8 2 4 2 4 2
Z
GE = 1 − sin θW GE + −1 + sin θW GE + −1 + sin θW GsE (2.24)
u d
3 3 3
     
8 2 4 2 4 2
GM = 1 − sin θW GM + −1 + sin θW GM + −1 + sin θW GsM (2.25)
Z u d
3 3 3

where the parameter sin2 (θW ) is known with a high degree of accuracy. Its on-shell

value is given by [43]:

sin2 (θW ) = 0.23117 ± 0.00016. (2.26)

Assuming charge symmetry between the up and down quarks in the neutron

and proton:

Gu,p d,n
E = GE Gu,p d,n
M = GM (2.27)

Gd,p u,n
E = GE Gd,p u,n
M = GM (2.28)

allows for the u and d quark contributions to be eliminated. The weak form factors

can then be written as

GZ,p 2 γ,p γ,n s

E,M = (1 − 4sin θW )GE,M − GE,M − GE,M (2.29)

solving for GsE,M yields

GsE,M = (1 − 4sin2 θW )Gγ,p γ,n Z,p

E,M − GE,M − GE,M . (2.30)
 30
In the Q2 = 0 limit, one may assume that

GsM = µs (2.31)

GsE = 0 (2.32)

where µs is the strange magnetic moment of the nucleon in terms of the nuclear mag-

neton. Relation 2.32 follows from the fact that the nucleon has no net strangeness.

Strange quarks could contribute to GpE and/or GpM . In order to contribute to

the charge form factor, there must be a spatial polarization between strange and

anti-strange quarks. In order to contribute to the magnetic form factor, the strange

and anti-strange quarks must have spatial distribution of angular momenta.

An alternative formulation is found, again, by considering only SU(3) flavors,

which allows one to write an expression for the SU(3) flavor singlet form factor

G0E,M (and hence the name of this experiment). This form factor characterizes the

difference between the corresponding electromagnetic and weak form factors for the

nucleon. In terms of the proton form factors

  
0,p 1 2 p,γ p,Z
GE,M = 4 − sin θW GE,M − GE,M . (2.33)
2
This can be written in terms of the quark flavors
1 h u,p i
G0,p
E,M = G E,M + G d,p
E,M + G s,p
E,M . (2.34)
3
It should be noted that the charge symmetry assumption (see Equations 2.27

and 2.28) between the up and down quarks in the neutron and the proton demands

a closer look. Under this charge symmetry, the up and down quark wave functions

in the proton describe the down and up quark wavefunctions in the neutron. This

charge symmetry is broken by the mass differences between the up and down quarks

and from electromagnetic effects. This charge symmetry breaking manifests itself

as an additive term to Equation 2.29

GZ,p 2 γ,p γ,n s u,d

E,M = (1 − 4sin θW )GE,M − GE,M − GE,M − GE,M . (2.35)
 31
The effects of Gu,d 2
E,M along with any possible Q dependence has been investi-

gated using non-relativistic quark models [46]. In these non-relativistic quark mod-

els, the largest effect has been calculated to alter the values of the electromagnetic

form factors by less than 1%.

2.5 Elastic Parity-Violating Electron-Nucleon

Scattering

In order to measure the strange electric and magnetic form factors of the proton

it is necessary to measure the weak neutral form factors. Since the electromagnetic

interaction dominates over the weak interaction, direct measurement of the weak

neutral form factors is challenging. In order to observe the weak neutral form

factors parity-violating electron-nucleon scattering may be employed.

In elastic electron-nucleon scattering, polarized electrons are scattered off of

the unpolarized nucleon. A polarized electron can come in two states: right handed

electrons (denoted by a ‘+’ or ‘r’) are electrons whose spin and momentum vectors

are parallel and left handed electrons (denoted by a ‘-’ or ‘l’) are electrons whose

spins and momentum vectors are anti-parallel. The scattering interaction can occur

via the exchange of a virtual photon or a virtual Z 0 . This gives rise to two scattering

amplitudes, MZ and Mγ . The total scattering amplitude, M, is given by

M = M γ + MZ . (2.36)

The cross section is proportional to the square of the scattering amplitude

|M2 | = |M2γ | + 2Re{(Mγ )∗ MZ } + |M2Z |. (2.37)

The weak interaction violates parity. A parity transformation involves an inversion

of a physical system through the origin of coordinates. The handedness of a particle,

 32
the relative orientation of its momentum and spin, reverses under mirror reflections.

Since the weak interaction violates parity, its scattering amplitude, M Z , will be

different for scattering of left or right-handed electrons. An asymmetry, A, can be

formed by taking the difference over the sum of helicity dependent scattering cross

sections, σR and σL

σR − σL
A= . (2.38)
σR + σL

Writing in terms of the helicity-dependent scattering amplitudes gives:

|Mγ + MZ |2R − |Mγ + MZ |2L

A = (2.39)
|Mγ + MZ |2R + |Mγ + MZ |2L
M∗γ MZ
= 2 . (2.40)
|Mγ |2

Using Equations 2.5 and 2.17 in Equation 2.40 yields

  γ,p Z,p γ,p Z,p γ,p
−GF Q2 GE GE + τ GM GM − 12 (1 − 4sin2 θW )0 GM GeA
A= √ (2.41)
4 2πα (Gγ,p 2 γ,p 2
E ) + (GM )

where

1
 = (2.42)
1 + 2(1 + τ )tan2 ( 2θ )
p
0 = τ (1 + τ )(1 − 2 ). (2.43)

The electromagnetic nucleon form factors have been measured and are fairly well

known (see Section 2.2) and these values can be used as inputs in the above equation.

Measurements of different linear combinations of weak neutral form factors must be

done. By selecting different kinematics, one is able to access different combinations

of weak neutral form factors. Three independent measurements are needed for a

complete determination of the three weak form factors: G Z,p Z,p e

M , GE , and GA . This

can be done by varying the kinematical variables  and  0 at a fixed Q2 (similar

to Figure 2.3). Forward angle measurements correspond to large  and small  0

while backward angle measurements are most sensitive to large  0 and small . Note
 33
that the axial term is suppressed relative to the vector electric and magnetic terms

because of the factor (1 − 4sin2 θW ) ≈ 0.08. For the third measurement to get to

GeA , it appears most effective to measure the asymmetry in quasi-elastic scattering

off of deuterium.

In this quasi-elastic scattering off of deuterium, the impulse approximation can

be used. This approximation is one in which the deuteron as a whole is described

as a linear superposition of the single nucleon observables. This allows one to write

the asymmetry as

σ p Ap + σ n An
AQE = (2.44)
σd

where σp(n) is the cross-section for elastic e-p(n) scattering, σd = σp + σn , the

asymmetry on the proton Ap is given by Equation 2.41 and the asymmetry on the

neutron is given by
 
−GF Q2 Gγ,n Z,n γ,n Z,n 1 2 0 γ,n Z,n
E GE + τ GM GM − 2 (1 − 4sin θW ) GM GA
An = √ . (2.45)
4 2πα (Gγ,n 2 γ,n 2
E ) + (GM )

Effects associated with the deuteron wave function and different potential models

have been explored in [47] and were shown to be quite small. Corrections for final

state interactions and exchange currents must be taken into account for a reliable

separation of the axial and magnetic form factors. These issues have been addressed

in [48].

2.6 Strange Form Factors and Physics

Asymmetry

Using Equations 2.41 and 2.29, the parity-violating asymmetry, to leading or-

der, can be written in terms of strange (unknown), axial (unknown), proton (known)
 34
and neutron (known) form factors. This asymmetry equation can be written as
 
−GF Q2 1
A = √ 2 γ,p (2.46)
4 2πα (G )E + τ (GM )2
γ,p

× Gγ,p 2 γ,p γ,n γ,p 2 γ,p

E [(1 − 4sin θW )GE − GE ] + τ GM [(1 − 4sin θW )GM

− Gγ,n 2 0 γ,p ep γ,p s γ,p s

M ] − (1 − 4sin θW ) GM GA + GE GE + τ GM GM .

It is useful to write Equation 2.46 as a combination of the three unknowns, G sE , GsM ,

and GeA :

A = η + ξGsE + χGsM + ψGeA (2.47)

where
 
−GF Q2 1 γ,p 2 γ,p γ,n
η = √ γ,p 2 γ,p 2 GE ((1 − 4sin θW )GE − GE )
4 2πα (G ) + τ (G M )
+ τ GM ((1 − 4sin θW )GM − Gγ,n
γ,p 2 γ,p
M ) (2.48)
  γ,p
−GF Q2 GE
ξ = √ γ,p (2.49)
4 2πα (G )2 + τ (GM )2
γ,p
 
−GF Q2 τ Gγ,p
M
χ = √ γ,p (2.50)
4 2πα (Gγ,p )2 + τ (GM )2
 
−GF Q2 (1 − 4sin2 θW )0 Gγ,p
ψ = √ γ,p . (2.51)
4 2πα (Gγ,p )2 + τ (GM )2

From Equations 2.48 and 2.51, there is expected to be a non-zero asymmetry,

even if the strange quarks do not contribute to the properties of the nucleon. This

non-strange quark asymmetry is determined by the non-strange electric, magnetic,

and axial form factors of the nucleon and by the electroweak parameter sin 2 (θW ).

2.7 Survey of Previous Experiments

HAPPEX I [14] was an experiment that ran in 1998 and 1999 at Jefferson Lab.

It measured the parity-violating electroweak asymmetry in elastic electron-proton

scattering. In order to be most sensitive to the strange electric form factor G SE ,

 35
kinematics of θ = 12.3◦ and Q2 =0.477 (GeV/c)2 were used. A 3.36 GeV polarized

electron beam scattered off an unpolarized 15 cm long liquid hydrogen target. In the

first run, the beam current was 100 µA with an average beam polarization of 38.8 ±

2.7% from a bulk GaAs polarized source. In the second run, the beam current was 35

µA with a beam polarization of about 70% from a “strained” GaAs polarized source.

A Compton polarimeter was used to continuously measure the beam polarization.

Two spectrometers in Hall A with a small acceptance ∆Ω = 5.5 msr detected the

scattered electrons at this extreme forward angle. The detected signals were inte-

grated from a lead/lucite scintillator calorimeter which only accepted elastically scat-

tered particles. The measured asymmetry was A = −15.05 ± 0.98(stat) ± 0.56(syst)

ppm. Due to limited kinematics, HAPPEX I was unable to perform the Rosenbluth

separation to disentangle GSE and GSM . Thus HAPPEX I effectively measured a

combination of strange electric and magnetic form factors at their kinematics of

GSE + 0.39GSM = 0.014 ± 0.020 ± 0.010 (2.52)

which is consistent with zero contribution of strange quarks, or a cancellation by

GSE and GSM at these kinematics.

The SAMPLE experiment [13] was performed at the Bates Linear Accelerator

Center in 1995-1999. It measured the parity-violating asymmetry in elastic scat-

tering of polarized 200 MeV electrons from a 40 cm long unpolarized hydrogen

(deuterium) target. The beam polarization was about 37% from a bulk GaAs po-

larized electron source. The scattered electrons were detected in a large solid angle

of approximately ∆Ω ≈ 1.5 sr using air Cerenkov detector at backward angles of

130◦ to 170◦ . Since measurements were made at backward angles, the asymmetry

is most sensitive to GsM and GeA . The scattered electrons with an average Q2 ∼ 0.1

(GeV/c)2 were detected by Cerenkov light produced in air absorbed by ten photo-

multiplier tubes via ten mirrors positioned around the beam axis. The measured
 36

FIG. 2.8: SAMPLE uncertainty bands of GsM vs. GsE at Q2 = 0.1 (GeV/c)2 . Also
shown is the uncertainty in a theoretical calculation of G eA by Zhu [49] at the same
momentum transfer. The smaller ellipse (yellow) corresponds to a 1σ overlap between
the SAMPLE results on hydrogen and the theoretical calculation. The larger ellipse (ma-
genta) corresponds to a one sigma overlap between the SAMPLE hydrogen and deuterium
results.

asymmetry for the proton was A = −5.61 ± 0.67 ± 0.88 ppm. Making measurements

on both hydrogen and deuterium allowed the SAMPLE collaboration to measure

the electron-proton axial form factor, GeA = −0.83 ± 0.26. This in turn, at the

SAMPLE kinematics, allows one to isolate the strange magnetic form factor. The

strange magnetic form factor was found to be

GsM (Q2 = 0.1) = 0.37 ± 0.20(stat) ± 0.26(syst) ± 0.07(theory) (2.53)

which is consistent with zero.

Using Chiral Perturbation Theory [50] to extrapolate to Q 2 = 0 yields a con-

 37
tribution to the magnetic moment due to strange quarks:

µs = 0.37 ± 0.20 ± 0.26 ± 0.15. (2.54)

FIG. 2.9: The PVA4 (represented as ’A4’ in the plot) results are shown as a solid line
representing all possible combinations of GsE + 0.225GsM at Q2 = 0.230 (GeV/c)2 . The
densely hatched region represents the PVA4 uncertainty. The HAPPEX result is shown
as a dashed line representing all possible combinations of G sE + 0.395GsM at Q2 = 0.477
(GeV/c)2 . The less densely hatched region represents the HAPPEX uncertainty.

The PVA4 experiment [51] was performed at MAMI at Mainz. It measured the

parity-violating asymmetry in elastic scattering of 854.3 MeV polarized electrons

from an unpolarized hydrogen target at a Q2 of 0.230 (GeV/c)2 . The scattered

electrons between 30◦ and 40◦ were detected by a large acceptance (∆Ω = 0.62

sr) calorimeter. This calorimeter consists of 512 PbF2 crystals. The beam current

was 20 µA with a polarization of about 80% produced using a “strained” GaAs

polarized electron source. The polarization was measured by both Compton and
 38
Moller polarimeters. The measured asymmetry was A = −5.44 ± 0.54(stat) ±

0.26(syst). The extracted linear combination of the strange and magnetic form

factors was found to be

GsE + 0.225GsM = 0.039 ± 0.034. (2.55)

This result is 1.2σ away from zero. It should be noted, with caution, that [51]

averages the PVA4 result (Q2 =0.230 (GeV/c)2 ) with the HAPPEX result (Q2 =0.477

(GeV/c)2 ) at different kinematics. They use a value for GsM based on theoretical

estimates and SAMPLE’s result to obtain GsM (0.1 < Q2 < 0.5 (GeV/c)2 ) = 0.066 ±

0.26 which leads to a non-zero contribution of strange quarks to the strange electric

form factor.
CHAPTER 3

Experimental Apparatus

The G0 engineering run was performed in Hall C at Jefferson Lab in the Fall of

2002 through January 2003. A second engineering run is scheduled for the Fall 2003

with the forward production run occurring in early 2004. This chapter will describe

the G0 spectrometer, Jefferson Lab’s electron accelerator, the G 0 beam structure,

and polarimetry.

3.1 The G0 Magnet

The superconducting magnet system (SMS) is the largest component of the G 0

apparatus. The purpose of the magnet is to bend elastically charged particles from

the target of the same momentum onto the focal plane detectors, independently of

the interaction point along the target length (see Figure 3.1). The magnet produces,

at a maximum, a 1.6 T·m field. This toroidal magnet consists of 8 superconducting

coils in a single cryostat. Each coil is made from 144 turns of integrated supercon-

ductor. The superconducting coils in series are cooled by four parallel liquid helium

convection circuits. Two additional parallel cooling paths are used to cool the su-

perconducting electrical buss through which power is supplied to the coils. The
39
 40

FIG. 3.1: Elastically scattered protons, of the same momentum, are focused onto the
Focal Plane Detectors (FPD). This is independent of where along the target the scatter-
ing occurs. This is shown schematically by looking at three different momentum values
(denoted by red, green, and blue) at opposite ends of the target being focused on the de-
tectors. The collimators are used to reduce the background and to set the acceptance of
the detectors.

coils, electrical buss, and collimators (which define the spectrometer acceptance and

provide shielding for the target and detectors) make up what is called the cold mass.

This sits inside a stainless steel shell and makes up the bulk of the magnet. A liquid

nitrogen shield surrounds the cold mass. The cryogens, lN 2 and lHe are fed into

the magnet via a manifold at the bottom of the magnet from reservoirs located in

the control dewar at the top of the magnet. The cryogens percolate back up to

the reservoirs through the cooling circuit, losing density and absorbing power along

the way. Aluminum end caps cover both the front and back of the main shell that

houses the cold mass. There are eight trapezoidal holes, 0.51 m 2 in area, on the

downstream end cap that are covered by an 0.020 inch titanium plate. These are

the exit windows that provide a path of low energy loss and multiple scattering from

particles emanating from the target to the detectors. Mounted to the beam line at

the upstream magnet end cap is the target service module that contains the target

and its positioning mechanism. Valves are connected upstream of the target service
 41
module and at the downstream end of the exit beam line to be isolated. When these

valves are closed, the entire experimental apparatus can be disconnected and moved

to the left of the beam line. This is necessary since the G 0 experimental program

will take several years to complete and will not be the only experiment running in

Hall C, thus the G0 apparatus must be able to be removed from the Hall C beam

line.

3.2 The G0 Hydrogen Target

FIG. 3.2: A schematic of the G0 target. The G0 target has been designed to minimize
false asymmetries. The target cell is 20 cm long and filled with liquid hydrogen. The
target operates at 450 watts.

The G0 experiment uses a liquid hydrogen target (see Figure 3.2) based on the

SAMPLE experiment’s design [15]. The target is optimized to reduce energy loss

along the scattered particles’ path and to minimize density fluctuations, caused by

up to 250 W of beam power deposited on the target. The target is connected to a

cryogenic loop to recirculate and cool the liquid hydrogen. The target and cryogenic

loop sit inside the 77K liquid nitrogen shield of the SMS. The hydrogen cell is 20

cm long and 5 cm in diameter. The radius of curvature of the endcap of the target

is 7.6 cm. The outer wall and endcap are 7.0 ± 0.5 mils of aluminum. A manifold
 42
inside the lH2 cell has been designed to direct the fluid flow down the center of the

target cell and back near the cell walls. The target is fronted by a helium cell that

has two purposes.

• To first order, it eliminates variations in target thickness with beam position

by matching the radius of curvature of the entrance and exit windows of the

hydrogen cell.

• It extends the hydrogen cell beyond the part of the cryoloop which is not sym-

metric about the beam axis allowing the target-beam interaction region to be

axially symmetric.

The heat exchanger uses gaseous helium with an inlet temperature of about 15 K

and pressure of 20 atm, supplied by the Jefferson Lab End Station Refrigerator.

With a beam current of 40 µA impinging on it, the target requires a flow of 17 g/s

of 15 K coolant. The coolant flows inside the finned tube and liquid hydrogen flows

over the fins on the outside of the tubing. The heat exchanger has 2 layers of finned

tubing through which the liquid hydrogen flows in parallel. The heat exchanger

removes about 450 W of heat from the target with 250 W coming from 40 µA beam

heating, 100 W from the pump motor and 100 W from connections to the outside

world.

Two internal heaters in the G0 cryogenic target are used to maintain a constant

heat load and/or temperature in the target. The high-power heater consists of

three heater coils in parallel. The high-power heater is a feedback loop that reads a

signal proportional to the beam current and calculates the heat load of the beam,

and then changes the heater such that there is a constant heat load on the target

and thus ensures a constant temperature in the cell and coolant loop. A second

80 W low-power heater works in parallel with the high-power heater to maintain

constant target temperature against small time-dependent drifts from other heat
 43
load sources (such as the motor, coolant inefficiencies and radiant heating). The

5.5mm Small Hole Target

11.1mm Large Hole Target 4.9mm Carbon Target

3.1mm Al Frame Beam

Target Cell

FIG. 3.3: Schematic of the dummy target looking downstream of the beam. There are
three dummy targets located above the target cell. Two of these dummy targets (large and
small hole) are used in studying beam halo. The third dummy target is a 12 C radiator
target.

target is mounted onto an aluminum frame that is 3.05 mm thick. On this aluminum

frame there are three dummy targets: two blank hole targets for studying beam halo
12
and one C target (see Figure 3.3). The big hole target has an inner diameter the

same as the hydrogen target but with an outer diameter of 19.05 mm and a thickness

of 3.912 mm to match the radiation length of 1 inch of aluminum. The small hole is
12
5.46 mm diameter. The C target sits behind a hole of 9.562 mm diameter in the
12
aluminum frame. Five slabs of ∼ 1 mm thick carbon make up the 4.9 mm thick C

target. The entire target can be removed from the beam trajectory for diagnostic

purposes and can be warmed independently of the spectrometer.

One issue with the target that must be addressed in a parity-violation experi-

ment is that of target boiling. If the target boils (e.g. density fluctuations due to a

high amount of energy being deposited on the target by the electron beam) this will

add noise to the experimental system. This noise will then increase the statistical

error of the experiment. This additional noise from target boiling, (σ b2 ), is added in
 44

Asym_width_raw (ppm) σ0 : 299.371 +/- 7.183, σρ:612.091 +/- 10.855, exp: 1.614 +/- 0.067
σ0 : 326.193 +/- 3.343, σρ:580.686 +/- 10.024, exp: 2
800

600

400

1 1.5 2 2.5 3
Raster (mm)

FIG. 3.4: Fluctuations in the statistical width (ppm) as a function of raster size for target
boiling studies. These data come from measurements with the LUMI detector.

quadrature to the statistical width due to counting statistics, (σ 02 ):

σ 2 = σ02 + σb2 (3.1)

where the overall statistical width is given by σ 2 . The beam’s noise σb2 is parame-
σρ2
terized by rx
where r is the raster size and x is the exponent which can either be

fixed at 2 or allowed to be variable. The LUMI monitors (see Section 3.7) are used

to measure the target boiling because they are at extreme small angles relative to

the beam line. This means that the LUMIs see an electroweak asymmetry of zero

and their detected rate is very high allowing one to make faster measurements than

with the Focal Plane Detectors. A test that is performed is to boil the target by

depositing more energy on the target by changing the raster (area) of the electron

beam (see Figure 3.2). The LUMIs are then used to measure the contribution to

the statistical width from the target. Two separate fits were performed to the data

and the noise contribution at the G0 raster size of 3 mm was found to be about
 45
300 ppm. This is small compared to the statistical width due to counting statistics

(found to be about 1400 ppm averaged over all the detectors). Taking the ratio

of the measured statistical width (σmeas ) compared to the statistical width from

counting statistics (σstat ) and using the fact that the noise contribution from the

beam monitors and target are ≈ 300 ppm each:

√
σmeas 14002 + 2 × 3002
= = 1.05. (3.2)
σstat 1400

This implies that the error has been increased by 5% past counting statistics (and

thus the error is dominated by counting statistics).

