51824”
9% SW
_ u . .

W

Y"'0ESIS

This is to certify that the
thesis entitled
NEUTRON/GAMMA-RAY PULSE SHAPE DISCRIMINATION
IN LIQUID ORGANIC SCINTILLATORS

presented by
John Edward Yurkon

has been accepted towards fulfillment
of the requirements for

_M..S_.__ degree in _Eh¥.si(:5_

wawg

Major professor

Date?)m!;/T77

0-7539

 

(WM FINES:
25¢ per ‘0 per it.
RETURNUS LIBRARY MATERIALS:

Place in book return to move
chum fro- circulation records

 
   
   
     
     
 

‘gvlq_’£shnl_‘
' m“
(5‘61...
‘tir‘izz'fl. .. v

1"
r“

  

 

 

 

NEUTRON/GAMMA-RAY PULSE SHAPE DISCRIMINATION
IN LIQUID ORGANIC SCINTILLATORS

BY

John Edward Yurkon

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Physics

ABSTRACT

NEUTRON/GAMMA-RAY PULSE SHAPE DISCRIMINATION
IN LIQUID ORGANIC SCINTILLATORS

BY
John Edward Yurkon

Origins of pulse shape differences are discussed.
A method of using these differences for neutron/gamma-ray
discrimination is also discussed and the factors affecting
the qualities of the pulse shape discrimination are shown
on theoretical grounds and then measured.

Optimal pulse shape discrimination is measured for the
liquid organic scintillators NE213 and NE224.

Problems with a large volume neutron time-of-flight
detector at Michigan State University are presented along

with suggested solutions.

ACKNOWLEDGMENTS

I would like to thank Dr. A. Galonsky for his careful
reading of this thesis and for his guidance and help in re-
searching the material, and in performing the associated
experiments.

The assistance of Mr. M. Wallace was very helpful in
preparing the figures in this thesis. Also the patient ex-
planations of the use of the computer facilities by Mr. R.
Melin and Mr. R. Fox were helpful in the preparation of
this thesis. The glass-blowing shop of the Department of
Chemistry constructed the various cells used in researching
this thesis. I would like to thank Ms. D. Barrett for her
help in organizing the material used in preparation of this
thesis and for her encouragement to stick with it.

The typing of this thesis by Mrs. Betty McClure is
greatly appreciated.

Finally, I would like to thank my parents and friends

for their support during my graduate study.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

l.
2.

INTRODUCTION
THEORY OF PULSE SHAPE DISCRIMINATION

2.1 Excitation Process and Decay

2.2 Zero-Crossover Technique

2.3 Model of System Performance

2.4 Calculation of System Performance

FIGURE OF MERIT FOR NE213 AND NE224 UNDER IDEAL
CONDITIONS

3.1 Figure of Merit for NE213
3.2 Figure of Merit for NE224

LIGHT COLLECTION IN THE SCINTILLATION COUNTER

4.1 Effects of Light Attenuation on Figure of
Merit

4.2 Light Collection Problems with a Large
Neutron Time-of—Flight Detector at Michigan
State University

DEOXYGENATION OF SCINTILLATION COUNTERS

5.1 Method of Deoxygenation
5.2 Diffusion Model

A NEUTRON TIME-OF-FLIGHT DETECTOR WITH IMPROVED
PULSE SHAPE DISCRIMINATION

6.1 The Detector
6.2 Performance

6.2.1 Detector filled with NE213
6.2.2 Detector filled with NE224

6.3 Comments

Page
iii

iv

10
15

22
22
33
36
36
36

45

45
45

50
50
'51

51
56

56

7. PHOTOMULTIPLIER PROBLEMS

7.1 Gas Ionization
7.2 After-Pulses in the Present System

8. CONCLUSION

APPENDIX A - COMPUTER PROGRAM FOR CALCULATING E,
D and 8

APPENDIX B - COMPUTER PROGRAM FOR GENERATING PULSE
SHAPE SIGNATURES

APPENDIX C - COMPUTER PROGRAM FOR CALCULATING FIGURE
OF MERIT D AS A FUNCTION OF PULSE HEIGHT FRACTION K

APPENDIX D - COMPUTER PROGRAM FOR CALCULATION LIGHT
ATTENUATION IN A RECTANGULAR DETECTOR

APPENDIX E - LIGHT ATTENUATION IN LIQUID ORGANIC
SCINTILLATORS

LIST OF REFERENCES

ii

Page
62

62
62

68
70

72

74

76

79

100

Table

1.

