You are on page 1of 317

Laboratory Manual for Nuclear Physics

s
sic
hy
rP
Shobhit Maha jan
lea
shobhit.mahajan@gmail.com
uc
lN
ua
an
bM
La

Version 2.1
Last modi ed on January 1, 2018
Contents
Note to the Reader 4

s
sic
4
1 STATISTICS & ERROR ANALYSIS 7

hy
1.1 Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.1 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Uncertainty in Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

rP
1.2.1 Uncertainity, Accuracy & Precision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.2 Systematic & Random Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.3 Signi cant Digits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2.4 Reporting of Uncertainties & Rounding O
1.3 Statistical Analysis of Data . . . . . . . . . . . . .
lea .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
14
16
1.3.1 Histograms & Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
uc
1.3.2 Parent & Sample Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.3.3 Mean & Deviation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
lN

1.4 Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.4.1 Binomial Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.4.2 Poisson Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.4.3 Normal Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
ua

1.5 Error Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46


1.5.1 Propagation of Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
an

1.6 Estimation and Error of the Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60


1.6.1 Method of Maximum Likelihood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
1.6.2 Estimated Error in the Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
bM

1.7 Method of Least Squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64


1.8 Goodness of Fit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
1.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
La

2 RADIOACTIVITY 75
2.1 Radioactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.1.1 Measure of radioactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.1.2 Activity Law & Half Life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
2.2 Nuclear Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2.2.1 Alpha Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.2.2 Beta Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
2.2.3 Gamma Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
2.4 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

1
LABORATORY MANUAL FOR NUCLEAR PHYSICS 2

3 INTERACTION WITH MATTER 91


3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.1.1 Cross Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.2 Interaction of Charged Particles with Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.2.1 Interaction of Heavy charged particle with matter . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.2.2 Interaction with matter of electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.2.3 Interaction of gamma rays with matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.4 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4 G-M COUNTER 123
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.2 Detector Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

s
4.3 Ionisation of Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

sic
4.3.1 Townsend Avalanche . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.3.2 Kinds of Detectors & Detector Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
4.4 GM Counter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

hy
4.4.1 Geiger Discharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.4.2 Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

rP
4.4.3 Dead Time & Recovery Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.4.4 Geiger Counting Plateau & Operating Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.4.5 Counting Eciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.5 References . . . . . . . . . . . . . . . . . . . . . . . . . lea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.6 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
5 Experiment: GM Characteristics 144
uc
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.2 Precautions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
lN

5.2.1 Health E ects of Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145


5.3 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.3.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
ua

5.3.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147


5.3.3 Sample Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.4 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
an

6 Experiment: GM Counter: Counting Eciency for & rays 165


6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
bM

6.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166


6.2.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.2.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.2.3 Sample Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
La

6.2.4 Error Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173


6.3 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
7 Experiment: Absorption of rays in Iron 182
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
7.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.2.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.2.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.2.3 Sample Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
7.3 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
LABORATORY MANUAL FOR NUCLEAR PHYSICS 3

8 Experiment: Veri cation of the Inverse Square Law for rays 188
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
8.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
8.2.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
8.2.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
8.2.3 Sample Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
8.3 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
9 Experiment: To Determine the Range of rays in Aluminum and to determine the End Point
Energy 197
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
9.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
9.2.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198

s
9.2.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

sic
9.2.3 Sample Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
9.3 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

hy
10 Experiment: Gamma Ray spectrum using a Scintillation Counter 207
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207

rP
10.2 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
10.2.1 Inorganic Scintillators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
10.2.2 Organic Scintillators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
10.2.3 Photomultiplier Tube . . . . . . . . . . . .
lea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
10.2.4 Gamma Ray Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
10.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
10.4 Sample Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
uc
10.5 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
lN

A p-Value Tables 248


B Using MS Excel 252
ua

C Using Gnuplot 266


C.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
an

C.2 Plotting with inbuilt functions of GNUPLOT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267


C.2.1 Interactive plotting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
C.3 Saving Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
bM

C.3.1 Customization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269


C.4 Plotting using data from a le . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
C.5 Plotting using data from le and tting to a smooth curve . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
C.5.1 Curve Fitting & Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
La

D Radioactive Decay Equilibrium 279


D.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
D.2 Bateman Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
D.3 Di erent Kinds of Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
D.4 Numerical Integration of Bateman Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
E Theory of Alpha Decay 289
E.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
E.2 1-d Tunnel E ect For Rectangular Barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
E.3 Tunnel E ect with Nuclear Potential Barrier: Geiger-Nuttall Law . . . . . . . . . . . . . . . . . . . . . . . 293
LABORATORY MANUAL FOR NUCLEAR PHYSICS 4

F Fermi's Theory of Beta Decay 297


F.1 Fermi's Golden Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
F.2 Fermi's Theory of Beta Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
F.2.1 Matrix Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
F.2.2 Density of Final States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
F.2.3 Decay Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
G Semi-Classical Theory of Gamma Decay 309
G.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
G.2 Density of Final States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
G.3 Interaction Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
G.3.1 Dipole Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
G.4 Transition Rate & Lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

s
sic
Index 314

hy
rP
lea
uc
lN
ua
an
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 5

A Note to the Reader

This Manual is intended for use in the Nuclear Physics Laboratory of the M.Sc (Previous) class.

The Manual is organised into 10 Chapters. The rst 4 Chapters provide the theoretical background
for the experiments which are performed in the laboratory. Chapter 1 is a fairly detailed introduction
to the Statistical Tools which are required to analyse the experiments. It includes topics like distri-

s
butions, Error Analysis and Goodness of Fit etc. Most of these topics are familiar to you from your

sic
undergraduate days. However, they are presented in enough detail here and they do not assume any
prior knowledge of statistics.

hy
rP
Chapter 2 and 3 are a review of nuclear physics concepts which are required for the experiments.
These include radioactive radiation (alpha, beta and gamma rays) as well as their interaction with mat-
ter. This material, once again, should be familiar to you and it is presented here for review. However,
lea
again, the material is complete and does not assume any prior knowledge. Most of the experiments
carried out in the laboratory use a Geiger Muller counter. This is discussed in detail in Chapter 4.
uc
Although the rst 4 chapters provide you with enough information to be able to do the experiments,
lN

they are Not meant to be a substitute for the books which discuss each of these topics in detail. There
are many excellent books available on these topics and you are encouraged to go through them.
ua

Chapters 5 10 are detailed discussions of the experiments which are available in the laboratory.
an

For each experiment, the procedure is discussed and sample data is given. This sample data is then
analysed and the errors and results are obtained. It is important for you to understand well the cal-
bM

culations given here so that you can do the same with the data that you obtain in the laboratory yourself.

Each Chapter has some questions in the end which you are encouraged to attempt to answer to test
La

your understanding of the theory and the experiments.

There are 7 Appendices in the Manual. There is an appendix which discusses the use of Microsoft
Excel program to do data analysis and plot graphs etc. There is also an appendix which discusses the
use of GNUPLOT to plot graphs in case you want to use a Linux platform. In addition, the detailed
theories for radioactive equilibrium and alpha, beta and gamma decay are also given in the Appendices.

We would very much like to get your suggestions regarding how to improve this Manual. In addition,
if there are any errors or misprints that are spotted in the Manual, we would like to hear from you.
Please send a mail with the suggestions/errors etc. to shobhit.mahajan@gmail.com making sure
LABORATORY MANUAL FOR NUCLEAR PHYSICS 6

you quote the version number of the Manual as well as the Modi cation date of the Manual you are
using. The version number and date are on the title page of the Manual.

Finally, it is important to realise that this manual can only help you with your experiments. Ulti-
mately, it is essential that you perform the experiments yourself and then do the data analysis with the
help of the tools discussed in the Manual. Unless you do the actual experiment and the calculations
yourself, you will never learn the subject.

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La
Chapter 1

STATISTICS & ERROR ANALYSIS

s
sic
hy
1.1 Probability

rP
Since we are going to be dealing with probability in the subsequent sections, let us give a brief back-
ground. The probability of an event refers to the likelihood that the event will occur. Mathematically,
the probability that an event will occur is expressed as a number between 0 and 1. The sum of prob-
lea
abilities in any statistical experiment is always 1, a statement of the fact that something will certainly
happen. Let us illustrate how one can calculate probabilities.
uc
lN

Consider rst the case of an experiment with n possible outcomes which are each equally likely. Now
if we take a subset r of these and call them successes then clearly the probability of success in the
experiment is nr . Thus, if there are 10 balls, 7 white and 3 black in a bag, then the probability of getting
ua

a black ball if one is picked out at random from the bag is 103 .
an

There is another approach to probability where one talks about relative frequencies. Suppose I count
bM

the number of cars passing a particular point on a road at a particular interval of time and notice how
many of them are white. Suppose on the rst day, I see 5 white cars out of a total of 20 cars, while
on the second day I count 9 white cars out of 30 while on the third day I nd 3 white cars out of 5
La

and so on. Clearly, the relative frequency of white cars is di erent on di erent days. However, one
could nd for instance that if I repeat this experiment many many times, then the relative frequency is
0:26. Then the Law of Large Numbers says that the relative frequency of an event will converge on
the probability of that event as the number of trials increases. This is what we anyway feel using our
common sense. If we have, for instance, a fair coin, then we know that the probability of getting a head
or a tail is 12 . However, when we start tossing the coin, it might happen that we get a series of heads
or tails which might lead one to believe that the probabilities of these events (that is getting a head or
a tail) are not equal. But we do know intuitively that if we perform a large number of coin tosses, then
the number of heads (or tails) is going to be 12 of the total number of tosses. This is precisely what the
Law of Large Numbers says- the average of the results of a large number of trials will be close

7
LABORATORY MANUAL FOR NUCLEAR PHYSICS 8

to the expected value and the two will converge as the number of trials increases.

Some de nitions in probability theory are useful:

1. Two events are mutually exclusive or disjoint if they cannot occur at the same time.
2. The probability that Event A occurs, given that Event B has occurred, is called a conditional
probability. The conditional probability of Event A, given Event B, is denoted by the symbol
P (AjB ).

s
sic
3. The complement of an event is the event not occurring. The probability that Event A will not
occur is denoted by P (A0 ).

hy
4. The probability that Events A and B both occur is the probability of the intersection of A and B.
T

rP
The probability of the intersection of Events A and B is denoted by P (A B ). If Events A and B
T
are mutually exclusive, P (A B ) = 0.
lea
5. The probability that Events A or B occur is the probability of the union of A and B. The probability
S
of the union of Events A and B is denoted by P (A B ) .
uc
6. If the occurrence of Event A changes the probability of Event B, then Events A and B are depen-
dent. On the other hand, if the occurrence of Event A does not change the probability of Event
lN

B, then Events A and B are independent.


ua

These de nitions allow us to write down the rules for probability.


an

Subtraction: The probability that event A will occur is equal to 1 minus the probability that event
A will not occur.
bM

P (A) = 1 P (A0 )
La

Multiplication: The probability that Events A and B both occur is equal to the probability that
Event A occurs times the probability that Event B occurs, given that A has occurred.

\
P (A B ) = P (A)P (B jA)
Addition: The probability that Event A or Event B occurs is equal to the probability that Event
A occurs plus the probability that Event B occurs minus the probability that both Events A and B occur.

[ \
P (A B ) = P (A) + P (B ) P (A B )) = P (A) + P (B ) P (A)P (B jA)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 9

These rules are fairly obvious intuitively and the easiest way to prove them is to use Venn diagrams
where the results quoted above are immediately clear.

1.1.1 Random Variables


When the value of a variable is determined by a chance event, that variable is called a random vari-
able. Random variables can be discrete or continuous.

Discrete Random Variable

s
sic
Within a range of numbers, discrete variables can take on only certain values. Suppose, for example,

hy
that we ip a coin and count the number of heads. The number of heads will be a value between 0
and +1. Within that range, though, the number of heads can be only certain values. For example,

rP
the number of heads can only be a whole number, not a fraction. Therefore, the number of heads is a
discrete variable. And because the number of heads results from a random process - ipping a coin - it
is a discrete random variable. lea
uc
Continuous Random Variable
lN

Continuous variables, in contrast, can take on any value within a range of values. For example,
suppose we randomly select an individual from a population. Then, we measure the age of that person.
ua

In theory, his/her age can take on any value between 0 and +1, so age is a continuous variable. In
this example, the age of the person selected is determined by a chance event; so, in this example, age
an

is a continuous random variable.


Note that discrete variables can be nite or in nite. Thus, for instance the number of heads in coin
bM

ips can be in nite while the number of aces that I can choose from a deck of cards is nite (0; 1; 2; 3; 4).
Continuous variables can always take an in nite number of values while some discrete variables can take
in nite number of values.
La

1.2 Uncertainty in Measurement


1.2.1 Uncertainity, Accuracy & Precision
All measurements that we do have some degree of uncertainty. This uncertainty might come from a
variety of sources about which we will talk later. But the fact that needs to be emphasised is that all
measurements have some uncertainty and an analysis of this uncertainty is what we call error analysis.
Any measured value that we quote must be accompanied by our estimate of the level of certainty or con-
dence associated with that measurement. This fact is absolutely essential since without this the basic
LABORATORY MANUAL FOR NUCLEAR PHYSICS 10

question of science, namely \does the result of our experiment agree with the theory" can not be an-
swered. To decide whether the proposed theory is valid or not, this question would need to be answered.

When we carry out an experiment to measure a quantity, we of course assume that some `true' or
exact value exists. We of course may or may not know this value but we always attempt to nd the
best value possible given the limitations of our own experimental setup. Typically, every time we carry
out the experiment, we will nd a di erent value and so the question is how do we report our best
estimate of this `true' value? Usually, this is done as

s
measurement = best estimate  uncertainty

sic
For example, let us assume you want to nd the weight of your mobile phone. By simply putting

hy
in your hand, you can estimate it to be between 100 and 200 grams. But that is not good enough. So
you go to a balance in the laboratory and it gives you a reading of 145:55 grams. This value is much

rP
more precise than the original estimate you obtained, but how does one know that it is accurate?
One way is to repeat your measurement several times and suppose you get the values 145:59; 145:53
lea
and 145:51 grams. Then one could say that the weight of the phone is 145:55  :04 grams. But
now suppose you go to another balance and nd a value of 144:15 grams? Now one is faced with a
uc
problem since your original best estimate is very di erent from this measurement. So what does one do?
lN

To understand this, we need to understand rst the di erence between precision and accuracy .

R
ua

Precision & Accuracy


an

 Accuracy is how close the measured value is to the true or the accepted value of
a quantity.
bM

 Precision is a measure of how well the result can be measured, irrespective of the
La

true value. It re ects the degree of consistency and agreement between repeated,
independent measurements of the same quantity as well as the reproducibility of
the results.

 Any statement of uncertainty associated with a measurement must include factors


which a ect both accuracy and precision. After all, it is a waste of time if we
determine a result which is very precise but highly inaccurate or its converse, that
is, a result which is very accurate but highly imprecise.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 11

In our example above, we have no way of knowing whether our result is accurate or not unless we
compare it with a known standard. For instance, we could use a standard weight to determine if the
balances used in our measurement are accurate or not. Suppose one performs two experiments and get
two values of g, the acceleration due to gravity. One experiment gives g = 9:7 m s 2 while the other
other gives g = 9:6345 m s 2 . Clearly the rst result is more accurate while the second one is more
precise.

Precision is often reported experimentally as relative uncertainty de ned as

s

uncertainty

sic
Precision: Relative Uncertainty =
measured value (1.1)

hy
Thus in our example of the weight of the mobile phone, the relative uncertainty is

rP
0:04
Relative Uncertainty = = 0:00027 = :027%
145:55
lea
Accuracy on the other hand is reported usually as relative error which is de ned as

R
uc
measured value - expected value
Accuracy: Relative Error = (1.2)
expected value
lN

In our example above, if we think that the expected value of the weight of the mobile phone is 145:50
grams, then the relative error is
ua

145:55 145:50 0:05


an

Relative Error = = = 0:034%


145:50 145:50
Thus we see that any measurement needs to be both precise and accurate for it to be good. The
bM

idea of making good measurements is directly related to errors in measurement. Errors in measurement
can be broadly classi ed into two categories- random and systematic.
La

1.2.2 Systematic & Random Errors


Systematic Errors are errors which will make the results obtained by us di er from the \true" value
of the quantity under consideration. They are reproducible inaccuracies which are dicult to detect and
also cannot be analysed statistically. An important thing to realise is that systematic errors cannot
be reduced by repeated measurements.

Random Errors are errors which are uctuations in the observations which are statistical in nature.
They can be analysed statistically and furthermore, they can be reduced by repeated measurements
LABORATORY MANUAL FOR NUCLEAR PHYSICS 12

and taking averages as we shall see later.

To illustrate this distinction think of the following experiment. Suppose I wish to nd the time
period of a pendulum by timing some number of oscillations with the help of a stop watch. There could
be several sources of error- One source of error will be my reaction time, that is the time between my
seeing the pendulum bob reaching the extreme position and my starting the watch and again at a later
point my observing the bob and my stopping the watch. Obviously, if my reaction time was always
exactly the same, then the time delay wouldn't matter since they would cancel. However, we know
that in practice, my reaction time will be di erent. I may delay more in starting, and so underestimate

s
the time of a revolution; or I may delay more in stopping, and so overestimate the time. Since either

sic
possibility is equally likely, the sign of the e ect is random. If I repeat the measurement several times,
I will sometimes overestimate and sometimes underestimate. Thus, my variable reaction time will show

hy
up as a variation of the answers found. By analyzing the spread in results statistically, I can get a very

rP
reliable estimate of this kind of error.

Now suppose that the watch that I use is slow or fast. Then, no matter how many times I re-
lea
peat the experiment (of course with the same watch), I can never know the amount of such an error.
Further, the error's sign will always be the same- either the watch will be fast or slow leading to ei-
uc
ther an overestimate or underestimate of the rate of revolution. This is an example of a systematic error.
lN

In general, there is no set prescription for eliminating systematic errors and mostly one has to use
common sense to know if there are any systematic errors and to eliminate them. Random errors on the
ua

other hand are usually easier to study and eliminate or reduce. But one should remember that in many
an

situations, the accuracy of a measurement is dominated by possible systematic errors in the instrument
rather than the precision with which we can make the measurement.
bM

To summarise

R Systematic & Random Errors


La

Systematic Errors are reproducible inaccuracies that are dicult to detect and cannot
be analyzed statistically. If a systematic error is identi ed when calibrating against a
standard, applying a correction or correction factor to compensate for the e ect can
reduce the bias.

Random Errors are statistical uctuations in the measured data due to the precision
limitations of the measurement device. Random errors can be evaluated through
statistical analysis and can be reduced by averaging over a large number of observations
LABORATORY MANUAL FOR NUCLEAR PHYSICS 13

The fundamental aim of an experimentalist is to reduce as many sources of error as s/he can, and
then to keep track of those errors that cant be eliminated.

1.2.3 Signi cant Digits


Whenever one writes the results of an experiment, the precision in the experiment is normally indicated
by the number of digits which one reports the result with. The number of signi cant gures depends

s
sic
on how precise the given data is. The following rules are helpful:

hy
Signi cant Figures

rP
1. The leftmost NONZERO digit is ALWAYS the MOST signi cant digit.
2. When there is NO decimal point, the rightmost NONZERO digit is the least
lea
signi cant digit.
uc
3. In case of a decimal point, the rightmost digit is the least signi cant digit EVEN
IF IT IS A ZERO.
lN

4. The number of digits between the most and least signi cant digits are the number
of signi cant digits.
ua
an

Thus for instance, 22:00; 2234; 22340000; 2200: all have four signi cant digits while 2200 has only
bM

two signi cant digits because of the absence of the decimal point. When one is adding, subtracting,
multiplying or dividing numbers, then the result should be quoted with the least number of signi cant
gures in any one of the quantities being used in the operation of adding, multiplying etc. In your in-
La

termediate calculations, always keep ONE MORE signi cant digit than is needed in the nal answer.
Also when quoting an experimental result, the number of signi cant gures should be one more than is
suggested by the experimental precision.

In reporting measurements of experiments, one has to exercise special care. Thus for instance, if
one sees a result as 8120, we are not sure whether the result actually has 3 or 4 signi cant digits. By
the rules above, the 0 in the end is just a place value. However, it might actually transpire that the
experiment actually did measure the last digit to be 0. To get around this problem, one always uses
the scienti c notation. Thus if the expreimental result is only precise to three signi cant gures, one
would report it as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 14

8:12  103
or if it was precise to 4 signi cant digits, we would report it as

8:120  103

R Two things that need to be always avoided are

s
 Writing more digits in an answer (intermediate or nal) than justi ed by the

sic
number of digits in the data.

hy
 Rounding-o , say, to two digits in an intermediate answer, and then writing three

rP
digits in the nal answer.
lea
While dropping o some gures from a number, the last digit that one keeps should be rounded
uc
o for better accuracy. This is usually done by truncating the number as desired and then treating
lN

the extra digits (which are to be dropped) as decimal fractions. Then, if the fraction is greater than
1 1
2 , increment the truncated least signi cant gure by one. If the fraction is smaller than 2 , then do
nothing. If the fraction is exactly 12 , then increment the least signi cant digit only if it is odd.
ua
an

What about reporting results which arise from addition or multiplication of two measurements with
di erent signi cant gures. For instance, consider a measurement of two masses of 9:9 gm and 0:3163
bM

gm. How does one report the total mass? To illustrate this, let us write the numbers as 9:9     gm
and 0:3163 gm. Then if we add them up we see that result we get is 10:2     gm that is to say that
we are only sure of the rst decimal place number and so we should report this as 10:2 gm. Similarly,
La

if we had two numbers which we need to multiply, say 3:413 and 2:3, the product will be 7:8    
which will be reported as 7:8. When multiplying two quantities with di erent signi cant
digits, the product has the same number of signi cant digits as that of the number with
the least number of signi cant digits.

1.2.4 Reporting of Uncertainties & Rounding O


We are now in a position to give some basic rules and conventions of how experimental results should
be reported. We have already seen that every measurement will have an uncertainty or error associated
with it. This error could be random or systematic. This uncertainty indicates to us that precision or
LABORATORY MANUAL FOR NUCLEAR PHYSICS 15

reliability of the measurement. However, it is obvious that the magnitude of uncertainty by itself does
not give us the complete picture. An uncertainty of :1 seconds for instance, would be a very precise
measurement if the quantity being measured is 100 seconds say but not very precise if we were mea-
suring 1 second. Thus a more useful concept was that of fractional uncertainty or relative uncertainty
that we saw in Eq(1.1). The percentage uncertainty then is a better indicator of the precision involved
in the particular experiment.

A more serious question is that of reporting uncertainties. Although there is no de nite theoretical
reason to expect that uncertainties should be reported to a certain number of signi cant gures, the

s
general consensus is that uncertainties in measurements should be reported to 1 signi cant

sic
gure. Note that this is NOT something that one can prove or disprove. It is more of a convention
which is followed. Of course, if the leading signi cant gure in the uncertainty that we are reporting

hy
has the value 1 or 2 then by keeping only one signi cant digit, we might be making a mistake. In these

rP
cases, it is best to report it to two signi cant digits.

Once we know how the uncertainty in the measured quantity is to be reported, then this also tells us
lea
how we need to report our answers. For instance, if the uncertainty is 0:4 meters in the measurement
of some length L , then it would be meaningless to report the result as
uc
L = 100:4135  0:4 m
lN

Clearly, the number of digits we retain in the result is governed by the uncertainty in the measure-
ua

ment. In this example,we should report it as


an

L = 100:4  0:4 m
bM

Next suppose the quantity we are measuring has a value x = 12:0349 cm with an uncertainty of
x = 0:153 cm. How does one report this result? The most signi cant digit in the uncertainty is 1 and
hence we should keep two signi cant digits. Thus x = 0:15 cm and the result should be
La

x = (12:03  0:15) cm
Next suppose we have a measurement as z = 1:43  106 seconds with an uncertainty of z = 2  104
seconds. The correct way to report this is

z = (1:43  0:02)  106 s


Similarly, the mass of the electron is measured as

m = 9:11  10 31 kg
LABORATORY MANUAL FOR NUCLEAR PHYSICS 16

with

m = 2:2345  10 31 kg
By our rule above, we see that we should report the uncertainty as 2  10 31 . This means that the
measured quantity should also retain only one signi cant digit, that is 9. Thus we should report this as

m = 9  2  10 31 kg
R Please keep this in mind when you report your results. Typically, since your cal-

s
culators give you many more digits for any calculation, please use these rules to determine

sic
how you should report the results.

hy
rP
1.3 Statistical Analysis of Data
1.3.1 Histograms & Distribution lea
The fundamental problem in reporting the results of an experiment is to estimate the uncertainty in a
uc
measurement. It is reasonable to think that the reliability of an estimate of uncertainty in a measure-
ment can be improved if the measurement is repeated many times. The rst problem in reporting the
lN

results of many repeated measurements is to nd a concise way to record and display the values obtained.
ua

Suppose we measured the weights of all the new one rupee coins minted since they were introduced.
It is clear that all the coins will not have the same weight. The actual weight would depend on several
an

things including when it was minted and how much it has been in use etc. One way to display the
results of our measurement would be a histogram as shown in the Figure 1.1. This is for a sample of
bM

100 coins and we have divided the weights into bins of width  = 0:01 gm.
La

Figure 1.1: Binned histogram

We have plotted the data in such a way that the fraction of measurements that fall in each bin is
LABORATORY MANUAL FOR NUCLEAR PHYSICS 17

given by the area of the rectangle above the bin. That is to say that the height P (k) of the kth bin is
such that

Area = P (k)   = fraction of measurements in the kth bin


Thus, for instance, the area of the rectangle between 2:50 2:51 is 20  0:01 = :2. This means that
20% of the coins fall in this weight range.

We can see that such a plot gives us a good way to represent the data namely how the weights of the
coins in our sample are distributed. In most experiments, as the number of measurements increases, the

s
sic
histogram begins to take on a de nite simple shape, and as the number of measurements approaches
in nity, their distribution approaches some de nite, continuous curve, the so-called limiting distri-

hy
bution as in Fig(1.2).

rP
lea
uc
lN
ua

Figure 1.2: Limiting Distribution


an

An important concept that we will need to understand is that of a probability distribution.


bM

R A probability distribution is a table or an equation that links each outcome of a


statistical experiment with its probability of occurrence.
La

Recall that when the value of the variable is an outcome of a statistical experiment, then it is a
random variable. Thus, for instance we can think of a statistical experiment of tossing a coin twice.
We can get 4 possible outcomes- HH,HT,TH and TT. Now let the variable X represent the number
of heads in this experiment. Then X is a random variable and it can take 3 values, 0; 1; 2. We can
construct a table for the values x of the random variable X and the probability associated with that
value.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 18

x p
0 0.25
1 0.50
2 0.25

Table 1.1: Discrete Probability Distribution

This is a probability distribution. Clearly we can see that there will be discrete and continuous prob-
ability distributions depending on whether the variable is discrete or continuous. In the example above

s
of the coin toss, the variable X is a discrete variable and hence this is a discrete probability distribution.

sic
Examples of discrete distributions are the Binomial distribution and the Poisson distribution which we
shall examine shortly.

hy
On the other hand, if the random variable is continuous then the the probability distribution as-

rP
sociated with it is called a continuous probability distribution. Note that a continuous probability
distribution di ers from a discrete probability distribution in that for a continuous distribution, the
lea
probability that the variable assumes a particular value is zero and hence it can't be represented by
a table. Instead, we represent it with a function. Such a function is called a probability distribution
uc
function.
lN

Properties of a Probability Distribution Function


ua

1. Since the continuous random variable is de ned over a continuous range of values
an

(called the domain of the variable), the graph of the density function will also be
bM

continuous over that range.


2. Since the continuous random variable can take an in nite number of values, the
probability that it takes a speci c value, say a is zero.
La

3. Furthermore, the area bounded by the curve of the density function and the x-axis
is equal to 1, when computed over the domain of the variable.
4. Finally, the probability that a random variable assumes a value between a and b is
equal to the area under the density function bounded by a and b. Note that the
area below the line x = a say, is equal to the probability that the variable X can
take any value value less than or equal to a.

An example of a continuous probability distribution that we will study will be the Normal or Gaus-
sian distribution.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 19

1.3.2 Parent & Sample Distribution


Any measurement of a quantity is usually expected to approximate the quantity though not be exactly
equal to it. We have already seen that every time we make a measurement, we expect some discrepancy
between them because of random errors and so we expect every measurement to be di erent. However,
as we increase the number of measurements, we see that the data is more and more distributed around
the correct value of the quantity being measured. (Of course all this is true if we can neglect or correct
for systematic errors).

s
sic
Suppose we make an in nite number of measurements. Then in principle, we would know the ex-

hy
act nature of the distribution of the data points. If we had such a hypothetical distribution, then we
could determine the probability of getting any particular value of the measurement by doing a single

rP
measurement. This hypothetical distribution is called the parent distribution . Thus, in the above
example, the parent distribution for the one rupee coins minted in a particular period would be the
lea
weights of ALL such coins. However, in practice, we always have a nite set of measured values, as
in the example above where we have a sample of 100 coins on which we carried out the measurements.
uc
The distributions that are obtained from the samples of the parent distribution are called sample dis-
tribution. Of course, in the limit of in nite observations, the sample distribution becomes the parent
lN

distribution.
ua

We can de ne the probability distribution function P (x) which is normalised to a unit area. This
function is de ned in such a way that for the limiting distribution (that is in the limit that the number
an

of observations N is very large), the number of observations of the variable x between x and x + x is
given by
bM

N = NP (x)x
La

1.3.3 Mean & Deviation


The parent and the sample distributions discussed above can be characterised by several quantities.
We can de ne a mean of the sample distribution in exactly the same way as we understand it- as the
average value of the quantity. Thus the mean of the sample distribution, x is

R Mean of Sample Distribution : x 


1X N
x
N i=1 i
(1.3)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 20

where xi are the di erent observed values of the variable x. Clearly, the mean of the parent distri-
bution,  is

R Mean of Parent Distribution :   Nlim


1X N
x
!1 N i=1 i
!
(1.4)

If the measurement of interest can be made with high precision, the majority of the values obtained
will be very close to the true value of x, and the limiting distribution will be narrowly peaked about
the value . In contrast, if the measurement of interest is of low precision, the values found will be

s
sic
widely spread and the distribution will be broad, but still centered on the value . Thus, we see that
the breadth of the distribution not only provides us with a very visual representation of the uncertainty

hy
in our measurement, but also, de nes another important measure of the distribution. How can we
characterise this measure?

rP
The most often used parameter for characterising the breadth or dispersion of the distribution is
lea
called the standard deviation, . We can de ne the variance 2 of the parent distribution as

R
uc
 
1 X
Variance of Parent Distribution : 2 = lim (xi )2 (1.5)
lN

N !1 N

which is easily seen to be


ua

 
1 X
2 = lim x2 2 (1.6)
an

N !1 N i

from the de nition of  in Eq(1.4).


bM

This is the measure of dispersion for the parent distribution. What about the sample distribution?
The variance here is de ned in an analogous way, except that the factor in the denominator is N 1
La

instead of N .

R Variance of Sample Distribution : s2 =


"

N
1 N
X
1 i=1
(xi x)2
#
(1.7)

Note that as N approaches 1, N 1 and N are the same. But for any nite N the di erence comes
in because though the initial set of measurements were N independent measurements (since all the N
xi were independent), however, in calculating x, we have used one independent piece of information.
The rigorous proof of this statement is a bit tricky though not required for our purposes.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 21

The importance of these two parameters, the mean  and the standard deviation  (or variance 2 ) is
that this is precisely the information we are trying to extract from the experiment that we perform. For
the sample distribution, s2 characterises the uncertainty associated with our experiment to determine
the true and actual values. As we shall see, the uncertainty in determining the mean of the parent
distribution is proportional to the standard deviation. Thus we conclude that for distributions which
are a result of statistical or random errors, these two parameters describe the distribution well.

How do we determine the mean and standard deviation of distributions? For this, we de ne a

s
quantity called the expectation value .

sic
The expectation value of any function f (x) of x is de ned as the weighted average
of f (x) over all the values of x weighted by the probability density function p(x).

hy
rP
It is easy to see that the mean is the expectation value of x and the variance is the expectation value
of square of the deviations from . Thus for a discrete distribution from Eq(1.4), we need to replace
the observed values xi by a sum over the values of possible observations multiplied by the number of
lea
times we expect the observation to occur. Thus
uc
N N
1X 1X X
  E (X ) = Nlim x = lim
!1 N i=1 i N !1 N j =1 j
[ x NP (xj )] = lim
N !1
[xj P (xj )] (1.8)
lN

In a similar way, the variance can be written as


ua

N 
X  N 
X 
2 = Nlim )2 P (xj ) = Nlim
(xj x2j P (xj ) 2 = E (X 2 ) E (X )2 (1.9)
an

!1 j =1
!1 j =1
For a continuous distribution, the analogous quantities are
bM

Z1
  E (X ) = xp(x)dx (1.10)
La

1
and
Z1 Z1
2 = (x )2 p(x)dx = x2 p(x)dx 2 = E (X 2 ) E (X )2 (1.11)
1 1

Example 1.3.3.1
I throw a dice and get 1 rupee if it is showing 1, get 2 rupees if it is showing 2, get 3 if it is showing
LABORATORY MANUAL FOR NUCLEAR PHYSICS 22

3, etc. What is the amount of money I can expect if I throw the dice 150 times?

For one throw, the expected value is


X
E (X ) = xi P (xi )
The probability of getting any of the digits in one roll is 16 . Thus

1 1 1 1 1 1 7
E (X ) = 1  + 2  + 3  + 4  + 5  + 6  =

s
6 6 6 6 6 6 2

sic
Thus if I roll the dice 150 times, my expected payo is 72  150 = 525 rupees.

hy
Example 1.3.3.2

rP
I am given a probability distribution as below
lea
X P(X)
8 1
uc
4
12 1
6
lN

16 3
8
20 1
5
1
ua

24 8
an

Find the variance of this distribution?


bM

We know that the variance is given by Eq(1.9). But to use this, we rst need to nd the expectation
value of x or the mean. This is
La

X 1 1 3 1 1
E (X ) = xi P (xi ) = 8  + 12  + 16  + 20  + 24  = 17
4 6 8 5 18
Now
X
2 = [xi E (X )]2 P (xi ) = 32:71

Example 1.3.3.3
At a pediatrician's clinic, the age of the children who come to the clinic, x years, is given by the
following probability distribution function:
LABORATORY MANUAL FOR NUCLEAR PHYSICS 23

3
f (x) = x(2 x) 0 < x < 2
4
= 0 otherwise (1.12)

If on a particular day 100 children are brought to the clinic, how many are expected to be under
16 months old?

16 months is 43 years. So the probability of nding a child under 16 months is given by the area

s
under the curve of the probability distribution function between 0 and 43 . This is

sic
hy
4
Z3
4
P (x < ) = f (x)dx

rP
3
0
4
Z3
lea
=
3
4
x(2 x)dx
0
uc
4
Z3
3
= x(2 x)dx
lN

4

0 
3 80
=
ua

4 81
20
=
an

27
(1.13)
bM

Thus for 100 children, the number we expect to be under 16 months is


La

20
100   74:07
27
What about the mean age of the children brought to the clinic? For this,we need to use Eq(1.10).
Thus
LABORATORY MANUAL FOR NUCLEAR PHYSICS 24

Z2
E (X ) = xf (x)dx
0
Z2  
3
= x x(2 x) dx
4
0
= 1
(1.14)

s
sic
This result is not surprising if we try to see how the distribution looks graphically as in Figure
1.3.

hy
rP
1
f(x)

0.8

0.6

0.4
lea
uc
0.2
lN

0
0 0.5 1 1.5 2

Figure 1.3: Graph of function in Eq(1.12)


ua
an

We can see that the distribution is symmetrical about x = 1 and therefore the mean must be in
the middle of the range that is x = 1. This is always true and can be used when we know that
bM

the distribution is symmetrical.

Finally, what about the variance of the distribution of the age of the children?
La

We know that the variance for a continuous distribution is given by Eq(1.11). Thus

2 = E (X 2 ) E (X )2
But
LABORATORY MANUAL FOR NUCLEAR PHYSICS 25

Z2
E (X 2 ) = x2 f (x)dx
0
Z2  
3
= x2 x(2 x) dx
4
0
6
=
5
(1.15)

s
sic
Thus

hy
6 1
2 = 12 =

rP
5 5

1.4 Distributions lea


We have already seen that the results of a statistical experiment result in a distribution (either discrete
uc
or continuous). We would be interested in three kinds of distributions, viz. Binomial, Poisson and
Gaussian or normal distributions. It is important to note where these distributions are used.
lN

The Gaussian distribution is the one which we encounter most frequently since this describes the
ua

distribution of random observations in many experiments.


an

The Poisson distribution is generally used for counting experiments of the kind used in nuclear
physics where the data is the number of events per unit interval. In the study of random processes like
bM

radioactivity, Poisson distribution is important as it is whenever we sort data in bins to get a histogram.

Finally, the binomial distribution is a discrete distribution which is used whenever the possible
La

number of nal states is small. This is true for instance in coin tossing experiments or even in scattering
experiments in particle or nuclear physics.

1.4.1 Binomial Distribution


A Binomial or Bernoulli trial is basically a statistical experiment which has the following properties:

1. The experiment consists of n repeated trials.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 26

2. Each trial can result in just two possible outcomes. We call one of these outcomes a success and
the other a failure.
3. The probability of success, denoted by P , is the same on every trial.
4. The trials are independent, that is, the outcome on one trial does not a ect the outcome on other
trials.
A typical example would be repeated tosses of a coin and counting the number of heads that are
turn up. Clearly the properties mentioned above are satis ed.

s
sic
We can de ne a Binomial random number as the number of successes, say x in a binomial
experiment with n trials. The probability distribution of such a binomial random number

hy
is called the binomial distribution. An example of such a distribution we have already seen in the
discussion of the discrete distribution where the probability distribution is given as a table in Table 1.1.

rP
We can nd an expression for the probability P (x; n) for x successes in a binomial experiment with n
lea
trials by analysing an experiment of coin toss. Suppose we want to know the probability of x coins with
heads and n x coins with tails. For this purpose, we know that there are n Cx di erent combinations
uc
in which we can get the set of observations. In each of these combinations the probability of x heads

coming is px which in this case is 21 x and the probability for n x tails is (1 p)n x = qn x which here
lN


is 12 n x . With this, we can write down the probability P (x; n) of getting x successes in an experiment
with n trials, each with probability p as
ua

R  
an

n x n x n!
P (x; n; p) = pq = n Cx px qn x = px (1 p)n x (1.16)
x x!(n x)!
bM

Mean of Binomial Distribution


La

To nd the mean of the binomial distribution, recall that the mean is just the expectation value of
the random variable. In this case, the random variable is x and the probability distribution function
Eq(1.16) is the probability of x successes in n independent trials when the probability of success in each
trial is p. Thus by the de nition of expectation value, we have
LABORATORY MANUAL FOR NUCLEAR PHYSICS 27

n
X
E (x) = xP (x; n; p)
x=0
n
X n!
= x px (1 p)n x
x=0
x !(n x)!
n
X n!
= px (1 p)n x (1.17)
x=1
( x 1)!(n x )!
since the x = 0 term does not contribute. Now substitute m = n 1 and y = x 1. Then

s
sic
m
X (m + 1)! y+1

hy
E (x) = (p) (1 p)m y
y=0
y!(m y)!

rP
m
X m!
= (m + 1)p (p)y (1 p)m y
y=0
y!(m y)!

= np
m
X m!
y!(m y)!
lea
(p)y (1 p)m y (1.18)
y=0
uc
But we know that the binomial theorem states that
lN

m
X m!
(p + q)m = py qm y
y=0
y!(m y)!
ua

Thus
an

m
X m!
(p)y (1 p)m y = (p + 1 p)m = 1
bM

y=0
y!(m y)!
and so the mean of the Binomial Distribution is given by

R
La

E (x) = np (1.19)

Variance of Binomial Distribution

Recall that the variance of a distribution is de ned as the di erence of the Expectation value of the
square of the random variable and the square of the expectation value , that is

2 = E (x2 ) E (x)2
LABORATORY MANUAL FOR NUCLEAR PHYSICS 28

Consider

n
X
E (x(x 1)) = [x(x 1)]n Cx px (1 p)n x
x=0
n
X n!
= x(x 1) px (1 p)n x
x=0
x!(n x)!
n
X n!
= px (1 p)n x
x=2
( x 2)!(n x )!

s
n
X (n 2)!

sic
= n(n 1) px (1 p)n x
x=2
(x 2)!( n x )!
n

hy
X (n 2)!
= n(n 1)p2 p(x 2) (1 p)n x
x=2
(x 2)!(n x)!

rP
m
X (m)!
= n(n 1)p2 p(y) (1 p)m y
y=0
(y )!(m y )!
lea
= n(n 1)p2 (p + 1 p)m
= n(n 1)p2 (1.20)
uc
where we have used the substitution, y = x 2 and m = n 2. Now variance is
lN

2 = E (x2 ) E (x)2
ua

= E (x(x 1)) + E (x) E (x)2


an

= n(n 1)p2 + np n2 p2
bM

= np(1 p) (1.21)

Thus we have the important result that the variance of a Binomial Distribution is given by
La

R 2 = np(1 p) (1.22)

Note that here we have used the fact that the Expectation value of the sum of two random variables
is the sum of the Expectation values of the variables. That is if X and Y are two random variables,
then

E (X + Y ) = E (X ) + E (Y )
In our case we have used the fact that
LABORATORY MANUAL FOR NUCLEAR PHYSICS 29

E (x2 ) = E (x(x 1)) E (x)


To show this in general is fairly straightforward. Consider two random variables X and Y which can
take values, x1 ; x2 ;    and y1 ; y2 ;    . Now

XX
E (X + Y ) = (xj + yk )P (X = xj ; Y = yk )
j k
XX XX
= xj P (X = xj ; Y = yk ) + yk P (X = xj ; Y = yk )

s
j k j k

sic
X X
= xj P (X = xj ) + yk P (Y = yk )
j k

hy
= E (X ) + E (Y )

rP
where we have used the fact that
X

k
lea
P (X = xj ; Y = yk ) = P (X = xj )

and
uc
X
P (X = xj ; Y = yk ) = P (Y = yk )
lN

j
because the sum over all values of y implies that the joint probability distribution P (X; Y ), that is
P
k P (X = xj ; Y = yk ) will reduce to P (X = xj ).
ua
an
bM

Example 1.4.1.1
An unbiased coin is tossed ten times. What is the probability of getting less than 3 heads?
La

The probability of nding less than 3 heads in 10 tosses, is the probability of nding less than or
equal to 2 heads, P (H  2). This will be the sum of probabilities of nding no heads, 1 head and
two heads. Thus

P (H  2) = P (H = 0) + P (H = 1) + P (H = 2)
Now the probability of nding x heads in n tosses is given by Eq(1.16). In our case, n = 10,
p = q = 21 . Thus
LABORATORY MANUAL FOR NUCLEAR PHYSICS 30

     10
n x n x 10 1 1
P (H = 0) = pq = =
x 0 2 1024
Similarly
       9
n x n x 10 1 1 10
P (H = 1) = pq = =
x 1 2 2 1024
     2  8
n x n x 10 1 1 45
P (H = 2) = pq = =

s
x 2 2 2 1024

sic
Thus the total probability of getting less than 3 heads in 10 tosses is

hy
1 7
P (H = 0) + P (H = 1) + P (H = 2) = [1 + 10 + 45] =
1024 128

rP
Example 1.4.1.2
Here is a game to test sixth sense. Take 4 cards numbered 1 to 4. One person picks a card at
lea
random and another person tries to identify the card. What is the probability distribution that
the second person would identify the card correctly if the test is repeated 4 times?
uc
lN

Let P (X = x) be the probability of correctly identifying x cards after 4 attempts. Then, by the
binomial probability distribution function, this is given by
ua

   
n x n 4
P (X = x) = pq x == (0:25)x (0:75)4 x
an

x x
since the probability of identifying the correct card in 4 attempts, out of 4 cards is 14 = 0:25. Here
bM

x = 0; 1; 2;    ; 4. Thus the probability of getting one card right in 4 attempts is


 
4
P (1) = (0:25)(0:75)3 = 0:4219
La

1
The probability distribution is given by

x P(x)
0 0.3164
1 0.4219
2 0.2109
3 0.0468
4 0.0039
LABORATORY MANUAL FOR NUCLEAR PHYSICS 31

If this is done with say 100 people, we can see that the number of people getting 1 card correct is
100  0:4219  42.

Example 1.4.1.3
A biased that is an unfair dice is thrown fty times and the number of sixes seen is ten. If the
dice is thrown a further fteen times nd:

(a)the probability that a six will occur exactly thrice;

s
(b)the expected number of sixes;

sic
(c)the variance of the expected number of sixes.

hy
The experiment is clearly a Bernoulli or Binomial trial. If the success is taken to be getting a six,
then the probability p is given by

rP
10
p= lea= 0:2
50
Now if X is de ned as the number of sixes in 15 trials, then
uc
lN

X = B (15; p)
We want the probability for getting exactly 3 sixes in 15 trials. Thus, x = 3 and
ua

   3  12
10 15 1 4
 0:2475
an

B (15; ) =
50 3 5 5
bM

The expected number of sixes will be the expectation value E (X ). This is, from Eq(1.19),

1 15
E (X ) = np = 15  = = 3
5 5
La

Finally, the variance is given by Eq(1.21)

1 4
2 = np(1 p) = 15   = 2:4
5 5
1.4.2 Poisson Distribution
A Poisson distribution is the probability distribution that results from a Poisson experiment. A Poisson
experiment has the following properties:
1. The experiment results in outcomes that we can call successes or failures.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 32

2. The average number of successes  that occur in a speci ed region is known.


3. The probability that a success will occur is proportional to the size of the region.
4. The probability that a success will occur in an extremely small region is virtually zero.
A Poisson random variable is the number of successes that result from a Poisson ex-
periment. The probability distribution of a Poisson random variable is called a Poisson
distribution. A Poisson distribution is an approximation to the binomial distribution when the aver-
age number of successes, that is  is much smaller than the possible number, that is when   n. In
such cases, evaluation of the binomial probability is extremely complicated and tedious.

s
sic
We have already discussed the Binomial Distribution which is the result of a Bernoulli trial. Recall

hy
that in a Bernoulli trial, we take a sample of de nite size and count the number of times a certain event
occurs. We thus know the number of times the event did occur and the number of times the event did

rP
not occur. However, there are some instances when we do know the number of times the event occurs
but do not know the number of times that the event did not occur. As an example, think of watching
lea
a thunderstorm for an hour. I can count the number of times that I see a lightning ash, say 15 times.
However I cannot say how many times the lightning ash did not occur. This is the case where isolated
uc
events are occurring in a continuum of time. Or for instance, suppose under a microscope I look for
the number of malarial parasites in a blood sample. Here what we have is isolated events (the sighting
lN

of a malarial parasite) in a continuum of volume or area of the slide. In such cases, we cannot use the
Binomial distribution because we do not know the value of n in (p + q)n . It is in these kinds of cases
ua

that we can use the Poisson Distribution.


an

As an example of a Poisson distribution, consider the ux of cosmic rays reaching the earth. This
is known to be around 1 per cm2 per minute. Now consider a detector with a surface area of 40 cm2 .
bM

We expect to detect 40 cosmic rays per minute in this detector. Now suppose we record the number
of cosmic rays detected in a 20 second interval. On an average we then expect about 13:3 cosmic rays.
However, when we do the experiment over many 20 second intervals, we will detect something like
La

13; 15; 14; 12 etc and occasionally even 9 or 18 cosmic rays. We can plot a histogram of this, that is
plot the number of times nx that we observe x rays in this xed interval of time. Or, if we divide
the number of times nx by the total number of intervals N , then we can get the probability Px of
observing x cosmic rays in this experiment. If our number of intervals N is large, then this probabil-
ity distribution will be a Poisson distribution. Whenever we observe independent random events that
occur at a constant rate such that the expected number of events is , we get a Poisson distribution.
In the case of cosmic rays, the events are obviously random and clearly independent since the arrival
of one cosmic ray does not depend on the arrival of others. Further, the rate of arrival is almost constant.

Another example can be a scattering experiment with a beam of B particles incident on a thin foil
LABORATORY MANUAL FOR NUCLEAR PHYSICS 33

and the probability p of any one interaction taking place is very small. Then we know that the number
of observed interactions r will be binomially distributed where we will take the number of trials as B
and the probability of success (interaction) as p. It turns out that when p is very small, then the values
of Px will be like a Poisson distribution with a mean given by Bp (which we have seen is the mean of
the binomial distribution Eq(1.19)).
Thus we see that a binomial distribution goes to a Poisson distribution when the number of trials
N increases while the probability of success p decreases in such a way that Np is a constant.
Consider the case of a binomial distribution where p  1 and we consider the situation where n ! 1
but np remains nite. Recall that np is the mean of the binomial distribution (Eq1.19). The probability

s
function for the binomial distribution is

sic
1 n!
P (x; n; p) = px (1 p) x (1 p)n

hy
x! (n x)!
But

rP
n!
= n(n 1)(n 2)    (n x + 2)(n x + 1)
(n x)! lea
Now since x  n, each of these x factors is very nearly n and hence this becomes
uc
n!
(n x)!
 nx
lN

Now
ua

(1 p) x  (1 + px)  1
an

since p ! 0. Thus we now have the probability function as


bM

1 x x
P (x; n; p) = n p (1 p)n
x!
Consider now
La

 
1 1
(1 p)n = plim
!0
[(1 p) p ] = =e 
e
since the mean,  = np for a Binomial distribution.

Thus we get the Probability Distribution function for the Poisson distribution

R PP (x; ) = plim P (x; n; p) =


!0 B
x
x!
e  (1.23)

This is the probability of obtaining x events in the given interval. Remember that x is positive
LABORATORY MANUAL FOR NUCLEAR PHYSICS 34

integer or zero.

Mean of Poisson Distribution

The mean of the distribution is the expectation value of the random variable. Thus

1  
X x 
E (x) = x e
x=0
x!

s
1  
X x 1

sic
= e 
x=1
(x 1)!

hy
1 y
X 
= e 
y=0
y!

rP
=  (1.24)
lea
Thus we see that the mean of the Poisson distribution is

R
uc
E (x) =  (1.25)
lN
ua

Variance of the Poisson Distribution


an

We know that the variance is de ned in terms of di erence of the expectation value of the square of
the variable and square of the expectation value. That is
bM

2 = E (x2 ) E (x)2
La

Consider
LABORATORY MANUAL FOR NUCLEAR PHYSICS 35

1  x 
X 
E (x2 ) = x2 e 
x=0
x!
1  
X
2 x 
= 0+ x e
x=1
x!
1  
X y+1
= (y + 1)2 e 
y=0
(y + 1)!
1  
X y 
= (y + 1)2 e 

s
(y + 1)(y)!

sic
y=0
1  
X y
=  (y + 1) e 

hy
y=0
(y)!
" #
1   1  y
X y  X 

rP
=  (y) e + e 
y=0
(y )! y=0
(y )!
= E (x) +  lea
= ( + 1) (1.26)
uc
Since the sum of the probability distribution function over all x is unity and therefore the second
term in the square bracket is unity. Thus we see that the variance is simply
lN

R
ua

2 = E (x2 ) E (x)2 = ( + 1) 2 =  (1.27)


an

This is a remarkable result. The mean and the variance of the Poisson distribution is the
same. This gives rise to the famous square root rule. In an experiment where the distribution
bM

satis es Poisson distribution conditions, for instance in the counting of N independent events in a xed
p mean of the distribution to be N . Then, as pwe saw above, the variance is
interval, we can estimate the
also N and therefore
p  = N . The statistical errors in them would be N and we would quote our
La

results as N  N . We will return to this later in the Chapter when we discuss error estimation.

Example 1.4.2.1
A factory produces resistors and packs them in boxes of 500. If the probability that a resistor is
defective is 0:005, nd the probability that a box selected at random contains at most two resistors
which are defective.

If we take X as the `number of defective resistors in a box of 500', then


LABORATORY MANUAL FOR NUCLEAR PHYSICS 36

X = B (500; 0:005)
since the trials are obviously Binomial. Now in this case, we can see that the number of trials,
n = 500 is large and the probability of success p = 0:005 in each trial is low, so the Binomial
distribution can be approximated by a Poisson distribution with a mean

 = np = 500  0:005 = 2:5


Then the probability of nding two or less defective resistors in a box of 500 is

s
sic
P (X  2) = P (0) + P (1) + P (2)

hy
where

rP
2:50 e 2:5
P (0) =
lea 0!
2:5 e 2:5
1
P (1) =
1!
uc
2:52 e 2:5
P (2) =
lN

2!
or
ua

P (X  2) = 6:625  e 2:5  0:543


an

Example 1.4.2.2
bM

A bank manager opens on an average 3 new accounts per week. Use the Poisson distribution to
calculate the probability that in a given week she will open 2 or more accounts but less than 6
accounts.
La

To use the Poisson distribution function, we need to know the mean. In this case, the mean is
given as 3 accounts per week. Thus we can use Eq(1.23) with  = 3. Then the probability of
opening 2 or more accounts but less than 6 in a week is simply

e 3 32 e 3 33 e 3 34 e 3 35
P (2 < x < 6; 3) = P (2; 3) + P (3; 3) + P (4; 3) + P (5; 3) = + + + = 0:7167
2! 3! 4! 5!
Example 1.4.2.3
Thirty sheets of plain glass are examined for defective pieces. The frequency of the number of
LABORATORY MANUAL FOR NUCLEAR PHYSICS 37

sheets with a given number of defects per sheet was as follows:

No. of defects Frequency


0 8
1 5
2 4
3 7
4 4
5 2

s
Table 1.2: Distribution of Defects in Glass sheets

sic
hy
rP
What is the probability of nding a sheet chosen at random which contains 4 or more defects?

lea
From Table 1.2 we see that the total number of sheets is 30 while the total number of defects in
these 30 sheets is 60. To use the Poisson distribution function, we need to nd the mean number
uc
of defects. We know that there are 60 defects in 30 sheets of glass. Thus the mean number of
defects per sheet is
lN

60
= =2
30
ua

The probability then of nding a sheet with 4 or more defects is


an
bM

P (x  4) = 1 P (x < 4)
= 1 P (0) P (1) P (2) P (3)
 2 0 2 (2)1 e 2 (2)2 e 2 (2)3 
e (2) e
= 1 + + +
La

0! 1! 2! 3!
= 0:1431

Example 1.4.2.4
At the ITO intersection, vehicles pass through at an average rate of 600 per hour.

(a)Find the probability that none passes in a given minute.


(b)What is the expected number passing in ve minutes?

