IIHHHWIHHW

—1_._.
101—:
0300\1

 

Ci

-- L F L‘\.‘.-,_u‘_

_ _ k
LIBRARY

 

Michigan State
University 6:; 3
M- 5!:

' This is to certify that the

thesis entitled
On Improving Ge Detector Energy Resolution
and Peak-to-Compton Ratios by
Pulse-Shape Discrimination
presented by

Nobuo Matsushita

has been accepted towards fulfillment
of the requirements for

M. S . dggree in Physical Chemistry

Almcm

Major professor \

 

 

Date Marma—

07639

 

 

 

 

II
‘I

V ovcnougzrrues: -
1‘25“”!- duper ite- ..
«manna L'I'amv MATERIALS:

' Place in book return to remove
charge fmu circulation records

 

ON IMPROVING GE DETECTOR ENERGY RESOLUTION AND

PEAK-TO-COMPTON RATIOS BY PULSE-SHAPE DISCRIMINATION

By

Nobuo Matsushita

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

MASTER OF SCIENCE

Department of Chemistry

Program in Physical Chemistry

1980

0

-‘\
.0
I

I

6 I/Aéf

ABSTRACT
0N IMPROVING GE DETECTOR ENERGY RESOLUTION AND
PEAK-TO-COMPTON RA'rIos BY PULSE-SHAPE DISCRIMINATION
By

Nobuo Matsushita

The rise-time discrimination of pulses from Ge detectors can be
used to improve the spectra on two levels: First, by discriminating
against slower-rising pulses, both the energy resolution and peakpto-
Compton ratios can be improved significantly, especially for detectors
that have suffered neutron damage. Second, by adding a pulse—height
correction to compensate for effects of varying rise-time, an improved
composite spectrum can be obtained without significant loss in detec-

tor efficiency.

ACKNOWLEDGMENTS

I wish to thank Dr. Wm. C. Menarris for his encouragement and
guidance during the preparation of this thesis. I would also like
to thank Dr. R. B. Firestone for his valuable advice and helpful
time-consuming discussions.

I also wish to extend special thanks to Drs. J. Kasagi and
J. A. Nolan, Jr., for their invaluable assistance and encouragement
during the data acquisition and analysis of this work. Many of their
ideas led to solutions to apparently insurmountable problems.

Thanks to Mr. R. Au, Mr. W. Bentley, Mr. R. Fox and
Mr. B. Jeltema for their programing help.

Ms. Rose Manlove aided greatly by typing the final version of
this manuscript.

Much of the financial assistance for this research has been
provided by the National Science Foundation and Michigan State

University.

ii

TABLE OF CONTENTS

Page No.

LIST OF FIGURES. . . . . . . . . . . . . . . . . . . . . . . . iv—v
Chapter

I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . 1

II. TWO DIMENSIONAL EXPERIMENT . . . . . . . . . . . . . . . . 4

1. Experimental Techniques. . . . . . . . . . . . . . . . 4

2. Results and Discussion . . . . . . . . . . . . . . . . 5

III. THREE DIMENSIONAL EXPERIMENT . . . . . . . . . . . . . . . 17

1. Experimental Techniques

and Data Processing. . . . . . . . . . . . . . . . . . 20

2. Results and Discussion . . . . . . . . . . . . . . . . 22

IV. CONCLUSIONS. . . . . . . . . . . . . . . . . . . . . . . . 31

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . 33
APPENDICES

A. Event Rec. . . . . . . . . . . . . . . . . . . . . . . 36

B. Auto Cen . . . . . . . . . . . . . . . . . . . . . . . 38

c.2DGATE........................40

iii

LIST OF FIGURES

Figure Page No.

2-1 Block diagram of the fast-slow
coincidence circuit used . . . . . . . . . . . . . . . . 6

2-2 Rise-time discriminated spectra with

Ge planar detector . . . . . . . . . . . . . . . . . . . 7

2-3 Output of the TAC for the
spectra shown in fig. 2-2. . . . . . . . . . . . . . . . 8

2-4 Summary of the analysis. . . . . . . . . . . . . . . . . 10

2-5 Plots of the centroid of four energies

as a function of the pulse rise-time . . . . . . . . . . 12

2-6 The correlation between the rise-time
and the centroid of the 1332.5-keV peak. . . . . . . . . 14

2-7 Rise-time discriminated spectrum (top)
compared with a raw spectrum (bottom). . . . . . . . . . 15

2-8 Blow up of a portion of fig. 2-7 . . . . . . . . . . . . 16
3-1 A Coaxial Detector Structure . . . . . . . . . . . . . . 18
3-2 An expected correlation of the time

difference between the 0.1-0.5 fraction
and the 0.1-0.7 fraction 0 o o o o o o o o o o o o o o o 19

iv

Figure

3-4

3-5

3-6

3-10

The block diagram for the three

parameter experiment . . . . . . . . . . . .

The correlation between the centroid of
the 1332.5-keV peak.and the two TAC signals.

90° clockwise rotation of fig. 3-4 . . . . .

The relative yield for each 1332.5-keV
peak.vs the two TAC signals. . . . . . . . .

90° clockwise rotation of fig. 3-6 . . . . .

The two dimensional plots displaying
the gated regions for 2D GATE. . . . . . . .

Spectrum with the 2D GATE (bottom)
compared with the raw data (top) . . . . . .

Blow up of a portion of fig. 3-9 . . . . . .

No.

21

23

24

25

26

28

29

3O

CHAPTER I

INTRODUCTION

Until semiconductor detectors were introduced in nuclear science,
scintillation counters were used in nuclear spectroscopy. The use of
the scintillation counter in its various forms has made possible the
investigation of a large number of problems that would otherwise have
been extremely tedious if not impossible. As an example we may con-
sider the field of gamma (Y)-ray spectroscOpy, whose present state of
development is due in large part to the use of scintillators. Prior
to about 1949, y-ray energies were generally measured either by ob-
serving the beta spectrum of associated Compton recoils or photo-
electron recoils or by interposing absorbers of various materials
between the Y source and detector (Geiger counter). The development
of scintillation counters, especially the NaI(T1) detector, have
provided a convenient, high-efficiency spectrometer such that mea-
surements of Y-ray spectra from radioactive nuclei have become a
fairly common laboratory technique. In the best scintillation systems
one photoelectron is produced at the photocathode for each 110 ev of
energy lost [C071].