10.

11.

El.

E2.

E3.

LIST OF TABLES

Data for Decay Times and Relative Contributions of
Decay Modes in NE213. Data taken from Sabbah and
Suhami13

Theoretical Figure of Merit (D) for a Given Number
of Photelectrons (P)

Data for Number of Photoelectrons per KeV for an
NE213 Scintillator Mounted on an RCA 8850 Photo-
multiplier

Theoretical Figure of Merit (M) at a Given Energy
for NE213

Figure of Merit (M) for a 45 mm x 50 mm NE213
Scintillator

Figure of Merit (M) for a 45 mm x 50 mm NE224
Scintillator

Theoretical Light Transmission, in Percent, through
a Rectangular Scintillator of Dimensions 2.7 cm x
12.8 cm x 83.8 cm. The Source Positions X, Y, and
Z are Defined in Figure 13.

Figure of Merit (M) for a 1.24 m x 45 mm NE213
Scintillator with Source at Near End of Scintil-
ator

Figure of Merit (M) for a 1.24 m x 45 mm NE213
Scintillator with Source at Center of Scintillator

Figure of Merit for a 1.24 m x 45 mm NE224 Scintil-
lator with Source at Near End of Scintillator

Figure of Merit for a 1.24 m x 45 mm NE224 Scintil-
lator with Source at Center of Scintillator

Attenuation Lengths in Meters as Measured with
Laser Light

Wavelength-Averaged Attenuation Length in Meters
for Unpainted Cells Containing NE213

Practical Attenuation Lengths in Meters for Black-
Painted Cells containing NE213

iii

Page
17
19

28

29
32
35

40

52

54
58
6O
85
93

95

Figure

1.
2.

17.
18.
19.
20.

21.

LIST OF FIGURES

Energy transfer process from solvent to solute.

Production of T1 states. The arrow represents
the spin of the electron.

Block diagram of pulse shape discrimination
electronics.

Unipolar and bipolar pulse shaping.

RC shaping circuit for dynode pulse.

Plots of expected PSD signatures for NE213.
Figure of merit versus discrimination level.
Scintillator cell, 50 mm x 45 mm.

Schematic of LED pulser of Hagen and Eklund.19
Electronics for determining Pel'

NE213 PSD spectra 45 mm x 50 mm cell.

NE224 PSD spectra 45 mm x 50 mm cell.

A large neutron time-of-flight detector.
Placement of scintillation cell.

Electronics for measuring light pipe attenuation.

Compton edge spectra without light pipe (a) and
with light pipe (b). .

PSD for NE213 cell 1.24 m, near end.
PSD for NE213 cell 1.24 m, middle.
PSD for NE224 cell 1.24 m, near end.

PSD for NE224 cell 1.24 m, middle.

After-pulses in anode signal (a), singly differen-

Page

11
20
21
23
25
26
31
34
37
41
42
43

53
55
59
61
63

tiated dynode signal (b), and doubly differentiated

dynode signal (c) with 1 usec delay line shaping.

iv

Figure Page

22. After-pulses in singly differentiated dynode 65
pulse (a), and doubly differentiated dynode
signal (b), with 400 nsecs delay line shaping.

23. PSD spectra with 1 usec double delay line shaping 66
(a), PSD spectra with 400 nsecs double delay line
shaping.

El. Details of scintillator cell with dimensions 81
2 e 1:.

45 t 1 mm 2.0 t 0.2 mm 2 mm

22.0 i .6 mm 1.5 0.2 mm 2 mm

H-

constructed of Pyrex glass.

E2. Photograph of scintillator cell and laser setup. 82

E3. Transmitted light intensity vs. path length for 83
NE213.

E4. Transmitted light intensity vs. path length for 84
NE224. -

E5. The setup for measurements of the "wavelength- 87

averaged attenuation length".

E6. Block diagram of the electronics for measurements 88
of the "wavelength-averaged attenuation length".

E7. Compton edge channel vs. path length for NE213 in 91
the unpainted 45 mm and 22 mm cell (darkened line
indicated region over which the least squares fit
was performed).