The average number of vehicles per minute is simply


LABORATORY MANUAL FOR NUCLEAR PHYSICS 38

600
= = 10
60
Thus the probability that no vehicle passes in a given minute is

e 10 100
P (0; 5) = = 4:53  10 5
0!
The expected number of vehicles passing in ve minutes is

E (X = 5) = 10  5 = 50

s
sic
hy
1.4.3 Normal Distribution

rP
The Normal Distribution is extremely important in probability theory and statistics and forms the
cornerstone of most of statistical analysis. For our purposes, its importance lies in the fact that many
lea
real-world phenomena involve random quantities that are distributed in an approximately normal fash-
ion. For instance, the errors in a scienti c measurement are approximately normal. It is often called
uc
Gaussian distribution and also referred to as \bell-shaped distribution", because the graph of its prob-
ability density function resembles the shape of a bell.
lN

R The Normal or Gaussian Distribution is an approximation to the Binomial dis-


ua

tribution for the limiting case when the number of possible observations, that is n goes
to in nity AND the probability of success in each measurement is nite and remains con-
an

stant, that is when np  1. It is also the limiting case of the Poisson distribution when
the mean  becomes large.
bM

The Gaussian probability density is de ned as


La

R 1
PG = p exp
 2
"

2 
 #
1 x  2

(1.28)

As one can see this is a continuous distribution function and thus describes the probability of getting
a value x from a random observation from a parent distribution of mean  and standard deviation .
Properly de ned, we should talk about a Gaussian Probability Distribution Function, such that
the probability dPG of the random observation having a value between x and x + dx i.e.

dPG (x; ; ) = pG (x; ; )dx


LABORATORY MANUAL FOR NUCLEAR PHYSICS 39

With this probability distribution function, we can now see how a Poisson distribution goes to a
Gaussian distribution for large mean. Consider rst a Poisson distribution and a normal distribution
both with mean 1. Thus  = 1 and in the case of the Gaussian distribution,  = 1. Then the probability
that n  2 that is n   + 1 for the Poisson distribution is

P (1  1) = P (0) + P (1) + P (2) = :736


where P (n) is the Poisson distribution function of Eq(1.23). With the Gaussian distribution function
(Eq(1.28)), we get

s
sic
P (0 < x < 2) = 0:68
Thus we see that for small means, the two distributions are fairly di erent. What about for large
p

hy
mean? Let us take  = 152 = 7:5. Then  = 7:5 = 2:7. Then the corresponding probabilities for
 + 1 are

rP
P (7:5  2:7) = P (0) + P (1) +  + P (10) = 0:64
lea
and for the Gaussian distribution
uc
P (0 < x < 10:2) = 0:68
lN

which is very similar. In general, for  > 5, the Gaussian distribution is a good approximation to the
Poisson distribution. This fact will be important to us since we will see that in counting experiments,
ua

it will become easier to use the Gaussian distribution in cases where the mean number of counts is large.
an
bM

Properties of a Normal Distribution


1. The total area under the normal curve is equal to 1.
La

2. The probability that a normal random variable x equals any particular value is 0.
3. The probability that x is greater than some value a equals the area under the normal curve
bounded by a and 1.
4. The probability that x is less than a equals the area under the normal curve bounded by a and
1.
5. About 68% of the area under the curve falls within 1 standard deviation of the mean.
6. About 95% of the area under the curve falls within 2 standard deviations of the mean.
7. About 99:7% of the area under the curve falls within 3 standard deviations of the mean.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 40

A convenient form of the normal distribution is the Standard Normal Distribution. To obtain
this, we simply use the substitution
x 
z= (1.29)

in Eq(1.28). Then the probability distribution function for the standard normal distribu-
tion becomes

R 1
pG (z )dz = p exp
 
z2
dz (1.30)

s
2 2

sic
It is important to see that since for all the values of X , the normal variable falling between x1 and

hy
x2 have corresponding Z (the standard normal variable) values between z1 and z2 , it means that the
area under the X curve between X = x1 and X = x2 equals the area under the Z curve between Z = z1

rP
and Z = z2 . Therefore, we have, for the probabilities,

lea
P (x1 < X < x2 ) = P (z1 < Z < z2 )
uc
Mean of the Standard Normal Distribution
lN
ua

Z1
E (x) = xpG (x)dx
an

1
Z1  
bM

1 x2
= p x exp dx
2 2
1
2 3
Z0   Z1  
x2 x2
La

= p1 4 x exp
2
dx + x exp
2
dx5
2
1 0
= 0 (1.31)

Thus we see that the mean of standard Normal distribution is 0.

Variance of the Standard Normal Distribution

We know that
LABORATORY MANUAL FOR NUCLEAR PHYSICS 41

2 = E (x2 ) E (x)2
Now

Z1
E (x2 ) = x2 pG (x)dx
1
Z1  
1 x2
= p x2 exp dx

s
2 2

sic
1
2 3
Z0  
 Z1   
1 4 x2 x2
= p x x exp dx + x x exp dx5

hy
2 2 2
2
1 0 3
0 1 Z1

rP
1
 
x2  0 Z 
x2   
x2 
x2 
= p 4 x exp + exp dx + x exp + exp dx5
2 2 1 1 2 2 0 2
lea 0
Z1  
1 x2
= p exp dx
2 2
uc
1
= 1 (1.32)
lN

R R
using Integration by parts ( udv = uv vdu) and also the fact that the probability distribution
function is normalised to 1.
ua
an

Therefore

R
bM

2 = E (x2 ) E (x)2 = 1 0 = 1 (1.33)

Thus we see that the standard normal distribution has a mean equal to 0 and a variance
La

equal to 1.

The di erence between the standard normal distribution and the normal distribution can be seen
from the curves for probability distribution. Consider a normal distribution with  = 2;  = 13 . The
probability distribution function will look like Figure 1.4
LABORATORY MANUAL FOR NUCLEAR PHYSICS 42

s
sic
hy
rP
Figure 1.4: Normal Distribution with mean = 2 and  = 1
3

lea
The corresponding standard normal distribution with  = 0;  = 1 will resemble Figure 1.5
uc
lN
ua
an
bM
La

Figure 1.5: Standard Normal Distribution with mean =0 and  = 1

The two graphs obviously have very di erent  and  but have identical shapes and a shifting of
the axes will give one from the other. It is also easy to see that the area under the two curves be-
tween two equivalent points is the same. Thus, for instance, the area of the Normal distribution (with
 = 2;  = 13 between 0:5 to 2 to the right of the mean will be the area from x1 =  + 2 = 2 + 16
to x2 =  + 2 = 2:66. The area under the Standard normal distribution would be the area from
z1 =  + 2 = 0 + 0:5 = 0:5 to z2 =  + 2 = 0 + 2 = 2:0.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 43

The Standard Normal distribution with the probability distribution function given by Eq(1.30), gives
us the probability of nding the value. This is also the area under the curve from 0 to the value. This is
usually tabulated in a z-table which can be looked up as a standard reference as given below in Figure
1.6.

s
sic
hy
rP
lea
uc
lN

Figure 1.6: z-tables for Standard Normal Distributionx


x(Source: http:// www.katyanovablog.com/ picsgevs/ normal-distribution-table )
ua

Example 1.4.3.1
an

The mean weight of 1000 parts produced by a machine was 30:05 gm with a standard deviation
of 0:05 gm. Find the probability that a part selected at random would weigh between 30:00 gm
bM

and 30:15 gm?


La

30:00 is 1 that is 0:05 below the mean. Similarly, 30:15 is 2 = 0:1 above the mean. Thus

P (30:00 < X < 30:15) = P ( 1 < Z < 2)


since recall that the area under the normal gaussian curve between two points x1 and x2 is the
same as under a standard normal curve between two points z1 and z2 which are related to x1 and
x2 by the transformation Eq(1.29). And 1 for the standard normal distribution is 1 from the
mean that is 0 and 2 is 2. These values can be looked up from the standard tables.

P ( 1 < Z < 2) = 0:3413 + 0:4772 = 0:8185


LABORATORY MANUAL FOR NUCLEAR PHYSICS 44

So the probability is 0:8185.

What about the Gaussian distribution? The Mean of the Gaussian distribution is obtained easily
now either by a substitution as given above into the standard normal distribution or by direct calcula-
tion.

Z1  
1 (x )2
E (x) = p

s
x exp dx
 2 2 2

sic
1
Z1   Z1  
1 (x )2 1 (x )2
= p dx + p

hy
(x ) exp  exp dx
 2 2 2  2 2 2
1 1

rP
Z1  2
(x )
= 0+ p1 exp
2 2
dx
 2
1 lea
=  (1.34)
uc
since the distribution function is normalised to 1.
lN

We can also easily see this by using the Standard Normal distribution. Recall that the standard
Normal distribution and a Gaussian distribution are related by Eq 1.29,
ua

x 
z=
an


Thus for a normal random variable X , we can write it as a linear combination of the standard normal
bM

variable Z as

x = z + 
La

Now using the linearity of expectation values, we have

E (x) = E (z + )
= E (z ) + 
= 

since E (z ) = 0 as we saw above.


Thus we see that for a Gaussian distribution,the mean is given by
LABORATORY MANUAL FOR NUCLEAR PHYSICS 45

R E (x) =  (1.35)

We can also nd the Variance of Gaussian Distribution easily.

2 = E ((x )2 )
Z1  
1 2 (x )2
= p (x ) exp dx
 2 2 2

s
1

sic
Z1  
2 (y)2
= p y2 exp dy
2 2

hy
1
= 2 (1.36)

rP
since the integral is simply the variance of the Standard Normal Distribution given in Eq(1.32) which
is 1. Thus we see that the Variance of the Gaussian distribution is 2 .
lea
The probability distribution function for the Normal or Gaussian distribution (Eq(1.28) is normalised
uc
to 1. That is
lN

Z1
PG (x)dx = 1
ua

1
From the de nition of the probability distribution function, we know that the probability that any
an

one x value lies between the limits x =   and x =  +  is simply the the area under the Gaussian
bM

curve between these limits. If one computes (by integration) such areas for various choices of , one
can show that the probability of nding any one measurement of x between various limits, measured
as multiples of the standard deviation, , is given by the data given in Table 1.3.
La

Probability Interval
0:50  0:674 < x <  + 0:674
0:68   <x<+
0:80  1:282 < x <  + 1:282
0:90  1:645 < x <  + 1:645
0:95  1:960 < x <  + 1:960
0:99  2:576 < x <  + 2:576
0:999  3:291 < x <  + 3:291
LABORATORY MANUAL FOR NUCLEAR PHYSICS 46

Table 1.3: Normal Distribution: Probabilities with intervals

How can one interpret this table? The table indicates that we can be 95% con dent that any one
measurement that we make in the experiment (assuming all measurements are distributed normally)
will lie within approximately 2 of the mean. Thus, the probability column can be taken as the
con dence level and the interval column as the con dence interval.

s
This interpretation and analysis looks very straightforward. However, there is a problem- the prob-

sic
lem is that the  and the  that we are using in the Gaussian distribution is the parent mean and
standard deviation. This means, as we have discussed above, that this will only be valid if we make an

hy
in nite number of measurements!

rP
We will address this issue in Section 1.6.
lea
1.5 Error Estimation
uc
lN

The basic aim in any experiment is to measure a quantity and also estimate the uncertainties in the
measurements. We also need to understand the sources of the uncertainties. Lastly we need to know
how to combine uncertainties in measurements of more than one quantity into the error in the quan-
ua

tity which is calculated from these measurements. This is what we now discuss. Throughout this
an

section, we will only be dealing with statistical or random errors. Systematic errors will
be assumed to have been taken care of.
bM

First of all, it is important to understate a crucial fact which allows one to use the statistical methods
discussed above to analyse the errors in any experiment. The crucial result is that any measurement
La

subject to many small random errors will be distributed normally. This follows from the
Central Limit Theorem which states that the distribution of the sum of a large number of random
variables will tend towards a normal distribution. We may think of a measurement as being the result
of a process namely our carrying out many small steps in the experiment. Each step in the process
may lead to a small error with a probability distribution. When we sum the error over all steps to get
nal error, the Central Limit Theorem guarantees that this will lead to a normal distribution no matter
what the error on the individual steps works out to be. So as a result of this, we generally expect
normal distributions to describe errors. Note that this simple, yet powerful fact allows us to use the
whole machinery of normal distributions and statistics to analyse errors.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 47

In the case of observations that we are taking are collections of nite number of counts over nite
intervals, then the underlying distribution we know is Poisson. In this case, we know that the observed
values would be distributed around the mean in a Poisson distribution. (Recall that the random variable
in a Poisson distribution can only take positive values, including zero since it is de ned as the number
of successes in a Poisson experiment.) In fact, in any experiment where data is grouped in bins to form
a histogram, the number of events in each bin will be distributed according to a Poisson distribution.

This allows us a tremendous simpli cation. We know that for a Poisson distribution, the standard
deviation is, Eq(1.27), simply the square root of the mean of the Poisson distribution.

s
sic
= 
p

hy
Recall that we have de ned relative uncertainty, Eq(1.1), as

rP
uncertainty
Relative Uncertainty =
measured value

The measured value for us is the mean, that is . The uncertainty in this is obviously the standard
lea
deviation, that is . Thus, we can say that
uc
 1
relative uncertainty = =p
 
lN

In our counting experiments, for instance, this means, that as we increase the number of counts per
interval (that is increase the mean ), the relative uncertainty goes down as the square root of the mean.
ua

This is actually referred to in general as the Square root rule for Counting Experiments which
states that the uncertainty in the any counted number of random events, which is used as an estimate
an

of the true average number, is equal to the square root of the counted number. For our purposes, this is
bM

extremely important in our counting experiments since the process which we are measuring, namely the
counts are random and distributed in a Poisson distribution in any time interval. However, care should
be taken to note that this uncertainty is only in the counted variable and not in any derived
La

quantity. Thus, for instance if we were to measure N counts in an interval of T seconds, to get the
rate R per second,
N
R=
T
p the uncertainty in R, we know that the uncertainty is in the measured random variable
Now to nd
N and it is N . Thus the number of counts in time T is really
p
N N
From this the rate can be seen to be simply
LABORATORY MANUAL FOR NUCLEAR PHYSICS 48

p
N N
R=
T
and NOT
r
N N
R= 
T T
since only the quantity N is counted and hence the uncertainty in N is the square root of N .

This example above then leads us to the issue of how to estimate the error in any derived quantity

s
from the errors in the measured quantities?

sic
hy
1.5.1 Propagation of Errors

rP
Consider an experiment where we measure some quantities, for instance the number of counts and the
distance and use these measured quantities to determine a quantity which is a function of the two
lea
measured quantities. Let the desired quantity by x and the measured quantities, u and v be such that
uc
x = f (u; v) (1.37)
lN

Suppose further that the most probable value of x, x is such that

x = f (u; v) (1.38)


ua

To determine the most probable value of x, we take the measurements of u and v, that is ui and vi and
an

determine the di erent xi . That is


bM

xi = f (ui ; vi ) (1.39)
We have already seen that in the limit of an in nite number of measurements, the sample distribution
La

will go to the limiting distribution or parent distribution and the average of x, that is x will be the
mean of the distribution. In that limit, we can use the calculated sample mean, x to nd the variance
x2 as
 
1X
x2 = lim
N !1 N
(xi x)2 (1.40)
Also, expanding the function in Eq(1.37) in a Taylor series around the averages of u and v,
   
@x @x
xi x ' (ui u) + (vi v) (1.41)
@u @v
Of course the partial derivatives have to be evaluated when the other variable is kept xed at the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 49

mean value.

Now combining Eq(1.40) and Eq(1.41), we have

@x 2
    
1 X @x
2 x ' lim
N !1 N
(ui u) + (vi v)
@u @v
"  2  2   #
1 X
2 @x 2 @x @x @x
' lim
N !1 N
(ui u)
@u
+ (vi v)
@v
+ 2(ui u)(vi v)
@u @v
(1.42)

s
Clearly, the rst two terms are related to the variance of u and v, that is

sic
 
1 X
u2 = lim (ui u)2 (1.43)

hy
N !1 N
and

rP
 
1X
N !1 N
v2 = lim
(vi v)2 (1.44)
lea
We also de ne a new quantity covariance between the two variables u and v as
uc
 
2 = lim 1 X
uv N !1 N
(ui u)(vi v) (1.45)
lN

Thus the variance for x is given by


ua

R 2 2
x ' u
 2
@x 2
 2
@x 2
  
@x @x
an

+ v + 2uv (1.46)
@u @v @u @v
bM

This is known as the error propagation equation.

The covariance term which is a measure of the correlation of the variations in u and v. In most
La

experiments the uctuations in the measured variables are uncorrelated and so on the average the
covariance term vanishes for a large number of observations. We shall neglect this in our further
discussion. Thus we have in general, the error propagation equation as

R 2 2
x ' u
 2
@x 2
+ v
@u
 2
@x
+ 
@v
(1.47)

where the quantity x could be a function of any number of uncorrelated, independent


variables.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 50

Let us see what the error propagation equation looks like in some common cases.

1. Sums & Di erences


Let

x=u+a (1.48)
where a is a constant. Then since @x = 1, we get
@u

s
sic
x = u (1.49)

hy
We can also nd the relative uncertainty (Eq(1.1))

rP
x u 
= = u (1.50)
x x u+a
2. Weighted Sums & Di erences lea
Suppose x is the weighted sum of two variables u and v.
uc
x = au + bv
lN

where a and b are constants. Then, since


ua

 
@x
=a
an

@u
 
bM

@x
=b
@v
we get using Eq(1.46)
La

x2 = a2 u2 + b2 v2 (1.51)


assuming no correlation.
3. Multiplication & Division
Suppose the quantity of interest x is de ned by

x = uv
where u and v are measured quantities. In this case, we can see that
LABORATORY MANUAL FOR NUCLEAR PHYSICS 51

 
@x
=v
@u
and
 
@x
=u
@v
and therefore the error propagation equation tells us that

s
x2 = v2 u2 + u2 v2 (1.52)

sic
or

hy
x2 u2 v2

rP
= + (1.53)
x2 u2 v2
For the case of division, we have lea
u
uc
x=
v
lN

then

@x 1
ua

=
@u v
an

@x u
= 2
@v v
bM

and therefore

2 u2 2
La

x2 = u2 + 4 v
v v
or

x2 u2 v2


= + (1.54)
x2 u2 v2
Note the very important di erence between Eqs(1.54, 1.53) and Eq(1.51). In the case of weighted
sums and di erences the absolute errors are relevant while in this case, it is only the fractional
errors in u and v which are related to the fractional error in x.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 52

4. Powers
Suppose

x = aub
where a and b are constants. Then
 
@x bx
= abub 1 =
@u u

s
and, using Eq(1.47)

sic
bx
x = u 

hy
u

rP
The relative uncertainty in x is

x blea
=  (1.55)
x u u
uc
To illustrate the use of the Error Propagation Equation, let us consider a few examples.
lN
ua

Example 1.5.1.1
Consider an experiment where we count N1 = 945 counts in a 10 second interval in an experiment
an

and then N2 = 19 counts in a 10 second interval. We have already found a background reading of
NB = 14:2 counts for the same 10 second interval by carrying out a separate experiment carefully.
bM

Thus we assume that there is no uncertainty in the background reading. Now in the
rst time interval, the corrected counts are
La

x1 = N1 NB = 930:8 counts
with an uncertainty of
p
x1 = N1 = 945 ' 30:7 counts
and a relative uncertainty of

x1 30:7
= = 0:032 ' 3:1%
x 930:8
On the other hand, in the second interval, the gures are
LABORATORY MANUAL FOR NUCLEAR PHYSICS 53

x2 = 19 14:2 = 4:8 counts


with an uncertainty of
p
x2 = N2 = 19 ' 4:4
and a relative uncertainty of

x2 4:4
= = 0:91

s
x 4:8

sic
Example 1.5.1.2

hy
Now suppose in the previous example, the background reading also was subject to some uncer-
tainty. Then we have the formula

rP
x=u+v
and thus the error propagation becomes
lea
uc
x2 = u2 + v2
lN

since the partial derivatives are unity. The uncertainty in x is


ua

p
x = u2 + v2
an

In the above experiment for the rst reading, if the background was not error free, then we will
have the error in the net count to be
bM

q p
x1 = N2 1 + N2 B = (30:7)2 + (3:7)2 = 30:97
La

and so we should report our net counts as

930:8  30:9
Example 1.5.1.3
Suppose we want to nd the ratio of the activity of two sources by measuring the counts from
them. Then suppose the counts from the rst source is N1 = 586 and that from the second is
N2 = 265. Then the ratio of activity is given by

N 586
R= 1 = = 2:211
N2 265
LABORATORY MANUAL FOR NUCLEAR PHYSICS 54

The error in R by Eq(1.54) is given by

x2 u2 v2 N1 N2 1 1


= + = 2 + 2 = + = 2:3  10 2
x2 u2 v2 N1 N2 586 265
and therefore
p
R = R  2:3  10 2 = 2:211  0:151 = 0:333
and therefore the ratio of activities should be quoted as

s
sic
R = 2:21  0:33

hy
Example 1.5.1.4

rP
We now consider a simple example where we can repeatedly use the error propagation formulas
given above to get the error in a compound quantity. Consider the simple experiment of nding
g, the acceleration due to gravity. For this, I throw a mass from top of the telescope tower and
lea
measure h the height of the tower and also t, the time it takes for the mass to reach the ground.
I get
uc
lN

t = 1:9  0:1 s
and
ua

h = 18:5  0:2 m
an

Now I can use the formula h = 21 gt2 to determine g as


bM

2h
g = 2 = 10:24 m s 2
t
La

I need to know the uncertainty in h and t which are given and then use the error propagation
rules to nd the uncertainty in g.
We can write

x
g=
y
where

x = 2h
LABORATORY MANUAL FOR NUCLEAR PHYSICS 55

and

y = t2
Now we can use Eq (1.54) to get

g2 x2 y2


= +
g 2 x2 y 2
To get the error in x, we have to use Eq(1.51) to get

s
sic
x2 = 4h2

hy
Similarly for the error in y, we can use Eq(1.55) to get

rP
y 2t
=
y t
Thus we have
lea
uc
2t 2
 
g2 4h2 4t2  h 2
= + = +
g2 4h2 t2 h t
lN

Now
ua

h 0:2
= = 0:011 = 1:1%
h 18:5
an

and
bM

t 0:1
= = 0:052 = 5:2%
t 1:9
Therefore,
La

g 2  h 2 2t 2
   
= + = :0112 + 4  :0522 = :0109
g h t
and therefore

g = :11  10:24 m s 2 = 1:07 m s 2


Therefore we should quote our result as

g = 10:24  1:07 m s 2
LABORATORY MANUAL FOR NUCLEAR PHYSICS 56

We can see that the accepted value of g, that is 9:8 ms 2 is within the margin of error. Also, since
the contribution to the error in g from t is much more, one should improve the measurement of t
rather than try to make the measurement of h better since that contributes very little to the error.

What we see from this example is that we can repeatedly use the various error propagation formulas
discussed above to get the error in any compound expression. However, this is something that
we need to be careful when using. The error propagation formulae that we have used
work ONLY when the quantities are independent. Here for instance, we could write
the quantity g as xy because we know that the errors in x (which is basically the

s
error in h) are independent of the error in the denominator y (which is basically the

sic
error in t). In another situation where we have the SAME variable appearing in the

hy
numerator and the denominator, this assumption is NOT true. In that case, we need
to use the general error propagation equation Eq(1.46). We shall encounter such a

rP
situation in Chapter 6.
Example 1.5.1.5
lea
A radioactive sample is counted for one minute and we observe 900 counts. We then remove the
sample and count the background rate for 10 minutes to observe 100 counts. What is the net
uc
count rate and the uncertainty in it?
lN

There are two measured quantities in this experiment- the gross count rate and the background
count rate. (We assume that the time has no uncertainty). The uncertainty in each of these
ua

measured quantities, we have seen is simply the square root of the measurement. Thus
an

p
Gross count rate = RG = 900  900 = 900  30 counts per minute
bM

p
Background count rate = RB = 100 100 = 10010 in 10 minutes = 101 counts per minute
La

Net count rate = RN = RG RB = 890 counts per minute


What about the uncertainty in RN ? For this we need to use the error propagation equation since
remember, RN is a derived quantity and NOT a measured quantity. The de ning relation for
RN is given above and so we get
q p
RN = R2 G + R2 B = (30)2 + (1)2  30
LABORATORY MANUAL FOR NUCLEAR PHYSICS 57

Thus we should quote our result as 890  30.

Example 1.5.1.6
A Geiger counter is placed near a suspected source of radioactivity and it records 58 counts in 30
seconds. The source is removed and the background count is found to be 48 counts in 30 seconds.
Can we be sure that the source is truly radioactive?

The total count rate is

s
sic
NT 58
RT = = = 116 counts per minute
T 0:5

hy
The uncertainty in this measurement is
p p

rP
 NT 58
RT = N =
T T
=
0:5
 15 counts per minute
The background count rate is lea
uc
NB 48
RB = = = 96 counts per minute
T 0:5
lN

and the uncertainty is


p p
ua

 NB 48
RB = N =
T T
=
0:5
 14 counts per minute
an

Our net computed source count rate is thus


bM

RN = RT RB = 20 counts per minute


Once again, this net count rate is NOT a measured but a derived quantity. So we need to use
La

the error propagation equation to nd the uncertainty in this given the uncertainties in the total
count rate and the background count rate. Thus
q p
RN = R2 T + R2 B = 152 + 142  21 counts per minute
Thus we have to report our result as 20  21 counts per minute and conclude that we can not say
for sure whether the source is radioactive or not.

Example 1.5.1.7
Finally, let us consider a very simple example where there are both systematic and random errors.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 58

Suppose I want to measure an unknown resistance R2 . I can for instance use the following circuit.
(Adapted from `A Practical Guide to Data Analysis for Physical Science Students' by L. Lyons)

s
sic
hy
rP
lea
uc
Figure 1.7: Resistance Measurement
lN

We measure the resistance R1 , voltages V1 and V2 using meters and then deduce the value of R2
using Ohm's law
ua

V V1
R2 = 2
an

R1 (1.56)
V1
bM

Clearly, the experiment involves measurement of 3 quantities and each of these will have random
and systematic errors. Random errors we already know happen because of statistical uctuations.
What about the systematics? In this particular case, there could be many sources of systematic
La

errors. Thus, for instance, the voltmeters V1 and V2 may not be calibrated properly and/or the
resistance meter might also not be calibrated. We could also have the resistance being temper-
ature dependent and hence not be very accurate. Or there could be some stray capacitances in
case we are using an AC source. We would need to determine or estimate the amount of these
systematic errors.

Now suppose, we performed the experiment and obtained the readings for R1 ; V1 and V2 as follows:

R1 = (2:0  0:1
)  1%
V1 = (1:00  0:02V)  10%
LABORATORY MANUAL FOR NUCLEAR PHYSICS 59

V2 = (1:30  0:02V)  10%


The results of the measurements are reported with the average value plus/minus the random error
and the second error in percentage is the estimate of the systematic errors. The random errors
come as previously discussed from statistical uctuations and are known accurately. The system-
atic errors quoted above are assumed to have been estimated using a variety of methods.

Now as we have been discussing, the nature of random and systematic errors is di erent. However
sometimes we may want the error in our measurement as a single gure. In this case, since the

s
two sources of error, random and systematic are uncorrelated, we should add them in quadrature

sic
just as we saw in the error propagation equation. Thus we get

hy
R1 = 2:0  0:1

rP
V1 = 1:00  0:10V
V2 = 1:30  0:13V
lea
But note that the quantity of interest is not V2 or V1 but V2 V1 . What about the error in
uc
this? This depends crucially on whether we measure the voltages using the same equipment or
separate equipment. If the same equipment is used, then the errors are obviously correlated while
lN

if separate instruments are used, then the errors are uncorrelated. If di erent meters are used,
then
ua

V2 V1 = (1:30  0:13) (1:00  0:10) = 0:30  0:16 V


an

where we have added the errors in quadrature. Now if the same meter is used then the errors
bM

are correlated and we have to include this e ect. Thus, if there is a 10% systematic uncertainty
in each measurement, then the overall result will also have a 10% uncertainty. Thus,
La

V2 V1 = (1:30  0:02) (1:00  0:02)  10% = 0:30  0:03  10% = 0:30  0:04 V
where the errors have been added in quadrature.

We are nally ready to see what the error in the quantity of interest, namely R2 is.
 
V2 V1 V
R2 = R1 = 2 1 R1
V1 V1
h i
Think of this expression as the product of two measured quantities, VV12 1 and R1 . If we know
the errors in both these, then we can use Eq(1.53) to nd the error in R2 . The error in R1 is
LABORATORY MANUAL FOR NUCLEAR PHYSICS 60

h i
known as above. The error in VV21 1 is simply the error in VV21 . Again the error in this quantity
can be found from Eq(1.54) once the errors in V2 and V1 are known.

The error in VV12 can be calculated assuming that the same meter is used to measure both these
quantities. Then

V2 (1:30  0:02)
= = 1:30  3% = 1:30  0:03
V1 (1:00  0:02)
since the systematic errors cancel out using the same meter. Thus the value of

s
sic
 
V2
1 = 0:30  0:03
V1

hy
Therefore

rP
R2 = [0:30  0:03] [2:0  0:1] = 0:60  0:07

lea
What if we had done the following: We have the error in V2 V1 above. We have the errors in R1
and V11 . Why couldn't we have used the error propagation equation for these three quantities and
uc
found the error in R2 ? This would be wrong because we cannot assume, under any circumstances
lN

that the errors in V2 V1 and V11 are uncorrelated since V1 comes in both of these and no matter
how we measure V1 , the errors will always be correlated. Thus one has to be careful when using
ua

related quantities in determining the errors in a calculated quantity.


an

1.6 Estimation and Error of the Mean


bM

Recall that when we do any measurement, we typically end up with a sample distribution of the data
points of the quantity being measured. We saw that in the limit of an in nite number of measure-
La

ments, this sample distribution goes to the parent distribution. Of course, the aim of any experiment
is ultimately to determine the parameters of the parent distribution. We also saw that a random set of
observations or measurements are distributed according to a Gaussian (or Poisson distribution). The
question we ask is what is the best estimate of the mean  of the parent distribution?

1.6.1 Method of Maximum Likelihood


Suppose we conduct an experiment where we obtain N data points which are randomly selected from
the parent distribution. This set of N points of course is our sample distribution. The question
we need to answer is how are the parameters, x and s, that is the average and standard
LABORATORY MANUAL FOR NUCLEAR PHYSICS 61

deviation of the sample distribution related to  and , the mean and standard deviation of
the parent population? If the parent distribution is Gaussian with mean  and standard deviation
, then the probability of nding x1 (one of the data points in the sample distribution) in the range x1
and x1 + dx1 , which we will de ne as the probability of nding x1 , is simply
"  #
1 x1  2

1
P (x1 ) = p exp (1.57)
 2 2 
Similarly, the probability of nding x2 in the range x2 and x2 + dx2 is
"  #
1 x2  2


s
1
P (x2 ) = p exp

sic
(1.58)
 2 2 

hy
and so on. Note that each of these observations are independent and hence these probabilities are
uncorrelated. Thus, the probability of nding x1 between x1 and x1 + dx1 , x2 between x2 and x2 + dx2

rP
etc is simply

lea "  #
1 X xi  2
 N 
1
P (x1 ; x2 ; x3 ;    ; xN ) = P (x1 )P (x2 )P (x3 )    P (xN ) = p exp (1.59)
 2 2 
uc
It is important to stress that this probability is for a speci c value of  and . But we don't know
lN

what that value is! So we start o by guessing a value of  and , say 0 and . We compute the
probability P0 ; given Eq(1.59). Then we guess another value of , say 00 with the same  and compute
ua

P00 ; . We do this for several values of  with the same . Then we choose that value of  for which
the probability is a maximum. This is known as the Method of Maximum Likelihood.
an

Now it is clear that for the probability (as a function of ) in Eq(1.59) to be a maximum, the
bM

argument of the exponential should be a minimum since the probability is a constant times a negative
exponential. Let
"  #
La

 2

1 X xi
Y=
2 
Then, we want to nd the value of  for which
"  #
d 1 X xi  2

dY
= =0
d d 2 
or
 
X xi 
=0

or
LABORATORY MANUAL FOR NUCLEAR PHYSICS 62

X
xi = N
which gives us

R =
P

N
xi
= x (1.60)

Thus, we see that the most probable value of the population mean  is precisely the aver-
age or sample mean x of the sample distribution.

s
sic
This is an extremely powerful result. It basically allows us to get an estimate of the parent popu-

hy
lation mean from a sample distribution which is all what we can possibly obtain from any experiment.
What about the error in this mean?

rP
1.6.2 Estimated Error in the Mean lea
We know that the sample standard deviation, s, is the average uncertainty associated with each of the
uc
measurements x1 ; x2 ;    ; xN . A legitimate question from our point of view is also, what is the estimated
uncertainty or standard deviation of in our determination of the mean  for the parent distribution or
lN

x for the sample distribution? Recall that our assumption was that each of the data points xi is from
the same parent distribution and hence it is characterised by the same standard deviation .
ua

To nd the uncertainty in the mean, we can use the error propagation equation Eq(1.47) to nd 
an

using Eq(1.60). Thus we have


bM

"  2 #
X @
2 = i2 (1.61)
@xi
La

Now, as we have supposed, all the data points have the same , that is i =  for all i. We also have
 
@ @ 1X 1
= xi =
@xi @xi N N
Thus, we get

R 2 =
X
" 
i2
1
N
2 #
=
2
N
(1.62)

Thus we see that the uncertainty in the mean is


LABORATORY MANUAL FOR NUCLEAR PHYSICS 63

R  = p
N
(1.63)

The quantity  is called the standard deviation of the mean or the standard error . Thus we
see that if we take N measurements of some quantity x and obtain x1 ; x2 ;    xN , then we can state
our result as the best estimate of x which we have seen is the mean of the sample population and the
uncertainty as Eq(1.63).

s
Best estimate of a variable from N measurements, xi

sic
x = x  

hy
where

rP
1 X
x = xi
N
lea
and
uc
 = px
N
lN

x is simply the sample standard deviation s.


ua

This result has an enormous signi cance for us. The standard deviation x or s represents the av-
an

erage uncertainty in the individual measurements x1 ; x2 ;    ; xN . Thus, if we were to make some more
measurements (using the same technique), the standard deviation x = s would not change apprecia-
bM

bly. On the other hand, the standard deviation of the mean,  would slowly decrease as we increase
N . This decrease is just what we would expect. If we make more measurements before computing an
average, we would naturally expect the nal result to be more reliable, and this improved reliability is
La

just what the denominator guarantees. Stated another way, by taking more and more measurements,
we smoothen the distribution (the histogram of the data points) and also can determine the peak (the
mean) in an improved fashion. But note that the increase in precision is only growing as the square
root of the number of measurements and so there is a limitation in by how much we can increase the
precision by taking more and more measurements.

We can now return to the question that we had raised in the Section on Gaussian distribution? How
does one establish con dence limits in the absence of an in nite number of observations? Now that we
have the best estimate of the standard deviation of the mean, we can see that the best way to report our
experimental observations of a parameter x whose mean x has been determined experimentally from
LABORATORY MANUAL FOR NUCLEAR PHYSICS 64

taking N measurements is

best value of x = x  
Using the table for Gaussian probability intervals, we can say that this way of reporting the result
is at 68% con dence level. We can increase the con dence level to 95% by reporting it as

best value of x = x  2


Note once again that the values for con dence limits used are for the Gaussian distribution and not

s
for the sample distribution which is obtained by a nite set of measurements whose standard deviation is

sic
s. The relationship between con dence limits and the sample deviation s was obtained by W.S. Gosset
and is known as the Student's t-distribution. Basically, there exists a function which can only be

hy
evaluated numerically that allows us to compute a single number which relates s to the con dence level.
What we can then say is that the true value of the parameter x fell in the interval

rP
x pt < true value of x < x + p
lea t
N N
where N is the number of measurements and t can be found from the tables at various con dence
uc
levels and values of N . We will not go into the details of the t- distribution but only note that in
the limit that N ! 1, the values of t at various con dence levels approaches that of the Gaussian
lN

distribution as it should.
ua

Basically, for our purposes it is enough to note that for a small number of observations, the pre-
dictions from Student's t-distribution are more accurate since the Gaussian probability distribution
an

overestimates the con dence level associated with a given range. To put it another way, the Student's
t-distribution probability requires a larger uncertainty estimate than the Gaussian probability and the
bM

two only coincide for a very large number of measurements.


La

1.7 Method of Least Squares


In an experiment suppose we nd a collection of data points (xi ; yi ) and we want to nd a relationship
between them. The theoretical relationship between these two quantities, x and y is supposed to be
linear. On plotting the points (xi ; yi ), we see that the relationship looks somewhat linear. In this case,
we can ask two questions- rstly, assuming the relationship to be linear, can we nd the actual rela-
tionship, that is y = Mx + C which best ts the data? This of course means nding the best estimates
of the constants M , the slope of the straight line and the C , the y intercept. The nding of the best t
straight line to the data is called linear regression or the least square t for a line. Once we have
found the best t straight line, then we can also ask the question about the goodness of t, that is to
LABORATORY MANUAL FOR NUCLEAR PHYSICS 65

say, how well does our data t the assumed straight line? Let us rst see how to nd the best t linear
relation between the two quantities x and y.

We assume that the uncertainty in measuring the variable x is negligible while there is an appreciable
uncertainty in the measuring the variable y. In most experiments, this assumption is a reasonable one
since usually, the uncertainty in one of the variables is much larger than that in the other one. For in-
stance, when we measure speed, the distance measurement has much larger uncertainties than the time
variable. Or, in our case for instance, when we plot the characteristic of the number of counts versus
voltage, the error in the voltage is negligible. We also assume that the measurements of the variable

s
y is governed by a Gaussian distribution of the same standard deviation, y for all measurements yi of y.

sic
Now the linear relationship we have assumed between x and y is

hy
rP
y = Mx + C
If we knew M and C , then for any value of x, say xi , we would know the true value of the quantity
y, that is lea
yiT  True Value of yi = Mxi + C
uc
But we have assumed that the measurements of y are distributed according to a Gaussian distribution,
lN

and therefore all the measurements of yi will be distributed in a Gaussian distribution centered on this
\true value" which would be the mean, . It also follows that the probability of nding a particular
ua

value of yi is
an

"  #
C 2

1 1 yi Mxi
PM;C (yi ) = p exp 2 y
(1.64)
bM

y 2
Similarly, we can obtain the probabilities of obtaining all the values of y, namely yi ; y2 ;    ; yN . Since
these values are all independent, the probability of obtaining our complete set of measurement, that is
La

yi ; y2 ;    ; yN is simply the product of each of these probabilities.


" #
1 1X N
(yi Mxi C )2
PM;C (y1 ; y2 ;    ; yN ) / exp (1.65)
yN 2 i=1 y2
Let us de ne a quantity 2 as
N
X (yi Mxi C )2
2 = (1.66)
i=1
y2
Then the probability can be easily written as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 66

 
1 2
PM;C (y1 ; y2 ;    ; yN ) / N exp (1.67)
y 2
We now use the same technique as we did in the previous section to nd the best values of M and
C , that is the method of Maximum Likelihood. The best values of these quantities would be those for
which the probability in Eq(1.67) is a maximum. Or, since the probability is proportional to a negative
exponential, those values of M and C for which 2 in Eq(1.66) is a minimum. To get those values of
the parameters M and C , we di erentiate 2 w.r.t them and equate to zero. Thus
 
@2 2 X N

s
= 2 x (y Mxi C) = 0 (1.68)

sic
@M C y i=1 i i
and

hy
 
@2 2 X N
= (y Mxi C ) = 0 (1.69)
@C y2 i=1 i

rP
M
These equations can be rewritten in a more suggestive form as

X X
lea X
C xi + M x2i = xi yi
uc
X X
CN + M xi = yi (1.70)
lN

Solving these simultaneous equations, we get


P P P
ua

N xi yi xi yi
M= P 2 P (1.71)
N xi ( xi )2
an

and
bM

P P P P
x2i Pyi xi xi yi
C= P (1.72)
N x2i ( xi )2
De ning a quantity  as
La

X X 2
=N x2i xi
we get
P P P
N xi yi xi yi
M= (1.73)

and P P P P
x2i yi xi xi yi
C= (1.74)

With these estimates of the slope M and the intercept C , we get the Least Square t of the straight
LABORATORY MANUAL FOR NUCLEAR PHYSICS 67

line or the line of regression of y on x.

The question now remains as to what is the uncertainty in the measurements of the quantity y?
Before we do this, it is important to remember that the measurements y1 ; y2 ;    ; yN are NOT the
measurements of the same quantity. So, a spread in their values is NOT a measure of the uncertainty in
their values. The measurement of each yi is, we have already assumed normally distributed around the
best t value which we have seen to be Mxi + C with a standard deviation which is assumed to be y .
The deviations of the measured value from the true value or best t value would thus be yi Mxi C
and these too would be normally distributed around a mean of 0 and with a standard deviation of y .

s
To obtain the best estimate for y , we again need to use the Method of Maximum Likelihood with

sic
the probability of nding the measured values y1 ; y2 ;    ; yN which is given in Eq(1.67). But this time,
the parameter we are interested in is y and so we need to nd the most likely value of y which will

hy
maximise the probability in Eq(1.67). Thus we di erentiate Eq(1.67) with respect to y to get

rP
X
Ny2 = (yi Mxi C )2
or
r
lea
1X
uc
y = (yi Mxi C )2 (1.75)
N
lN

It turns out that just as in Eq(1.7), we needed to divide by N 1 instead of N because the number
of independent values had been reduced in computing the mean, in this case too, since we need to nd
ua

two parameters M and C (from the measurements) we need to divide by N 2 instead of N . In any
statistical calculation, we can nd the number of degrees of freedom by taking the number of independent
an

measurements and subtracting the number of parameters determined using these measurements. Thus
we nally get the expression for y , the uncertainty in the measurements yi ; y2 ;    ; yN is
bM

r
1 X
y = (yi Mxi C )2 (1.76)
N 2
La

This nally leaves us with the issue of the uncertainty in the quantities that we have determined,
that is M and C . For this, we know the expressions for M and C in terms of yi , namely Eq(1.73) and
Eq(1.74). Thus, knowing the uncertainty in yi , that is y which we have just determined, we can easily
use the error propagation equation Eq(1.46) to determine the uncertainty in M and C . These are given
by
r
N
M = y (1.77)

and
LABORATORY MANUAL FOR NUCLEAR PHYSICS 68

rP
x2i
C = y (1.78)


1.8 Goodness of Fit


Let us consider a typical counting experiment where we obtain counting statistics. The experimental
data consists of a series of measured or observed quantities. Let us assume that we have N independent
measurements of the same quantity. These could be anything- in our case, these could be the number
of counts in a speci ed time interval that we have taken repeatedly. This set of observations allow us to

s
sic
prepare a sample distribution which as we have already seen can be characterised by two quantities-
the sample mean, x and the sample variance s2 .

hy
The underlying distribution which describes the process is, as we have seen, called the parent dis-

rP
tribution which is characterised by the mean,  and the variance 2 . This distribution is either Poisson
or Gaussian, depending on the value of the mean  since we recall that for large values of  typically
lea
larger than 20, these two distributions become almost identical. But the question is that we don't know
these parameters of the parent distribution. So how do we proceed?
uc
We take the sample mean x and assume it to be the mean of the underlying parent distribution.
lN

This is something which is a good approximation as we saw in the section on Method of Maximum
Likelihood. With this mean for the parent distribution, we compute the actual variance 2 . We also
ua

have the sample variance s2 . If our estimate of the sample distribution being a good representative of
the parent distribution is good, then we expect that the two, that is the sample variance and the true
an

variance should be close to each other. This comparison is done quantitatively by a procedure known
as the Chi-squared test.
bM

Basically, what we are trying to see is whether an obtained or measured set of frequencies in a random
sample and what we expect from an assumed statistical hypothesis match and how well they match. In
La

our present case, the Chi Squared test allows us to determine how well the observed sample distribu-
tion and the assumed parent distribution (in this case a Poisson distribution with a mean  = x ) match.

We de ne 2 as
N
1X
2 = (x x)2 (1.79)
x i=1 i
If we recall the de nition of the sample variance s2 from Eq(1.7), we can rewrite this in terms of the
sample variance as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 69

R 2 =
(N
x
1)s2
(1.80)

In our case of the underlying distribution being a Poisson distribution,we know that the variance
of the parent distribution 2 is simply the mean of the distribution . But we have already chosen
2
the mean  to be the same as the sample mean x. Thus, the deviation of the ratio sx from unity is
a measure of how much the sample variance di ers from the variance. In terms of the 2 , we can say
that if the sample distribution is truly Poisson, then 2 = N 1. Any departure from this would be a

s
measure of how much the sample distribution di ers from a Poisson distribution.

sic
Thus we see that 2 is a statistic that tells us about the dispersion of the observed frequencies from

hy
the expected frequencies. It is usually convenient to de ne a quantity called the degrees of freedom
 as

rP
=N lea Nc (1.81)
where N is the number of sample frequencies and Nc is the number of constraints. One way to think
uc
of the number of constraints is that it is the number of parameters which have been calculated from
the data to determine the probability distribution. Thus, in the case above of the Poisson distribution,
lN

we have calculated one parameter, the sample mean x from the data and therefore in that case, the
number of degrees of freedom  is simply N 1. With this, we can de ne a quantity called reduced
Chi sqaured or ~2 as
ua

R
an

2
~2 = (1.82)
bM


This reduced chi squared clearly has an expectation value equal to 1. If the calculated values of ~2
are much larger than 1 then we can say that either our measurements are not good, or the underlying
La

probability distribution that we have assumed is incorrect. If the value of ~2 is very small then again
there is some problem with the experiment.

Another way to think of the the 2 squared test is to think of trying to t a model to some given
data. Recall the quantity 2 that we encountered while discussing the Method of Least Squares in
Eq(1.66). There, we had de ned
N
X (yi Mxi C )2
2 =
i=1
y2
for the case of the least square tting of a straight line. It is obvious that in the general case, suppose
LABORATORY MANUAL FOR NUCLEAR PHYSICS 70

we have some observations yobs and we assume an underlying model for the data which yields the values
yth for the data points, then we can de ne 2 as

2 XN
(yobs yth )2
 =
i=1
i2
Typically, we can have various competing models for our data or, what is the same thing, several dif-
ferent values of some model dependent parameter. To decide which model (or the value of a parameter)
ts the data best, we compute the 2 from the data and from that determine the reduced chi squared,
~2 . From our discussion above, it is clear that we should choose that model, or the value of the model

s
parameter, for which the reduced chi squared ~2 is closest to 1.

sic
The 2 probability distribution function is given by

hy
1
P (x2 ;  ) = 2   2 2 e x22

rP
 x
2 2 ( 2 )
We can show easily that with this probability distribution function, that
lea
2  E (2 ) = 
uc
and
lN

V ar(2 )  E ((2 )2 ) (E (2 ))2 = 2


ua

This is evident from Fig 1.8.


an
bM
La

Figure 1.8: Probability distribution for 2 for di erent degrees of freedom

We can look up tables of 2 (for instance in Appendix A or at en.wikibooks.org/wiki/Engineering


Tables/Chi-Squared Distibution or www.pd.infn.it/lunardon/didattica/docsper2/TavoleChi2.pd) to
nd out the probability associated with any value of ~2 . The tables tabulate the values of
LABORATORY MANUAL FOR NUCLEAR PHYSICS 71

Z1
Probability (~2  ~2o ) = P (x2 ;  )dx2
2o
or
Z1
1  2 x2
Probability (~2  ~2o ) =
  x2 2 e 2 dx2 (1.83)
22 (2) 2
o
where  is the number of degrees of freedom, ~2o is the calculated value of the reduced chi-squared
from the observed data and ~2 is the expected value of the reduced chi-squared from our model.

s
Clearly, from this we can say that if our model is correct, we expect

sic
p
2    2

hy
Given this, we can ask the question that we started with- do our observations correspond to our

rP
underlying model (in our case, a Poisson distribution)? Or to put it another way, what is the probability
that our observed value of 2 or a larger one, could arise purely by chance? This is the probability in
lea
Eq(1.83) and is called the p-value. It is basically the area of the probability distribution function curve
from the observed value of 2 to 1 as can be seen in the Fig 1.9.
uc
lN
ua
an
bM

Figure 1.9: p-value and area under the curve


La

Basically, we know that if the observations were a good approximation to the underlying model, then
2
~ should be close to 1 and from the graph is it clear that the value of the probability is around 0:5.
If the t is not good, the value of ~2 will be larger and the probability smaller. Thus suppose in our
experiment, our data gives an observed value of 2 = 220:1 and there are 199 degrees of freedom. Then
looking up the tables, we see that the value of Probability (~2  ~2o ) is 0:12 or 12% roughly. What this
means is that if our observations were indeed from a parent Poisson distribution, then if we repeat our
experiment many times, and get di erent data, in roughly 12% of the experiments, we will get these
values.

Another way to see this is as follows: Suppose we carry out an experiment and the observations give
LABORATORY MANUAL FOR NUCLEAR PHYSICS 72

us a 2 value of 1:80. Suppose that the number of degrees of freedom in the experiment is 1. Then we
see that the reduced chi-squared, ~2 is 1:80. If we believe that the underlying distribution is Gaussian
(or Poisson with a large mean, which as we have seen goes to a Gaussian distribution), then can we say
that the underlying assumption of a Gaussian distribution is ruled out?

If our assumption about the Gaussian nature of the underlying distribution is correct, then we can
see from the tables, that probability of obtaining a reduced chi-squared of 1:80 or larger is simply

Probability (~2  1:80)  18%

s
sic
That is, if our results were governed by an underlying Gaussian distribution, then there is a 18%
probability that we would obtain a value of ~2  1:80. This seems like a reasonable agreement and we

hy
can accept the hypothesis that the underlying distribution is Gaussian.

rP
In general, we need to decide on a cut-o below which we will not accept the hypothesis as true.
Usually, this is taken as either 5% or 1% and the result is quoted as being at 5% or 1% signi cance
level. To reiterate, what this means is that if lea
Probability (~2  ~2o ) < 5%
uc
we reject our expected or hypothesised distribution at 5% signi cance. As an example of reading
lN

the tables, given in Appendix A, let us assume we have an experiment with 10 degrees of freedom and
suppose we obtain a value of reduced chi-squared as 2:2. Then from the table in Appendix I, we see
ua

that the probability of obtaining 2  2:2 is slightly less than 2%. With this, we can safely say that we
an

can reject the hypothesis at 5% signi cance but not at 1% signi cance.
bM

Let us consider another example of the use of p-values. Consider a counting experiment where we
take the number of counts in 100 , 1 minute intervals. We know that the counting statistics should
follow a Poisson distribution. The data is as follows:
La

Counts/minute (xi ) Occurrences (fi ) Expected Number of Occurrences


0 7 7.5
1 17 19.4
2 29 25.2
3 20 21.7
4 16 14.1
5 14 12.1
P
We can see that the sample mean, or the average is simply x = 2:59 that is x = Pxfi fi i . If we
assume that the underlying distribution is Poisson and choose the mean of the Poisson distribution as
our sample mean, we can calculate the expected number of occurrences as shown in Column 3 of Table
LABORATORY MANUAL FOR NUCLEAR PHYSICS 73

1.8. To calculate the reduced chi-squared, we also need the number of degree of freedom. There are
6 bins that we have with 0; 1; 2; 3; 4; 5 counts. We have used two degrees of freedom to calculate the
mean and the variance (which in the case of a Poisson distribution is the same, that is i2 = x since
we have assumed that the best guess for the mean of the underlying Poisson distribution is the sample
mean.) and thus the number of degrees of freedom is  = 4. We can now easily calculate the reduced
chi-squared, ~2 using Eq 1.79 and Eq 1.82 as

2 2 1 X N
(yobs yth )2
~ = =
  i=1 i2

s
We nd that ~2 = 0:35. This value is less than one and so we believe that the agreement with our

sic
hypothesis is good and the data seems to support that the underlying distribution is indeed Poisson.

hy
To see this using p-values, we use Appendix A for  = 4 to see that

rP
Probability(~2  0:35)  85%
This con rms what we have seen namely that the agreement between experiment and hypothesis is
very good.
lea
uc
The p-value is the probability (measured from 0 to 1, or 0% to 100%) that the hypothesis that
the data corresponds to the model is true. You can reject the hypothesis if the p-value found for a
lN

calculation is less than 0:05(5%) or less than 0:01(1%). For example, a p-value of 0:03 when comparing
the observed data and the theoretically expected data (Poisson distribution in our case) means that
ua

there is only a 3% chance that the two samples are NOT similar. Therefore, this means there is a 97%
chance they ARE similar. To emphasise, if we get a p-value of 0:01 then this means the following: that
an

there is a 1% chance of obtaining a set of measurements at least this di erent (that is this di erent or
more di erent) from the model, assuming the model is true. It does NOT mean that the probability
bM

that the model is true is 1% or that the probability that the model is false is 99%.
La

One should however keep certain things in mind while using the Chi-squared test. One, we are as-
suming the errors in the data are Gaussian. If the errors have been under-estimated then an improbably
high value of chi-squared can be obtained. On the other hand, if the errors have been over-estimated
then an improbably low value of chi-squared can be obtained. In normal experiments, some errors can
sometimes be non-Gaussian, a model is typically only rejected for very low values of p such as 0:001.

1.9 References
1. \Data Reduction & Error Analysis for the Physical Sciences", D. Keith Robinson & Phillip R.
Bevongton, Mcgraw Hill (2003).
LABORATORY MANUAL FOR NUCLEAR PHYSICS 74

2. \An Introduction to Error Analysis: The Study of Uncertainties in Physical Measurements", John
R. Taylor, University Science Books (1997).

3. \Radiation Detection & Measurement", Glenn F. Knoll, Wiley India (2009). Chapter 3.

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La
Chapter 2

RADIOACTIVITY

s
sic
hy
Learning Objectives

rP
1. To understand the nature of radioactive decay.
lea
2. To study quantitatively the phenomenon of radioactivity including half life, decay
constant etc.
uc
3. To learn about the nature and properties of di erent kinds of radiation in radioac-
lN

tivity.
ua
an

2.1 Radioactivity
bM

Radioactivity, discovered in 1896 by Becquerel has played an important role in our understanding
of the nature of matter at the subatomic level. Radioactivity has certain characteristic features which
were inexplicable when it was discovered but can be easily explained within the context of quantum
La

mechanics and relativity. Thus, for instance, the fact that a radioactive nucleus can spontaneously
decay and liberate energy without any excitation from outside can only be understood by thinking of
the equivalence of mass and energy. The completely random or probabilistic nature of the decay process
cannot be understood classically but comes out naturally within the framework of quantum mechanics.

There are basically ve kinds of radioactive decay as listed in the Table (2.1).

75
LABORATORY MANUAL FOR NUCLEAR PHYSICS 76

Decay Transformation Example Reason for Instability


Alpha Decay A
ZX ! A{4 4
Z-2Y + 2He
238
92U ! 234
90Th + 2He
4
Nucleus is too large
Beta Decay A
ZX ! A {
Z+1Y + e +   14
6C ! 14
7N + e
{
Nucleus has too many neutrons relative to protons
Positron Emission A
ZX ! A +
Z-1Y + e +  64
29Cu ! 64
28Ni + e
+
Nucleus has too many protons relative to neutrons
Electron Capture A
ZX + e
{
! A
Z-1Y
64
29Cu + e
{
! 64
28Ni Nucleus has too many protons relative to neutrons
Gamma Decay A *
ZX ! A
ZX + 38 Sr !
87 * 87
38 Sr + Nucleus has excess energy

Table 2.1: Radioactive Decaysx

x(Adapted from \Concepts of Modern Physics" by Beiser,Mahajan & Rai Choudhury )

s
sic
We shall discuss these di erent kinds of phenomenon in some detail later.

hy
2.1.1 Measure of radioactivity

rP
A quantitative measure of the radioactivity of a sample is activity which is de ned as the rate at
which the atoms of the sample decay. If N is the number of atoms (or nuclei) present at time t, then
lea
its activity R is de ned as
uc
R dN
lN

R= (2.1)
dt
Clearly, since the derivative is negative, the negative sign in the de nition makes the activity a
ua

positive quantity. The SI unit of activity is becquerel (Bq) which is de ned as


an

1 Bq = 1 decay s 1
bM

The traditional unit of activity is the curie (Ci) which was de ned originally as the activity of 1
gram of radium 236
88Ra, but is now de ned as
La

1 Ci = 3:70  1010 decays s 1 = 37 GBq


As we shall see, the radiations which come out when a radioactive nucleus decays are ionizing in
nature and when they pass through living tissue, they can damage the tissue. Sometimes the damage
maybe slight and the body heals itself. But sometimes the damage can be severe and have long term
disastrous consequences which include cancer and other illnesses.

Radiation dose is measured in a unit called sieverts (Sv) which is de ned as the amount of radiation
(of any kind) which has the same biological e ect as the absorption of 1 joule of X-rays or gamma rays.
Typically, a safe exposure to radiation is taken to be about 1 milliSv per year. This does not include the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 77

background radiation to which we are anyway exposed. The background radiation that we experience
is from the radionuclides in the rocks and earth and buildings as well as due to cosmic rays hitting the
atmosphere. To put this in perspective, a typical X-ray exposes us to about 0:02 mSv while a CT scan,
which is basically several X-ray exposures, typically can expose us to 5 8 mSv.