In the early 1960's, semiconductors were introduced in nuclear
science. Since then, almost constant effort has been expended to
improve both their energy resolution and their peak-to-Compton ratio
[Ie71] to some extent, mutually incompatible goals. For example, the
larger the detector, the better the peak-to-Compton ratio, but in

1

general, the poorer its resolution. Such panaceas as very large

volume intrinsic Ge detectors [6074] or successful fabrication of detec-
tors from semiconductors having both a high Z and a small band gap
[Pe72], (i.e., InSb) have not yet appeared on the scene. Thus, we
often find a necessary compromise: a Ge Ybray detector with the heat
energy resolution practicable (and often of medium to small size) used
in conjunction with one of the more or less elaborate Compton-suppres-
sion spectrometers [Au67,Be77]. The latter are necessarily expensive,
cumbersome, and require elaborate coincidence electronics.

Additional concern arises when Ge detectors have been used for
in-beam ybray experiments where they often suffer from neutron damage.
The neutrons can cause lattice defects in the Ge crystal which act as
traps for the charge carriers [Go74,Kr68,Ma68]. These trapping centers
not only contribute to poor resolution because of incomplete charge
collection, but also slow down the rate of charge collection [Ma68].
Often in-beam ybray experiments extend over periods of days. In such
cases neutron damage may be occurring continuously and the resolution
will degenerate accordingly.

One way to repair these neutron damaged detectors is to regenerate
the Ge crystal. One can attempt to regenerate the crystal by the Badder
Method [Ba74]. This method is not always successful, so the detector
is often sent back to the company.

Our original purpose was to investigate the different rise-time
signals from Ge detectors, since this information might suggest some
ways of improving the energy resolution of the detectors. Also, at the

same time, we were interested in how the relative efficiency and the
t

peakrto-Compton ratios depended on the rise-times. In general,

the poor energy resolution is associated with pulses having slower
than normal rise-times. By discriminating against slower pulses one
should be able to improve the energy resOlution of a detector, albeit
at some expense in efficiency.

The improvement in the energy resolution has been borne out by
my experiments. I found that I can improve the energy resolution of
a detector by approximately 50%, the exact amount of depending on the
amount of neutron damage the detector has suffered. For a brand new,
state-of-the-art, high-resolution detector, the effect is not very
great; for a badly damaged detector, the improvement can exceed 50%.
Unexpectedly, however, I also found that I can use pulse-height
correction techniques to gain these improvements without any signi-

ficant loss in efficiency.

CHAPTER II

TWO DIMENSIONAL EXPERIMENT

1. Experimental Techniques

I have performed experiments using several Ge and Ge(Li)
detectors--from different manufacturers, of various sizes, and of both
planar and coaxial configurations. In this thesis, however, I present
results only for the two detectors studied most extensively: a Ge
planar detector manufactured by Princeton Gamma-Tech and a Ge(Li)
true coaxial detector manufactured by ORTEC.

The planar detector had an active area of 1000 mm?, giving it
an active volume of 13 cm3, or an efficiency of 2.52 relative to a
7.6 X 7.6_cm NaI(Tl) detector (for the 60Co 1332.5-keV peak, with a
source-to-detector distance of 25 cm). The operational bias was 2500 V,
and the original energy resolution (again, for the 60Co 1332.5-keV)
was 1.7 keV full width at half maximum (FWHM), although the resolution
had deteriorated slightly to 1.84 keV FWHM at the time these measure-
ments were performed.

The true coaxial detector was 44.4 mm long, 49.6 mm in diameter,
and with a drift depth of 19.3 mm, resulting in an active volume of
78.3 cm3, or an efficiency of 16%. Its Operational bias was 4800 V,
and its original warranteed resolution was 2.0 keV FWHM. This detec-
tor, however, had been used extensively for in-beam experiments and
had suffered extensive neutron damage, resulting in a poorer energy

resolution of 4.92 keV.

Standard 60Co, 13703, 152Eu sources were used for my measurements.

In order to perform our rise-time experiments, I used a fast-
slow'"megachannel" [Gi7l] circuit with branched signals from the single
detector under investigation for the coincidence inputs. A block
diagram of the experimental set-up is shown in fig. 2—1. The ORTEC
Model 473 Constant Fraction Timing Discriminators (CFTD), together
with associated Model 425 Delays, were used to start and st0p the
Model 467 Time-to-Amplitude Converter (TAC), whose output was recorded
on.magnetic tape along with the pulse-height information.

Although the rise-time distributions vary from detector to detec-
tor, 40 nsec on the external delays for the CFTD's was determined to
be an appropriate setting for these experiments (one has to be wary,
in particular, of using much shorter delays-these can sometimes cause
the CFTD's to act somewhat as lower-level discriminators for the
slower-rise pulses, thus artificially eliminating some of the lower-
energy events for slow rise times). The 0.1 fraction CFTD was used to
start the TAC; the 0.5 fraction CFTD, to stop the TAC. Thus the rise-
time is the time difference between the output of 0.1 and 0.5 fraction
CFTD. All data were recorded event by event on magnetic tape and
later sorted, using the II-Event [Au72] data taking and II-Event

Recovery [Au75] programs on the MSU NSCL Sigma-7 computer.

2. Results and Discussion

Eramples of 5000 + 152Eu spectra taken with the Ge planar detector
are shown in fig. 2-2, and the output of the TAC for these spectra is
shown in fig. 2-3. The "all-event" spectrum shown at the bottom of

fig. 2-2 in A) was obtained by integrating over all the TAC signals.