E8. Compton edge channel vs. path length for NE213 in 97
the black painted 45 mm and 22 mm cells (darkened
line indicated region over which the least-squares
fit was performed).

I . INTRODUCTION

Neutron time-of-flight detectors typically employ liquid
organic scintillators due to their fast response which gives
good time resolution. They are, however, as most scintil-
lators, sensitive not only to neutrons but also to gamma-
rays. It is necessary to discriminate between the two.
Liquid organic scintillators such as NE213 and NE2241provide
a means of accomplishing discrimination due to their differ-
ing responses to neutron and gamma-ray induced scintillations.
These responses and their origins will be discussed.

The quality of neutron/gamma-ray discrimination depends
on many factors. Among these are the purity of the scintil-
lator, light collection in the detector, optimization of the
electronics, system noise, and the properties of the scintil-
lator itself.

It is the hope of this thesis to show the theory behind
neutron gamma-ray pulse shape discrimination, and to provide
the information necessary to design good pulse shape discrim-
ination into a neutron time-of-flight detector. This will
be done by investigating the factors mentioned and testing
them on a detector system designed with these factors in

mind.

2. Theory of Pulse Shape Discrimination

2.1. Excitation Process and Decay

Liquid organic scintillators, as used in neutron/gamma-
ray discrimination, are a two part system. They consist of
a solvent, which is the primary energy abosrber of the nuclear
radiation, and a solute, which is efficient in accepting the
energy from the solvent and converting it into photons. The
scintillators NE213 and NE224, which are discussed here, are
xylene and psuedo cumine solvent based scintillators respec-
tively. These two solvents are used because they have low
lying energy levels, and non-bonding electrons (n) which need
little energy to attain higher energy levels.2 Also, these
higher levels are long lived compared to the migration time
of the excitation between the solvent and solute. This is
necessary if the solute is to be efficient in converting the
absorbed energy into photons.

The solvent molecule, after ionization by radiation,
recombines with an electron and forms a neutral molecule in
an excited state. Generally, these are the upper excited
singlet states 83, $2 and 81' The S1 excited state is the
state of primary interest, since this is the level that is
involved in the transfer of energy from the solvent to the
solute, at normal solute concentrations. The upper excited

singlet states are depopulated through internal conversion

3

and other deexcitation processes. Most of the S1 states
are produced directly by ion recombination. For p-xylene
based scintillators, approximately 43% of the S1 states are
produced by internal conversion of the S3 states.3 The S1
states then transfer their energy to the solute promoting
it to its excited singlet state Fl as shown in the energy
level diagram of Figure 1.

An excited triplet state Tl can also be formed. However,
since the transition 50 + T1 requires a change of spin, it
is not very likely to occur. Instead an S1 state is formed
which undergoes an intersystem crossing to the T1 state shown
in the energy level diagram of Figure 2. In the intersystem
crossing process the S1 state interacts to produce the T1
state. Therefore, the yield of the T1 states is proportional
to the density of S1 states which is inturn proportional to
the ionization density. The T1 + 80 transition, also being
forbidden, is unlikely to occur. When two Tl states interact
an intersystem crossing occurs yielding an excited singlet
state, Tl + T1 + Sl + SO. This Sl state then inturn trans-
fers its energy to the solute producing the scintillation.“
The time for the decay then is long compared to that of the
S1 + 50 transition since the T1 + T1 + Sl + S0 transition

depends on the migration time of the molecules in the T1

state. The S1 + S decay has a mean life on the order of

0

4 nsecs. The T1 + 30 decay has a mean life time of the order
of 100 nsecs.
When the scintillator is used for neutron/gamma-ray

discrimination, the neutron produces an ionization track by

. Dwa
.muDHOm ou udm>a0m Eonw mmmUOHd Hmmmcmuu mwumcm H mm. .m

 

 

 

 

 

 

4, o
f m
1 z 111111111
J ........ 34w
mm

 

 

.couoooam mno mo swam ecu mocomoudou Sousa 05B .mouoom HE mo defiooopoum ..m ouswwm

e E \Q ,_l om
\

x
x

\

 

 

.
wammouo .
Al _
_

.

a

.