2.1.2 Activity Law & Half Life


When we measure the activities or the rate of decay of radionuclides, we nd that the rate falls o
exponentially with time. This can be encapsulated mathematically as

s
sic
R(t) = R0 e t (2.2)

hy
where the constant  is characteristic of the radionuclide and is called the decay constant. It is
convenient to de ne another quantity called the half-life. T1=2 as the time in which the activity drops

rP
to one half of its initial value. Thus, we see that by de nition,
R lea
R(T1=2 ) = 0 = R0 e
2
T1=2
uc
or
lN

T1=2 = ln 2
and so we get
ua

R
an

ln 2
T1=2 = (2.3)

bM

It is clear that for a radionuclide which has a large decay constant, the half life is small
and vice versa.
La

The empirical activity law (Eq(2.2)) can be obtained if we assume a constant probability  per unit
time for the decay of every nucleus in the sample. Then, in time dt, the probability for decay of any
one nucleus is simply dt. This is the probability for any one nucleus to decay. If we have N nuclei,
the number dN that will decay in time dt will be clearly

dN = Ndt
Now since the number of nuclei (of a particular kind which we had initially) is decreasing, this
expression must have a negative sign for it to make sense. Thus we get
LABORATORY MANUAL FOR NUCLEAR PHYSICS 78

dN = Ndt (2.4)
We can de ne another quantity called activity A of a radionuclide. This is simply the rate of decay
of the sample. Thus clearly

R A=
dN
dt
= N (2.5)

s
Integrating this expression gives us the Radioactive decay law as

sic
R

hy
N (t) = N0 e t (2.6)

rP
It is important to note that the whole phenomenon of radioactivity is statistical in nature. As noted
above, every single nucleus has a de nite probability of decay but which particular one decays in a
lea
particular interval of time is essentially random in nature. All we can say is that if we had many nuclei
present (which is always the case in any experiment that we do), the fraction that will decay in any time
uc
period will be approximately the same as the probability of an one nucleus to decay. As we have seen in
Chapter 1, we can model this statistical phenomenon as a Poisson distribution and the probabilities will
lN

then be given by the distribution function for the Poisson distribution. Thus if we say that a particular
sample has a half life of 1 hour, all it means is that every single nucleus in that sample has a 50%
ua

probability or chance of decaying in 1 hour. And recall that the decay probability is constant and thus
if a particular nucleus does not decay in 1 hour, it has a 75% probability of decaying in 2 hours and
an

NOT a 100% probability.


bM

We can also de ne a quantity called Mean Lifetime which is simply the reciprocal of the decay
constant. Thus
La

R T=
1 T1=2
=
 ln 2
= 1:44T1=2 (2.7)

Finally, a quantity which is used sometimes to describe a radioactive source is speci c activity
which is de ned as the activity per unit mass. The mass of the radioactive nuclide with N nuclei and
having a molecular weight (or atomic weight) M , will be given by
NM
mass = (2.8)
AAv
Thus, using Eq(2.8), Eq(2.1), Eq(2.2) and Eq(2.4), we have
LABORATORY MANUAL FOR NUCLEAR PHYSICS 79

NAAv
speci c activity =
NM
AAv
= (2.9)
M
where AAv is the Avogadro's number or 6:02  1023 nuclei mole 1 .

Example 2.1.2.1
Consider a sample of 113

s
49In weighing 2g with a half life of 1:6582 hours. Calculate the number of

sic
atoms remaining in the sample after 4 hours as well as the speci c activity of the sample. Assume
that the daughter nuclides are stable.

hy
rP
We rst need to calculate the number of atoms initially in the sample. Given the mass (M ) and
the atomic weight (Aw ) of the sample, this is easily done

N0 =
M
lea
A = 1:066  1016 atoms
Aw Av
uc
We next need the decay constant . Knowing the half life  or T1=2 , we know from Eq 2.3 that
lN

ln 2
= = 1:16  10 4 s 1

ua

Then using the Radioactive Decay Law (Eq 2.6), we get the number of atoms after 4 hours as
an

N = N0 e t = 2:006  1015 atoms


bM

Speci c Activity can be found from the activity since we know that the speci c activity is simply
the activity per unit mass. Activity is given by Eq 2.5 as
La

A = N = 2:32  1011 decays per second


Thus

A
Speci c Activity = = 1:16  1011 decays per second per microgram
M

We now have the de nitions of quantities used to describe radioactivity.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 80

It is also important to remember that typically, a radioactive nuclide decays and produces a daughter
nuclide and some radiation( alpha, beta and gamma radiation). The daughter nuclide is also usually
radioactive and decays itself producing another nuclide which also may or may not be radioactive. Thus,
typically there is a radioactive series or a decay chain in which di erent nuclides are being produced
and are decaying. Of course, the time scales of the production of the nuclides depends on the half lives
of the parent nuclides while their decay time scales are related to their own half lives. This decay chain
can be analysed and leads to di erent kinds of behaviour with time. We address this in Appendix D.

We next turn to a discussion of the nature of particles and radiation emitted by the radionucleus

s
during the process of its decay by radioactivity. These we have seen can be of several types as given in

sic
Table 2.1.

hy
rP
2.2 Nuclear Decay
There are several facts that we know about the nucleus. Let us recall them.
lea
uc
1. Nuclei consist of positively charged protons and neutral neutrons. These are collectively known as
nucleons.
lN

2. The size of the nucleus is of the order of 1 femtometer or 10 15 m. This unit is also called a Fermi.
ua

3. The density of nucleons is roughly the same in the inside of the nucleus and hence the nuclear
radius is proportional to A1=3 where A is the mass number. This relationship is usually written as
an

R = R0 A1=3 with R0  1:2  10 15 m


bM

4. Nuclear densities are enormous- a typical nucleus will have a mass density of  1017 kg m 3 .
5. Protons and neutrons are fermions and carry spin 12 . They also possess a magnetic moment. The
La

unit of nuclear magnetic moment, in analogy to the Bohr magneton is the nuclear magneton
which is de ned as
e~
N = = 3:15  10 8 eV T 1
2 mP
and is smaller than the Bohr magneton because of the presence of the proton mass mP in the
denominator. The proton magnetic moment is P = 2:793N and the neutron magnetic moment
is n = 1:913N .
6. The nucleons experience two kinds of forces- the positively charged protons experience the normal
electrostatic repulsive force which is a long range force. Neutrons and protons between themselves
also experience another type of force arising out of `strong interactions. These forces between
LABORATORY MANUAL FOR NUCLEAR PHYSICS 81

protons and neutrons are of short range. They are practically negligible at interparticle separations
of a few Fermis. Below such separations, the forces are attractive all the way down to about half
a Fermi beyond which they become repulsive. This complicated nature of the internucelon force
keeps the nucleons together . Between protons they balance out the repulsive coulomb force and
also prevents nucleons from collapsing into a much smaller size object because of the repulsive
nature of the nuclear force at very short distances.
7. The nucleons also experience weak interactions which is also a short range force and which has a
range almost a thousand times smaller than nuclear forces.

s
8. The sum total of the masses of the nucleons in a nucleus is more than the mass of the nucleus.

sic
The balance is called binding energy which can be thought of as the energy required to keep the
nucleus together. The range of binding energies is from a few MeV (as in the case of deuterium)

hy
to more than 1:5 GeV (in the case of an isotope of Bismuth).

rP
9. Some combinations of neutrons and protons form stable nuclei. For light nuclei, generally the
number of protons and neutrons are equal. As we go to heavier nuclei, the number of neutrons
lea
becomes greater since neutrons only experience the short range strong nuclear force and this is
required to balance the electric repulsion of protons. Of the stable nuclei, lightest one of course is
uc
the H-nucleus which is just a proton. As we go up from hydrogen, the light nuclei tend to have
more or less the same number of neutrons and protons
lN

10. Nuclear force is short range and nucleons interact via this force only with their nearest neighbours.
On the other hand, the electric force is present throughout the nucleus and so there comes a point
ua

when the neutrons cannot prevent the break up of the nucleus. This is the limit of stable nuclide
which is 209
83Bi.
an
bM

2.2.1 Alpha Decay


Alpha particles are basically helium nuclei which are emitted by some radionuclides during radioactivity
decay. The process is
La

AX ! A 4Y + 4He
Z Z 2 2
The fundamental reason for alpha particle emission is the fact that some nuclei are too
large to be stable since the short range nuclear force cannot suciently counteract the
electric repulsion. Nuclei which contain more than 210 nucleons are such nuclei and they decay by
emitting alpha particles to reduce their size and thereby increase their stability. The disintegration
energy or the Q factor is basically the mass di erence between the parent nuclei and the sum of masses
of the daughter nucleus and all the other decay products. That is, the Q value is the di erence in the
kinetic energies T of the initial and nal states. As an example, consider a simple nuclear reaction
LABORATORY MANUAL FOR NUCLEAR PHYSICS 82

where a projectile a strikes a target nuclei A and produces two products, b and B . Then the Q value
for the reaction is given by

R Q = [ma + mA (mb + mB )] c2 = T nal Tinitial (2.10)

Since the nucleus, both the parent and the daughter nucleus in the case of alpha decay are so much
heavier than the alpha particle, there is very little recoil of the daughter nucleus. Energy momentum
conservation then gives us the kinetic energy of the alpha particle as

s
A 4
K:E 

sic
Q
A
This should be obvious from Eq 2.10 and the fact that alpha decay entails a nucleus of mass number

hy
A decaying to a daughter nucleus of mass number A 4 and a helium nucleus of mass number 4. The
energies are small enough that non-relativistic momentum conservation can be used though for energy

rP
conservation we clearly need to take into account the mass di erence since that is the only source of
kinetic energy for the decay products. Clearly, since almost all alpha particle emitters have A > 210,
lea
the kinetic energy of the alpha particle is roughly the disintegration energy. Typical alpha particle
energies are in the range of 4 6 MeV. Since the energy of the alpha particle is directly related to the
uc
Q value, the alpha particles are monoenergetic.
lN

Although a heavy, unstable nucleus can become more stable by emitting an alpha particle, the prob-
lem still remains as to how the alpha particle can escape from the nucleus. The strong nuclear forces
ua

dominate at very short distances inside the nucleus and this leads to a potential barrier. If we model
an

this barrier and the alpha particle as a particle in a box, the height of the box (or the potential barrier )
turns out to be around 25 MeV. This is much more than the kinetic energy of the alpha particle. Thus,
bM

classically, there is no possible mechanism for the alpha particle to escape from the nucleus as shown
in Figure 2.1.
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 83

s
sic
hy
rP
Figure 2.1: Nuclear Potential Barrier and particle tunnelling

lea
However, we know that alpha particles do come out and are observed. It was Gamow who rst
proposed the mechanism whereby this could be possible. Essentially, we can think of the alpha particle
uc
tunnelling through the potential barrier, something which is permitted by quantum mechanics. Under
some very reasonable assumptions, the Gamow Theory of Alpha decay agrees remarkably well with
lN

experimental observations regarding the behaviour of the decay constant  with the energy of the alpha
particle. Although a detailed derivation of the relationship of  and the energy of the outgoing alpha
ua

particle is fairly complicated, we can attempt to give a simpli ed derivation of the relationship. This is
done in Appendix E.
an

The higher the energy of the alpha particle emitted, the higher is the decay constant or the shorter is
the half life of the parent nuclide. This is obvious since as we know from elementary quantum mechanics
bM

(for details see Appendix E) that the tunnelling probability is inversely proportional to the energy of
the tunnelling particle. (Actually the relationship between the tunnelling probability and energy is a
bit more complicated. It turns out that the ln T / E 1=2 . See Eq E.18. ) If the tunnelling probability
La

is low, the decay constant will be low (Eq E.21). Typically, when the energy is more than 6:5 MeV,
the half life is of the order of a few days. On the other hand, if the energy is less than 4 MeV, the
probability of the alpha particle tunnelling through the nuclear potential barrier is very small and the
half life is very long. This is shown in Table 2.2.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 84

Source Half-Life Alpha particle Energy (MeV) Branching Ratio


148
Gd 93 years 3.1827 100%
238
U 4:5  109 years 4.196 77 %
238
U 4:5  109 years 4.149 23 %
240
Pu 6:5  103 years 5.168 76 %
240
Pu 6:5  103 years 5.124 24 %
244
Cm 18 years 5.80 76.4 %
244
Cm 18 years 5.76 23.6 %

Table 2.2: Half Life of Alpha Particle Sourcesx

s
x(Adapted from \Radiation Detection & Measurement" by Knoll )

sic
2.2.2 Beta Decay

hy
Another process by which an unstable nucleus can become more stable is by the process of beta decay.

rP
This is

Z Z+1
lea
AX ! AY + e + 

The basic decay in this process is the decay of a neutron in the parent nuclei via
uc
n ! p + e + 
lN

Initially, it was thought that the only particles produced in this reaction were a proton and an
ua

electron. However, it was observed that the beta particles come out with a range of energies from 0
upto a maximum energy called the end point energy instead of having a monoenergetic spectrum.
an

This was surprising since if the only particles produced were the proton and the electron, then energy
momentum conservation tell us that the electron will be monoenergetic and that the electron and the
bM

recoiling nucleus would be moving back to back. This was not the case. Further, the neutron, proton
and electron are all fermions and have spin 21 and hence the reaction with a neutron going to a proton
and electron would not conserve spin. Finally, as we know, the electron carries a `charge' called lepton
La

number. The nucleons have 0 lepton number and thus producing only an electron in the decay would
violate lepton number conservation.

All these anomalies were solved by introduction of a new particle called the neutrino by Pauli.
The neutrino is assumed to be massless, neutral and carries spin 21 and lepton number of 1. Thus the
process that we have is a neutron going to a proton, an electron and an antineutrino (the antiparticle
of the neutrino). With this, all the anomalous observations could be accounted for. Thus, the Q value
of the reaction, that is the mass di erence between the parent nuclei and the daughter nuclei is the end
point energy of the electron. The anti neutrino carries some kinetic energy and the actual energy of
the electron is thus the Q value minus the kinetic energy of the anti neutrino. This being a three body
LABORATORY MANUAL FOR NUCLEAR PHYSICS 85

decay now (that is there are three particles in the nal state), we get a continuous energy spectrum
for the electron upto the maximum of Q value. Again, since there are now three particles in the nal
state, there is no reason for the electron and the recoiling nucleus to be moving back to back since the
antineutrino carries some momentum too. Finally, lepton number conservation and spin conservation
is also taken care of because of the anti neutrino.

n ! p + e + 

s
sic
hy
rP
lea
uc
Figure 2.2: Energy Distribution in Beta decay of 210Bi x

x(Adapted from http://hyperphysics.phy-astr.gsu.edu/hbase/Nuclear/beta2.html )


lN

Each beta decay is characterised by a xed Q value as we have seen above. This is normally the
ua

case when the transition of the nucleus takes place between the ground states of both the parent and
the daughter nuclei. However, if the transition is between excited states and/or ground states, then the
an

spectrum will change because the Q value will be di erent. In practice, in some beta emitters, because
bM

of the presence of excited states which could be populated, one gets several components with di erent
end-point energies.
La

Source Half-Life Endpoint Energy (MeV)


14
C 5730 years 0.156
32
P 14.28 days 1.710
36
Cl 3:08  105 years 0.714
45
Ca 165 days 0.252

Table 2.3: Beta particle Sourcesx

x(Adapted from \Radiation Detection & Measurement" by Knoll )


The accurate theory of Beta Decay was given by Fermi and we will not describe it here. A short
description is given in Appendix F.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 86

Another process which can take place inside the nucleus is positron emission. This is the conversion
of a proton into a neutron and a positron and a neutrino.

p ! n + e+ + 
Obviously, since a proton is lighter than a neutron, this process can only take place inside a nucleus
and not for free protons. On the other hand, a free neutron does decay by emitting a proton and an
electron and antineutrino with a half life of around 10 minutes.

s
sic
hy
rP
lea
uc
Figure 2.3: Energy Distribution in Beta decay of 64Cu x

x(Adapted from "Concepts in Modern Physics" by Beiser, Mahajan & Rai Choudhury )
lN

Positron emission typically occurs in proton rich nuclei and the daughter nuclei has an atomic num-
ua

ber less by one than the parent nuclei. There is another interesting thing about positron emission.
When a proton converts into a neutron and a positron and a neutrino inside the nucleus, the positive
an

charge of the atom decreases by one. To balance this, the daughter atom must get rid of one of its
orbital electrons to maintain neutrality. Thus what we have is that positron emission is only pos-
bM

sible energetically if the parent atom is atleast as heavy as the daughter atom plus two
electron masses. What this means is that isotopes which decrease in mass by less than 2me cannot
spontaneously decay by positron emission. This also means that the Q value for any positron emission
La

emission process is the di erence in the masses of the parent atom and the sum of the masses of the
daughter atom and 2me .

Positron Emission is used extensively nowadays in a very sophisticated imaging process called
Positron Emission Tomography . The idea here is to introduce an isotope, Fluorine-18 into a
compound of glucose ( uorodeoxyglucose). Flourine-18 decays by positron emission . The glucose com-
pound is administered to the patient and the glucose is taken up by the cells. Concentrations of tumour
cells take up more of the glucose than normal cells. Now when the Fluorine isotope decays by positron
emission, the positron annihilates an electron in the body and this pair annihilation gives rise to 2 pho-
tons moving back to back because of energy momentum conservation. These photons detected by using
LABORATORY MANUAL FOR NUCLEAR PHYSICS 87

a scintillator detector and photomultiplier tubes. The image is then reconstructed using sophisticated
algorithms. The tumor cells, because of their high concentration of the positron emitting isotope, give
rise to higher positrons which show up as di erent from healthy tissue.

Finally, another related process is electron capture. This is when a nucleus absorbs an inner shell
electron and a proton and the electron go to a neutron and a neutrino.

p+e !n+
Usually, since the K shell electrons are the ones which are captured, an electron from the outer

s
sic
shell falls to ll the vacancy thereby releasing X-rays which are characteristic of the daughter nuclide.
Electron capture is more likely than positron emission in heavy nuclides since the inner shell electrons

hy
are closer to the nucleus in heavy elements.

rP
2.2.3 Gamma Decay
lea
We are familiar with atoms existing in excited states as well as ground state. When an electron in an
excited state returns to the ground state, a photon is emitted. In essentially the same way, a nucleus
uc
can also exist in ground as well as excited states. When an excited nuclei returns to its ground state, it
emits photons of energies equal to the di erence in the energies of the excited and ground state. We can
lN

easily estimate this energy by using the Uncertainty Principle. A typical nucleus size is x  10 15 m,
while a typical mass for a nucleon would be 1 GeV c 2 . This gives us an estimate of the energy as
ua

 2(~x2)2m  few MeV. Incidentally, a similar calculation for an atom, with x  10 10 m and mass
 1 MeV c 2 for an electron would give us the energy estimate to be a few eV. Clearly, atomic transi-
an

tions occur between energies separated by a few eV and those in the nucleus by a few MeV. Thus, as
bM

we have seen, alpha and beta particles also carry energies in the MeV range.

Electromagnetic radiation which carries a few MeV of energy is said to be in the Gamma ray region
La

of the electromagnetic spectrum.

To study the process of gamma decay ( that is an excited nucleus giving out a gamma ray photon
and making a transition to a lower energy state) we need to use quantum mechanics. A semi-classical
theory of gamma decay using Fermi's Golden Rule is given in Appendix G.
In most cases, some form of beta decay results in the creation of an excited state of the daughter
nuclei. This process happens in a time scale which is characteristic of the half life of beta emitters as
in Table 2.3. However, the excited states of the daughter nuclei thus created are short lived states and
decay to the ground state by emitting gamma rays. Thus what we see typically is that a parent nuclei
emits a beta particle and gamma ray photons with a time of the order of the half life. However, and
LABORATORY MANUAL FOR NUCLEAR PHYSICS 88

this is crucial, the energy of the beta particle is characteristic of the parent nuclei while that of the
gamma rays is determined by the energy levels of the daughter nuclei.

Two typical examples of gamma decay schemes are shown in Fig(2.4) and Fig(2.5).

s
sic
hy
rP
Figure 2.4: Decay Scheme for 60Co- Gamma decayx
x(Source: Wikicommons )
lea
uc
lN
ua
an
bM

Figure 2.5: Decay Scheme for 27Mg - Gamma decayx


x(Adapted from \Concepts of Modern Physics" by Beiser, Mahajan & Rai Choudhury )
La

For instance, in the decay of 27


12Mg shown in Fig (2.5), the half-life of the decay is 9:5 minutes and
13Al. The excited aluminum nucleus, 13Al* can
it can take place to either of the two excited states of 27 27
then decay by emitting one or two gamma rays and come to the ground state.
An excited nucleus can sometimes also give its energy to one of the atomic electrons around it. In that
case, the electron then is emitted with a kinetic energy equal to the nuclear excitation energy minus
the binding energy of the electron. This process, a sort of photoelectric e ect for nuclear photons is
called internal conversion.

Thus we see that certain nuclei are radioactive and emit one or more of the above mentioned par-
ticles/radiation. These radiations and particles are what we use to study the properties of the parent
LABORATORY MANUAL FOR NUCLEAR PHYSICS 89

and daughter nuclei. However, to detect and measure the properties of these radioactive emissions, we
need to have them interact with matter. This is what we shall now turn to.

2.3 References
1. \Radiation Detection & Measurement", Glenn F. Knoll, Wiley India (2009).

2. \ Concepts of Modern Physics", Arthur Beiser, S. Mahajan & S. Rai Choudhury, Mcgraw Hill

s
sic
(2015).

hy
rP
2.4 Questions
lea
1. Why are some nuclei stable and others unstable?
2. What are the various reasons for nuclei to be unstable?
uc
3. What is the activity law in radioactivity? What is the essential assumption in obtain-
lN

ing the empirical activity law?


4. What are the di erent forces operative inside an atomic nucleus? What are their
ua

properties? How do they explain the stable or unstable nature of a nucleus?


an

5. What is Binding energy of a nucleus?


bM

6. What is an alpha particle? Why do nuclei decay by emitting an alpha particle?


7. What are the typical energies of an alpha particle emitted by a radioactive nucleus?
Why?
La

8. The alpha particle emitted by a particular nucleus are mono-energetic. Why is this
the case?
9. How does an alpha particle come out of the nucleus given that there is a potential
barrier for it to come out? What is responsible for the potential barrier?
10. How is the energy of the alpha particle related to the half life of the nuclide? Why?
11. What are beta particles and by what process are they produced?
12. Why does a nucleus decay by emitting beta particles?
LABORATORY MANUAL FOR NUCLEAR PHYSICS 90

13. Given that the neutron in a laboratory decays with a half life of about 10 minutes,
why don't all the neutrons in the nuclei decay?
14. Are beta particles mono-energetic? If not, why not?
15. How do we know that another particle apart from the beta particle must be coming
out in beta decay?
16. Why does positron emission happen in some nuclei and not in others?
17. What is electron capture? What is the characteristic signal for electron capture?

s
18. Why is electron capture more likely in heavier atoms than in lighter atoms?

sic
19. What are gamma rays and what is the process by which they are produced?

hy
rP
lea
uc
lN
ua
an
bM
La
Chapter 3

INTERACTION WITH MATTER

s
sic
hy
Learning Objectives

rP
1. To understand the concept of cross section.
lea
2. To study the interaction of a heavy charged particle with matter.
uc
3. To derive the formula for energy loss by a heavy charged particle in its interaction
with matter.
lN

4. To study the interaction of electrons with matter.


ua

5. To derive the energy loss formula for electrons moving through matter.
an

6. To study the di erent ways in which radiation interacts with matter.


bM

3.1 Introduction
La

The experiments that we perform in this laboratory are all concerned with the detection and measure-
ment of radiation and particles, namely gamma rays and beta particles (We do not carry out experiments
with alpha particle emitters in this laboratory). Clearly, we need detectors for this purpose- in our labo-
ratory, we use two kinds of detectors. Geiger-Muller counters (GM counters) and Scintillation counters.
We shall be studying in detail about the workings of these detectors in a later Chapter. In this Chapter,
we would like to understand in general how particles and radiation interact with matter. After all, if
radiation or a particle is to be detected, it must interact with the material of the detector. This might
be a gas, a liquid or even a solid. Before we do this, let us de ne some terms which we shall be using.

91
LABORATORY MANUAL FOR NUCLEAR PHYSICS 92

3.1.1 Cross Section


Cross section is a way to express the probability of interaction. Consider a beam of incident particles
impinging on a target material. We assume that each particle in the target material has a certain area,
which we call cross section. Any incident particle which passes within this area will interact with the
target particle. Clearly, the larger the area, the larger is the possibility of interaction. Of course, this
cross section which we think of as an area of in uence in a way, depends on the nature of the process,
the nature and energy of the incident particle etc. In principle, it could be very di erent from the
geometric cross section.

s
sic
We know that ux of a beam is de ned as the number of particles crossing a unit area in unit time.
Suppose now we have a slab of some target material with area A and thickness x. Let the number of

hy
atoms per unit volume of the target be n, that is the total number of nuclei or atoms in the slab are
nAx. If each nuclei has a cross section of  , then the aggregate cross section for the slab is nAx.

rP
Now consider a beam of incident particles, N of them hitting the slab. The number N which will
interact with the nuclei in the slab will be
N
=
lea
Aggregate cross section nAx
= = nx
N Target area A
uc
since one incident particle is assumed to interact only once with any nuclei, and get de ected, it is
lN

removed from the beam. Then for a nite slab thickness, we have
ua

dN
= ndx
N
an

ZN Zx
dN
= n dx
N
bM

N0 0
N = N0 e nx (3.1)

We see that the number of surviving particles decreases exponentially with the slab thickness. Recall
La

that this is exactly what we see when we pass a beam of light through some absorber. The intensity
falls exponentially with thickness.

We can think of the cross section as in Figure (3.1) where an incident beam of ux F hitting a target.
The di erential cross section is dd
(; ).
LABORATORY MANUAL FOR NUCLEAR PHYSICS 93

s
sic
Figure 3.1: Di erential Cross Sectionx

hy
x(Source: "ScatteringDiagram" by Original uploader was JabberWok at en.wikipedia - Transfered from en.wikipedia.
Licensed under CC BY-SA 3.0 via Wikimedia Commons

rP
- https://commons.wikimedia.org/wiki/File:ScatteringDiagram.svg#/media/File:ScatteringDiagram.svg )

Then, lea
uc
d 1 dNs
(; ) =
d
F d
 
lN

d
dNs = F d
(; ) (3.2)
d

ua

We should think of this equation as the number of particles that scatter, Ns , into a portion of solid
angle per unit time is equal to the ux of incident particles per unit area per unit time multiplied by
an

the probability (represented by a cross section area) that would scatter into that portion of solid angle.
We can integrate Eq(3.2) over all solid angles to get the total cross section . The usual unit for cross
bM

section is a barn which is de ned as


La

1 barn = 10 28 m2
Please remember that this discussion of cross section, though in the context of scattering, is equally
applicable for absorption of incident particles or radiation by matter.

If Nt is the number density of the target particles, then the probability that a single interaction occurs
through a volume with thickness x is simply Nt x. A convenient parameter in this discussion is the
mass thickness. This quantity is often more convenient to use instead of thickness. Thus
LABORATORY MANUAL FOR NUCLEAR PHYSICS 94

R mass thickness = x gm cm 2 (3.3)

This is convenient because it tells us directly the e ect of the absorber on the incident beam. Thus,
for instance, a beam travelling through 2 gm cm 2 of air (of density  = 0:012 gm cm 3 ) has the same
e ect as the beam passing through 2 gm cm 2 of water, even though it passes through 1:67 m of air
and just 2 cm of water. Essentially, what we have done is to factor out the density of the absorber and
encapsulate that in the mass thickness.

s
sic
We are interested in the passage of alpha, beta and gamma radiation through matter. Thus we can
divide our discussion into two parts- interaction of charged particles with matter and interaction of

hy
radiation with matter. For charged particles, once again, we need to consider two cases of passage of
heavy particles (like alpha particles) and the passage of electrons or beta particles.

rP
3.2 Interaction of Charged Particles with Matter
lea
uc
When a charged particle interacts with matter, two things happen-
a) the particle loses energy traversing matter and
lN

b) particle is de ected from its initial direction.


ua

In general, there are two main processes which cause this. These are inelastic collisions with atomic
electrons in the material and also elastic scattering o the nuclei. There are some other processes which
an

contribute to the energy loss namely, Bremsstrahlung and nuclear reactions which are extremely rare.
bM

As mentioned above, the processes which are dominant causes of energy loss are di erent for light and
heavy charged particles. For heavy charged particles it is basically the inelastic collision with atomic
electrons which are responsible for energy loss.
La

3.2.1 Interaction of Heavy charged particle with matter


Consider a charged particle like an alpha particle entering a medium. The atoms of the absorbing
medium will have atomic electrons which will interact with the incoming charged particle. It is im-
portant to remember that the alpha particle interacts simultaneously with many atomic electrons. In
this process, some energy is transferred to the atomic electron and depending on the amount of energy
and the nature of the absorber, the atomic electron either gets into an excited state or in some cases,
may even become free, that is, the atom gets ionised. This energy transferred is of course exactly the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 95

energy that is lost by the incoming particle. However, as we shall see below, the maximum amount of
energy that is transferred in such a collision is very small compared to the kinetic energy of the incom-
ing particle. This means that the incoming particle continuously loses small amounts of energy as it
passes through the absorber. However, the number of such encounters in any macroscopic length is large.

Let us consider one such encounter in some detail. Consider a particle of mass M and velocity V
incident on another particle at rest of mass m, M  m. This could be the case, for instance of an alpha
particle colliding with an electron. Classically, using non-relativistic energy momentum considerations,
we know that

s
sic
1 1 1
MV 2 = MV12 + mv22
2 2 2

hy
MV = MV1 + mv2
where V is the initial velocity of the incoming particle, V1 is the velocity of the alpha particle after

rP
the collision and v2 is the velocity of the electron after the collision. Solving these two equations for the
velocity after collision of the incoming particle, V1 , we get
lea
M m
V1 = V
uc
M +m
Thus, the loss of energy of the incoming particle is
lN

1   4MmE
E = M V 2 V12 =
2 (M + m)2
ua

where E is the initial kinetic energy of the incoming particle. Thus, we see that in the case when
M  m, we get that the maximum energy transferred to the electron is 4Mm E . This justi es our state-
an

ment above that in the case of alpha particles (or protons) and electrons, each collision only diminishes
bM

the incoming particle's energy by a small amount, in this case about 2000 1 of the initial energy. It also
implies that the electron's velocity after the collision can be at most 2V since the electron was at rest
initially and all this energy E is the electron's kinetic energy after collision.
La

Given that in any one collision, the energy transferred is small, we can also see that the momentum
transferred in any one collision, p  m(2V )  2m MP which is again small since M  m where P is
the initial momentum of the incoming heavy particle. Thus we can assume that the incoming particle
does not get de ected in any one collision and continues along the unde ected.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 96

s
sic
Figure 3.2: Heavy particle in matter

hy
Consider a heavy particle of mass M and charge ze entering a cylinder of absorber material with a

rP
velocity v along the x direction. Consider a cylinder of radius b and length dx placed along the x axis as
shown in Fig 3.2. Let us look at the interaction of the heavy particle with a single atomic electron at a
lea
distance b from the path of the incoming particle. This interaction is the Coulomb interaction between
the heavy particle and the electron and so we need to nd out the eld at the surface of the cylinder,
uc
which is the location of the electron. The eld is easily computed using Gauss's Law as the symmetry
considerations tell us that the x component vanishes.
lN

Z
ze
Ey dA =
ua

Z
0
ze
Ey 2b dx =
an

Z
0
2ze
Ey dx =
bM

40 b
k2ze
= (3.4)
b
La

where k = 41 0 .

Consider the impulse produced on the electron as a result of this Coulomb force.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 97

Z
I = Fy d t
Z
= e Ey dt
Z
dx
= e Ey
v
2ze2 k
= (3.5)
vb

s
using Eq(3.4).

sic
Now the energy gained by the electron (which is at a distance b from the incoming particle) is simply

hy
I2

rP
E (b) =
2 me
4z 2 e4 k2
lea
=
2v2 b2 me
2z 2 e 4 k 2
= 2 2 (3.6)
uc
b v me
lN

This is the energy loss to a single electron. We need to nd out the energy loss of the incident
particle when it travels a distance dx in the absorber material. For this we need to add the energy lost
to all the electrons in the annular thickness of the cylinder between b and b + db. If Ne is the electron
ua

density in the material, then this energy loss is given by


an
bM

dE (b) = E (b)Ne dV
= E (b)Ne 2b db dx
2z 2 e4 k2
= 2 2 Ne 2b db dx
La

b v me
4z 2 e4 k2 db
= N dx (3.7)
me v2 e b
To nd the total loss, we need to integrate over values of b and thus we get
 
dE 4z 2 e4 k2 b
= 2 Ne ln max (3.8)
dx me v bmin
Clearly, the values of bmax and bmin will be determined by physical considerations. These are eas-
ily obtained. Consider the maximum energy lost by the incoming particle (or the maximum energy
2
transferred to the electron). This, we have already seen is given by me (22 v) . This will happen when the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 98

impact parameter is the minimum. Thus we have, using Eq(3.6),


2z 2 e4 k2
Tmax = 2me v2 = 2 2 (3.9)
bmin v me
where Tmax is the maximum energy that can be transferred.
kze2
bmin =
me v2
Now, for the maximum impact parameter, bmax , clearly, if the distance is very large and the energy
transfer is smaller than the ionization or excitation energy Ie , then no energy is transferred. Thus,

s
again using Eq(3.6), we get

sic
hy
2z 2 e 4 k 2
Ie =
b2max v2 me

rP
(3.10)

Thus we have for the energy loss lea


R
uc
   
dE 4k2 z 2 e4 Tmax 4k2 z 2 e4 2me v2
= N ln = N ln (3.11)
dx me v 2 e Ie me v2 e Ie
lN

since as we have seen Tmax = 2me v2 . This is the classical formula for the energy loss of
ua

a charged particle in an absorbing material as given by Bohr. Of course this formula


assumes that classical, non-relativistic, particle must be heavy compared to me , that the
an

interaction time is short thus the electrons are stationary and also does not account for
binding of atomic electrons. We can nd the number density Ne of the electrons in the material as
bM

ZNA
Ne =
AmN
La

where  is the density of the absorber material with atomic weight A and atomic number Z and NA
is the Avogadro's number.

The classical formula of Bohr was extended by Bethe and Bloch to include relativity and other e ects
like density e ects and the e ect of atomic electrons etc. When one uses relativity to nd the maximum
energy transferred (and hence bmin ) by a particle with mass M interacting with a particle of mass m,
we get instead of Eq(3.9),
2m 2 2
Tmax = m + m 2
(3.12)
1 + 2 M M
LABORATORY MANUAL FOR NUCLEAR PHYSICS 99

where = p11 2 and = vc .

The basic result is the Bethe-Bloch formula

R dE
dx
=
4k2 e4 z 2 ZNA
me c2 2 A
B (v) (3.13)

where  
2 me v 2
B (v) = ln ln(1 2 ) 2 (3.14)
Ie

s
sic
Clearly, as we have seen, for non-relativistic particles (that is  1), only the rst term in the factor
B (v) is signi cant .

hy
This is the basic formula to estimate the energy loss by a charged particle in matter. We can rewrite

rP
the Bethe-Block formula in another way. Remember that the classical electron radius re is de ned as
lea ke2
re =
me c 2
uc
In terms of re , we can rewrite Eq(3.13) as

R
lN

1 dE z2 Z
= 4re2 me c2 NA 2 B (v) (3.15)
 dx A
ua

The quantity Ie in the Bethe-Bloch formula is the mean ionisation energy of the material. This is
an

usually determined empirically though one can nd it out in principle by taking the average over all the
ionisation and excitation processes in the atom. Experimentally, it is usually found to be Ie  10Z eV.
bM

Thus for instance, in air, Ie  85 eV while in aluminum, it is Ie  160 eV.

Let us see what this equation for the energy loss by a charged particle in matter tells us in general.
La

Note that most materials have the same AZ  12 except for hydrogen for which it is 1. Thus we can
2
immediately say that apart from the corrections due to B (v), the energy loss depends on z 2 since all
the other factors are constants. That is, for a given non-relativistic particle, the energy loss
varies inversely with particle energy since it varies inversely with v2 . This can be easily
understood since if the particle has higher energy, it spends less time near any one electron
and hence the impulse on the electron is less and the energy transfer is lower. Slow mov-
ing particles lose more energy and as their momentum increases and thus their velocity
approaches c, we expect a attening of the dE dx curve. We can see this in the energy loss
curves for various charged particles as shown in Fig 3.4. The second point is the variation with
z 2 . Energy loss depends directly on z 2 . Thus alpha particles with the greatest charge will have
LABORATORY MANUAL FOR NUCLEAR PHYSICS 100

the highest energy transfer (among particles with the same energy or velocity). Finally,
when we compare di erent absorbers, the quantity which is entering the energy loss is the
product NZ . This obviously is a measure of the electron density in the absorber. Hence,
materials with high NZ or higher electron density will have a higher stopping power or
energy loss for a given beam.

Another interesting thingabout the Bethe-Bloch formula is the logarithmic term. The rst term in
2
the factor B (v) is ln 2mIeev which shows that there is a slow rise in the energy loss with increased
momentum or velocity. The reason for this is actually the behaviour of a relativistic particle's electric

s
eld as the velocity increases. We know that the electric eld of the incoming charged particle becomes

sic
more and more squashed as the velocity increases. The reason for this is that the transverse eld, Ey , is
larger by Ey relative to the isotropic case of the charge at rest while the longitudinal eld Ex decreases

hy
by a factor of 12 relative to the isotropic case as shown in Figure 3.3. The typical whiskbroom pattern

rP
of the electric eld of a moving charge is shown in the Figure 3.3. This leads to a slow increase in
ionization at farther and farther distances from the particle track.
lea
uc
lN
ua
an

Figure 3.3: Electric eld of moving charge x

x(Source: http:// sernam.ru/ lect f phis6.php? id=65 )


bM

From Figure 3.4, we see that the typical speci c energy loss at the minimum ( for most particles
except the electron) is around 2 MeV cm2 gm 1 . Since most relativistic particles have very similar
behaviour (in terms of energy loss as a function of velocity or energy), we refer to these relativistic
La

particles ( > 3) as minimum ionizing particles.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 101

s
sic
Figure 3.4: Speci c energy loss in air versus energy of charged particlesx

hy
x(Source: Review of Modern Physics, Vol 24, page 273 )

rP
The Bethe-Bloch formula is a good guide to understand the energy loss of charged particles in matter.
However, there is another factor which comes into play as the charged particle moves through matter.
This is reduction in the e ective charge. To understand this, consider Figure 3.5.
lea
uc
lN
ua
an
bM

Figure 3.5: Energy loss of Alpha particles of energy 5.49 MeVx


La

x(Source: Wikicommons )

This kind of plot of the energy loss of a particle in a medium is called a Bragg curve . There
are several things about this gure that one needs to understand. Firstly, as the alpha particles move
through matter, we expect from the Bethe-Bloch formula that the energy loss increases as E1 . We see
this in the initial part of the plot. However, as the particle keeps moving in the material, its velocity
and hence energy decreases and the energy loss increases to a maximum near the end of the track.
Just before the particle comes to a complete stop, the energy loss reaches a maximum which is called
a Bragg peak . At that point, we see that the energy loss sharply falls o . This is because the alpha
particles are now travelling slow enough that they can pick up electrons from the material and thus the
e ective charge is reduced. We expect that charged particles with the maximum nuclear charge will
LABORATORY MANUAL FOR NUCLEAR PHYSICS 102

pick up the electrons earlier than those with less nuclear charge. This is indeed seen as we can see in
Figure 3.6.

s
sic
hy
Figure 3.6: Energy loss for helium & hydrogen ions.x

rP
x(Source: B.Wilken, T.A. Fritz, Nuclear Instrumentation Methods, 138, pp 331 (1976))

lea
Incidentally, the energy loss curve that we have discussed is for charged particles only. For photons,
for instance rays or light, the fall in intensity as we know is exponential. The energy loss curve is of
uc
fundamental importance in medicine also. Proton beams of a xed energy from a linear accelerator are
used to treat certain kinds of tumours. As we have seen, the maximum energy loss will occur at the end
lN

of the proton trajectory (the Bragg peak) and it is a sharp maximum. To increase the target tumour
volume for the protons, the monoenergetic proton beam is widened to a spectrum of energies which
ua

lead to a broadening of the Bragg peak thereby increasing the e ect of the treatment on the tumour as
shown in Figure 3.7
an
bM
La

Figure 3.7: Radiation dose produced by a monoenergetic proton beam and a modi ed beam.x
x(Source:CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=469545 )

Finally, we consider the phenomena of straggling. When we pass a beam of charged particles of a
xed energy (instead of a single particle) through di erent materials and study the number transmitted,
LABORATORY MANUAL FOR NUCLEAR PHYSICS 103

we nd an interesting feature. We see that energy loss is not continuous but statistical or stochastic
in nature and some particles undergo less/more energy loss and their range will be larger/smaller than
the typical, expected range. This is known as straggling.

s
sic
hy
rP
Figure 3.8: Energy Stragglingx
x(Source: `DEVELOPMENT And IMPLEMENTAZIONE IN C++ OF ALGORITHMS FOR The CALCULATION OF PLANS
lea
OF TREATMENT IN ADROTERAPIA' Giovanni Nicco, Bachelor's Thesis, University of Turin, 2000. )
uc
As we can see from Figure 3.8, the variation in the amount of collisions/energy loss is approximately
Gaussian (because of the central limit theorem). We de ne the mean range as the distance at which
lN

50% of the particles are transmitted, as shown in Figure 3.8. We can nd the range of transmission by
extrapolation to the distance at which we expect zero transmission as in the Figure.
ua

The discussion above is strictly valid for heavy charged particles and hence will be a good description
an

for the energy loss for a beam of alpha particles. However, things are di erent for beta particles.
bM

3.2.2 Interaction with matter of electrons


La

In our discussion of the Bethe-Bloch formula, we had assumed that the incoming charged particle is
much heavier than the atomic electron in the absorber and therefore does not su er any signi cant de-
viation from its straight line path. However, when one is interested in the interaction of beta particles
or electrons with matter, this assumption clearly is invalid. Here, the beam and the target particles are
identical and hence large deviations can be expected from the beam path. In addition, the interaction
between the beam and the nucleus of the absorber can lead to sudden changes in the path. The target
and the beam particles are identical and indistinguishable and this needs to be taken into account.
Finally, at the energies that we are interested in, namely nuclear energies (remember the beta particles
are in the MeV range), the electrons are always relativistic.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 104

The above di erences relate to the considerations of energy loss by collisions with atomic electrons
as in the case of alpha particles. However, since the electrons are light and therefore su er substantial
acceleration (or deceleration) because of their interaction with the absorber, they emit radiation as we
know from the well known Larmor formula in classical electrodynamics. The emission of radiation be-
cause of de ections (and hence acceleration) leads to energy loss. These are of two kinds: Cherenkov
radiation and Bremsstrahlung. To nd out the energy loss of an electron beam in matter, we need
to take into account all these processes.

To nd out the energy loss, we need to modify the Bethe-Bloch formula to take into account all the

s
above mentioned di erences between a heavy particle interaction and electron interaction. Recall the

sic
Bethe-Bloch formula Eq (3.13) which we had for heavy particle interaction with matter,

hy
dE 4k2 e4 z 2 ZNA
= B (v )
dx me c 2 2 A

rP
with
 
2 me v 2lea
B (v) = ln ln(1 2 ) 2
Ie
uc
This gets modi ed in the case of electrons to
lN

dE 4k2 e4 z 2 ZNA
dx
=
m e c 2 2 A B (v )
ua

 2 
m v2E p 
1 p
B (v) = ln 2 e (ln 2)(2 1 2 1 + 2 ) + (1 2) + 1 1 2 (3.16)
2I (1 2 ) 8
an

This is a much more complicated formula for energy loss due to collisions of electrons in a beam
bM

with atomic electrons in an absorbing material. Another interesting di erence is in the straggling
for electrons as compared with heavy charged particles. Since the electron mass is small, there is a
signi cant fractional energy loss in each collision than for heavier particles. Hence we see a lot more
La

straggling for electrons, Figure 3.9.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 105

s
sic
Figure 3.9: Energy Straggling for particlesx

hy
x(Source: www.physics.queensu.ca/ phys352/ )

rP
Apart from the collisions su ered by the beta particles, they also radiate and hence lose energy
because of the two processes mentioned above. Cherenkov radiation is the electromagnetic shock wave
generated by a particle in a medium when it moves faster than the velocity of light in that medium.
lea
That is when the velocity of the particle, v > nc where n is the refractive index of the material. This radi-
ation, which is the familiar blue radiation that one sees in pictures of nuclear reactor's water shield, goes
uc
as 12 and thus peaks at small wavelengths or in the ultraviolet. In the case of electrons in a medium, the
lN

loss due to Cherenkov radiation is small, around 1% and so we will ignore this. As it turns out, in the full
Bethe-Bloch formula for energy loss, this radiative loss due to Cherenkov radiation is already accounted.
ua

The main radiative loss for an electron is due to Bremsstrahlung. This braking radiation is what
an

the electron emits when it is accelerated due to the strong nuclear electric eld, though electron-electron
Bremsstrahlung is also possible. One can use the classical theory of radiation by an accelerated charge
bM

(Larmor's formula) to calculate the radiative loss due to this process.

The treatment of the radiation loss in any collision process is easy to analyse in the case when the
La

particle is non-relativistic. This may be the case of an alpha particle, an ion or an electron. When
the particle passes near a target atom, it experiences an acceleration due to the eld of the target.
We can think that the particle does not experience this eld when it is far away from the target and
only when it is close to it. In e ect, it receives an impulse or experiences a force for a short time.
(remember that impulse is the product of force and time that it acts). The time scale of this impulse, 
can be estimated to be   vb0 where b is the impact parameter and v0 is the incoming particle velocity.
Of course the direction of the force and hence the impulse will be determined by various factors but
one can assume for simplicity that on the average it is perpendicular to the initial velocity of the particle.

Now we know that an accelerated charge radiates energy and the power radiated from it in the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 106

non-relativistic case is given by Larmor's formula


q1 2a2
P= (3.17)
40 3c3
where q1 is the charge of the incoming particle, and a is its acceleration. To determine the accelera-
tion, we can simply use the electrostatic force acting between the target atom and the incoming particle
as
F q1 q2
a= = (3.18)
m1 40 m1 b2

s
where q2 is the charge of the target atom and m1 is the mass of the incoming particle and b is the

sic
impact parameter. This allows us to get an estimate of the radiated energy since we know the time for
which this impulse is acting and therefore the charge is radiating , that is  . This gives us

hy
q1 2a2 b q14 q22 2

rP
W  P   4 = 2 (3.19)
0 3c v0 (40 ) 3m1 v0 b3 c3
3 3
Clearly, this is the energy radiated by the incoming charge in one collision with impact parameter
lea
b. To get the total energy radiated per unit length of the target, we need to integrate over the whole
range of impact parameters and multiply by the number density of target atoms, n2 .
uc
lN

bZmax
dW q14 q22 2
= n2 2 2b db (3.20)
dl (40 ) 3m1 v0 b3 c3
3
ua

bmin
 bmax
q4q2 4 1
= n2 1 2 3 2 3
an

(40 ) 3m1 v0 c b bmin

Here it is important to point out the di erence from the case of energy loss by collision that we
bM

discussed in the alpha particle case, that is Bohr's formula for the energy loss, Eq.3.11 where we could
not take the maximum limit of b, that is bmax to be 1 because of the lograithmic dependence. In this
case, bmax can be taken to be in nity. On the other hand, for bmin we need to be careful and cannot
La

take it to 0. Instead, we use the minimum value of b to be where the wave nature of the incoming
particle becomes important, that is the de-Broglie wavelength of the incoming particle. Thus we take
~
bmin =
m1 v0
With this approximation, and taking bmax = 1, we get the energy lost per unit length by radiation
as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 107

R dW
dl
q 4 q 2 4 1
= n2 1 2 3 3
(40 ) 3c m1 ~
(3.21)

This non-relativistic expression allows us to say several things about radiation loss due to Bremsstrahlung
in collisions. First, notice that the incoming particle velocity cancels out but the mass of the incoming
particle enters in the denominator. This means that the loss for electrons is much more than that for
alpha particle or ions. Secondly, there is however the fact that if the incoming alpha particle collides
with a free electron, the free electron will be accelerated and radiate energy. The energy loss will be

s
exactly the same as in Eq 3.21 except that q1 and q2 will be interchanged and m1 will be

sic
replaced by m2 .

hy
From Eq 3.21 it is also obvious that targets of heavy nuclei, that is with high Z will cause a higher
radiative loss because q2 will then be Ze and the loss is / Z 2 e2 . Also, one might think that in case of

rP
incoming electrons interacting with the target electrons, we would get radiation from both the incoming
and target electrons. It turns out electron-electron collisions do not produce signi cant bremsstrahlung
lea
because both the incoming and target electrons experience equal accelerations in opposite directions
and thus produce radiation which is out of phase and thus interferes destructively to cancel each other.
uc
This is the case for non-relativistic electrons. In the case of relativistic electrons, this is not the case
lN

and one needs to take this into account. Also note that for the case of electron-positron collisions, this
cancellation does not occur and therefore even for the non-relativistic case, we need to take it into
account.
ua
an

We can write Eq 3.21 more instructively in terms of constants like the ne structure constant =
e2 2
and the classical electron radius re = 40em2 c2 .
40 ~c
bM

dW 4
= n2 Z 2 me c2 re2 (3.22)
dl 3
The important question that we need to address is about the relative importance of the energy lost
La

due to radiation versus the collisional energy loss that is given by the Bethe-Bloch formula. If we
take non-relativistic particles, we need to compare the loss due to bremsstrahlung to the loss due to
collisions. We know that the loss due to collisions is given by Eq 3.11
 
dE 4k2 z 2 e4 2me v2
= N ln (3.23)
dx me v2 e Ie
In this expression, the dimensionless argument of the logarithm can be taken as . Then we can
write Eq 3.23 as
dE 4k2 z 2 e4
= N ln  (3.24)
dx me v2 e
LABORATORY MANUAL FOR NUCLEAR PHYSICS 108

To nd the relative importance of these two energy losses, we need to take the ratio of the energy
loss per unit length due to the two processes. Thus

dW me v02 1
= z 2 Za (3.25)
dE m1 c2 3 ln 
where Za is the atomic number of the target nucleus, and therefore Ne = n2 . m1 is the mass of the
incoming particle and is the ne structure constant.

Eq 3.25 allows us to see that for non-relativistic case, bremsstrahlung is never important. In the
best case of electrons interacting with the heaviest nuclei, Za = 92, the factor Za  0:67. However,

s
sic
1 and v202 are always less than one. In the case of heavy particles like alpha particles impinging on a
3ln c
target, the mass of the incoming particle m1 in the denominator makes the ratio very small and thus

hy
we can conclude that the energy loss due to radiation for alpha particles is very small. We might think
that the heavy alpha particles could accelerate the atomic electrons which could radiate. However,

rP
note that this loss can never exceed the total energy transferred by the alpha particle to the electrons
since that energy transfer is shared by collissional loss and bremsstrahlung. This can also be seen if we
lea
take the energy loss by bremsstrahlung expression (Eq 3.21) and change m1 and m2 and q1 and q2 and
take the ratio of this energy loss and the collisional energy loss. We see that the denominator in the
uc
equation corresponding to Eq 3.25 will be me and thus will cancel the me in the numerator. Thus we
can conclude that in the non-relativistic case, energy loss by radiation of a charged particle
lN

passing through matter is negligible compared to the loss due to collisions.


This situation changes in the relativistic case. The calculation for the energy loss in the relativistic
ua

case is quite complicated. It turns out that the energy loss for the relativistic case is given by
an

   
dE NEZ 2 e4 2E 4
= 2 4 4 ln 2 (3.26)
dx r mc mc 3
bM

Clearly then, the total energy loss is the sum of the energy loss due to collisions (Eq(3.16)) and that
due to Bremsstrahlung (Eq (3.26)). That is
La

   
dE dE dE
= + (3.27)
dx dx c dx r
If we compare the two expressions for energy loss of beta particles, Eq(3.16) and Eq(3.26), we notice
some important features. Firstly, the collisional energy loss increases logrithmically with the energy E
and linearly with the atomic number of the absorber, Z while the energy loss due to radiation increases
linearly with the energy E and as the square of the atomic number Z . The radiative energy loss also
has a mass factor in the denominator which implies that the radiative energy losses are most for lighter
particles (like beta particles) than for heavier particle (like alpha particles). Also radiative energy losses
are most for high energy particles and in materials of high Z .
LABORATORY MANUAL FOR NUCLEAR PHYSICS 109

This also allows us to de ne a critical energy, Ecrit for which


   
dE dE
=
dx c dx r
An approximate formula for Ecrit given by Bethe & Heitler is
1600mc2
Ecrit 
Z
The value of the critical energy for various materials is given in Table 3.1.

s
Material Ecrit (MeV)

sic
Cu 24.8
Pb 9.51

hy
air (STP) 102
plastic 100

rP
water 92

Table 3.1: Critical energy for various materials


lea
As we have discussed above, the electron being of the same mass as the scattering particles (atomic
uc
electrons), can su er large deviations from its original path. Thus, if one has a beam of mono-energetic
lN

electrons from a source and we pass it through an absorber, even a thin absorber can lead to a loss of
electrons from the detector. This scattering therefore means that the intensity of the transmitted beam
drops immediately and goes to zero as the absorber thickness is increased.
ua
an

A corollary to this random motion of the electron is therefore that the range of the electron is hard
to de ne since it might have travelled a much larger distance within the absorber than simply the
bM

thickness of the absorber. What is normally done is to extrapolate the range from the linear portion of
the graph.
La

To summarise then, we can say that the energy loss of electrons is much smaller than that of heavy
charged particles of the same energy. This means that they have a much larger range. What is observed
experimentally is that for a wide variety of absorber materials, the product of the range and the density
of the absorber is a constant for any particular electron energy.

The situation is very di erent for the beta particles emitted by a radioactive source. This is because,
as we have seen in Section 2.2.2 in Fig 2.2, the energy spectrum of the beta particles is continuous.
What is seen therefore is that the beta particles at the lower end of the spectrum are absorbed even
with a very thin absorber. However, for the most part of the spectrum, the transmission of the beta
particles shows an exponential decrease with thickness. This is an experimental fact which cannot be
LABORATORY MANUAL FOR NUCLEAR PHYSICS 110

derived easily from fundamental physics. What we see is that the counting rate (or intensity) falls
o exponentially with an attenuation coecient which depends on the end point energy of the beta
particle.