 

Ge(Li)
Detector

 

 

 

 

 

474
Timing F ii
AMP

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-——J 473 Dual Sum
C F T D and Invert.
(Fraction 0. I) A .
425 425
Delay Delay
425
Delay
46?
5-) TAC/SCA
START STOP
TAC "“"‘
427A
Delay Amp Generator
e fl) A
ADC ADC/SCA DC
AL.____
Figure 2-1

    

 

 

 

 

  

433A

 

455
SCA

 

    

 

 

 

 

 

 

441
Raterneter

 

Block diagram of the fast-slow coincidence circuit used

 

 

7
Planar Detectgrr_

__Y___ ,—

f_. I
I05 D) All Events after Correction

 

 

 

55
~99

 

 

 

 

 

 

 

lO
IO
IO5~ A) All Events
|C33 4,1
I02
I0’
2000 ‘ 4000 ‘ 6000 A 8000
Channel Number
Figure 2-2

Rise-time discriminated spectra with Ge planar detector

Counts per Channel

8

Planar Detector TAC Spectrum

 

 

 

 

 

 

 

 

 

 

 

:03
'02” M
10‘»-

32.9% l5.2% l|.l% 9.0% 7.8% 72% 6.7%

GATE GATE GATE GATE GATE GATE GATE GATE<3UE

I 2 3 4 5 6 7 8
g

300 500

Channel Number
Figure 2—3

Output of the TAC for the spectra shown in fig. 2-2

    

 

Nine separate gates were taken, covering a 4S-nsec range. In B) a
fast rise-time spectrum is shown (Gate 1), while in C) a considerably
slower one is shown (Gate 6). Finally, at the top of fig. 2-2, in D),
I show an "all-events" Spectrum that has been pulse-height corrected
in the manner described below;

In fig. 2-4 I show a summary of my analysis of the nine TAC-gated
spectra: The energy resolution versus TAC channel is shown in the
middle. The resolution (for the 6000 1332.5-keV peak) has been
improved by approximately 10% in the faster rise-time slices. Corres-
pondingly, the peak-to-Compton ratio is shown as a function of rise-
time, where it can be seen there is a ten-fold improvement between
the poorest (Gate 9) and the best (Gate 1). At the bottom of the
figure I show the fraction of total events occurring in each slice.

As anticipated, better energy resolution and peakvto-Compton
ratios can be obtained by discriminating against pulses having slower
rise-times, as demonstrated in fig. 2-4. Somewhat more surprising,
however, is the fact that, for most of the slices, the individual
energy resolution and peakpto-Compton ratios are better than in the
raw "all-events" Spectrum. Encouraged by this result, I made a more
elaborate investigation of each peak shape and centroid position.

The lattice defects contribute to degraded resolution because
they can act as traps for the charge carriers, leading to incomplete
charge collection. At the same time they slow down the rate of the
charge collection. It would not matter if all the charge were not
collected, provided a constant percentage of the charge were collected

for every event. This would change the normalization, but, in

Peak - to - Compton Ratio

Energy Resolution (keV)

Fraction of Events

10
Planar Detector “332.5 keV)

 

26

24

2C)

l6

 

I“
a

N
to

P
o

h.

.0
on

(L2

(Ll

 

 

 

 

 

1 J lfl ! .L__l__.l_l_u
360 450 630 720
TAC Channel Number

Figure 2-4

Summary of the analysis

11

principle, one could obtain quite good resolution in such a system
even without complete charge collection. If there is a direct
correlation between the rise-time of a pulse and the percent of charge
collected, one has a means with which to work on the problem.

In fig. 2-5 I have plotted the centroid of four different Y-ray
peaks versus the TAC channel: The peaks shift linearly as a function
of the rise-time, and the slopes for the lines are energy dependent,
being greater for the higher-energy peaks. The slope of the line for

each peak can be fitted to relation
A I dB (1)

Here a is a constant and E is the energy. This indicates that
the amount of charge trapped in the damaged portions of a detector is
proportional to the energy. 3

This linear dependence makes it easy to develop computer software
to shift each pulse-height value by an amount determined by the rise-

time and the pulse-height. The amount of each shift is given by

Eshift ' AT (2)

After this correction, the "all-events" resolution for the
1332.5-keV 6°Co peak improved from 1.84 keV to 1.77 keV mer. This
is not a remarkable improvement; however, this detector originally had
quite good resolution. Had it suffered more extreme neutron damage,

I assume that the improvement would have been greater. In principle,
the use of more and narrower rise-time slices should make a further

improvement up to the theoretical limit of time resolution.

Number

Centroid Channel

12

Centroid vs TAC Gate (Planar Detector)

 

 

l2l.8 keV 244.7 kev
683.0 — I37eo —
682.0 355.94% |378.0 ELFPM
68l.0 - l376.0 -
680.0 - l375.0 ,.
1:! 7524‘)
6625.0 - \ "73.2 keV - l332.5 keV

6624.0

6623.0

6622.0

- \ 7523.0 -

__ \i 7622.0: \
- R x l

l
: \g 7...“: I

 

 

 

 

662l.O—- ‘—
élllélll174l [[11111]]
C) (D C)
s . .8 R a r:

TAC Channel Number

Figure 2-5

Plots of the centroid of four energies
as a function of the pulse rise-time.

13

The same method was applied for the coaxial Ge(Li) detector which
had indeed suffered considerable neutron damage. Fig. 2-6 shows the
correlation between the rise-time versus the 1332.5-keV y-ray line
shape. This figure shows that with increasing rise-time (i.e.,
increasing the gate number), the peaks get broader and Split into a
doublet. This splitting phenomenon has also been seen in a previous
study [Kr68].

The pulse-height correction method is only suitable if there is
some relation between the pulse-height and the rise-time. The indi-
vidual energy resolution of the gated spectra must be better than the
raw data energy resolution. Thus, for the coaxial detector case, even
if we could correct for the pulse-height, it would not be expected to
improve the energy resolution. With 52 efficiency, by choosing the
rise-time gate, I could get 2.31 keV FWHM (vs 4.92 keV for the raw
data), or an energy resolution improvement of 53%. The "all-events"
and fast rise-time Spectra (corresponding approximately to Gates 1-2
in fig. 2-3) are Shown in fig. 2-7. An expanded comparison of these
two spectra is given in fig. 2-8, where the spectra have been normal-

ized to the height of the 1173.2-keV peak.

COUNTS PER CHANNEL

14

 

l5th Gate

 

 

 

 

 

 

 

I A

 

st Gate

 

a AAA; L- h A. _.__ IL 4

 

CHANNEL NUMBER

Figure 2—6

The correlation between the rise-time and
the centroid of the 1332.5-keV peak.

 

15

AEoSuoev Enuuoonm 3mm a cue: nounnEou Anouv Esuuuonm pouanfiawuumfin uefiulumam

 

 

EDS: Z7N Manganese
ooom . ooom . oooe _ 000m

.0_

No.
mg m
nu
.VO_ W

mEm>m __< 69

 

 

|8uu0u3 lad s

 

2573;. to“.
-60.