 

EmumxmpmucH

 

 

elastically scattering a proton. A gamma-ray, on the other
hand, Compton scatters an electron. The difference in the
ionization densities of the two particles then forms the
basis for distinguishing whether a neutron or a gamma-ray

has been detected, since the light decay of the scintillation
will have a different shape. The photon number per unit

time for the electron and proton, respectively, are then

generally considered to be given by:5

(l) Ae(t)=(Aefexp(-t/r)/r + Aesexp(-t/IeS)/Tes),

(2) Ap(t)=(Apfexp(-t/T)/r + A exp(-t/Tps)/t ),

P5 P8

where Aef and Apf are the relative contributions of the fast
part of the decay produced by the electron and proton respect-
ively. Aes and Aps are the relative contributions of the

slow part of the decay of the scintillation. T is the fast
decay time for both the proton and electron induced scintil-
lations. Tes and Tps are the slow decay times for the
electron and proton induced scintillations. Ae(t) and Ap(t)

are normalized such that ofmA(t)dt=l.

2.2. Zero-Crossover Technique

The zero-crossover technique measures the time it takes
for the linear signal from the photomultiplier to reach one
half of its peak value. The block diagram in Figure 3 shows

one method of measuring this time.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

brash

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.mowconuooao cowumCHEHHomwp madam omHSQ mo Emuwmwp xoon .m ouswflm
+|||L, o<e seams omHomo
awe «ems unmum .ae< «Home mas
omouo <ooH z< 0mm: H oouuo
acum
comm o
Ana .dE< .dEmmum mcoc< mom .z.m w
<ome .A.n.n mag omuuo ommm<om w
omqa Ho illll_ HHcH Ho oouuo mcoc>o . m
an
<om9 _ m
<omq owwoq
oouuo

 

The anode signal from an RCA 8850 photomultiplier pro-
vides a fast start signal which is passed into an ORTEC 473
constant fraction discriminator. The constant fraction
discriminator provides a signal when the anode pulse reaches
a fixed fraction of its peak value, thus reducing time jitter
due to amplitude variation. This signal is then delayed by
cable to compensate for other circuit delays and fed to the
start input of an ORTEC 437A time to amplitude converter.

The linear dynode signal is amplified by an ORTEC 113
preamplifier and sent through a Canberra 1411 double delay
line amplifier. The double delay line amplifier first delays
the incoming pulse 1 usec., inverts it, and adds it to the
original pulse, thus producing a unipolar pulse of l usec.
duration, as shown in Figure 4a. The trailing edge is an
inverted image of the leading edge. This unipolar pulse
is further delayed l usec., inverted, and added to the orig-
inal unipolar pulse as in Figure 4b. This produces a bipolar
pulse which crosses through zero at approximately 1 usec.
plus the time it takes for the original signal to reach 1/2
peak value.6 This zero-crossing time is determined by a
Canberra 1436 timing single channel analyzer in the zero
crossing mode. The fast signal from the timing single
channel analyzer is used as a stop signal for the time to
amplitude converter.

The output of the time to amplitude converter is then
related to the rate of decay of the scintillation. To know

exactly how it relates we must know what happens to the

f

 

V(t)

....... Lkl -- ...... Ll ........

1186C

 

 

 

F
F
F

 

 

 

 

 

 

 

 

Unipolar and bipolar pulse shaping.

Figure 4.

10

original light pulse as it is detected and the resulting

signal passes through the system.‘

2.3 Model of System Performance
Let us choose the case of the electron induced scintil-

1ation. The current output of the dynode will then be
(3) i(t)=Ae(t)PGq,

where P is the number of photoelectrons produced at the photo-

7 Since G is constant

cathode and G is the multiplier gain.
it will be considered to be one. Therefore, if at time t=0
a light pulse occurs at the dynode, the voltage V(t) at time

t is given by 7
(4) V(t)=Rofti(t’)g(t,t’)dt’,

where g(t,t’) is the transfer function of the RC dynode
circuit, a portion of which is shown in Figure 5. The trans-

fer function for this circuit is‘

l/(RC)exp(-(t-t’)/(RC) for t>>t’
(5) g(t,t’)=
0 t>>t’ .