C = C0 e nd (3.28)
where C is the counting rate with the absorber material, C0 is the counting rate without the absorber
and d is the mass thickness in units of mass per unit area. The coecient n is the attenuation coecient.
This behaviour is shown in Fig 3.10.

s
sic
hy
rP
lea
uc
lN

Figure 3.10: Range-Energy for 1:17 MeV beta particles from 210Bi in Al absorbersx
x(Source: `Radiation Protection', course at Oregon State University Extended Campus )
ua

3.2.3 Interaction of gamma rays with matter


an

The interaction of electromagnetic radiation like gamma rays with matter is fundamentally di erent
than what we have studied so far. The interaction in this case is between radiation and the charged
bM

particles like atomic electrons unlike between two charged particles that we have been investigating
so far. The interaction of electromagnetic radiation with matter depends on its frequency (and hence
energy). Gamma rays, as we have seen in Section 2.2.3 have typical energies in the MeV range since their
La

origin is in the de-excitation of the nucleus. In general, the interaction of electromagnetic radiation with
matter can be of four kinds depending on the energy of the radiation. These are Rayleigh Scattering,
Photoelectric Absorption, Compton Scattering and Pair Production. The relevant photon
energies for which these processes are operative are given in Table 3.2 ( EI is the ionization energy of
the atom).
LABORATORY MANUAL FOR NUCLEAR PHYSICS 111

Rayleigh Scattering Photoelectric Absorption Compton Scattering Pair Production


h < EI h > EI h  me c2 h > 2me c2
 eV  keV  MeV  MeV
Visible X-rays rays Hard rays

Table 3.2: Interaction of Photons with matter

The fact that the three processes listed above are dominant in di erent energy regimes can be clearly
seen if we plot the variation of the cross section for the three processes with the gamma ray energy.

s
sic
This is shown in Fig 3.11.

hy
rP
lea
uc
lN
ua

Figure 3.11: Photon interaction cross sections for uranium as a function of photon energy. Solid line, PE ab-
sorption; dashed line, Compton scatter; dash-dotted line, pair production; dotted line, total attenuationx
an

x(http://rsif.royalsocietypublishing.org/content/7/45/603 )
bM

To understand these processes of course one needs to develop a fully relativistic theory of the interac-
tion of photons and electrons, which is known as quantum electrodynamics. However, for our purposes,
a simple classical description is sucient to provide some insights into the mechanisms. We think of
La

the atom as a dipole with the electron in the atom attached to the nucleus by a spring. The electron
is oscillating with a frequency !0 around its mean position. This `natural' frequency of oscillation can
be modelled by a linear restoring force kx on the electron, with the `spring constant' k and !0 being
related as
k
!02 =
me
Of course, k is related to the attractive electric force between the electron and the nucleus and is
related to the binding energy of the electron. Now when the photon or electromagnetic wave is incident
upon such an oscillating dipole, the electric eld of the wave exerts an added force on the electron,
LABORATORY MANUAL FOR NUCLEAR PHYSICS 112

eE (t) where E (t) is the oscillating electric eld of the electromagnetic wave given by E (t) = E0 sin !t
where ! is the frequency of electromagnetic wave. The equation of motion for the electron is then

me x = kx eE (t)
e
x + !02 x = E (t) (3.29)
me
The harmonic solutions to this equation of a forced oscillator are well known as x(t) = A sin !t which
we can substitute in Eq(3.29) and get

s
sic
e
(!02 !2 )A = E
me 0

hy
1 e
A = 2 E
! !0 me 0
2

rP
1 e
x(t) = 2 E sin !t (3.30)
! !0 me 0
2
lea
We know from classical electromagnetic theory that an accelerated charge radiates energy. The
power radiated is given by Larmor's formula as
uc
2 e2 2
P= a
lN

3 c3
where a is the acceleration. In our case, the time averaged acceleration squared is simply (the only
ua

time dependent quantity is


sin!t which time averages (over a period) to 21 .
an

 2

2 !2 e 1
a = E0
bM

!2 2
!0 me 2
which gives us the radiated power as

1 e2 2
 
!4
La

P= 2 2 2 2 cE02 (3.31)
3 me c (! !0 )
Now we can think of the power radiated

P = I
where  is the cross section of the interaction and I is the intensity of the incoming radiation. We
know that
cE02
I=
8
LABORATORY MANUAL FOR NUCLEAR PHYSICS 113

2
since I = uc with u = E20 being the energy density of the radiation. Thus we get
cE 2
P = 0
2
or

8 e2 2
 
!4
= (3.32)
3 me c2 (!2 !02 )2
This classical cross section can be expressed in terms of the classical radius of the electron re = e2
me c2
as

s
sic
R 
!2
2

hy
2 2
 = 4re (3.33)
3 (!2 !02 )

rP
This is the expression which describes the interaction of an electromagnetic wave with an electron.
Please remember that this expression is classical ( no quantum) and is non-relativistic. We can use this
lea
to get some idea about the various processes which take place when a photon interacts with an atom
in matter.
uc
Consider Rayleigh Scattering rst. This, as we have seen happens when the incoming energy is
lN

much less than the binding energy of the electron. Now the incoming energy of the photon is related
to the frequency of incoming radiation, that is !. What about the binding energy? The electron is
ua

bound to the nucleus by the electrostatic force which in this model is what provides the restoring force
with a force constant k. This leads to a "natural" frequency of the dipole which is !0 . Thus, Rayleigh
an

scattering regine is when !  !0 . In this limit,


bM

 2
!2 !4
(!2 !02 )
 !4
0
and the cross section is simply
La

R Ray =
8re2 !4
3 !04
(3.34)

This is the famous Rayleigh scattering cross section where we can see the dependence on the
wavelength of the electromagnetic wave. This is what gives us that blue colour of the sky where blue
light is scattered the most.

Rayleigh Scattering occurs when the electromagnetic wave has a much smaller energy than the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 114

binding energy. What about if the electromagnetic radiation is energetic enough to ionize the atom but
not energetic enough for it to accelerate it to relativistic speeds? This is the domain ~!0  ~!  me c2 .
This kind of scattering is called Thomoson Scattering In this limit,
 
!2 1
(!2 !02 )
 !02  1
1 !2
Thus the cross section for Thomson Scattering becomes

R 8re2

s
T = (3.35)

sic
3
This is a remarkable expression since it is totally independent of the frequency of the incoming ra-

hy
diation provided the condition above is met. In fact, it has a xed value of about T  32 barn. In fact,
Thompson scattering is basically the low energy limit of Compton scattering which we shall explore a

rP
little later. In Thompson scattering, the photon is scattered elastically while Compton scattering is the
inelastic scattering of the photon from an electron. (Recall that in elastic scattering of particles, the
lea
kinetic energy of the particle is conserved in the center of mass frame though not in the lab frame.)
uc
Though we have discussed Rayleigh and Thomson scattering, we know that these processes do not
play an important role in the interaction of gamma rays with matter. This is simply because the ener-
lN

gies what we have encountered above are clearly too low for the gamma rays that we are interested in
since we have seen that the gamma ray energies are typically in the MeV range (Figure 2.4 for instance)
ua

while the binding and ionisation energies of atoms is in the eV range and so Rayleigh scattering is
unimportant for gamma rays. Similarly, for Thomson scattering, though the energy of the photon is
an

more than the ionisation energy, it is less than the rest mass energy of the electron since the assumption
bM

is that the electron is non-relativistic. Thus, the energy of the photon is in the keV range. Thus,
neither of these two processes of elastic collision of photon and electron are important for gamma rays.
For gamma rays, there are three processes which play a signi cant role in the energy loss in matter.
La

These are Photoelectric Absorption, Compton E ect and Pair Production. It is important to
note that unlike the case of alpha and beta particles, where the energy of the incoming particle is lost
gradually due to continuous interaction with the electrons in the material, in the case of gamma rays,
the photon can simply disappear or get scattered signi cantly abruptly.

Photoelectric Absorption

The classical theory that we have described above of the interaction of the photon and the electron
breaks down when !  !0 since the denominator blows up. This is the resonance condition where
the theory outlined above is not valid. In this case, the photon interacting with the atomic electron
LABORATORY MANUAL FOR NUCLEAR PHYSICS 115

transfers all its energy to the electron and disappears. If the photon energy was more than Eb , the
binding energy of the electron it interacts with, then of course the balance energy manifests itself as
the kinetic energy of the free electron, Ee . That is

E = h = Ee + Eb
It turns out that to properly understand the process, one needs quantum mechanics to calculate the
cross section for photoelectric absorption. If a proper quantum mechanical calculation is done, we see
that the cross section PE  Z 5 where Z is the atomic number of the absorber. Thus we can see that
if we want an absorber to absorb gamma rays of the relevant energies, then we need to use high Z

s
sic
materials like lead etc.
One can use the classical electromagnetic theory together with the correspondence principle to get

hy
a fairly accurate expression for the cross section for photoelectric absorption. If one does this for both
the K shell electrons, for the non-relativistic case, one obtains (for the cross section per atom)

rP
R P E
p 5
= 4 2Z 4

leame c2 3:5
h

T (3.36)
uc
where Z is the atomic number of the atom,  is the frequency of the incoming photon, and T = 83 re2
is the Thomson scattering cross section.
lN

For the strongly relativistic case, that is E  me c2 , the corresponding expression is


ua

R
an

 
5 4 me c2
P E = 1:5Z T (3.37)
h
bM

An alternate formula which is applicable when the energy of the incoming photon is higher than the
binding energy of the K shell electrons is given by
La

R 4
3
3
P E = p T Z 4

me c 2 3
h

(3.38)

There are several things to note about these expressions. One is that the cross section shows an
absorption edge- when the energy of the photon is below the K shell binding energy, photoionization of
the K shell electrons cannot happen. Thus the cross section would drop at that energy. However, since
the L shell electrons require less energy to be ionized ( 14 of the K shell energy), they would be ionized
till the energy drops to below that level which again gives rise to an absorption edge. Below that, the
M shell electrons would be ionized and so on. Also note that the cross section goes as  (h ) 3 and
LABORATORY MANUAL FOR NUCLEAR PHYSICS 116

therefore the lower the energy, the higher is the cross section.

When a photoelectron is ejected, typically from an inner shell, it creates a vacancy in that shell.
This vacancy is lled either by a free electron which is captured or by an outer electron falling into the
vacant position. This results in the production of characteristic X-rays which it turns out are mostly
absorbed in the material itself.

Although we will study Compton Scattering next, it is instructive to point out the di erences between
the two processes.

s
sic
1. In Compton scattering, the combined momentum of the electron and the photon is conserved while
in photo absorption, momentum is transferred to the nucleus also.

hy
2. In Compton scattering, the outgoing photon carries some energy while in photo absorption, all

rP
the energy of the incoming photon is transferred to the electron as binding and kinetic energy.
3. Finally, Compton scattering becomes important only when the photon energy is atleast equal to
lea
the rest mass energy of the electron while for photo absorption, the cross section increases sharply
at low energies and for energies less than  100 keV, it dominates the total cross section as is clear
uc
from Fig 3.11.
To conclude, photoelectric absorption is the main process by which gamma rays of relatively low
lN

energy interact with matter (the cross section in Eq(3.36) varies inversely with energy and so at high
energies is small). The cross section depends strongly on the atomic number of the absorber and also
ua

at the binding energies of the various shells in the absorber atoms.


an

Compton Scattering
bM

For gamma rays of energies that we typically encounter in radioactive decay, the dominant process
for interaction is Compton Scattering. This, as we know, is the interaction of an atomic electron with
La

the incoming photon, in which the electron acquires enough energy to become free and also relativis-
tic. The scattering is therefore inelastic ( as opposed to Rayleigh and Thomson scattering which were
elastic). The photon transfers a part of its energy to the atomic electron and thereby loses energy
to the recoil electron. The process can be analysed using relativistic energy momentum conservation
as follows. We perform the calculation for the case of a free, unbound electron for simplicity.

Let the initial 4-momentum of the electron be Q~ i .

Q~ i = (me ; 0)
.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 117

Note that the electron is assumed to be at rest and hence its energy is simply the rest mass energy.
Here and in the following treatment, we take c = 1. Let the nal 4-momentum of the electron
after being hit by the photon be Q~ f . We know that the sqaured of the four momentum is an invariant
quantity. Hence

Q~ 2i = Q~ 2f = m2e
.
Let P~i and P~f be the initial and nal four momenta of the photon. Then

s
P~i = (pi ; pi n^ i )

sic
and

hy
P~f = (pf ; pf n^ f )

rP
where n^ i and n^ f are the unit vectors in the direction of the initial and nal photon 3-momentum
respectively.
For the photon, we know that lea
P~i2 = P~f2 = E 2 p2 c2 = 0
uc
Now energy momentum conservation implies that
lN
ua

Q~i + P~i = Q~f + P~f


Q~ f = Q~ i + P~i P~f
an

Q~ 2f = P~i2 + P~f2 + Q~ 2i 2P~i  P~f + 2Q~ i  (P~i P~f )


bM

P~i  P~f = Q~ i  (P~i P~f )


pi pf pi pf cos  = me (pi pf )
 1 1
2 sin2 ( ) = me
La

2 pf pi
  1 1 
2 sin2 ( ) = me c 2 (3.39)
2 hf hi
where  is the scattering angle of the photon and in the nal expression, we have not taken c = 1. In
the derivation above, we have used the fact that Q~ 2i = Q~ 2f ; P~i2 = P~f2 = 0 and the fact that Q~ i = (me ; 0)
and therefore Q~ i  (P~i P~f ) = me (pi pf ).

We can use the above expressions to get the energy of the scattered photon as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 118

R hf = 
h
i
1 + mhe ci2 (1 cos )
(3.40)

or, the recoiling electron energy which is simply the kinetic energy

R
0   1
hi (1 cos )
mc 2
Ee = hi hf = hi @ e  A (3.41)
1 + mhe ci2 (1

s
cos )

sic
The calculation of the cross section for Compton Scattering once again needs to take into account

hy
relativistic and quantum mechanical e ects. The result is a very complicated formula called the Klein-
Nishina formula which is

rP
 
d 2 1 + cos2  1 lea hi2 (1 cos )2
=r 1+ (3.42)
d
e 2 [1 + hi (1 cos )]2 (1 + cos2 )[1 + hi (1 cos )]
uc
This is of course a very complicated expression. A useful way to remember this is to approximate it
by
lN

R Comp  T
me c 2
ua

(3.43)
h
an

Thus we expect the cross section to decrease at high energies.


bM

It is instructive to consider two extreme cases for Compton scattering- one in which the photon is
hardly scattered, that is   0 and the other of back scattering, that is   . For   0, we can see
from Eq(3.40) and Eq(3.41), that
La

hf 
= hi (3.44)
and
Ee  0 (3.45)
There is hardly any energy transfer as is expected in such a case.

For   , the photon is back scattered and


hi
hf = (3.46)
1 + 2 mhe ci2
LABORATORY MANUAL FOR NUCLEAR PHYSICS 119

and
2hi2 !
me c
Ee = hi (3.47)
1 + 2hi2
me c
This is the case for the maximum energy transfer from the photon to the electron. Figure 3.12 shows
how the energies of the scattered photon and recoil electron vary.

s
sic
hy
rP
lea
uc
Figure 3.12: Compton E ect energies as a function of scattering anglex
x(Source: Wikipedia )
lN

Of course, when gamma rays pass through any absorber, scattering will occur at all angles and thus if
ua

one was to measure the energy spectrum of the recoiling electron, we would nd a continuous spectrum
till the maximum energy given by Eq(3.47). If one takes a monoenergetic gamma ray, then the energy
an

spectrum will look like that in Figure 3.13.


bM
La

Figure 3.13: Compton E ect- Energy spectrum of recoiling electronx


x(Source:By utefreek (Flutefreek) (Own work) [GFDL (http://www.gnu.org/copyleft/fdl.html), CC-BY-SA-3.0
(http://creativecommons.org/licenses/by-sa/3.0/)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 120

or CC BY-SA 2.5-2.0-1.0 (http://creativecommons.org/licenses/by-sa/2.5-2.0-1.0)],


via Wikimedia Commons )

In this gure, the di erence between the Compton edge and the full energy peak is Ec and is given
by
!
2hi hi
me c2
Ec = hi hi = (3.48)
1 + m2he ci2 1 + m2he ci2
or for large gamma ray energies (hi  me c2 =2), we get

s
sic
me c2
Ec  = 0:256 MeV
2

hy
Pair Production

rP
For high energy gamma rays, another process can result in the loss of gamma ray photons. This
is the production of an electron-positron pair by the photon in the presence of a nucleus. An isolated
photon cannot produce a particle-antiparticle pair because of conservation of energy and momentum.
lea
To see this, just go to the center of mass frame in which the electron and positron y back to back
uc
with equal momentum. The nal momentum in this frame is zero. However, the photon, in any frame
always moves with the velocity c and has a momentum equal to the energy of the photon (in units
lN

where c = 1). Thus it is not possible. If there is another nucleus or another particle then the situation
is completely di erent and that nucleus or particle can recoil and balance the momentum.
ua

By energy considerations, the minimum energy of the photon to be able to produce a pair of electron
an

and positron would be equal to twice the mass of the electron, that is 1:02 MeV. Of course, this is
the threshold energy and it is only when the photon energies are of several MeV that pair production
bM

becomes signi cant. The excess energy of the photon, above 1:02 MeV is shared by the electron-positron
pair as kinetic energy.
La

The pair production cross section can only be obtained using relativistic quantum mechanics. How-
ever, the cross section turns out to be, in the range of photon energy, 2me c2  h  me c2 Z 1=3 1 ,
   
28 2h 218
PP  Z 2 re2 ln 2 (3.49)
9 me c 27
This is of course a very complicated expression. However, note the weak dependence on the photon
energy in this range. Of course, for h < 2me c2 , the cross section goes to zero as it should. The
shielding of the nucleus by the atomic electrons basically for energies h  me c2 Z 1=3 1 means that
the cross section becomes constant beyond these energies.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 121

We also see that the cross section in the eld of an atomic nucleus varies as Z 2 and so absorbers with
high atomic number will give an increased energy loss due to this process. Further, the expression Eq
3.49 tells us that the cross section at low energies goes as the logarithm of the photon energy. Finally,
since we have seen above that the Compton Scattering cross section varies as  !1 , pair production is
signi cant at higher energies when the the Compton e ect cross section falls o .

In practice, when a high energy gamma ray passes through an absorber and pair production takes
place, what we see are two photons emerging as secondary products of the interaction. This is because
the positron which is produced in the pair production process, typically interacts with an electron soon

s
after being produced and produces two photons (again, it cannot produce one photon due to energy

sic
momentum conservation).

hy
Thus far, we have studied ve processes which are operative when electromagnetic radiation is inci-

rP
dent on a material. We have already seen that for gamma rays produced in radioactive decays, Rayleigh
and Thomson scattering are unimportant. The three process of importance are the photoelectric ab-
sorption, Compton E ect and Pair Production. All of these are operative but depending on the energy
lea
of the incoming gamma rays, their cross sections vary. This is shown in Figure 3.14.
uc
lN
ua
an
bM
La

Figure 3.14: Relative importance of interactions of gamma raysx


x(Source: Encyclopedia of Occupational Health & Safety, www.icocis.org ) We now have some understanding of how
alpha, beta and gamma rays interact with matter. We next turn to a study of the Geiger-Muller counter
which is the mainstay of our laboratory. We will study its working and see how it can be used to detect
these radioactive emissions.

3.3 References
1. \Radiation Detection & Measurement", Glenn F. Knoll, Wiley India (2009).
LABORATORY MANUAL FOR NUCLEAR PHYSICS 122

3.4 Questions
1. What is mass thickness? Why is it important and what does it tell us about the
absorber?
2. What is the mechanism for an alpha particle to lose energy while interacting with
matter?
3. When an alpha particle travels through matter, why does it not get de ected from its
initial path?

s
4. How does the energy loss by an alpha particle depend on its energy?

sic
5. What is a Bragg peak and why does it occur?

hy
6. How is the energy loss of an electron in matter di erent from that of alpha particles?
Why?

rP
7. What are the various processes by which an electron can lose energy in matter?
lea
8. What is Bremsstrahlung? Do alpha particle travelling through matter also lose energy
by Bremsstrahlung? If not, why not?
uc
9. In the case of electrons, what is the relative importance of energy lost by collisions
lN

and that by radiation in the non-relativistic and relativistic case?


10. What is the di erence in the range of a beam of mono-energetic electrons and that
ua

of the beta particles emitted from a radioactive nucleus? What is the di erence in
change of intensity with distance in both the cases?
an

11. What are the di erent processes by which electromagnetic radiation interacts with
bM

matter? What determines which process is dominant?


12. What is the di erence between Rayleigh scattering and Thomson scattering? Are
these two processes important for the interaction of gamma rays with matter? If not,
La

why not?
13. What are the processes which are dominant in the energy loss by interaction of gamma
rays with matter? How does the relative dominance of each of these processes vary
with the energy of the gamma rays?
14. How does the Compton e ect cross section vary with the energy of the incoming
gamma rays?
15. Can a gamma ray photon of energy 5 MeV in space produce an electron-positron pair?
If not, why not?
Chapter 4

G-M COUNTER

s
sic
hy
Learning Objectives

rP
1. To study various kinds of detector models.
lea
2. To understand the concept of ionisation of gases and Townsend avalanche.
3. To study the working of a GM counter.
uc
lN

4.1 Introduction
ua

In our laboratory, we use the Geiger-Muller counter (GM Counter) to study the radiation from
an

radioactive sources. The GM counter is a kind of counter or detector which uses the ionization produced
in a gas by the radiation to detect and study the radiation from the sources. It is a very convenient
bM

detector or counter but can not be used for studying the energy characteristics of the radiation and is
only used to detect and count.
La

Before we study the GM counter, it is instructive to understand the general principles of ionisation
of a gas by radiation since these are used in the GM counter as well. In addition to the GM counter,
other kinds of detectors also use this phenomenon. Thus, ion chambers and proportional counters are
also based on this process. We rst look at a generalised model of a detector and then investigate the
process of ionisation by radiation in gas lled detectors before studying the GM counter in detail.

123
LABORATORY MANUAL FOR NUCLEAR PHYSICS 124

4.2 Detector Models


Fundamentally, all detectors of nuclear radiation work on the principle of the radiation (alpha, beta
or gamma rays) interacting with matter as discussed in Chapter 3 and depositing its energy in some
form in the material. This could lead to an ionisation of the material, or the production of photons etc.
Typically we detect and measure this change in the detector material and from that infer the properties
of the incoming radiation.

In most detectors of interest to us, the net result is the production and collection of charge in the

s
detector. The charge Q produced by the interaction at some time t = 0 is collected by the presence of the

sic
eld which separates the two kinds of charges produced in the detector and collects them. Depending
on the detector, the time taken to collect the charge will obviously vary. The ow of this charge will

hy
lead to a current which lasts till the time the charge is collected and whose integral over the time from
t = 0 to the charge collection time tc will give us the amount of charge deposited and collected.

rP
Ztc
Q= lea I (t)dt
0
uc
lN
ua
an
bM
La

Figure 4.1: Current versus time

A pulse produced in the detector is due to one single interaction and we normally assume that the
rate of the ionising radiation entering the detector is low enough that we can distinguish between various
pulses as shown in Figure 4.2. Clearly, the size and the duration of each pulse depends on the actual
interaction taking place between the detector and the ionising radiation.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 125

s
sic
Figure 4.2: Current versus time

hy
rP
We can distinguish between three di erent modes of the operation of detectors:

1. Current Mode
lea
2. Pulse Mode
uc
3. Mean Square Voltage Mode
lN

In the Current Mode, one basically measures a detector signal as a current measurement. In this
mode, a current meter is connected to the detector output. Since the current level is in picoAmperes
ua

or nanoAmperes, a precise meter is required. Given that the current meter has a response time T , the
an

observed current from a sequence of events at time t will be


bM

Zt
1
I(t) = I (t0 )dt0
T
t T
La

The response time is usually longer than the time between individual detection events, so that an
average current is recorded at a time t. The current mode is used when event rates are very
high, which makes a stable current.

In the Pulse Mode, the information on energy and timing of individual events, i.e. the
information on the signal amplitude and time of occurrence is usually recorded. In this
mode, the detector records the charge from each individual ionising event. This mode is used when
we need to get the information on the timing and the amplitude of individual pulses. Thus, given this
property, this mode is not suitable for high count rates since then the time between neighbouring events
may not allow for getting that information. Because the information on the charge collected in each
LABORATORY MANUAL FOR NUCLEAR PHYSICS 126

individual event is recorded, this mode is used for energy measurements and spectrum.

The signal shape from a radiation detector depends on the electronics to which the detector is
connected as well as the detector response. In most cases, the the input stage of the electronics is
an RC circuit where the resistance is the total input resistance and the capacitance C is the total
capacitance of the detector, cables, electronics etc as shown in Figure 4.3. Here V (t) is the time
dependent voltage across the load R and is the signal produced.

s
sic
hy
rP
lea
uc
Figure 4.3: Detector Circuit
lN

It is easy to see that the time dependent voltage V (t) is given by


ua

V (t) = V0 (1 e t=RC ) (4.1)


an

where  = RC is the time constant of the RC circuit.


bM

We can distinguish two kinds of operations : When the time constant  = RC  tc , the charge
collection time. In this case, the current through R is the instantaneous value in the detector and
La

V (t) = I (t)R
In this case, the detector can collect charge from a single event with time tc .

In the second case, we have  = RC  tc . Now we can see that very little current ows through
R during the time tc and the current from the detector gets integrated in the capacitor which charges
to Vmax = QC . If the time between the two events is long, this charge will discharge through R. In this
case, note that for a xed C (which is determined by the electronics and the geometry and construction
of the detector etc.), Vmax is directly proportional to the charge Q deposited by the event. Thus,
measuring the height of the pulse that is Vmax , we can determine the charge deposited in
the detector and hence the energy produced by the incoming radiation in its interaction
LABORATORY MANUAL FOR NUCLEAR PHYSICS 127

with the detector material. Furthermore, a measurement of the pulse rate gives us a rate for the
interactions of the incoming radiation with the detector material. The two cases are shown in Fig 4.4

s
sic
hy
rP
lea
Figure 4.4:
uc
Finally, the Mean Square Voltage Mode (MSV) Mode is used for certain kinds of measure-
lN

ments where we need to measure radiation from sources which produce charges which very di erent
from each other. Basically, what we do is that in a current mode operation, the varying current which
can be regarded as the superposition of a steady current and a varying component, we block the steady
ua

component. The varying component is then squared and the signal is thus proportional to the square
an

of the charge that is created by the incident radiation. This enhances the di erences between the two
kinds of radiation entering the detector. We shall not discuss this much since it is not used in our lab.
bM

4.3 Ionisation of Gases


La

The basic idea in any ionisation based detector is that of gas multiplication. In its most rudimentary
form, consider a cylindrical enclosure lled with some gas. The enclosure has an electric eld which
increases from the center to the walls of the cylinder. A radiation in the form of alpha, beta or gamma
rays enters the gas and interacts with the atoms or molecules of the gas, ionizing the atom and produc-
ing a positive ion and a free electron. The main characteristic of any gaseous ionisation detector is to
amplify this initial ionization event. Let us see how this happens.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 128

4.3.1 Townsend Avalanche


The ionizing radiation upon interacting with the gas produces an ion and an electron. If there was no
eld present, then of course the positive ion and the electron would just drift till they lose all their
energy. However, in the presence of the eld, they move towards the electrodes, the ion towards the
cathode and the electron towards the anode.

The motion of the electrons and the ions is of course very di erent because of their di erent masses.
Both the ions and the electrons at any given temperature are always in some random thermal motion.
In the presence of the eld, there is a force on the charges and they also get a drift velocity. (This is

s
sic
very similar to the motion of electrons in a conductor when a voltage is applied between the two ends
of the conductor). It is seen that in a gas, the drift velocity v in the presence of a eld is given by

hy
/ EP

rP
v
E
v = lea (4.2)
P
where E is the electric eld and P is the pressure. The proportionality constant  is called the
uc
mobility and it depends on the nature of the gas and it turns out to be fairly independent of the value
of E and P . For ions, typical value of the mobility is around  10 4 m2 s 1 atm V 1 , For electrons,
lN

because of their much smaller mass, the mobility is typically higher by a factor of 103 than the ions.
The variation of the drift velocity with EP for some gases is shown in Figure 4.5.
ua
an
bM
La

Figure 4.5: Variation of drift velocity of electrons in various gases with EP x .


x(Source: http:// encyclopedia2.thefreedictionary.com/ Mobility+of+Ions+and+Electrons )

Now suppose we increase the eld. The ions being heavier move somewhat faster than before but
still have low mobility. The electrons on the other hand could gain signi cant kinetic energy. In their
path towards the anode, the electron would collide with other neutral atoms and if it transfers more
LABORATORY MANUAL FOR NUCLEAR PHYSICS 129

than the ionisation energy to the electron in a neutral atom, it will ionise it. Of course, exactly when
this secondary ionisation caused by the electrons in motion actually happens, depends on the density
and pressure of the gas as well as the electric eld strength. It is seen that in most gases at atmospheric
pressure, the threshold value of the eld required for this is around 106 V m 1 .

So after the rst ionisation by the radiation, we now have a second ionisation produced by the ionizing
electron from the rst ionisation. The two electrons now again move towards the anode because of the
electric eld and gain kinetic energy. When this energy is large enough, they could collide with two other
neutral atoms and ionise them thereby producing 4 electrons now. These four electrons move towards

s
the anode and after gaining enough energy could ionise four neutral atoms and thereby producing 4 more

sic
electrons and so on. This process continues as a chain reaction or a cascade as depicted schematically
in Figure 4.6 and is known as the Townsend Avalanche.

hy
rP
lea
uc
lN
ua
an

Figure 4.6: Schematic Diagram of Townsend Avalanche x

x(Source: \Electron avalanche" by Dougsim - Own work. Licensed under CC BY-SA 3.0 via Wikimedia Commons http:
bM

//commons.wikimedia.org/wiki/File:Electron avalanche.gif#/media/File:Electron avalanche.gif)

We can quantify this process of cascade ionisation by the Townsend Equation which gives us the
increase in the number of electrons per unit length as
La

R dN
N
= dx (4.3)

where is the rst Townsend coecient which depends on the eld strength and the nature as well
as the density and pressure of the gas. It is clearly zero for elds below the threshold value of the eld
referred to above. The solution to this equation is as expected, an exponential increase in the number
of electrons
LABORATORY MANUAL FOR NUCLEAR PHYSICS 130

R N (x) = N (0)e x (4.4)

The charges are nally deposited on the electrodes and this leads to a pulse in the associated elec-
tronics and whose characteristics give us some information on the nature and properties of the original
ionizing radiation.

4.3.2 Kinds of Detectors & Detector Regions

s
sic
We mentioned that there are several kinds of detectors which use the principle of ionisation. Propor-
tional Counters and GM counters are two examples which though using the same basic principle of

hy
ionisation, work in di erent regions as we shall see below. Both these work in the pulse mode.

rP
Consider a gas lled detector where, as we have seen above, we have a chamber lled with a gas
and an electric eld. Suppose in such a detector, we have one ionisation event due to the interaction
lea
of radiation. If we increase the applied voltage to the electrodes, hence increasing the electric eld
between them and note the amplitude of the pulse generated (which is a measure as we have seen of
uc
the charge which has been generated by the ionisation and is collected by the electrodes), we can see
the di erences in the response of the detector as shown in Figure 4.7.
lN
ua
an
bM
La

Figure 4.7: Detector Regions x

x(Source: \Radiation Detection & Measurement", G. Knoll, Wiley India, 2009 )

Starting from very low values of the voltage, we see that the electric eld experienced by the charges
is so small that some of the original ions can recombine and the charge that is collected by the detector
is less than that produced by the original radiation. This region is of course of no use for detection of
radiation. As we increase the applied voltage, we reach a saturation. This is when the charge collected
by the electrodes is equal to the charge created by the original radiation. This is the region where
LABORATORY MANUAL FOR NUCLEAR PHYSICS 131

another class of detectors called ion chambers operate.

When the voltage is increased even further, at some point the eld becomes strong enough that the
threshold is reached for gas multiplication which as we have seen above, then produces secondary ions
and electrons by collisions. The charge collected by the detector increases and is proportional to the
energy of the ionising radiation. As the voltage increases, the collected charge increases and we see an
increase in the pulse amplitude. Initially, for a range of applied voltage, the gas multiplication is linear
and we see that the charge collected is proportional to the original number of ion pairs (and hence to
the energy of the incoming radiation). This region is called the region of true proportionality and is

s
the region in which proportional counters operate. We can achieve a multiplication of about 106 in

sic
this region.

hy
If we increase the eld further, then this proportionality breaks down. The reason for this is the

rP
accumulation of a sheath of positive charges formed from the ions which are produced in the primary
and secondary ionisation. The positive ions being much heavier have smaller mobility and hence take a
much longer time to move towards the cathode. The space charge that is formed, distorts the electric
lea
eld experienced by the electrons in the gas and this leads to some nonlinearity. Thus in this region of
the applied voltage, we will not see a strictly linear increase of pulse amplitude with voltage though there
uc
is still an increase. This region is known as the region of limited proportionality. We can say that
lN

in this region, the pulse amplitude becomes more dependent on the voltage than on the initial ionisation.

As the eld is increased even more, we reach what is called the Geiger Region. Now the conditions
ua

are there for an avalanche to occur in the gas and a maximum number of electrons are produced in the
an

gas due to the cascade e ect as described above in Section 4.3.1. This is the region where the ion pair
count is now independent of the original ionisation and the detector cannot distinguish between di erent
bM

ionising radiations or their energies. However, the detector eciency is very good in this region. What
we see because of this independence is that the pulse amplitude is independent of the applied voltage
over a range of voltage and we get a plateau in the graph.
La

Finally, when the voltage increases beyond a certain value ( which depends on the nature and the
pressure of the gas as well as the geometry of the detector), the gas breaks down into a plasma and we
get a continuous discharge. This is bad for the detector and can damage the detector.

4.4 GM Counter
We have already seen how in general the detectors which are based on ionisation of gases work. The
GM counter, which is what we use in the laboratory is a kind of ionisation based detector which has
LABORATORY MANUAL FOR NUCLEAR PHYSICS 132

certain special characteristics which we shall examine now.

4.4.1 Geiger Discharge


The Townsend avalanche which we discussed above (Section 4.3.1) consists of secondary ions and elec-
trons forming after the initial ionisation event. However, the electrons produced in the avalanche also
sometimes produce excited gas molecules, that is when the energy transferred to the molecule is less
than that required for ionisation. These excited molecules deexcite with a characteristic time of around
10 9 seconds and in doing so, emit a photon. This photon could be in the visible or the ultraviolet,

s
sic
depending on the excitation state of the molecule. Now when this photon interacts (by photoelectric
absorption as we discussed in Section 3.2.3) with a less tightly bound electron somewhere else in the gas,

hy
the electron might be released. This electron will move towards the anode and once again be a source
of a new avalanche. The photon could also hit the walls of the tube and release an electron when it is

rP
absorbed. Typically, in a GM counter, the gas multiplication factor is very large,  109 1010 , and thus
we have an increasing number of avalanches created after a single ionisation event. The avalanches begin
lea
when the electron is close to the anode wire and thus the time needed for producing all the ions and
electrons is less than a microsecond. The ultraviolet photons produce more electrons near the original
uc
avalanche and these electrons also move towards the anode and create further avalanches. Thus, what
we see is that the discharge grows along the central anode wire in both directions from the position of
lN

the original ionising event. The speed of this growth along the anode is very high,  2 4 cm ( s) 1 .
This process is shown in Figure 4.8. The electrons collected by the anode constitute a pulse that is an
ua

increased voltage across the external load resistor and this is counted as an event or a count.
an
bM
La

Figure 4.8: Geiger Discharge x

x(Source: \Spread of avalanches in G-M tube" by Dougsim - Own work. Licensed under CC BY-SA 3.0 via Wikimedia Commons
- http://commons.wikimedia.org/wiki/File:Spread of avalanches in G-M tube.jpg#/media/File:Spread of avalanches in G-M tube.
jpg)

So we have a scenario where a single ionizing event has in a very short time, created many avalanches
LABORATORY MANUAL FOR NUCLEAR PHYSICS 133

throughout the tube. How does this process end? To understand this, we need to realise that whenever
a free electron is created in the avalanche or otherwise, a positive ion is also created. The positive ions
have a much lower mobility because of their larger mass. Thus they hardly move while all the electrons
created in the various avalanches are collected by the anode. Clearly, this concentration of positive
charges near the anode causes the electric eld to reduce since we know that in a cylindrical tube the
electric eld at a distance r from the anode will be
V
E (r ) = (4.5)
r ln ab

s
where V is the constant voltage between the cathode and the anode and b and a are the inner radius

sic
of the cathode and the radius of the anode wire respectively. (To see this, think of the positive space
charge acting as a sheath around the anode, thereby increasing the e ective radius of the anode wire , a.

hy
Now obviously, with an increase in a, the eld at a xed r will increase. However, and this is important,
the r where the eld is relevant now for ionisation, increases because of the space charge. This e ect is

rP
much larger and so overall the eld will decrease. To demonstrate this, let V = 100 V, b = 10 cm and
a = 1 cm. Then the eld at r = 1 cm, that is just above the anode wire, will be 43:4 V cm 1 . Now let
lea
a = 2 cm, then the eld at r = 2 cm, that is just above the e ective anode radius now is 31:1 V cm 1 .
) With the decrease in the electric eld below some threshold value, no further avalanches are possible
uc
and the Geiger discharge ends. For a particular tube (that is a particular geometry and composition of
the gases), with a xed external voltage the termination of the Geiger discharge is always after a xed
lN

number of avalanches or, what is the same thing, a xed amount of total charge. This does not depend
on the original ionizing event at all. To put it another way, the size of the pulse is always the same no
ua

matter what the nature of the incoming radiation which caused the initial ionizing event is.
an
bM

4.4.2 Quenching
What we have seen till now is that incoming radiation generates an ionisation induced Geiger discharge
in the GM counter. We saw that the mechanism for this is a Townsend avalanche which, in a GM tube
La

generates multiple avalanches till the discharge ends because of the space charge due to the positive
ions. In all of this, it is clear that the role of the gas in the GM tube is very important. Firstly, we must
make sure that the gas used in the counter, called the ll gas is such that there is no possibility of it
forming negative ions. This is obvious since the whole operation depends on the generation of electrons
and positive ions. Thus, we need the ll gas to have a low electron attachment coecient like hydrogen
or the rare gases and not a high electron attachment coecient like oxygen. Secondly, the properties
of the ll gas should be such that the energy gained by the free electrons is enough for the multiple
avalanches to be generated. Recall that the drift velocity and hence the average energy depends on the
ratio EP (Eq(4.2)). Thus we try to have a gas mixture which gives the maximum of this ratio so that
we can generate the minimum electron energy to generate multiple avalanches. Typically, we use argon
LABORATORY MANUAL FOR NUCLEAR PHYSICS 134

in the GM counters used in our labs.

The discussion above on Geiger discharge began with the single ionising event creating a cascade of
avalanches and then the avalanches travel down the anode wire very quickly and a pulse is created by
the charge collected at the anode which then essentially gets stored in the capacitance of the circuit and
discharges through the load resistor. We also discussed how the discharge ends because of the positive
space charge formed by the low mobility positive ions which reduces the electric eld near the anode as
we discussed above (Eq(4.5)). Now let us consider what happens to the positive ions. These ions, say of
argon or some other inert gas with a high ionisation potential, are much heavier and therefore have low

s
mobility. Their motion towards the cathode is slow but they at some point will arrive at the cathode.

sic
In this motion, they will gain kinetic energy. Now when they hit the cathode they will neutralise by
picking up an electron from the cathode. But suppose their energy was larger than twice the work

hy
function of the cathode, that is the energy required to liberate an electron from the surface of the

rP
cathode. In that case, the positive ion could, with some probability, liberate more than one electron
from the cathode. If the total number of positive ions is large (which it will be as we saw above since in
a GM tube, a large number of ion-electron pairs are produced) there could be at least one extra electron
lea
liberated from the cathode. This electron, like the other electrons in the tube, would accelerate towards
the anode and can cause another Geiger discharge producing more positive ions and the process can go
uc
on endlessly producing a continuous pulse.
lN

There are several ways of dealing with this problem of multiple pulses. In the initial days of the
GM counter, a method known as External Quenching was used. In this, the electronics of the device
ua

was arranged in such a way so as to reduce the voltage for some time after a pulse. The lower voltage
an

is such that no gas multiplication can take place and therefore no Geiger discharge happens after the
initial pulse. The time that the voltage needs to be reduced obviously depends on the time that the
bM

positive ions take to travel to the cathode, typically  10 1 milliseconds. One way to do this is to
choose a large enough R in the RC circuit external to the tube. Then, as we have seen, (Eq(4.1)), the
time constant of the RC circuit becomes much larger than the charge collection time. Clearly, this also
La

means that the GM counter will only be able to record events which are separated by more than a few
milliseconds. This method therefore can only be used with low count rates.

In the GM counters used in our lab, we use Internal Quenching. In this method, a small percent-
age, typically 5 10% of another gas, called the quench gas is added to the tube. The quench gas
has the property that it has a lower ionisation potential than the ll gas and is also a more complex
molecular structure. Popular quench gases are ethyl alcohol and halogens, both of which satisfy these
conditions compared to argon, a typical ll gas. Now as the positive argon ions are drifting towards the
cathode, they will collide with the halogen molecules and since these have a lower ionisation potential,
the argon ions will transfer their energy to the quench molecules and neutralise. The quench gas ions
LABORATORY MANUAL FOR NUCLEAR PHYSICS 135

will now start drifting towards the cathode. When they strike the cathode, they will neutralise by
picking up an electron but, and this is the crucial point, the extra energy instead of liberating another
electron, will go towards dissociating the quench gas molecules.

As mentioned above, both halogen gases and complex organic gases like ethyl alcohol are used as
quench gases. However, the organic gases are generally consumed and so the tubes which use them
have a nite lifetime. Halogens on the other hand can recombine spontaneously and so can be used
since they are self replenishing.

s
sic
4.4.3 Dead Time & Recovery Time

hy
After a Geiger discharge takes place, as mentioned above, we still have the positive ions of the ll gas
drifting away from the anode towards the cathode. Initially of course, when they are closer to the

rP
anode, they reduce the electric eld experienced by the electrons to below the level required for creating
another avalanche. At this time, if another particle of radiation enters the GM tube, and causes a single
lea
ionising event, that will not lead to a Geiger discharge because the eld is lower than that required by
the ionising electron to create an avalanche. This time is called the dead time and if another ionising
uc
event occurs in this period, it will not be recorded. However, slowly the positive ions drift towards
the cathode and so the eld near the anode rises. In this time, there could be some avalanches formed
lN

and therefore some pulses may be observed depending on the sensitivity of the counting circuit. These
pulses would not be of the same amplitude as the original Geiger discharge pulse though. Finally, all
ua

the positive ions do reach the cathode and at this point the electric eld near the anode becomes strong
enough again for the Geiger discharge of the same amplitude to take place. The behaviour is shown in
an

Figure 4.9
bM
La

Figure 4.9: Time behaviour of a GM tube & Dead Time x

x(Source: \Dead time of geiger muller tube" by Dougsim - Own work. Licensed under CC BY-SA 3.0 via Wikimedia Commons -
http://commons.wikimedia.org/wiki/File:Dead time of geiger muller tube.png#/media/File:Dead time of geiger muller tube.png)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 136

In most tubes, the dead time is around  50 100s. As mentioned above, the tube takes a longer
time than the dead time to return to its original con guration, that is when it can produce a second
pulse of the same magnitude as the rst one. This time is called recovery time.

Clearly, in cases where the count rates are high, determination of dead time is of critical importance.
There are several methods of determining the dead time. Here we discuss the commonly used two
source method.

s
sic
Let us de ne the following variables:
 , the dead time of the counter

hy
tr , the real or actual time the counter is in operation. This is the total time that we take to measure
the counts. It is clearly something which does not depend on  .

rP
tl , the live time of the detector, or the time that the detector is able to record the counts. This obviously
depends on  . lea
N , the total number of counts recorded by us during tr .
m, the measured counting rate, that is Ntr .
uc
n, the true counting rate, that is Ntl .
lN

So we have
ua

N
m=
tr
an

N
n=
tl
bM

Thus
m tl
= (4.6)
La

n tr
The interpretation of this is clear- the fraction of the counts we record is equal to the ratio of the live
time to the real time or to put it another way, it is the fraction of the time that the detector can actually
record the counts. The fraction of the time that the detector cannot record the counts is given by m .
It turns out that to a very good approximation, the live time is simply the real time minus N . This is
easy to see since the detector is unable to record any event during the total counting time tr is simply N .

So we have

tl = tr N (4.7)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 137

Using Eq(4.6) and Eq(4.7), we see that


tl N
=1  = 1 m (4.8)
tr tr
Or

tl m
=
tr n
= 1 m
m

s
n = (4.9)
1 m

sic
Eq(4.9) tells us the relationship between the measured count rate m , the dead time  and the true

hy
count rate n. Note that n is always greater than m as it should be. Further, for small values of m , the
dead time is not important since the di erence between n and m is small as a percentage. Finally, note

rP
that the parameter of importance is not just  but m . So, we can keep the product small by either a
small value of  or that of m.
lea
To nd the dead time of the counter, we use the fact that the count rate from two sources individually
uc
do not add up to the count rate from both of them together. This is because when the count rate is
high, the counter is dead for a longer period and therefore the counts dont add up. If m1 and m2 are
lN

the measured counts for the two sources and m12 is the measured count rate for both of them together,
ua

m12 6= m1 + m2
an

Of course, the true count rates do add up, that is


bM

n12 = n1 + n2
In carrying out any counting experiment with a source, we also need to take into account the measured
background count rate, that is the count rate caused by ambient sources of activity. Let the true count
La

rate for the background be nb and the measured one be mb . Then we have

n12 nb = n1 nb + n2 nb
n12 + nb = n1 + n2 (4.10)

Using Eq(4.9) for each of these true count rates n12 ; n1 and n2 , we get
m12 mb m1 m2
+ = + (4.11)
1 m12  1 mb  1 m1  1 m2 
In this equation all the quantities except  is a measured quantity and therefore we can use this
LABORATORY MANUAL FOR NUCLEAR PHYSICS 138

equation and the measured values of the count rates for the sources individually, together and the
background to get the value of  . Solving for  , we get

p
A(1 1 B)
 = (4.12)
C
A = m1 m2 mb m12
C = m1 m2 (m12 + mb ) mb m12 (m1 + m2 )
C (m1 + m2 m12 mb )
B =
A2

s
This is obviously a very complicated expression. An approximate solution to the equation (Eq(4.11))

sic
is

hy
  m1 +2mm2m m12 (4.13)
1 2

rP
The above expression, in the presence of signi cant background counts is replaced by
lea
R  2(mm1 + mm2 )(mm12 mmb)
uc
 (4.14)
1 b 2 b
lN

Similarly, many other approximations are sometimes used instead of the more complicated expres-
sion Eq(4.12).The experiment is usually carried out by measuring the counts from source 1, placing the
ua

second source 2 close by to count the combined counts and nally removing source 1 to get the counts
for source 2. To take care of absorption and scattering, when measuring counts from individual sources
an

1 or 2, a dummy source without activity is placed in place of the other source. One needs to be sure
bM

that the positions of the sources does not change when measuring them together. We also know that
since there is not much di erence between m12 and m1 + m2 , the count rates need to be measured very
accurately. We already know from Chapter 1 that the error in counting experiments goes as  p1N
La

where N is the count rate. Thus to get good results we need to have the fractional dead time m12  in
Eq(4.9) to be around 15 20%.

4.4.4 Geiger Counting Plateau & Operating Voltage


Now that we have discussed the general working of a GM counter, we need to understand in detail as
to how exactly one needs to use the GM counter in our experiments. The rst thing to determine is the
Operating Voltage of the tube. From the discussion above, we know that the GM tube will only work
if the electric eld inside it is above a certain value. If the electric eld (or the applied voltage) is too
low, there will no counts recorded because the eld cannot produce any pulse. As we raise the applied
LABORATORY MANUAL FOR NUCLEAR PHYSICS 139

voltage above the starting voltage , which is de ned as the voltage at which the GM counter just
begins to count, we start observing counts. As the voltage is increased further, we observe a very rapid
increase in the counting rate which is almost like a step function. This rising region is called the knee
because of its resemblance to the knee. The counting rate then rises till we reach a threshold voltage
Vt . Above this voltage, all ionising events produce the same output pulses. The threshold voltage de-
pends on the GM tube and the circuit components of the counter. Above the threshold value, the graph
levels o for a broad range of applied voltage. This region, after the knee or threshold value, where
the counting rate is level is called the plateau region. As we increase the voltage, the counting rate
remains essentially constant, that is the shape of the counting rate versus applied voltage is a straight

s
line almost parallel to the x axis. This continues till the voltage is high enough that a continuous

sic
discharge takes place in the tube and there is a breakdown. The reason for the continuous discharge are
clearly related to the inability of the quenching mechanism to stop runaway avalanches because of the

hy
high energy gained by the positive ions. The operating voltage is a value of the voltage in the middle

rP
region of the plateau, roughly equidistant from the knee and the point where continuous discharge starts.

To illustrate this, let us consider a GM counter where the starting voltage is Vs . If the applied voltage
lea
is less than Vs , then the pulse is not recorded as we have seen. As the voltage is increased to beyond
Vs , to reach the plateau region, the pulse can be measured. However, if the voltage is somewhat close
uc
to the knee, then the low amplitude tail of any pulse will cause a slight rise in the slope of the plateau
lN

region. The plateau region could also have a nite slope because of the quenching mechanism failing
sometimes.
ua
an
bM
La

Figure 4.10: Counting Plateau of GM tube

The operating voltage of the GM tube depends therefore on the ll gas and the quenching gas used.
For instance, the typical operating voltage for an argon lled tube with alcohol as the quenching gas is
around 1000 1200 V while those with argon as a ll gas and bromine as a quenching agent is around
200 400 V.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 140

4.4.5 Counting Eciency


Now that we have looked at how the GM counter works in theory, let us see what are the design con-
siderations that go into the making of a GM counter. Firstly, we need a gas lled enclosure with a
window. The window and the walls of the enclosure need to be strong enough to withstand the pressure
di erence between the inside and outside the tube since most GM tubes operate below the atmospheric
pressure. The window while being strong enough must be thin enough so as not to have signi cant
e ect on the ux of radiation which one is trying to measure. The gas, as we have seen is typically a

s
sic
rare gas along with a quenching agent.

hy
The operation of the tube depends crucially on the existence of a high enough eld near the anode.
This to a large extent determines the geometry of the counter. Thus, for instance, a parallel plate kind

rP
of arrangement of electrodes will be less ecient than a cylindrical geometry. This is evident from
the expressions for the elds in the two geometries. For a parallel plate geometry, the eld would be
lea
uniform and inversely proportional to the distance between the electrodes. On the other hand, the eld
for a cylindrical geometry, for the same voltage would be inversely proportional to the distance from
uc
the anode (Eq(4.5)) and so near the anode, where we need a large eld to ensure the avalanche, we can
use a much smaller external voltage.
lN

With the above mentioned design of the GM tube, when a charged particle like an alpha or a beta
ua

particle enters the tube, we can detect it. The reason for this, as we have seen is that in a GM tube,
a single ionising event can cause a pulse. So if the charged particle enters the tube, the eciency in
an

its detection will be close to 100%. However, we know that alpha particles have a low penetrating
power and so if we need to detect alpha particles, we need to ensure that the window is extremely thin.
bM

For beta particles, we use a thicker window though here too some particles could be re ected or back
scattered from the window material also.
La

The situation with gamma rays is very di erent from that for charged particles. The gamma ray
photons do not cause direct ionisation of the ll gas atoms. Instead, the photons interact with the
walls of the detector to produce a secondary electron. If the electron is able to penetrate the mate-
rial of the wall (this depends on the thickness and the material used for the walls), and enter the gas
in the tube, it will essentially behave like an ionising electron from an ionising event produced by a
charged particle in the GM counter and cause a Geiger discharge as discussed above. Thus we need
to make the walls of the tube thinner than the range of the electrons which are produced by the in-
teraction of the gamma ray photons with the wall material. Further, to increase the cross section of
this interaction, material with a higher atomic number needs to be used. Even with all this, for high
energy gamma rays, the typical GM counter eciency is a few percent. However, for low energy gamma
LABORATORY MANUAL FOR NUCLEAR PHYSICS 141

rays, the probability of a direct interaction with the ll gas atoms increases and if we use appropri-
ate gasses of high atomic number (Xenon for instance) at high pressures, the eciency can be very high.

Gamma sources are also are typically isotropic, that is the radiation is emitted isotropically in all
directions. Since the GM counter has a window of a nite area, only a fraction of these emitted gamma
rays will hit the counter window. Geometrically, it is clear that if we take the counter window to be
a circle of radius r and place the counter at a distance d from the gamma source, then the fraction of
gamma rays hitting the counter window will be the ratio of the area of the spherical cap of radius r
and the total area of the sphere of radius R. The area of the spherical cap with height x can be easily

s
seen to be 2Rx using double integrals or by any other method. Thus we have

sic
hy
2Rx x R d 1 d
fraction of gamma rays incident on counter window = =
4R2 2R
=
2R
= p
2 2 r2 + d2
(4.15)

rP
This expression for the fraction of gamma rays incident on the counter window is valid for all
distances of the source from the counter. We can intuitively see it to be true since when the distance
lea
of the source is 0, that is the source is next to the window, we expect one half of the emitted gamma
rays to enter the counter, which is what we get. On the other hand, when the source is very far away,
uc
then d  r and we get no gamma rays entering the counter.The geometry of the arrangement is shown
in the Figure 4.11.
lN
ua
an
bM
La

Figure 4.11: Isotropic source with GM counter


LABORATORY MANUAL FOR NUCLEAR PHYSICS 142

We can de ne two kinds of eciency of the counter/detector in principle. Absolute Eciency and
Intrinsic Eciency . Absolute eciency, Abs can be thought of as the number of particles detected
as a fraction of the total number of particles emitted. That is

R Abs =
Number of particles detected
Number of particles emitted
(4.16)

Clearly, the absolute eciency will depend on the counter geometry and the distance from the source.
Intrinsic eciency  int factors in these parameters and gives us a measure of eciency for that particular

s
detector

sic
R

hy
Number of particles detected
int = (4.17)
Number of particles incident on the detector

rP
We are now familiar with the mechanism by which a GM counter works as well as its operational
details. We also have learnt about the various parameters associated with the GM counter like operat-
lea
ing voltage, dead time, recovery time and its eciency. With this knowledge, we can now proceed to
use the GM counter for various experiments.
uc
lN

4.5 References
ua

1. \Radiation Detection & Measurement", Glenn F. Knoll, Wiley India (2009).


an
bM

4.6 Questions
1. What is a Townsend avalanche and how is it produced?
La

2. How does the number of electrons in a Townsend avalanche increase with time?
3. What are the various regions in which detectors operate?
4. What are the di erences between the proportional region and the Geiger plateau?
5. Explain what happens when a beta particle enters a GM counter.
6. How does the avalanche once started in a GM tube end?
7. Explain the process of quenching in a GM tube.
8. What is dead time and how is it di erent from recovery time of a GM counter?
LABORATORY MANUAL FOR NUCLEAR PHYSICS 143

9. Are the counting eciencies for beta and gamma rays di erent in a GM counter? If
yes, why?
10. Consider a very energetic beta particle and a very energetic gamma ray photon. What
are the di erences between the way in which these two interact with a GM counter?

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La
Chapter 5

Experiment: GM Characteristics

s
sic
hy
Things to know before you do the experiment

rP
1. Familiarity with radioactivity (types, properties, safety and precautions, etc.) and
lea
radioactive sources (Natural and arti cial) and proper handling of the radioactive
sources. For safety precautions, please see Section 5.2.
uc
2. Concept of half life, activity, decay rate, decay law of the radioactive sources.
lN

3. Familiarity with the various sources of background radiation.


4. The di erent process by which radiation interacts with matter.
ua

5. Basic interactions and their properties.


an

6. Properties of Inert gases and molecular gases with regard to ionisation.


bM

7. Electric eld distribution for various con gurations like a cylinder, sphere etc.
8. Concept of a Solid angle.
La

9. Knowledge of statistics. Familiarity with probability, Central Limit theorem, vari-


ous kind of basic probability distribution functions (Binomial, Poisson, Gaussian),
concept of mean, average, variance and standard deviation of a distribution.
10. Curve tting techniques like least square t and 2 t.
11. The meaning of p-value in statistics.
12. Di erence between interpolation and curve tting.
13. Concept of dead time, recovery time, RC circuit, time constants, etc.