 

 

0902.5. sesame: as.

l6

KIN .wfiw mo newuuom a “0 a: 30am

mum ounwfim
cmoEDZ ECCDfiO

 

 

 

 

_ we; SE SE 1
a.» If; a

55%....

,. 265 __<

lauuoug Jed slunog

 

 

 

 

CHAPTER III

THREE DIMENSIONAL EXPERIMENT

In general, the pulse Shape of a coaxial Ge(Li) detector is
determined by the electric field as a function of the position inside
the detector's sensitive region. The following equation gives the

integrated voltage (V) for a coaxial detector as Shown in fig. 3—1.

 

.9._L__ 2 2-2 t _ 2
V C log(b/a) [(log(Xo + (b x°)Tion) log x0)
2 2 2 2.5...
+(logXo - log(Xo - (KO-a )T )1 (3)
el
(t 5 Tel and/or Tion)

Here, Tion and Tel are collection time for ion and electron respec-
tively. Q is the charge, and C is the capacitance. The other terms
are referred to in fig. 3-1. The pulses have different rise-times,
from 60 nsec to 150 nsec, depending on the interacting positions.
This creates a much more complicated relation between the pulse-height
and the rise-time. For example, consider the two pulses A and B:
A is a Slow rise-time pulse, the Slow rise-time resulting from neutron
damage, and B is a naturally slow rise-time pulse without neutron
damage. These pulses, A.and B, arrive at the same.time at the 0.5
fraction level. Since A is the neutron damaged pulse, its pulse—
height is smaller than that of B.

However this complication may be solved by adding another CFTD.
Fig. 3-2 Shows the expected correlation of the time difference

17

18

 

I|l||'||'l"||J

 
 

oitntwllllll ..........i

I

I II!+...!.. .illllllllll

 

 

Figure 3-1

A Coaxial Detector Structure.

19

.20w000am >.oua.o ecu new :OMSSmLm m.¢ta.o one :oozuoa
au:o.~n;aD..C .23.; E: ac :Otmurmogioa 1.33.3.ix0 :<
mum can: w m

04... cote—Eu h.0.._.o

.035 00 on ON 0

 

— u q

1||||"||'

hllllllll

 

ON

.00»:.

3V1. "0!”on 9°O'I'O

20

between the 0.1-0.5 fraction and the 0.1-0.7 fraction. The solid

line indicates the calculated correlation of the output pulse from the
timing filter amplifier with a shaping time constant of 50 nsec [Ka80].
For ideal conditions, we expect all events to be distributed on the
line in fig. 3-2. For neutron damaged cases, I expect those linear
distributions to become area distributions because it takes more time

to collect pulses.

1. Experimental Techniques and Data Processing

 

The block diagram for the triple coincidence experimental elec-
tronics is Shown in fig. 3-3. This diagram.is Similar to that used
previously, except that-there is another CFTD and TAC. The detector
was the 16% Ge(Li) detector previously used, and I used the same
radiation sources.

The 0.1 fraction level of the CFTD was used to start both TAC's;
the 0.5 fraction and the 0.7 fraction levels were used to stop them.
Therefore, for one ADC the input signal is proportional to the rise-
time differences between the 0.1 and 0.7 fraction levels of the CFTD,
and for the other ADC the input is the time difference between the
0.1 and the 0.5 fraction levels. All data were recorded event by
event on magnetic tape using II-event program as before and sorted
off-line.

In order to see the centroid movement for short calculation time,
it was necessary to develop a new sorting program: Event Rec. This
program allows me to accumulate the three dimensional (20 X 25 X 200
channels) energy Spectrum. I used 200 channels for an energy Spectrum,

25 channels for the 0.5 TAC, and 20 channels for the 0.7 TAC. Then,

 

 

21

 

 
    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

The block diagram for the three parameter experiment.

 

 

 

 

 

 

 

 

 

 

 

 

FL\L
Detector Supply
TVNngfi
{ IVnp.
[CFTD ' cho ‘DuoISum
Errnunoml autumn”. EEEEFEEE
Eff: °"
.JL.
Ddoy
START START
AC/SCA TAC/SCA
. ‘ STOP
AC SCA .~ ‘ TAC
um
Invert. '
‘ Delay 15;: 3' ‘Dolay
Mo. W“. Amp. ‘v
GATE l GATE GATE
ADC ADC ADC
Figure 3-3

SCA

 

22

after data were sorted by Event Rec, the centroid of all 500 TAC
gated spectra were obtained by a program called Auto Cen. This

program outline is as follows:

1. Find the highest count channel.

2. Find the channels of the beginning and
the end of the peak.

3. Calculate the centroid. The background
is determined by averaging the number of
counts per channel found in the five
channels immediately preceeding the peak
and five channels immediately following
the peak. This is subtracted from the
area under the peak.

4. Assume that peaks are a Gaussian dis-
tribution. Thus, FWHM - 2.3540

After completely analyzing the data, it was necessary to develop

another sorting program which allows me to gate in two dimensions:

2D GATE. These three new programs are listed in the appendix.

2. Results and Discussion

Figs. 3-4 and 3-5 Show the correlation between the centroid of
the 1332.5-keV y-ray peak and two TAC'S: the 0.7 and 0.5 fractions.
On the 0.5 fraction axis the interval between each line is about 1.5
nsec. On the 0.7 fraction axis, it is 3.3.nsec. The centroid axis is
39 channels from the bottom to the tap (on a log scale). Fig. 3-5
was obtained by rotating fig. 3-4 by 90° clockwise about the centroid
axis. The relative yield for each 1332.5-keV peak.versus the two
TAC settings is shown in figs. 3-6 and 3-7. Again, Fig. 3-7 was
obtained by a similar rotation of fig. 3-6.

AS can be seen in figs. 3-6 and 3-7, the observed events are not

23

 

 

 

 

 

 

 

 

 

 

 

 

(ch)
7509
7467t< _>
740M)
0
.l‘o." ‘3’
A)- (“63“
‘ifia, J EH‘
0 2/
)~ u/ A
40 ’66 5 0 Fred)“ T

Figure 3-4

The correlation between the centroid of
the 1332.5-keV peak and the two TAC signals.