3.3 The G0 Detector

The G0 detector system consists of eight octants surrounding the beam line

(see Figure 3.5). Four octants were designed and mounted by French collaborators

and four by North American collaborators (see Figure 3.6). Each octant encloses

an array of 16 scintillator pairs along the focal plane of the G 0 magnet. Each end

of a scintillator is coupled to a lightguide, at the end of each lightguide is a PMT.

These detectors are named the focal plane detectors (FPDs). The FPDs are the

only detectors necessary for the forward angle measurement in the G 0 experiment.

In this configuration, the scintillators detect elastically recoiled protons. The output

PMT signals are used to measure the Time-of-Flight of the particles between the

target and the focal plane of the magnet, providing particle identification. Each

scintillator provides a selection in momentum transfer (Q 2 ) which is supplemented

by a selection on the Time-of-Flight necessary for particle identification.

 46

FIG. 3.5: The G0 detector system, supported by an aluminum support known as the
‘Ferris Wheel‘, as seen upstream on the left picture and downstream on the right picture.
The octants are numbered 1 though 8 starting from the top and moving clock-wise. Odd
(even) numbered Octants are North American (French). The G0 target and magnet are
not pictured here; they were placed directly upstream of the Ferris Wheel.

Detector System

An octant is an array of 16 BC408 Bicron scintillator pairs located at the focal

plane of the G0 magnet. They are labeled from 1 to 16 where the larger scintillator

number corresponds to the scintillators located further away from the beam line and

further from the target. Larger scintillator number also corresponds to the larger

scintillator size and, in general, the larger measured value of Q 2 (with detectors 14

and 15 containing two Q2 points and detector 16 containing no elastic Q 2 points so

it is used to measure background). The FPDs have been divided into 16 detectors

in order to keep the individual count rate for each detector around 1 MHz. The

arc shape of the scintillator was defined to follow the iso-Q 2 phase space. The
 47

FIG. 3.6: Pictured on the left is the North American Octant 7. On the right is pictured
the French Octant 2. Both octants are shown without their light tight box.

dimensions and areas of the scintillators are given in Table 3.1. The scintillators are

paired together to increase background rejection, moreover two scintillators from the

same pair are separated by a thin material (aluminum for French and polycarbonate

for NA). Each FPD is a pair of two light pipes attached to the aluminum support

of one octant. Each light pipe is the assembly of a scintillator and two light guides

glued at each of the ends of the scintillator with photo-multiplier tubes attached to

each end of the light guides. Optical fibers are air coupled to each end (right and

left) of the scintillator. The fibers shine UV light to the scintillator and allow gain

monitoring of the PMTs. The whole assembly is enclosed inside a light-tight box.

Further discussion of the calibration and testing of the NA detectors system

can be found in Appendix C.

 48

Scintillator 1 2 3 4 5 6
Length (cm) 60.1 61.3 62.1 62.6 64.6 69.2
Length (ns) 4.4 4.3 4.3 4.4 4.3 4.8
Area (cm2 ) 177.8 237.1 254.0 238.9 293.5 353.9
Thickness (cm) 0.5 0.5 0.5 0.5(f)-1.0(b) 1.0 1.0

Scintillator 7 8 9 10 11 12
Length (cm) 74.7 80.8 88.6 95.8 104.5 112.3
Length (ns) 5.6 5.1 6.0 6.4 7.0
Area (cm2 ) 441.6 441.8 581.2 689.5 878.2 864.5
Thickness (cm) 1.0 1.0 1.0 1.0 1.0 1.0

Scintillator 13 14 15 16
Length (cm) 120.1 136.8 135.0 136.0
Length (ns) 8.0 9.5 9.3 9.3
Area (cm2 ) 871.4 1146.7 1209.9 1218.8
Thickness (cm) 1.0 1.0 1.0 1.0

TABLE 3.1: Lengths and areas of the NA G0 scintillators. The scintillators are arc
shaped. The length (in centimeters) refers to the perimeter of the inner arc. The statis-
tical precision of the measurement (in nanoseconds) is 0.3 ns. Note that the size of the
scintillator is not the same for French and North American octants. Also note that the
thickness of scintillator 1 to 3 is 0.5 cm. The thickness of scintillator 4 back (b) is 0.5
cm, the thickness of scintillator 4 front (f ) is 1 cm, and the thickness of scintillator 5 to
16 is 1 cm.
 49
Light Pipe

The North American light guides are made out of Lucite. The French light

guides are made of Polymethyl Matacrylate (PMMA). The North American de-

tectors used Philips XP2262B (12 stages) PMTs powered by custom made Zener-

resistor bases [52]. The French detectors used Photonis XP2282B04 (8 stages) PMTs

powered by a custom made Zener base with an amplifier with a gain of 20. The

light pipes are wrapped in µ-metal to reduce any remaining fringe magnetic field

influence from the G0 magnet.

Light-Tight Box

The North American light pipes are individually wrapped in aluminized mylar.

The French light pipes are wrapped in aluminum foil. These light pipes are not

light-tight. For operating the PMTs, the assemblies are enclosed in a light-tight box

known as an octant. Each octant is roughly 2 m by 2 m by 3 m long. The octant

consists of aluminum plates on four sides with a black cover on the remaining fifth

side. The inside of the octant support as well as the aluminum plates have been

covered by black Tedlar to minimize reflection of light leaks. The front fifth side of a

North American (French) octant is covered by a sheet of Herculite (plastic sheeting).

No access is possible to the inside of the octant without destroying the integrity of

the light tight-box.

Detector Superstructure Support Frame

The North American and French octants “plug in” to a detector superstruc-

ture known as the “Ferris Wheel” (see Figure 3.5). Within the Ferris Wheel, the

North American octants are located at the 12 o’clock, 3 o’clock, 6 o’clock and 9

o’clock positions. The French octants are located in between the North American
 50
octants. Each octant is attached to the Ferris Wheel by means of three bolts on the

downstream face of the structure. The Ferris Wheel is about 7 m above the floor

at its highest point with the symmetry axis (the beam line) 4 m from the floor.

The detectors are shielded from the beam line by 9.525 cm of lead and 15.716 cm

of poly-boron. A rail system permits the Ferris Wheel to be retracted from the

Superconducting Magnet System. This is done so that the G 0 apparatus can easily

be moved out of the beam line so other experiments in Hall C can be performed.

3.4 The G0 Beam

G0 requires a beam current of 40 µA pulsed at 31.2 MHz instead of the typical

CEBAF 499 MHz. This produces a charge bunch that is 16 times larger than the

normal operating mode bunch. These high charge bunches require special beam

optics due to space charge effects. Each micropulse contains on average 1.28 pC,

yielding an average current of 40 µA.

In order to measure the small asymmetry between the two helicity states the

same experimental conditions must exist for the two different helicity states. Devi-

ations in the experimental conditions between the two helicity states can induce a

false asymmetry.

The number of detected particles measured in each helicity state depends on

the electron-proton scattering cross section. This cross section is sensitive to energy,

beam position, and beam angle. Any systematic difference between these properties

for the two helicity states can manifest itself as a false asymmetry.

Helicity-correlated differences typically originate in the injector due to the way

the circularly polarized laser light emerges from the Pockels cell. The beam will not

be perfectly circularly polarized but will be elliptically polarized due to some residual

linear component; this, coupled to the analyzing power of the cathode, produces
 51

Aq (ppm) ∆x (µm) ∆y (µm) ∆θx (µrad) ∆θx (µrad) ∆E/E (×106 )

specified 1 20 20 2 2 10
measured 1 7 5 2 2 6.0

TABLE 3.2: The measured run averaged helicity-correlations for January 18-26, com-
pared to the G0 specifications for 700 hours of data taking.

helicity-correlated differences in the beam. Helicity-correlated laser motion coupled

to spatial variations in the quantum efficiency of the cathode will also produce

helicity-correlated differences in the beam. Beam scraping in the injector (and

possibly in the main machine) can cause helicity-correlated beam differences.

Charge asymmetries1 must also be minimized since charge asymmetries can in-

duce helicity-correlated differences in other beam parameters. Via beam loading,

charge asymmetries become energy differences. Energy differences, in turn, man-

ifest themselves as position differences due to achromatic transport through the

accelerator.

The G0 experiment has specified how large these helicity-correlated beam dif-

ferences can be allowed to be (see Table 3.2). This is because the size of these

quantities determines how accurately the helicity-correlated beam properties are

measured. This in turn determines the errors on the correction procedures to cor-

rect for the helicity-correlated beam properties.

3.4.1 The G0 Beam Structure

The Time-of-Flight technique requires that the polarized electron beam be

pulsed at 31.2 MHz rather than the typical CEBAF 499 MHz. This translates

into buckets of polarized electrons arriving every 32 ns at the target. The helicity

of the beam is flipped at 30 Hz. One illustration of why this time scale is chosen
1 Q+ −Q−
The charge asymmetry is defined as Aq = Q+ +Q−
 52

A Quartet(+--+) A Quartet(+--+)

+ - - + - + + -

1/30 Hz = 33 ms 500 µ s

FIG. 3.7: The G0 beam structure. The G0 beam is pulsed at 31.2 MHz. The G0 beam
flips helicity every 30 Hz, known as a macropulse (MPS). The G0 beam is divided into
quartets of four MPSs. The first MPS is chosen pseudo-randomly with the next two MPSs
the complement of the first MPS and the fourth MPS is the same as the first MPS.

can be seen if there is 60 Hz noise in a power supply causing a discriminator level

to oscillate up and down at 60 Hz. Then during one half of the cycle more counts

would be collected than in the other cycle. If the spin flipping was not a multiple

of the 60 Hz period, there would be non-statistical fluctuations from one data col-

lection period to the next. By holding each helicity state for 1/30 s, the experiment

averages over any changes in the experiment caused by the 60 Hz noise and any

of its harmonics. 30 Hz is also a quiet frequency in regards to noise, with higher

frequencies typically populated by noise from electronics and lower frequencies sen-

sitive to mechanical vibrations. Each 33 ms helicity state of the beam is referred to

as a macropulse (MPS). The helicity is reversed during a 500 µs interval (see Figure

3.7). The macropulses are grouped into quartets with the sign of the first MPS

being chosen pseudo-randomly. The next two MPSs are the helicity complement of

the first MPS. The last MPS helicity in the quartet is the same as the first MPS

helicity. The beam helicity is flipped quickly to insure that short-lived changes in

the experimental conditions are experienced by both helicity states.

Since there is a 500 µs settle time for the Pockels cell, this scheme allows the
 53
phase of each new helicity states to slip with respect to the 60 Hz power cycle.

This phase slip is beneficial to the experiment. This will allow the helicity states to

precess through all the phases and thus sample different aspects of the 60 Hz line

noise.

3.5 The CEBAF Injector and the Polarized Source

This section describes the various pieces of equipment that are in the injector

and associated with the polarized source. A description of the helicity devices on

the polarized source laser table in the injector are covered along with a description

of the performance of the feedback systems that use these helicity devices. The G 0

TIGER laser used to excite the photo-cathode along with the GaAs photo-cathode

are also described in this section.

3.5.1 The G0 TIGER Laser

The G0 laser, known as the TIGER, is a Ti-Sapphire laser. Ti:Sa lasers can

achieve both high current and high polarization, unlike the typical diode lasers

used at CEBAF that can achieve either high current or high polarization but not

both. The TIGER laser is a commercial laser produced by Time-Bandwidth. It

has passive mode-locking to reference the 31.2 MHz RF source. The TIGER laser

produced more than 300 mW of power at a wavelength of 840 nm. The TIGER

laser is tunable from 770-860 nm.

 54

FIG. 3.8: Schematic of the CEBAF polarized source laser table in the injector. This
schematic shows the path the laser light produced by the G0 TIGER laser follows. Im-
portant optical elements to note are the IA cell used for controlling intensity differences
and the PZT used to control helicity-correlated position differences. There is a quadrant
photo-diode (QPD) that is used to understand the optical elements upstream of the QPD.
The insertable halfwave plate is a device to passively change the handedness of the laser
light and thus used as a diagnostic for understanding helicity-correlated noise in the sys-
tem. The helicity Pockels cell is used to flip and produce circularly polarized light before
striking the photo-cathode.

3.5.2 Optical Elements

Intensity Attenuator Cell

The Intensity Attenuator (IA) cell is the central device in the injector used

by the charge feedback system from Hall C to null the charge asymmetry without

inducing large helicity-correlated beam differences. The IA cell is an electro-optic

modulator (see Appendix B for more details). This device modulates the laser light

in a helicity-correlated manner before it reaches the photo-cathode. The IA is made

 55
up of a Lasermetrics model 1059-10 Pockels cell with a 3 ◦ wedge on both faces and a

broadband anti-reflective coating at 850 nm. The Pockels cell is between two parallel

linear polarizers followed by bare mica model WPUM-10-850 tenth-wave plate for

850 nm from the Karl Lambecht Corporation.

Linearly-polarized light from the laser is transmitted through the first linear

polarizer, then is transmitted through the tenth-wave plate. This produces slightly

elliptically polarized light. This elliptically polarized light is transmitted through a

low-voltage Pockels cell (operating at about 50 V in one helicity state). If the Pockels

cell is at 0 V, then the elliptically-polarized light is transmitted unchanged through

the cell. The light is then transmitted through the second polarizer (parallel to the

first polarizer) and linear light is produced. If the IA Pockels cell is at some non-zero

voltage, more elliptically polarized light emerges from the Pockels cell. When this

elliptically polarized light is transmitted through the second parallel polarizer only

the linear component will survive, thus modulating the intensity of the laser beam.

The IA cell responded well during the commissioning, offering a large (∼ 400

ppm/V) intensity difference calibration, though it did generate large position dif-

ferences for reasons that were not always clear (see Table 3.3). The IA calibration

constants were fairly stable over time, but did require new measurements of this

constant roughly every couple of days, or whenever the insertable halfwave plate

was inserted or retracted. The main problem with the IA cell was variations in time

of the calibration constant (ppm/V). This was caused by the IA cell inducing po-

sition differences that created charge asymmetries at apertures in the injector (see

Figure 3.9). This occurred as the IA would steer the beam in a helicity-correlated

way causing the electron beam to clip along the sides of apertures in the injector.

These apertures’ diameters were subsequently increased to reduce this problem.

 56

IA Scan Comparison
IB=38 µA 12/15/02
1000
IB=20 µA 12/15/02

IB=20 µA 12/17/02
800
IB=40 µA 12/12/02

apertures IB=20 µA 12/05/02

Q A (ppm) / V

600

400

200

0 QPD 1i02 1i04 1i06 0L02 BCM0L02 0L03 Hall

100 KeV BPM 5 MeV

FIG. 3.9: The IA calibration constants (ppm/V) as a function of beam position monitor
in the injector to Hall C. A quadrant photo-diode (QPD) is placed on the laser table
to determine what is happening on the laser table. Note that the values from 1i02 to
1i06 are consistent with one another. Between 1i06 and 0L02 are two apertures. It is
hypothesized that, due to scraping on these apertures, the calibration constant did not
remain the same between the 1i06 and 0L02 regions.

Intensity Feedback

The automated intensity feedback uses the IA cell (see section 3.5.2). It was

used to vary the laser intensity in a helicity-correlated way to insure that the helicity-

correlated electron beam intensity measured in Hall C at BCM1 (Beam Current

Monitor) is kept small. BCM1 is used to measure beam currents above 20 µA and,

another BCM, BCM2, is used to measured beam currents from 5-20 µA. Typically

a change to this feedback system was made every 5 minutes (see Figure 3.10).

The nature of the feedback system is to cancel the contributions arising from the

statistical jitter of the charge asymmetry. In the absence of feedback, the statistical

reduction in the charge asymmetry goes like √1 where N is the number of 5 minute
N

measurements. With the charge asymmetry feedback active, the charge asymmetry
 57

date charge (ppm/V) x position (nm/V) y position (nm/V)

Jan. 13 ’03 -468 840 ± 100 958 ± 100
Jan. 16 ’03 -452 645 ± 323 1117 ± 497
Jan. 25 ’03 -237 46 ± 334 14 ± 241
Jan. 15 ’03 -313 N/A N/A
Jan. 17 ’03 -344 N/A N/A
Jan. 19 ’03 -348 N/A N/A
Jan. 23 ’03 -537 231 ± 219 95 ± 267
Jan. 26 ’03 -231 -58 ± 225 6 ± 183
Jan. 26 ’03 -201 -21 ± 174 -75 ± 134
Jan. 27 ’03 -197 83 ± 213 -4 ± 162

TABLE 3.3: Table of the IA calibration slopes measured in Hall C. Overall, the IA
feedback was moderately stable during the engineering run. Entries marked ’N/A’ do not
have data available.

1
convergence goes like N
.

Rotating Halfwave Plate

The rotating halfwave plate (RHWP) is also used to minimize the intensity dif-

ference. The halfwave plate operates on the principle that the light from the Pockels

cell will not be perfectly circularly polarized, but will instead be elliptically polar-

ized. Elliptically polarized light can be decomposed into two components: circular

polarized and linearly polarized. The RHWP takes the residual linear component

of the light and rotates it with respect to the cathode depending on the adjustable

rotation angle (see Figure 3.11). Scans of the halfwave plate are done on the order

of days (versus the IA, which is used to minimize the intensity difference on the

order of minutes using an automated feedback system) to find the minimum inten-

sity difference (known as the “sweet spot”). From results in Hall C, the setting for

the sweet spot of the RHWP is very stable and the null setting only needed to be

checked on a daily basis.

 58

FIG. 3.10: Typical performance of the charge feedback system using the IA cell. The
top panel shows the intensity difference as measured every 5 minutes. The bottom panel
shows the run-averaged intensity difference converging to zero as time goes on.

PZT

The PZT is a device used to correct for helicity-correlated beam position. This

device is a mirror mounted on a Thor Labs KC1-PZT kinematic mount. The PZT

is an electro-ceramic lead-zirconae-titanate ceramic which is piezoelectric. The PZT

can be set to different independent positions in both the X and Y directions. The

PZT oscillates between these set positions and the null position at the helicity-flip

frequency (each 1/30 s). This allows any helicity-correlated position differences in

the electron beam to be corrected by changing the angle that the laser beam strikes

the photo-cathode in a helicity-correlated manner.

 59

15000

10000
Intensity Asymmetry (ppm)

5000

-5000

-10000

-15000

0 20 40 60 80 100 120 140 160 180

Angle (degrees)

FIG. 3.11: Measurement of the intensity difference as a function of rotatable halfwave

plate angle. The rotatable halfwave plate is set to an angle to null the charge asymmetry.
In this plot the null point would have been chosen to be near 100 ◦ .

Another issue with the PZT was related to the orthogonality of the PZT X and

PZT Y motion. If this motion is not orthogonal, one may not be able to effectively

and independently correct for x and y position differences. For example, if the PZT

X and PZT Y motion was exactly at 180◦ then it would be impossible to correct for

both X and Y motion. Note that rotations are fine since some combination of PZT X

and PZT Y would still correct for X and Y beam motions. Hence the calibration of

the PZT requires not just measuring the calibration slope for PZT X in X (denoted

as XX) but also what affect PZT X has on Y (denoted at YX), motion and how

PZT Y not only affects Y (denoted at YY) motion but X motion (denoted as XY)

of the beam. Table 3.5.2 lists the calibration slopes of the PZT.
 60

date XX (nm/V) YX (nm/V) XY (nm/V) YY (nm/V)

Jan 24 541.3 ± 25.5 573.2 ± 25.1 -206.7 ± 25.1 150.3 ± 24.8
Jan 14 606.5 ± 37.9 843.4 ± 37.0 -253.5 ± 40.7 521.9 ± 41.1
Jan 12 -478.1 ± 33.8 -510.4 ± 34.4 356.3 ± 34.2 490.6 ± 35.2

TABLE 3.4: Table of the limited number of PZT calibration slopes in nm/V. The PZT
did not perform well during the engineering run. XX (YX) denotes the affect of PZT X
on the X (Y) beam motion. YY (XY) denotes the effect of PZT Y on the Y (X) beam
motion.

Position Feedback

This feedback system was intended to be used to vary the position of the laser

beam on the photo-cathode in a helicity-correlated manner. This feedback loop

insures that the helicity-correlated beam position as measured in Hall C by beam

position monitors G0 and G0B is kept small. The PZT’s calibration constants were

unstable on time scales of the order of minutes during the 2002-2003 G 0 commis-

sioning run. This meant that the position feedback to control the helicity-correlated

position differences was not used. Even though the position feedback was not used,

the measured mean position differences (∆X and ∆Y) were on the order of 50 nm

with a σ of 6 µm.

3.5.3 Helicity Generation

The electronics (known as the G0 Helicity Digital Controls) that determine

the helicity are located in the Injector Service Building above the polarized source.