144
LABORATORY MANUAL FOR NUCLEAR PHYSICS 145

5.1 Introduction
Now that we have discussed the theoretical basis for the experiments in the laboratory, let us look at
the actual experiments that we perform in this laboratory. The basic equipment that we use in the
laboratory is the GM counter. We use it to nd out the count rates from various nuclear sources and
thus try and understand the characteristics of the radiation coming out from them. For this purpose,
we rst need to study the operational characteristics of the GM counter itself. This experiment
investigates the working of a GM counter and nding its parameters like the optimum
operating voltage, dead time etc. In addition, it also demonstrates the statistical nature
of radioactive processes.

s
sic
hy
5.2 Precautions

rP
Since all the experiments in this laboratory deal with radioactive sources, it is important to follow cer-
tain guidelines and precautions for working in this laboratory. Radioactive sources pose a substantial
lea
health hazard and so it is essential that you follow these guidelines very carefully.
uc
5.2.1 Health E ects of Radiation
lN

As we have already seen in earlier chapters, radioactive materials decay spontaneously to produce ion-
izing radiation like alpha, beta and gamma rays. These radiations have sucient energy to strip away
ua

electrons from atoms (as we saw in Sections 3.2.1, 3.2.2 & 3.2.3) or to break some chemical bonds. Any
an

living tissue in the human body can be damaged by ionizing radiation in a unique manner. The body
attempts to repair the damage, but sometimes the damage is of a nature that cannot be repaired or it
bM

is too severe or widespread to be repaired. Also mistakes made in the natural repair process can lead to
cancerous cells. All types of radiation to which the person is exposed and the pathway by which they
are exposed in uence health e ects. Di erent types of radiation vary in their ability to damage di erent
La

kinds of tissue. Radiation and radiation emitters (radionuclides) can expose the whole body (direct
exposure) or expose tissues inside the body when inhaled or ingested. All kinds of ionizing radiation
can cause cancer and other health e ects. The main di erence in the ability of alpha and beta particles
and gamma and x-rays to cause health e ects is the amount of energy they can deposit in a given space.
Their energy determines how far they can penetrate into tissue. It also determines how much energy
they are able to transmit directly or indirectly to tissues and the resulting damage. Although an alpha
particle and a gamma ray may have the same amount of energy, inside the body the alpha particle will
deposit all of its energy in a very small volume of tissue. The gamma radiation will spread energy over
a much larger volume.
There are two kinds of health e ects that radioactive sources can cause - Stochastic and Non-
LABORATORY MANUAL FOR NUCLEAR PHYSICS 146

stochastic.
Stochastic health e ects are those that are associated with long term and low level exposure to radia-
tion. Increased levels of exposure make these health e ects more likely to occur, but do not in uence
the type or severity of the e ect. Cancer is considered by most people the primary health e ect from
radiation exposure. Simply put, cancer is the uncontrolled growth of cells. Ordinarily, natural processes
control the rate at which cells grow and replace themselves. They also control the body's processes
for repairing or replacing damaged tissue. Damage occurring at the cellular or molecular level, can
disrupt the control processes, permitting the uncontrolled growth of cells cancer This is why ionizing
radiation's ability to break chemical bonds in atoms and molecules makes it such a potent carcinogen.

s
Other stochastic e ects also occur. Radiation can cause changes in DNA, the \blueprints" that ensure

sic
cell repair and replacement produces a perfect copy of the original cell. Changes in DNA are called
mutations. Sometimes the body fails to repair these mutations or even creates mutations during repair.

hy
The mutations can be teratogenic or genetic. Teratogenic mutations are caused by exposure of the

rP
fetus in the uterus and a ect only the individual who was exposed. Genetic mutations are passed on
to o spring.
Non-stochastic health e ects appear in cases of exposure to high levels of radiation, and become more
lea
severe as the exposure increases. Short-term, high-level exposure is referred to as `acute' exposure.
Many non-cancerous health e ects of radiation are non-stochastic. Unlike cancer, health e ects from
uc
`acute' exposure to radiation usually appear quickly. Acute health e ects include burns and radiation
lN

sickness. Radiation sickness is also called `radiation poisoning.' It can cause premature aging or even
death. If the dose is fatal, death usually occurs within two months. The symptoms of radiation sick-
ness include: nausea, weakness, hair loss, skin burns or diminished organ function. Thus, for instance,
ua

some medical patients receiving radiation treatments often experience acute e ects, because they are
an

receiving relatively high \bursts" of radiation during treatment.


The unit of e ective or equivalent dose or radiation is the rem or Roentgen Equivalent in Man. This
bM

is the older though much used unit of the radiation dose exposure. A more modern unit is the Sievert
which is de ned as 100 rem. A rem is a very large unit and so typically millirem is used for ordinary
exposures. 1 rem or 0:01 sievert translates to typically a 0:055% chance of the development of cancer
La

eventually. Typically,the exposure from background sources is around 0:01 millisieverts (mSv) per day
while the exposure to a chest x-ray is 0:06 mSv during the exposure to the x-ray. For a CT scan,
the exposure can be as high as 2 mSv. To compare, during the Fukushima nuclear accident, near the
accident site, a level of 1000 mSv hour 1 has been reported.

General Precautions

1. Handle the radioactive sources with utmost care and respect. Dont bend or try to break them.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 147

2. Although the sources in the laboratory are always in their holders, it is in general important
to never touch the source using bare hands. Always use forceps to handle sources.
3. Do not eat or drink during the lab. Please do not keep any edible material or even drinking
water on your work table. Keep your bags with the food and water on the table on the side.
4. When bringing or returning the source to the source room, please be extremely careful to not
let it fall.
5. Do not leave the source lying around the work table. Always keep it carefully while using it

s
and return it promptly after you are nished.

sic
6. Do not keep the source in your pocket or in close contact with your body.

hy
7. Wash your hands after the experiment and before eating or drinking anything.

rP
5.3 Experiment
lea
uc
5.3.1 Purpose
lN

The purpose of this experiment is to study the GM counter and determine its characteristics. Speci -
cally, we will be plotting the counts versus the external voltage curve to determine the plateau region
ua

(Section 4.4.4) and hence the operating voltage. We will also be determining the dead time (Section
4.4.3) of the counter. Finally, we will investigate the statistical nature of the radioactive process. For
an

this purpose we would need a GM counter (which comprises of the GM tube, associated electronics for
counting purposes and an external high voltage source), two radioactive sources (which in our case are
bM

60Co) and a dummy source. The decay scheme for 60Co is given in Figure 2.4. The reactions are
27 27
60Co ! 60Ni + e + 
27 28 e
La

60Ni ! 60Ni +
28 28
60Ni ! 60Ni +
28 28

5.3.2 Method
1. GM Characteristics:
To determine the characteristics of the GM counter, we rst set the voltage to zero. Place the
source provided near the counter window and set the preset time to 50 seconds. This is the
time during which the counter will measure the counts. We increase the voltage gradually, initially
increasing it by a larger amount (say 50 V) till the counts start. This voltage at which the counts
LABORATORY MANUAL FOR NUCLEAR PHYSICS 148

start is the threshold voltage. Once the counts start, the voltage should be increased in smaller
steps (say 10 V). You will note that after a sharp sudden rise in the counts (the knee of the char-
acteristic), the counts remain constant with an increase in the voltage. This is the plateau region
discussed in Section 4.4.4. Finally, at some point, the count rate starts to increase. This is the
region where the continuous discharge starts and one should not increase the voltage beyond this
to avoid damage to the counter.

2. Measurement of Dead Time:


We use the two source method to determine the dead time of the GM counter (Section 4.4.3). For

s
sic
this we need two sources. First, we set the voltage in the GM counter at the operating voltage.
This, as we have seen is the value at which the GM counter should be operated and is the value

hy
in the middle of the plateau region of the GM characteristic. After setting the voltage, we set the
preset time. For this experiment now, we set the preset time to a large value, say around 500 1000

rP
seconds. Recall that for determining the dead time, we need to nd the counts from two sources
separately and then together and also need to know the background counts (Section 4.4.3, Eq
lea
(4.14)). First we remove all sources and measure the background count rate only with the dummy
source. Then we place one source and the dummy source (to ensure that all the absorption and
uc
other e ects from the source holder are equalised) and measure the counts. We repeat the same
with the second source and a dummy. Finally, we take both the sources together and determine
lN

the combined count rate for the two sources.


ua

3. Counting Statistics:
an

To see the statistical nature of the counts, we remove all sources. We set the operating voltage
and then for various preset times (say 5; 10; 20 seconds), we note the number of counts to get the
bM

background reading. We can repeat the measurement for 150 200 times to get a viable statistical
sample.
La

5.3.3 Sample Data


1. GM Characteristics:

Given below is the sample data for determining the GM characteristics.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 149

Voltage (V) Counts (N) Error


320 0 0
324 16 4
326 111 11
328 411 20
330 539 23
332 795 28
334 930 31
336 909 30

s
338 950 31

sic
340 912 30

hy
345 998 32
350 933 31

rP
360 1005 32
370 964 31
390 947
lea 31
400 983 31
420 1009 32
uc
430 980 31
lN

440 993 32
450 1035 32
ua

460 1037 32
470 971 31
an

480 991 31
490 1054 32
bM

500 938 31
510 1029 32
520 987 31
La

530 1043 32
540 1085 33
550 952 31
560 998 32
570 1026 32
580 1049 32
590 1058 33
600 1069 33
Table 5.1: Sample Data for GM Characteristics
LABORATORY MANUAL FOR NUCLEAR PHYSICS 150

The rst two columns give us the voltage (V) in Volts and the number of counts N . Since this is
a counting experiment with a xed time interval (the preset time) and the events are random (ra-
dioactive decay is a random event), the conditions of the distribution being a Poisson distribution
are satis ed as we saw in Section 1.4.2. Therefore,
p square root rule (Section 1.5) tells us that the
error in each measurement of the counts is N . This is tabulated in column 3. As expected, the
estimated error is lower for lower count rates.

Notice how the errors are reported. The voltage that we use is assumed to be accurate
and hence by assumption has no error. The counts
p N , that we measure at any voltage,

s
as discussed earlier have an error which is N . But we have already seen that the

sic
error or uncertainty should be reported to 1 or 2 signi cant gures with appropriate

hy
rounding o (See Section 1.2.4). Hence, for instance, the square root of N = 1058 is
32:52 and that of N = 1049 is 32:38, and so we report the uncertainty in these as 33 and

rP
32 respectively. Further, though in this case it is not relevant, the data itself should
only be reported to a precision which is the same as the uncertainty. This is very
important since your calculators will give you a precision of 8 or more digits which is
lea
meaningless in our experiment set up.
uc
We can plot the date as shown in Fig 5.1.
lN

GM Characteristic
ua

1200
an

1000
bM

800
Counts

600
La

400

200

0
300 350 400 450 500 550 600
Voltage

Figure 5.1: Sample GM Characteristics

We see that the graph looks very similar to what we expect theoretically as in Figure 4.10. The
LABORATORY MANUAL FOR NUCLEAR PHYSICS 151

sudden rise after the threshold voltage, the knee, the plateau are all clearly visible in this sample
data. The plateau region is not a line of zero slope as expected in an ideal GM tube but has some
nite slope. Nevertheless, the constancy of the count rate beyond the knee of the characteristic is
clearly visible in the graph. The error bars represent our estimate of the error in each measurement
of N . We can draw a smooth line through the points and we obtain the characteristic. Alternatively,
we can assume that the plateau region is a straight line and t the points to a straight line using
Method of Least Squares (Section 1.7, Eq(1.71) & Eq(1.72)). We can use the linear part of the
curve to nd the operating voltage of our GM tube. For this, recall that the operating voltage is
typically taken to be in the middle of the plateau region. We can take two values near the extreme

s
ends of the straight line in the plateau region and compute the middle point. In this case, the

sic
operating voltage turns our to be  450 V. The slope of the plateau region should ideally be 0.
However, we can determine the slope from the data given as

hy
y 952 933 19

rP
m= = =
x 550 350 200
 0:1
2. Dead Time: lea
Preset time is set at 600 seconds. The observed counts for source 1 + dummy (N1 ), source 2 +
uc
dummy (N2 ), source 1 + source 2 (N12 ) and the background (NB ) are are follows.
lN

N1 N2 N12 NB
ua

28401 23926 51957 268


an

28589 23970 51827 236


N1 =p28495 N2 =p23948 N12 =
p
51892 NB p
= 252
bM

N1 = N1 = 168 N2 = N2 = 154 N12 = N12 = 227 NB = NB = 15


Table 5.2: Dead Time Determination
La

Once again, note the reporting of the uncertainty in the count rates N1 ;    ; NB . We are again taking
the square root of the counts and reporting it to 3 signi cant gures here. We can and should report
these to two signi cant gures as per our convention. However, in this case we have retained more
than 2 signi cant gures because as we will see below, we dont actually use these numbers to calculate
the error in the dead time.

We can calculate the dead time using Eq(4.14)


LABORATORY MANUAL FOR NUCLEAR PHYSICS 152

  2(mm1 + mm2 )(mm12 mmb)


1 b 2 b
To get the proper dead time,we need to multiply  with the preset time since the counts m1 ; m2 ;   
are the counts per unit time. In our case N1 ; N2 ;    are the total number of counts in the preset time
interval.

Putting in the numbers we get,

s
 1:34  10 4 seconds

sic
 (5.1)

The error in  can also be calculated. We know that  is a derived quantity from the measured

hy
quantities N1 ; N 2; N12 and Nb . We can use the error propagation equation (Eq(1.46) to calculate the

rP
error in  if the errors in the measured quantities are known. But the measured quantities are simply
the counts in a xed time interval and the events are random, the distribution will be Poissononian
and hence the error in the measurements will go as the square root of the counts.
lea
To calculate the error in the dead time, we use the error propagation expressions. We know that
uc
 2(mm1 + mm2 )(mm12 mmb)  uv
lN


1 b 2 b
ua

u = m1 + m2 m12 mb
an

v = 2(m1 mb )(m2 mb )
Then by Eq(1.54), we have
bM

2 u2 v2


= +
 2 u2 v2
La

But

u2 = m2 1 + m2 2 + m2 12 + m2 b


using Eq(1.51).

Similarly using Eq(1.53) and Eq(1.51) on the denominator, we get


v2 m2 1 + m2 b m2 2 + m2 b
= +
v2 (m1 mb )2 (m2 mb )2
Putting it all together, we get the expression for the error in the dead time as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 153

 2 
2 2 m1 + m2 2 + m2 12 + m2 b m2 1 + m2 b m2 2 + m2 b
 =  + +
(m1 + m2 m12 mb )2 (m1 mb )2 (m2 mb )2
In this expression, we know all the quantities. One way to think about the errors in m1 ; m2 ; m12 ; mb or
equivalently in N1 ; N2 ; N12 ; NB would be that they are given by
q
m1 = N1
q
m2 = N2
q

s
m12 = N12

sic
q
mb = NB

hy
However, there is a subtle point here that one needs to understand. Let us see what we are doing- we

rP
are taking the total number of background counts in 600 seconds and reporting it as a number. Then
we are repeating the same procedure 2 times. Note that in each of the values of N1 ; N2 ; N12 &NB ,
lea
there is an error. This is the inherent statistical error that is associated with the event which as we
know is a result of a Poisson distribution. We can therefore think of each value of N1 ; N2 ; N12 &NB
uc
as a mean of a Poisson distribution. The associated error with each value of N1 ; N2 ; N12 &NB is thus
pN ; pN ;    ; N . Therefore, the correct procedure to exhibit this inherent statistical error would
1 2 B
lN

be to take the expression for the mean value of N1 ; N2 ;    NB , that is N1 ; N2 ; N12 ; NB and apply the
appropriate error prorogation equation to it. In the case of N1 , for instance, this means
ua

2
an

P
(N1 )i
i=1
N1 =
2
bM

1 2 
N2 1 = (N1 )1 + (2N1 )2
4
1
= [28401 + 28589]
La

4
56990
=
p4
56990
N1 =
2
= 119:4 ' 120 (5.2)

Similarly we can calculate the errors in N2 ; N12 ; NB . They are


p
47896
N2 = = 109:4 ' 110
2
LABORATORY MANUAL FOR NUCLEAR PHYSICS 154

p
103784
N12 = = 161:1 ' 160
2
p
504
NB = = 11:2  eq11
2
With these values of the errors in the counts, we can calculate the error in the deadtime.To get the
actual dead time, the above expression will need to be multiplied by the preset time.

3. Counting Statistics:

s
Preset Time: 5 seconds

sic
Operating Voltage: 460 Volts

hy
In Table 5.3, xi is the number of counts in the preset interval and fi is the number of times out of a

rP
P
total of fi = 200 that we get xi counts.

xi fi xi fi
lea
(xi x) (xi x)2 pe = Pfifi (xi x)2 p
uc
0 27 0 -1.94 3.76 0.135 0.507
1 55 55 -0.94 0.880 0.275 0.242
lN

2 60 120 0.06 0.086 0.300 0.010


3 31 93 1.06 1.12 0.155 0.173
ua

4 16 64 2.06 4.24 0.080 0.339


5 06 30 3.06 9.86 0.030 0.280
an

6 05 30 4.06 16.48 0.025 0.412


P P P P
fi = 200 xi fi = 387 pe = 1 E = (xi x)2 p = 1:96
2
bM

Table 5.3: Counting Statistics for Preset time = 5 seconds


La

P
We note that the sum of the probabilities pe = 1 as it should be.

We can easily calculate the sample mean as in Eq(1.3) as


P
x i fi
387
x = P = 1:94
=
fi 200
The corresponding sample variance can be calculated from the expectation value of the square of the
deviations from the sample mean
X
E2 = (xi x)2 p = 1:96
LABORATORY MANUAL FOR NUCLEAR PHYSICS 155

or

E = 1:40
Our theoretical expectation is that the distribution of the counts would follow a Poisson distribution.
However, we do not know the mean of the distribution. We can estimate the mean of the parent
distribution by taking it to be the sample mean given above. This means that

 = x

s
Then we expect that the probability distribution function to be (Eq(1.23))

sic
xi
PPoisson  P (xi ; ) = e 

hy
xi !
We can tabulate these values also

rP
xi PPoisson pe
0 0.144 lea 0.135
1 0.279 0.275
2 0.271 0.300
uc
3 0.175 0.155
4 0.084 0.080
lN

5 0.030 0.030
6 0.011 0.025
ua

Table 5.4: Experimental and Poisson probabilities: Preset Time 5 seconds


an
bM

The standard deviation of the Poisson distribution is simply


p
La

Poisson =  = 1:39
The graph of the experimental data and the theoretical Poisson distribution with the same mean is
shown in Figure 5.2.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 156

GM Counting Statistics: Preset time 5 seconds


0.35
Poisson
Data

0.3

0.25

0.2
p

s
0.15

sic
0.1

hy
0.05

rP
0
0 1 2 3 4 5 6
x

lea
Figure 5.2: Counting Statistics : Preset 5 seconds
uc
We can see that the theoretically expected Poisson distribution and the experimental distribution are
lN

fairly similar though there are deviations. However, the error bars on the experimental values of the
derived probabilities are such that the theoretical distribution falls within the range. The error bars
ua

are the error on pe . However, we know that pe is actually a derived quantity, since
an

fi
pe = P (5.3)
fi
bM

P
Now fi = 200 and obviously there is no error in this. We take the error in f as
pf and get the
pf i i
error bars by taking the error in pe to be 200 .
i
La

We also saw in Chapter 1, Section 1.8 that there is another way to determine if the experimental data
conforms to our assumed statistical model. The way to do this through the 2 test. Recall that we
de ne a quantity 2 as (Eq.(1.79))
N
2 1X
 = (x x)2
x i=1 i
Here x is simply the sample mean. The numerator in the above expression can be easily related to the
sample variance (Eq(1.7)) as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 157

(N 1)s2
2 =
x
We can also de ne another quantity, the reduced Chi squared (Eq(1.82)) as
2
~2 =

where  is simply the degrees of freedom de ned as

=N Nc

s
sic
Here N is the number of observations and Nc is the number of constraints.

hy
The number of constraints is 1 , coming from the fact that we have already used the data to determine
x . Therefore

rP
=N 1
This implies
lea
uc
s2
~2 =
x
lN

If our data was a true Poisson distribution, then we know that

s2 = x
ua

and therefore, for a true Poisson distribution, the value of ~2 will simply be 1. If this value of ~2 turns
an

out to be too di erent from 1 then we can say that either our experimental data is not quite correct or
bM

our underlying statistical model is faulty.

In our case, we have already calculated both the sample mean and the sample variance above. We thus
La

get
2 s2 E2 1:96
~ = = = = 1:01
x x 1:94
The value of the 2 is then simply

2 = 1:01   = 1:01  (200 1) = 201:05


We can look up tables of 2 (for instance at en.wikibooks.org/wiki/Engineering Tables/
Chi-Squared Distibution or www.pd.infn.it/lunardon/didattica/docsper2/TavoleChi2.pd) to nd
out the probability associated with this value of ~2 . To use these tables, one rst determines the
degrees of freedom in the experiment. In our case, the number of observations is 200 and so the degrees
LABORATORY MANUAL FOR NUCLEAR PHYSICS 158

of freedom are 199. Then one looks in the row corresponding to the number of degrees of freedom (199
in our case) and nds the column where the number is the closest (next smallest) to the value of 2
obtained from your data. The value of p corresponding to that column is the probability that a random
sample chosen from a Poisson distribution would have a larger value of 2 than that shown in the
table. In our case, we see that this probability is less than 0:5 which is a very good match. A perfect t
to the Poisson distribution should yield a value of p = 0:5 exactly. In our case, it is slightly less than 0:5.

In fact, one can get the p-value directly from MS Excel by using the function CHISQ.DIST.RT(2 ;  ).
If we use this we get a p-value of 0:445989. If we plot the distribution for 199 degrees of freedom with

s
this observed value of 2 , we see that the graph will look like

sic
hy
rP
lea
uc
lN

Figure 5.3: p-value for  = 199 and 2 = 201:05

Preset Time: 10 seconds


ua

Operating Voltage: 460 Volts


an

xi fi x i fi (xi x) (xi x)2 pe = Pfifi (xi x)2 p


bM

0 06 0 -3.81 14.52 0.030 0.435


1 18 18 -2.81 7.890 0.090 0.710
2 33 66 -1.81 3.280 0.165 0.541
La

3 38 114 -0.81 0.660 0.190 0.125


4 36 134 0.19 0.036 0.180 0.006
5 27 135 1.19 1.420 0.135 0.191
6 18 108 2.19 4.790 0.090 0.431
7 13 91 3.19 10.180 0.0650 0.661
8 05 40 4.19 17.560 0.0250 0.439
9 04 36 5.19 26.940 0.020 0.538
10 02 20 6.19 38.320 0.010 0.383
P P P P
fi = 200 xi fi = 762 pe = 1 E = (xi x)2 p = 4:46
2

Table 5.5: Counting Statistics for Preset time = 10 seconds


LABORATORY MANUAL FOR NUCLEAR PHYSICS 159

We can easily calculate the sample mean as in Eq(1.3) as


P
xi fi
762
x = P = 3:81
=
fi 200
The corresponding sample variance can be calculated from the expectation value of the square of the
deviations from the sample mean
X
E2 = (xi x)2 p = 4:46
or

s
sic
E = 2:11

hy
Our theoretical expectation is that the distribution of the counts would follow a Poisson distribution.
However, we do not know the mean of the distribution. We can estimate the mean of the parent

rP
distribution by taking it to be the sample mean given above. This mean that

lea
 = x
Then we expect that the probability distribution function to be (Eq(1.23))
uc
xi
PPoisson  P (xi ; ) = e 
lN

xi !
We can tabulate these values also
ua

xi PPoisson pe
0 0.022 0.030
an

1 0.083 0.090
bM

2 0.159 0.165
3 0.202 0.190
4 0.193 0.180
La

5 0.147 0.135
6 0.093 0.090
7 0.050 0.065
8 0.024 0.025
9 0.010 0.020
10 0.003 0.010
Table 5.6: Experimental and Poisson probabilities: Preset Time 10 seconds

The standard deviation of the Poisson distribution is simply


LABORATORY MANUAL FOR NUCLEAR PHYSICS 160

p
Poisson =  = 1:95
The graph of the experimental data and the theoretical Poisson distribution with the same mean is
shown in Figure 5.4.

GM Counting Statistics: Preset time 10 seconds


0.25
Poisson
Data

s
sic
0.2

hy
0.15

rP
p

0.1

0.05
lea
uc
0
0 2 4 6 8 10
lN

x
ua

Figure 5.4: Counting Statistics : Preset 10 seconds


an

We can see that the theoretically expected Poisson distribution and the experimental distribution are
bM

fairly similar though there are deviations. However, the error bars on the experimental values of the
derived probabilities are such that the theoretical distribution falls within the range.

The reduced 2 can again be calculated as


La

E2 4:46
~2 = = = 1:17
x 3:81
and the 2 as

2 = 232:95
Again, referring to the tables for 2 for 199 degrees of freedom, we get a value between 0:05 < p < 0:1
indicating a reasonable though not very good t to our theoretical model. The exact value from Excel
is 0:0492.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 161

s
Figure 5.5: p-value for  = 199 and 2 = 232:95

sic
Preset Time: 20 seconds

hy
Operating Voltage: 460 Volts

rP
xi fi xi fi (xi x) (xi x)2 pe = Pfifi (xi x)2 p
2 5 10 -6.155 37.88lea0.025 0.947
3 5 15 -5.155 26.57 0.025 0.664
4 15 60 -4.155 17.26 0.075 1.294
uc
5 12 60 -3.155 9.95 0.060 0.597
lN

6 23 138 -2.155 4.64 0.115 0.533


7 25 175 -1.155 1.33 0.125 0.166
ua

8 21 168 -0.155 0.024 0.105 0.002


9 30 270 0.845 0.714 0.150 0.107
an

10 26 260 1.845 3.4 0.130 0.442


11 15 165 2.845 8.09 0.075 606
bM

12 09 108 3.845 14.78 0.045 0.665


13 05 65 4.845 23.47 0.025 0.586
14 02 28 5.845 34.16 0.010 0.341
La

15 04 60 6.845 46.55 0.020 0.937


16 02 32 7.845 61.54 0.010 0.615
17 01 17 8.845 78.23 0.005 0.391
P P P P
fi = 200 xi fi = 1631 2
pe = 1 E = (xi x)2 p = 8:893

Table 5.7: Counting Statistics for Preset time = 20 seconds

We can easily calculate the sample mean as in Eq(1.3) as


LABORATORY MANUAL FOR NUCLEAR PHYSICS 162

P
xi fi
1631
x = P == 8:155
fi 200
The corresponding sample variance can be calculated from the expectation value of the square of the
deviations from the sample mean
X
E2 = (xi x)2 p = 8:893
or

s
E = 2:98

sic
Our theoretical expectation is that the distribution of the counts would follow a Poisson distribution.

hy
However, we do not know the mean of the distribution. We can estimate the mean of the parent
distribution by taking it to be the sample mean given above. This mean that

rP
 = x
lea
Then we expect that the probability distribution function to be (Eq(1.23))
uc
xi
PPoisson  P (xi ; ) = e 
xi !
lN

We can tabulate these values also


xi PPoisson pe
ua

2 0.009 0.025
3 0.025 0.025
an

4 0.052 0.075
bM

5 0.080 0.060
6 0.117 0.115
7 0.136 0.125
La

8 0.139 0.105
9 0.130 0.150
10 0.100 0.130
11 0.076 0.075
12 0.051 0.045
13 0.032 0.025
14 0.010 0.010
15 0.010 0.020
16 0.010 0.010
17 0.003 0.005
LABORATORY MANUAL FOR NUCLEAR PHYSICS 163

Table 5.8: Experimental and Poisson probabilities: Preset Time 20 seconds

The standard deviation of the Poisson distribution is simply


p
Poisson =  = 2:85
The graph of the experimental data and the theoretical Poisson distribution with the same mean is
shown in Figure 5.6.

s
sic
GM Counting Statistics: Preset Time 20 seconds
0.18
Poisson

hy
data
0.16 Gaussian

rP
0.14

0.12

0.1

0.08
lea
p

uc
0.06

0.04
lN

0.02

0
ua

-0.02
2 4 6 8 10 12 14 16 18
x
an
bM

Figure 5.6: Counting Statistics : Preset 20 seconds

The value of the reduced 2 is simply


La

8:893
~2 = = 1:09
8:155
and of 2 is

2 = 217
The tables tell us that the value of p is around 0:2 which is a reasonably good t to the assumed
theoretical model. The exact value is 0:182.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 164

s
Figure 5.7: p-value for  = 199 and 2 = 217

sic
We can see that the theoretically expected Poisson distribution and the experimental distribution are

hy
fairly similar though there are deviations. From our theoretical considerations (Section 1.4.3), we know
that when the mean of the Poisson distribution is large, the distribution tends towards a Gaussian or

rP
normal distribution. In Figure 5.6, we have also plotted the Gaussian distribution with the sample
mean as the mean and the sample standard deviation as the standard deviation. We can see the
lea
similarities or di erences between the experimental distribution and the normal distribution. Indeed
one sees that the normal distribution approximates the experimental distribution much better than the
uc
Poisson distribution.
lN

5.4 Questions
ua

1. Describe the working of a GM counter.


an

2. What are the various regions of a GM characteristic curve?


3. What is the meaning of threshold region, knee region, plateau region and breakdown
bM

region? Explain the reasons behind their formation.


4. With reference to the experiment, explain the concept of threshold voltage, knee
La

voltage, Operating voltage and breakdown voltage.


5. What does one mean by primary and secondary ionisation?
6. Explain the formation of a Townsend avalanche?
7. What is quenching (internal as well as external) and how does it work?
8. What are statistical uctuations and what is their relationship to the number of
events?
9. What are the two kinds of errors encountered in any experiment? What is the im-
portance of statistical errors on the measurements?
Chapter 6

Experiment: GM Counter: Counting

s
Eciency for & rays

sic
hy
rP
Things to know before you do the experiment
lea
1. All the concepts in Chapter 5 .
uc
2. Concept and importance of eciency and its types (intrinsic and extrinsic).
lN

3. Concept of solid angle.


ua

4. Error propagation and estimating errors on the derived quantities.


an
bM

6.1 Introduction
We have already studied the GM counter and recorded its characteristics. These include nding the
operating voltage for the GM counter by plotting its characteristic and taking the middle point of the
La

at plateau region. We have also determined its dead time. In this experiment we will study the
eciency of the GM counter for detecting gamma and beta radiation. As already discussed
in Section 4.4.5, alpha particles are dicult to detect with the GM counters available to us because
of the thickness of the window (which needs to be thick for structural reasons to be able to withstand
the di erence in pressure inside and outside the tube). Beta particles on the other hand have a larger
penetrating power and so can penetrate the window and being charged can easily cause an ionising
event in the tube which leads to a Gieger discharge. Gamma rays on the other hand, rarely interact
directly with the ll gas in the tube but instead cause the emission of a photoelectron from the inner
walls of the tube which leads to an avalanche as explained in Section 4.4.5.

165
LABORATORY MANUAL FOR NUCLEAR PHYSICS 166

6.2 Experiment
6.2.1 Purpose
The purpose to this Experiment is To estimate the eciency of the GM counter for the detec-
tion of and rays. We determine the eciencies for a set of various and rays and study the
dependence of eciency as a function of distance between the source and the counter.

To determine the eciency of the detector, we clearly need to measure the number of particles
emitted by the source and the number detected. The eciency can be de ned in two ways, Absolute

s
Eciency, Abs which does not take into account geometric factors like the distance from the source

sic
etc. (Eq(4.16)) and Intrinsic eciency , int which does take into the account the number of parti-
cles which actually enter the counter, Eq(4.17). For gamma ray sources, since the emission is roughly

hy
isotropic, we need to take into account the number of particles actually striking the counter window

rP
while for beta rays, we assume that all the particles emitted by a source are actually striking the window.

For this experiment, we would need a GM counter setup, a range of beta and gamma sources, which
lea
in our case are 60Co; 57Co; 133Ba; 204Tl and 147Pm. We would also need some aluminum sheets.
uc
The decay schemes for these sources are given below.
lN
ua
an
bM
La

Figure 6.1: Decay scheme for 60Co


LABORATORY MANUAL FOR NUCLEAR PHYSICS 167

s
Figure 6.2: Decay scheme for 57Co

sic
hy
rP
lea
uc
lN

Figure 6.3: Decay scheme for 133Ba


ua
an
bM
La

Figure 6.4: Decay scheme for 204Tl


LABORATORY MANUAL FOR NUCLEAR PHYSICS 168

s
Figure 6.5: Decay scheme for 147Pm

sic
6.2.2 Method

hy
As mentioned above, we need to basically know three numbers- one, the total number of particles

rP
emitted by the source; the number of particles detected by the source and nally, the number of
particles entering the detector (which, as discussed in Section 4.4.5 may be di erent from the number
lea
emitted.). The total number of particles emitted by the source depends on the activity of the source.
The activity of a source is de ned as the rate of decay of the radionuclides in a source and is of course
uc
a time dependent quantity as we know from Section 2.1.2. The activity follows an exponential law
(Eq(2.2))
lN

R(t) = R(0)e t
ua

where  is the decay constant. The decay constant is related to the half life T 12 by
an

ln 2
=
T 21
bM

For the sources mentioned above, we are given the activity of the source at the time of its fabrication,
R(0). The time of fabrication ti is also given. From this, we can determine the time elapsed since the
source with the given activity R(0) was fabricated. This time is t. Knowing the source, we know the
La

half life T 12 of the particular radioisotope. This allows us to calculate , the decay constant. Finally,
knowing R(0), t and , allows us to calculate the present activity R(t).

For the purely gamma sources, 133Ba; 57Co, we need to take into account the geometric factor also
as discussed in Section 4.4.5 since the sources are isotropic and the number of particles entering the
counter window will depend on the distance from the source and the radius of the counter window.
In addition to the pure gamma sources, we can also use 60Co which is a gamma and beta source. We
place an aluminum sheet between the source and the detector window which e ectively blocks all the
beta particles, its thickness being more than the range of the 0:514 MeV beta particle but is not thick
enough to signi cantly attenuate the gamma rays.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 169

For the pure beta sources, 204Tl; 147Pm, the geometry is not a signi cant issue since we assume all the
beta particles emitted enter the window. This is because the source is placed in a holder which has only
a small aluminum window through which the beta particles can escape. The source is otherwise enclosed
by thick plastic casing which absorbs the beta particles. Nevertheless, we will use the geometric
factor for determining the number of particles striking the window of the counter, instead
of taking it to be equal to the number emitted. Of course as we increase the distance between
the source and the counter, there is some absorption of the beta particles in the air.
We start by determining the operating voltage of the GM tube as discussed in Chapter 5 and choosing
an appropriate preset time, say 180 seconds. The radius of the GM counter window r is known to be

s
7:5 mm.We also determine the background count rate NB in the absence of all sources. With the GM

sic
counter set up at the appropriate voltage, we then use the pure gamma sources at various distances d,
and measure the count rate N . We repeat this for various distances for the gamma sources and beta

hy
sources. The detected count rate per second has to be decreased by the background count rate per

rP
second to get the net counts detected per second (cps). The counts emitted per second or the
number of decays per second (dps) which are hitting the counter window are obtained from the
calculation of the activity today R(t). The geometric factor is what we obtained in Equation 4.15 and
lea
so we get for the number of particles hitting the window as
uc
 
1 d
dps = R(t)  p
2 2 r + d2
2
lN

where the geometric factor is important for the gamma sources but not so important for the beta
sources. The eciency is then simply
ua

cps
=  100% (6.1)
an

dps
bM

6.2.3 Sample Data


Operating Voltage: 425 V
Preset Time: 180 seconds
La

nB = 85 counts
85
NB ( average) = = 0:47 counts per second
180
Error estimation and the correct method to determine the error in the eciency, is
discussed in the next section.

sources:
LABORATORY MANUAL FOR NUCLEAR PHYSICS 170

1. 133
Ba

R(0) 111 KBq


ti August 2013
t 1:833 years
T 12 10:5 years
 0:066 years 1
R(t) 98:35  103 Bq

s
r 0:75 cm

sic
Table 6.1: Data for 133Ba

hy
rP
 
R(t) Bq d (cm) dps = R(t) 12 2prd2 +d2 N
N (counts) cps = 180 cps  100%  
NB  = dps 
98:35  103 2.7 1794 lea 3045 16.45 0:91  0:02
98:35  103 4.7 614 1201 6.20 1:01  0:03
98:35  10 3 1:08  0:04
uc
7.0 279 629 3.02
lN

Table 6.2: Count rate and eciency data for 133Ba


ua

2. Co
57
an
bM

R(0) 76:2 KBq


ti August 2013
t 1:833 years
La

T 12 0:74 years
 0:936
R(t) 13:64  103 Bq
r 0:75 cm
Table 6.3: Data for 57Co
LABORATORY MANUAL FOR NUCLEAR PHYSICS 171

 
R(t) Bq d (cm) dps = R(t) 12 2prd2 +d2 N
N (counts) cps = 180 cps  100%  
NB  = dps 
13:64  103 2.7 240.1 239 0.86 0:35  0:02
13:64  103 4.7 83 117 0.18 0:21  0:02
13:64  10 3 7.0 38 101 0.091 0:23  0:02
Table 6.4: Count rate and eciency data for 57Co

3. Co with Aluminum sheet


60

s
sic
R(0) 17 KBq

hy
ti August 2013
t 1:833 years

rP
T 12 5:27 years
 0:131
lea
R(t) 13:37  103 Bq
r 0:75 cm
uc
Table 6.5: Data for 60Co
lN
ua

 
R(t) Bq d (cm) dps = R(t) 12 2prd2 +d2 N
N (counts) cps = 180 cps  100%  
NB  = dps 
an

13:37  103 2.7 235.3 1010 5.14 2:11  0:07


13:37  10 3 4.7 81.5 426 1.89 2:23  0:11
bM

13:37  10 3 7.0 37.4 277 1.07 2:81  0:17


Table 6.6: Count rate and eciency data for 60Co with aluminum sheets
La

sources:

1. 204
Tl
LABORATORY MANUAL FOR NUCLEAR PHYSICS 172

R(0) 0:11 Ci = 0:11  3:7  104 = 4070 Bq


ti August 2011
t 3:89 years
T 12 3:9 years
 0:178
R(t) 2036:48 Bq
r 0:75 cm
Table 6.7: Data for 204Tl

s
sic
hy
 
R(t) Bq d (cm) dps = R(t) 12 2prd2 +d2 N
N (counts) cps = 180 cps  100%  
NB  = dps 
28:22  0:64

rP
2036:48 2.7 35.84 1972 10.48
2036:48 4.7 12.42 579 2.74 21:59  0:89
2036:48 7.0 5.70 lea 406 1.78 30:81  2:16
Table 6.8: Count rate and eciency data for 204Tl
uc
lN

2. 147
Pm
ua

R(0) 13 KBq
an

ti Decemeber 2014
t 0:493 years
bM

T 12 2:6 years
 0:266
R(t) 11:40  103 Bq
La

r 0:75 cm
Table 6.9: Data for 147Pm

 
R(t) Bq d (cm) dps = R(t) 12 2prd2 +d2 N
N (counts) cps = 180 cps  100%  
NB  = dps 
11:40  103 2.7 200.6 3238 17.52 8:42  0:15
11:40  10 3 4.7 69.5 907 4.566 6:42  0:21
11:40  103 7.0 31.9 341 1.420 4:39  0:24
LABORATORY MANUAL FOR NUCLEAR PHYSICS 173

Table 6.10: Count rate and eciency data for 147Pm

From the data and the results obtained above, we see that the eciency of the counter for rays is
indeed very small as expected while that for the particles is much higher. Theoretically, we expect an
almost 100% eciency for beta particles. There could be several reasons why this was not seen in this
experiment. One, the eciency depends on the dead time and that might have reduced the eciency for
some source. This would be particularly noticeable for high activity sources where the time between the

s
particles entering the detector is small. Secondly, as the distance increases, the probability of the beta

sic
particles being absorbed in the air or being scattered o increases. This also might be a contributing
factor which our experiment does not take care of. It only takes care of the distance via the geometric

hy
factor.

rP
6.2.4 Error Estimation lea
The experiment aims to determine the eciency  of the GM counter in measuring the counts for beta
uc
and gamma rays. We calculate the eciency by nding the counts per second and the detected counts
per second as in Eq(6.1). Clearly, both these quantities are derived quantities and so to estimate the
lN

errors in these, we need to estimate the errors in the measured quantities from which they are derived
and then use the error propagation equation Eq(1.47).
ua

We have
an

cps
=
bM

dps
or
c
La

e=
dp
Thus
2
e2 c2 dp
= +
e2 c2 dp2
Let us consider the numerator rst.
N nB
c=
180
Each of these quantities is a measured quantity, that is the total number of counts in 180 seconds
with the source and without the source (with the background only). We have seen already in Chap 1
LABORATORY MANUAL FOR NUCLEAR PHYSICS 174

that the errors will be simply the square root of the counts. Thus
p
N2 + ( nB
2 )
c =
180
We assume that there is no uncertainty in the time measurement. Then
p
N = N
p
nB = nB
Therefore we get

s
sic
N + nB
c2 =
1802

hy
and

rP
N +nB
c2 N + nB
2 = (N180n2B )2 =
c 1802 (N nB )2
lea
What about the denominator? Here again, the quantity is a derived quantity and we need to estimate
the error in each of the measured quantities. We have
uc
 
1 d
dps = dp = R(t) p
lN

2 2 d2 + r2
ua

The error calculation for dps is a little more complicated. Let us use our knowledge of error
propagation and see how it works. If we were to use the standard formulae for error propagation in
an

products and quotients etc. as discussed in the discussion around Eq(1.53) and Eq(1.54), then we
bM

will need to rst write

dps = R(t)B
La

where
 
1 d
B= p
2 2 d2 + r2
Now by Eq(1.53), we have
2
dp R2 B2
= +
dp2 R2 B 2
But we know that R(t), which we have calculated from the original activity and the time elapsed
since the fabrication of the source does not have any error since we assume we know these quantities
LABORATORY MANUAL FOR NUCLEAR PHYSICS 175

precisely. Thus
2
dp B2
=
dp2 B 2
Now
 
1 d 1 G
B=
2
p 2 2 =
2 2
2 d +r
Using Eq(1.49), we have

s
sic
1
B2 = G2
4

hy
or

rP
2
dp G2
=
dp2 4B 2
To calculate G2 , we have
lea
p 2d F
uc
G= =
d + r2 J
lN

where
p
ua

F = d; J = d2 + r2
an

Using Eq(1.54), we have


G2 F2 J2
bM

= +
G2 F 2 J 2
But F = d and so
La

F2 = d2

G2 d2 J2


= +
G2 d2 J 2
What about J ? Here we have
p p
J = d2 + r2 = M
with
M = d2 + r2
LABORATORY MANUAL FOR NUCLEAR PHYSICS 176

Using Eq(1.55), we have


J M
=
J 2M
or
J2 2
M
=
J 2 4M 2
Writing

s
sic
M = d2 + r2 = P + S

hy
Now r, the radius of the GM counter window also is speci ed by the manufacturer and so we
assume no error in it. Thus

rP
M = P
But
lea
uc
P = d2
lN

Thus, using Eq(1.55), we have


 P 2 d
ua

=
P d
or
an

4P 2 d2
P2 = = 4d2 d2
bM

d2
So
La

2 =  2 = 4d2  2
M P d

But
J2 2
M 4d2 d2 d2 d2
= = =
J 2 4M 2 4(d2 + r2 )2 (d2 + r2 )2
and so
G2 d2 J2
= +
G2 d2 J 2
gives us
LABORATORY MANUAL FOR NUCLEAR PHYSICS 177

G2 d2 d2 d2


= +
G2 d2 (d2 + r2 )2
But
2
dp G2
=
dp2 4B 2
Putting in the value of G2 , we get
2  
dp G2 G2 d2 d2 d2

s
= = +
dp2 4B 2 4B 2 d2 (d2 + r2 )2

sic
or

hy
2   
dp d2 d2 1 d2
= +
dp2 4B 2 d2 + r2 d2 (d2 + r2 )2

rP
Putting in the value of B , we get
2
dp
=
d2

d2

1
lea
d2
+ 2 22 h

1
dp2 4 d2 + r2 d (d + r ) 1 p d i2
2
uc
2 2 d2 +r2
Finally, we get the expression for the error in the eciency as
lN

2
e2 c2 dp
= +
e2 c2 dp2
ua

or
an

  
e2 N + nB d2 d2 1 d2 1
= + + i2
e2 (N nB )2 4 d2 + r2 d2 (d2 + r2 )2 h 1
bM

d
2 2pd2 +r2
However, there is a problem with this method in this case. The issue is basically
that whenever we have expressions like the one above, involving the same variable
La

in the numerator and the denominator, then the above method of calculating error
misses out on the cancellation of errors between the numerator and the denominator.

To see this, consider a simple case of a function of three variables, u; v; x as


u+v
f (u; v; x) =
u+x
Now suppose all of these variables are positive numbers. Now suppose we overestimate the error
in one of them, say u. Now this overestimate will a ect BOTH the numerator and denominator
and these overestimates will cancel each other to a large extent. Similarly if we underestimate the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 178

error in u, it will a ect both the numerator and denominator and these underestimates will cancel
each other largely. This kind of cancellation is missed when we do the error calculation by the above
method of splitting the function f (u; v; x) by repeated use of di erent error formulae Instead, what
we need to do is to use the basic error propagation equation to get the correct error.
Let us see how this is done.
We know that
c
e=
dp

s
sic
Thus 2
e2 c2 dp
= +
e2 c2 dp2

hy
The error in the numerator c is not a problem and we have already computed the error in c. This

rP
is simply
N +nB
c2 1802 = N + nB
lea
2 = n B )2
c ( N
1802 (N nB )2
uc
The problem is with the error in the denominator. For the error in dp, we have
 
lN

1 d
dp = R(t) p
2 2 d2 + r2
or
ua
an

dps = R(t)B
bM

where
 
1 d
B= p
2 2 d2 + r2
La

We can use the error formula for the product (Eq(1.53)) here since again, the two functions B
and R are independent. There is no error in R(t). So we have
2
dp B2
=
dp2 B 2
 
1 d 1 G
B= p
2 2 d +r 2 2 =
2 2
For nding the error in B , again there is no problem in using the expression for weighted sums
(Eq(1.51)). Using this, we have
LABORATORY MANUAL FOR NUCLEAR PHYSICS 179

1
B2 = G2
4
or
2
dp G2
=
dp2 4B 2
where

p 2d

s
G=
d + r2

sic
To nd the error in G, we might think of using the error formula for division (Eq(1.54)). But as

hy
noted above, this will give us the wrong result since both the numerator and the denominator contain
the variable d and doing this, we will miss out on the errors cancelling out in both the numerator

rP
and denominator. To calculate G , we use the basic error propagation equation Eq(1.47). We know
that there is no error in r. Thus
Then
lea
@G 2
 
uc
2 2
G = d
@d
lN

But
   
ua

@G d2
@d
= p 21 2 (d2 + r2 )3=2
d +r
an

Thus
bM

 2
d2
G2 = d2 p 21 2 (d2 + r2 )3=2
d +r
or
La

 2
1 d2
B2 = d2
4
p 21 2 (d2 + r2 )3=2
d +r
Finally
2  2
dp B2 1 d2
dp 2 =
B 2 = 2 d2
4B
p 21 2 (d2 + r2 )3=2
d +r
or
2
e2 c2 dp
= +
e2 c2 dp2
LABORATORY MANUAL FOR NUCLEAR PHYSICS 180

that is h i2
e2 d2
e2 = (NN nB
n B )2 + 4B1 2 d2 pd21+r2 (d2 +r2 )3=2 (6.2)
where
 2
1 d
B2 = p
2 2 d2 + r2

s
The only estimation we need therefore to determine the error in the eciency  is the error in d.

sic
We assume that the error is equal to the least count of the scale used to measure the distance, namely
0:1 cm. As an example, consider the measurement for the beta source 204Tl. We have

hy
N = 1972

rP
nB = 85
lea
r = 0:75 cm
d = 2:7 cm
uc
d = 0:1 cm
lN

 = 0:2822
Then a calculation of e2 gives us
ua
an

e2 = 0:00004
bM

or

e = 0:0064
La

Thus we have for this case

R  = 28:22  0:64% (6.3)

A similar analysis can be carried out for all the other readings to get an estimate of the error involved
in our experiment.
It is easy to write a program for instance in C language to do the calculation for the eciency and
the error in the eciency. A sample program is given below
LABORATORY MANUAL FOR NUCLEAR PHYSICS 181

#include < s t d i o . h>


#include <math . h>
main ( )
f float se , e , n , nb , sd =0.1 , d , r =0.75 , se1 , se2 , x , y , z ;
float r t , d1 , c , b ;
p r i n t f ( " i n p u t N" ) ;
s c a n f ( "%f " ,&n ) ;
nb = 0 . 4 7 ;
p r i n t f ( " i n p u t d" ) ;
s c a n f ( "%f " ,&d ) ;

s
p r i n t f ( " i n p u t R" ) ;

sic
s c a n f ( "%f " ,& r t ) ;
c=(n / 1 8 0 . 0 ) nb ;

hy
y= d / ( 2 . 0  s q r t ( d  d+r  r ) ) ;
d1=0.5 y ;

rP
d=r t  d1 ;
e=c /d ;
p r i n t f ( "N=%f n t NB= %f n t d=%f n t R( t ) = %f n t c= %f n t e = %f n n" , n , nb , d , r t , c , e ) ;
p r i n t f ( " input e p s i l o n " ) ;
s c a n f ( "%f " ,& e ) ;
lea
x = ( n+nb ) / ( ( n nb )  ( n nb ) ) ;
uc
z =1/( s q r t ( d  d+r  r ) ) ;
s e 1=x+(sd  sd / ( 4  d1  d1 ) )  ( z d  d  z  z  z )  ( z d  d  z  z  z ) ;
lN

s e 2=s e 1  e  e ;
s e=s q r t ( s e 2 ) ;
p r i n t f ( " e = %f n t s e= %f " , e , s e ) ;
ua

g
an
bM

6.3 Questions
1. How does one de ne eciency and what is meant by intrinsic and extrinsic eciency.
Which of these eciencies is a more practical quantity?
La

2. Why are the eciencies of a given GM counter for and rays not the same?
3. How does the eciency of a GM counter depend on the distance between the source
and detector?
4. What are the sources of systematic errors in this experiment?
5. How is the error in the eciency evaluated given that it is a derived quantity?
Chapter 7

Experiment: Absorption of rays in Iron

s
sic
hy
Things to know before you do the experiment

rP
1. All the concepts in Chapter 5 .
lea
2. The interaction of rays with matter and the dependence of these interaction
processes on the energy of the ray.
uc
3. The concept of mass thickness.
lN

4. Knowledge and use of semi-log graphs.


ua

5. Concept of intensity of radiation and probability of interaction.


an
bM

7.1 Introduction
We have already seen how gamma rays interact with matter in Chapter 3, Section 3.2.3. We saw there
La

that at energies of  MeV, the three processes of Photoelectric absorption, Compton E ect and Pair
Production are operative. The relative importance of these depends on the energy of the gamma rays
(Figure 3.14). We also saw that when gamma rays pass through an absorber, there is an exponential
attenuation of the incoming beam (Section 3.1.1). The number of transmitted particles after crossing
a distance x is (Eq(3.1))

N = N0 e nx = N0 e x (7.1)


where  is called the linear attenuation coecient. Clearly, this attenuation coecient, being
a product of the cross section  and n, the number of scatterers per unit volume, will depend on the
density  of the scattering material. A more convenient measure to quantify absorption is the mass

182
LABORATORY MANUAL FOR NUCLEAR PHYSICS 183

attenuation coecient which is simply  . Consequently, we replace Eq(7.1)by


x
N = N0 e  (7.2)
where x is the mass thickness of Eq(3.3). This, as we have seen is much easier to measure and
also much more informative in comparing di erent absorbers.

We can de ne a quantity called the half thickness, or the mass thickness where the intensity of the
incoming gamma rays is reduced to one half the initial value. Thus

s
N
N = 0 = N0 e m d1=2

sic
(7.3)
2
where m is the mass attenuation coecient and d1=2 is the half thickness of that particular

hy
absorber for the particular energy gamma rays used. Clearly

rP
0:693
d1=2 = (7.4)
m

7.2 Experiment
lea
uc
7.2.1 Purpose
lN

The aim of the experiment is to nd the half thickness of monoenergetic gamma ray source 137Cs, which
emits gamma rays with energy 0:6617 MeV, using iron plates of varying thickness.
The experiment uses a GM counter, a gamma ray source, 137Cs and iron plates of varying thickness.
ua
an

7.2.2 Method
bM

As with all experiments using a GM counter, the rst step is always to nd the operating voltage of
the counter by plotting its characteristic as in Chapter 5. We also obtain the background counts for a
La

xed preset time, say 180 seconds.

We then put the source in the lead stand and note down the number of counts during the preset time.
The lead absorber stand is then placed between the source and the counter window and an increasing
number of iron plates of known thickness are placed in the absorber stand and the counts noted. A
plot of the logarithm of the net number of counts versus the mass thickness of the absorbers gives us a
straight line whose slope allows us to calculate m and d1=2 .
LABORATORY MANUAL FOR NUCLEAR PHYSICS 184

7.2.3 Sample Data


Operating Voltage = 425 V
Preset Time = 180 seconds
Background counts NB = 97 in 180 seconds
Density of Iron = 7:86 gm cm 3
Atomic Number of Iron = 56

Mass Thickness (ti = xi ) gm cm 2 Counts (Ni ) Net Counts (Ni


p
NB ) Error ( Ni + NB )

s
sic
0 337 240 20.83
2.28 273 176 19.23

hy
3.76 280 183 19.41
4.60 250 153 18.62

rP
6.08 250 153 18.62
8.36 216 119 17.69
9.86
11.41
192
172
lea 95
75
17.00
16.40
uc
13.92 173 76 16.43
15.26 167 70 16.24
lN

16.20 179 82 16.61


19.12 141 44 15.42
ua

23.72 132 35 15.13


25.29 117 20 14.62
an

27.61 131 34 15.09


bM

Table 7.1: Count rate and thickness data for 137Cs with iron plates
La

We need to plot the graph between ln(N NB ) and t = x. This, as we expect will be straight line.
However, given the statistical nature of the data, we would need to nd the best t straight line using
the Method of Least Squares (Section 1.7, Chapter 1).