 

 

 

 

 

 

 

 

 

25

.maaswfim o<9 o3u onu m> anon >oxlm.NMMH some now paafih o>wunaou one

sum mcawum

 

 

 

 

i...

   

 

As...

.2503

26

 

 

 

 

 

; , . 3% ,/ H.
age... a. _  

Room»
.2508

 

 

 

 

 

 

 

27

distributed on a line, but rather in a broad area. The centroid
versus the rise-time, figs. 3-4 and 3-5, Show that the higher yield
areas of figs. 3-6 and 3-7 also have higher centroids. This result
agrees with my predictions: These lower yield and centroid regions
result from the neutron-damaged pulses and the higher yield and
centroid regions, from undamaged pulses. A

The energy resolution in the non-neutron damaged area is about
2.59 keV FWHM and in the neutron damaged area about 3 keV. After
accumulating events from the non—neutron area (A in fig. 3-8) by
2D GATE program, we are able to improve the energy resolution to
2.61 keV (at 1332.5 keV) with 10% efficiency, corresponding to a 46%
improvement in energy resolution. With 28% efficiency in the gated
area A and B in fig. 3-8, we get 2.96 keV energy resolution, 40%
improvement in energy resolution. The Spectrum resulting from area
A in fig. 3-8 is Shown in fig. 3-9 with the original spectrum for
comparison, and I have blown up a portion of fig. 3-9 in fig. 3-10.
Since each individual gated energy resolution is much better than the
raw data, in principle it is possible to make the pulse-height correc-
tion for a coaxial detector by finding a pulse-height function which

is a function of both the rise-time 0.1-0.5 and 0.1-0.7 fractions.

28

.MH<U cm HOW meowwou poumw use wcfimaannfie muoan HmaofimcmEHp 030 0:9

mum muswsa

US. 5:02.... ho ....o
.035 om on on om

0?

 

- a - fit

ON

 

035

aw. uouaoas S'o -r0

29

.AQODV name any one cows nonmeEoo AEODDOLV MH<U ON ecu no“: Enuuoonm

. arm unawwm

mum—>52 IEZZ<IQ

 

ooom Doom ooom doom Door ooom ooom cog

    

 

emEo am new: 52 8am.

  

€2.0on 30.. we...

 

 

 

 

 

TENNVHU 83d SlNl'lOCJ

30

mum .wfim m0 cofiuuoa a No a: Scam

calm muswae

mmmZDZ meZ<Iu

 

 

 

 

PEG 0N 05m: £28on

£2.38 38 of.

 
 

 

 

 

WENNVHCJ 83d SINHOU

CHAPTER IV

CONCLUSIONS

It is evident from these results that neutron damage changes the
normal pulse-heights, and these different pulse-height signals result
in lower overall energy resolution. It is also possible to correct
for this effect and improve the overall energy resolution. When the
energy resolution of individual rise-time gated spectrum slices is
better than the composite "all-events" Spectrum, one can significantly
improve the resolution of overall spectrum by either employing the
pulse-height correction method or collecting data only from non-
neutron damaged areas. By using the pulse-height correction method,
the improvement in energy resolution can be obtained without an
appreciable loss in efficiency. In other words, one can salvage the
"degraded" spectrum parts, slice by slice, by Shifting them to their
undegraded equivalent and then adding them up to obtain an improved
spectrum.

For a coaxial detector, it is necessary to use two different
rise-time discriminators because of its more complicated structure.
Also, it may be possible to apply the pulse-height correction to this

type of detector by finding the function,

Here 8 is energy dependent constant, and f(t1,t2) is a function of

the two rise-times. The energy dependent constant can be obtained by

31

32

investigating the peak centroids versus the energy relationship as
described in chapter 2.

Finally, this phenomenon should be studied using gaseous
proportional counters, where conditions can be controlled more readily.
We may ultimately discover that the sequence of events for collection
of charge carriers within the detector producing a pulse, particularly
a phot0peak pulse, from a Ge(Li) detector are considerably more complex
than we had realized. In the meantime, using fast pulse rise-time
discrimination technique can be a practical way for significantly

improving the quality of y-ray spectra.

BIBLIOGRAPHY

[Au72]

[Au75]

[Au67]

[Ba74]

[Be77]-

[Be72]

[C071]

[G171]

[G074]

BIBLIOGRAPHY

II—Event, written for the MSU NSCL Sigma-7 computer
by R. Au (1972).

II-Event Recover, written for the MSU NSCL Sigma-7
computer by R. Au (1975).

R. L. Auole, D. B. Berry, G. Berzins, L. M. Beyer,

R. C. Etherton, W. H. Kelly, and am. C. McHarris,,
Nucl. Instr. and Meth._§l (1967) 61.

R. Baader, W. Patzner, and H. Wohlfarth, Nucl. Instr.
and Meth. 111 (1974) 609.

R. Beetz, W. L. Posthumus, F. W. N. Deboer, J. L.
Maarleveld, A. Van der Schaaf, and J. Konijn, Naclr.
Instr. and Meth. 144_(1977) 353.

D. Belusa, E. Fretwurst, F. Jaber, and G. Lindstrom,
Nucl. Instr. and Meth._191 (1972) 171.

B. L. Cohen, "Experimental Methods of Nuclear Physics",

Chap. 9 :hi Concepts of Nuclear Physics, McGrawbHill,

 

1971.

See, e.g., G. C. Giesler, wm. C. McHarris, R. A. warner,
and W. H. Kelly, Nucl. Instr. and Meth. 21 (1971) 313.
F. S. Goulding and R. H. Pehl, "Semiconductor Radiation

Detectors", Chap. III.A in Nuclear Spectroscopy and

 

Reactions, Ed. by J. Cerny, Academic Press, 1974.

33

[Ie71]

[10:80]

[Kr68]

[L174]

[Ma68]

[Pe72]

[Wh79]

34

"An American National IEEE Standard Test Procedure for
Germanium Gamma-Ray Detectors", ANSI IEEE N42.8-l972
Std. 325-1971.

Private communication with J. Kasagi.

H. W. Kraner, C. Chasman, and K. W. Jones, Nucl. Instr.
and Meth 62 (1968) 173.

G. Lindstrom and R. Lisclat, Nucl. Instr. and Meth.
.11§_(1974) 181.

J. W. Mayer, "Search for Semiconductor Materials for
Gamma-Ray Spectrosc0py", Chap. 5 in Semiconductor
Detectors, Ed. G. Bertoleni and A. Coche, Wiley (1968).
Eg., R. H. Pehl, R. C. Cordi, and F. S. Goulding

IEEE Trans. Nucl. Sci. N§:19,'No. 1 (1972) 265.