The helicity of the electrons is determined by the helicity Pockels cell. The G 0

Helicity Controls adjusts the Pockels cell, thus controlling the helicity of the electrons

injected into the accelerator. The make-up of the G 0 Helicity Digital Controls

include a pseudo-random bit pattern stored in two 1Mbit EPROM’s and control

code, which includes multiplexers, registers, counters, and a state machine [53].
 61
The output of the G0 Helicity Digital Controls is either 1 or 0. The helicity state of

the electrons in turn is based on whether the output from the G 0 Helicity Controls

is a 1 or 0. In order to reduce helicity-correlated crosstalk and ground loops, the

helicity signal from the helicity electronics is fed to the G 0 electronics and DAQ in

Hall C via a fiber optic cable. Furthermore, this signal sent to Hall C is delayed by

eight helicity windows, thereby ensuring no in-time helicity correlation between the

helicity Pockels cell and the signals sent to Hall C.

3.5.4 Photocathode

FIG. 3.12: A schematic of the strained GaAs band structure and energy level diagram.
The circled numbers indicate the relative transition strengths.

The strained GaAs crystal acts as the photo-cathode for the polarized source.

GaAs is a direct band-gap crystal, which means that the valence band maximum

and the conduction band minimum are aligned in momentum space, allowing for

optical transitions between the energy bands that follow the angular momentum se-

lection rules for optical transitions in atoms. These transitions are shown in Figure

3.12. The strained layer GaAs photo-cathodes are produced by Bandwidth Semi-

conductor. By photo-emission, polarized electrons can be liberated from the crystal

 62
to be used in the polarized electron beam for the accelerator. In photo-emission,

an electron in the valence band (P3/2 and P1/2 levels) absorbs a circularly polarized

photon and is excited to the conduction band. The crystal has been specially treated

with Cesium Fluoride to increase the quantum efficiency of the cathode by reducing

the work function of the GaAs. This also produces a negative electron affinity to

allow the electrons to escape from the crystal. A typical quantum efficiency for the

photo-cathode is about 1% for light with a wavelength of 780 nm. The cathode is

held at a bias voltage of -100 kV in order to liberate the electrons from the cathode.

The GaAs crystal is “strained” in order to the electron beam polarization.

The maximum polarization of the strained GaAs crystal is about 80% compared

to typical polarizations of 40% for bulk GaAs crystal. This is done by growing a

thin layer of about 100 nm of GaAs on a substrate of GaAsP, which breaks the

degeneracy of the P3/2 and P1/2 levels.

An issue with the strained GaAs crystal is that a large quantum efficiency

anisotropy is produced, which in turn can produce a charge asymmetry. The light

emerging from the Pockels cell is not typically perfectly circularly, but has a resid-

ual linear component. The axis of the residual linear component of the light can

be different for the two helicity states. In this case there will be two different ori-

entations of the light’s polarization axis with respect to the “strain axis” of the

crystal. Because these axes are different, the number of electrons liberated from the

crystal is different in a helicity-correlated way, producing a charge asymmetry for

the resulting electron beam.

The crystal is located in the electron gun and is kept under very high vac-

uum. Still, residual gases can contaminate the crystal surface thereby lowering the

quantum efficiency. This residue needs to be removed periodically by performing a

“bake-out”. This is when the temperature of the gun is raised until the contaminated

molecules have evaporated away.

 63
3.5.5 The CEBAF Accelerator at Jefferson Lab

The Continuous Electron Beam Accelerator Facility at Jefferson Lab can deliver

up to 5.5 GeV of 200 µA polarized electron beam. A Wien filter in the injector sets

the launch angle of the polarization vector to compensate for g-2 precession, thereby

assuring that longitudinally-polarized electrons arrive at the experimental halls. The

polarized electrons are injected at 45 MeV into the main accelerator. The electrons

are accelerated in two linacs that make up the straight portions of the accelerator

racetrack. The linacs are connected by recirculating arcs located at both ends of

the linac. Acceleration is provided by the CEBAF-Cornell superconducting radio

frequency cavities that operate at 1497 MHz. The electron beam can take up to

5 passes around the accelerator before being sent to one of the three experimental

halls. This is done by using an electromagnetic “kicker” to send every third electron

bunch to the appropriate hall, resulting in a beam structure of 499 MHz in the halls.

3.5.6 Beam Instrumentation

Beam Current Monitors

Hall C is equipped with three different beam current monitors (BCMs). Two

of these BCMs, BCM1 and BCM2, are cylindrical resonant cavities [54, 55]. The

current is monitored by using the electron beam to excite the resonant modes in the

cylindrical waveguide. The cavity is sensitive, by design, to the TM 010 mode. Inside

the cavity is a loop antenna which couples to the resonant modes. The measured

signal is proportional to the beam current. In addition to the two cavity monitors,

Hall C is also equipped with a parametric current transformer (also known as the

Unser) [56]. The Unser has a very stable and well-measured gain but it suffers from

large unstable offsets. The Unser is not used to measure the beam current but due

to its stable gain it is used to calibrate the BCMs. The noise in BCM1 is found to
 64
be about 300 ppm which is much smaller than the statistical width of the measured

asymmetry from one quartet (see Section 4.2).

Beam Position Monitors

The beam position monitors (BPMs) used during the first engineering run in

both the injector and Hall C were standard strip-line BPMs [57,58]. These BPMs are

made up of four antennae situated symmetrically around the beam at 45 ◦ angles.

The BPMs operate at 1500 MHz and inductively pick up the RF signals of the

electron beam as it passes through the device. The signals are then amplified and

down-converted to 1 MHz. The signals for beam position can then be computed

knowing that the signal in the antenna is proportional to the beam position and

beam intensity:

BPM Antenna Signals ∝ (Beam position) × (Beam Intensity)

The relative X 0 and Y 0 beam positions can be calculated

(X+ − Xoffset+ ) − αX (X− − Xoffset− )
X0 = k (3.3)
(X+ − Xoffset+ ) + αX (X− − Xoffset− )
and similarly for Y 0 . Xoffset+(−) is the offset for the X+(−) antenna. Since the gain

of each antenna may be different, αX is a measure of the different gain between the

X+ and X− antenna. This measure of this gain difference is

X+ − Xoffset+
αX = . (3.4)
X− − Xoffset−
Since the BPMs are oriented at 45 degrees, the position of X is given by
   
 
X  1 1 −1
 = √   X 0Y 0 . (3.5)
Y 2 1 1

The position (or position difference) calculated from the beam monitors has a

certain amount of noise associated with it. This noise is due both to beam noise

and to electronic noise. The measured noise can be written as:

 65
2 2 2
σmeasured = σbeam + σinstrumental .

The instrumental noise can be found by using three BPMs along the beam line

without magnetic optics between the monitors. The first two BPMs can be used to

determine the position of the beam in the third monitor. This predicted behavior

can then be removed from the measured signal of the third BPM leaving behind

only the instrumental noise. The noise of the BPMs was found to be about 2 µm.

3.6 Halo Monitors

FIG. 3.13: Plot of normalized rates as a function of run number for the 2003 G 0 Engi-
neering run as measured by the Lucite Halo Monitor. The red (blue) points correspond
to the halo target ‘in’ (‘out’) data. The baseline value for ’good’ beam is approximately 5
Hz/µA)
 66
Electrons outside the core of the electron beam are defined as the beam halo.

This halo could be produced in a variety of processes. The halo could be formed

by space-charge effects, beam scraping on an aperture, interactions of residual gas

in the beam line, etc. This halo can cause numerous troubles. These halo electrons

could cause an increase in the dead time by producing higher singles rates in the

G0 detectors from inelastic scattering and they could also contribute to the inelastic

background. This higher unwanted radiation could also cause radiation damage to

the PMTs. It is for these reasons that the G0 experiment requires a limit on the

beam halo of 1 ppm outside a 6 mm radius of the electron beam. The halo monitors

are used to detect beam halo and time-dependent halo behavior of the electron beam

in Hall C. This system was designed to be non-invasive, so that the beam halo could

be monitored continuously during data taking. This detector system was made up

of a bare Hamamatsu 931B low gain PMT, a Phillips XP2262 PMT coupled to a

piece of 10 cm long, 5 cm diameter Lucite cylinder, and a PMT coupled to a lead

glass brick that measured 4 cm × 4 cm × 43 cm. Multiple detector types were used

as a part of a study to determine which was the optimal design for the halo detector.

A 2 mm thick carbon target with a 6 mm diameter square hole was located at the

hall C pivot. The main portion of the electron beam would pass through the hole

but halo electrons would interact with the target. Besides the ability to read out

the beam halo rates, the halo target also had a Fast Shut Down output so that if

the beam halo rate became too high the electron beam would be turned off.

From the engineering run, it was found that not all the prototype halo monitors

were suitable for use. The bare PMT was damaged by work done at the Hall C target

pivot. The lead glass halo monitor had a large amount of noise associated with it

and was insensitive to the target ‘in’ and ‘out’ differences. The Lucite halo monitor

worked the best. It was fairly insensitive to low-energy background and was sensitive

to the position of the halo target. It was found that for good beam, the Lucite halo
 67
monitor observed a baseline value of 5 Hz/µA (see Figure 3.13).

3.7 Lumi Monitors

FIG. 3.14: Plot of Lumi asymmetry as a function of run number for the 2003 G 0 com-
missioning run. Due to the extreme forward angle of the Lumi monitors, measured
asymmetries may be due to beam boiling effects on helicity correlated properties of the
beam. The red circles (blue triangles )indicate when the insertable halfwave plate was in
the ’out’ (’in’) position.

Downstream of the G0 target is a detector package that measures the luminos-

ity. The luminosity is the product of the beam current and target density. This

detector package consists of a pair of bare PMTs, a pair of PMTs with Lucite

(Cerenkov light is produced in the Lucite and the signal is picked up by the PMT),

and a prototype water Cerenkov detector from Mainz experiment PVA4. After

dividing out the beam charge, target density fluctuations can be monitored along

with helicity-correlated properties of the electron beam. The luminosity monitors

are at extreme forward angles and thus the parity-violating asymmetry should be
 68
near zero. If an asymmetry is measured by the detectors, this implies that a false

asymmetry is being generated, presumably due to helicity-correlated beam param-

eter differences or electronics artifacts. Results from the LUMI monitors are shown

in Figure 3.14 indicating that the target was neither boiling nor were there large

effects from helicity-correlated beam properties.

3.8 Moller Polarimeter

FIG. 3.15: The Hall C Polarimeter. The top schematic shows the collimators and
quadrupole magnets used to select scattered electrons of interest. The bottom schematic
shows the scattered electrons focused onto the polarimeter detectors.

The beam polarization was measured by the Hall C Moller (polarized ~e + ~e →

e + e ) polarimeter [59] which is located in the beam alcove upstream of Hall C. Ob-
 69
serving differences in the scattering rates, depending on whether the beam and target

electrons are polarized parallel or anti-parallel with one another, provides a mea-

surement of the beam polarization. Since this is a QED process, the cross sections
4
has been calculated precisely (up to αQED ). The cross section for a longitudinally-

polarized electron beam striking a polarized target electron is:

dσ dσ0 h k k
i
= 1 + Pt Pb AZZ (θ) (3.6)
dΩ dΩ
dσ0
where the unpolarized cross section dΩ
is given by
 2
dσ0 α(4 − sin2 θ)
= (3.7)
dΩ 2me γ sin2 θ

and in the high energy limit, the analyzing power is given by

− sin2 θ(8 − sin2 θ)

AZZ (θ) = . (3.8)
(4 − sin2 θ)2

One can measure the beam polarization by comparing the cross section asymmetry

for the beam and target spins aligned parallel and anti-parallel:

dσ k dσ ⊥
−
 = dΩk dΩ = A (θ)P k P k .
ZZ t b (3.9)
dσ dσ ⊥
+
dΩ dΩ
k
Knowing the target polarization Pt allows one to isolate the beam polarization.

The target electrons are provided by atomic electrons associated with the iron

atoms in the target. Typically, about 2 electrons of the iron’s 26 electrons are po-

larized, leading to a target polarization ∼ 8%. The target is a thin foil of iron

oriented perpendicular to the electron beam which is magnetized by a supercon-

ducting solenoid producing a 4 T field. The scattered electron and recoiled target

electron that emerge in the horizontal plane between 1.83 ◦ and 0.75◦ in the lab frame

are focused by a quadrupole magnet Q1. The desired scattering angles are set by

collimators. The electrons are then defocused using another quadrupole magnet,
 70
Q2, and the electrons detected in coincidence using two symmetrically placed ho-

doscope counters and lead glass counters. This system of movable collimators and a

pair of quadrupoles allow this device to be tuned to operate at any beam momentum

between 0.9 and 6.0 GeV/c.

The need to correct the elastic asymmetry for the polarization can be demon-

strated as follows from considering the scattering cross sections. One of these com-

ponents is a parity-conserving electromagnetic part (σ EM ) that is equal for both

helicity states. Then there is a parity-violating part, σ P V , which is caused by the in-

terference between the weak and electromagnetic amplitudes. This parity-violating

part has opposite signs depending upon the helicity, thus the contribution of (σ P V )

will scale with the beam polarization. The right-handed component of the cross

section can be written as

σ+ = σEM + Pb σP V (3.10)

while the left handed component can be written as

σ− = σEM − Pb σP V . (3.11)

Assuming that σP V << σEM , the asymmetry equation can be written as

(σEM + Pb σP V ) − (σEM − Pb σP V ) σP V
Ael = ≈ Pb = Pb Aphysics. (3.12)
(σEM + Pb σP V ) − (σEM + Pb σP V ) σEM

3.9 Data Acquisition

The data acquisition system (DAQ) used by the G0 experiment is CODA (CE-

BAF Online Data Acquisition system) [60]. CODA was developed by Jefferson Lab.

The G0 DAQ runs on a 1 GHz Pentium III computer running Linux (kernel 2.4.18).

The DAQ reads out the time-encoding scaler data (see Section 3.10) at a rate of

30 Hz during the 500 µs window for the Pockels cell to settle down after flipping
 71
helicity. The time encoding data is read from the North American Scalers and from

the French DSP concentrator on the DMCH-16X boards (see Section 3.10). Dur-

ing an MPS ∼ 50 kB of data is collected. Besides the 30 Hz time-encoding data,

other types of data events are interleaved in the data stream. FASTBUS ADC and

TDC data (see Section 3.10) were collected at a rate of 1/30 Hz. These FASTBUS

data are useful for monitoring and calibrating the detector system. In addition,

“slow control” EPICS [61] events are taken at 30 Hz; these events record data from

the beam position monitors, beam current monitors, temperature and pressure of

the target, temperature and current in the SMS, etc. The DAQ computer running

CODA communicates with each of the electronic crates via a single board computer

on each crate called a ROC (Read Out Controller). There are six ROCs: ROC1,

ROC2, and ROC3 all contain North American scalers, ROC3 contains the French

DMCH-16X modules, ROC5 contains the FASTBUS monitoring electronics ADCs

and TDCs, and the last ROC known as TS0 is the trigger supervisor that also reads

out beam and slow control electronics. The CODA datafiles are copied to the tape

silo system and to a group of three computers where G0Analysis, the replay engine,

produces ntuples and histograms to be read by ROOT and fills a MySQL database.

3.10 The G0 Electronics

The signals from the G0 detectors in Hall C are routed upstairs to be processed

by the G0 electronics. The signals coming from the North American (French) detec-

tors are handled by a North American (French) subset of the electronics. Having two

different sub-systems allows for a powerful cross check between the North American

and French data.

Both North American and French electronics can be described as falling into two

classes: monitoring/cross calibration FASTBUS electronics and the Time Encoding

 72
Electronics (TEE). The data obtained from the TEE is in the form of Time-of-Flight

(ToF) histograms accumulated over a macropulse. The other set of electronics are

used to monitor detector efficiencies, to calibrate the gains of the detectors, etc.

Common to both sets of electronics is the need to minimize both dead time and

helicity-correlated systematics.

As mentioned above, the North American (French) electronic sub-systems re-

ceives 256 input signals from the North American (French) octants 1,3,5,7 (2,4,6,8)

and the implementation of the electronics follow two different philosophies with the

North American electronics being highly modular and the French electronics being

highly compact.

3.10.1 North American electronics

A schematic of the NA electronics chain is shown in Figure 3.16. The follow-

ing sections describe in further detail the signal’s journey from the PMT signal

to accumulated data. Signals from the PMTs first go to a patch panel in Hall C

through 36 meters of RG58 cables. They are then sent up to the electronic counting

room in 107 meter long RG8 cables for reduced attenuation. Due to the high rate,

event-by-event data collection is excluded in the Time Encoding Electronics (though

the monitoring electronics record a sample of event-by-event data). The hardware

must then be chosen to reduce the time resolution of the signals arriving at the

Time Encoding Electronics. The nominal 1 ns wide bins are determined by a clock

signal. Commercial constant fraction discriminators minimize the walk in time of

logic pulse, and meantimers are employed to average the pulse times of the PMTs

at opposite ends of the scintillator.

.
 73

FIG. 3.16: Schematic of the North American electronics for the G 0 forward running
mode. Custom built electronics are denoted by dark boxes.

Splitter

There are 16 custom built splitter modules. Each splitter module asymmet-

rically (70:30) and passively splits 16 PMT signals to provide 16 inputs to the

Constant Fraction Discriminator and 16 inputs to the FASTBUS ADCs.

Constant Fraction Discriminator

The high counting rates on the detectors implies that the signals must be of

low amplitude in order to reduce the instantaneous and integrated currents on the

PMTs. With these low amplitude signals, a good precision on the elastic proton

time information is required and time-walk could be a problem, hence the use of
 74
the constant fraction discriminators. There are 16 Lecroy 3420 Constant Fraction

Discriminators (CFDs). These CFDs take the PMT analog signal from the splitter

and produce a logic signal if the signal is above some threshold, which was typically

about 35 mV. This is done by generating the logic signal at some constant fraction

of the peak height to produce a nearly walk-free signal. The output signals from

the CFDs are then sent to the FASTBUS TDCs and to the TEE meantimers.

Meantimer

There are 8 custom built modules with 16 meantimer channels in each module.

Each meantimer (MT) has inputs from two CFDs. One input is for each PMT

signal from each end of a detector. The output signal is the mean time of the two

PMT input signals. The mean timing of the input signals is performed by custom

Application Specific Integrated Circuits.

This is done because the proton can hit anywhere along the scintillator. This

will produce variations in the signals on an event-by-event basis. In order to produce

a Time-of-Flight histogram that is constant, the meantime of the signals is produced.

This can be seen by considering when a detector is hit by a particle. Let t 1

and t2 be the time of the signal from each PMT, d1 and d2 the distance from where

the particle hits along the scintillator to the PMT, and d = d 1 + d2 . By taking the

average of the propagation times:

t1 + t2 d1 + d2 d
= =
2 2 2c

where the result should be a constant.

In total there are 128 meantimed signals that are sent to the FASTBUS TDCs

and the TEE Latching Time Digitizer.

 75
Latching Time Digitizers

The purpose of the Latching Time Digitizers (LTD) is to accumulate time

spectra for data rates of several MHz. There are 32 custom built LTD modules.

The LTDs have a cycle time of 32 ns. Twelve clock pulses are used to clock a shift

register. The LTDs take 2 signals from the front and back meantimers of a detector

in coincidence and latches for a single beam burst. The time since the beam sync

signal of the latched input is determined by using two 16 bit shift registers which

are clocked 180◦ out of phase with respect to one another. The status of the shift

registers are latched into another set of registers at the end of the beam pulse. These

registers then drive 24 VME scaler channels (see below) which count how many times

the bits were set. The depth of penetration of the input signal into the shift register

during the shifting sequence then depends upon the time of the coincidence within

the 32 ns cycle. This depth of penetration of the signal thus encodes the time of

the coincidence. The LTDs also monitor the quality of the clock train for errors of

“too many” or “too few” clock pulses. These gated clock trains for the LTDs are

produced and distributed by two other sets of electronics:

• Clocking Gating Board: The custom built Clocking Gating Board (KGB) takes

the 499 MHz clock signal from the CEBAF accelerator master oscillator and the

“YO” signal provided by a beam pickoff monitor upstream of the G 0 target. This

produces an ∼ 2 ns sync and the twelve ∼ 1 ns pulses of the clock train.

• Signal Duplication Boards: The custom built Signal Duplication Boards (SDBs)

take a single copy of the gated clock train and a copy of the sync signal and

provides 9 copies of each to be fanned out to the LTDs.

 76
Scalers

The scalers which capture the time spectra from the LTDs are custom built

VME latching scalers designed by LPSC-Grenoble. Data are accumulated for a

1/30 s macropulse. During the settle time of 500 µs between macropulses, the scaler

data is latched into on-board memory and the scaler channels are cleared. The G 0

DAQ system then reads out the latched data for a macropulse while the scalers are

accumulating data for the next macropulse.

3.10.2 French electronics

Splitter

There are 4 custom built splitter modules. Each splitter module symmetrically

(1:1) and actively splits 64 PMT signals to provide inputs to the Constant Fraction

Discriminator on the DMCH-16 X and FASTBUS ADCs.

DMCH-16X

The heart of the French time encoding electronics are 8 custom built boards

known as the DMCH-16X (Discriminator, Mean-timer, time digital Converter, His-

togramming, 16 channels within the vXi standard). Each board receives 32 PMT

analog signals and builds 8 Time-of-Flight histograms associated with the detectors

of two quarters of two octants. Most of the board’s settings such as the CFD thresh-

olds and differential non-linearity (DNL) for the TDCs are controlled via software.

Each DMCH-16X board contains 3 daughter boards containing: 16 CFD-MTs,

1 S-DMCH, and 1 G-DMCH. The G-DMCH is a generator used occasionally to

check CFD thresholds and MT outputs. The S-DMCH keeps track of individual

counts for the CFDs and MTs. Each DMCH-16X motherboard holds 2 TDCs which

distribute data through FiFo buffers to 4 front-end digital signal processors (DSP)
 77
where the ToF spectra are accumulated in different memory registers associated

with each detector. At the end of an MPS, the 4 front-end DSPs and the DSP on

the scaler S-DMCH send their data to another DSP known as the DSP concentrator,

which transfers the block of data associated with one DMCH-16X board over the

VME bus to the CPU board.

CHAPTER 4

Data Analysis

The first G0 engineering run took place between October 2002-January 2003.