To determine the slope and the intercept of the best t straight line, we need to use Eq(1.73) and
P P P P P P P
Eq(1.74) and therefore need xi = ti ; yi = ln(Ni NB ); x2i = t2i and xi yi =
P
ti ln(Ni NB ). We can calculate them as below.

xi = ti x2i = t2i yi = ln(Ni NB ) xi yi = ti ln(Ni


p
NB ) ln( N + NB )
0 0 5.48 0 3.03
LABORATORY MANUAL FOR NUCLEAR PHYSICS 185

2.28 5.10 5.17 11.79 2.58


3.76 14.13 5.21 19.59 2.60
4.60 21.16 5.03 23.14 2.51
6.08 36.97 5.03 30.58 2.51
8.36 69.89 4.78 39.96 2.39
9.86 97.22 4.55 44.86 2.28
11.41 130.19 4.32 49.29 2.16
13.92 193.77 4.33 60.83 2.17

s
15.26 232.83 4.25 64.09 2.12

sic
16.20 262.44 4.41 71.44 2.20
19.12 365.57 3.78 72.27 1.89

hy
23.72 562.64 3.46 82.07 1.78

rP
25.29 639.58 2.99 75.62 1.50
27.61 762.31 3.53 97.46 1.76
P P 2 P P
xi = 187:47 xi = 3393:80 yi = 66:34
lea xi yi = 742:99

Table 7.2: Least Square Fitting for graph of ln(N NB ) vs t


uc
lN

The number of observations, N = 15.


ua

With these values, we can compute the best t straight line using Method of Least Squares. The results
an

are
bM

m = 0:0819; c = 240
The graph of N NB vs t is plotted on a semi-log scale with the error bars as indicated. We have also
La

drawn the best t line with the above mentioned slope and intercept.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 186

1000
Gamma Ray absorption
data
best fit

100
N-NB

s
sic
10

hy
rP
1
0 5 10 15 20 25 30
t (gm cm-2)

lea
Figure 7.1: ray absorption
uc
With the given slope, we can calculate the mass attenuation coecient m , which is precisely
lN

the slope calculated as 0:0819 gm cm 2 . The half thickness can be calculated from Eq(7.4) as
d1=2 = 0:693 = 8:55 gm cm 2 .
ua

What about the error in our calculation? The error in N NB is clear. We p know that a weightedpsum
an

or di erence has an error given by Eq(1.51). The error in N is simply N and that in NB is NB .
p
bM

These two quantities are uncorrelated and hence the overall error in N NB will be N + NB . The
p
error bars are then simply ln( N + NB ). But what about the errors in the slope and the intercept of
the graph which is what we are using to determine the desired derived quantity, d1=2 ? As you would
La

recall from Section 1.7, the uncertainty in the slope determined by this method can be calculated using
Eq(1.77). Since we have the values of m and c, for each xi = ti , we calculate y = mxi + c . We also
P P
have the observed yi as well as the values of  = N x2i ( xi )2 . Putting it all together, we then
know the uncertainty in y which is given by
rP
mxi c)2
(yi
y = (7.5)
N 2
With this uncertainty, we can calculate the uncertainty in the slope determined by the Least Square
method as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 187

r s
N N P
m = y = y P 2 (7.6)
 N xi ( xi )2
The values for the uncertainty in the slope turns out to be

m = 0:00018
Thus, our value for the mass attenuation coecient is

m = 0:0819  0:0002

s
sic
The corresponding error in the half thickness can also be calculated using Section 1.5.1 is

hy

d = d1=2 m = 0:017
m

rP
Thus we can quote the half thickness as

R lea
d1=2 = 8:55  0:02 gm cm 2 (7.7)
uc
7.3 Questions
lN

1. What is half thickness and how it is measured?


ua

2. How does the half thickness depend upon the energy of the incoming rays, intensity
of the rays, nature of the absorber and nature of the source?
an

3. What are the various ways in which radiation can interact with matter?
bM

4. How does the relative dominance of each of these processes vary with the energy of
the gamma rays?
La

5. How does the Compton e ect cross section vary with the energy of the incoming
gamma rays?
6. Can a gamma ray photon of energy 5 MeV in space produce an electron-positron pair?
If not, why not?
7. How do we estimate the errors on the measurements of derived quantities (error
propagation).
Chapter 8

Experiment: Veri cation of the Inverse

s
Square Law for rays

sic
hy
rP
Things to know before you do the experiment
lea
1. All the concepts in Chapter 5 .
uc
2. Concept of solid angle.
lN

3. Error propagation and estimating errors on the derived quantities.


ua

4. The inverse square law for radiation.


an
bM

8.1 Introduction
The intensity of radiation from a point source emitting isotropically is known to fall of as the square of
the distance from the source. We know that this is a purely geometric e ect since the area of the surface
La

increases as d2 and the same energy therefore spreads over four times the area leading to a decrease in
intensity by a factor of four. This can be clearly seen in Figure 8.1.

188
LABORATORY MANUAL FOR NUCLEAR PHYSICS 189

s
sic
Figure 8.1: Inverse Square Law for radiation x

hy
x(Source: http:// hyperphysics.phy-astr.gsu.edu/ hbase/ forces/ isq.html )

rP
We expect that gamma rays being electromagnetic radiation will exhibit similar behaviour.

8.2 Experiment
lea
uc
8.2.1 Purpose
lN

The objective of this experiment is to determine the relationship between the intensity
of gamma radiation and the distance from the source to the detector. We expect that the
ua

inverse square law would be valid.


an

8.2.2 Method
bM

If we have a point source of gamma radiation, which we assume emits isotropically at a rate of N0
particles per second, and we observe the intensity, I at a distance r from the source, we expect a
La

relationship like
N0
I= (8.1)
4r2
If we use a long lived source such that N0 is constant through the duration of the experiment, we then
will see that as we increase r, the intensity will go down as the r12 . That is a graph of I vs r12 will be
a straight line with a slope N40 . Alternatively, a graph of log I vs log r will give us a straight line with
slope 2.

We use a variety of long lived gamma sources and determine the relationship of the intensity and
distance.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 190

The sources we use are 137Cs, 60Co, 22Na and 133Ba. The decay schemes for these are given below.

s
sic
hy
Figure 8.2: Decay scheme for 137Cs

rP
lea
uc
lN
ua

Figure 8.3: Decay scheme for 60Co


an
bM
La

Figure 8.4: Decay scheme for 133Ba


LABORATORY MANUAL FOR NUCLEAR PHYSICS 191

s
Figure 8.5: Decay scheme for 22Na

sic
We also need a GM counter to measure the intensity at various distances as well as a scale to measure

hy
the distance from the source to the counter window.

rP
As with all experiments which use a GM counter, we rst need to determine the GM characteristics
and nd the operating voltage. We then choose a preset time and determine the background radiation
lea
rate so that it can be subtracted from the count rate with the sources to get the net or true count
rate. Finally, with each of the sources placed at di erent distances, the count rate is measured and
uc
tabulated to see if the inverse square law is valid or not.
lN

8.2.3 Sample Data


ua

Operating Voltage = 425 V


Preset Time = 60 seconds
an
bM

Background Count

S.No. NB (NB NB ) (NB NB )2


La

1 18 -0.8 0.64
2 17 -1.8 3.24
3 20 1.2 1.44
4 19 0.2 0.04
5 20 1.2 1.44
P
NB = 18:8 (NB NB )2 = 6:80
Table 8.1: Background Count rate for 60 seconds

With these 5 values of the background counts, we determine the mean to be 18:8 counts in 60 seconds.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 192

What about the error in the background count rate? Once again, we could think of using the standard
deviation Eq(1.63) to determine the error in the mean. For the  of the parent distribution, we can use
the sample standard deviation s given by
rP
(x x)2
s=
N 1
In our case, we get s = 1:30. From this, we can estimate the error in the mean (from Eq(1.63)) to be

 = ps
N

s
or

sic
1:3
p

hy
NB = = 0:58
5
Thus one would think that one should report the result as

rP
NB = 18:80  0:58
lea
However, there is a subtle point here that one needs to understand. Let us see what we are
doing- we are taking the total number of background counts in 60 seconds and reporting
uc
it as a number. Then we are repeating the same procedure 5 times. Note that in each of
lN

the values of NB , there is an error. This is the inherent statistical error that is associated
with the event which as we know is a result of a Poisson distribution. We can therefore
think of each value of NB as a mean of a Poisson distribution. The associated error with
ua

p
each value of NB is thus NB . Therefore, the correct procedure to exhibit this inherent
an

statistical error would be to take the expression for the mean value of NB , that is NB and
apply the appropriate error prorogation equation to it. In our case this means
bM

5
P
(NB )i
i=1
La

NB =
5
1  2 
N2 B = (NB )1 + (2NB )2 +    + (2NB )5
25
1
= [18 + 17 + 20 + 19 + 20]
25
94
=
25
p
94
NB =
5
= 1:93 (8.2)

Thus our background count should be reported as


LABORATORY MANUAL FOR NUCLEAR PHYSICS 193

NB = 18:8  1:9

137
Cs source with Activity  3.1  Ci

qN +2
d (cm) Counts (N) N NB Error= 60 NB
2.0 8728 8709.2 1.55
2.5 4858 4839.2 1.16

s
3.0 2976 2957.2 0.91

sic
3.5 1969 1950.2 0.74

hy
4.0 1418 1399.2 0.63
4.5 1061 1042.2 0.55

rP
5.0 867 839.2 0.50
5.5 711 692.2 0.45
6.0
6.5
572
485
553.2
466.2
lea 0.41
0.37
7.0 407 388.2 0.34
uc
Table 8.2: Measured Count rates at various distances from 137Cs with error estimates
lN
ua

d (cm) Counts (N) N NB yi = R = N 60NB xi = d12 m 2 x2i xi yi


625  104
an

2.0 8728 8709.2 145.15 2500 362875


2.5 4858 4839.2 80.65 1600 256  104 129040
bM

3.0 2976 2957.2 49.28 1111 123:4321  10 4 54751


3.5 1969 1950.2 32.50 816 66:5856  10 4 26520
4.0 1418 1399.2 23.32 625 39:0626  104 14575
La

4.5 1061 1042.2 17.37 493 24:3049  10 4 8563


5.0 867 839.2 13.98 400 16  10 4 5592
5.5 711 692.2 11.53 330 10:89  104 3937
6.0 572 553.2 9.22 278 7:7284  10 4 2563
6.5 485 466.2 7.77 236 5:5696  10 4 1834
7.0 407 388.2 6.47 204 4:1616  104 1320
P P P 2 4 P
yi = 397:29 xi = 8593 xi = 1178:74  10 xi yi = 611571
Table 8.3: Count rates at various distances from 137Cs & Least Square tting
LABORATORY MANUAL FOR NUCLEAR PHYSICS 194

We can plot the curves for R vs d as shown in Figure 8.6.

160
Rate vs distance
data
best fit
140

120

100

s
Rate

sic
80

60

hy
40

rP
20

0
2 3 4
lea
d (cm)
5 6 7
uc
Figure 8.6: Graph of R vs d
lN

We can also plot a graph of R vs d12 using the Method of Least Squares. Using Eq(1.73) and Eq(1.74),
ua

we get
an

m = 0:0593; c = 10:250
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 195

160
Rate vs 1/d^2
data
best fit
140

120
Rate counts/sec

100

80

s
sic
60

40

hy
20

rP
0
0 500 1000 1500 2000 2500
1/d^2 (m)^-2

lea
Figure 8.7: Graph of R vs d12
uc
lN

Finally, we need to compute the error in our estimation of the slope of this graph. In Section 1.7, we
ua

saw that the uncertainty in the slope determined by this method can be calculated using Eq(1.77).
Since we have the values of m and c, for each xi , we calculate y = mxi + c . We also have the observed
an

P P
yi as well as the values of  = N x2i ( xi )2 . Putting it all together, we then know the uncertainty
in y which is given by
bM

rP
mxi c)2 (yi
y = (8.3)
N 2
La

With this uncertainty, we can calculate the uncertainty in the slope determined by the Least Square
method as
r s
N N P
m = y = y P 2 (8.4)
 N xi ( xi )2
Putting in the values, we get a value of

m = 0:002
We can thus quote the result for the slope of the best t line as
LABORATORY MANUAL FOR NUCLEAR PHYSICS 196

m = 0:0593  0:002
A similar exercise would need to be done with all the other 3 sources and similar graphs would be
obtained.

8.3 Questions
1. Why do rays follow the inverse square law?

s
sic
2. Do rays follow the inverse square law? If not, why not?

hy
3. Do all kinds of radiation follows the inverse square law?
4. Is the inverse square law valid in vacuum only or in matter also?

rP
5. What could be the sources of systematic errors in this experiment?
lea
uc
lN
ua
an
bM
La
Chapter 9

Experiment: To Determine the Range of

s
rays in Aluminum and to determine the

sic
hy
End Point Energy

rP
lea
Things to know before you do the experiment
uc
lN

1. All the concepts in Chapter 5 .


2. Concept of solid angle.
ua

3. The decay process and its description with the help of Fermi theory of decay.
an

4. The concept of continuous energy spectrum of particles and the end point energy.
bM

5. Error propagation and estimating errors on the derived quantities.


La

9.1 Introduction
In Chapter 3, we saw how electrons, both mono-energetic and those which are produced in nuclear
beta decay interact with matter. Section 3.2.2, we saw that for a beam of mono-energetic electrons the
energy loss of electrons is much smaller than that of heavy charged particles of the same energy. This
means that they have a much larger range. What is observed experimentally is that for a wide variety
of absorber materials, the product of the range and the density of the absorber is a constant for any
particular electron energy.

The situation is very di erent for the beta particles emitted by a radioactive source. This is because, as

197
LABORATORY MANUAL FOR NUCLEAR PHYSICS 198

we have seen in Section 2.2.2 in Fig 2.2, the energy spectrum of the beta particles is continuous. What
is seen therefore is that the beta particles at the lower end of the spectrum are absorbed even with
a very thin absorber. However, for most part of the spectrum, the transmission of the beta particles
follows an exponential curve with thickness. This is an experimental fact which cannot be easily derived
from fundamental physics. What we see is that the counting rate (or intensity) falls o exponentially
with an attenuation coecient which depends on the end point energy of the beta particle.

C = C0 e d (9.1)
where C is the counting rate with the absorber material, C0 is the counting rate without the absorber

s
sic
and d is the mass thickness in units of mass per unit area. The coecient  is the attenuation
coecient. Thus, a graph between ln CC0 and d would give us a straight line whose slope will be

hy
the attenuation coecient. This behaviour is shown in Figure 3.10 where the at part of the curve
corresponds to the count rate going to the constant background value.

rP
We also know from Chapter 2, that the beta particle spectrum is a continuous one with a maximum
lea
energy determined by the Q value of the nuclear reaction producing the beta particle. Thus, we can infer
that beta particles from a radioactive source will have di erent penetrating power and the maximum
uc
penetration depth, the range will be for particles with the maximum or end-point energy. It turns out
that for aluminum absorbers, a single range energy relationship for both monoenergetic electrons and
lN

beta particles with energy range 0:01  E  2:5 MeV exists as was shown empirically by Katz and
Penfold (Reviews of Modern Physics, 24, page 28, 1952). They found that the relationship is given by
ua
an

R = 412E0n
n = 1:265 0:0954 ln E0 (9.2)
bM

Here E0 is in MeV and the range R is in units of mass thickness, that is mg cm 2 . Eq(9.2) allows
us to determine the end point energy of the beta rays from a radioactive source. If we can nd the
La

maximum range R0 experimentally, then that will give us the value of Emax which will be the end point
energy since the maximum energy will correspond to the maximum range.

9.2 Experiment
9.2.1 Purpose
In this experiment, we will study the absorption of rays in aluminum and investigate the exponential
attenuation of Eq(9.1). We will also determine the end point energy of beta rays from 90Sr using the
range-energy relationship Eq(9.2).
LABORATORY MANUAL FOR NUCLEAR PHYSICS 199

9.2.2 Method
For this experiment, we would need a GM counter, a radioactive source ( 90Sr), and aluminum absorber
foils of varying thickness.

We rst need to determine the operating voltage of the GM counter as in Chapter 5. We then determine
the background counts for a preset time of say 120 seconds. Next we take a 90Sr source with known
activity. The decay scheme for the source is given in Figure 9.1.

s
sic
hy
rP
lea
uc
lN
ua
an

Figure 9.1: Decay scheme for 90Sr


bM

We see that the dominant emission is with a maximum energy of around 2:28 MeV and so we can
assume that the range energy relation given above will be valid.
La

We rst nd the count rate without any absorber and then use di erent aluminum foils to block the
beta particles and measure the count rate with the varying thickness of the absorber which is aluminum.
We continue this process till the number of counts reaches a constant which is the background value
obtained earlier.

9.2.3 Sample Data


Operating Voltage = 420 V
Preset Time = 120 seconds
LABORATORY MANUAL FOR NUCLEAR PHYSICS 200

Sr source with Activity  3.7 KBq


90

Background Count

S.No. NB (NB NB ) (NB NB )2


1 45 -1.8 3.24
2 47 -0.2 0.04
3 47 -0.2 0.04

s
4 49 1.8 3.24

sic
5 48 0.8 0.64
P
NB = 47:2 (NB NB )2 = 7:16

hy
Table 9.1: Background Count rate for 120 seconds

rP
With these 5 values of the background counts, we determine the mean to be 47:2 counts in 120 seconds.
lea
To determine the error in this, we can think of using use Eq(1.63) to determine the error in the mean.
uc
For the  of the parent distribution, we can use the sample standard deviation s given by
rP
lN

(x x)2
s=
N 1
ua

In our case, we get s = 1:33. From this, we can estimate the error in the mean (from Eq(1.63)) to be

ps
an

 =
N
or
bM

1:33
NB =p = 0:59
5
La

Thus one would think that one should report the result as

NB = 47:20  0:59
Here again, as in the discussion around Eq(8.2), there is a subtle point that one needs to understand.
Let us see what we are doing- we are taking the total number of background counts in 120 seconds and
reporting it as a number. Then we are repeating the same procedure 5 times. Note that in each of the
values of NB , there is an error. This is the inherent statistical error that is associated with the event
which as we know is a result of a Poisson distribution. We can therefore think of each value of NB as
p
a mean of a Poisson distribution. The associated error with each value of NB is thus NB . Therefore,
the correct procedure to exhibit this inherent statistical error would be to take the expression for the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 201

mean value of NB , that is NB and apply the appropriate error prorogation equation to it. In our case
this means

5
P
(NB )i
i=1
NB =
5
1  2 
N2 B = (NB )1 + (2NB )2 +    + (2NB )5
25
1
= [45 + 47 + 47 + 49 + 48]
25

s
236

sic
=
p25
236

hy
NB =
5
= 3:072 (9.3)

rP
Thus our background count should be reported as
lea
NB = 47:20  3:07
uc
Next we take several values of the counts with the source and without any absorbers.
lN

S.No. N0 (N0 N0 ) (N0 N0 )2


ua

1 8082 47.8 2284.84


2 7975 -59.2 3504.64
an

3 8010 -24.2 585.64


4 8188 153.8 23654.54
bM

5 7916 -118.2 13971.24


P
N0 = 8034:2 (N0 N0 )2 = 44000:9
La

Table 9.2: Count rates without absorbers for 90Sr

The sample standard deviation is therefore


rP r
(x x)2 44000:9
s= = = 104:9
N 1 4
Thus the error in the mean is

 = ps =
104:9
p = 46:9
N 5
Thus our count rate is
LABORATORY MANUAL FOR NUCLEAR PHYSICS 202

N0 = 8034:2  46:9
Again, here too the error calculation given above is not quite correct for exactly the same reason as
we discussed for the error in the background count. We follow the same procedure for calculating the
actual error in N0 .

5
P
(N0 )i
i=1
N0 =
5

s
1  2

sic

N2 0 = (N0 )1 + (2N0 )2 +    + (2N0 )5
25
1

hy
= [8082 + 7975 + 8010 + 8188 + 7196]
25
40171

rP
=
p25
40171
N0 =
= 40:08
5 lea (9.4)
uc
Thus we should report the value of the counts in the given preset time without any absorbers as
lN

N0 = 8034:2  40:08
The data for various thicknesses of aluminum plates is as below. The density of aluminum is
ua

2:7 gm cm 3 .
an

 
Thickness(cm) t ( mg cm 2 ) No. of Counts N N NB Error Transmission Coecient N NB
bM

N0 NB
0 0 8034.2 7987.2 47.00 1
0.037 100 4396 4349 66.37 0.5445
La

0.077 207 2768 2721 52.70 0.3407


0.095 256 2346 2299 48.53 0.2878
0.114 307 1938 1891 44.13 0.2368
0.123 332 1946 1899 44.22 0.2378
0.132 356 1796 1749 42.49 0.2190
0.160 432 1188 1141 34.60 0.1428
0.164 443 1041 994 32.41 0.1244
0.172 461 910 863 30.32 0.1081
0.200 539 556 509 23.77 0.0640
0.201 543 586 539 24.40 0.0670
LABORATORY MANUAL FOR NUCLEAR PHYSICS 203

0.204 551 513 466 22.86 0.0580


0.218 588 402 355 20.28 0.0440
0.241 650 276 229 16.89 0.0290
0.259 699 212 165 14.88 0.0210
0.262 707 199 152 14.44 0.0190
0.281 758 136 89 12.06 0.0110
0.287 775 110 63 10.93 0.0080
0.299 807 103 56 10.60 0.0070

s
0.327 883 62 15 8.45 0.0019

sic
0.329 888 72 25 9.02 0.0031
0.339 914 65 18 8.63 0.0022

hy
0.357 963 59 12 8.27 0.0015

rP
0.366 988 51 4 7.77 0.0005
0.368 994 50 3 7.71 0.0004
0.385 1039 51 lea 4 7.77 0.0005
0.426 1151 53 6 7.90 0.0006
0.466 1258 51 4 7.77 0.0005
uc
0.493 1331 52 5 7.84 0.0005
lN

Table 9.3: Count rates with absorbers from 90Sr, Error Estimation & Transmission Coecient
ua
an

We can plot the graph of \N NB vs t" on a semilog scale. The error bars for N NB are obtained as
bM

discussed in Section 1.5.1. It is importantphowever to be careful when calculating the error in N NB .


p
We know that the error in N is simply N . But the error in NB is NOT NB as discussed above.
Instead, it is 3:07. So to calculate the error in N NBpwe need to add the errors in quadrature. Thus,
La

for instance when N = 4396, the error in N NB is 4396 + (3:07)2 = 66:37. We see that as the net
count rate becomes small, the errors increase drastically and at very small net count rates, the errors
are much more than the count rate itself. We also plot the transmission coecient (multiplied p by a
100, to get a percentage) against t. We also plot the graph for N vs t with the error bars given by N .
The graphs are shown in Figure 9.2.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 204

10000
Beta Rays Range
N-N_B: data
N-N_B data
N-N_B: best fit
Tramsission Data
N, N-N_B, transmission coeff 1000 Transmission Best Fit
N: data
N: data with errorbars

100

10

s
sic
1

hy
0.1

rP
0.01
0 200 400 600 800 1000 1200 1400
t( mg cm^-2)
lea
Figure 9.2: Graph of N NB vs t
uc
lN

We can see that the net count rate approaches a constant value which is close to zero while the count
rate approaches the background value after a point. This is the point where essentially all the beta
ua

particles have been stopped by the absorber and thus gives us the range for the beta particles from
this particular 90Sr source. To determine the range, we take the N vs t graph and determine the point
an

where it turns to become the constant background value. The error in the determination of the
p range
is then the di erence in the values between the range obtained from N and that from N  N . For
bM

this sample data, we obtain

R
La

R = 900  15 mg cm 2 (9.5)

Using this range, we can now calculate the end point energy for the beta rays from this source, using
Eq(9.2). Thus
LABORATORY MANUAL FOR NUCLEAR PHYSICS 205

R = 412E0n
n = 1:265 0:0954 ln E0
ln R = ln 412 + n ln E0
= ln 412 + (1:265 0:954 ln E0 ) ln E0
 
R
0:0954(ln E0 )2 1:265 ln E0 + ln = 0
412
1:265  1:141
ln E0 =

s
0:1908

sic
E0 = 1:92; 299539
E0 = 1:92 MeV (9.6)

hy
The equation to determine ln E0 is a quadratic and the two solutions are given above. Clearly, the only

rP
reasonable value is E0 = 1:92 MeV.

lea
We can also determine the lower and upper limits of the end point energy corresponding to the error
in the determination of the Range. Thus
uc
R = 900 15 mg cm 2 ; E0 = 1:88 MeV
lN

R = 900 + 15 mg cm 2 ; E0 = 1:94 MeV


ua

Thus

R
an

E0 = (1:92 + 0:02 0:04) MeV (9.7)


bM

The theoretical value of the end point energy for this source is known to be 2:28 MeV (Figure 9.1).
Thus the percentage error in our determination is simply
La

2:28 1:92
% error = = 15:8%
2:28

9.3 Questions
1. Why do rays have a continuous energy spectrum?
2. Is the emission of particles from a radioactive source isotropic?
3. Can one say something about the energy spectrum of the  which is emitted in the
decay process?
LABORATORY MANUAL FOR NUCLEAR PHYSICS 206

4. What is meant by the range or maximum range of particles?


5. How does the range of the particles depend on their energy?
6. How do we derive the range from the graph while minimizing systematic errors?

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La
Chapter 10

Experiment: Gamma Ray spectrum using

s
a Scintillation Counter

sic
hy
rP
Things to know before you do the experiment
lea
1. All the concepts in Chapter 2 .
uc
2. Interaction of Gamma Rays with matter.
lN

3. Least Square Fitting, Section 1.7.


ua

4. Goodness of Fit- Section 1.8.


an

10.1 Introduction
bM

In all the previous experiments, we have been using a GM counter to detect and measure radiation
from radioactive nuclides. In this experiment we will use a di erent kind of detector, the Scintillation
La

Counter. This works on a very di erent principle than a GM counter which is a gas lled detector
working on the principle of an ionising particle or radiation causing an avalanche. We shall study the
working of this counter as well as its method of operation. The basic purpose of the experiment is
to detect and study the energy characteristics of gamma rays. Recall that the GM counter
does not allow us to measure the energy of the incoming radiation. The scintillation
counter on the other hand allows us to measure the energy as we shall see. In this
experiment, we shall be studying the energy spectrum of gamma rays. For this purpose,we
shall be using a scintillation counter and several gamma sources.

207
LABORATORY MANUAL FOR NUCLEAR PHYSICS 208

10.2 Theory
The Scintillation counter basically can be thought of a scintillating material which emits light when
radiation or an ionising particle interacts with it, and a mechanism for collecting and measuring the
light which is produced. We shall study these separately. The incoming particle loses its energy and it
is this energy which is ultimately converted into light.

Actually the use of light produced in certain kinds of materials when radiation or a particle interacts
with it is one of the oldest ways of detection of such radiation. In the historic Rutherford experiment for

s
instance, the scattered particles were detected by the light ashes they produced on a Zinc Sulphide

sic
screen. In fact, the rst detection of X-rays by Roentgen also used scintillation- the platino-barium
cyanide crystals began to glow when the rays from his apparatus interacted with them. So how is this

hy
light produced and what are the properties that we want the material of the detector to have in order
for it to be useful?

rP
The emission of light by a material which is excited can be of several kinds. The most common one,
lea
uorescence is what we get when a material which has been excited, emits light immediately after
excitation and the light is in the visible region of the electromagnetic spectrum. Phosphorescence
uc
is basically the same as uorescence but here the light emitted is of a much longer wavelength and is
emitted usually on a time scale much longer than uorescence.
lN

We can de ne some terms which would be useful later on:


ua
an

1. Luminescence is the process of exciting a material (not thermally) and the subsequent emission
of light.
bM

2. How the material is excited determines the type of luminescence (e.g. photolumines-
cence, chemiluminescence, triboluminescence)
La

3. Fluorescence is photoluminescence or scintillation (i.e. excitation produced by ionizing radiation)


that has a fast decay time (nanoseconds or s)
4. Phosphorescence is the same as uorescence, but with a much slower decay time (milliseconds
to seconds)
There are basically two types of scintillating materials- inorganic and organic . These two kinds of
materials have very di erent mechanisms of production of light. In our laboratory, we use inorganic
scintillators. We shall study these two kinds separately.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 209

10.2.1 Inorganic Scintillators


To understand the phenomenon of scintillation in inorganic materials, we rst need to understand the
energy structure of crystalline materials since the mechanism for scintillation depends crucially on the
structure of the crystal. Recall that an atom has discrete energy levels or orbitals. When several
atoms form a molecule we get molecular orbitals which as molecules aggregate into a solid, combine
and become more and more dense. Finally, when a large number of molecules `combine' to form a
solid, the energy levels become so close to each other that they can be considered to form a continuum
which is called an energy band. The width of energy bands depends on the atomic orbitals which are
superposed to form the band. It also happens that there are some energies where there is no overlap at

s
sic
all and we then get band gaps.The width of the bands of course can vary and depends on the overlap
between the underlying atomic orbitals.

hy
Typically, a solid has an in nite number of allowed bands but most of the them have very high
energies to be of any relevance. It turns out that the electronic properties of solids depends on

rP
the bands which are near the Fermi level. Fermi level (NOT Fermi energy) in the band theory
is a hypothetical level such that at equilibrium, it has a 50% probability of being lled. It is not
lea
necessarily an actual energy level. In the language of Fermi-Dirac statistics, it corresponds to the total
chemical potential of the electrons. The closest band above the Fermi level is called the conduction
uc
band and the one closest below is called the valence band. The band gap is large in insulating
materials, somewhat smaller in semiconductor materials and very small in conductors, as seen in Figure
lN

10.1. The electrons will never have energy in the forbidden region or the band gap. Electrons in a
lower energy state are in the valence band and these are tightly bound to lattice sites. On the other
ua

hand, electrons in the conduction band have a higher energy and are more mobile throughout the crystal.
an
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 210

s
sic
hy
rP
Figure 10.1: Band gap in di erent class of materials
lea
Now let us consider what happens when radiation (we shall use the term radiation to denote both
ionising radiation and charged particles which can ionise the material), hits a pure inorganic crystal
uc
like Sodium Iodide. The radiation can deposit its energy in the crystal and one of the electrons in the
valence band can gain enough energy to move to the conduction band, leaving a hole in the valence
lN

band. When this excited electron returns to the valence band, it will emit a photon though in a pure
crystal, this is a very inecient process. Furthermore, the energy gap between the conduction and
ua

valence band is typically so large that the photon emitted is of short wavelength and not in the visible
region.
an
bM

To get the crystal to emit light in the visible range, we obviously need to decrease the energy gap that
the electron experiences when it falls or de-excites from the conduction band. This is usually done by
adding a trace amount of impurity to the pure crystal. These impurities are called activators. The
La

activator sites modify the energy band structure of the pure crystal locally while the overall energy
band structure remains unmodi ed. The net e ect of the activator sites is to create energy levels in
the forbidden region (the region between the conduction and the valence band, where ordinarily in a
pure crystal, the electrons cannot be). This is shown in Figure 10.2.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 211

s
sic
Figure 10.2: Energy Band Structure of Pure crystal & Activated Crystal

hy
Now let us consider what happens when radiation or a charged particle enters the crystal with activator

rP
sites. The incoming particle deposits its energy and creates an electron hole pair in the valence band.
This primary electron-hole pair, through a cascade e ect creates many secondary electron-hole pairs.
lea
When the energy of the electronic excitations becomes below the ionization energy, thermalization
takes place. At the end of this stage, all the electrons are at the bottom of the conduction band and
uc
the holes at the top of the valence band. This whole process takes place on a time scale of a picosecond.
lN

After the thermalization stage, the free electron hole pairs migrate through the material. The hole will
migrate to the activator site and ionise it. The electron in the conduction band continues migrating
ua

till it meets one of the ionised activator site and neutralises it. Now we have a neutral activator, but
one which depending on the energy of the electron will be an excited state (of the neutral activator).
an

This excited state will de-excite to its ground state and in the process give out a photon which, now
since the energy di erence between the activator excited and ground state is smaller than the original
bM

band gap of the crystal, will be in the visible region. This process takes place on a time scale of around
10 10 10 11 seconds. This is essentially then the time scale of the scintillation which one observes.
La

Sometimes, the electron in the conduction band when it encounters the ionised activator or impurity
site, neutralises it but goes into an excited state from where the transition to the ground state is
forbidden. In that case, the electron typically gains more energy from thermal motion and moves to
a higher excited state from which it can de-excite to the ground state emitting light. Obviously, this
process is on a much longer time scale and sometimes therefore we get an after glow in scintillating
materials. This component of light emitted, as we saw above is phosphorescence.

It can also happen that the de-excitation of the electron from the activator excited state to its ground
state is such that no visible light is emitted. This process is called quenching.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 212

The processes we have outlined above depend on the dynamics of an electron-hole pair created in the
valence band, which are essentially independent of each other. In semiconductors, there is another kind
of process which can take place. Instead of the electron and hole moving independently, sometimes it
happens that the electron and the hole form a loosely bound state called an exciton. The exciton
band is typically just below the conduction band and therefore the energy of the electron in an exciton
is somewhat lower than that in the conduction band. Here again, the electron hole exciton moves
together to the activator site and the hole gets neutralised while the electron in the exciton band
neutralises the ionised activator and when it de-excites to the ground state, can emit visible light. It
should be noted that the electron hole pair in an exciton can also recombine at the site of impurities or

s
traps in a crystal without any external doping of the kind used to create activator sites. The time scale

sic
for the exciton recombination is typically much faster than that of the electron-hole pairs which move
independently since here the electron and hole move together. So we have two kinds of components

hy
in most materials- the fast component which is caused by the recombination of excitons and the slow

rP
component which is when the electrons in the conduction band and holes in the valence band are
captured successively by the activator sites. The fast and slow components can be resolved for most
scintillating materials. lea
What we have thus seen is how an incoming radiation interacting with a crystal which has been doped
uc
with an activator produces light through luminescence. One can do an elementary analysis of the
lN

energy eciency of such materials and we nd that for every electron-hole produced by the incoming
radiation, there is roughly one photon produced. Further, note that the emitted light can essentially
pass right through the bulk of the crystal. This is because remember that the energy di erence in the
ua

pure crystal bulk between the conduction and valence bands was such that a de-excited electron from
an

the conduction gap produced short wavelength or high energy photons. The energy di erence between
the activator excited states and its ground state is much less and so the light produced is of a longer
bM

wavelength or smaller energy. There is thus usually no absorption of this light by the bulk of the crys-
tal material since its emission(and hence absorption spectrum ) is peaked at a much shorter wavelength.
La

10.2.2 Organic Scintillators


The scintillation mechanism in organic materials is quite di erent from the mechanism in inorganic
crystals that we studied in the previous section. In inorganic scintillators, we saw that the scintillation
arises because of the structure of the crystal lattice and the impurities which we introduce. The
uorescence mechanism in organic materials arises from transitions in the energy levels of a single
molecule and therefore the uorescence can be observed independently of the physical state. Practical
organic scintillators are organic molecules which have symmetry properties associated with the electron
structure.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 213

The molecular energy levels of organic molecules which exhibit scintillation are separated by a few
electron volts and they get closer to each other as we go up. The singlet energy levels are subdivided
into a series of levels with a much ner structure because of the vibrational modes of the molecule.
The typical spacing of these is around a tenth of an electron volt.

Now let us consider the case when radiation interacts with an organic scintillator. The average energy
at room temperature is around 0:025 eV (recall that an energy of 1 eV is roughly equal to the thermal
energy at 104 K, thus a room temperature of  300 K is about 0:03 eV) and so all the molecules are in
the ground state (called the S00 state where the rst subscript indicates the singlet spin state and the

s
second the ne structure state). When radiation deposits its energy into the material, the electrons

sic
are excited to the upper levels. The higher states like S2 ; S3 etc. de-excite in a matter of picoseconds
to the S1 state via transitions which do not produce any radiation. The S1 states like S11 ; S12 etc

hy
with higher vibrational energy also lose energy and soon we have all the excited molecules in the S10

rP
state. When these electrons in the S10 state de-excite to the S0 state, we get luminescence. Again,
essentially all the emission light is of a lower energy than that required for absorption and therefore
the organic material is transparent to the luminescence produced like in the case of inorganic scintillator.
lea
In addition to the transitions in the singlet states, there are also transitions from the triplet states to
uc
the ground state. The triplet states are typically longer lived and therefore the typical time scale of
lN

this transition is longer leading to a phosphorescence. A schematic illustration of the states is shown
in Figure 10.3.
ua
an
bM
La

Figure 10.3: Energy States in an organic scintillatorx


x(Source: "Pistates" by Napy1kenobi - Own work. Licensed under CC BY-SA 3.0 via Wikimedia Commons -
https://commons.wikimedia.org/wiki/File:Pistates.svg#/media/File:Pistates.svg )
LABORATORY MANUAL FOR NUCLEAR PHYSICS 214

10.2.3 Photomultiplier Tube


We have seen now how ionising radiation or a charged particle from a radionuclide when it interacts with
a scintillator would produce light because of uorescence. However, to convert this light into something
which can be detected or its properties measured requires some kind of sensor which is sensitive to the
light. The most often used sensor is a Photomultiplier Tube (PMT). Let us now see how this works.

The basic purpose of a photomultiplier tube is to convert the light signal into an electrical signal
that is, ultimately convert a photon into one or more electrons which can be detected and whose
properties measured. The PMT basically consists of three components- a photocathode which

s
sic
will produce the initial or primary electron on interaction with the photon; an arrangement to
accelerate the electron(s) produced and an arrangement to measure the current produced by these

hy
electrons. The whole arrangement has to be in a vacuum tube. As we will see, the eciency of the
photocathode to produce the primary electrons is not very high. Thus, what is usually done in a

rP
PMT is to have an arrangement which multiplies the number of electrons that is produce a number of
secondary electrons from the primary electrons. Let us see how each of these three components function.
lea
uc
lN
ua
an
bM

Figure 10.4: Schematic of a Photomultiplier Tube


La

x(Source: Wikipedia )

The light from the scintillator is made to fall on a photocathode. This is, as the name suggests, made
of a material which by using the photoelectric e ect, produces electrons when photons impinge on it.
Obviously, we need to choose the material and the design of the photocathode in such a way that the
energy of the photon is transferred eciently to the primary electron which is produced. Firstly, recall
that when a photon transfers its energy to an electron, for the electron to emerge from the material
and be detected, it needs to overcome not only the collisions with other electrons and the lattice within
the material but also the surface potential barrier. Clearly then, there has to be a minimum energy of
the photon when this is possible. This means that any photocathode will have a long wavelength cuto
LABORATORY MANUAL FOR NUCLEAR PHYSICS 215

depending on the material and the geometry of the photocathode. For our purposes, typical light
given o in the scintillation is in the blue region and thus has an energy of around 3 eV. (  400 410
nanometers). It turns out that the surface potential barrier or work function for most metals is more
than 3 eV while that for some semiconductors is only around 2 eV. Thus we see that to detect the light
from the scintillator, we need a photocathode made from a suitably prepared semiconductor.

But this is not enough- as we saw above, the electron on absorbing the energy from the photon,
needs to travel to the surface of the material to be ejected out. In its motion to the surface, it loses
energy also by collisions with other electrons. Clearly, the probability of collisions with other electrons

s
increases with increased electron density in the material as well as the distance travelled. Thus, metals

sic
have a high energy loss and so the electrons travel only a small distance before losing enough energy to
be unable to escape from the surface. The maximum depth from which an electron can travel to the

hy
surface of a material and still escape is called the escape depth. In metals it is only a few nanometers.

rP
This fact then determines the geometry of the photocathode since we thus need a very thin layer of the
photosensitive material or else most of the electrons produced will not be able to escape. The situation
in semiconductors is better where the escape depth is a few tens of nanometers. Here again, we need to
lea
have a very thin layer in order to maximise the probability of the electrons produced to escape from the
surface. This however, leads to another issue which e ects the eciency of the photocathode- a very
uc
thin layer of photosensitive material means that it will allow a large fraction of the incident radiation
lN

through, thereby reducing the eciency of the photocathode. Thus, there is a trade o which needs to
be made between these two factors while determining the thickness of the photosensitive material.
ua

The eciency of the photocathode is usually described by a quantity called quantum eciency which
an

is de ned as
number of photoelectrons emitted
bM

Quantum Eciency =
number of incident photons
Clearly, this is a function of the wavelength of the incident light. Most PMTs have an eciency of
around 15 25%.
La

The photomultiplier tube, as the name suggests, does more than just produce photoelectrons- it
also has a multiplying e ect which we now turn to. First the primary electrons produced in the
photocathode layer are focussed onto a narrow region. In this process, the electrons are also accelerated
in an electric eld of a few hundred volts. The focussing is done by using a focussing electrode.
The accelerated primary electrons are then made to produce secondary electrons by the process of
secondary emission using a series of electrodes called dynodes.

A dynode is basically an electrode in a vacuum tube that serves as an electron multiplier through
secondary emission. When the accelerated primary electrons are focussed on a dynode, they transfer
LABORATORY MANUAL FOR NUCLEAR PHYSICS 216

their energy to the electrons in the dynode material which are ejected out. Of course, just like the
photoelectric e ect, the electrons in the dynode material need to have enough energy to overcome the
surface potential barrier which, as we have seen is around 3 4 eV. However, the primary electrons
produced by the photocathode were emerging with very low energies, ( 1 eV) but these are accelerated
through around 100 V and so when they strike the dynode, have an energy of 100 eV. Thus, if all this
energy was transferred to the electrons in the dynode, we could in principle get around 30 secondary
electrons for each primary electron striking the dynode. Clearly, this is the maximum number of
secondary electrons that can be produced per primary electron. The actual number is signi cantly less
because once again, for the electrons to emerge from the material, they need to travel to the surface

s
and in this process lose energy. Only those electrons which reach the surface with energy more than the

sic
work function can escape. Typically, around 6 8 secondary electrons are produced for each primary
electron impinging on the dynode.

hy
rP
The next stage in the PMT is to multiply this number of secondary electrons. This is done by using
di erent geometries of an array of dynodes. Each dynode when struck by the secondary electrons from
the previous dynode produces more secondary electrons which then impinge on the next dynode to
lea
produce an even larger number, leading to a cascade e ect. Various kinds of geometries are used to
achieve this. A fairly simply, box and grid type of arrangement is shown in Figure 10.4. Finally, the
uc
multiplication achieved is around 107 by a PMT. The electrons which nally emerge from the dynode
lN

arrangement are collected and analysed using the associated electronics to which the PMT is connected.

A simpli ed model will give us an idea of the amount of charge produced by the PMT. Let us de ne:
ua
an

p: The number of light photons produced in the scintillating crystal.


k : Optical eciency of the scintillating crystal, that is the fraction of transmitted photons.
bM

l : Quantum eciency of the photoelectrode, that is the number of electrons produced per photon
striking it.
n: The number of dynodes in the setup.
La

R: The dynode multiplication factor that is the number of secondary electrons produced in a dynode
per primary electron absorbed.
e: The charge of an electron.

Then, we can simply write the amount of charge which comes of the photomultiplier when a gamma
ray photon hits the scintillator crystal. This is simply

Q = pklRn e
To get an estimate of this, assume that the number of light photons produced in the crystal is 1000 and
its optical eciency is 50%. Further assume that the quantum eciency of the photocathode is about
LABORATORY MANUAL FOR NUCLEAR PHYSICS 217

10%. Let the number of dynodes by 10 in the setup and each dynode have a multiplication factor of 5.
Then

Q = 103  0:5  0:1  51 0  1:9  10 19 C  92 picoC


which is an extremely small amount of charge.

The PMT output as we have seen is a very small pulse since the amount of charge, even with the
multiplication by the dynode stages is very small. A very sensitive ampli er is connected to the PMT
output to amplify the signal. This is called a pre-ampli er. The ampli er makes the pulse narrower

s
sic
and the Pulse Height Analyzer then creates an output pulse for input pulses of acceptable height. A
Single Channel Analyzer (SCA) or a Multi Channel Analyzer (MHA) are two kinds of Pulse Height

hy
Analyzers.

rP
10.2.4 Gamma Ray Spectrum
lea
The basic purpose of the experiment, apart from studying the working of a Scintillation counter, is
to detect and study the gamma ray emission from various nuclei. To detect and measure the energy
uc
of the gamma rays, we rst need to see how an incoming gamma ray photon would interact with the
scintillation detector. In our case, the detector is a Thallium activated Sodium Iodide (NaI)
lN

scintillating crystal.
ua

We have already studied the interaction with matter of gamma rays in Section 3.2.3. Recall that there
are basically three di erent ways for a gamma ray photon to interact with matter- photoelectric e ect,
an

Compton scattering and pair production. Which process dominates the overall cross section depends
bM

on the energy of the incoming gamma rays. Note that in all three processes, a high energy electron is
produced by the incoming gamma ray. (In Compton e ect, the free electron is provided with energy
by the gamma ray in elastic scattering.) The production of this high energy electron is crucial since
La

gamma ray photons being charge neutral, will otherwise not be detected by the scintillator. Since the
spectral distribution of the light produced by the scintillator will depend on the interaction of the
gamma rays with the crystal and we now discuss this.

When a gamma ray photon strikes an atom of the scintillator material, it is absorbed completely and
all of its energy is transferred to one of the bound electrons. Since the energy of the gamma rays is
typically of the order of MeV, it is much larger than the binding energies of the bound electrons. The
electron is therefore released from the atom and moves rapidly through the crystal since it is carrying
the balance energy of the gamma ray photon (the original photon energy minus the binding energy).
This fast moving electron produces the scintillation as we have seen above. However, another process is
LABORATORY MANUAL FOR NUCLEAR PHYSICS 218

also operative- when the bound electron is freed by the gamma rays, another bound electron in the atom
will fall into the vacancy left by the bound electron which has been freed. Typically, a K-shell electron
is produced by the gamma rays and another bound electron falls into the vacancy in the K-shell. This
process produces x-rays which in turn will free more loosely bound electrons and these too shall produce
the scintillation light pulses. The whole process, that is the original ash produced by the gamma ray
initially and the subsequent ashes will happen within the resolution time of the counter and so usually
cannot be distinguished by the counter. In the end, if the photoelectron stops in the crystal and no
light escapes the crystal, then the total incoming gamma ray energy would have been converted into
the photomultiplier output pulse. This pulse is called the photopeak which has an energy equal to

s
the incoming gamma ray photon.

sic
However, we know that photoelectric e ect is not the only interaction that the gamma rays have with

hy
matter. Compton e ect is also operative. We know that when a gamma ray photon scatters o a free

rP
electron, the scattered electron takes away some of the energy of the incoming gamma ray. This electron
then interacts with the scintillator crystal and is detected. What happens to the scattered gamma ray?
The scattered gamma ray escapes from the crystal without any further interaction. This is because
lea
the probability of Compton scattering of a typical gamma ray in a typical scintillator crystal is about
1 10%. Thus it is extremely unlikely that the scattered gamma ray will have a second scattering
uc
event. We know that the energy of the scattered electron is given by Eq 3.41
lN

R
0   1
hi (1 cos )
m c2
ua

Ee = hi hf = hi @ e  A (10.1)


1 + mhe ci2 (1 cos )
an

or the energy of the scattered photon is given by


bM

R E 0 =
1+ E
E
me c2 (1 cos )
(10.2)
La

Here E is the energy of the incoming gamma ray and E 0 is the energy of the scattered gamma photon.

The energy of the scattered electron (which remember is what the scintillator is detecting) is a function
of the angle by which the gamma ray is scattered, . The energy of the electron goes from 0 when
2
 = 0 that is the gamma ray is not scattered, to me c22E+2 E when  = 180 that is the gamma ray photon
is backscattered. This maximum energy of the scattered electron is called the Compton Edge. Also
note that the angular dependence of the scattered electron energy is a slowly varying w.r.t . Hence
the energy of the of the electron is essentially constant till it falls o sharply at the Compton Edge.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 219

In addition to the Compton edge and the Photopeak, there is another e ect which can take place.
Suppose a gamma ray encounters Compton scattering outside the material of the scintillator. This can
be in the shielding of the detector for instance. In this case, the scattered gamma ray enters the crystal
and produces a peak by photoelectric e ect. However, since the geometry of the detector is such, only
those scattered gamma rays with large values of  can enter the detector and hence these gamma rays
will have energies close to the maximum photon energy at  = 180, that is
m e c 2 E
E 0 =
me c2 + 2E

s
We can easily see that this is lower than the Compton edge. So typically, one would also see this peak

sic
in the detector at a lower energy. This peak is called the backscatter peak.

hy
Finally, we know that for gamma ray energies greater than 2me c2 , pair production is possible. Thus
for high energy gamma rays, we will get a pair of electron and positron. If both of these lose all their

rP
energy within the scintillator, then we will see a peak at the energy of the gamma ray minus the rest
mass energy of the electron and positron. However, it might happen that the positron, before it loses all
lea
its energy, annihilates with an electron and produces two gamma ray photons. Each of these photons
will have an energy of atleast me c2 . Usually the gamma ray photons will have a higher energy since
uc
the positron will have some kinetic energy before it annihilates.
lN
ua
an
bM
La

Figure 10.5: Typical Cs-137 spectrum with a NaI(Tl) scintillator

We also know from Section 3.2.3 that the cross section for these three processes of the interaction of
gamma ray photons with matter depend on the energy of the gamma rays. For low energies, we have
seen that photoelectric e ect dominates the cross section. And since the photopeak gives us the total
energy of the incoming gamma ray, most scintillators are optimised for this energy range. For the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 220

NaI:T1 scintillator that we use, a small amount of Thallium is added to the sodium iodide. Thallium
being a heavy metal, has a lot of electrons and the photoelectric cross section depends strongly on the
number of electrons in the atom. Of course, as the energy of the gamma ray increases, the photoelectric
cross section decreases fast but the Compton scattering cross section decreases more slowly and above
a few hundred keV, it is the Compton scattering which dominates. Finally, above 1:02 MeV, pair
production becomes operative.

The idea behind the experiment is to use a scintillator counter to determine the energy spectrum
of gamma rays from di erent emitters. However, the set up of the scintillator counter will give us

s
information on the output voltage or the pulse height obtained from the counter. This needs to

sic
be correlated to the energy of the gamma ray entering the counter. For this purpose calibration
needs to be done. As we have seen above, gamma ray sources typically give a photopeak when

hy
they interact with the material of the scintillator. These are at distinct energies which depends on
the energy of the gamma ray produced by the source. Thus, for instance, 137Cs gives a gamma ray

rP
photon with energy E  0:662 MeV. So we use various sources to obtain the photopeak voltages
and to calibrate the counter. The relationship between the energy of the incoming gamma ray and
lea
the pulse height (or voltage in our case) is a linear one. By tting the experimentally obtained
points, we can nd the coecients (the slope and the y-intercept) and therefore get the relation-
uc
ship. This then will allow us to nd the energy spectrum for any source by measuring the output voltage.
lN

An analysis of the spectrum, specitically nding the peaks requires a Pulse Height Analyser (PHA).
This is usually done with two kinds of circuits, a Lower Level Discriminator (LLD), which as
ua

the name suggests, allows only voltages which are higher than the setting and an Upper Level
an

Discriminator (ULD) which allows only voltages which are lower than the setting. It is easy to see
that if we choose the two settings of LLD and ULD appropriately, we will be able to get a window in
bM

which we can determine the peak at any part of the spectrum. We can, for instance, start with the
lower end of the spectrum and then keep adjusting the LLD and ULD so as to scan the whole spectrum
at various energies.
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 221

s
sic
hy
rP
Figure 10.6: Use of LLD and ULD to obtain a voltage/energy windowx
x(Source: Wikicommons )
lea
Once we have a window with a Pulse Height Analyser, then the output is fed into a Single Channel
Analyser (SCA). A SCA basically consists of a Ratemeter which measures the number of pulses
uc
produced in unit time in a particular channel or window. A more sophisticated instrument is an or a
Multi Channel Analyser (MCA). This replaces the PHA and the ratemeter with a single instrument
lN

and has an electronic circuit which allows us to look at many channels simultaneously and therefore
get the complete spectrum in one go. Typically, an MCA has 1024 channels.
ua

A simpli ed diagram of a Scintillation detector setup with a Single Channel Analyser is given in Figure
an

10.7.
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 222

s
sic
hy
rP
Figure 10.7: Block Diagram of a scintillation counter with a Single Channel Analyserx
x(Source: Wikicommons )
lea
For the Multi Channel Analyser, the set up is shown in Figure 10.8.
uc
lN
ua
an
bM
La

Figure 10.8: Block Diagram of a scintillation counter with a Multi Channel Analyserx
x(Source: Wikicommons )
We can now summarise the whole operation then as follows:
Gamma ray photons from a source placed near the scintillator interact with the crystal and produce
photons. These pass through the Photo Multiplier Tube and produce electrons which are fed into a
Pre Ampli er. The Pre Ampli er output is fed into an ampli er which ampli es it. The output of
the Ampli er is then passed through a Pulse Height Analyser where the LLD and ULD settings are
chosen appropriately. Finally, the PHA output goes to a SCA to get the counts in a particular channel
LABORATORY MANUAL FOR NUCLEAR PHYSICS 223

which can then be correlated to a corresponding voltage. Alternatively, the ampli er output can
go directly to a MCA which allows us to look at many channels simultaneously and obtain the spectrum.

The decay schemes for some common gamma emitters are given below.

s
sic
hy
rP
Figure 10.9: Decay Scheme for 60Co- Gamma decay
lea
uc
lN
ua
an
bM

Figure 10.10: Decay Scheme for 22Na- Gamma decay


La

Figure 10.11: Decay Scheme for 137Cs- Gamma decay


LABORATORY MANUAL FOR NUCLEAR PHYSICS 224

s
Figure 10.12: Decay Scheme for 57Co- Gamma decay

sic
hy
Thus, we can tabulate the gamma ray energies for 4 known sources which can be used to calibrate the
instrument.

rP
Parent Nucleus Lifetime(years) Daughter Nucleus Gamma Energy (MeV)
57Co 271.8 (days) 57Fe
lea 0.122
22Na 2.605 22Ne 1.2746
137Cs 137
uc
30.17 Ba 0.6616
60Co 5.272 60Ni 1.1732
lN

1.3325
Table 10.1: Nuclear Decay Data for Gamma Ray Sources
ua

In the case of 22Na, we see from the decay scheme above, that a positron is also emitted. We have
an

already seen that a positron will annihilate with an electron to produce two gamma ray photons. The
bM

typical energy of these gamma ray photons is 511 keV each (corresponding to the rest mass energy of
the electron/positron). When we measure the gamma ray spectrum of using nuclides which produce
positrons, then we typically see this annihilation peak at 511 keV.
La

Another important characteristic of the counter is energy resolution. This as the name suggests, is a
measure of the ability of the detector to resolve adjacent peaks in the gamma spectrum. We can de ne
it is
E
R=
E
 100% (10.3)
Here E is the Full Width at Half Maximum of the photopeak and E is the energy of the photopeak.
The resolution is basically controlled by the statistical uctuations of the number of photoelectrons
produced at the photocathode of the photomultiplier tube in the counter. In addition, some of the
other factors controlling the resolution are:
LABORATORY MANUAL FOR NUCLEAR PHYSICS 225

 The number of photons per scintillation event.


 The number of photons that strike the photocathode.
 The number of photoelectrons released from the photocathode per photon hitting it.
 The number of photoelectrons striking the rst dynode.
 The multiplication factor of the photomultiplier tube.
Typically, the resolution of Na(Tl) detectors is seen to be

s
R pk  100%

sic
E
where k is a factor which is characteristic of the particular detector.

hy
rP
10.3 Procedure
lea
Before starting the experiment, it is important to set up the electronics that are used. For this purpose,
the following steps need to be followed:
uc
1. Ensure that the voltage is at a minimum, that is the power supply knob is at its lowest position.
lN

2. High Voltage (HV) power supply is switched o .


ua

3. NIM-BIM is o .
4. Switch on the mains, that is the AC input.
an

5. Switch on the NIM-BIM.


bM

6. Switch on the power supply module.


The operating voltage of NAI(Tl) scintillator used along with the PMT is 500 Volts. Before applying
La

the operating voltage to the counter, connect the output of the detector to DSO.

Place the source at the top of the detector mount. Slowly increase the voltage in the High Voltage
power supply from 0 to 500 Volts and simultaneously observe the shifting base line in DSO. This
ensures that the connections are done properly as bias voltage is reaching the detector.

Connect the output of the detector to the linear ampli er.

Settings for the ampli er:


LABORATORY MANUAL FOR NUCLEAR PHYSICS 226

 Fine Gain: 0
 Coarse Gain: 1
 Shaping s: 2
 Atn: 1
 Polarity: +
Set the ampli ed voltage to around 6 Volts.

s
DSO setting:

sic
Voltage Scale: 1 V/div

hy
Timing in s
Use trigger to observe the output properly.

rP
Connect the output of shaping ampli er to the input of the Single Channel Analyzer (SCA).

SCA Settings:
lea
Mode: WIN
uc
LLD: 0:1 V (Range: 0-10 V)
lN

V : 0.1 V (Range: 0-1 V)


ua

Connect output of SCA to the input of the counter. Polariy : + ve. Preset time: 50 seconds.
an

After setting up the electronics.


bM

1. Source 137Cs is placed on the PMT setup. The voltage is set at 500 V.
2. A preset time of 50 seconds is set, with the polarity of the counter timer set at + ve.
La

3. ULD window is xed at 0:1 V and the LLD is varied in steps of 0:1 V and the corresponding counts
are noted for 50 seconds each.
4. The graph of Counts versus Voltage of pulses is plotted
5. The position of the photopeak is obtained from the graph.
6. This procedure is repeated for all the sources.
7. Using the photopeaks from these sources, an energy calibration graph is plotted and t to a straight
line.
8. Using the energy calibration equation, graphs of the energy spectrum are plotted for each source.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 227

9. From the energy spectrum graphs, the energy resolution of the counter for di erent energies (of
the photopeaks of the di erent sources) is obtained.
10. Using the energy spectrum graphs, the back scatter peaks are obtained and compared with the
theoretical values for the sources.