11

Part of
Preamplifier

Part of
Tube Base

Dynode

 

< >

 

 

 

 

 

 

O
-_———————-——_—_———-— --—-—————_——-

Figure 5. RC shaping circuit for dynode pulse

12

Performing the integral in equation 4 then yields

(6) V(t)=£%-T:f ] [exp (- g ) - exp ( ‘ %5 H -+
—— - 1
RC

Aes [exp (- f: ) - exp ( - RT:- )] .
Tes _ es
RC

Since RC is typically several orders of magnitude larger than

Tes or r and much larger than the time t at which V(t) is

evaluated, equation 6 may be simplified to

_ P _ _ E _ _ t
(7) V(t)—Eg Aef [1 exp( 1)] +Aes 1 exp[ .125]

To find the distribution of V(t) we can first find the

distribution for the integral light output N=Pf(t) given by
(8) N=Pf(t)=of”PAe(t)dt.

Integrating yields

”k?

(9) Pf(t)=P Aef [1 - exp ("%J + Aes l - exp [- es] .

Since the number of photoelectron N arriving at time t=T has
a Poisson distribution about the average value P, we
can assume that the probability that the Nth pulse arrives

between time 0 ant t is given by the Poisson distribution9

13

exp(-Pf(t)[Pf(t)]N Pf(t)
N! °

 

(10) PN(t)=

Note that equation 6 can be expressed, by using equation 9,

as
(11) V(t)= g9 f(t) or g3 .

Since V(t) and N(t) only differ by a constant factor

the same distribution holds for V(t), namely,

 

 

Q G 1
N
exp [_ cvm [cvm [cvm

- Q. q J as.
(12) Pv(t)- N! .

 

 

 

 

The probability that the voltage has reached a value

of V(t) between t and t+dt is

exp [_ CV(t)

_ ' q
(13) Pv(t)dt- ,

 

q

[cvmi N [dim]
(N-l)! ‘ i 7 °

 

This, as stated earlier, assumes a Poisson distribution
of the Pulse of P photelectrons. This, inturn, implies the
voltage pulse V is normally distributed about the average
value of V at time t=T. If the voltage is to be exactly
V(t) then the pulse must have P photoelectrons so that (P-N)
photoelectrons must arrive in the remaining time T-t. Then

a similar derivation as for equation 13 yields the probability

_ _ emu“ [ __ chflJP'N
exp [ [1 q , 1 -—7i—— .

(P-N)!

for this to happen as

V-V(T

14
Thus, the combined probability PV V(T)(t)dt that the
I
voltage reaches V(t) between t and t+dt in a voltage pulse
of exactly V(T) is
(15) P

)(t)dt= P )(T-t) or

V,V(T V'PV(T

q q
(N-l)! (P-N)!

CV(t) 13"“ [CV(t)]N"1 v(t)

 

exp(-1) [l -
dt

 

r

This is equivalent to the treatment by Kuchnir and Lynch.1°
For a fixed fraction k, V(t)=KV(T), or N=kP, the mean

time at which this occurs is

othP

)(t)dt

(16) E: V,V(T

 

T
of PV,V(T)(t)dt

and the variance of t is

T 2

_ t P (t)dt _
(17) €:=<t:>_t:= Qf v,V(T) _ t2

T
of PV:V(t)dt

A similar derivation yields the same result for the proton
induced scintillation.
One method of characterizing the ability to distinguish

pulse shape is the figure of merit D defined by

15

2_ 2 2 11
where - e+ p'

of merit M defined by the relation

In practice, one measures the figure

(19) M: te ' t2_ .

sum of FWHM

If the t distribution were Gaussian, then 2.36M;D_<=3.34M.12

2.4. Calculation of System Performance.

The figure of merit D can be calculated from various
levels of photoelectron numbers, and the fraction of pulse
height at which one chooses to measure the time difference
of the proton and electron induced scintillation by eval-
uating equation 18. First, expanding equation 16 one obtains

P .

T . F t
f t l- A 1- exp - -fl+
(20) t= ° L L ef L

 

 

 

 

of 1— LAef [-1- exp L— ?]]+

 

 

 

 

 

 

 

 

 

 

-' N-l
Lgl x
es
» ' N-l
-5121) x
L es‘

 

 

F
A A
_Te:f._ exp [- E:l+ —§§- exp [- L] dt .
r T 1
es es
VA A L
ef t es t
——— exp [- -]+-——— exp - — dt
T 'r Tes [ res]

A similar result for {2 is obtained by changing the first

t to a t2 in the top integral. One could find an analytical
solution if t were assumed to approach infinity. However,
little insight would be gained from a very tedious task.
Instead equations 16, 20 and 18 were evaluated using the
program listed in Appendix A, on an XDS Sigma 7 computer
located at Michigan State University's Cyclotron Laboratory.