The data analysis from this run was performed in several steps. First the raw

measured asymmetries must be formed from the detector yields. Then the dead time

must be corrected. False asymmetries due to helicity-correlated beam properties

must be calculated and corrected. There is a correction for the background dilution

factor and the background has an asymmetry associated with it that must be taken

into account. Then the physics asymmetry can be extracted after correcting for the

beam polarization, and radiative corrections can be applied.

4.1 Measured Asymmetry

The detector signals are accumulated during a helicity state (MPS). When the

helicity is reversed by the Pockels cell that flips the helicity at 30 Hz, the data

is read out by the electronics. The detectors count the desired elastically scattered

protons as well as particles from other processes such as inelastic scattering from the

aluminum windows of the target [62] and production of pions in the hydrogen target.

An example of a time-of-flight spectrum is shown in Figure 4.1. These histograms

78
 79
count how many particles have been detected in a detector. These histograms

contain either 24 (NA electronics) or 128 (French electronics) time bins. Other types

of data are also recorded during this spin-flip period, such as the integrated charge

by the beam current monitors and integrated positions from the beam position

monitors. For these ToF histograms the yield (Y ) is the number of events (N)

measured in each time bin normalized to the beam charge (Q) accumulated during

that MPS. This charge normalization is needed since the event rate is a function of

the beam intensity.

×10
3

8000
Elastic
7000 Protons
6000

5000
evts

4000 Inelastic
Pions Protons Elastic Cut
3000

2000

1000

0
0 20 40 60 80 100 120
ToF(1/4ns)

FIG. 4.1: An example of a typical French Time-of-Flight spectrum from detector 8 with
data taken during a one hour run..

From the quartet, an asymmetry can then be computed for events within the

elastic cut window, which is about 4 ns wide. This asymmetry for each detector
 80
measured over a quartet can be written as:
 +   − 
N1 N2+ N N2−
+ +
Q1 Q2+ − Q1− + Q−
Ameas,det,qrt =  +   1− 2
 (4.1)
N1 N2+ N N2−
Q+
+ Q+ + Q1− + Q−
1 2 1 2

where Nis is the number of counts recorded for the ith MPS of this quartet with beam

helicity of sign s, and Qsi is the beam charge incident on the target during that MPS.

Since the yield is the number of counts normalized to the charge, Equation 4.1 can

be written as:

Y1+ + Y2+ − Y1− − Y2−

Ameas,det,qrt = . (4.2)
Y1+ + Y2+ + Y1− + Y2−

The measured asymmetry from all quartets within a run are averaged in each de-

tector to give Ameas,det,run . The precision σmeas,det,run on the measurement is the

root mean square of the standard deviation of the quartet asymmetry distribution

divided by the square root of the number of quartets.

The next step is to take the weighted average and error of the measured asym-

metries for all the runs for each detector:

P Ameas,det,run
2run
σmeas,det,run
Ameas,det = P (4.3)
1
run 2
σmeas,det,run
1
σmeas,det = P (4.4)
1
run 2
σmeas,det,run

However, this is not the full story. Within the elastic proton peak cut window,

there is some contamination, since not all the events are from elastic protons. Some

of these events are inelastic protons produced in scattering within the target. This

contamination must be removed in order to evaluate A physics for the elastic protons.

The measured asymmetry for each detector must be corrected to subtract the

inelastic contribution (note the quartet helicity ordering denoted at “1” and “2”
 81
above will now be suppressed for readability):

Yel+ − Yel− + Yinel

+ −
− Yinel
Ameas,det = (4.5)
Yel + Yinel
Yel Yinel
= (Ael ) + (Ainel ) (4.6)
Yel + Yinel Yel + Yinel
+(−)
where Ainel(el) is the inelastic(elastic) asymmetry and Yel(inel) is the number of

right(left) handed elastic (inelastic) events. One can define the so-called ‘inelastic
Yinel
dilution factor’ as d = and solve for the elastic asymmetry for each detector:
Yel

Ael = (1 + d)Ameas − dAinel . (4.7)

The error on the elastic asymmetry is given by:

q
∆Ael = (Ameas − Ainel )2 ∆d2 + d2 ∆A2inel + (1 + d)2 ∆A2meas . (4.8)

After correcting for the inelastic contributions, other corrections must be applied to

the elastic asymmetry to obtain the physics asymmetry, Aphy . The physics asym-

metry must be corrected for not having 100% beam polarization, radiative effects,

and false asymmetries due to helicity-correlated beam systematic effects.

The error on the elastic asymmetry has several contributions, as can be seen

in Equation 4.8. For the G0 forward angle production run, an error on the elastic

asymmetry of ≤ 0.5 ppm is desired.

4.2 Statistical Widths

The statistical distribution of quartet asymmetries have a minimum variance

width, σstat (Aqrt ), determined solely from counting statistics, given by

1
σstat = p (4.9)
Nproton
where Nproton is the number of detected protons in one quartet. Thus if the measure-

ment is purely statistical, the width of the distribution of the measured asymmetries,
 82

σmeas/ σstat σmeas/ σstat

Entries 9544
Mean 1.06
1600 RMS 0.02136

1400

1200

1000
Number

800

600

400

200

0
0.9 0.95 1 1.05 1.1 1.15 1.2 1.25
σmeas/ σstat

FIG. 4.2: The ratio of the measured quartet asymmetry distribution widths to the width
σmeas
expected solely from counting statistics for the detectors . Deviations from 1.0
σstat
indicate the presence of instrumental noise in the measurements.

σmeas , is:

1
σmeas = p + (4.10)
N1 + N2+ + N1− + N2−
for each helicity period.

A comparison between the measured quartet asymmetry distribution widths

with the statistical width is an indicator of systematic effects, since instrumental

noise can broaden the width of the measured distribution. The number of counts

used in this calculation has been corrected for dead time (see Section 4.6). The

measured statistical widths due to the number of counts for each detector is on the

order of 3800 ppm. This makes sense, since the detector rate is about 0.5 MHz, the

measurement interval is 1/30 second, so the number of scattered protons/macropulse

0.5 MHz × 1/30 second = 16667. (4.11)

Obtaining the number of events in a quartet requires the result from Equation 4.11 to

be multiplied by 4 for four helicity windows in one quartet. Using Equation 4.2 yields
 83
σmeas
Ameas,qrt,det ∼ 3800 ppm. All detectors were measured to have ∼ 1.06 (see
σstat
Figure 4.2) implying that all measurements are dominated by counting statistics.

4.3 Cuts

The measured raw asymmetry had several cuts applied event by event in or-

der to improve the statistical properties of the results. These were applied by the

G0Analysis replay engine which produces the histograms and NTuples, to be read by

ROOT, from the raw CODA data files. The first check is to be sure that an event in

the MPS make up a good quartet. The code will check the reported helicity (which

has been delayed by 8 helicity windows) against an algorithm for helicity prediction

in G0Analysis. These cuts defined a good MPS. A good quartet requires that all

MPSs in a quartet are good and have the correct helicity pattern. Asymmetries are

calculated for good quartets only. After this cut, two more sets of cuts were applied:

- Beam Cuts:

If the beam current read out by the beam current monitor (see Section 3.5.6)

is less than 5µA, that MPS along with the next 2000 MPS after beam recovery

are removed. This is because it has been noted, by looking at the luminosity

monitors, that the target takes approximately a minute to settle down after

beam is restored.

- Detector Cuts

Both North American and French electronics have error indicators in the data

stream. The North American electronics (see Section 3.10.1) have error indica-

tors on the LTD boards to note if the LTD is seeing “too many” or “too few”

micropulses within the 30 Hz helicity window. The French electronics (see Sec-

tion 3.10.2) are armed with “alerts” which count the number of overflow words in
 84
a DMCH module. These counters are looked over first, then data from a detector

is tagged as “bad” when the associated counter reads a non-zero value.

Sometimes a detector yield would “jump” up or down in a non-statistical manner

which thereby creates a large systematic asymmetry. To identify this, G0Analysis

takes the first 100 MPS of each run for each detector and calculates the mean

and width of the yield distribution, for each detector and for each ToF bin, as a

reference value. A ±10σ cut around the reference value is then applied to each

ToF bin and a detector is tagged as “bad” if any ToF bin fails the cut.

4.4 Raw Asymmetry

The raw asymmetry is calculated for each quartet for each detector from the

normalized yield

Y1+ + Y2+ − Y1− − Y2−

Aqrt,det = + . (4.12)
Y1 + Y2+ + Y1− + Y2−

The normalized yield is the yield for one MPS divided by the beam charge accumu-

lated during that MPS

(dσ/dΩ)L∆ΩTM P S
Y+MorP −
S=1 or 2
= (4.13)
QM P S
= NM P S /QM P S (4.14)

where dσ/dΩ is the differential cross section for →

−e p scattering, L is the luminosity,

∆Ω is the solid angle acceptance, NM P S is the number of counts in a detector during

a macropulse, QM P S is the charge accumulated during a macropulse and T M P S is

the length of time for one macropulse. Figure 4.3 is a plot of the measured raw

asymmetries as a function of detector number.

 85

FIG. 4.3: North American and French (out-in) raw asymmetries. The insertable halfwave
plate reverse the sign of the raw asymmetry. By subtracting the ’in’ asymmetries from
the ’out’ asymmetries, one can combine the two sets of results. The error bar is found
by adding the ’in’ and ’out’ errors in quadrature

4.5 Passive Sign Reversal

In order to establish the validity of the measured raw asymmetry in a parity

violation experiment it is important to demonstrate that the results are being pro-

duced by helicity-dependent dynamics in the cross section instead of by spurious

electronic effects. Passively reversing the sign of the measured raw asymmetry is a

powerful way of searching for potentially unforeseen systematic effects.

By periodically (every two or three days) inserting a halfwave plate at the

source, the helicity of the circularly polarized light produced by the G 0 TIGER laser

(see Section 3.5.1) is passively reversed. This in turn reverses the electron helicity

pattern without changing any other device systematics. If the analysis software does

not take into account the insertion of the halfwave plate, the opposite sign for the
 86
asymmetry will be calculated.

NA in+out χ2 / ndf 15.75 / 14

15 Prob 0.3287
Asymmetry (ppm)

p0 0.3339 ± 1.027

10

-5

-10

-15
0 2 4 6 8 10 12 14 16
Ring

FR in+out χ2 / ndf 9.118 / 14

15 Prob 0.8234
Asymmetry (ppm)

p0 0.6342 ± 0.7087

10

-5

-10

-15
0 2 4 6 8 10 12 14 16
Ring

FIG. 4.4: North American (NA) and French (FR) (in+out) raw asymmetries. Under
the raw asymmetry sign reversal, due to the insertable halfwave plate, adding the ‘in’ and
‘out’ states should yield a zero result.

It is clear from Figure 4.3 and Table 4.1 that there is a correlation between the

presence of the halfwave plate and the sign of the asymmetry. From Figure 4.4, it

can be seen that adding the ‘in’ and ‘out’ data results in zero, as expected. When

the halfwave plate is inserted, the sign of the asymmetry is corrected when combined

with the data set.

 87
Det NA OUT NA IN NA OUT-IN Fr OUT Fr IN Fr OUT-IN
(ppm) (ppm) (ppm) (ppm) (ppm) (ppm)
1 -1.7 ± 2.3 6.1 ± 2.6 -3.63 ± 1.71 -2.1 ± 2.0 3.8 ± 1.9 -3.00 ± 1.37
2 -3.2 ± 3.0 1.8 ± 3.0 -2.48 ± 2.14 -4.5 ± 1.8 3.1 ± 1.7 -3.78 ± 1.24
3 -2.6 ± 2.2 3.7 ± 2.3 -3.11 ± 1.60 -4.3 ± 1.8 2.9 ± 1.7 -3.60 ± 1.24
4 0.6 ± 2.3 6.9 ± 2.5 -2.89 ± 1.71 -3.6 ± 1.9 6.0 ± 1.9 -4.80 ± 1.32
5 -6.2 ± 3.1 6.3 ± 3.1 -6.26 ± 2.17 -3.5 ± 1.8 1.1 ± 1.8 -2.31 ± 1.29
6 -6.9 ± 3.0 -2.2 ± 3.1 -2.57 ± 2.14 -3.4 ± 1.8 1.1 ± 1.8 -2.22 ± 1.27
7 -6.3 ± 2.3 7.2 ± 2.5 -6.70 ± 1.68 -5.9 ± 1.8 6.2 ± 1.8 -6.04 ± 1.26
8 -1.9 ± 3.1 5.7 ± 2.9 -3.92 ± 2.12 -5.1 ± 2.0 6.7 ± 1.9 -5.90 ± 1.37
9 -5.5 ± 2.6 5.5 ± 2.6 -5.49 ± 1.81 -4.5 ± 1.9 9.0 ± 1.9 -6.80 ± 1.33
10 -9.6 ± 3.0 6.8 ± 3.0 -8.17 ± 2.12 -8.4 ± 2.0 9.1 ± 1.9 -8.76 ± 1.37
11 -5.8 ± 2.3 5.7 ± 2.3 -5.74 ± 1.65 -5.5 ± 2.0 7.7 ± 1.9 -6.65 ± 1.38
12 -6.7 ± 2.8 8.4 ± 3.0 -7.45 ± 2.04 -6.6 ± 2.2 4.9 ± 2.2 -5.75 ± 1.53
13 -9.4 ± 4.3 -0.3 ± 4.4 -4.57 ± 3.08 -4.5 ± 2.4 7.4 ± 2.4 -5.95 ± 1.68
14 0.4 ± 4.4 -2.6 ± 4.8 1.39 ± 3.24 -3.9 ± 2.5 5.4 ± 2.4 -4.65 ± 1.74
TABLE 4.1: Comparison of North American and French raw asymmetries under reversal
of the halfwave plate. When the halfwave plate is inserted into the laser path, the hand-
edness of the light reverses thus causing the handedness of the liberated electrons to be
opposite to that of the Pockels cell. If there aren’t any systematic effects, the asymmetry
magnitudes should be the same but with opposite sign.

4.6 Dead Time Corrections

The asymmetry had to be corrected to take into account electronic dead time.

The electronic dead time is caused when the electronics are still busy after/during

the processing of an event. The G0 custom electronics have been designed to measure

high rates on the order of 1 MHz with a controlled dead time. Dead time from the

North American electronics can come from the LTDs and the CFDs. Both North

American and French electronics are set to neutralize the next pulse after an event

(known as Next-Pulse-Neutralization or NPN) and allows for an exact calculation

of dead time. The probability of being dead, f , is proportional to the rates of events

triggering the electronics. This will cause the measured yield Y meas , to differ from

the true yield, Ytrue as

Ymeas = (1 − f )Ytrue (4.15)

 88
Controlling the dead time is mandatory since the number of events are expected to

vary in a helicity-correlated manner arising from the charge asymmetry A q . Thus

the dead time can introduce a false asymmetry

Af alse = −f × Aq (4.16)

The North American LTDs and French time encoding electronics record only a four-

fold coincidence and therefore cannot be used to correct for single events, where

singles occur when one to three (but not all four) PMTs fire. The singles are mainly

pions that come before the protons in the time spectrum, and so a singles hit will

then cause the loss of the proton count. This means that on average the electronics

are dead to protons for one micropulse.

In the North American electronics, the dead time was dominated by the CFDs.

The CFDs should work by taking the original signal and producing two more signals.

One signal is a duplicate of the original but delayed in time and the other signal

is a fraction of the original signal amplitude. The CFD then will fire when the

original signal is above a certain threshold and the two duplicate signals intersect.

What happened during the 2002-2003 engineering run was that sometimes a smaller

detector signal would precede a larger detector signal, and this would cause the

duplicate signals to intersect before the original signal would cross threshold. This

produces an apparent dead time in the electronics. This effective dead time was

about 70 ns for the North American electronics. This problem will be corrected in

future running by adjusting the delay and fraction in the CFDs to prevent these

sub-threshold particle signals from affecting the larger particle detector signals.

The dead time correction is

Nraw
Ntrue = (4.17)
1−f
where Ntrue is the true yield, Nraw is the raw yield, and f is the dead time fraction.

The overall dead time fraction for the NA detectors was found to range from 3%
 89
to 11% for detectors 1 through 14. The technique for extracting the dead time

fraction is to plot the detector yield versus beam current or detector asymmetry

versus charge asymmetry. The overall dead time fraction is the sum of the dead

time due to singles and the dead time due to coincidence

f = Rcoinc τcoinc + Rsingles τsingles (4.18)

where Rcoinc (Rsingles ) is the rate due to the coincidence(singles), and τ coinc (τsingles )

is the dead time due to coincidence(singles). The dead time due to coincidences is

given by

τcoinc = 0.5 + (j − 1) + 6.5 + 32 (4.19)

where j is the timebin of the detected particle. The dead time due to singles was

found to be 70 ns (as explained above) and this dominates the dead time, thus the

dead time fraction can be written as

f = (Rcoinc + Rsingles ) × 70 ns (4.20)

where Rcoinc is determined from the LTDs and Rsingles is determined from FASTBUS

data.

The French electronics singles dead time from the CFDs was found to be 35 ns.

The dead time due to the meantimers was found to be 32 ns.

4.7 Correcting Helicity-Correlated Beam

Properties

Helicity-correlated fluctuations in the beam parameters such as position, angle,

energy and charge can cause false asymmetries to appear in the data. This can be

seen if one of the beam parameters mentioned above is on average, different between
 90
the two different helicity states. If this is the case, then the measured yield will be

different between the two different helicity states, thus producing a false asymme-

try. These fluctuations are the results of reversing the voltage on the helicity Pockels

cell. Besides producing circularly polarized light, this voltage reversing might pro-

duce some helicity-correlated systematic, e.g. the angle of the emerging laser light

might be different in one state versus another. This systematic then translates to

slightly different types of helicity-correlated laser light hitting the GaAs crystal dif-

ferently in a helicity-correlated manner which in turn produces an asymmetry in

the electron beam between helicity states (see Figure 4.5). These helicity-correlated

fluctuations in beam parameters manifest themselves as false asymmetries that ap-

pear in the data. This false asymmetry requires a systematic correction to the

measured asymmetry.

This can be illustrated assuming a linear relationship between the measured

yield Y and the beam parameter xi , where the beam parameter can be any of the

six mentioned above. The yield due to beam fluctuations can be written as

Y = αxi . (4.21)

+,−
The measured yield, Ymeas , is then a combination of the parity-violating yield, Y +,− ,

and the yield due to the correlation with the beam parameter

Y +,− = Y +,− + αxi+,− . (4.22)

The beam parameter difference can be written as

−
δxi = x+
i − xi (4.23)

where the superscript +(-) indicates positive (negative) helicity state.

Forming the asymmetry

Y+−Y−
Ameas = (4.24)
Y++Y−
 91
substituting in the yield from Equation 4.22, one obtains
−
Y + − Y − + α(x+
i − xi )
Ameas = + − . (4.25)
Y + Y − + α(x+ i + xi )

Assuming Y ± >> αx±

i

Y + − Y − + αδx
Ameas =
Y++Y−
Y −Y−
+
δx
= −
+α +
+
Y +Y Y +Y−
δx
= Acorr + α
2<Y >
(4.26)

where the average yield is given by < Y >= 21 (Y + + Y − ), and α is identified as the

detector response to the beam parameter. This can finally be written as

1 X ∂Y
Ameas = Acorr + δxi . (4.27)
2 < Y > i=1,6 ∂xi

∂Y
The detector yield response to the various beam parameters, , must be
∂xi
extracted from the data. This can be done by looking at the natural motion of the

beam by assuming a linear response of the yields to the beam parameters

X ∂Y
Ycorr = Ymeas − δxi (4.28)
i=1,6
∂xi

where Ymeas is the raw yield of the detector, xi is one of the six beam parameters (x

and y position at target, x and y angle at the target, charge at the target and the

energy). The helicity-correlated beam position and angle are calculated using two

BPMs closest to the G0 target. The energy difference is measured using a BPM in

a dispersive region in the arc. The beam charge is measured from the standard Hall

C BCMs.
∂Ymeas
The slopes, , are found by inverting the covariance matrix in the follow-
∂xi
ing equation that relates the slopes to the mean correlation between the yield and
 92
the various beam parameters

X  
∂Y
(< δY · δxi >) = (< δxi · δxj >) (4.29)
i=1,6
∂xi

where (δxi · δxj) is a 6 × 6 beam parameter covariance matrix, and (δY · δx i ) is

the covariance of the yield and beam parameters. This variation in the yield to

the beam parameters characterizes the sensitivity of the G 0 apparatus to changes in

beam properties. Figure 4.6 shows the percent yield on detector 1 for all the octants

as the beam properties are being changed. In general, diametrically opposing octants

should have opposing sensitivities and thus opposing signs. It is due to this behavior

that the sensitivities due to changes in beam position and angle are largely canceled

out between diametrically opposing octants when one takes the mean.
 93

FIG. 4.5: Plots of the helicity-correlated beam properties. The position differences and
angle differences are the values projected onto the target from the BPMs G0 and G0B.
The energy difference is measured from BPM 3C12.
 94

FIG. 4.6: Detector sensitivity slopes plotted as a function of octant.

 95
Beam Parameter x position y position x angle y angle Energy charge
Octant Average 0.04 0.17 -1.73 -1.08 -0.08 -0.001
slopes %/mm %/mm %/mrad %/mrad %/MeV %/nC
helicity-correlated -3.24 -0.9 0.43 1.02 1.15 -0.001
difference nm nm nrad nrad eV nC
False Asymmetry -0.015 -0.015 -0.074 -0.11 -0.01 0.013
correction ppm ppm ppm ppm ppm ppm
TABLE 4.2: IHWP out false asymmetries. The data are from 49 asymmetry runs.