A summary of the present state of the art with respect
to certain "exotic" detector materials is given by R. C.
Whited and M. M. Schieber, Nucl. Instr. and Meth. 162

(1979) 113.

APPENDICES

OWO'JOMANNPOWO‘IOUI5IIJNP

lunnmuauumnduw

N
p

22.
3.

25
26

17

450
150

451
151

452
152

50
500

301
310

200

250

260
410

36

APPENDIX

A. Event Rec

DIMENSION IDATAItZOB.ZS.1B).IDATAZtZBBp25aia).ID11(50000):ISEO(4)
+.IDIZ(50000):ITITLE(Z0):IARRYt4)oJTITLE(20)
EOUIURLENCEtIDAT81(1p1p1);1011(1))
EOUIUALENCE(IDATAZ(1.1.1):1012(1))
RERD(105:10)ITC1nHAX1uICEl
READ<1BS:10)ITC2.HAX2:ICFZ
READ(1BS.10)IENG.HAX3:ICFB
REQD(105:15)HTSTOP

READ(105.100)JTITLE

FORMATtBI)

FORN9T(1I)

READIIBS.20)ITYPE NUH

FORH9T(ZI)

1550(1):0

1520(2):1

ISEQ!3):3

HRITE(108:2511TC1oHAX1pICF1
HRITE<108;25)ITCZpHAX2:ICFZ
HRITEtlBszellfiNGpHnXBpICFB

FORM9T(1HO.’TAC’:IS.5X:'HRX':IS.5X-'1CF';IS)
OUTPUT ITYPE.NUM.ISEO

FORMRT(1H0:'ENG':IS:5X:'HRX'IISaSXp'ICF'115)
DO 17 J:1:50000

IDIZ(J)=0

IDIl(J)=0

CRLL EUSET(110:ITYPE:NUH:ITITLEpHnISEO)
IFtH.EQ.0)GO TO 50

IF(H.EQ.1)GO T0 450

1F(H.EO. ZJGO TO 451

IFtH.EQ. 3)GO TO 452

HRITEfiaa 150)

GO TO 999

FORMA ATti Hflo'TfiPE READ ERROR')
GO TO 999

NRITE(108:151)
FORHQT (1H0 'PARQHETER ERROR')
GO TO 999
HRITE!109:152)
GoFORMAT(1H0p'ILLEGL REQUEST RT SET')

TO 999
HRITE(108:500)ITITLE
FORMAT(1H0 TITLE'oSXp2004)

1
REO:JREO+1
T 3 1
ADCE:JQDCE+1
TO 301

ICOIN1:0
ICOINZ:
IEUENT:
JREO:0
JRDCE:
ILEG=0
IF(MTSTOP.LE.0)GO TO 310
IF(IEUENT.EO.HTSTOP) GO TO 400
IEUENT:IEUENT+1
CRLL EUGET(IRRRY:N)
IF(N.EO. 0)GO TO 200
IF(N.E0. -1)GO TO 400
IF(N.EO.1)GO TO 260
IF(N.E0. 2)GO T0 410
IF(N.EQ.3)GO TO 420
IF(IRRRY(1).LT. ITC1.0R.IRRRY(1).GT.HRX1)GO TO 301
IF(IRRRY(3).LT.IT CZ. OR. IARRY(3). GT. MAXZ)GO TO 301
IF(IARPY(2).LT. IENG. OR. IARRY(2). GT. HAX3) GO TO 301
K:(IRRRY(1)-ITC1)/IC71+1
J:(IfiRRY(3)-ITCZ)/ICF2+1
I:(IRRPY(2)-IENG)/ICF3+1
IFIK.GT.20) GO TO 301
IF(J.GT.25) GO TO 301
IF(I.GT.200)GO TO 301
IF(K.GE.11) GO TO 250
ICOIN1:ICOIN1+1
GO TO 301
KzK-10
IDATAZ(I:J:K):IDAT92(IIJIK)§1
ICOINZ:ICOIN2*1
G? TO 30
GO
J
GO

420
400

100

700
701
999

37

1L$G:ILEG+1

0)JT1TLE

NR1TE(201:100)JTITLE
OUTPUT ICOINlaICOINZ
HRITEI2021100)JTITLE
NRITE(203,100)JTITLE
URITE(2047100)JTITLE
HRITE(205;100)JTITLE
HRITE(206p100)JTITLE
HRITE(207.100)JTITLE

FORMRT<20A4)
CALL CPUNCH(IDIl(1):16200o1aNERRp01200)
CRLL CPUNCH(IDIl(16201)o16200 2 NERRo0-Z01)
CALL CPUNCH(ID11(32401);16200 3 NERRo0o202)
CALL CPUNCH(ID11(48601).1400a4;NERR 0:203)
CALL CPUNCH(IDIZ(1);1620075pNERRa 0 204)
CRLL CPUNCH(ID12(16201):16200 6: NERR.0:205)
CRLL CPUNCH(ID12(32401)116200o7. NERRa0o206)
CALL CPUNCH(IDIz(48661)01‘80081NERR000207)
HRITE(100:700) JR 0

FORMAT!1H0:'ERROR ON RERD OPERRTION'oZXaIB)
HR!TE(100:701) JADC

FORMATt1H0p'RDC GROUP ERROR'oZXaIB)
OUTPUT IEUENT

STOP
END

N
Neooowmuawmwoomqmmawm»

mmntrmvuuwnanuwnaww
wow

26.
2

”M
MD

25
26
500

30

27

GOO

000

160

170

+
mahflHXFWHflHHU

38

B. Auto Cen

DIHENSION ITIT(16).IFNRH(10);IDT116384)pY(16384)o9TIT(20)
INTEGER Y

RERD(105050) ATIT

FORHHTCZOQ4)

READ(105D21DEND:9°0)0(IFNAH(I):I:Z:10)

¥3n5I1:1.16364

Y(I):)
CONTI
FORHR
RERD( )ISTRRT ISTOPoKCHAN

5
6) INTER

.IFNflHoB)
5)ITIT

ART CHIS 'o15p10Xp'STOP CH IS 'o15o10x.'THIS IS A PE

menu»

FORMAT(1H0.