The values of Ae , A listed in Table 1

s ef’ Aps’ A

pf’ T’ Tef
were obtained from the paper by Sabbah and Suhami.13 These
were used in all of the following calculations. Table 2
lists the results as a function of the photoelectron number
P for a fixed fraction of pulse height. The constant k was
chosen to be 0.5. Since the distribution is Poisson, the
figure of merit D should increase as the square root of P.
Inspection of Table 2 shows the expected increase in the
figure of merit as the photoelectron number is increased.

One is also interested in the actual appearance of the
pulse shape signatures. These were obtained by summing the
probability distributions for the electron and proton induced
scintillations assuming an equal photoelectron yield for
both types of scintillations. The program listing for this

is in Appendix E. Figure 6 shows the expected pulse shape

signatures as a function of increasing photoelectron number.

17

Table 1. Data for Decay Times and Relative Contributions
from Sabbah and Suhami.13

 

 

Aef = 0.5 Tef = 4 nsecs
Aes = 0.5 Tes = 25 nsecs
A f = 0.8 Tpf = 4 nsecs

nsesc

W
'U '0
II
C
N
,4
ll
.5
\l

5 ps

 

18

The fraction of the integrated pulse height at which
one chooses to do the timing affects the figure of merit.
For a fixed number of photoelectrons, 300, the figure of
merit was calculated for various values of k using the
program listed in Appendix C. The results are graphed in
Figure 7. One can see that the best discrimination between
the electron and proton induced scintillation occurs at
k=0.87. This treatment assumes no system noise.

Inspection of Equation 7 shows that the rate of change
of the pulse decreases as time increases. The uncertainty
of the timing of the signal due to noise jitter is given

approximately by

_/2G2.35AV

(21) At— Q2, ,
dt t

 

where G is the delay line amplifier gain, AV is the RMS
noise at the input, and dv/dt is the time rate of change of

1“ The factor of 2.35 converts from

the signal of interest.
the RMS value to FWHM, and the /2 factor occurs because of
two level measurements. Thus, as the time rate of change

of the singal increases, the timing jitter decreases. There-
fore a compromise may have to be made to minimize the effects
of amplifier noise and fraction of the pulse height discri-

mination level on the quality of the pulse shape signature.

19

Table 2. Theoretical Figure of Merit (D) for a given Number
of Photoelectrons (P).

 

 

 

P D

500 4.74
1,000 6.70
1,500 8.23
2,000 9.50
2,500 10.64
3,000 11.67
3,500 12.56
4,000 13.45
4,500 14.25
5,000 15.04
5,500 15.81
6,000 16.44

 

COUNTS/CHANNEL

20

 

 

 

"l?"
p
)—
I
l

 

 

 

 

l l l L J L A/k l

 

 

d—
q

1

[d] Number of Pho+oe|ec+rons
- 0. “+8
b. 93
_ c. 180
d. 397
e. 871
- f. 1.295
g. 2,500

 

l l 1

 

Figure 8.

CHANNEL NUMBER

P|o+s of expec+ed P.S.D. Signo+ures for NE213.

21

._®>®_ COIOEECQQU mDme> :LmE 05 OLD?“—

g Jm>mj ZOHH<ZHZHQOHQ
m6 0.6 71.0 Nd

. _ . . 1

.m 959.1

0.0

 

om;

oo.m

Omar

cod

[G] 1183») ac Hanan

3. FIGURE OF MERIT FOR NE213 AND NE224 UNDER IDEAL CONDITIONS

3.1. NE213

A test cell was constructed of Pyrex glass as shown in
Figure 8. The cell was painted with NE56015 reflector paint
to optimize light collection. The cell was cleaned with
nitric acid and rinsed with distilled water and allowed to
dry. A teflon cap was machined to seal the cell. Teflon
was chosen since it is nonreactive and highly insoluble,
thus lessening the chances of contamination.