10.4 Sample Data


Source: Cs-137 (Gamma Source)
Half-Life: 30.08 years

s
sic
Activity: 163 kBq
Date of Manufacturing: 01-08-13

hy
rP
LLD (Volts) Counts (Ni ) LLD (Volts) Counts(Ni )
0.1 8235 3.6 2065
0.2
0.3
12699
25814
lea3.7
3.8
2061
2067
uc
0.4 9132 3.9 1934
0.5 3874 4.0 1834
lN

0.6 3862 4.1 1730


0.7 3711 4.2 1391
ua

0.8 3623 4.3 1041


0.9 3506 4.4 757
an

1.0 3367 4.5 552


1.1 3510 4.6 405
bM

1.2 3501 4.7 311


1.3 3416 4.8 276
1.4 3500 4.9 200
La

1.5 3380 5.0 238


1.6 3563 5.1 242
1.7 3873 5.2 270
1.8 4216 5.3 344
1.9 4225 5.4 438
2.0 4095 5.5 694
2.1 3763 5.6 1008
2.2 3424 5.7 1747
2.3 3170 5.8 2618
2.4 2904 5.9 3446
LABORATORY MANUAL FOR NUCLEAR PHYSICS 228

2.5 2662 6.0 3985


2.6 2512 6.1 3727
2.7 2445 6.2 2752
2.8 2276 6.3 1724
2.9 2224 6.4 774
3.0 2240 6.5 316
3.1 2137 6.6 126
3.2 2057 6.7 57
3.3 2035 6.8 46

s
3.4 2073 6.9 37

sic
3.5 2070

hy
Table 10.2: LLD Voltage and Counts for 137Cs

rP
lea
uc
We can plot the counts versus voltage. The plot is shown in Fig 10.13.
lN

Gamma Spectrum for Cs-137


ua

4500
"cal1.txt" u 1:2
"cal1.txt" u 1:2
4000
an

3500
bM

3000

2500
Counts

2000
La

1500

1000

500

0
0 1 2 3 4 5 6 7
Voltage (V)

Figure 10.13: Voltage vs Counts for 137Cs

Source: Na-22 (Gamma Source)


LABORATORY MANUAL FOR NUCLEAR PHYSICS 229

Half-Life: 2.602 years


Activity: 174 kBq
Date of Manufacturing: 01-08-13

Note: The gain of the ampli er was changed to accommodate the higher energy pulses.Gain: x2.3674

LLD (Volts) Counts (Ni ) LLD (Volts) Counts(Ni )


0.0 15632 3.2 1702

s
0.1 15624 3.3 1741

sic
0.2 15660 3.4 1591
0.3 15557 3.5 1633

hy
0.4 15456 3.6 1592

rP
0.5 16332 3.7 1573
0.6 16926 3.8 1672
0.7 18841 lea3.9 1702
0.8 19093 4.0 1688
0.9 17334 4.1 1679
uc
1.0 14721 4.2 1528
lN

1.1 13490 4.3 1227


1.2 12246 4.4 881
1.3 10341 4.5 535
ua

1.4 7072 4.6 441


an

1.5 4832 4.7 39


1.6 4414 4.8 409
bM

1.7 4994 4.9 477


1.8 6514 5 705
1.9 10700 5.1 1168
La

2.0 19735 5.2 1639


2.1 30084 5.3 1820
2.2 21687 5.4 1367
2.3 6027 5.5 621
2.4 2324 5.6 236
2.5 2114 5.7 125
2.6 2070 5.8 109
2.7 1896 5.9 147
2.8 1839 6.0 107
2.9 1795 6.1 122
3.0 1788
LABORATORY MANUAL FOR NUCLEAR PHYSICS 230

3.1 1768

Table 10.3: LLD Voltage and Count rate for 22Na

We can plot the counts versus voltage. The plot is shown in Fig 10.14.

s
sic
Gamma Spectrum for Na-22

hy
35000
"cal3.txt" u 1:2
"cal3.txt" u 1:2

rP
30000

25000

20000
lea
Counts

uc
15000
lN

10000

5000
ua

0
an

0 1 2 3 4 5 6 7
Voltage (V)
bM

Figure 10.14: Voltage vs Counts for 22Na


La

Source: Co-60 (Gamma Source)


Half-Life: 5.2712 years
Activity: 134 kBq
Date of Manufacturing: 01-08-13
Note: The gain of the ampli er was changed to accommodate the higher energy pulses.Gain: x2.3674

LLD (Volts) Counts (Ni ) LLD (Volts) Counts(Ni ) LLD (Volts) Counts(Ni )
0 1077 2.1 579 4.1 349
0.1 1128 2.2 543 4.2 333
0.2 1201 2.3 554 4.3 306
LABORATORY MANUAL FOR NUCLEAR PHYSICS 231

0.3 1311 2.4 553 4.4 299


0.4 1321 2.5 525 4.5 273
0.5 1273 2.6 532 4.6 223
0.6 1187 2.7 509 4.7 263
0.7 1136 2.8 520 4.8 417
0.8 1054 2.9 502 4.9 546
0.9 1222 3.0 516 5.0 477
1.0 1231 3.1 523 5.1 212
1.1 1142 3.2 550 5.2 75

s
1.2 1000 3.3 578 5.3 69

sic
1.3 911 3.4 523 5.4 155
1.4 795 3.5 525

hy
5.5 257
1.5 702 3.6 587 5.6 356

rP
1.6 655 3.7 576 5.7 238
1.7 599 3.8 603 5.8 96
1.8 617 3.9 lea 455 5.9 30
1.9 567 4.0 440 - -
2.0 572 - - - -
uc
lN

Table 10.4: LLD Voltage and Count rate for 60Co


ua
an
bM

We can plot the counts versus voltage. The plot is shown in Fig 10.15.
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 232

Gamma Spectrum for Co-60


1400
"cal4.txt" u 1:2
"cal4.txt" u 1:2

1200

1000

800
Counts

s
600

sic
400

hy
200

rP
0
0 1 2 3 4 5 6
Voltage (V)

lea 60Co
Figure 10.15: Voltage vs Counts for
uc
Source: Co-57 (Gamma Source)
lN

Half-Life: 271.8 days


Activity: 149 kBq
ua

Date of Manufacturing: 17-08-15


an
bM

LLD (Volts) Counts (Ni )


0.0 5594
0.1 5768
La

0.2 2700
0.3 2109
0.4 2012
0.5 2073
0.6 3350
0.7 6197
0.8 12020
0.9 19935
1.0 29708
1.1 52454
1.2 94828
LABORATORY MANUAL FOR NUCLEAR PHYSICS 233

1.3 87560
1.4 31426
1.5 7605
1.6 2127
1.7 1403
1.8 1363
1.9 1267
2.0 1385
2.1 1329

s
2.2 1479

sic
2.3 1364

hy
Table 10.5: LLD Voltage and Count rate for 57Co

rP
lea
uc
We can plot the counts versus voltage. The plot is shown in Fig 10.16.
lN

Gamma Spectrum for Co-57


ua

110000
"cal2.txt" u 1:2
"cal2.txt" u 1:2
100000
an

90000

80000
bM

70000

60000
Counts

50000
La

40000

30000

20000

10000

0
0 0.5 1 1.5 2 2.5
Voltage (V)

Figure 10.16: Voltage vs Counts for 57Co

Energy Calibration
LABORATORY MANUAL FOR NUCLEAR PHYSICS 234

Voltage (Volts) Energy (keV)


1.24 122
5.05 511
6.00 662
12.535 1275
11.644 1173
13.402 1333

s
Table 10.6: Voltage vs Energy of Photopeaks (and annihilation peak for 22Na) for di erent gamma sources

sic
hy
rP
One can see that the relationship is roughly linear. On doing the least square t to a straight line, one
gets
lea
m = 99:454 c = 19:353
uc
and therefore the energy-voltage relationship is given by
lN

E = 99:454V + 19:353
where V is in Volts and E is in keV. The points and the best t line are plotted in Figure 10.17.
ua
an
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 235

Energy Calibration
1400
"res4.txt" u 1:2
99.454*x+19.353

1200

1000
Energy(keV)

800

s
600

sic
400

hy
200

rP
0
0 2 4 6 8 10 12 14
Voltage

lea
Figure 10.17: Energy Calibration
uc
lN

One can also use Eq(1.77) and Eq(1.78) to nd the errors in the slope and the intercept. This gives us
ua

m = 2:143 c = 20:247
an

The error as we see, in the slope is not high ( 2%) but that in the intercept is high, almost 100%.
However, remember that the scale on the y axis is greater and therefore an error of 20 in the y-intercept
bM

is not signi cant.

To test how good our t is, we can use the 2 test. Recall from Eq(1.66) that is 2 is de ned as
La

N
X (yi Mxi C )2
2 =
i=1
y2
where
r
1 X
y = (yi Mxi C )2
N 2
The quantity of interest for us is the reduced 2 , that is ~2 , which is de ned as
2
~2 =

LABORATORY MANUAL FOR NUCLEAR PHYSICS 236

where

=N Nc
where N is the number of sample frequencies and Nc is the number of constraints. In our case, N is
the number of data points that is 6 and the number of constraints is 1. A t is considered good if the
value of the reduced 2 , that is ~2 is close to one.

In our case, we can calculate the reduced 2 and we get a value of ~2 = 0:8 which indicates that the
t of our energy calibration equation is a very good one.

s
sic
With the calibration done, we can now plot the counts versus energy to get the actual gamma ray

hy
spectrum.

rP
Source: Cs-137 (Gamma Source)
Half-Life: 30.08 years
Activity: 163 kBq
Date of Manufacturing: 01-08-13
lea
uc
lN

LLD (Volts) Counts (Ni ) Energy (keV)


0.5 3874 69.080002
ua

0.6 3862 79.02


0.7 3711 88.97
an

0.8 3623 98.91


0.9 3506 108.86
bM

1.0 3367 118.80


1.1 3510 128.75
1.2 3501 138.69
La

1.3 3416 148.64


1.4 3500 158.58
1.5 3380 168.53
1.6 3563 178.47
1.7 3873 188.42
1.8 4216 198.37
1.9 4225 208.31
2.0 4095 218.26
2.1 3763 228.20
2.2 3424 238.15
LABORATORY MANUAL FOR NUCLEAR PHYSICS 237

2.3 3170 248.09


2.4 2904 258.04
2.5 2662 267.98
2.6 2512 277.93
2.7 2445 287.87
2.8 2276 297.82
2.9 2224 307.76
3.0 2240 317.71
3.1 2137 327.66

s
3.2 2057 337.60

sic
3.3 2035 347.55
3.4 2073 357.49

hy
3.5 2070 367.44

rP
3.6 2065 377.38
3.7 2061 387.332
3.8 2067
lea 397.27
3.9 1934 407.223
4.0 1834 417.16
uc
4.1 1730 427.11
lN

4.2 1391 437.05


4.3 1041 447.00
4.4 757 456.95
ua

4.5 552 466.89


an

4.6 405 476.84


4.7 311 486.78
bM

4.8 276 496.73


4.9 200 506.67
5.0 238 516.62
La

5.1 242 526.56


5.2 270 536.51
5.3 344 546.45
5.4 438 556.40
5.5 694 566.35
5.6 1008 576.29
5.7 1747 586.24
5.8 2618 596.18
5.9 3446 606.13
6.0 3985 616.07
LABORATORY MANUAL FOR NUCLEAR PHYSICS 238

6.1 3727 626.02


6.2 2752 635.96
6.3 1724 645.91
6.4 774 655.85
6.5 316 665.80
6.6 126 675.74
6.7 57 685.69
6.8 46 695.64
6.9 37 705.58

s
sic
Table 10.7: LLD Voltage, Count rate & Energy for 137Cs

hy
rP
4500
lea
Energy Spectrum for Cs-137
uc
"cal11.txt" u 3:2
"cal11.txt" u 3:2
4000
lN

3500

3000
ua

2500
Counts

an

2000

1500
bM

1000

500
La

0
0 100 200 300 400 500 600 700 800
Energy (keV)

Figure 10.18: Energy Spectrum for 137Cs

We can use this graph to nd the energy resolution. For this we need the Full Width at Half Maximum.
From the graph, the peak is at E = 618 keV. The Full Width at Half Maximum, E is 53:8 keV. Then,
by using Eq(10.3), we get
E
R=
E
 100% = 8:71%
LABORATORY MANUAL FOR NUCLEAR PHYSICS 239

Source: Na-22 (Gamma Source)


Half-Life: 2.602 years
Activity: 174 kBq
Date of Manufacturing: 01-08-13

Note: The gain of the ampli er was changed to accommodate the higher energy pulses.Gain: x2.3674

LLD (Volts) Counts (Ni ) Energy (keV)

s
0.0 15632 19.35

sic
0.1 15624 29.29
0.2 15660 39.24

hy
0.3 15557 49.18

rP
0.4 15456 59.13
0.5 16332 69.08
0.6 16926lea 79.02
0.7 18841 88.97
0.8 19093 98.91
uc
0.9 17334 108.86
lN

1.0 14721 118.80


1.1 13490 128.75
1.2 12246 138.69
ua

1.3 10341 148.64


an

1.4 7072 158.58


1.5 4832 168.53
bM

1.6 4414 178.47


1.7 4994 188.42
1.8 6514 198.37
La

1.9 10700 208.31


2.0 19735 218.26
2.1 30084 228.20
2.2 21687 238.15
2.3 6027 248.09
2.4 2324 258.04
2.5 2114 267.98
2.6 2070 277.93
2.7 1896 287.87
2.8 1839 297.82
2.9 1795 307.76
LABORATORY MANUAL FOR NUCLEAR PHYSICS 240

3.0 1788 317.71


3.1 1768 327.66
3.2 1702 337.60
3.3 1741 347.55
3.4 1591 357.49
3.5 1633 367.44
3.6 1592 377.38
3.7 1573 387.33
3.8 1672 397.27

s
3.9 1702 407.22

sic
4.0 1688 417.16
4.1 1679 427.11

hy
4.2 1528 437.05

rP
4.3 1227 447.00
4.4 881 456.95
4.5 535
lea 466.89
4.6 441 476.84
4.7 39 486.78
uc
4.8 409 496.73
lN

4.9 477 506.67


5.0 705 516.62
5.1 1168 526.56
ua

5.2 1639 536.51


an

5.3 1820 546.45


5.4 1367 556.40
bM

5.5 621 566.35


5.6 236 576.29
5.7 125 586.24
La

5.8 109 596.18


5.9 147 606.13
6.0 107 616.07
6.1 122 626.02

Table 10.8: LLD Voltage, Count rate & Energy for 22Na

While plotting the energy spectrum, note that we need to factor in the gain of 2:3674. Consequently,
the voltage values need to be multiplied by this factor before one nds the energy using the calibration
LABORATORY MANUAL FOR NUCLEAR PHYSICS 241

equation.

Energy Spectrum for Na-22


35000
"cal31.txt" u 3:2
"cal31.txt" u 3:2

30000

25000

s
sic
20000
Counts

hy
15000

10000

rP
5000

0
0 200 400 600
lea 800
Energy (keV)
1000 1200 1400 1600
uc
lN

Figure 10.19: Energy Spectrum for 22Na

We can use this graph to nd the energy resolution. For this we need the Full Width at Half Maximum.
ua

From the graph, the peak is at E = 520 keV. The Full Width at Half Maximum, E is 67:4 keV. Then,
an

by using Eq(10.3), we get


bM

E
R=
E
 100% = 12:97%
Source: Co-60 (Gamma Source)
Half-Life: 5.2712 years
La

Activity: 134 kBq


Date of Manufacturing: 01-08-13
Note: The gain of the ampli er was changed to accommodate the higher energy pulses.Gain: x2.3674

LLD (Volts) Counts (Ni ) Energy (keV)


0.0 1077 19.35
0.1 1128 29.29
0.2 1201 39.24
0.3 1311 49.18
LABORATORY MANUAL FOR NUCLEAR PHYSICS 242

0.4 1321 59.13


0.5 1273 69.08
0.6 1187 79.02
0.7 1136 88.97
0.8 1054 98.91
0.9 1222 108.86
1.0 1231 118.80
1.1 1142 128.75
1.2 1000 138.69

s
1.3 911 148.64

sic
1.4 795 158.58
1.5 702 168.53

hy
1.6 655 178.47

rP
1.7 599 188.42
1.8 617 198.37
1.9 567
lea 208.31
2.0 572 218.26
2.1 579 228.20
uc
2.2 543 238.15
lN

2.3 554 248.09


2.4 553 258.04
2.5 525 267.98
ua

2.6 532 277.93


an

2.7 509 287.87


2.8 520 297.89
bM

2.9 502 307.76


3.0 516 317.71
3.1 523 327.66
La

3.2 550 337.60


3.3 578 347.55
3.4 523 357.49
3.5 525 367.44
3.6 587 377.38
3.7 576 387.33
3.8 603 397.27
3.9 455 407.22
4.0 440 417.16
4.1 349 427.11
LABORATORY MANUAL FOR NUCLEAR PHYSICS 243

4.2 333 437.05


4.3 306 447.00
4.4 299 456.95
4.5 273 466.89
4.6 223 476.84
4.7 263 486.78
4.8 417 496.73
4.9 546 506.67
5.0 477 516.62

s
5.1 212 526.56

sic
5.2 75 536.51
5.3 69 546.45

hy
5.4 155 556.40

rP
5.5 257 566.35
5.6 356 576.29
5.7 238
lea 586.24
5.8 96 596.18
5.9 30 606.13
uc
lN

Table 10.9: LLD Voltage, Count rate & Energy for 60Co
ua
an

While plotting the energy spectrum, note that we need to factor in the gain of 2:3674. Consequently,
the voltage values need to be multiplied by this factor before one nds the energy using the calibration
bM

equation.
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 244

Energy Spectrum for Co-60


1400
"cal41.txt" u 3:2
"cal41.txt" u 3:2

1200

1000

800
Counts

s
600

sic
400

hy
200

rP
0
0 200 400 600 800 1000 1200 1400 1600
Energy (keV)

lea 60Co
Figure 10.20: Energy Spectrum for
uc
We can use this graph to nd the energy resolution. For this we need the Full Width at Half Maximum.
lN

From the graph, the peak is at E = 1336 keV. The Full Width at Half Maximum, E is 69 keV. Then,
by using Eq(10.3), we get
ua

E
R=  100% = 5:10%
an

E
Source: Co-57 (Gamma Source)
bM

Half-Life: 271.8 days


Activity: 149 kBq
Date of Manufacturing: 17-08-15
La

LLD (Volts) Counts (Ni ) Energy (keV)


0.0 5594 19.35
0.1 5768 29.29
0.2 2700 39.24
0.3 2109 49.18
0.4 2012 59.13
0.5 2073 69.08
0.6 3350 79.02
0.7 6197 88.97
LABORATORY MANUAL FOR NUCLEAR PHYSICS 245

0.8 12020 98.91


0.9 19935 108.86
1.0 29708 118.80
1.1 52454 128.75
1.2 94828 138.69
1.3 87560 148.64
1.4 31426 158.58
1.5 7605 168.53
1.6 2127 178.47

s
1.7 1403 188.42

sic
1.8 1363 198.37
1.9 1267 208.31

hy
2.0 1385 218.26

rP
2.1 1329 228.20
2.2 1479 238.15
2.3 1364
lea 248.09

Table 10.10: LLD Voltage, Count rate & Energy for 57Co
uc
lN
ua
an
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 246

Energy Spectrum for Co-57


110000
"cal21.txt" u 3:2
"cal21.txt" u 3:2
100000

90000

80000

70000

60000
Counts

50000

s
sic
40000

30000

hy
20000

10000

rP
0
0 50 100 150 200 250
Energy (keV)

lea 57Co
Figure 10.21: Energy Spectrum for
uc
We can use this graph to nd the energy resolution. For this we need the Full Width at Half Maximum.
lN

From the graph, the peak is at E = 143:6 keV. The Full Width at Half Maximum, E is 26:5 keV.
Then, by using Eq(10.3), we get
ua

E
 100% = 18:5%
R=
an

E
Once we have the energy calibration done and have the energy spectrum for the sources that we have
bM

used, we can now nd the back scatter peaks in the spectrum and compare with the theoretical values
from the tables.
La

Backscatter peaks- Experimental and Theoretical values

Source Theoretical Value (keV) Experimental Value (keV) % Error


137Cs 184.4 204.9 11.1 %
22Na 199.8 198.8 0.5 %
60Co 210 & 214 248.7 17.3 %

Table 10.11: Backscatter peaks- Theoretical & Experimental values for various Sources
LABORATORY MANUAL FOR NUCLEAR PHYSICS 247

10.5 Questions
1. What is the di erence between Fluorescence & Phosphorescence
2. How are bands formed in a solid?
3. What is band gap? How is the band gap di erent between an insulator, a conductor
and a semiconductor?
4. What are activators? What is their function in a crystal?
5. Describe the functioning of a photomultiplier tube.

s
sic
6. What is a dynode and why is it needed in a PMT?

hy
7. Describe the various interactions of a gamma ray photon with matter.
8. What is the origin of the Compton Edge in a gamma ray spectrum?

rP
9. How are backscatter peaks formed in a gamma ray spectrum?
lea
10. What is a photopeak and how is it formed?
uc
lN
ua
an
bM
La
Appendix A

p-Value Tables

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La

248
La
bM
an
ua
lN
uc
lea
rP
hy
sic
s
La
bM
an
ua
lN
uc
lea
rP
hy
sic
s
La
bM
an
ua
lN
uc
lea
rP
hy
sic
s
Appendix B

Using MS Excel

s
sic
hy
The experimental data that we obtain in our experiments needs to be analysed. Frequently, this
means computing the errors, plotting the data with the error bars, nding the best t curve through

rP
the data etc. Sometimes, we also need to compare the data with some expected theoretical values,
as for instance in the case of Counting Statistics for G.M Counter (Chapter 5). All this analysis,
lea
including plotting of data can be easily done using the Microsoft Excel package. This appendix will give
some basic tips on how to use MS Excel for the purposes of analysis of data obtained in the experiments.
uc
Microsoft (MS) Excel is a simple spreadsheet program.

R
lN

SOME HANDY TIPS ON MS EXCEL


ua

You can highlight a group of cells by clicking on one cell, holding the mouse button
an

down, and dragging the mouse over the spreadsheet. The cells that are highlighted
will appear black with a black cell border, except for the rst cell highlighted, which
bM

will remain white.

To move the contents of a cell (or many cells) from one place to another, highlight the
La

cell or group of cells, place your cursor on the sides of the cell (on the black outline
of the cell) { the mouse cursor will change to an arrow from a fat cross, click and
hold the mouse button on the border, and drag the cell to its nal destination. The
destination cell will be overwritten!

To copy the cell or group of cells, highlight the cell(s), click Edit Copy. Then highlight
the destination cell(s) and click Edit Paste. If you want to paste the formulas in the
cells or just the values of the cells, you can select Edit Paste Special.

252
LABORATORY MANUAL FOR NUCLEAR PHYSICS 253

You can't solve symbolic equations in Excel. This means that whatever complicated
mathematical expression you type in as an equation must return a number. You can
have cell references in an equation, but the cell that is referenced must contain a
number. If you reference a blank cell, the number 0 is automatically inserted. The
standard operations are +; ; ; =; ab

When you want to type in an equation into a cell, the rst character you type must be
an "=" (equals sign). This tells Excel to evaluate whatever comes after it, otherwise
Excel will just treat it like a string (a bunch of letters) and not evaluate the equation.

s
sic
You can also enter in a cell reference into an equation. While typing the equation,

hy
you can either manually type in the appropriate cell reference or click on the cell
with your mouse. You can drag and ll equations to make a series of equations as

rP
well as dragging and lling in numbers . For example, let's say you had a series of x
values from 0 to 10, incremented by 1. You want to multiply each of those cells by
lea
2 and put those new values in a separate column. The A column has the rst series
of numbers. The B column is where you type in the equation that multiplies the A
uc
column by 2. You can manually type in the equation into B1, referencing A1 in the
equation. Now, you hit enter to enter in the equation into the cell. B1 shows the
lN

value that the equation returns (0  2) = 0. Now select the lower right-hand corner of
the highlighted B1, hold the mouse button down, and drag down to B11.There are
ua

now similar equations in all of the B column cells.


an
bM

The rst step for any analysis is to enter the data. As an example to illustrate some of the things one
can do with MS Excel, we will choose the data for counting statistics from Chapter 5.
La

The data for the preset time of 5 seconds is as follows:


LABORATORY MANUAL FOR NUCLEAR PHYSICS 254

xi fi
0 27
1 55
2 60
3 31
4 16
5 06
6 05

s
Table B.1: Raw Data

sic
hy
rP
lea
uc
lN

Figure B.1: Raw Data


ua

The data in Table B.1 gives us the frequency fi for the number of counts x, with a total number of
P
an

readings that we have taken, which in this case is fi = 200. We need to determine the probability
pei , of getting the various number of counts. Clearly, this is simply
bM

fi
pei = P
fi
La

We can now start using Excel. When you start Excel, you start with a blank "sheet" with rows and
columns identi ed as numbers for rows and Capital letters for columns. Enter the data above in a
blank worksheet. We now have a table where each entry can be identi ed by a number and a letter.
Thus, for instance, A2 will be 0, B 5 will be 31 etc as shown in Figure B.1.
P
Next we need to determine the probability of each of those counts. For this we need fi . It is easy
to do this in Excel. Simply select all the data in column B (which contains the frequencies fi ) and use
the AUTOSUM function which you can see on the top right hand side of the screen. This will give
P
you fi for the data in column B which will be displayed in cell B 9.
To nd the probabilities for each count, we need to use the formula above. Excel provides a way to do
this calculation easily. In the cell C 2, write
LABORATORY MANUAL FOR NUCLEAR PHYSICS 255

= (B2/200)
and press Enter.

What this will do is to take the value of the cell B 2 and divide the value by 200 and display the re-
sult in cell C 2. However, we need to do this for all the data in column B . There is an easy way to do this.

First one needs to enable a functionality of Excel called AUTOFILL. To do this,

s
1. Click the File tab, and then click Options.

sic
2. Click Advanced, and then under Editing options, select or clear the Enable ll handle

hy
and cell drag-and-drop check box to show or hide the ll handle.

rP
3. To avoid replacing existing data when you drag the ll handle, make sure that the
Alert before overwriting cells check box is selected. If you don't want to see a message
about overwriting nonblank cells, you can clear this check box.
lea
Now, you need to simply select the cell C 2, and take the mouse to the LOWER right corner of the
uc
cell. Then while HOLDING the LEFT button of the mouse, simply move down to C 8 (since the
data is from B 2 to B 8). This will automatically perform the SAME operation for the di erent values
lN

of the column B , that is in C 3 it will give B 3=200, in C 4, it will give B 4=200 etc. Thus we have all
the probabilities now with the count rates. To check, we can use AUTOSUM to nd the sum of the
ua

probabilities in column C and display it in C 9. The table will now look like Table B.2
fi pe = Pfifi
an

xi
0 27 0.135
bM

1 55 0.275
2 60 0.300
3 31 0.155
La

4 16 0.080
5 06 0.030
6 05 0.025
200 1
Table B.2: Raw & Probability Data
LABORATORY MANUAL FOR NUCLEAR PHYSICS 256

s
Figure B.2: Raw & Probability Data

sic
However, we need the sample mean and sample variance for this data to analyse and compare the

hy
experimental data with the theoretically expected result. Though Excel has inbuilt statistical functions
for mean and variance AVERAGE & STDEV, one cannot use them with a frequency table. However,

rP
this is easy to do using the de ning formulae for the sample mean and the sample variance. The sample
mean is given by

x =
lea
P
P
xi fi
fi
uc
and the sample variance is given by (Equation 1.7)
lN

" #
N
1 X
E2 = (xi x)2
N1 i=1
ua

To calculate the sample mean, we use the inbuilt formulae in Excel called SUMPRODUCT and
an

SUM. SUMPRODUCT(array 1, array 2) has arguments array 1, array 2, array 3, etc. In our
case, we want to multiply the corresponding data of the column A with the data of column B and then
bM

sum it up. So our array 1 is A2:A8 and array 2 is B2:B8. The function SUM(number 1, number 2,
...) simply sums up numbers number 1, number 2, etc. Therefore to nd the sample mean, we do the
following. We click on any empty cell, say F 10 and enter the formula
La

=SUMPRODUCT(A2:A8,B2:B8)/SUM(B2:B8)
Here remember that we dont need to write A2:A8 by hand. All we need to do is to select the cells from
A2 to A8 and this gets entered automatically. Similarly for B2:B8. To calculate the sample variance,
we use these formulae in another empty cell , say F 11, by typing
=(SUMPRODUCT(A2:A8^ 2,B2:B8)-SUM(B2:B8)*F10^ 2)/(SUM(B2:B8)-1)
where recall that F10 has the sample mean x = 1:94.
p
We thus have the sample mean x and the variance E2 and also E = E2 .
LABORATORY MANUAL FOR NUCLEAR PHYSICS 257

The theoretical expectation is that the distribution follows a Poisson distribution whose probability
distribution function P (xi ; ), is given by
xi 
P (xi ; ) = e
xi !
We take the mean  of the underlying distribution to be the sample mean x and then try to see how
closely the sample distribution follows the expected Poisson distribution with the SAME mean. To do
this, we need to calculate the Poisson distribution function P (xi ; x) for various xi . This is easy to do
in Excel. We select an empty cell, say D2 and enter

s
=(D10^ (A2)*exp(-D10))/fact(A2)

sic
This will give us

hy
1:940  exp( 1:94)=(0!) = 0:1437

rP
Now we need to calculate this for all the values of xi . For this, we again point the cursor to the
BOTTOM RIGHT corner of the cell D2 and pull it down using the LEFT button on the mouse.
lea
Excel then calculates this formula for each value of A2:A8 and displays it in the cells D2:D8. We can
also use AUTOSUM to nd the sum of the Poisson probabilities in D2:D8 and display it in D9 The
uc
table will now look something like this
lN

xi fi pe = Pfifi Poisson
0 27 0.135 0.144
ua

1 55 0.275 0.279
2 60 0.300 0.271
an

3 31 0.155 0.175
bM

4 16 0.080 0.084
5 06 0.030 0.030
6 05 0.025 0.011
La

200 1 0:9961
Table B.3: Data & Poisson Distribution
LABORATORY MANUAL FOR NUCLEAR PHYSICS 258

s
Figure B.3: Raw & Probability Data

sic
Now we are ready to plot this data to get the shape of the probability distribution. Excel also allows us

hy
to do this easily. Go to the Tool Bar and click on INSERT. In the INSERT Menu, click on Scatter in
the CHARTS section. Within the SCATTER section of the CHARTS, click on SMOOTH LINES

rP
WITH MARKERS. An empty window will open on the screen. You can move and resize the window
if you want. Next, click SELECT DATA. This will open a dialog box. Now, remember we want to
lea
plot two curves. One, the plot of the experimental probabilities versus the counts, that is the values in
column A and column C. We also want to plot the theoretically expected Poisson probabilities versus
uc
counts that is the values in column A and column D. So after clicking on SELECT DATA, take the
cursor to A1 and press the LEFT button on the mouse. Press the CTRL key while KEEPING THE
lN

MOUSE BUTTON PRESSED. Now scroll the mouse down to A8 and leave the Mouse Key BUT
KEEP THE CTRL key pressed. Take the cursor to C1 and press the left button on the mouse and
ua

scroll down to C8 and then while keeping the CTRL key pressed once again go to D1 and press the
mouse button and scroll down to D8. Now we have selected three columns, A, C and D. Pressing OK
an

in the dialog box will give you the two graphs in the empty window. That is it! It is as simple as that.
bM
La

Figure B.4: Experimental & Poisson Graph & Data

The graph will look like Figure B.5.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 259

s
Figure B.5: Experimental & Poisson Graph

sic
What about the error bars? The error bars are the error on pe . However, we know that pe is actually

hy
a derived quantity, since

rP
fi
pe = P
fi
P
Now fi = 200 and obviously there is no lea
error in this. We take the error in f as
pf and get the
p i i
error bars by taking the error in pe to be 200fi .
uc
So we need to calculate this quantity for each value of pe . Once again, this is easy to do. In the cell
lN

E2, we can write the formula


=(SQRT(B2)/200)
ua

This will take the value in the cell B 2, which in our case is 27, take its square root and divide by 200.
an

This will give us the error in the corresponding pe . Once again, we need to repeat this for every value
of fi and so we select the cell E2, take the pointer to its lower right corner and while pressing the left
bM

button on the mouse, scroll down to E8. This will take the corresponding values in the cells B3,B4,
etc. and calculate the error by the formula used in E2. Now we have the errors in our experimental
distribution.
La

The screen will look like Figure B.6.


LABORATORY MANUAL FOR NUCLEAR PHYSICS 260

s
Figure B.6: Error Bar Calculation

sic
Next, we need to plot these on the graph we have obtained earlier in Figure B.5. This is easy to do in

hy
MS Excel. Go to the Layout tab on the top. There you will see a tab for Error Bars. When you press
it, it gives you several options like None, Error Bars with Standard Error etc.. Press the button

rP
for More Error Bar Options. This will open a dialog box with Add Error Bars and in our case
give you two options- Adding error bars to the series pE and to Poisson distribution graph. Choose the
lea
series pE . This will open another dialog box with Format Error Bars. Choose the option Vertical
Error Bars and in Error Amount, choose Custom and press the button Specify Value. This will
uc
open another dialog box with specify positive and negative values of the errors. In BOTH choose the
column E2:E8. To do this, take the cursor to E2 and with the mouse button pressed, move down to
lN

E8. This will automatically put the values in these cells in the positive values of errors. DO the same
for the negative values of the errors and press OK. This will draw the error bars on the experimental
ua

data points. The graph will look look Figure B.7.


an
bM
La

Figure B.7: Error Bars on Experimental Curve

You can also change the shape, thickness, color etc of the error bars using the format error bars option.

Now that we have the basic graph done,we can play around with Excel and try to make it better. We
can resize it by just going to one of the corners and clicking on the mouse and dragging the mouse to
make the graph bigger or smaller. Or we can click on the graph and then on the TOOLBAR open
LABORATORY MANUAL FOR NUCLEAR PHYSICS 261

the LAYOUT tab. This allows us various options of choosing the Chart Title (the size, the placement
etc.), Titles for the axis ( the size, the positions, the orientations etc.), Gridlines both in the horizontal
and vertical directions (No lines, major lines in one or both directions, minor lines in one or both
directions, both major and minor lines in one or both directions), changing the colors of the graphs etc.
You are encouraged to try out various options and see how the graph changes. When you are nished,
you can right click on the graph and copy the graph and paste it in one of the imaging programs like
MS Paint and then save it in the format you want (PNG, JPEG, BMP etc.). The nal output could,
for instance, be like in Figure B.8.

s
sic
hy
rP
lea
uc
lN
ua

Figure B.8: Experimental & Poisson Graph with Error Bars


an

We have seen how to plot data graphs in Excel from the given data. Most of the times in our
bM

experiments, we need to t a straight line graph to the data and determine the slope and the intercept
of the data. This too is very simple in Excel. Consider the data that we obtain from the experiment
on veri cation of inverse square law for gamma rays, Chapter 8. We see from Table 8.3, that the data
La

for the count rate and the distance is certainly not linear. However, we expect that the plot between
count rate and d12 will be a straight line. So let us see how to obtain the graph and also the least square t.

Our data is basically as follows :


NB = 18:8= minute
LABORATORY MANUAL FOR NUCLEAR PHYSICS 262

d(cm) N(counts/minute)
2 8728
2.5 4858
3.0 2976
3.5 1969
4.0 1418
4.5 1061
5.0 867
5.5 711

s
6.0 572

sic
6.5 485

hy
7.0 407
Table B.4: Count rates at various distances from 137Cs

rP
We need to rst nd out the count rate net of background per second and then also value of d12 ( in
lea
units of m 2 ) for various d in the data. This is easily done in Excel. Open a new sheet in Excel and
in column B, enter the values of d and in column C, enter the values of N . Next in column A, in the
uc
cell A2 (A1 we will have the labels), enter the formula
lN

= (10000*1/(B2*B2))
This as we know will take the value in the cell B2 and do the operation as speci ed above. But we want
ua

this to be repeated for all the values of column B. So again we do the same AUTOFILL technique
an

described above by clicking on the right bottom corner of A2 and pulling the mouse down all the time
keeping the left mouse button pressed. Now we have the values of d12 . Next we need to nd N 60NB . Now
bM

in the cell D2, enter the formula


= (C2-18.8)/60
La

This will take the value in the cell C2 and perform the operation described. Again we want to repeat
it for all the values in column C. So we use AUTOFILL again to get the values. Now we have all the
data we need. The sheet will look like Figure B.9
LABORATORY MANUAL FOR NUCLEAR PHYSICS 263

s
sic
hy
rP
Figure B.9: Data for Inverse Square law

lea
An important thing to note is that the variable you want to plot on the x-axis should come in the rst
column as we have done above. Now with the data at hand, we are ready to plot the curve and also the
uc
least square t straight line. First, as usual, we go to INSERT Menu on the Toolbar. We then click
the SCATTER chart button and get a choice of charts/graphs we can plot. Choose the SCATTER
lN

WITH SMOOTH LINES AND MARKERS. This will open an empty chart box on the sheet.
Once again, you can move the position and resize the cart area as you wish. Next press the SELECT
ua

DATA button on the Toolbar. This will open a dialog box with Select Data Source. We want to plot
a graph of Rate vs d12 . So we select the cells (by pressing the CTRL key and using the mouse) from
an

A1 to A12 and from D1 to D12. You will see that the chart area gets lled up with the points which
are joined with curve. We next need to label the graph and the axes. To do this, take the mouse to
bM

any point in the Chart Area and press the left mouse key. This will open the CHART TOOLS in
the toolbar at the top. In this toolbar, press LAYOUT. Now double click on the Title in the Chart
Area and enter any text. In our case, we enter the title of the graph, namely "R vs d12 ". Next we want
La

to label the axes. In the LAYOUT toolbar, press the AXIS TITLES button. It will ask you about
the placement and the orientation of the titles. Enter the titles that you want on the x and y axis. In
our case it is Rate and d12 . Lastly, we see that the graph is approximately a straight line. So we want
to t a straight line to the data using the Method of Least Squares. This is easy to do with Excel.
In the LAYOUT toolbar, there is a button TRENDLINES. When you press this, you get many
options of choosing the kind of curve you want to t. Choose a MORE OPTIONS and in this choose
LINEAR. Also check the box at the bottom of the dialog box which says DISPLAY EQUATION
ON CHART. This will plot the best t straight line and also give you the equation of the line. In our
case it is
LABORATORY MANUAL FOR NUCLEAR PHYSICS 264

y = 0:0593x 10:237
The sheet would now look something like Figure B.10.

s
sic
hy
rP
lea
uc
Figure B.10: Graph of Rate vs d12 and the Best t line
lN

You can copy the chart area with the graphs and then use any image processing program like MS Paint
or Photoshop etc to resize and do other things to make the graph look better. The graph will nally
ua

look like Figure B.11


an
bM
La

Figure B.11: Graph of Rate vs d12 and the Best t line


LABORATORY MANUAL FOR NUCLEAR PHYSICS 265

MS Excel is a very powerful tool to do many other things like statistical analysis etc. You can try out
the various functionalities of the software if you want.

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La
Appendix C

Using Gnuplot

s
sic
hy
C.1 Introduction

rP
Most of you are familiar with MS-Excel and we have seen how to use it to analyse data and plot graphs
in the previous chapter. However, if one is using a Linux based machine, then the chances are that one
does not have access to MS Excel. However, Linux has a package called Gnuplot which comes with
lea
Linux. It is a very powerful package which allows on to generate two- and three-dimensional plots of
functions and data. The program runs on all major computers and operating systems (Linux, UNIX,
uc
Windows, Mac OSX...).
lN

The software is copyrighted but freely distributed (i.e., you don't have to pay for it). It was originally
ua

intended to function as a software for plotting mathematical functions and data but has outgrown
its credentials. It now supports many non-interactive uses, including web scripting and integration
an

as a plotting engine for third-party applications like Octave. Gnuplot has been supported and under
development since 1986.
bM

Gnuplot supports many types of plots in either 2D and 3D. It can draw using lines, points, boxes, con-
tours, vector elds, surfaces, and various associated text. It also supports various specialized plot types.
La

Gnuplot supports many di erent types of output: interactive screen terminals (with mouse and hotkey
functionality), direct output to pen plotters or modern printers (including postscript and many color
devices), and output to many types of graphic le formats (eps, g, jpeg, LaTeX, metafont, pbm, pdf,
png, postscript, svg, ...). Gnuplot is easily con gurable to include new devices.

R THIS CHAPTER ASSUMES YOU ARE USING A LINUX MACHINE OR A


LINUX EMULATOR ON A WINDOWS MACHINE

266
LABORATORY MANUAL FOR NUCLEAR PHYSICS 267

C.2 Plotting with inbuilt functions of GNUPLOT


We will rst demonstrate gnuplot using built-in functions.

C.2.1 Interactive plotting


In the following example, we plot cos(x) between 2 < x < 2 with labels on x and y axis.

Open a terminal and type at the prompt

s
$gnuplot

sic
You will see a screen like this

hy
rP
lea
uc
lN
ua
an
bM

Figure C.1: Screen Shot of opening GNUPLOT


La

On the gnuplot prompt type the following lines one by one (self explanatory)

gnuplot> set xlabel `x'


gnuplot> set ylabel `Cos(x)'
gnuplot> set grid
gnuplot> set title `Cos Function'
gnuplot> plot [-2*pi : 2*pi] cos(x) w lp
LABORATORY MANUAL FOR NUCLEAR PHYSICS 268

You will get a plot like this

s
sic
hy
Figure C.2: Screen Shot of Cos(x)

rP
To turn o the grid, you can \unset grid", to turn o the xlabel, you can type \set xlabel ' ' ". Type
set at the gnuplot prompt to see all of the options you can turn on and o . To turn on auto-scaling
lea
(without any `x' or `y' labels: default), type \set auto" at gnuplot> prompt.

R
uc
Note that many of the gnuplot keywords including: using, title, and with can be
abbreviated with a single alphabet as: u, t and w but should be avoided by beginners.
lN

Also, each line and point style has an associated number.


ua

In order to draw two plots on top of each other, you can replace the last line by
an
bM

gnuplot>plot [-2*pi : 2*pi] sin(x) t `Sine Wave' with linespoints, cos(x) t `Cosine Wave'
with linespoints
La

Figure C.3: Screen Shot of Sin(x) & Cosine(x)


LABORATORY MANUAL FOR NUCLEAR PHYSICS 269

You can exit from gnuplot by typing \exit" (or \quit") on the gnuplot prompt.

gnuplot>exit

C.3 Saving Plots


To save the above image as an enhanced postscript le with \.eps" or as a postscript le with \.ps"
extension, instead of displaying it to the screen, enter the following commands:

s
sic
gnuplot> set terminal postscript
gnuplot>set output `plot.ps'

hy
gnuplot>set xlabel `x'
gnuplot>set ylabel `y'

rP
gnuplot>set title `Sine Wave'
gnuplot> plot [-2*pi : 2*pi] sin(x) w lp lea
This will create a postscript image le called \plot.ps" of the previous plot. It will be placed in the
uc
same folder in which you are working. You can then use any postscript viewer program like \gv" (or
\evince") to open your saved graphics le. Remember, the plot will not appear on the screen
lN

when you redirect the terminal type to postscript ( rst line of the example above), so it
may appear as if nothing has happened. Exit from gnuplot prompt, and then type on the terminal
ua

$gv plot.ps &


an
bM

You can also use any converter available on the Internet (like ps2pdf.com or online2pdf.com) to convert
the postscript (.ps) les to PDF les which can then be used.
La

C.3.1 Customization
Customization of the axis ranges, axis labels, and plot title, as well as many other features, are speci ed
using the set command. Speci c examples of the set command follow. (The numerical values used in
these examples are arbitrary.) To view your changes type: replot at the gnuplot> prompt at any time.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 270

Action Command
Create a title: > set title "Beta Particle Spectrum"
Put a label on the x-axis: > set xlabel "Energy (keV)"
Put a label on the y-axis: > set ylabel "Intensity"
Change the x-axis range: >set xrange [0.001:0.005]
Change the y-axis range: > set yrange [20:500]
Have Gnuplot determine ranges: > set autoscale
Put a label on the plot: > set label "Q-Value" at 0.003, 260
Remove all labels: > unset label
Plot using log-axes: > set logscale

s
Plot using log-axes on y-axis: > unset logscale; set logscale y

sic
Change the tic-marks: > set xtics (0.002,0.004,0.006,0.008)
Return to the default tics: >unset xtics; set xtics auto

hy
Other features which may be customized using the set command are: arrow, border, clip, contour,

rP
grid, mapping, polar, surface, time, view, and many more. The best way to learn is by reading the
on-line help information, trying the command, and reading the Gnuplot manual.
lea
FUNCTION RETURNS
uc
abs(x) absolute value of x, jxj
acos(x) arc-cosine of x
lN

asin(x) arc-sine of x
atan(x) arc-tangent of x
ua

cos(x) cosine of x, x is in radians.


cosh(x) hyperbolic cosine of x, x is in radians
an

erf(x) error function of x


exp(x) exponential function of x, base e
bM

inverf(x) inverse error function of x


invnorm(x) inverse normal distribution of x
log(x) log of x, base e
La

log10(x) log of x, base 10


norm(x) normal Gaussian distribution function
rand(x) pseudo-random number generator
sgn(x) 1 if x > 0; 1 if x < 0; 0 if x = 0
sin(x) sine of x, x is in radians
sinh(x) hyperbolic sine of x, x is in radians
sqrt(x) the square root of x
tan(x) tangent of x, x is in radians
tanh(x) hyperbolic tangent of x, x is in radians

Table C.1: Some Inbuilt Functions in Gnuplot


LABORATORY MANUAL FOR NUCLEAR PHYSICS 271

Bessel, Beta and Gamma functions are also supported. Many functions can take complex arguments.
Binary and unary operators are also supported. The supported operators in Gnuplot are the same
as the corresponding operators in the C programming language, except that most operators accept
integer, real, and complex arguments. The ** operator (exponentiation) is supported as in FORTRAN.
Parentheses may be used to change the order of evaluation. The variable names x, y, and z are used as
the default independent variables.

C.4 Plotting using data from a le

s
sic
This is the section which will be most important for us. We normally will have our data and we would
like to plot it. First we need to create a le with the data in it. Let us suppose that the le is called

hy
data1.txt and it looks something like Table C.2.

rP
#n n2 n3
1 1 1
2
3
lea
4
9
8
27
4 16 64
uc
5 25 125
lN

6 36 216
7 49 343
8 64 512
ua

9 81 729
10 100 1000
an

Table C.2: Experimental Data


bM

Again start gnuplot in a terminal by writing \gnuplot" on the terminal prompt and type the following
La

line

gnuplot> plot \data1.txt" u 1:2 w lp,\data1.txt" u 1:3 w lp

gnuplot ignores lines starting with # (comment lines.) Also, you can combine any number of plots in
one gure. Thus, in this example, we have plotted n vs n2 and n vs n3 by the command u 1:2 and
then again u 1:3. It should be clear that the 1 refers to the rst column of the data le and the 2 and
3 refer to the second and third column. Similarly, one can plot data from di erent data les in this
LABORATORY MANUAL FOR NUCLEAR PHYSICS 272

manner. The above command will create a plot like that in Fig C.4.

s
sic
Figure C.4: Screen Shot of Plot using data le

hy
rP
You can modify these plots again using the x and y-labeling.

You can also save the above plot in a .ps le using the following commands:
lea
uc
gnuplot>set terminal postscript
lN

gnuplot>set output `plots2.ps'


gnuplot>plot \data1.txt" u 1:2 w lp, \data1.txt" u 1:3 w lp
ua
an

Exit the gnuplot and then type gv plots2.ps on the terminal command prompt to see the output.
bM
La

C.5 Plotting using data from le and tting to a smooth curve


Frequently, in our experiments we will get data which visually looks scattered. Now if we join the
points on a plot, we get a set of connected straight lines. This obviously is not how nature would
behave. We all know that in nature there would be no discontinuities and so we would like to join
the points with a smooth curve rather than a collection of straight lines. Of course, some plots are
straight line plots and these can be obtained easily by using the Method of Least Squares as we saw in
Section 1.7 in Chapter 1. For other kinds of plots, we basically need to interpolate between points to
get a smooth curve. Gnuplot allows one to do this easily. To demonstrate this, consider the example
of Counting Statistics from Section 5.3.2 in Chapter 5. For this example we will take the data for a
preset time of 5 seconds as given in Table 5.3.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 273

xi fi xi fi (xi x) (xi x)2 pe = Pfifi (xi x)2 p


0 27 0 -1.94 3.76 0.135 0.507
1 55 55 -0.94 0.880 0.275 0.242
2 60 120 0.06 0.086 0.300 0.010
3 31 93 1.06 1.12 0.155 0.173
4 16 64 2.06 4.24 0.080 0.339
5 06 30 3.06 9.86 0.030 0.280
6 05 30 4.06 16.48 0.025 0.412
P P P P

s
fi = 200 xi fi = 387 pe = 1 E = (xi x)2 p = 1:96
2

sic
Table C.3: Counting Statistics for Preset time = 5 seconds

hy
rP
From this we can derive the probabilities since we know the frequencies and the total number of readings
as explained in Eq(5.3).
f
lea
pe = P i (C.1)
fi
The error in the experimental data can also be easily calculated since we know that pe is actually
uc
a derived quantity and therefore the error has to be evaluated using the error propagation equa-
P
lN

tion. However, we know that the denominator, that is fi has no error since it is xed at 200p in
p
our experiment. We take the error in fi as fi and get the error bars by taking the error in pe to be 200fi .
ua

Further, we expect that the counting statistics follows a Poisson distribution with a mean given by the
an

sample mean which can be calculated. This table is given below.


bM

xi PPoisson pe
0 0.144 0.135
1 0.279 0.275
La

2 0.271 0.300
3 0.175 0.155
4 0.084 0.080
5 0.030 0.030
6 0.011 0.025
Table C.4: Experimental and Poisson probabilities: Preset Time 5 seconds

Now let us see how to plot these. First let us use Gnuplot to plot the experimental data which is stored
in a le called \nuc1.dat" for instance which you have created. We will also need the error bars on the
experimental data. The data with the error bars thus becomes
LABORATORY MANUAL FOR NUCLEAR PHYSICS 274

xi PPoisson pe error in pe
0 0.144 0.135 0.026
1 0.279 0.275 0.037
2 0.271 0.300 0.038
3 0.175 0.155 0.028
4 0.084 0.080 0.020
5 0.030 0.030 0.012
6 0.011 0.025 0.011

s
Table C.5: Experimental and Poisson probabilities with errors: Preset Time 5 seconds

sic
Now we are ready to plot this data using Gnuplot. First let us plot the raw data for a plot of pe vs xi .

hy
rP
gnuplot> plot \nuc1.dat" u 1:3 w lp

since the experimental data is in the rst and third column in the Table C.5. This will generate a graph
lea
as in Figure C.5.
uc
lN

0.3
"nuc1.dat" u 1:3
ua

0.25
an

0.2
bM

0.15
p

0.1
La

0.05

0
0 1 2 3 4 5 6
x

Figure C.5: Counting Statistics : Preset 5 seconds: Experimental Data

Next we need to get the graph with the errorbars. For this, we need to just do the following
LABORATORY MANUAL FOR NUCLEAR PHYSICS 275

gnuplot> plot \nuc1.dat" u 1:3 w lp, \nuc1.dat" u 1:3:4 w yerrorbars

The graph will look like that in Figure C.6

0.35
Exp. data
Exp. data with errorbars

0.3

s
0.25

sic
0.2

hy
y

0.15

rP
0.1

0.05

0
lea
uc
0 1 2 3 4 5 6
x
lN

Figure C.6: Counting Statistics : Preset 5 seconds: Data with errorbars


ua

But this is not a smooth curve. So we need to use some interpolation to generate a smooth curve.
an
bM

C.5.1 Curve Fitting & Interpolation


Typically in any experiment, we will get a set of data points (x; y). It is a good assumption that there
is some function f (x) which will generate this set of data. The function f (x) does not only generate the
La

set of data points, but should represent all the non-data points that could be generated by the partic-
ular process that the experiment is using. The problem is that we don't know what the function f (x) is.

In Interpolation , we try to nd a function, say g(x) which approximates the function f (x) as best as
we can. This function should pass through all the data points that we have got from our experiment.
In addition, we believe that the interpolating function also will give us the values of the variables at
the non-data points. This is what makes interpolation useful.

In Curve Fitting we are NOT doing this. Instead, when we determine the best- t line, say by using
the Method of Least Squares, we are basically only nding the best t curve to the set of data points.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 276

In this case, the approximating function does not have to pass through all the data points. This point
is very important to remember. Curve tting usually gives a good idea of the trend of the data. It
cannot be used to determine the values in between the data points. The Method of Least Squares
that we learnt in Section 1.7 only used a linear t. Recall that in the case of a linear t, we needed
to determine 2 parameters, namely, the slope m of the best t line and c its y-intercept. We can
easily generalise the discussion in Section 1.7 to use instead of a straight line with two parameters, a
higher order polynomial. A typical Least Square tting can, for instance, use a fth order polynomial.
Normally, in the case of higher order polynomials, the determination of the coecients becomes a bit
dicult algebraically. Therefore one uses a standard mathematical package like Scilab or Matlab

s
which does it for you.

sic
Now let us consider Interpolation. The simplest interpolation would of course be a linear interpolation

hy
where we just join the data points with a straight line. An example of this, for our case is Figure C.5.

rP
Obviously we want a better curve rather than simply joining the adjacent points with a straight line.
This is usually done with a cubic polynomial instead of a polynomial of order 1 that is a linear function.
If one uses a cubic polynomial, then one will need to determine the coecients of the polynomial to
lea
get it. This is usually done by using the data points and solving a system of simultaneous equations.
uc
However, by simply using a cubic interpolation, one might have a problem of smoothness. This is
lN

overcome by spline functions. These are basically di erent polynomials between the datapoints which
are piecewise continuous and have a high degree of smoothness where the polynomials meet. The most
often used spline functions are cubic splines or csplines. As the name suggests, these use cubic
ua

polynomials between data points.


an

A natural smoothing spline approximation can be done by using csplines between adjacent points but
bM

weighting the coecients with not just the adjacent datapoints but the ones which are further o also.
Obviously, the weight of the nearest data points in the cubic polynomial will be the most and that of
the farthest will be the least. Once again, spline functions don't have to be cubic though these are the
La

most often used. The spline function can be a polynomial of any order.