Beam Parameter x position y position x angle y angle Energy charge

Octant Averaged 0.06 0.090 0.92 0.30 0.01 -0.001
slopes %/mm %/mm %/mrad %/mm %/MeV %/nC
helicity-correlated 2.21 -0.91 -0.69 0.49 -1.19 -0.001
differences nm nm nrad nrad eV nC
False Asymmetry 0.014 -0.008 -0.063 0.015 -0.001 0.008
correction ppm ppm ppm ppm ppm ppm
TABLE 4.3: IHWP in false asymmetries. The data are from 58 asymmetry runs.

The resulting octant averaged slopes as well as the associated asymmetries are

shown in Table 4.2 for the insertable half wave plate out state and in Table 4.4

for the insertable halfwave plate in state. The false asymmetry A f alse due to the

helicity-correlated beam parameters as a function of insertable halfwave plate is

found to be:

IHWP OUT: Af alse = −0.210 ± 0.616 ppm (4.30)

IHWP IN: Af alse = −0.037 ± 0.522 ppm (4.31)

IHWP IN+OUT Af alse = −0.11 ± 0.40 ppm. (4.32)

The measured asymmetry is then corrected for helicity-correlated false asymmetry

by

Acorr = Ameas − Af alse (4.33)

 96
where Acorr is the corrected measured asymmetry.

Det NA Acorr Fr Acorr

1 -3.74 ± 1.76 -3.11 ± 1.43
2 -2.59 ± 2.18 -3.89 ± 1.30
3 -3.22 ± 1.65 -3.71 ± 1.30
4 -3.00 ± 1.75 -4.91 ± 1.38
5 -6.37 ± 2.20 -2.42 ± 1.35
6 -2.68 ± 2.18 -2.33 ± 1.33
7 -6.81 ± 1.73 -6.15 ± 1.32
8 -4.03 ± 2.16 -6.01 ± 1.43
9 -5.60 ± 1.85 -6.91 ± 1.39
10 -8.28 ± 2.16 -8.87 ± 1.43
11 -5.85 ± 1.70 -6.75 ± 1.43
12 -7.56 ± 2.08 -5.85 ± 1.58
13 -4.68 ± 3.10 -6.06 ± 1.73
14 1.28 ± 3.27 -4.76 ± 1.79
TABLE 4.4: The measured asymmetries corrected for false asymmetries due to helicity-
correlated beam properties for detectors 1-14.

4.8 Inelastic Dilution Factors

4.8.1 French Results

The French time-of-flight (ToF) spectra with its fine 128 binning (versus the

North American ToF with its more coarse and non-equal 24 binning) is the first

data set to be looked at for extracting the inelastic dilution factors for each detector.

Each spectrum was fit with one Gaussian for each of the three particle distributions:

pion, inelastic proton and elastic proton (see Figure 4.7). Numerical integration was

employed to find the contributions due to the elastic and inelastic protons within

the 4 ns wide elastic cut window. A total of 106 runs were fit. The inelastic dilution

factors of each detector were stable to 3% over the set of runs assuming the cut does

not change (see Figure 4.8).

 97

Elastic
Protons

Events
Elastic
Inelastic Cut
Pions
Protons

Time (.25ns/bin)

FIG. 4.7: An example of a 3 Gaussian fit to French Detector 9 Octant 2. Each par-
ticle distribution has been fit to a Gaussian. The vertical lines around the elastic peak
constitute the elastic cut for this spectrum.
Inelastic Dilution Factor

Cut Changed

Run Number

FIG. 4.8: An example of the stability of the inelastic dilution factor over 106 runs for
detector 14 Octant 2. The abrupt change near run 60 is due to the elastic cuts being
changed.

The contributions to the percent errors on the extracted inelastic dilution fac-

tors (which are added in quadrature to produce the final percent error) are :

• Error on the fit:

– Statistical precision of the fit: the dispersion between the inelastic dilution

factors for the same detector and same octant, with the same cuts for different

runs. This was found to be 0.3%.

• Error on the shape of the fit: the dispersion between using different plausible

mathematical equations to describe the same spectra. This was found to be 1%.
 98
• Error on the average over the octants: the dispersion between the dilution factors

for different octants. Having different cuts for the same detector but for a different

octant or having different signal amplitude compared to the thresholds might

cause this dispersion. This was found to be 4%.

Since there was a 5% error between dilution factors for each detector between

the octants, the inelastic dilution factors for each detector was averaged over the 4

French octants (see Figure 4.9 and Table 4.5).

FIG. 4.9: French inelastic dilution factors as a function of detector number, averaged
over 4 French octants, using the 3 Gaussian fit.

4.8.2 North American Results

The North American (NA) time-of-flight (ToF) spectra were plagued by differ-

ential non-linearity (DNL) (see Figure 4.10). The NA ToF spectra was expected to

have time bins that were 1 ns wide (except for the first bin which was 6.5 ns wide

by design). The DNL manifested itself as deviations from the 1 ns width. The DNL
 99
Det Inelastic Dilution Factor Error
1 0.148 0.007
2 0.171 0.009
3 0.180 0.009
4 0.160 0.008
5 0.194 0.010
6 0.202 0.010
7 0.216 0.011
8 0.200 0.010
9 0.210 0.010
10 0.222 0.011
11 0.259 0.013
12 0.267 0.013
13 0.307 0.015
14 0.401 0.020
TABLE 4.5: The average inelastic dilution factors for the French detectors for the cuts
on the elastic peak which were ∼ 4 ns. The error represents the final absolute error on
the dilution factor for each detector.

created time bins ranging from ∼ 0.5-2 ns. The DNL was significantly larger than

expected, and the cause(s) for this large DNL are only imperfectly understood at

the present, but appear to be at least partly due to the clock signal.

The North American electronics dictate the width of the bins of the North

American ToF spectra by forming the beginning and ending of the bins on the

rising and falling edge of a clock signal supplied by accelerator. If the clock train is

made up of perfect square waves, then the bins in the North American ToF spectra

will be 1 ns wide. If the clock train is asymmetric, then the bins will deviate from

1 ns, as observed during the 2002-2003 G0 commissioning run. This makes fitting

the North American ToF spectra to find the inelastic dilution factors more difficult

than for the French ToF.

The DNL can be measured with “white noise” runs. A white noise run con-

sists of an LED shining on a photo-multiplier tube whose signal is then fed into the

North American electronics. For bins of equal width, a flat ToF spectrum should

be produced for bins 2 though 24. Since the North American ToF does suffer from
 100

FIG. 4.10: An example of a time-of-flight spectra from a North American detector. The
jagged binning is due to the differential non-linearity that causes the bins not to be of
equal time widths.

the DNL problem, the bins are not of equal widths and by normalizing each of the

bins, the width of each bin can be found (see Figure 4.11). The qualitative pattern

of the DNL tends to alternate between ∼ 0.5 and ∼ 1.5 ns wide bins although there

are significant deviations from this pattern. Another difficulty with the DNL is that

it changes with time, on time scales of the order of a day, as can be seen in both

Figures 4.11 and 4.12.

To evaluate what effect the DNL had on extracting the inelastic dilution fac-

tors, a simulation of the ToF spectra without DNL for each detector was produced.

Using the fit parameters obtained from the French ToF for the elastic and inelastic

proton distributions, the ToF spectra for detectors 1-14 were simulated with very

fine binning (see Figure 4.13).

Three white noise spectra were taken during the G0 commissioning run and

three white noise runs were taken later after the commissioning run; with these six
 101

FIG. 4.11: An example of several white noise spectra for North American octant 3 detector
4. The ordinate axis is the width of each bin along the abscissa axis. Note that the bins
deviate from the expected 1 ns width. This deviation is the so-called ’differential non-
linearity’ or DNL. Also note that the DNL changes with time.

runs, a total of 212 measurements1 of the DNL exist. The simulated spectra were

then recast into North American binning and the 212 measured DNLs were applied

(see Figure 4.13).

These spectra were then fit to 2 Gaussians (one for the inelastic proton and one

for the elastic proton distributions) and numerical integration was applied to find

the inelastic dilution factors. When the results for each detector were histogrammed,

a bimodal distribution was found (see Figure 4.14). This bimodal distribution was

due to how the DNL was applied to the ToF spectra. The DNL creates a bias

weighting within the elastic cut (the first bin after the cut is typically either 0.5 ns

wide or 1.5 ns wide). The actual inelastic dilution factor 2 was distributed around

only one value (see Figure 4.14). This implied that the DNL must be corrected to

extract the inelastic dilution factors correctly.

1
[(16 × 4) detectors] × [6 measurements] = 256 DNL measurements of which 44 had to be
excluded because of detector malfunction.
2
Computed with the initial binning.
 102

Bin Width (ns)

1/16 1/21 1/24 4/16 4/17 4/18 4/21 4/22
Date

FIG. 4.12: An example of one bin’s variation over 8 measurements over 4 months for
detector 7 octant 1. Note that on April 17th the bin width changed by nearly a factor of
two from the day before and the day after.

Events

Time (ns)

FIG. 4.13: The plot on the left is a simulated North American ToF spectrum using
the fit parameters obtained from the French ToF. The plot on the right is a simulated
North American ToF spectra with DNL. Each simulated North American ToF spectra
was produced 212 times with different DNLs.

The DNL was corrected by using variable binning and normalizing each bin to

the variable bin width. Only the three sets of North American white noise runs taken

during the 2002-2003 G0 commissioning run were utilized to fix the North American

ToF spectra. An example of a corrected and uncorrected North American ToF

spectrum is shown in Figure 4.15.

Since the DNL changes with time, one set of white noise measurements will

not fix the whole North American data set. Figure 4.17 shows a ToF spectra for

the same detector with the same DNL correction but for different runs where one

run passes the subjective test (the ToF spectrum is “smoothed” out) and the other
 103

Number

Inelastic Dilution Factor

Number

Inelastic Dilution Factor

FIG. 4.14: The top histogram is an example of the simulated extracted inelastic dilution
factors for a North American detector. The bottom histogram are the results of the “true”
inelastic dilution factor from the simulated corrected North American ToF spectra.

fails (the ToF spectrum still appears “jagged”). To further complicate this, since

the DNL can change one day to the next day, one white noise spectra might fix the

DNL for many runs while failing for some of the intermediate runs within the run

set (see Figure 4.16). Looking at the relative minimum in relation to all other runs,

the sum of the differences between the fit and the data (as shown versus run number

in Figure 4.18) was a useful (though not definitive) guide in determining which runs

were fixed by which white noise spectra.

With only three sets of white noise spectra which correct the asymmetry runs

near it in time to the white noise run, only 46 runs out of 124 were able to be

corrected. There was an overlap of 8 runs between the 3 sets of data corrected by

the 3 different white noise runs.

After the DNL was corrected on the 46 runs, a three Gaussian fit to each

particle distribution and the standard 4 ns database cuts were applied. Numerical

integration was employed to find the inelastic dilution factors.

The contributions to the percent errors (added in quadrature to obtain the final
 104

FIG. 4.15: Corrected and uncorrected North American ToF spectrum.

percent error) on the inelastic dilution factors are:

• Error on the fit: the dispersion of the fit on a detector over 46 runs. This is

important since the dilution factors are sensitive to the size of the cut window

which is known to change over the data set. This was found to be 14%.

• Error on the knowledge of the DNL: the dispersion between the results from the

8 runs that are corrected by the 3 white spectra. This was found to be ∼ 1%.

• Error on the shape of the fit: the dispersion between different mathematical

equations to describe the same spectra. This was found to be 1%.

• Error on the average over octants: the dispersion between the dilution factors for

different octants. Having different cuts for the same detector but for a different

octant or having different set of thresholds might cause this dispersion. This was

found to be 12%. This is important in that it suggests that instead of quoting

an inelastic dilution factor for each detector for each octant, one can quote the

inelastic dilution factor for each detector averaged over the octants.
 105

Inelastic Dilution Number

The DNL Changes

Run Number

FIG. 4.16: Stability of the North American inelastic dilution factors over ∼50 runs for
a typical detector. Note that for ∼5 runs the inelastic dilution factors have changed in
a similar fashion as noted in Figure 4.12. This is presumably due to the DNL changing
with time. ToF spectra for these runs resemble the failing ToF spectra in Figure 4.17.
These failing runs were not used in calculating the inelastic dilution factors.

The results of the extraction of the North American inelastic dilution factors

can be found in Figure 4.19 and in Table 4.6.

Det Inelastic Dilution factor Error

1 0.118 0.022
2 0.145 0.027
3 0.182 0.034
4 0.216 0.040
5 0.203 0.038
6 0.253 0.047
7 0.277 0.051
8 0.272 0.050
9 0.287 0.053
10 0.284 0.052
11 0.304 0.056
12 0.297 0.055
13 0.311 0.057
14 0.378 0.070
TABLE 4.6: The average inelastic dilution factors and the final absolute errors for the
North American detectors.
 106

Events
Time (ns)

Events

Time(ns)

FIG. 4.17: Both ToF spectra have been corrected using the same white noise run. The
top ToF is taken to be a “passing” corrected spectrum. The bottom ToF spectrum is not
as smooth as the top ToF spectrum and it “fails” under the assumption that the DNL
has changed.

4.8.3 Inelastic Dilution Factor Results

The North American and French inelastic dilution factors have been extracted

(see Tables 4.5 and 4.6) for the 2002-2003 G0 commissioning run. The errors as-

sociated with the inelastic dilution factors allow for the extraction of the elastic

asymmetry with an error that is tolerable. The North American errors on the

inelastic dilution factors are considerably larger than the French results. This is

presumably, in part, due to the width of the elastic cut window which is affected by

the DNL.

The DNL that plagued the North American ToF spectra should be reduced in

the second G0 commissioning run with the replacement of the RF translator board.

White noise runs will be taken on a frequent basis (∼ 1/day) until the DNL is under

control and understood.

 107

Runs Corrected by DNL Run 15712

Fit-Data
Sum of

Run Number

FIG. 4.18: In deciding which runs are fixed by which white noise spectra, looking by eye
at the DNL corrected ToFs in conjunction with looking for minimums of the sum of the
fit-data was used. This quantitative method, though useful, was not definitive. Notice
fluctuations between runs 43 and 78 might be considered failing but when examined by
eye these ToFs appeared corrected.

FIG. 4.19: North American inelastic dilution factors as a function of detector number,
averaged over the North American octants.
 108
4.9 Background Inelastic Asymmetry

Asymmetry(ppm)

Bin

FIG. 4.20: Example of the coarser asymmetry binning to find the background inelastic
asymmetry for North American detector 4. The Time-of-flight asymmetries have been
broken down into 7 bins. The elastic proton bin is denoted in red. The asymmetry bins
above and below the elastic proton bin were interpolated to find the background inelastic
asymmetry under the elastic proton peak.

In order to extract the physics asymmetry, the background inelastic asymmetry

must be known, since inelastic events overlap with the elastic peaks and dilute the

elastic asymmetry. This is calculated as

Ael = (1 + d)Acorr − dAinel (4.34)

where Ael is the elastic asymmetry, Acorr is the corrected measured asymmetry from

false asymmetries due to helicity-correlated beam properties, A inel is the inelastic

asymmetry, and d is the inelastic dilution factor. The main contribution to the back-

ground comes from processes involved in scattering from the downstream aluminum

window of the target. This background represents 13-25% of the events within the

elastic cut window. The background fraction rises with higher detector number (and

thus momentum transfer, Q2 ). This background has a sizable asymmetry associated

with it, on the order of 10 ppm. This background is thought to be due mainly to

photo-disintegration of quasi-deuterons in the aluminum windows of the target.

 109

FIG. 4.21: North American and French extracted background asymmetries. These asym-
metries were obtained by dividing the ToF asymmetries into 7 bins and interpolating the
bin above and below the elastic cut window.

With only Time-of-Flight histograms, the background inelastic asymmetry was

extracted by dividing the Time-of-Flight spectra into 7 bins with the 5th bin being

the elastic proton cut (see Figure 4.20). A linear interpolation was made between

the side band bins below and above the elastic proton bin. The results can be seen

in Figure 4.21 and Table 4.7.

In the next commissioning run, the downstream window thickness will be re-

duced from 11 mils to 3 mils, which should reduce the background by nearly 60%.

The background asymmetries and yields will be directly measured during the next

run with dedicated dummy target runs with a 30 mil aluminum foil dummy target

(known as the “flyswatter”) and a 3.4 mil tungsten radiator. The purpose of the

flyswatter and radiator will be to confirm the expected fraction of events from the

downstream window and to be able to quickly collect asymmetry data on aluminum

to the level of a few ppm in a short amount of time.

 110
Det NA Ainel (ppm) FR Ainel (ppm)
1 -11.5 ± 6.9 -1.1 ± 5.8
2 5.9 ± 7.8 -12.7 ± 5.1
3 -10.8 ± 5.6 -8.9 ± 4.9
4 -13.2 ± 6.0 -5.9 ± 5.2
5 -5.3 ± 7.1 1.5 ± 4.9
6 -5.6 ± 6.9 -7.8 ± 4.6
7 -3.6 ± 5.3 -7.8 ± 4.4
8 -8.1 ± 6.3 -10.7 ± 4.6
9 -16.0 ± 5.4 -17.7 ± 4.3
10 -17.5 ± 5.9 -10.0 ± 4.3
11 -13.0 ± 4.4 -20.3 ± 4.1
12 -6.7 ± 5.4 -5.0 ± 4.4
13 10.7 ± 7.5 -0.3 ± 4.6
14 19.0 ± 7.4 6.0 ± 4.6

TABLE 4.7: The extrapolated background inelastic asymmetries for North American and
French detectors. These asymmetries were interpolated from averaging the N bins above
and N bins below the elastic cut window. These results are reported by detector number,
where the results for a detector number were averaged over the 4 detectors from the NA
(Fr) Octants, e.g. results reported for Fr Ainel Det 1 are the weighted average of detectors
1 from French Octants 2,4,6,and 8.

4.10 Polarimetry

The Moller polarimeter described in Section 3.8 was used to correct the physics

asymmetry for each detector

Ael
Aphy = (4.35)
Pb

where Pb is the beam polarization.

A limited number of measurements were made during the month of data taking

as can be seen in Figure 4.22. The average beam polarization was found to be (77.3

± 0.4)%.
 111

FIG. 4.22: Polarization measurements made during the engineering run. Polarization
measurements made with the insertable halfwave plate ‘in‘ must be multiplied by -1 to
compare to the insertable halfwave plate ‘out‘ measurements.

4.11 Radiative Corrections

Since the statistical error bars of this work are so large, the data have not been

corrected for electromagnetic radiative effects. This is because the radiative effects

are expected to be small, on the order a few percent of the measured asymmetry.

In order to carry out a complete analysis, radiative corrections should be taken into

account. In order to do this the following references are invaluable [63, 64].

There are two types of radiative corrections: External Bremsstrahlung correc-

tions and internal Bremsstrahlung corrections. External corrections are when the

beam electrons lose energy by bremsstrahlung from the target aluminum entrance

window or in the hydrogen itself before scattering off of a second proton and into

the detector. Internal corrections are when a beam electron interacts by more than

one photon with the proton. These internal corrections are further divided into

”real” and “virtual” processes. In real processes the photon is a real photon that
 112
is emitted during the scattering. In virtual processes, during the scattering virtual

photons are emitted and re-absorbed in the scattering process.

These higher-order interactions have two effects on the measured parity violat-

ing asymmetry

• The electron energy is reduced leading to a lower value of Q 2 and asymmetry for

a given scattering angle.

• The spin of the electron can be flipped, yielding a net depolarization.

These effects will reduce the measured asymmetry with respect to the tree

level asymmetry. Emission of the photons after the parity-violating interaction will

reduce the energy of the scattered proton leading to a reduction in the detector

signal. The effect of the internal and external bremsstrahlung is to remove protons

from the elastic peak and put them into a long tail.

In order to calculate the radiative correction, R c, it is necessary to calculate

two different parity-violating asymmetries: At , the tree level asymmetry from single

boson exchange and AR , the asymmetry including electromagnetic radiative effects.

The radiative correction is the ratio of the two asymmetries:

At (Q2 )
Rc = . (4.36)
AR (Q2 )

4.12 Q2 Determination

Knowing the elastic electromagnetic form factors from other experiments allows

for the extraction of the strange electric and magnetic form factors. In order to

perform this extraction, the Q2 must be known. The total error on the extracted

strange form factors should be smaller than 10%; this requires that Q 2 be known

to the 1% level. In order to reach this precision, one should know the absolute ToF

with an accuracy better than 50 ps.

 113
The average momentum transfer, < Q2 >, was determined by comparing the

detector ToF at different magnetic fields with a Monte Carlo simulation [65]. As

the magnetic field is varied, the pion peak will remain stationary while the proton

trajectory will shift and may even reach another detector system.

As mentioned above, this study relied on the results from simulation (G0GEANT,

G0TRACE, and GRAAL). The simulation takes into account the electronics, the

effects of the spectrometer magnetic field and the detector positions. These results

allow one to determine the value of all these parameters directly from the measured

ToF. Using these values allows one to find the < Q2 > for each detector.

When the magnetic field varies, the particle trajectories are modified; they

might even reach another detector. The elastic proton peak is modified by different

field strengths though the pion peak remains unchanged. The idea is to study

the magnetic field variation using the relative position between the pion and elastic

proton peak. From simulation one can see how the ToF should change with magnetic

field and then compare this to measurements.

With the French electronics, the proton peak may be determined within a few

ps (since the French electronics has 250 ps bins). Unfortunately, in the case of the

North American electronics, it is not possible to know the peak positions to better

than 60 ps for the Time Encoding Electronics; this was further complicated by the

North American DNL.

In conclusion, the simulation correctly reproduces the experimental data. The

< Q2 > were found (see Table 4.8) with a 1% precision.