+AK STRRTING AT THE 'vIS)

CRLL CRERD(IDT'NCHRNoNRUNpNREEoNSTQRTo11°)
HR!TE(108 505,15T9RTI ISTOPoKCHflN
HR!TE(103 27’1NHE

FORHRT(1HQD'INTERUSL IS 'vIS)

IF(NREE. NE. 0) 3‘0

CRLL S"OOTH(YDIBTT ISTRRTDISTOP)
“RITE(10605°)RTIT

CQLL CPUNCH(Y.16200.001.NERR.0.106)

FIND PEAK

1K=1NTER
18:200

13:0

ISTRRT:ISTART+46

DO 100 1:15TART.ISTOP
ICHQN:0

IHICOUNT:-1

O
p
p
0
(4
I
H
m
4
D
4”
4
AH

A
J

H
I
H
O
O
”5
4
O
Q.

( )) GO TO 110
ICOUNT: Y(J)
H9N:J
NTI NUE

(IHICOUNT.LT.1) GO TO 150
ICHAN

L K-IK

R.

K+IK
B=(IDT(K-IK-1)+IDT(KeIK-2)+IDT(K-IK-3)+IDT(K-IK-4)+IDT(K-IK-S))/
SRB:(IDT<K+IK+1)+IDT(K+IK+Z5+IDT(K+IK+3)+IDT(K+1K+4)+IDT(K+IK+S)J/
5

F
H
C
O
r
E
p
L
FIND CENTROID

DCEN: .

FK: (SRB- SLO)/(SR-SL)

D0 160 II: IPLulPR

N:N+1
BRCKG:SLO+FKI(FLOAT(N))
A:IDT(II)-0ACKG

AREA: AREA+G

CEN: III +CEN

SHACK: SBACK+DACKG
CONTINUE
ERR:SORT(2.ISBRCK+AREA)
CEN:CEN/AREA

D0 170 JK:IPL.IPR
DCEN:DCEN+(FL09T(JK)-CEN)332*(IDT(JK)+BACKG)

nnn

200

39

DCEN:SQRT(DCEN)/AREA
FIND RESULTION FHHN

N:-1
59:0
53:0
SUNB:0
SUNA:0
BRCK: 0

E: 0

DONZBD NH: IPL IPR

BACK: SLB+FKS(FLOAT(N))
AS: IDT(HM) -39

AR: ((HH- CEN)*$2)snS
BB=((HM-CEN+DCEN)sa2)tAS
SUNA:SUNA+AR
SUHB:SUNB+BB

CONTINUE

OUTPUT SUN“! SUHB 7 “RE“

59:2.354k$0RT<SUHA/(ARER-1))
58:2.354*SORT(SUHB/(RREfl-1))

FE:SB-SA
SCHRN:KCHRN+(CEN-IB‘Z00)

. 0 I I 0 U 0 l 0 O O O .

wan-Hnwnawnwnannrnnu
NNNVNRHOflHH‘HrMaHVH‘H

GNTGURWHDOGHNDGIHDNHNIAQNUHGHMDQUNLIUHNNDW

nwn»nrnau

YCHAN:KCHAN+((CEN-DCEN)-ID‘200)
ZCHQN: KCHRN+((CEN+DCEN)- 10.200)
URITE(10B 605
605 FORHRT(1H0 3X:'PL'vGXp'PR'pGXp'SLD'AGXp'SRD'aGXp'ARER'pSXr'ERR'aSX
+‘ggfiCfiéfigfir'CEN'aSXo'DCEN':$X;'FHHH"6Xp’DFH'oGXp'CHfiN’oSX:'LCHAN'
+. .'
$3525‘209m 610)1PL zen. SL8 SLR.AREA.ERR ssncx.crn nc:n.sn. rt.scun~.
*
610 FORMAT(1H0N HIS 2X.15 2X F5. 2:2XoF5. 2 3X: F10.3 3XoF5. 4.3X:F10. 3 3X
+Dr1oo‘fi1XDF7S4OZXDF703DZXIF60‘IZXDFé03'2XDF90‘ sz' F9:‘
150 CONTINUE
IB:IB*1
ISTART:IB¥200
19:1 STQRT+200
ISTART: ISTART +50
IF(ISTART. GE. ISTOP) GO TO 15
100 CONTINUE
800 HRITE(10Bp 550)NREE
550 FORMRT<$SOo12)

GO TO 9
900 STOP
C END

SUBROUTINE SHOOTH(Y:IDT:ISTART.ISTOP)
DIMENSION Y(1)pIDT(1)

IST:IST P-l
15:0
DO 99 1: ISTARToISTOP
99 IS:IS+IDT(I)
HRITE(10B:100)IS
100 FORMAT<1H0'16)
D0 10 I:IST9R
10 Y(I):(IDT(I-1)+2. 0*IDT(I)+IDT(I+1))/4. 0

l‘OUHDflUHflb“WWHGGMDVOHlbuHUH

Nflhufluwwunwuwu‘

22:

50

55

106
15
20
15

26
25

27

28

100

101

132

600

2(1024).IGNI3(1024).IGHXI(1024).ISEG(3).
RTA1(0192)oIDGTAZ(BI9Z):IDGTA3(8192)oIflR

INZoIDYHZoIDHXZoIDXHINBoIDYHBpIDHX3

DIMENSION
+IGHX2(102
ORY(3)'IT

REAL I

DO

HOHHHHHH
UCHBOCHDOC)
D :IIJHIZZ
(HHFU*X)CKW
DUHNUFQHW*°
AA AA AA

TITLE
JTITLE

J
)
A4)

TSTOP
TC1oIHAX1'ITCZoIHAXZoITC3oIHAX3
I

A

I

X

TCIoIHflX1pITCZpIHNXZ.ITC3.IHAX3
gl'pISpSXv'TRC HAXl'oIly 9Xu'TRCZ'vI5 5Xp'TfiC HAXZ'
1