Gnuplot has several di erent techniques of interpolation. These are for instance, acsplines, csplines,
bezier etc. Acsplines uses a cubic polynomial piecewise where the coecients are weighted with
weights to generate a smooth curve. A bezier smoothing uses a bezier curve to smooth the data while
csplines uses a natural cubic polynomial. A bezier curve is a curve based on Bernstein polynomials
and is used to generate smooth curves frequently in computer graphics applications. To generate a
smooth curve, we do the following

gnuplot> plot \nuc1.dat" u 1:3:4 w yerrorbars, \nuc1.dat" u 1:3 smooth csplines,


LABORATORY MANUAL FOR NUCLEAR PHYSICS 277

This will generate a graph as in Figure C.7

0.35
"nuc1.dat" u 1:3:4
Exp.data

0.3

0.25

s
sic
0.2
p

0.15

hy
0.1

rP
0.05

0
0 1 2
lea 3
x
4 5 6
uc
lN

Figure C.7: Counting Statistics : Preset 5 seconds: Data with errorbars & smoothed
ua

Finally we need to see how well this experimental data ts the expected Poisson distribution. For
this,we plot the probabilities of the Poisson distribution with the sample mean as in Column 2 of Table
an

above and draw a smooth line between them. For this, we do the following
bM

gnuplot> plot \nuc1.dat" u 1:3:4 w yerrorbars, \nuc1.dat" u 1:3 smooth csplines,


\nuc1.dat" u 1:2, \nuc1.dat" u 1:2 smooth csplines
La

This will generate a graph as in Figure C.7


LABORATORY MANUAL FOR NUCLEAR PHYSICS 278

0.35
"nuc1.dat" u 1:3:4
Exp.data
"nuc1.dat" u 1:2
Poisson
0.3

0.25

0.2
p

0.15

s
sic
0.1

hy
0.05

rP
0
0 1 2 3 4 5 6
x

lea
Figure C.8: Counting Statistics : Preset 5 seconds: Data with errorbars & smoothed & Poisson distribution
uc
One can do many other things with Gnuplot. You can explore the other features of the software and
lN

learn more about it from any of the references on the Internet.


ua

R Some of the common errors & good practices while using gnuplot
an
bM

1. In the plot statement, the name of the data le to be plotted should always be in
\ ".
La

2. Before you plot any data from a le, make sure you check the le and con rm
that it has the data that you expect.
3. Always rst see the plot without any extra labels, titles etc. on your screen. If
it is of the form that you expect, then only go and add all the extra things like
labels, titles, legends etc and save as .ps le.
4. Do not use very long and complicated names for your data les or the graphic
les (.ps) since then there is a chance you will make a mistake when trying to
enter the names. Short, descriptive names are best.
Appendix D

Radioactive Decay Equilibrium

s
sic
hy
D.1 Introduction

rP
We saw in Section 2.1.2 that typically a radioactive nuclide is part of a decay chain consisting of parent
and daughter nuclides etc. This process can be represented by
lea
A ! B ! C ! 
uc
The question we need to ask is, given the half lives of the di erent nuclides in the decay chain, what can
we say about the behaviour with time? In particular, we will talk about di erent kinds of equilibrium
lN

con gurations that can be analysed depending on the half lives.


ua

D.2 Bateman Equation


an

Let us consider a decay chain as shown above. Let the nuclides A; B; C;    have decay constants
bM

1 ; 2 ; 2    . Let the number of nuclides of these types at time t be N1 ; N2 ; N3 ;    and their number
at time t = 0 be N10 ; N20 ; N30 ;    . Then we know that the number of nuclei of type A , that is the
parent nuclide, at any time t is given by the Activity Law, that is
La

N1 = N10 e 1 t (D.1)
and its rate of change is given by
dN1
= 1 N1 (D.2)
dt
Similarly, the nuclide of type B which is being produced by A in its decay can be analysed. However,
remember that while it is being produced by the decay of A it itself is also decaying into the grand-
daughter nuclide C simultaneously. The rate of production of B is obviously equal to the rate of decay
of A and its rate of decay is given by its own decay constant 2 . Thus we can write

279
LABORATORY MANUAL FOR NUCLEAR PHYSICS 280

dN2
= 1 N1 2 N2 (D.3)
dt
Substituting Eq D.1 into Eq D.3, we get
dN2
= 1 N10 e 1 t 2 N2 (D.4)
dt
or
dN2
+ 2 N2 = 1 N10 e 1 t (D.5)
dt

s
sic
Multiplying both sides by e2 t we get
dN2 2 t
+ e 2 N2 = 1 N10 e(2 1 )t

hy
e2 t (D.6)
dt
The left hand side is a complete di erential and we get

rP

d N2 e2 t = 1 N10 e(2 1 )t dt
lea (D.7)
This can now be integrated to give us
uc
1  
N2 e2 t = N10 e(2 1 )t + C (D.8)
(2 1 )
lN

where C is the integration constant. To determine C we need to use the initial condition namely,
N2 (t = 0) = N20 . Then
ua

1
C = N20 N
an

(2 1 ) 10
Thus we get for the number of nuclide B at any time t to be
bM

R N2 (t) =
1 
N10 e 1 t

e 2 t + N20 e 2 t (D.9)
La

(2 1 )
In terms of activity, A, which we recall is simply
dN
A= = N
dt
we can write Eq D.9 as

R A2 (t) =
(2
2
1 )

A10 e 1 t

e 2 t + A20 e 2 t (D.10)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 281

The interpretation of this equation is obvious. The rst term on the right is the number of atoms
produced from the decay of A which have not yet decayed. The second term is the number of atoms of
B which remain from the initial number N20 that is the number that has not decayed since remember
that B is decaying to C .

Now we turn to the nuclide C . This is being produced by the decay of B and is also decaying simulta-
neously. Thus its number at any time is given by
dN3
= 2 N2 3 N3 (D.11)
dt

s
sic
But N2 is given by Eq D.9. Substituting, we get
dN3 1 2  

hy
+ 3 N3 = N10 e 1 t e 2 t + N20 e 2 t (D.12)
dt (2 1 )

rP
At this point,we can make a simpli cation and assume that there are no daughter (B ) or granddaughter
nuclide (C ) at time t = 0, that is N20 = N30 = 0. In other words, we assume a pure sample of the
parent nuclide. Then Eq D.12 becomes lea
dN3 1 2  
+ 3 N3 = N10 e 1 t e 2 t (D.13)
uc
dt (2 1 )
Once again multiplying by e3 t on both sides, we get
lN

1 2
  
d N3 e3 t = N10 e(3 1 )t e(3 2 )t dt (D.14)
ua

(2 1 )
Integrating this and using the initial condition that N30 = 0, we get the number of nuclides of type C
an

at any time t as
bM

R N3 (t) = N10 1 2


(2
e 1 t
1 )(3 1 )
+
(1
e 2 t
2 )(3 2 )
+
(1
e 3 t
3 )(2 3 )

(D.15)
La

This equation and its solutions are called Bateman Equations . They were proposed rst by H.
Bateman in 1910.

In the general case, for the situation where all the daughter nuclides are not present at time
t=0, that is N20 = N30 =    = Nn0 = 0, the solutions are given by

Nn = C1 e 1 t + C2 e 2 t +    + Cn e n t (D.16)
where the constants C1 ; C2 ;    are given by
LABORATORY MANUAL FOR NUCLEAR PHYSICS 282

1 2    n 1
C1 = N
(2 1 )(3 1 )    (n 1 ) 10
1 2    n 1
C2 = N
(1 2 )(3 2 )    (n 2 ) 10
..
. (D.17)
1 2    n 1
Cn = N (D.18)
(1 n )(2 n )    (n 1 n ) 10
One can immediately see how useful these equations and their solutions will be. For instance knowing

s
sic
the initial activity of a pure sample of a radioactive material, we can easily nd the activities of all the
daughter nuclides in the decay chain at any time.

hy
Example D.2.0.1
Consider the decay of a sample weighing 2 g of pure 208O radioisotope. The isotope decays by

rP
beta decay by the radioactive chain

8 9 10
lea
20O ! 20F ! 20Ne stable
uc
The half life of the 208O is 13:51 seconds and that of 209F is 11:163 seconds. Calculate the activity
lN

of the sample after 1 minute.


ua

The rst thing we need to do is to calculate the decay constants of the two isotopes. This is simply
an

ln 2
1 = = 0:0513sec 1
1
bM

ln 2
2 = = 0:0621 sec 1
2
Now we need to calculate the activity of the initial sample.
La

 
2  10 6 
A10 = 1 N10 = 0:0513   6:023  1023 = 3:09  1015 Bq = 8:35  104 Ci
20
Now we can use Eq D.10 to calculate the activity at time t = 60 seconds since we know that the
initial activity of nuclide 2 , A20 = 0. Then

2  
A2 = A10 e 1 t e 2 t
2 1
Putting in the numbers we get
LABORATORY MANUAL FOR NUCLEAR PHYSICS 283

A2 (t = 60) = 1:44  104 Ci

D.3 Di erent Kinds of Equilibrium


Looking at the Activity equations, we can think of three di erent situations in radioactive decay chains
depending on the relative values of the half lives of the parent and daughter nuclides. These are
1. No Equilibrium

s
sic
2. Transient Equilibrium

hy
3. Secular Equilibrium
When the half life of the parent nuclide is shorter than that of the daughter nuclide, we have a situation

rP
of No Equilibrium. Remember that a shorter half life means a higher decay constant. In this case,
the parent decays much faster than the daughter and therefore the build up of the daughter nuclide is
lea
much faster than its own decay. Essentially, after a few half lives, all of the parent has decayed and
all the subsequent activity is only due to the daughter nuclide. Examples of this kind of situation are
uc
131Te ! 131I; 210Bi ! 210Po; 92Sr ! 92Y which are all examples of decay towards higher stability. An
example for 138 138
54Xe ! 55Cs is shown in Figure D.1.
lN
ua

No Equlibrium
an

1
Parent
Daughter
0.9
bM

0.8

0.7

0.6
Activity
La

0.5

0.4

0.3

0.2

0.1

0
0 10 20 30 40 50 60 70 80 90 100
Time(m)

Figure D.1: No Equilibrium

Second case is that of Transient Equilibrium. Here the half lives of the parent and the daughter
are of the same order, though the half life of the parent is of the order of 10 times the half life of the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 284

daughter. Now the parent controls the decay chain. This can be easily seen from Eq D.9.

Now we have

2 > 1
Then as t ! 1, we have

e 2 t  e 1 t

s
and therefore

sic
N20 e 2 t ! 0

hy
Therefore in Eq D.9

rP
1 
N2  N10 e 1 t = N1 1
2 1
lea 2 1
Thus
uc
N1 2 1
=
N2 1
lN

We see that at long times, the ratio of the daughter and parent activity becomes a constant. Example
of transient equilibrium is 140Ba ! 140La where the half life of the parent is 12:8 days and that of the
daughter is 40 hours. An example for 99 99m
ua

42Mo ! 43Tc is shown in Figure D.2.


an
bM

Transient Equlibrium
1
Parent
Daughter
0.9
La

0.8

0.7

0.6
Activity

0.5

0.4

0.3

0.2

0.1
0 10 20 30 40 50 60 70 80 90 100
Time(h)

Figure D.2: Transient Equilibrium


LABORATORY MANUAL FOR NUCLEAR PHYSICS 285

Finally we have a situation of Secular Equilibrium. Here the half life of the parent is many times
( 104 ) that of the daughter and therefore there is no signi cant change in the parent population during
observation. In this case, as in the case of Transient Equilibrium, we can see that

2  1
and therefore
N1 2 1
N2
=
1
 2
1

s
or

sic
A1 = A2

hy
Examples of these are the naturally occurring heavy element chains like 238U ! 206Pb or 235U ! 207Pb

rP
because of very long half lives of the parent nuclides. An example for 144 144
58Ce ! 59Pr is shown in Figure
D.3.
lea
uc
Secular Equlibrium
1.2
lN

Parent
Daughter

1
ua

0.8
Activity

an

0.6

0.4
bM

0.2

0
La

0 50 100 150 200 250 300


Time(m)

Figure D.3: Secular Equilibrium

D.4 Numerical Integration of Bateman Equations


Although we have given above the exact solutions of the Bateman equations in the general case, these
may not be easy to obtain. A simpler method is to numerically solve the equations governing the
number of nuclides (like, for instance, in the case of 3 nuclides, Eq D.2, Eq D.3 and Eq D.11)using
any standard method to solve di erential equations. A powerful method to do this is to use the
LABORATORY MANUAL FOR NUCLEAR PHYSICS 286

Runge-Kutta (RK) method. In our case we use RK-4 method. The input in this method is the rate
equations that is the equations governing the number of nuclides, the decay constants of the various
nuclides and the initial conditions, that is the number of nuclides of di erent kinds present at time
t = 0. Once these are used as inputs, we can write a program to solve for any number of nuclides in a
decay chain.

As an example consider a pure sample of a nuclide X with a half life of 14:08 minutes. This implies
that the initial sample has only nuclide X and the initial amounts of other nuclides is 0. The nuclide
X decays to another nuclide Y which has a half life of 33:41 minutes to a nuclide Z . The nuclide

s
Z is stable. The rate equations are obviously coupled since the solution of one enters into the other

sic
equations. The program below uses RK-4 method to solve three coupled equations for the change of
numbers of nuclides X; Y and Z . It also evaluates the exact solutions obtained above (EqD.1, EqD.9

hy
and EqD.15). The evolution of the numbers of the three nuclides are then plotted in both cases. These

rP
are shown in Fig D.4 and Fig D.5 respectively. We can see that the numerical solution and the exact
solutions are identical.
lea
uc
/  S o l v i n g Bateman Equations Numerically  /
/  s o l v i n g dx / d t = l 1  x ,
lN

dy/ d t = l 1  x l 2  y ,
and dz / d t = l 2  y l 3  z u s i n g RK4 method  /
ua

/  l1 , l2 , l 3 are t h e decay c o n s t a n t s o f parent , d a u g h t e r and granddaughter  /


/  t1 , t 2 are h a l f l i v e s  /
/  x1 i n i t i a l amount o f parent . i n i t i a l amounts o f d a u g h t e r and granddaughter taken t o be 0  /
an

/  x , y , z are t h e number o f parent , daughter , granddaughter a t time t from n


numerical i n t e g r a t i o n  /
bM

/  x2 , y2 , z2 are number o f parent , daughter , granddaughter from e x a c t s o l u t i o n  /


/  ax , ay , az are a c t i v i t i e s  /
#include< s t d i o . h>
#include<math . h>
La

#define f 1 ( t , x , y , z ) ( l 1 ( x ) )
#define f 2 ( t , x , y , z ) ( l 1  ( x) (y )  l 2 )
#define f 3 ( t , x , y , z ) ( l 2  ( y) l 3  ( z ) )
main ( )
f
float t 1 =14.08 , t 2 =33.41 , h = 0 . 0 1 ; /  h a l f l i v e s and s t e p s i z e  /
float t , x , y , z , k1 , k2 , k3 , k4 , m1, m2, m3, m4, n1 , n2 , n3 , n4 , l 1 , l 2 , l 3 =0;
float ax , ay , az ; /  a c t i v i t i e s  /
float x1 =1, y1 =0, z1 =0; /  i n i t i a l v a l u e s o f t h e n u c l i d e s  /
float x2 , y2 , z2 ;
FILE  f p=NULL;
f p=f o p e n ( " r e s . t x t " , "w" ) ;
LABORATORY MANUAL FOR NUCLEAR PHYSICS 287

FILE  f p 1=NULL;
f p 1=f o p e n ( " r e s 1 . t x t " , "w" ) ;
l 1=l o g ( 2 ) / t 1 ;
l 2=l o g ( 2 ) / t 2 ;
t = 0 . 0 ; x = x1 ; y = y1 ; z = z1 ;

do
f k1 = h f 1 ( t , x , y , z ) ;
m1 = h f 2 ( t , x , y , z ) ;
n1 = h f 3 ( t , x , y , z ) ;
h  f 1 ( t+h / 2 , x+k1 / 2 , y+m1/ 2 , z+n1 / 2 ) ;

s
k2 =

sic
m2 = h  f 2 ( t+h / 2 , x+k1 / 2 , y+m1/ 2 , z+n1 / 2 ) ;
n2 = h  f 3 ( t+h / 2 , x+k1 / 2 , y+m1/ 2 , z+n1 / 2 ) ;
h  f 1 ( t+h / 2 , x+k2 / 2 , y+m2/ 2 , z+n2 / 2 ) ;

hy
k3 =
m3 = h  f 2 ( t+h / 2 , x+k2 / 2 , y+m2/ 2 , z+n2 / 2 ) ;
n3 = h  f 3 ( t+h / 2 , x+k2 / 2 , y+m2/ 2 , z+n2 / 2 ) ;

rP
k4 = h  f 1 ( t+h , x+k3 , y+m3, z+n3 ) ;
m4 = h  f 2 ( t+h , x+k3 , y+m3, z+n3 ) ;
n4 = h  f 3 ( t+h , x+k3 , y+m3, z+n3 ) ;
x
y
=
=
x+(k1 +2.0  ( k2+k3)+k4 ) / 6 . 0 ;
y+(m1+2.0  (m2+m3)+m4 ) / 6 . 0 ;
lea
uc
z = z+(n1 +2.0  ( n2+n3)+n4 ) / 6 . 0 ;
t = t+h ;
lN

/  Exact S o l u t i o n o f Bateman Equations  /


x2=x1  exp( l 1  t ) ;
ua

y2=( l 1 / ( l 2 l 1 ) )  x1  ( exp( l 1  t ) exp( l 2  t ))+ y1  exp( l 2  t ) ;


z2=l 1  l 2  ( ( exp( l 1  t ) / ( ( l 2 l 1 )  ( l 3 l 1 ) ) ) + ( exp( l 2  t ) / ( ( l 1 l 2 )  ( l 3 l 2 ) ) ) n
+(exp( l 3  t ) / ( ( l 1 l 3 )  ( l 2 l 3 ) ) ) ) ;
an

/  For A c t i v i t y  /
bM

ax=l 1  x ;
ay=l 2  y ;
az=l 3  z ;
f p r i n t f ( fp , "%f n t %f n t %f n t %f n n" , t , x , y , z ) ;
La

f p r i n t f ( fp1 , "%f n t %f n t %f n t %f n n" , t , x2 , y2 , z2 ) ;


g
while ( t < =100.0);
g
LABORATORY MANUAL FOR NUCLEAR PHYSICS 288

Numerical integration of Bateman Equations


1
Parent
Daughter
0.9 Granddaughter

0.8

0.7

0.6

Number
0.5

0.4

0.3

s
0.2

sic
0.1

hy
0 10 20 30 40 50 60 70 80 90 100 110
Time(m)

rP
Figure D.4: Numerical Integration of Bateman Equations

lea
uc
Exact Solution of Bateman Equations
1
Parent
Daughter
0.9 Granddaughter
lN

0.8

0.7
ua

0.6
Number

0.5
an

0.4

0.3
bM

0.2

0.1

0
0 10 20 30 40 50 60 70 80 90 100 110
Time(m)
La

Figure D.5: Exact Solution of Bateman Equations

Although it is easy to obtain the exact solutions to the rate equations in the case of a few nuclides, it
can get very tedious in the case of a larger number of nuclides in a decay chain. Numerical solutions
provide an easy method of solving for the number evolution of nuclides in any chain.
Appendix E

Theory of Alpha Decay

s
sic
hy
E.1 Introduction

rP
As we have seen in Chapter 2, Section 2.2.1, classically it is impossible to explain how an alpha particle
can escape from the radioactive nucleus. However, quantum mechanically, this is easily explained. We
give below a simpli ed derivation of the relationship between the decay constant  and the energy of
lea
the outgoing alpha particle. This is adapted from \Concepts of Modern Physics", by Beiser,
Mahajan & Choudhury.
uc
lN

The basic assumptions of the Theory of Alpha Decay are as follows:


 An alpha particle exists as an entity within the nucleus.
ua

 The particle is in constant motion but is held inside the nucleus because of a potential barrier
an

 There is a small but de nite probability that the particle can tunnel through the barrier every
bM

time it collides with the barrier.


With these assumptions, notice rst that the decay probability per unit time,  can be expressed as
La

 = T (E.1)
where  is the number of times the alpha particle strikes the potential barrier per unit time and T is
the probability that the particle will transmit through the barrier. We further assume that at any one
moment, one and only one alpha particle exists inside the nucleus and this moves back and forth inside
the nucleus, along the diameter. Then we can easily see that the collision frequency,  is
v
= (E.2)
2 R0
where v is the velocity of the alpha particle and R0 is the nuclear radius. If we put in typical values of
these quantities, v  2  107 m s 1 and R0  10 14 m, we get

289
LABORATORY MANUAL FOR NUCLEAR PHYSICS 290

  1021 s 1
This is remarkable since this implies that even though the alpha particle strikes the nuclear barrier
1021 times per second, it still has to wait a long time to come out (depending on the half life of the
nuclide in question).

E.2 1-d Tunnel E ect For Rectangular Barrier

s
sic
To understand alpha particle decay, we need to revisit Tunnel E ect in Quantum Mechanics. Recall
that in 1-d quantum mechanics, if we have a nite potential barrier, a particle with a lower energy

hy
than the height of the barrier can still have a nite probability of tunnelling through the barrier. This
is the basis, for instance of the semiconductor tunnel diode.

rP
Consider a barrier of height U and width L as in Figure E.1. A particle with energy E < U is incident
lea
on the barrier from the left, that is from Region I.
uc
lN
ua
an
bM
La

Figure E.1: Tunneling in 1-dimension

Now let us write down the Schrodinger equation for the particle in Region I and Region III where there
is no potential and so the particle is a free particle.
d2 I 2m
+ 2E =0 (E.3)
dx2 ~ I
LABORATORY MANUAL FOR NUCLEAR PHYSICS 291

and
d2 III 2m
+ 2 E III = 0 (E.4)
dx2 ~
The solutions are as expected, free particle, plane wave solutions of the form

I = Aeik1 x + Be ik1 x (E.5)


and

III = F eik1 x + Ge ik1 x (E.6)

s
sic
where
p
2mE

hy
k1 =
~

rP
Clearly, the two terms in I represent the right moving and left moving waves, or I + and I , where
the right moving wave is the incident beam of particles and the left moving is the re ected part of the
beam. That is lea
= I+ + I = Aeik1 x + Be ik1 x
uc
I

The incident ux is then simply


lN

SI = j I + j2 vI +
ua

where vI + is the velocity of the incident particles. Similarly, since in Region III, there can only be
transmitted particles moving right, the wavefunction in that region is simply
an
bM

III = F eik1 x
The transmission probability is what we are interested in. This is simply the transmitted ux divided
by the incident ux.
La

T=
j III + j2vIII + (E.7)
j I+ j2vI +
or

R T=
F F  vIII +
AA vI +
(E.8)

As always, the values of the constants A and F are to be determined by the boundary conditions.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 292

What about the Region II? Here, classically, there cannot be any particle since E < U . However,
quantum mechanically, there is a de nite probability of nding the particle. The Schrodinger equation
for this region is simply
d2 II 2m
+ 2 (E U) =0 (E.9)
dx2 ~ III

which has solutions

II = Ce k2 x + Dek2 x (E.10)

s
with

sic
p
2m(U E)
k2 =

hy
~
Now we are ready to apply the boundary conditions. Recall that the wavefunction and its derivative

rP
have to be continuous everywhere, and in particular at the boundaries, x = 0 and x = L. This gives us

A+B
lea = C +D
uc
ik1 A ik1 B = k2 C + k2 D
Ce k2 L + Dek2 L = F eik1 L
lN

k2 Ce k2 L + k2 Dek2 L = ik1 F eik1 L (E.11)


ua

This system of equations allows us to solve for A and F which we need for the transmission probability
in Eq E.8. Assuming that U  E and the barrier is wide enough that is k2 L  1, we get
an

 
AA 1 k22 2k2 L
= + e
bM

FF 4 16k12
Further, since vIII + = vI + (because the wave numbers are the same in both regions and thus the group
velocity of the de-Broglie waves is the same), we have
La

" #
16
T= k2 2 e 2k2 L (E.12)
4 + ( k1 )
where
p
2m(U E)
k2 =
~
Now consider the quantity kk21 . Substituting their values, we see that
k2 U E
=
k1 E
LABORATORY MANUAL FOR NUCLEAR PHYSICS 293

Thus this quantity is slowly varying compared to the exponential. Also, the quantity in front of the
exponential in Eq E.12 is of order 1. Thus to a good approximation, we can write

R T = e 2k2 L (E.13)

E.3 Tunnel E ect with Nuclear Potential Barrier: Geiger-Nuttall Law


All this analysis is standard for a rectangular barrier in quantum mechanics. However, for the alpha

s
particle, it does not face a rectangular barrier. Instead, the barrier is like in Figure E.2 as we have seen

sic
in Section 2.2.1, Figure 2.1.

hy
rP
lea
uc
lN
ua
an
bM

Figure E.2: Energy diagram for Alpha decay

Ze2 is the electric potential energy of the alpha particle of charge 2e at a distance
In the gure, U = 420r 2
r from the nucleus of charge Ze. R = 42Ze0 E is the distance where E , the energy of the alpha particle
La

is equal to the potential energy U . R0 is the radius of the nucleus. Z is the atomic number of the
daughter nuclei.

To solve for the transmission probability in this case is somewhat complicated. However, we will do a
simple analysis to get an expression for the transmission probability which will allow us to determine 
as a function of energy as in Eq. E.1. We rst write Eq E.13 as

ln T = 2k2 L (E.14)
or
LABORATORY MANUAL FOR NUCLEAR PHYSICS 294

ZL ZR
ln T = 2 k2 (r)dr = 2 k2 (r)dr (E.15)
0 R0
Thus,
p 1=2
2m 1=2 2Ze2

 
2m(U E)
k2 = = 2 E
~ ~ 40 r
But at r = R, U = E and thus

s
2Ze2

sic
E=
40 R
and therefore

hy
2mE 1=2 R
 
 1=2

rP
k2 = 1 (E.16)
~2 r
With this k2 , we can now evaluate the integral in Eq E.15. lea
uc
ZR
ln T = 2 k2 (r)dr
lN

R0
 ZR 
2mE 1=2
 1=2
R
= 2 1 dr
ua

~ 2 r
R0
1=2 "  #
R0 1=2 R0 1=2 R0 1=2
     
an

2mE
= 2 R arccos 1 (E.17)
~2 R R R
bM

But we have assumed that the barrier is suciently wide and hence R  R0 and therefore

R0 1=2  R0 1=2

  
La

arccos
R
2 R
and

R0 1=2
 
1
R
1
We nally get
"  #
2mE 1=2  R0 1=2
  
ln T = 2 R 2 (E.18)
~2 2 R
Substituting the value of R in this expression, we get
LABORATORY MANUAL FOR NUCLEAR PHYSICS 295

4e m 1=2 1=2 1=2 e2 h m i1=2


 
ln T = Z R0 ZE 1=2 (E.19)
~ 0 ~0 2
Putting in the numbers for the various constants, we get

R log T = 1:29Z 1=2 R01=2 1:72ZE 1=2 (E.20)

where the nuclear radius R0 is in fermis and the energy E of the alpha particle is in MeV. Finally, we
know that the decay constant is related to the transmission probability Eq E.1 as

s
sic
v
 = T = T
2 R0

hy
Substituting, we get

rP
R log  = log

v
2R 0

+ 1:29Z 1=2 R01=2
lea 1:72 p
Z
E
(E.21)

This is the famous Geiger-Nuttall Law relating the decay constant  with the alpha particle energy
uc
E . This law is usually written in the form
lN

Z
ln  = a1 p + a2
E
ua

where a1 and a2 are two constants depending on the nuclei in question. This law has been con rmed
experimentally as can be seen in Figure E.3. In this gure, the log of the half life of various alpha
an

emitters is plotted against E 1=2 . We see that as expected, it is a straight line.


bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 296

s
sic
hy
rP
Figure E.3: Geiger-Nuttall Law
x(Source: http://www.open.edu/openlearn/science-maths-technology/science/physics-and-astronomy/scattering-and-
tunnelling/content-section-5.2 ) lea
uc
lN
ua
an
bM
La
Appendix F

Fermi's Theory of Beta Decay

s
sic
hy
The properties of Beta decay can be understood by using quantum mechanics. In particular, one uses
Fermi's Golden Rule to nd the transition rate between the initial and nal states of the nucleus which

rP
are involved in Beta decay. Thus, we need to rst see how we can obtain Fermi's Golden Rule from
elementary quantum mechanics.
lea
uc
F.1 Fermi's Golden Rule
lN

Consider a system with Hamiltonian H0 . We assume that the the eigenvalues Ei and the eigenfunctions
ui (x) can be determined. That is
ua
an

H0 ui (x) = Ei ui (x) (F.1)


bM

Given this, we know that any state of the system will evolve as

X
(x; t) = ci (0)e i!i t u
i (x) (F.2)
La

Of course if the state is an eigenfunction of the Hamiltonian, it will be stationary, that is have no time
dependence. In our case, we take the initial state of the system, that is the state at t = 0 to be an
eigenfunction.

(x; 0) = ui (x)
Next we consider a time independent perturbation being applied to the system. Thus, if our initial state
was the ground state of an atom, we could think of a momentary pulse of laser light on the system. Of
course, the perturbation in general can be time dependent but we will consider only time independent

297
LABORATORY MANUAL FOR NUCLEAR PHYSICS 298

perturbations here. The perturbation can be written as a potential term V^ and the total Hamiltonian
of the system thus becomes

H = H0 + V^ (F.3)
To solve for the state of the system, we would need to nd the eigenvalues and eigenfunctions of this
composite Hamiltonian. This is not always easy to do. Thus we adopt a di erent strategy. We assume
that the state of the system with this new Hamiltonian can be expressed in terms of the
eigenfunctions ui (x) of the original, unperturbed Hamiltonian H0 . That is we guess a solution
of the form

s
sic
0 (x; t) = X c (t)e i!i t u (x)

hy
i i (F.4)
i

rP
Notice that Eq F.4 is almost the same as Eq F.2 except for the fact that now the coecients are time
dependent. This is because of the perturbation that we have introduced.
lea
The task then is to nd the time dependent coecients ci (t). We use the Schrodinger equation
uc
@ 0
= (H0 + V^ ) 0
i~ (F.5)
@t
lN

and substitute Eq F.4 in Eq F.5. Using the fact that the ui (x) are eigenfunctions of H0 and that
Ei = ~!i , we get
ua

X X
i~c_i e i!i t u (x) ci (t)e i!i t V^ [u
= i (x)] (F.6)
an

i
i i
Notice that the LHS of the above expression has the time derivative of the coecients ci and the RHS
bM

has the perturbing potential V^ . We now take the inner product of both sides with uj (x) and use the
orthogonality of the eigenfunctions to get
La

+1
Z +1
Z
X X
i~c_i e i!i t uj (x)ui (x)dx = ci e i!i t uj (x)V^ [ui (x)]dx (F.7)
i 1 i 1
But +1
Z
uj (x)ui (x)dx = ij
1
and therefore
LABORATORY MANUAL FOR NUCLEAR PHYSICS 299

+1
Z
X
i~c_i e i!i t uj (x)V^ ui (x)dx = i~c_j e i!j t (F.8)
i 1
We de ne
+1
Z
V^ij = uj (x)V^ [ui (x)]dx
1
Eq F.7 becomes

s
sic
iX
c_j = ck (t)ei(!j !k )t V^
jk (F.9)
~ k

hy
or

rP
Zt
iX !k )t0 V^ dt0 + c (0)
cj (t) = ck (t0 )ei(!j jk j (F.10)
~ k
0 lea
This is now the formal solution to the problem. Once we have the coecients cj (t), we can substitute
uc
them in Eq F.4 and nd the time evolution of the state of the system in the presence of the perturbing
potential. However, to actually solve Eq F.10 is not easy. Therefore we make a crucial approx-
lN

imation. We assume that the perturbing potential is such that its e ect on the system
is slow. That is the coecient ck (t) on the RHS of Eq F.10 doesnt change with time and
ua

therefore can be replaced with its initial value at t = 0. This then allows us to rewrite Eq F.10
as
an

Zt
iX !k )t0 V^ dt0 + c (0)
ei(!j
bM

cj (t) = ck (0) jk j (F.11)


~ k
0
But our initial assumption was that the unperturbed state of the system initially, that is the state at
La

t = 0 is an eigenfunction of the original unperturbed Hamiltonian, that is (x; 0) = ui (x). Therefore


the coecients ck (0) take a xed value, that is ck (0) = 0 for all k 6= i. We are now left with

Zt
i !i )t0 V^ dt0
cj (t) = ei(!j ji (F.12)
~
0
Zt
i 0
= ei!j t V^ji dt0
~
0
where
LABORATORY MANUAL FOR NUCLEAR PHYSICS 300

!j = !j !i
This integral can be solved easily and we get
1 ^  
cj (t) = Vji 1 ei!j t (F.13)
~!j
Once we know the coecients, we can nd the quantity of interest, that is the probability of transition
of the system from an initial state (x) = ui (x) to the nal state which we assume to be also an
eigenfunction of the original Hamiltonian H0 . For this, we simply need to take the modulus squared of

s
the coecients.

sic
P (i ! j ) = jcj (t)j2 (F.14)

hy
The probability of transition is then given by

rP
 
4jV^ji j2 2 !j t
P (i ! j ) = 2 2 sin (F.15)
~ !j lea 2
We can rewrite this in another way for reasons which will become apparent soon.
uc
0 1
jV^ j2 sin2 !j t
P (i ! j ) = ji2 @ !2 2 A (F.16)
lN

~ j
22
Now we are interested in the behaviour of our system at large times, that is t ! 1. In this limit, the
ua

function sinx x becomes very narrow until we can approximate it with a delta function. The exact limit
can be taken and we get
an

2jV^ji j2 t
bM

P (i ! j ) =  (!j ) (F.17)
~2
This allows us to nd the transition rate which is simply the probability of transition per unit time as
La

R Wij =
2jV^ji j2
~2
 (!j ) (F.18)

This is Fermi's Golden Rule which provides a reasonably accurate way of nding the transition rates.

This form of the Golden Rule is obviously valid if the transition is from an initial state of de nite energy
to another one of de nite energy. In the cases of interest to us, namely beta and gamma decay, the
nal particles (electron or photon) are free and therefore have a continuous range of energies. In that
case we need to modify the Golden Rule as state above to take into account this. To do this, recall that
in the derivation above, we used a delta function for the di erence between the initial and nal energy,
LABORATORY MANUAL FOR NUCLEAR PHYSICS 301

that is  (!j ) where !j = !j !i . If the nal state does not have a de nite energy, then we need
to consider transitions not to a de nite nal state but instead to states in a narrow interval centered
on Ej , that is Ej  dE . Obviously the transition rate will be proportional to the number of states to
be found in this energy interval. This is related to the density of nal states, (E ) by dN = (E )dE .
Thus we replace the delta function by the density of states centered around Ej = Ei . In doing this,
remember that

 (!j ) = ~ (~!j ) = ~ (Ej Ei )


by the properties of delta function. Thus the nal result has only one ~ in the denominator. We can

s
sic
nally write the Golden Rule as

hy
2jV^ji j2
Wij = (Ej ) (F.19)

rP
~
where it is understood that the density of states needs to be evaluated for Ej = Ei .
lea
uc
F.2 Fermi's Theory of Beta Decay
lN

The accurate description and theory explaining beta decay of course would mean that we use quantum
eld theory to analyse the process. In this, the particles in question, namely the electron and the
ua

anti-neutrino are described by eld operators. The interaction, as we have already seen which is
responsible for Beta Decay is the weak interaction. We will not attempt to do a full quantum eld
an

theoretic analysis of the process. Instead we will try to see if we can use Fermi's Golden Rule to get
some results.
bM

From Eq F.19, we see that the transition rate will be


La

R W=
2 j < i jV^ j f > j2 (E )
~ f (F.20)

where V^ is the interaction Hamiltonian relevant to the process. Thus we need two things to use Fermi's
Golden Rule: the matrix element, < i jV^ j f > and the density of nal states (Ef ). Let us look at
the Matrix element rst.
LABORATORY MANUAL FOR NUCLEAR PHYSICS 302

F.2.1 Matrix Element


We write the interaction term in terms of eld operators which create and destroy particles. In this
case, the initial state is a neutron while the nal state has a proton, electron and anti-neutrino. So we
take the interaction term to be

V = g ey y (F.21)
where g is the strength of the interaction and the two operators ey and y create an electron and
antineutrino.With this our matrix element can be written as (where for the moment, we are not explicitly

s
writing the nuclear initial and nal states)

sic
Z
Vif = g d3 x p e  (F.22)

hy
n

where we have replaced the dagger with a star since we consider our operators to be scalars. Now we

rP
know that the process results in a free electron and a free anti-neutrino. Thus as an approximation we
can take the operators to be plane waves. With this approximation, we get

Vif = g
Z
3
lea
ei~ke ~x ei~k ~x
dx p p
 (F.23)
n
p
V V
uc
p
Please remember that the V in the denominator above refers to the volume that we need to
lN

normalise the plane wave states and not the potential.


ua

Now we can make a further approximation. Typical kinetic energies of the electron are in the MeV range.
Thus if we take the kinetic energy of the electron to be 1 MeV, we can easily nd the corresponding k.
an

To do this, note that the kinetic energy of the electron, Te is given by


bM

p
Te = p2e c2 + m2e c4 me c 2
Solving for pe , with Te = 1 MeV and me = 0:51 MeV c 2 , we get
La

pe  1:4 MeV c 1
Thus by de-Broglie relation and the de nition of the wave number, we get
pe
ke =
h
 0:007 fermi 1
We can thus easily see that for nuclear dimensions,

ke r  1
With this, we see that
LABORATORY MANUAL FOR NUCLEAR PHYSICS 303

ke2 r2
eike r = 1 + ike r +   1
2
We thus get for the matrix element
Z
g g
Vif = d3 x p n = Mfi (F.24)
V V
where we have introduced a function Mfi which is some complicated function which takes into account
all the details of the nuclear states. We also know that there would be some electromagnetic interaction
between the electron and the proton which we have neglected above. So to take care of that, we replace

s
jMfij2 by jMfij2F (Z; Q) where the function F (Z; Q), called the Fermi function, incorporates the

sic
electromagnetic interactions. This function is tabulated extensively for various values of Q and Z . We
thus have the expression

hy
R
rP
2 g
W= jM j2F (Z; Q)(Ef )
~ V fi
(F.25)

F.2.2 Density of Final States


lea
uc
The two particles that we see in beta decay are the electron and the anti-neutrino. The density of
nal states should therefore re ect this as also the fact that both of them are free particles, that is
lN

they exist in a continuum of possible states. To nd the density of nal sates, we need to know the
number of states accessible to the electron and the anti-neutrino. Suppose the electron is emitted with
ua

a momentum pe and the anti-neutrino with q. Since we need to know only the shape of the energy
spectrum, the directions of these momenta are not relevant to us. Consider a coordinate system with
an

p
axes along pex ; pey ; pez . Then for a speci c value of the electron momentum p = p2ex + p2ey + p2ez , the
locus of points is a sphere of this radius. Thus the locus of all the points with momentum between p
bM

and p + dp is a shell with inner radius p and outer radius p + dp with a volume 4p2 dp.

Recall that phase space is an imaginary space consisting of 3 space dimensions (x; y; z ) and 3 momentum
La

coordinates (px ; py ; pz ). To specify the particle, say the electron, we need to specify its position and
momentum. Of course, uncertainty principle tells us that

xpx  h
Therefore
xyz px py pz  h3
This is basically saying that when I say an electron is at a point in phase space, it is basically in a
unit cell of phase space of \volume" h3 . Now if we say that an electron has momentum between pe and
pe + dpe , then we know that the uncertainty in the momentum of the electron will be
LABORATORY MANUAL FOR NUCLEAR PHYSICS 304

px py pz = 4p2e dpe


and if we also require that the electron is con ned to a volume V , then the corresponding volume in
phase space where the electron could be found will be

4p2e dpe V
Now the question is really how many ways can an electron exist in this volume of phase space. The
answer is simply the number of unit cells in this volume. Recall that each unit cell has a volume h3 and

s
therefore we get

sic
4p2e dpe V
dNe =

hy
h3
Similar arguments lead us to the expression for the number of neutrinos.

rP
4p2 dp V
dN =
h3
lea
We can rewrite these expressions in a more useful form as
 
uc
4V
dNe = 3 p2e dpe (F.26)
(2~)
lN

Similarly for the neutrino we get


 
ua

4V
dN = 3 p2 dp (F.27)
(2~)
an

Thus we write the number of states in a small energy volume as


bM

dN = dNe dN (F.28)


where we have for simplicity use the symbol for neutrino instead of anti-neutrino. We also know that
La

the total kinetic energy available in the reaction, that is the Q value is shared between the electron and
neutrino.

Q = Te + T
We will take the neutrino to be massless and therefore

T = p c
The electron kinetic energy is simply
p
Te = E me c2 = p2e c2 + m2e c4 me c2 (F.29)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 305

Now
T Q Te
p = =
c c
Therefore for a xed value of Te , we have
dQ
dp =
c
The density of nal states for a xed electron energy is thus given by

s
dN

sic
(pe )dpe = dNe
dT
16 V 2 2
2 dp
pe dpe p2 

hy
= 6 (F.30)
(2~) dT

rP
Now from the relationship between the kinetic energy and momentum for the neutrino, we know that

lea
T = p c
and
( Q Te )
uc
p =
c
lN

Therefore we get
V2 h p i2
(pe )dpe = 4 6 3 Q ( p2e c2 + m2e c4 mc ) p2e dpe
2
ua

(F.31)
4 ~ c
But
an

p
p2e c2 + m2e c4 mc2
bM

Te =
Therefore
Te + me c2
La

pe dpe = dTe
c2
Substituting in Eq F.31, we get

R (pe )dpe =
V2 
4  4 ~6 c 6
Q T 2 2 (pT 2 + 2me c2 Te )(Te + me c2 )
e e (F.32)

F.2.3 Decay Rate


We are now nally ready to nd the decay rate using Fermi's Golden Rule. We have
LABORATORY MANUAL FOR NUCLEAR PHYSICS 306

2
W= jVif j2 (E )
~
Notice that as far as the shape of the nal energy spectrum is concerned, all the factors in the decay
rate that are independent of the momentum can be collected into some constant, say C . This will
include the nuclear matrix element Mfi etc since we assume them to be independent of the electron
momentum. Then the number of electrons between momentum p and p + dp will be

Npe dpe  Cp2e p2 dpe

s
But we know that

sic
( Q Te )
p =

hy
c
Substituting, we get

rP
R Npe dpe  Cp2e (Q Te )2
lea (F.33)

and
uc

R
lN

 p 2
Npe dpe  Cp2e Q p2e c2 + m2e c4 mc2 (F.34)
ua

These functions can be plotted for any value of Q. We see that for pe = 0 and for Te = Q, the function
vanishes. A plot of for the shape for Q = 2:5 MeV is given in Fig F.1.
an
bM
La
LABORATORY MANUAL FOR NUCLEAR PHYSICS 307

Beta Spectrum
5
Beta Spectrum

4.5

3.5

3
N(T)

2.5

s
2

sic
1.5

hy
1

0.5

rP
0
0 0.5 1 1.5 2 2.5
T(MeV)

lea
Figure F.1: Beta spectrum plot for Q = 2:5 MeV
uc
The complete beta spectrum thus contains three contributions:
lN

1. The statistical factor coming from the density of nal states. This as we have seen is proportional
to p2e [Q Te ]2 .
ua

2. The contribution coming from the nuclear Coulomb eld, that is the electromagnetic interaction
an

of the nucleus with the beta particle. This as we saw above is proportional to the Fermi function
bM

F (Z; Q).
3. The contribution coming from the nuclear matrix element, Mfi which includes the e ects of the
particular initial and nal nuclear states of the particular decaying nucleus and could also contain
La

some momentum dependent factors.


Thus we can write the nal beta spectrum as

Npe / p2e (Q Te )2 F (Z; Q)jMfi j2 (F.35)


We can rewrite this as

R
s
Npe
(Q Te ) / (F.36)
p2e F (Z; Q)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 308

Note that if we plot the RHS of Eq F.36 against the electron energy, we get a straight line and the value
of the x intercept will give us the Q value or the end-point energy of the beta particles. This is the
famous Fermi-Kurie plot shown in Figure F.2. The Fermi-Kurie plot is a good way to test Fermi's
theory of Beta Decay. It also provides a very accurate way to determine the end-point energy.

s
sic
hy
rP
Figure F.2: Fermi-Kurie plot
x(http://physics-database.group.shef.ac.uk/phy303/phy303-4.html )
lea
uc
lN
ua
an
bM
La
Appendix G

Semi-Classical Theory of Gamma Decay

s
sic
hy
G.1 Introduction

rP
We have already seen in Section 2.2.3 that gamma radiation comes when a nucleus de-excites from an
excited state to a lower energy state. We have also seen that the typical energies of the gamma rays is
of order MeV. To understand the gamma emission, we can use a semi-classical approximation alongwith
lea
Fermi's Golden Rule as described in Section F.1. The transition rate from Fermi's Golden Rule is given
by Eq F.19 as
uc
2jV^ji j2
lN

Wij = (Ej ) (G.1)


~
where
ua

Vji =< j jV^ ji >


an

that is the matrix element of the interaction potential V^ with the initial and nal states and (Ej ) is
the density of nal states as discussed in Section F.2.2. Thus we see that if we need to apply this to
bM

gamma decay, we need two ingredients, the density of nal states and the matrix element.

G.2 Density of Final States


La

We have already seen in the case of beta decay that the number of electrons in a volume of phase space
is given by Eq F.26. Using similar arguments, we consider the nucleus and the gamma radiation as
existing in a box of volume V (or L3 ), we can easily nd that the number of nal states with the gamma
ray photon momentum between k and k + dk will be
V
dNs = 4k2 dk (G.2)
(2)3
If we consider a solid angle d
instead of the whole sphere, we get

309
LABORATORY MANUAL FOR NUCLEAR PHYSICS 310

V
dNs = k2 dkd
(G.3)
(2)3
But we know that for photons, p = ~k and E = pc = ~kc = ~!. Substituting we get

R (E ) =
dNs !2 V
=
dE ~c3 (2)3
d
(G.4)

G.3 Interaction Hamiltonian

s
sic
Since we are dealing with gamma rays, that is electromagnetic radiation, we need to consider the
interaction of a particle with an electromagnetic eld. This is expressed in terms of the vector potential

hy
A~ associated with the electromagnetic eld as

rP
e
V^ = A~^  p~^ (G.5)
mc
where the particle charge is e. Here A~^ is the quantum mechanical operator associated with the vector
lea
potential A~ and p~^ is the momentum operator for the particle. Now we consider the electromagnetic
uc
eld as a quantum eld and therefore expand it in terms of creation and destruction operators, a^k and
a^yk . These can be thought of as operators which create and annihilate photons of momentum k.
lN

s
X 2~c2  i~k~r y i~k~r 
A~^ = a^ e + a^k e ~k (G.6)
ua

k
V !k k
where ~k is the polarisation of the electromagnetic wave.
an
bM

G.3.1 Dipole Approximation


With the interaction potential now in this form, we can evaluate the transition rate using Fermi's Golden
Rule now. We have
La

2jV^ji j2
Wij = (Ej )
~
In our case, we note that in gamma decay, our initial state has no photon (there is only the excited
nucleus) and the nal state has one photon with an energy E = ~! = ~kc. If we substitute the
expression for A~^ from Eq G.6 in the expression for V^ in Eq G.5, we see that only one term will be
non-zero in the evaluation of the matrix element. This will be the term with one a^yk since this is the
only one which can connect an initial state of no photon with a nal state of one photon. Thus
s
e 2~c2 D ^ i~k~r E
Vji = ~  p~e (G.7)
mc V !k k
LABORATORY MANUAL FOR NUCLEAR PHYSICS 311

The expectation value of the momentum operator p~^ is obviously between the initial and nal states.
Now we can try to simply this. Recall that

[p~^2 ; ~r^] = 2i~p~^


Therefore
i ^ im p~^2 ^
p~^ = [p~^2 ;~r ] = [ ;~r ]
2~ ~ 2m
But we know that the nuclear Hamiltonian Hn is

s
p~^2 ^ ^

sic
Hn = + Vn (~r)
2m
where V^n (~r^) is the nuclear potential term. Since this only a function of r, it will commute with~r
^ and

hy
so we see that

rP
im p~^2 im
p~^ = [ + V^n (~r^); ~r^] = [Hn ; ~r^]
~ 2m ~
lea
This is an enormous simpli cation since we know that the initial and nal states in the matrix element
are eigenstates of the nuclear Hamiltonian. Thus
uc
im im h i
< j jp~^ji >= < j j[Hn ; ~r^]ji >= < j jHn~r^ji > < j j~r^Hn ji > (G.8)
lN

~ ~
But since the states are eigenstates of Hn , we can simplify this to
ua

j^ ji >= im (Ej
< j~p
~
Ei ) < j j~r^ji >= im!k < j j~r^ji > (G.9)
an

where ~!k = Ej Ei .
bM

Thus
s
e 2  ~c 2 D E
im!k~k  ~re
La

^ i~k~r
Vji = (G.10)
mc V !k
^ ~k~r . To simplify this further, let us consider a
We still need to evaluate the expectation value of ~re
typical gamma ray of energy 1 MeV. The wavelength of this radiation is  10 13 m. Nuclear size we
know is of the order of 1 fermi or  10 15 m. Thus the quantity kr  1. This allows us to approximate
the exponential by its rst term. In this approximation,which is called the dipole approximation,
we thus have

R Vji =
r
2~e2 !k D^ E
V
~k  ~r (G.11)
LABORATORY MANUAL FOR NUCLEAR PHYSICS 312

The reason for calling this the dipole approximation is obvious now since the dipole operator e~r^ is
what is relevant as can be seen from Eq G.11.

G.4 Transition Rate & Lifetime


The transition rate can now be calculated since we have the density of nal states (Eq G.4) and the
matrix element (EqG.11). Then the transition rate is given by

2jV^ji j2 !3 D ^E 2

s
W  (E 1) = (Ej ) = j ~  ~r j d
(G.12)

sic
~ 2c3 ~ k
Here E 1 signi es that this is only the dipole term.

hy
But we know that the angle between the polarization vector and the dipole moment vector is =2 .
To see this, consider a dipole aligned along the z axis. Now the position vector of a point (r; ; ), will

rP
have an angle of  with the z axis. But the E~ eld direction will be perpendicular to the direction of
propagation and therefore the angle between it and the z axis is 2 . Please do not confuse between
lea
the position vector ~r and the dipole operator e~r^ though they look the same. The angle  is between
the two in our example. The direction of the E~ is precisely the direction of the polarization vector ~k .
uc
That is, the angle between ~k and ~r^ is also 2 . Therefore,
lN

D E D E
~k  ~r^ = ~r^ sin 
Then the transition rate is simply
ua

e2 !3 D ^E 2 2
an

(E 1) =
2c3 ~
j ~r j sin d
(G.13)
bM

To get the total transition rate, we need to integrate over all directions. Then since
Z2 Z
8
d sin2  sin d =
3
La

0 0
we get

R (E 1) =
4e2 !3 D ^E 2
3c3 ~
j ~r j (G.14)

This is the transition rate. We can get a reasonable estimate of this quantity. We know that ! = E~ .
We can also approximate the value of the expectation value of the position operator by the nuclear
radius, Rn  r0 A1=3 where r0  1:25 fermi. Putting it all together we get
LABORATORY MANUAL FOR NUCLEAR PHYSICS 313

R (E 1) 
4e2 E 3 2 2=3
rA
3~(~c)2 0
(G.15)

If we put in representative numbers for the quantities above, E = 1 MeV and A = 65, we get (E 1) 
1:5  1015 s 1 or its reciprocal the time to be   10 15 seconds.

s
sic
hy
rP
lea
uc
lN
ua
an
bM
La
Index
Bateman Equations, 281 Propagation of errors, 48
Interaction with matter expectation value, 21
Rayleigh Scattering, 110

s
Fermi's Golden Rule, 87
Student's t-distribution, 64

sic
Fermi's Theory of Beta Decay , 297
Absorption of Gamma Rays in Iron, 182 Fermi-Kurie plot, 308

hy
accuracy, 10 ux, 92

rP
barn, 93 Gamow Theory of Alpha Decay , 289
Beta Ray end point energy, 197 Geiger-Nuttall Law, 295
Binding Energy, 81 leaGM Counter, 123
Binomial Distribution, 25 Absolute Eciency, 142
mean, 26
uc
Counting Eciency, 140
probability, 26 Dead Time, 135
lN

variance, 27 Detector Regions, 130


Central Limit Theorem, 46 External Quenching, 134
ua

Chi Squared test, 68 ll gas, 133


classical electron radius, 99 gas multiplication, 127
an

Continuous Random Variable, 9 Geiger discharge, 132


Geiger Region, 131
bM

Covariance, 49
Internal Quenching, 134
degrees of freedom, 69 Intrinsic Eciency, 142
Detector Models, 124 Ion Chamber, 131
La

Discrete Random Variable, 9 mobility, 128


error analysis, 9 Operating Voltage, 138
Error Estimation, 46 Plateau, 138
Error propagation equation, 49 proportional counters, 131
error propagation for multiplication & division, Quench gas, 134
50 Quenching, 133
error propagation for Powers, 52 region of limited proportionality, 131
error propagation for sums and di erences, 50 Region of true proportionality, 131
error propagation for weighted sums and dif- starting voltage, 139
ferences, 50 time constant, 126

314
LABORATORY MANUAL FOR NUCLEAR PHYSICS 315

Townsend Avalanche, 128 Rayleigh Scattering, 113


Townsend Equation, 129 Straggling, 103
two source method, 136 Thomson Scattering, 114
work function, 134 internal conversion, 88
GM Counter Characteristics, 144 Interpolation, 275
preset time, 147 Inverse Square law for Gamma Rays, 188
GM Counting Eciency, 165
Law of Large Numbers, 7
Goodness of Fit, 68
limiting distribution, 17
half thickness, 183 Linear attenuation coecient, 182

s
sic
Health e ects linear regression, 64
Non-Stochastic health e ects, 146
Mass attenuation Coecient, 183

hy
rem, 146
mean, 19
sievert, 146

rP
Method of Least Squares, 64
Stochastic health e ects, 146
Method of Maximum Likelihood, 60
Health e ects of radiation, 145
histogram, 16 lea
neutrino, 84
Normal Distribution, 38
Interaction with matter
uc
Mean, 44
Bethe-Bloch for electrons, 104
Mean of standard normal, 40
lN

Bethe-Bloch formula, 99
probability density, 38
Bohr formula, 98
probability density function, 38
ua

Bragg curve, 101


standard form, 40
Bremsstrahlung, 104
Variance, 45
an

Cherenkov radiation, 104


Variance of standard normal, 40
Compton Scattering, 110
nuclear magneton, 80
bM

critical energy, 109


cross section, 92 p-value table, 248
electrons, 103 parent distribution, 19
La

gamma rays, 110 Poisson Distribution, 31


Heavy particles, 94 Mean, 34
Klein Nishina Formula, 118 Probability function, 33
mass thickness, 93 Variance, 34
mean range, 103 Precautions, 145
minimum ionizing particles, 100 precision, 10
Pair Production, 110, 120 probability, 7
Photoelectric Absorption, 110 probability distribution function, 18
radiative loss non-relativistic, 107
Radioactive Equilibrium , 279
radiative loss relativistic, 108
Radioactivity, 75
LABORATORY MANUAL FOR NUCLEAR PHYSICS 316

Activity, 78 Photopeak, 218


Alpha decay, 81 secondary emission, 215
Becquerel unit, 76 Energy Resolution, 224
Beta decay, 84 Semiclassical Gamma , 309
curie unit, 76 signi cant gures, 13
decay constant, 77 Square root rule, 47
Decay law, 78 standard deviation, 20
de nition of activity, 76 standard error, 63
electron capture, 87 systematic errors, 11

s
Fermi's Theory of Beta Decay, 85

sic
Using Gnuplot
Gamma Decay, 87
Customization, 269
Gamow theory of alpha decay, 83

hy
Interactive Plotting, 267
Mean Lifetime, 78
Plots using data from le, 271

rP
Nuclear Decay, 80
Plots with tting, 272
PET, 86
Saving Plots , 269
positron emission, 86 lea
Using Gnuplot , 266
Q-value, 81
Using MS Excel , 252
sieverts unit, 76
uc
speci c activity, 78 variance, 20
lN

Activity Law, 77
Half Life, 77
random errors, 11
ua

random variable, 9
an

reduced Chi squared, 69


relative uncertainty, 11
bM

rules of probability, 8
sample distribution, 19
Scintillation Counter, 207
La

activators, 210
Backscatter peak, 219
Compton Edge, 218
dynode, 215
Energy Calibration, 234
uorescence, 208
Inorganic Scintillators, 208
Organic Scintillators, 208
Phosphorescence, 208
Photomultiplier Tube, 214

You might also like