 114

Det < Q2 > (GeV/c)2

1 0.12
2 0.13
3 0.14
4 0.14
5 0.15
6 0.16
7 0.18
8 0.19
9 0.21
10 0.23
11 0.26
12 0.30
13 0.34
14 0.40

TABLE 4.8: < Q2 > values determined by comparing Time-of-Flight differences between
pions and the elastic protons at various magnetic fields. Only data from the French
detectors were used to determine < Q2 > due to the fine time binning of the French
electronics (0.25 ns).
CHAPTER 5

Conclusions

Parity-violating electroweak asymmetries have been measured in elastic scatter-

ing of polarized electrons from the proton at forward angles. The asymmetries are

compared to the Standard Model assuming no strange quark contribution. Various

models for predicting strange quark contributions to the proton are discussed for

completeness. Future experiments using parity violation are described.

5.1 Discussion of Engineering Run

The first G0 engineering run (from October 2002 through January 2003) was

very successful. Each sub-system of the apparatus was commissioned.

Many of the challenges associated with generating and maintaining the unique

beam properties were met during the first engineering run. The time structure for

the electron beam was 32 ns which differs from CEBAF’s typical beam structure of 2

ns. This produces a higher bunch charge, due to having 16 times as many electrons

in one bunch, which in turn produces space-charge effects that complicate beam

transport through the injector. Most of the critical beam properties were delivered

in January 2003:
115
 116
• beam current of 40 µA,

• beam fluctuations in position ∆ x, ∆ y < 20 µm, beam fluctuations in intensity

∆ I/I < 2000 ppm.

Feedback systems used to minimize the helicity-correlated beam properties were

tested. The charge feedback system worked but the position feedback system re-

quires some investigation into its unstable behavior (the calibration slopes for the

PZT mirror seemed to change very quickly over a 3 hour period). The false asymme-

tries due to helicity-correlated beam properties were small and kept under control.

The G0 detectors performed well. The high voltages for the PMTs were set at

values that allowed for high detection efficiency and the PMTs were able to stand

rates at the nominal beam current of 40 µA. The gains were matched and their

stability was monitored and deemed satisfactory over time and for different beam

currents. The discriminator’s thresholds were adjusted to eliminate noise and low-

energy background while not rejecting the elastic proton signals. Typical detector

rates were on the order of 1-2 MHz with a typical dead time of 10%. This induces a

false asymmetry when coupled to a non-zero charge asymmetry with an uncorrected

effect of ∼ 15%; after correction ∼ 1%. The G0 detectors observed yield sensitivities

to six beam properties: helicity-correlated x and y beam motion, x and y beam angle,

the energy and beam charge.

There was higher background in Hall C than what was expected. This problem

had to be taken care of early on since the projected anode currents of the PMTs in

the higher numbered (larger Q2 ) detectors would have been too high at the nominal

current of 40 µA with the nominal gain settings. This higher background was due to

neutrals coming from the downstream beam pipe. This was taken care of by adding

a 4 inch thick lead box around the beam pipe.

The G0 magnet ran at full design current at 5000A for the first time on Decem-
 117
ber 18, 2002 and then throughout the January 2003 running.

The G0 target was well behaved with density fluctuations at 40 µA being neg-

ligible. Various target density studies were carried out to extract the contribution

from the target windows.

The background yields and asymmetries are needed to correct for the elastic

asymmetries. The inelastic yield under the elastic cut due to inelastic protons was

found to be about 13-25% of the signal. In order to cut down on the background

signal, in the forward angle run the downstream window thickness will be reduced

and an insertable dummy target will be added to help quantify the background.

5.2 Measured Physics Asymmetry

The physics asymmetry is given by

1
Aphy = ((1 + d)Ameas − dAinel ). (5.1)
Pe

Table 5.1 contains a list of the electroweak parity-violating asymmetries measured

by the first G0 engineering run.

The errors are determined by Equation 4.8 along with considering the error on

the polarization. Explicitly writing these out: the contribution to the error from

the measured asymmetry is given by

 2
2 1
σAmeas = (1 + d)2 ∆A2meas , (5.2)
Pe

the contribution to the error from dilution factor is given by

 2
2 1
σd = (Ameas − Ainel )2 ∆d2 , (5.3)
Pe

the contribution to the error from the background inelastic asymmetry is given by
 2
2 1
σAinel = d2 ∆A2inel , (5.4)
Pe
 118
and the error from the polarization is given by
 2
2 1
σ Pe = A2el ∆Pe2 . (5.5)
Pe

Table 5.2 contains a list of different error contributions to the physics asymmetry.

Det Q2 (GeV/c)2 NA Aphys (ppm) Fr Aphys (ppm)

1 0.122 -3.8 ± 2.5 ± 1.6 -4.9 ± 2.0 ± 1.6
2 0.128 -4.5 ± 3.2 ± 2.1 -3.9 ± 1.9 ± 1.5
3 0.135 -2.6 ± 2.5 ± 1.9 -3.8 ± 1.9 ± 1.5
4 0.143 -1.2 ± 2.7 ± 2.3 -6.6 ± 2.0 ± 1.5
5 0.152 -8.0 ± 3.4 ± 2.5 -4.1 ± 2.0 ± 1.7
6 0.163 -2.5 ± 3.5 ± 3.0 -2.0 ± 2.0 ± 1.6
7 0.177 -9.1 ± 2.8 ± 2.5 -7.3 ± 2.0 ± 1.7
8 0.192 -3.4 ± 3.5 ± 3.0 -6.7 ± 2.1 ± 1.6
9 0.209 -3.8 ± 3.0 ± 2.8 -6.6 ± 2.1 ± 1.6
10 0.231 -8.7 ± 3.5 ± 2.9 -11.3 ± 2.2 ± 1.7
11 0.260 -5.8 ± 2.8 ± 2.3 -5.6 ± 2.3 ± 1.9
12 0.298 -10.9 ± 3.4 ± 2.7 -9.1 ± 2.5 ± 2.0
13 0.341 -14.0 ± 5.2 ± 4.2 -10.3 ± 2.9 ± 2.4
14 0.404 -6.1 ± 5.8 ± 4.8 -10.3 ± 3.2 ± 3.0
TABLE 5.1: Extracted North American and French physics asymmetries. The
first(second) error is the statistical(statistical error on systematic effects) error. These
results are reported by detector number, where the results for a detector number were
averaged over the 4 detectors in a ring from the NA (Fr) Octants, e.g. results reported
for Fr Aphys Det 1 are the weighted average of detectors 1 from French Octants 2,4,6,and
8.

The data shows good agreement with the expected statistical properties. The

parity-violating asymmetries behave as expected. The asymmetries have the correct

sign and change sign under the influence of the insertable halfwave plate. The results

are consistent between the North American and French sets of detectors/electronics.

It is important to keep in mind when looking at the results of this work that the
1
amount of data taken during the first engineering run represents only ∼ of the
16
expected final statistics from the final forward production run. The statistical error

bars should be about 4 times smaller for the forward production run asymmetries.

Detectors 15 and 16 are not included in this analysis. Detector 15 contains

two Q2 points (0.45 < Q2 < 0.9 (GeV/c)2 ) in the elastic TOF spectrum. This
 119
complicates extracting the inelastic background. Decector 16 contains no elastic

protons in the TOF spectrum. This detector is used as a background detector. For

these reasons, detectors 15 and 16 are missing from the extracted asymmetry results.

NA Fr NA Fr NA Fr NA Fr
Det σd2 σd2 σA2 meas σA2 meas σA2 inel σA2 inel σP2 e σP2 e
(ppm)2 (ppm)2 (ppm)2 (ppm)2 (ppm)2 (ppm)2 (ppm)2 (ppm)2
1 0.04 0.00 6.21 4.17 2.06 2.13 0.00 0.00
2 0.05 0.00 10.16 3.55 3.86 2.02 0.00 0.00
3 0.09 0.00 6.064 3.60 3.17 2.06 0.00 0.00
4 0.24 0.00 7.29 3.94 4.83 2.00 0.00 0.00
5 0.00 0.00 11.47 3.97 5.94 2.48 0.00 0.00
6 0.03 0.00 12.19 3.95 8.57 2.35 0.00 0.00
7 0.00 0.00 7.79 3.95 6.03 2.48 0.00 0.00
8 0.11 0.00 12.31 4.55 8.22 2.25 0.00 0.00
9 0.41 0.01 9.15 4.37 7.29 2.22 0.00 0.00
10 0.14 0.00 12.56 4.73 7.89 2.51 0.00 0.00
11 0.12 0.02 7.80 5.06 5.05 3.13 0.00 0.00
12 0.04 0.00 11.84 6.37 6.96 3.80 0.00 0.00
13 2.15 0.01 27.51 8.17 14.86 5.44 0.00 0.00
14 2.07 0.04 3.72 10.04 20.55 8.87 0.00 0.00
TABLE 5.2: Individual contributions to the errors given by Equations 5.2 through 5.5.
Ameas contains corrections due to helicity-correlated beam properties and dead time.

This can be compared to the Standard Model strangeness-independent asym-

metry, Ath = η + ψGeA , taken from Equations 2.46 through 2.51, which is written

again here for convenience

 
GF Q2 2 Gγ,p γ,n γ,p γ,n 2 0 γ,p e
E GE + τ GM GM + 2(1 − 4 sin θW ) GM GA
Ath = − √ (1 − 4 sin θW ) − .
4πα 2 (Gγ,p 2 γ,p 2
E ) + τ (GM )

Deviations from AT h would imply the presence of strangeness in the proton. The

standard model strangeness-independent asymmetry in Figure 5.1 was calculated

using the dipole parameterization of the proton’s electric (with an uncertainty of 2%)

and magnetic form factors (with an uncertainty of 2%) , the dipole parameterization
 120

FIG. 5.1: Plot of the extracted North American and French asymmetries vs. momentum
transfer. The errors are statistical and systematic errors added in quadrature. The
dashed line represents the Standard Model calculation of the parity-violating asymmetry
assuming no contributions from the strange quarks.

of the neutron’s magnetic form factor (with an uncertainty of 3%) and the Galster

parameterization of the neutron’s electric form factor (with an uncertainty of 20%).

5.3 Theoretical Predictions

A proper description of the strange form factors should be based on QCD. The

problem is that the mass of the strange quark, ms ≈ 150 MeV is comparable to the

QCD scale factor λQCD , thus not easily allowing for a small expansion parameter as

used for the heavier quark calculations. This forces the theorist into the territory of

models and chiral perturbation theory. Several review articles on this subject can be
 121
found in the literature [66–68]. Even the applicability of chiral perturbation theory

is called into question here since the strange quark mass may not be light enough
mK 1
to make the SU(3) chiral perturbation valid since λχ
∼ 2
which is not particularly

small. Even if the SU(3) chiral expansion is well behaved, there appears to be various

counter terms (low energy constants) that have not been measured in experiment

and must be extracted from various models. Typically the form factors associated

with s̄γµ s are characterized by the strange magnetic moment µ s ,

µs ≡ GsM (Q2 = 0) (5.6)

and by the strangeness radius, rs2 ,

 
dGsE
rs2 ≡ −6 . (5.7)
dQ2 Q2 =0

Loop Models

Proton Proton

FIG. 5.2: Kaon loop diagram.

A set of models known as “loop models” describes the strangeness content of

the nucleon in terms of pairs of KΛ,KΣ, or ηN components. The nucleon fluctuates

into a q q̄ pair to form a meson and an intermediate baryon state. The meson

and baryon later recombine as the q q̄ pair annihilates and the original nucleon is
 122
left in the ground state. Only diagrams involving kaons and strange baryon states

contribute to yield non-zero strangeness. Koepf [69] first evaluated µ s and rs2 but

did not include the so-called “seagull” diagrams. These diagrams are needed to

satisfy the Ward-Takahashi identity. These were later added by M. Burkhardt et

al. [70]. The predictions of rs2 in the kaon loop calculation tends to be smaller than

the pole-fit analysis. To reconcile this difference, the kaon loop model was merged

with VMD and ω − φ mixing.

Lattice QCD

Lattice QCD computations can provide a means of obtaining values for the

low energy constants that have not been measured. These calculations are typically

carried out in the “quenched” approximation where the ss̄ pairs appear only via

operator insertion. To achieve a firm lattice QCD prediction requires resolving

several issues. One is to perform an unquenched calculation. Another issue is that

one would like to have light quark masses that one can extrapolate to physical values

using chiral perturbation theory.

Dispersion Relations (Pole-Fit Analysis)

This is a first-principles approach to calculating G sE and GsM . This analysis

involves various inputs such as form factors and experimental scattering amplitudes.

The nucleon strangeness arises from the nucleus coupling to a strange meson. In

this case, the exchanged vector boson (Z 0 or γ) fluctuates into an isoscaler meson

(ω or φ) which interacts with the nucleon. The ω and φ are linear combinations of

strangeness and non-strangeness components.

Jaffe [71] was the first to make theoretical predictions of µ s and rs2 . Jaffe based

his analysis on the work of Hohler [72]. Hammer et al. [73] updated this analysis

using a new dispersion-theoretic analysis of the nucleon electromagnetic form factors.

 123

φ/ω

Proton Proton

FIG. 5.3: Pole loop diagram.

A noticeable point with these analyses is that they typically yield a different sign of

the strange electric radius compared to most other models.

5.4 Future Parity-Violation Experiments

HAPPEX-II and HAPPEX-He

The HAPPEX II [84] experiment at Jefferson Lab, in Hall A, proposes to con-

strain the nucleon strangeness radius of the proton:

ρs + µp µs

GsE
where ρs is the strangeness radius defined as ρs =, µp is the proton magnetic
dτ
moment and µs is the strange magnetic moment. This will be done by measur-

ing parity-violating asymmetries in elastic scattering of 3.2 GeV polarized electrons

from an unpolarized hydrogen target. The measurement is made at a forward scat-

tering angle (θ = 6◦ ) corresponding to a Q2 of 0.11 (GeV/c)2 (see Figure 5.4). The

expected physics asymmetry will be about 1.7 ppm. This measurement is com-

plementary to the SAMPLE [13] measurement at MIT Bates at the same Q 2 but
 124

Model Author µs (n.m.) rs2 (f m2 )

Pole Jaffe [71] -0.31±0.09 0.11 - 0.22
Pole Hammer [73] -0.24±0.03 0.19±0.03
Pole Meissner [74] 0.003 0.002
Pole Forkel [75] -0185±0.075 0.14±0.06
Loop Burkhardt [70] -0.355±0.045 -0.0297±0.0026
Loop Geiger [76] 0.035 -0.04
Loop Koepf [69] -0.026 -0.01
Loop+VMD Cohen [77] -0.28±0.04 -0.0425±0.0026
Skyrme Park [78] -0.13 -0.11
Skyrme Park [78] -0.33 -0.19
Lattice QCD Dong [79] -0.36 ±0.20 -0.16±0.06
HBχPT Hemmert [80] 0.18±0.34 0.05±0.09
NJL Soliton Abada [81] 0.10 ±0.15 -0.15±0.05
χSM Goeke [82] 0.115 -0.095
PχQM Gutsche [83] -0.048±0.012 -0.011±0.003

TABLE 5.3: Some predicted values of strangeness radius rs2 and strange magnetic moment
µs .

different kinematics. At the same kinematics, there is another proposed experiment

in the HAPPEX family: the HAPPEX-He [85] experiment will measure the parity-

violating asymmetries of polarized electrons scattering from 4 He nuclei. Scattering

from 4 He will be sensitive only to GsE and not GsM or GeA due to the fact that 4 He is a

0++ nucleus. With the 4 He measurement of GsE (Q2 → 0) = ρs and the HAPPEX-II

proton measurement, HAPPEX will be able to separately extract both ρ s and µs .

PVA4

The PVA4 experiment [51], besides making forward angle measurements as

described in Section 2.7, will be performing backward angle measurements. The

detector system will be reversed relative to the target to measure back-scattered

electrons between 140◦ < θe < 150◦ . These measurements will be made at two

values of Q2 = 0.23 (see Figure 5.4) and 0.48 (GeV/c)2 in order to complement the

forward angle PVA4 and HAPPEX measurements.

 125
G0 Forward Angle

FIG. 5.4: Expected forward angle results from the G 0 , HAPPEX II [84], and A4 [51]
along with result from the HAPPEX [14] experiment. A linear combination of the strange
and electric form factors are accessible from forward angles. This linear combination is
of the form of GsE + α(Q2 )GsM where α(Q2 ) is dependent on kinematic factors.

In the winter of 2004, G0 will make forward angle measurements of electroweak

asymmetries on polarized electrons scattering from unpolarized hydrogen. These

measurements will be made over a momentum transfer of 0.1-1.0 (GeV/c) 2 (see

Figure 5.4). This will allow for an extraction of the vector neutral weak form

factors. Combining the neutral weak form factors with the known electromagnetic

form factors allows for an extraction of the linear combination of strange electric,

GsE (Q2 ), and strange magnetic, GsM (Q2 ), form factors over this momentum transfer

range.
 126
G0 Backward Angle

Beginning in 2006, G0 will begin the backward angle measurement phase of

the experiment. Electroweak asymmetries will be measured in backward-scattered

electrons from polarized electron scattering from hydrogen and deuterium targets.

This asymmetry allows for an extraction of a linear combination of strange mag-

netic, GsM (Q2 ), and the electron-nucleon axial, GeA (Q2 ), form factors. Three sets of

measurements will be made in order to obtain these asymmetries at three different

momentum transfers: 0.3, 0.5, and 0.8 (GeV/c)2 . Combining these linear combina-

tion of GeA (Q2 ) and GsM (Q2 ) with the G0 forward angle measurements, which extract

a linear combination of GsE (Q2 ) and GsM (Q2 ), will allow a separation of the strange

electric, strange magnetic, and axial form factors as a function of Q 2 .

APPENDIX A

G0 Abbreviation & Acronym

Glossary (GAAG)

Acorr : The measured elastic asymmetry that has been corrected for helicity

correlated false asymmetries due to beam properties.

Ael : The measured asymmetry.for elastic events.

Ainel : The asymmetry for inelastic protons that must be removed from the cor-

rected measured asymmetry.

Ameas : The experimentally measured raw elastic and inelastic proton elec-

troweak asymmetry.

Aphy : The final fully corrected measured elastic electroweak asymmetry.

Ath : Tree-level Standard Model prediction of the electroweak asymmetry.

127
 128

BCM: The Beam Current Monitor (BCM) is a cylindrical cavity whose reso-

nant frequency is adjusted to 1497 MHz (the frequency of the typical CEBAF beam).

Inside the cavity is a loop antenna located where the electric field is minimum and

the magnetic field is at a maximum. The antenna is coupled to one of the resonant

modes of the cavity and the output signal is proportional to the beam current.

BPM: The Beam Position Monitor (BPM) consists of four metal strips sur-

rounding the beamline. When the electron bunches pass through the BPM, a signal

is produced by induction. The output signals from the four strip lines are then

combined to yield beam position information.

CFD: The Constant Fraction Discriminator (CFD) are designed to produce ac-

curate timing information from analog signals of varying heights but with the same

rise time. This will reduce the “walk” of the output signal.

χPT: χPT is a short-hand notation for Chiral Perturbation Theory.

DMCH-16X: The DMCH-16X (Discriminator, Mean-Timers, time digital Con-

verter, Histogramming, 16 channels, and X is for VXI standard) is the French elec-

tronics.

DNL: The Differential Non-Linearity (DNL) is defined as deviations from the

nominal 1 ns wide bin structure of the North American time-of-flight spectra.

FPD: The Focal Plane Detectors (FPD) are 16 iso-Q2 double layered scintilla-

tor detectors located at the focal plane of the spectrometer.

 129

Fr: Fr is a short-hand notation for “French”.

GMS: The Gain Monitoring System (GMS) shines laser light onto the scintil-

lators in order to monitor the gain changes in the photo-multiplier tubes.

IA: The Intensity Attenuator (IA) is a charge feedback device that controls the

helicity correlated charge asymmetry.

IHWP: The Insertable Halfwave Plate (IHWP) is used to change the helicity

of the polarized light coming from the G0 laser on the laser table in the injector.

LTD: The Latching Time Digitizers (LTDs) are specialized electronics used to

bin detector signals into 1 ns time bins for inputs to the scalers.

MPS: A Macro-pulse (MPS) is one helicity state that lasts for 33 ms (1/30

second).

NA: “NA” is an acronym for North American.

NPN: Next Pulse Neutralization (NPN) is the disabling of the LTDs for the

next beam burst (32 ns) after a hit has been recorded from a detector. This allows

the mean-timers to clear and allows for an exact calculation of the deadtime.

PMT: “PMT” is an acronym for Photo-Multiplier Tube.

ppm: “ppm” represents parts-per-million.

 130

PZT: “PZT” is an acronym for Lean Zirconate Titante. This device is a mirror

attached to a piezo-electric mount that changes the angle of the laser beam before

entering the Pockels cell. This is used to minimize helicity correlated position dif-

ferences.

QRT: A quartet (QRT) is a sequence of 4 macro-pulses from which an asymme-

try can be formed. The helicity of the first macro-pulse is chosen pseudo-randomly

with the next two macro-pulse helicities the complement of the first macro-pulse.

The fourth macro-pulse is the same helicity as the first macro-pulse. This allows for

two different quartet patterns (-++-) and (+–+).

RHWP: The Rotatable Halfwave Plate (RHWP) is used to minimize the charge

asymmetry by rotating the residual linear component of the slightly elliptical light

emerging from the helicity Pockels cell..

SMS: The Superconducting Magnet System (SMS) is the 1.6 T·m magnet used

in the G0 experiment.

ToF: The Time-of-Flight (ToF) is amount of time it takes for a particle to reach

the detectors from the G0 target.

APPENDIX B

Detector Testing and Calibration

B.1 Output signals of the North American Focal

Plane Detectors

During the G0 experiment, a particle is detected if the signals it produces in

each of the four photo-multiplier tubes (PMT) attached to the light pipe pair are

above a certain discriminator threshold. The amplitude of the signal produced by

each PMT can be approximated by the following expression :

Amplitude = Nγe × gain(HV ) × Acable (B.1)

where :

- Nγe is the number of photo-electrons1 produced at the photo-cathode of the PMT,

- gain(HV ) is the gain2 which increases as the HV applied to the PMT increases,
1
A photo-electron is an electron produced by photo-electric effect when a (scintillation) photon
hits the PMT photo-cathode.
2
The gain of a PMT is the factor of amplification of the photo-electrons through the dynodes
chain of the PMT.