FORHRT(
HRITEIIO
FORH9T‘1H
+'“ .14.3X
HRITE(108:Z7)IXZI IXZZ IXZ3nIXZ4vIYZ

F0RH9T(1HG9'2ND GAT E“'0 ”X1-- 01‘93XO'XZ"'01‘03X0'X3"'01403XO x4
§-" 014'3XI'Y1-"! I4

HRITE(103 ZOlIX31 1X32 IX33oIX34nIY3

FORflflT(1H0v'3RD GATE--’ 'X1--' oI‘o3Xo'X2--'a14p3Xp'X3--':I4:3Xp'
‘--’DI4OBXO'Y1'-'DI4)

IDXHIN:IX1'IX2

IDYH: ITCI- IHRXI

IDHX: IX3-IX4

SLOPHI: IDYH/IDXHIN

SLOPHX= IDYH/IDHX

OUTPUT SLOPHIoSLOPHX

mu“. DIGH‘ p~ IO‘ sows-1H" u H II n u HJH
xovuvwnavvmnwu~puuu~HHHOOQ~H45

- Nvo—cmwv r-
-< ~0I -<v ~ «It

Ix1.IXZ.IX3.Ix4.IY1
ST GGIE""'XI-"II‘I3X"XZ"'I!"3XO'X3”':I4D3Xo'x4

--'

DO 100 I:ITC1.IH9X1
IGHXitI):I/SLOPHX-IY1/SLOPHX+IX3
IGHI1(I):I/SLOPHI-IYIISLOPHI*IX1
IDXHIN2:IX21-IXZZ
DYM2:ITC2-IHAX2

I

IDHX2:IXZ3-IX24
SLOPHIZ:IDYH2/IDXHINZ

SLOPHXZ: IDYflZ/IDHXZ

OUTPUT SLOPHIZ ISLOPHXZ

DO 101 J: ITCZ 2' X2

IGHX2(J): J/SLOPHXZ-IYZ/SLOPHX2+IX23
IGHIZ(J):J/SLOPHIZ-IYZISLOPHIZ+IX21
IDXHIN3:IX31-IX32

IDYHB: ITC3- IHAX3

IDHX3: IX33-IX “3

SLOPHI3: IDYH3/IDXHIN3

SLOPNX3: IDYH3/IDHX3

OUTPUT SLOPHI3'SLOPHX3

DO 102 K:ITC3:IHAX3
IGHX3(K):K/SLOPHX3- IY3/SLOPHX3+IX33
IGHI I3(K):K/SLOPHI3- IY3/SLOPNI3+IX31
HR ITE(108 600)IGHX1

HRITE(108 600)IGNI1
HRITE(108.600)IGHXZ
HRITE<108 600JIGHIZ
HRITE(108 600)IGH X3
HRITE(103:600)IGN13
FORMAT<1H0.ZOI5)
NUH:3

ISEO(1):0

ISEO(2):1

ISEG(3):3

CALL EUSET(110.ITYPE.NUH:ITITLE:H,ISEO)
IF(H.EQ.0)GO TO 51
IF(H.EQ.1)GO TO 450

(:)t\10~*15|\lllll“J17‘(Jl1h-(A)l\30"lll‘li

vIIr-»raut-»raI-ou-ra unsr- .1..........,.
[\ll\i|\)t\)ounra»ra~r-'plura-rI-r-rr-rr-¢||

r-»rd-r- rn~rn»r-»v-rauv-ro-»t-oa-v-Ipath-trn~r->rn~t-
gRMMUNhHO
UHDQOHRA

It 1|»It»(1)113(Altlllllllllnltlltsl
1‘) i‘*l!l£l)(l)'~l(li(Jl.ll 1|)l\)0-'

301
310

200

350

351

260

410

420
400

105

700
701

999

Ml

IF(H.E0.Z)GO TO 451
IF(N.E0.3)GO TO 452
NRITE(10B'150)
G TO 999
FORMAT(1H0:'TAPE READ ERROR')
GO TO 999
HRITE(10B:151)
FORMAT(1H0:'PARAHETER ERROR')
GO TO 999
HRITE<10B:152)
FORMAT<1H0»’ILLEGL REOUEST AT SET')

GO TO 999
HRITE(100:500)ITITLE
F8§:?Té1fl0p’TITLE':5X:29A4)

o-cl-q ¢-00-OO-OO-OO-n

l":l’il)l'll')¢')l')
I'I!:ll'!<:2{:>1:>
(I)¢')<:)t'io-oo-a
IflhHZZZ
II Ill-OIA)I\)

II II us an
Ilitlllli

-0
IFtMTSTOP. LE. 0)GO TO 310
IF(IEUENT. E0. HTSTOP) GO TO 400
IEUENTzIEUENT+1
CALL EUGET(IARRY;N)
ICH7:IARRY(1)
ICHE:IARRY(2)
ICH5: IARRY(3)
.0)GO TO 200

H
'I‘
A
2!"
f'I"
I)

IF(N. E0. -1)GO TO 400

IF(N.E0.1)GO TO 260

IF(N. E0. 2JGO TO 410

IF(N. E0.3)GO TO 420

IF(ICH?.LT. IGMI1(ICH5). OR. ICH7.GT.IGHX1(ICH5))GO TO 350
IDATA1(ICHE): IDATA1(ICHE)+1

ICOIN1:ICOIN1*1

IF(ICH7. LT. IGNI2(ICH5). OR. ICH7.GT.IGHXZ(ICH5))GO TO 351
IDATA2(ICHE): IDATAZ<ICHE)+1

ICOIN2:ICO IN N2+

IF(ICH7. LT. IGHI3<ICH5). OR. ICH7. GT. IGHX3(ICH5))GO TO 301
IDATA3£ICHE): IDATA3<ICHE)+1

ICOIN3:ICOIN3*1

GO TO 301

JREO:JREO+1
GO 301

JRDCE: JRDCE+1
GO TO 301
ILEG= ILEG+1
50 01
IT
IT
IT
CRLL CPUNCH(IDATA 92a1pNERR.B.201)
CQLL CPUNCH(IDATAZDB19202'NERRDBD282)
CQLL CPUNCH(IDATaa.B192;3.NERR.0.ZB3)
HRITE(1089700) JREO

FORNQT(1Har'ERROR ON RERD OPERATION'.ZX.IB)
HRITE(1980701) JADCE

FORMAT(1HBD'RDC GROUP ERROR'tZX:IB)

OUTPUT IEUENT

ggggUT ICOINI'ICOINZvICOINa

END

MINIMUM"?

312