131
 132
- Acable is the attenuation of the signal through the cables between the PMT output

and the discriminator input.

In order to produce a signal of large enough amplitude to pass the discriminator

threshold, one may consider increasing the gain. This option seems the most efficient

although its long term effect in a radiation harsh environment should be considered.

A large gain could produce a large anode current and as a consequence an early

aging of the PMT.

Another method is to increase the amplitude of the signal by maxiimizing the

number of photo-electrons. It has been determined that for the G 0 experiment a

minimum number of 40 photo-electrons must be produced by particles hitting the

scintillators [86]. The determination of this minimum output is described in the

following section.

B.1.1 Light Pipe Characteristics

A light pipe is characterized the number of photo-electrons, N γe , produced by

the PMT attached to it. This quantity is the following product:

Nγe = ∆E × C × P MTQE × P M Tcontact × Ascint × Aglobal (B.2)

where :

- ∆E is the energy lost by the particle passing through the scintillator, and C is

the conversion factor between energy lost and photon produced. Those quantities

are characteristic of the scintillator type and the energy of the detected particle,

and are not in the scope of this report.

- P M TQE is the quantum efficiency of the PMT photo-cathode, and P MT contact

is the fraction of photons exiting the light guide end that actually hit the PMT
 133
photo-cathode.

- Ascint is the attenuation of photons along the scintillator from the hit position to

the glue joint between the scintillator and the light guide. A global is the attenua-

tion of the photons going through the glue joint between the scintillator and the

light guide and the bulk attenuation along the light guide.

Measurements of the number of photo-electrons extracted from the G 0 NA FPD

have been performed during the assembly of the detectors; they took place at JLab

in the so-called Clean Room. They are described in the following section. For

clarity, it has been decided to separate the characterization of the light pipe and the

characteristic of the PMT. Therefore, the following results are given for an arbitrary

but constant value of P M TQE and P MTcontact as described later. Ascint and Aglobal

can vary from one light pipe to the other as they are a function of the quality of the

scintillator surface and the length of the scintillator (A scint ), or the global quality of

the light guide and the gluing between the scintillator and the light guide (A global ).

Setup and Calibrations

As soon as the light pipe have been mounted on one octant, the ends of the

light pipe were equipped with PMTs and the assembly was rolled into a dark space.

The signals were produced either by a radioactive source placed on the scintillator, a

LED illuminating the faces of the PMTs, or cosmic rays. The systematic calibrations

performed in order to estimate the number of photo-electrons produced by the light

pipes were :

1. The photo-electron calibration :

The signal produced by a particle passing through the scintillator is measured as

a charge in an ADC channel. In order to determine the number of photo-electrons

produced, it is necessary to know the charge produced by a single photon hitting

 134
the PMT photo-cathode. The calibration was performed by shining a LED in

front of the PMT face. The brightness of the LED was reduced by lowering the

voltage applied to the LED down to a point were only one photon at a time

was detected by the PMT. The ADC signal measured in that case was very small

and usually overlapped with the pedestal of the ADC. Two independent methods

were used to amplify the single photo-electron signal. In one hand the signal was

amplified using a calibrated analog amplifier. The precision of this determination

is ∼ 10%. In the other hand, the single photo-electron peak was measured with a

large HV applied to the PMT, the calibration was then extrapolated to “regular”

HV using a calibrated gain-HV curve. The precision of this calibration is also

∼ 10% due to the extrapolation. The two methods agree within error bars.

For the tests performed in the Clean Room, only eight PMTs were used. Each

PMT was mounted in the same position and tested. A photo-electron calibration

was performed for each of those PMT before almost all measurements using the

analog amplifier method. The gain of each tube was found to be constant in

the 10% error bar during the course of the measurements (almost a year). The

results in term of photo-electrons presented later are computed with the daily

calibrations, therefore their precision is 10%.

2. The determination of P M TQE and P MTcontact :

The number of photo-electrons detected at the end of the light pipe depends

on the quality of the PMT used for detection because of the quantum effi-

ciency (P M TQE ) of the cathode. It also depends on the quality of the con-

tact (P M Tcontact ) between the tube and the light-guide. The reproducibility of

P M Tcontact was measured in the following way. The radioactive source was placed

on a scintillator, and was not moved during the whole course of the test. The

signal produced by the source was measured in an ADC channel. The contact
 Minimum Maximum
135

3


1





  




 






 








 

2


Point of Observation
FIG. B.1: Positions of interest along the G0 scintillator. A signal produced in position 2
will undergo a maximum attenuation in its travel to the point of observation. Position 3
is the geometrical middle of the scintillator, a signal produced at this position therefore
undergo an average attenuation to the point of observation. Signals produced in position
1 undergo a minimal attenuation into the scintillator before being detected.

between the PMT face and the light pipe was broken by removing the PMT

from its housing. The silicon cookie was unstuck from the face of the PMT. The

PMT and the cookie were then put back in the housing therefore creating a new

contact between the PMT face and the light pipe. The signal of the radioactive

source was re-measured. The process was repeated ten times and the ADC value

was found to be consistent within 5%. The measurement of the relative tube

quantum efficiencies (P M TQE ) was performed using the same protocol. Again,

the radioactive source stays fixed. The output signal is measured in terms of

photo-electrons for different tubes inserted in the housing. It was found that

the number of photo-electrons produced by four different tubes used during this

testing differed by 15%. In the following, the different relative quantum efficien-

cies are corrected to normalize the signal measured with different tubes between

themselves. Nevertheless, one should consider this uncertainty when quoting the

number of photo-electrons produced at the end of the light pipes.

Measurements

To determine if the G0 particle going through the scintillator will produce at

least 40 photo-electrons, the signal in the worst case must be determined. Prelim-
 136
inary measurements [87] and simulation [88] have shown that the worst case is for

the particle hitting the far end of the scintillator at the bottom of the scintillator

(position 2 in Figure B.1). The attenuation of the signal along the scintillator as

well as the attenuation along the light pipe or through the glue joints were also

measured by measuring the amplitude of signals produced in position 1, 2 and 3

of Figure B.1. Those data allow us to differentiate between bad glue joints and

a bad scintillator. The data of attenuation along the scintillator can be compared

to data after data-taking with beam to indicate possible yellowing of the scintillator.

The most straightforward way to do such measurements is to use a collimated

radioactive source (known amount of energy deposit) located in different positions

along the scintillator. The ratio of the signal for position 2 and 3 (see Figure B.1)

gives the attenuation along the scintillator. The measurement in position 3 gives

the average number of photo-electrons, useful to compare from one scintillator to

another one. For this measurement, a Ru-106 source [89] was used. It emits betas

up to a maximum energy of 3.5 MeV. By using the appropriate trigger (requiring

the betas to traverse both scintillators and the plastic spacer) and discriminator

threshold, one is able to select only the most energetic part of the beta spectrum.

The energy deposited by the beta is 1.97 MeV per cm. Those measurements have

been carried out for 41 scintillator pairs out of the total of 64.

This method is tedious and cannot be used once the detector is enclosed in

the light tight box. Part of the calibration of the light pipes were carried out using

cosmic (µ) rays with the octants oriented such that the scintillators were concave

towards the earth and the scintillator faces perpendicular to the ground. The cosmic

trigger required all four PMTs on one detector to fire and two of the PMTs from an

adjacent detector to fire. This method is quicker as one can test many light pipes
 137
at a time. Also, this method can be used once the detectors are in the hall (and

enclosed in the light tight box). While the energy deposited (per cm) by cosmic

rays is known, the path the cosmic takes through the scintillators is not known.

For this reason, the signals produced by cosmic rays were calibrated by comparing

them with the signals produced by the Ru source located in position 3 on Figure B.1.

The calibration was performed on all scintillators of Octant 7. As a result it appears

that one can use a constant multiplicative conversion factor between the cosmic and

Ru-106 amplitudes. The ratio of Ru-106 to cosmic amplitudes is plotted in Figure

B.2. The 10% dispersion in the data, is mostly due to the precision which which

the cosmic peak was located on the ADC distributions. No significant variation

of this ratio was found as a function of the size of the scintillator. Also, the data

for cosmic runs were taken with the same HV being applied on the PMT, thus the

photo-electron calibration precision does not contribute to the dispersion. Using

this method, one can measure the average number of photo electrons produced by

the light pipes. It is possible to measure the attenuation of the signal along the

scintillator using the cosmic ray data. A careful measurement of the time arrival of

signals between each other allows us to locate the hit position. Though the principle

was demonstrated, the quality of the data taken at that time did not allow us to

extract this information. The cosmic method was applied to the 23 scintillator pairs

that were not measured using the source method.

Results and extrapolation to the G0 case.

Raw results of the testing are presented in the upper plots of Figure B.3. The

results are three-fold. Each scintillator end is characterized by :

1. the number Nγe (mid) of photo-electrons produced by a minimum ionizing particle

crossing it in its geometrical middle (position 3 on Figure B.1).

 138

Measurements
3.575 / 4
Constant 18.76 3.709
Mean 0.8256 0.1569E-01
20
Sigma 0.1031 0.1362E-01

17.5

15

12.5

10

7.5

2.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Ru/Cosmics

FIG. B.2: Calibration of the signal produced by non-collimated cosmic rays crossing the
scintillator against the signal produced by the Ru source located in the middle of the
scintillator. Those data represent the calibration of the signal for all scintillators of
Octant 7.

2. the attenuation (Afscint

ar
) of the signal from the far end to the geometrical middle

(signal of position 3 over signal on position 2 on Figure B.1)

3. the attenuation (Aclose

scint ) of the signal from the geometrical middle to the close

end (signal on position 1 to signal on position 3 on Figure B.1)

The analysis of the raw data shows that scintillators of the same size with the

same light-guide configuration3 gives in average similar results (see lower plots of

Figure B.3). Moreover the dispersion of 10% on Nγe can be interpreted as a result of

the 10% precision on the photo-electron calibration. In other words, no scintillator

3
Front light pipe and back light pipe do not have the same light-guide geometry.
 139
shows a significantly better or worse performance than equivalent other detectors.

This is also the case with Aclose

scint , moreover for this variable, the analysis of the data

shows that this attenuation is independent of the scintillator size and is found to be

1.44 ± 0.05. As a result, the average values (see Figure B.3: middle plots and Table

B.1) are going to be used to extrapolate data to the G 0 case.

The following equation describes the extrapolation of the Clean Room data to

the minimum number of photo-electrons Nγe (g0) produced by G0 particles hitting

the scintillators.

Nγe (g0) =< Nγe (mid) > × ∆E(g0)/∆E(Ru) × < Afscint

ar
> (B.3)
| {z }
energy normalization

For the G0 backward running, the energy normalization is 1 as electrons are

detected : ∆E(g0) = ∆E(Ru) is the energy lost by minimum ionizing particles. For

the G0 forward running, ∆E(g0) is lost by low-energy protons hitting the scintilla-

tors. The variation of energy lost by protons from one scintillator size to the other

one is significant [88]. For example the energy lost by the proton in scintillator 5

(1 cm thickness) is on average 9.3 MeV whereas the energy lost in scintillator 16 (1

cm thickness) is on average 3.0 MeV. Moreover, this energy loss can vary by up to

25% across the face of a single scintillator; the minimum energy lost is for protons

crossing the scintillator on the top as they are the more energetic. This minimal

energy loss is considered for the computation of N γe (g0). This is the absolute worst

case scenario, as one is combining the least energy deposited (top of the scintillator)

with the worst transmission (bottom far end of the scintillator). The energies taken

into account for the extrapolation to the G0 forward running are given in Table B.2,

they have been computed using the Bethe-Bloch formula. The computation takes

into account the different materials crossed by the protons before hitting the scintil-

lators ; the main losses occur in the LH2 target, the air gap and, when relevant, the
 140

G0 NA-FPD : Cleanroom data

Ascintfar
400 2.5
Nγe(mid)

2
300
1.5
200
1
100
0.5

0 0
0 5 10 15 0 5 10 15
scint. # scint. #

<Ascintfar>
400 2.5
<Nγe(mid)>

2
300
1.5
200
1
100 Front detectors
0.5
Back detectors
0 0
0 5 10 15 0 5 10 15
scint. # scint. #

100 100
events

events

8.254 / 10 5.889 / 5
Constant 61.99 5.296 Constant 64.86 7.219
Mean 0.9976 0.7218E-02 Mean 1.005 0.5681E-02
80 Sigma 0.1116 0.6241E-02
80 Sigma 0.6764E-01 0.5218E-02

60 60

40 40

20 20

0 0
0.5 1 1.5 0.5 1 1.5
Nγe(mid)/<Nγe(mid)> Ascintfar/<Ascintfar>

FIG. B.3: Results of the testing with minimum ionizing particles. The tests were per-
formed with the Ru radioactive source or with cosmic rays, following a procedure described
in Section B.1.1. The left plots refer to the average number of photo-electrons, while the
right ones refers to the attenuation of the signal from the far end of the detector to the
average signal. Upper plots show the raw data, middle plots show the average value on a
scintillator size to scintillator size basis, lower plots show the deviation of the actual data
(upper plot) to the average value (middle plot) again on a scintillator size to scintillator
size basis.
 141

FIG. B.4: Minimum number of photo-electrons detected by the G 0 NA FPD in the case
of the proton forward angle running running. Error bars on these predictions are 20%.
Note that detector 4 is made up of 1 cm thick front scintillator and 0.5 cm thick back
scintillator. The horizontal line indicates the design minimum value of 40 photoelectrons.

front scintillator and the polycarbonate. The minimum number of photo-electrons

produced in the NA-FPD light pipes in the cases of the G 0 running is presented

in Table B.2 as well as in Figure B.4. The error associated with this estimation

is the quadratic sum of the precision on the photo-electron calibration (10%), the

estimation on the stability (from one tube to another one) of the quantum efficiency

of the PMTs (15%) and the precision of the energy loss computation (10%). The

total precision is therefore 20%.

In this worst case estimation, the number of photo-electron produced by the

North American Focal Plan Detectors (NA FPD) is always larger than 40 photo-

electron. Moreover, the number of photo-electrons produced in the Forward case

running is always larger than 100. This minimum should allow for the time of flight

measurement. As a conclusion, the NA FPD produce more than enough light to

insure a good measurement when mounted in the Clean Room.

The next section explains how the PMT were matched to the light pipes.
 142

Clean Room data : minimum ionizing particle.

Scintillator 1 2 3 4 5 6 7 8
< Nγe (mid) > front 125 124 108 245 266 221 231 262
< Nγe (mid) > back 94 92 86 101 219 219 201 212
f ar
< Ascint > 1.19 1.17 1.22 1.25 1.18 1.16 1.18 1.23

Scintillator 9 10 11 12 13 14 15 16
< Nγe (mid) > front 240 213 198 207 232 166 170 166
< Nγe (mid) > back 189 187 152 165 200 145 148 147
< Afscint
ar
> 1.46 1.62 1.50 1.35 1.47 1.93 1.69 1.74
TABLE B.1: Results of the testing with minimum ionizing particles. N γe front,Nγe back
and Afscint
ar
(and their errors) are defined in the text.

Proton case : forward running for G0

Scintillator 1 2 3 4 5 6 7 8
∆E(g0) front 6.19 5.92 5.59 10.5 9.98 9.37 8.75 8.11
∆E(g0) back 7.68 7.17 6.59 6.62 12.1 11.0 10.0 9.06
Nγe (g0) front 660 637 502 1049 1142 906 870 877
Nγe (g0) back 616 572 472 543 1143 1058 866 793

Scintillator 9 10 11 12 13 14 15 16
∆E(g0) front 7.58 6.99 6.41 5.78 5.23 4.73 4.10 3.68
∆E(g0) back 8.31 7.53 6.79 6.03 5.41 4.84 4.16 3.71
Nγe (g0) front 633 467 430 450 419 207 209 178
Nγe (g0) back 546 441 349 374 374 185 185 159
TABLE B.2: Minimum energies left (in MeV) by the protons (G0 forward running) hit-
ting the G0 scintillators and subsequent number of photo-electrons detected. The energies
left take in account the variation of scintillator thickness.
 143
B.1.2 PMT attributes

To understand the characteristics of the scintillator and light pipes, the previ-

ous measurements were performed using the same eight PMTs. The characteristics

of those eight PMT were very well known : their gains were tracked on a regular

basis, and their relative quantum efficiencies measured. After these tests had been

performed, each detector had to be fitted with its own PMT.

Gain Matching

The goal of the gain matching process was to pair the PMTs to specific light

pipes such that if one applies a given HV value to all of them, the output signal will

be roughly equivalent. In other words, one tried to compensate for the variation in

the number of photo-electrons (Nγe (g0)) produced by the G0 particles (see Figure

B.4) by carefully choosing the gain (gain(HV )) of the PMTs at a given HV. That

is :

Nγe (g0) × gain(HV ) ∝ amplitude = constant. (B.4)

For this computation, the PMT gains measurements performed at JLab [90]

were used. Those measurements were performed using a regular resistive basis and

not the final Zener-resistance G0 basis. The number of photo-electrons (Nγe (g0))

used for this pairing are the ones corresponding to the forward angle setup,in which

protons will be measured. There was nearly a perfect gain match for detectors 1

through 12 (see Figure B.5). The last three detectors required a gain that could not

as easily be matched to a PMT of sufficient gain to compensate the lower number

of photo-electrons. The PMTs on these later detectors will have a higher voltage

applied to them compared to the lower detectors.

 144

Average Actual:Predicted Gain vs Detector

2

Ratio of Average Actual:Predicted Gain

1.8

1.6 Octant 1

Octant 3

1.4 Octant 5

Octant 7

1.2

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14 16 18
Detector

FIG. B.5: Ratio of the gain of the actual PMT attached to the light pipe to the gain
necessary to perfectly match the light pipe. For detectors 13 through 16, not perfectly
matched PMT have to be used. This implies that the high voltage will be higher on these
detectors in comparison with the other detectors.
events

35 6.984 / 8
Constant 28.87
Mean 0.9661
Sigma 0.3013
30

25

20

15

10

0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
predicted gain / measured gain

FIG. B.6: Dispersion of the ratio of the measured to the predicted cosmic amplitudes in
the Clean Room. The 31% error is due to the gain measurements, the assumed quantum
efficiency, and to the fitting of the cosmic and pe peak for the Ru.
 145
Gain Balancing

The optimization described in the previous section is based on the use of various

data sets. It also supposed that the quantum efficiency of all PMTs is equivalent.

In order to check the validity of the previous matching as well as the running of

the bases and the tubes, a cosmic measurement was performed on all the light pipe

equipped with their proper PMTs. In addition, cosmics were available in both the

Clean Room and down in Hall C. This meant that results in the Clean Room could

be compared with results after moving the octants into Hall C. The amplitude of

the signal produced by cosmics rays going through whatever scintillators is brought

to a constant value by adjusting the HV applied to the PMT.

Using the procedure described in Section B.1.2, one is able to compute the HV which

should be applied to the tube so that the cosmic peak will have a given amplitude 4 .

In this case one should take care of correction factors not considered in Equation

B.4, that is :

amplitude ∝ Nγe (g0) × gain(HV ) × P MTQE × Acable (B.5)

where P M TQE is the quantum efficiency of the tube and Acable is the attenuation

of the signal through the cable between the PMT and the ADC channel. For this

measurement Acable was known. The variation5 of the attenuation from one cable

to another one was corrected for.

Figure B.6 shows the ratio of the signal amplitude produced by the actual light

pipe with the amplitude expected. The ratio is distributed as a Gaussian of sigma

31%. The larger part of this dispersion can be explained by taking in account the

precision of the data used to compute this ratio :

4
Note: In this case minimum ionizing particles are used and not protons as discussed in Section
B.1.2
5
Different types of Lemo cables were used, the older set eventually transmit only 88%(± 2%)
of the portion transmitted by the newer set.
 146
- The actual amplitude is measured with a precision of 10%.

- The precision on the number of photo-electron produced by each light pipe :

10%.

- The dispersion in the quantum efficiency of the tubes: 15%

- The knowledge of the gain of each tube : 15%. This value is estimated by

comparing the gain of the eight tubes used for the initial calibration with the one

given in [90].

This yields an error of 25%. Thus the gain balancing process appears as a successful

cross check of our previous work and of the assumptions made during testing.

From these measurements the high voltage was then adjusted until the cosmic

peaks were aligned within 10%. From this data, the high voltage can also be adjusted

for the G0 proton case for the experiment. In that case, one should take in account

the appropriate energy loss in the scintillator.

APPENDIX C

Injector Studies

C.1 Overview of 2002 Results

The G0 commissioning run began in August 2002 until the end of January 2003.

This was an opportunity to test many of the systems under G0 running conditions.

This appendix discusses results of the commissioning run to understand the helicity

correlated devices at the source that were monitored by the injector DAQ system.

C.1.1 BPM Noise

The BPMs used in the injector are the standard JLab stripline BPMs. These

BPMs are made up of four antennae situated symmetrically about the beam pipe.

When the electron beam passes through the stripline BPM, rf signals are picked up

by the monitors:

BPM Antenna Signals ∝ (Beam position) × (Beam Intensity).

The position (or position difference) calculated from beam monitors have a

certain amount of noise associated with it. This noise is due to beam noise and to

electronic noise. The measured noise can be written as:

147
 148

FIG. C.1: The top two plots show the BPM resolution. The bottom two plots are the
profile plots of the top scatter plots. These profile plots can now be fitted and one axis
can be used to predict the position of one BPM from the measured position in another
BPM.

2 2 2
σmeasured = σbeam + σinstrumental .

The instrumental noise can be found by using three BPMs along the beamline

without magnetic optics between the monitors. The first two BPMs can be used to

determine the position of the beam in the third monitor. This predicted behavior

can then be removed from the measured signal of the third BPM leaving behind

only the instrumental noise. This analysis has been done when looking at absolute

positions and position differences. See Table C.2 and Table C.1 for results.

Another method for finding the instrumental noise of the BPM is to take a

run without beam but at a gain comparable to when beam is present. This is just

a typical pedestal run in ’Forced Gain Mode’, then in the analyzer code a typical

beam signal is injected into the BPM ntuple channels. Results of this analysis can
 149
be found in Table C.3.

FIG. C.2: These are plots of the instrumental noise of the BPMs. After predicting the
position of the beam in a BPM from the above plots, the predicted position of the beam
can be subtracted from the BPM signal leaving behind only the instrumental noise of the
BPM.

It is not known why these results do not agree with one another. One reason

why these results might be inconsistent is that it takes two BPMs to predict the

location of the beam in a third BPM. This was not done in the above analysis since

there are magnetic elements between many of the BPMs in the injector.
 150

Electronic BPM noise found by predicting and subtracting position differences

BPM X Instr Noise (µm) Y Instr Noise(µm)
1i02 7.419 ± 0.077 9.793 ± 0.113
1i04 16.380 ± 0.172 11.360 ± 0.133
1i06 22.400 ± 0.226 6.363 ± 0.067
0L02 14.790 ± 0.232 16.799 ± 0.172
0L03 16.610 ± 0.162 30.150 ± 0.322
0I05 6.449 ± 0.066 13.200 ± 0.144
0I02A - -

TABLE C.1: Electronic noise calculated from position differences.

Electronic BPM noise found by predicting and subtracting positions

BPM X Instr Noise (µm) Y Instr Noise(µm)
1i02 1.769 ± 0.016 1.932 ± 0.029
1i04 3.738 ± 0.037 3.702 ± 0.040
1i06 3.890 ± 0.037 5.277 ± 0.060
0L02 14.830 ± 0.207 3.584 ± 0.034
0L03 17.850 ± 0.209 11.970 ± 0.116
0I05 3.747 ± 0.034 2.875 ± 0.030
0I02A - -

TABLE C.2: Electronic noise from calculating positions.

Electronic BPM noise found by injecting a fake BPM signal

BPM X Instr Noise (µm) Y Instr Noise(µm)
1i02 1.682 ± 0.120 30.735 ± 1.004
1i04 41.70 ± 3.172 4.756 ± 0.191
1i06 18.330 ± 2.183 7.813 ± 0.289
0L02 0.536 ± .0300 2.026 ± 0.057
0L03 9.008 ± 0.291 0.041 ± 0.002
0I05 2.398 ± 0.123 3.431± 0.090
0I02A 4.451 ± 0.144 5.121 ± 0.156
TABLE C.3: Electronic noise by taking a pedestal run and injecting a fake signal.
 151
C.1.2 PZT

From the 2001-2002 G0 commissioning run, the following PZT calibration slopes

were measured by different monitors in the injector.

From December 19, 2002:

BPM QPD 1i02 1i04 1i06 0L02

∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y
X-PZT -464 -2393 3642 -6253 -2016 7916 -1403 4684 -213 -1162
Y-PZT 4876 -217 11750 4454 -1350 -2800 -7211 -984 -5596 -1424
TABLE C.4: The responses of the injector BPM’s position differences to the X and Y
motion of the PZT mirror on December 19, 2002. These slopes are given in nm/V

Device QPD 1i02 1i04 1i06 0L02 BCM

∆Q ∆Q ∆Q ∆Q ∆Q ∆Q
X-PZT 6 -147 -185 -183 -197 -251
Y-PZT 335 -577 -606 -523 -773 -955
TABLE C.5: The responses of the injector BPM’s charge asymmetries to the X and Y
motion of the PZT mirror on December 19, 2002. These slopes are given in ppm/V.

BPM QPD 1i02 1i04 1i06 0L02

θ 34 98 88 79 64
TABLE C.6: These results are the angle (in degrees) between the responses of the PZT
motion in X and in Y on December 19, 2002 . This is a method of observing the orthog-
onality of the PZT motion.
 152
From January 14, 2003:

BPM QPD 1i02 1i04 1i06 0L02

∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y
X-PZT -1375 3204 -2595 4910 2036 -7083 824 -2913 1393 2534
Y-PZT -1531 -818 9338 4684 -11550 -321 -4585 -346 -7029 -308
TABLE C.7: The responses of the injector BPMs to the X and Y motion of the PZT
mirror on January 14, 2003. These slopes are given in nm/V.

Device QPD 1i02 1i04 1i06 0L02

∆Q ∆Q ∆Q ∆Q ∆Q
X-PZT 256 6 2 4 -16
Y-PZT 603 -319 -331 -313 -162
TABLE C.8: The responses of the injector BPM’s charge asymmetries to the X and Y
motion of the PZT mirror on January 14, 2003. These slopes are given in ppm/V.

BPM QPD 1i02 1i04 1i06 0L02

θ 95 91 104 102 115
TABLE C.9: These results are the angle (in degrees) between the responses of the PZT
motion in X and in Y on January 14, 2003. This is a method of observing the orthogo-
nality of the PZT motion.
 153
From January 24, 2003:

BPM QPD 1i02 1i04 1i06 0L02

∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y
X-PZT 18 2295 -1936 2436 3856 2138 735 -3553 931 919
Y-PZT -2557 -810 5648 523 11570 500 -6078 -1067 -2397 -339
TABLE C.10: The responses of the injector BPM’s position differences to the X and Y
motion of the PZT mirror on January 24, 2003. These slopes are given in nm/V.

Device QPD 1i02 1i04 1i06 0L02

∆Q ∆Q ∆Q ∆Q ∆Q
X-PZT 852 -71 -88 -77 2
Y-PZT 565 -238 -364 -314 -245
TABLE C.11: The responses of the injector BPMs to the X and Y motion of the PZT
mirror on January 24, 2003. These slopes are given in ppm/V.

BPM QPD 1i02 1i04 1i06 0L02

θ 108 123 149 92 143
TABLE C.12: These results are the angle (in degrees) between the responses of the PZT
motion in X and in Y on January 24, 2003 . This is a method of observing the orthog-
onality of the PZT motion.

The PZT was calibrated with only G0 beam in the injector on three days during

the commissioning run. It is difficult to make a comparison with such a small

sample. The PZT response as measured in Hall C showed erratic behavior that was

not understood at the time. G0 will investigate further the response of the PZT on

the bench between the end of the February 2003 and the beginning of the second

2003 engineering run.

 154

FIG. C.3: These are plots of the helicity correlated position differences in X and Y as a
function of PZT X and Y motion and as a function of beam monitor in the injector.
 155
Adiabatic Damping (Transverse Magnification)

If the energy of the electron beam increases much slower (adiabatically) than

the betatron oscillations, then the normalized emittance will remain constant while

the unnormalized emittance (that is the actual beam size) will shrink. This means

that the transverse size of the beam as measured in Hall C should be smaller than

what is measured in the injector. During the HAPPEX experiment there was so

much adiabatic damping that they did not need to run position feedback. The adia-

batic damping should be about a factor of 20 in suppression between the injector and

Hall C. This was demonstrated during the HAPPEX running. Besides benefiting

from the smaller position differences in the hall, adiabatic damping is an indicator

of the quality of the accelerator setup.

The first measurements of the adiabatic damping factor was performed on

March 18,2002. At this time, the adiabatic damping was defined as:

centroid ∆X at Hall A target BP M

centroid ∆X at injector 0L02 BP M
.

Measurements were made with both the Hall A diode laser and a homemade

Ti:Sapphire laser, the following was observed for each laser respectively:

1100±100nm
• 24568±1373
∼ 0.045 ± 0.005 →∼ 22 ± 2

1200±70nm
• 34380±1584
∼ 0.035 ± 0.003 →∼ 29 ± 3.

During the G0 commissioning run, three opportunities were available to mea-

sure the adiabatic damping between the injector and Hall C. The reason for so few

opportunities was due to the fact that other experimental halls were operational

meaning the other hall’s beam would be present in the injector making the mea-

surement difficult to do. From April 2002, BPM 0L02 was the reference BPM in the

injector due to the fact that this BPM had the largest response. This response has
 156
changed during the commissioning run and all the BPMs in the injector that were

read out by the G0 injector DAQ were examined.

The method for observing the adiabatic damping had further been improved for

the commissioning run. Instead of only looking at the position differences between

BPMs in the injector and in the hall the G0 PZT mirror located on the laser table

was utilized. Now the ratio of the responses of the PZT mirror in X and Y are used

to find the adiabatic damping:

√
2 2
Adiabatic Damping in X = √XX +Y X
XX 2 +Y X 2

√
2 2
Adiabatic Damping in Y = √XY +Y Y
XY 2 +Y Y 2

where:

• XX = ∆ X as one varies PZT X

• YX = ∆ Y as one varies PZT X

• XY = ∆ X as one varies PZT Y

• YY = ∆ Y as one varies PZT Y

BPM QPD 1i02 1i04 1i06 0L02

Varying PZTx(12/19/02) 1.0 3.5 3.5 2.1 0.5
Varying PZTy(12/19/02) 3.3 8.6 9.4 4.9 3.9
Varying PZTx(01/16/03) 3.4 5.3 7.1 2.9 2.8
Varying PZTy(01/16/03) 3.0 18.0 20.0 7.9 12.2
Varying PZTx(01/24/03) 2.9 3.9 5.6 4.6 1.7
Varying PZTy(01/24/03) 10.4 22.1 45.1 14.4 9.4
TABLE C.13: Results of the adiabatic damping as measured during G 0 commissioning.

In conclusion, the adiabatic damping as measured at BPM 0L02 falls short of

the factor of 20 that is expected. This might be due to the fact that 0L02 now sits at
 157

FIG. C.4: Adiabatic damping is defined to be the ratio of position differences in the
injector (quadrant photodiode, BPM 1i02, 1i04, 1i06, and 0L02) to the hall monitor
(BPM G0B). The adiabatic damping as measured on the three dates indicated. Data
taken on December 19, 2002 was at a beam current of 20 µA. Data taken on January
16, 2003 was at a beam current of 5 µA. The data taken on January 24, 2003 was at a
beam current of 40 µA. The top plot is the adiabatic damping as one varies the x PZT.
The bottom plot is the adiabatic damping as one varies the y PZT.

the waist of the beta function of the machine when these measurements were made.

After a better tune of the machine can be found for the G 0 beam, work should be

done to find which BPM is most sensitive in the injector and use that as a reference.
 158
C.1.3 Intensity Attenuator Cell

The following are results from the 2002-2003 G0 commissioning run:

BPM QPD 1i02 1i04 1i06 0L02

∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y ∆X ∆Y
12/17 -383 55 -459 407 624 320 332 123 -238 -5
12/16 563 1149 -304 359 13 448 -342 229 -185 -32
12/15 23 9 -448 585 -129 460 -699 491 -1321 -354
12/15 181 210 -5343 4515 -178 -159 -439 613 -1104 -89
12/15 6 24 -401 506 n/a 619 -501 354 -258 71
12/12 123 116 -398 585 -533 535 -1164 781 -1894 -36
12/05 727 721 -1063 1260 -320 733 -842 756 157 131
TABLE C.14: The IA slopes for position differences in nm/V measured in December
2002.

Device QPD 1i02 1i04 1i06 0L02 BCM 0L03

∆Q ∆Q ∆Q ∆Q ∆Q ∆Q ∆Q
12/17 553 497 504 403 348 436 416
12/16 531 472 370 327 267 352 310
12/15 549 496 492 424 221 238 267
12/15 551 505 511 450 375 481 463
12/15 550 305 485 218 272 272 270
12/12 542 502 497 424 84 94 95
12/05 543 467 520 438 237 288 301

TABLE C.15: The IA slopes for charge asymmetry in ppm/V.

The IA responded well during the commissioning offering a large (∼ 400 ppm/V)

charge asymmetry calibration though it did generate large position differences. The

IA calibration constants were very stable over time requiring new measurements

of this constant every couple of days. Work to minimize the position differences

induced by the IA will begin after the 2002-2003 commissioning run.

 159

FIG. C.5: Charge asymmetry calibration constants (ppm/V) for the IA. Note that the
values at the QPD are consistent with one another over both months. Note the values
from 1I02 to 1I06 are consistent with one another. Between 1I06 and 0L02 are a variety
of apertures such as A1, A2, the Chopper, etc. It is due to scraping on these apertures
that
 BIBLIOGRAPHY

[1] J. Gasser, H. Leutwyler and M.E. Sanio, Phys. Lett. B253, 252 (1991).

[2] J. Gasser, H. Leutwyler, Phys. Rep 87, 77 (1982).

[3] J. Gasser, Ann. Phys. 136, 62 (1981).

[4] K. Abe et al., Phys. Lett. B405, 180 (1997).

[5] G. Altarelli et al., Nucl. Phys. B496, 337 (1997).

[6] G.P. Zeller et al., Phys. Rev. Lett. 88, 091802 (2002).

[7] S. Okubo, Phys. Lett. B5, 165 (1963); G. Zweig, CERN Report 8419/TH412
(1964); I. Iizuka, Prog. Theor. Phys. Suppl. 21, 37 (1966).

[8] A. Bertin et al., Phys. Lett. B388, 450 (1996).

[9] M.P. Rekalo et al., Z. Phys. A357, 133 (1997).

[10] J. Ellis and R. Jaffe, Phys. Rev. D9, 1444 (1974).

[11] J. Ashman et al., Phys. Lett. B206, 364 (1988).

[12] T. Adams et al.,Strange Content of the Nucleon (NuTeV). In Physics with

a High Luminosity Polarized Electron Ion Collideredited by L.C. Bland,
A.P.Szczepaniak, and J.T. Londergan, p. 337 (World Scientific, 2000).

[13] D.T. Spayde et al., Phys. Lett. B583, 79 (2004).

[14] K.A. Aniol et al., Phys.Rev. C69, 065501 (2004).

[15] E.J. Beise et al., Nucl. Instrum. Meth. A378, 383 (1996).

[16] M.N. Rosenbluth, Phys. Rev. 79, 615 (1950).

160
 161
[17] N.F. Mott, Proc. Roy. Soc. A124, 425 (1929).

[18] P.Bosted et al., Phys. Rev. Let. 68, 3841 (1992).

[19] A.F. Sill et al., Phys. Rev. D48, 29 (1993).

[20] J. Arrington, Phys. Rev. C68, 034325 (2003).

[21] E.J. Brash et al., Phys. Rev. C65, 051001 (2002).

[22] A.I. Akhiezer and M.P. Rekalo, Sov. J. Part. Nucl. 3, 277 (1974); R. Arnold,
C. Carlson, and F. Gross, Phys. Rev. C23, 363 (1981).

[23] J. Litt et al., Phys. Lett. B31, 40 (1970).

[24] Ch. Berger et al., Phys. Lett. B35, 87 (1971).

[25] L.E. Price et al., Phys. Rev. D4, 45 (1971).

[26] W. Bartel et al., Nucl. Phys. B58, 429 (1973).

[27] R.C. Walker et al., Phys. Rev. 49, 5671 (1994).

[28] L. Andivahis et al., Phys. Rev. D50, 5491 (1994).

[29] B. Milbrath et al., Phys. Rev. Lett. 80, 452 (1998); erratum Phys. Rev. Lett.
82, 221 (1999).

[30] M.K. Jones et al., Phys. Rev. Lett. 84, 1398 (2000).

[31] O. Gayou et al., Phys. Rev. Lett. 88, 092301 (2002).

[32] P.G. Blunden, W. Melnitchouk, J.A. Tjon, Phys. Rev. Lett. 91, (2003).

[33] P.A.M. Guichon, M. Vanderhaeghen, Phys. Rev. Lett. 91(2003).

[34] A. Lung et al., Nucl. Rev. Lett. 70, 1719 (1993).

[35] S. Galster et al., Nucl. Phys. B32, (1971).

[36] T. Eden et al., Phys. Rev. C 50, 1749 (1994).

[37] C. Herberg et al., Eur. Phys. J A5, 131 (1999).

 162
[38] M. Ostrick et al., Phys. Lett. 83, 276 (1999).

[39] D. Rohe et al., Phys. Rev. Lett. 83, 4257 (1999).

[40] J. Becker et al., Eur. Phys. J A6, 329 (1999).

[41] M.Meyerhoff et al., Phys. Lett. B327, 201 (1994).

[42] F. Halzen and A.D. Martin, Quarks and Leptons(John Wiley & Sons, New
York, 1984).

[43] Particle Data Group, Euro. Phys. J. C54, 1 (2000).

[44] W.M. Albenico et al., Phys. Rep. 3 58, 182 (1995).

[45] G.P. Ramsey et al., Int. J. Mod. Phys. A18 , 1211 (2003)

[46] G.A. Miller, Phys. Rev. C57, 1492 (1998).

[47] E. Hadjimichael, G.I. Poulis, and T.W. Donnelly, Phys. Rev. C45, 2666 (1992).

[48] L.Diaconescu, R. Schiavilla, and U. van Kolck, Phys. Rev. C63, 4007 (2001).

[49] S.-L. Zhu, S. J. Puglia, B. R. Holstein, and M. J. Ramsey-Musolf, Phys. Rev.

D62, 033008 (2000).

[50] H.-W. Hammer, S.J. Puglia, M.J. Ramsey-Musolf, and S.-L. Zhu, Phys. Lett.
B562, 208 (2003).

[51] F.E. Maas et al., Phys. Rev. Lett. 93, 022002 (2004).

[52] L.Lee, Report on North-American PMT-Base design,

http://burrhus.triumf.ca/g0/g0index.html/g0dsr6-pmtbase.pdf.

[53] R. Flood, private communication.

[54] G. Niculescu, Resonant Cavities used as Beam Current Monitors, CEBAF In-
ternal Report (unpublished).

[55] C. Armstrong, Beam Current Measured in Hall C, CEBAF Internal Report

(unpublished).

[56] K.B. Unser, The Parameteric Current Transfer, a Beam Current Monitor De-
veloped for LEP, CERN SL/91-42 (unpublished).
 163
[57] P. Gueye, Status of the Actual Beam Position Monitors in Hall C Beamline,
CEBAF Internal Report CEBAF-TN-93-004 (unpublished).

[58] G. Krafft and A. Hofler, How the LINAC Beam Position Monitors “Work”,
JLab Technical Note CEBAF-TN-93-004 (unpublished).

[59] M. Hauger et al., Nucl. Instrum. Meth. A 462, 382 (2001).

[60] CODA Users Manual (1995).

[61] S. Lewis, Overview of the Experimental Physics and Industrial Control System:
EPICS, LBNL Internal Report (2000).

[62] L. Hannelius, G0-Geant Simulation of the Inelastic Background from the Target
Windows, Internal report G0-03-044 (unpublished).

[63] L.W. Mo and S. Tsai, Rev. Mod. Phys. 41, 205 (1969).

[64] L.C. Maximon, Rev. Mod. Phys. 41, 193 (1969).

[65] G. Batigne, et al., Q2 Determination in the First Phase of the G0 Experiment,

Internal Report G0-03-001.

[66] D.H. Beck and R.D. McKeown, Ann. Rev. Nucl. Part. Sci. 51, 189 (2001).

[67] D.H. Beck and B.R. Holstein, Int. J. Mod. Phys. E10, 1 (2001).

[68] M.J. Musolf et al., Physics Reports 239 (1994).

[69] W. Koepf et al., Phys. Lett. B288, 11 (1992).

[70] M. Burkhardt et al., Z. Phys. C61, 433 (1994).

[71] R.L. Jaffe, Phys. Lett. B229, 275 (1989).

[72] G. Hohler et al., Nucl. Phys. B114, 505 (1976).

[73] H.-W. Hammer et al., Phys. Lett. B367, 323 (1996).

[74] U.-G. Meissner et al., Phys. Lett. B408, 381 (1997).

[75] H. Forkel, Phys. Rev. C56, 510 (1997).

 164
[76] P. Geiger and N. Isgur, Phys. Rev. D55, 299 (1997).

[77] T.D. Cohen et al., Phys. Lett. B316, 1 (1993).

[78] N.W. Park et al., Phys. Rev. D43, 869 (1991).

[79] S.J. Dong et al., Phys. Rev. D58, (1998).

[80] T.R. Hemmert et al., Phys. Rev. C60, (1990).

[81] A. Abada et al., Phys. Lett. B353, 20 (1995).

[82] K. Goeke et al., Phys. Rev. D65, (2001).

[83] Th. Gutsche et al., Phys. Rev. C66, (2002).

[84] Jefferson Lab experiment 99-115, K. Kumar and D. Lhuiller, spokespersons.

[85] Jefferson Lab experiment 00-114, D. Armstrong and R. Michaels, spokesper-

sons.

[86] K. McFarlane, Required number of photoelectrons, email to the g0-det mailing

list on Oct, 6th 1998.

[87] J. Roche et al., Status Report of the G0 Detector 8 Prototype Studies, Internal
report G0-99-032.

[88] H. Breuer, Photo-electrons for all detectors, Tosca 4.5 shapes, V7, Internal
report G0-99-012.

[89] Isotope Products Laboratories, 1800 N. Keystone Street, Burbank, CA 91504,

(818)843-7000, fax (818)953-9776.

[90] R. Woo and D. Armstrong, Spreadsheet of PMT Test Results, Internal report
G0-00-31.
 VITA

Jeffery Allen Secrest

Jeffery Allen Secrest was born in Cincinnati, Ohio on November 22nd 1973 to

Mike and Joy Secrest. He received a Bachelor of Science degree in physics at the

University of Cincinnati in 1997. In the fall of 1997 he begun his graduate career

at the University of Mississippi where he earned a Master of Arts in physics. In

1998 he won the Outstanding Teaching Assistant award. In the fall of 2000 he

entered the College of William and Mary as doctoral candidate in the Department

of Physics. Along the way, he earned a Masters of Science in 2001. This dissertation

was defended on December 17th 2004 at the College of William and Mary.

165