The cell was filled with NE213 and helium gas was then
bubbled freely through it for 24 hrs to remove oxygen contam-
ination. After bubbling the cell was capped. Care was
taken to ensure that the remaining void was filled with
helium.

The cell was then attached to an RCA 885016 photomulti-
plier tube with optical coupling grease. The number of
photoelectrons produced per Kev of energy deposited in the
scintillator was then measured using the method of Kelley
35 31.17 The average number of photoelectrons P in the

pulse was taken to be18
(22) p: El 2 1+ ii3§___
_ c Gl(G-l) '

22

 

 

 

 

 

 

 

)k
A
2 mm wall
50 mm Pyrex
\l \/

 

 

 

 

 

r< 45m >1

Figure 8. Scintillator cell, 50 mm x 45 mm.

24

where G1 is the average gain of the first dynode, G is the
average gain of the other dynodes, I1 is the centroid of the
peak in channels, and o is the standard deviation in channels
of the peak. Taking Gl=5 and G=3 we have P=1.36(Il/c)2.

If the channel number C of the Compton edge, of known
energy, is then measured, the number of photoelectrons per

KeV of energy deposited in the scintillator is then

9 constructed, the

A very stable LED light pulser was1
schematic of which is in Figure 9. The LED light pulser
provides a very stable pulse, so that the width of the
observed peak is due entirely to counting statistics and
not due to drifts in the source. The LED pulser was coupled
to the same photomultiplier that the NE213 cell was coupled
to with a fiber optic light pipe. This allows the measure-
ment of P to be made without disturbing the coupling of the
cell to the photomultiplier.

The electronics used are shown in block diagram in
Figure 10. The LED light pulser produces a light pulse
that is linear with respect to its input trigger pulse ampli-
tude. An Ortec 448 research pulser was used to trigger the
LED pulser. The Ortec pulser was set for a decay time con-
stant of 10 nsecs and a rise time of 20 nsecs. The output
was adjusted for an LED pulse such that the spectrum's

60

centroid would be roughly equivalent to the Co Compton

edge. The pole zero of the CI 1411 delay line amplifier

25

d4. 5.

ml.ocsaxm can comm: mo HmmHSQ 9mg mo owumaonom .m ounwwm

m7:
3,

 

 

 

 

 

4

 

 

 

Onmczm

OnN¢2N .P

dz.

 

r\A\

 

dim

 

k
/l..l “2:;

12min.

 

 

 

 

w/L
“.10.. “7&3;

.6on

4%

.hnooo

n_+

6
2

 

 

 

 

 

 

 

dE<
A .1H.D.Q
<oz cu Haea Ho
use anwfia
.dEmmum
mHH oouuo

 

.Hom wcHEuooop How mowcopoooam

mafia Laws; .nuliv

 

 

 

 

opocza

 

comm mash
mom omuno

.oH munwaa

 

 

uwmflsm
summomom
mqq oouuo

 

 

 

 

 

Homage ems

 

 

 

 

C

 

omww¢om

 

 

mHNmz WNHV

 

 

 

 

 

oUMDOm 00

or

27

was then adjusted so that the trailing edge of the singly
differentiated output pulse would return to the baseline.

Spectra were then taken for various settings of the
Ortec pulser. The LED pulser was then shut off and a
spectrum was taken using the 60Co source. The results are
listed in Table 3. For NE213 an average of 1.8 photoelec-
trons per Kev was obtained for Pel'

Using the data in Table 2, and knowing that there are
1.8 photoelectrons per KeV, the expected M value versus
equivalent electron energy can be tabulated. This is done
in Table 4.

The NE213 scintillator cell was then tested for its
neutron/gamma-ray pulse shape discrimination characteristics
using the electronics of Figure 3. The system was tuned
using a 60C0 source. The high voltage for the photomulti-
plier was set at approximately 1800 volts and the CI 1411
delay line amplifier's pole zero was then adjusted so that
the singly differentiated output pulse would return to the
baseline with little under or overshoot. The risetime
adjustment of the CIl436 timing single channel analyzer
was then adjusted for minimum walk. The Ortec 473 constant
fraction discriminator was set at the minimum threshold that
would not pass noise and was operated in the leading edge
node. The light, logic, and PSD signals were routed to a
multichannel analyzer system comprised of a PDP 11/45 computer

connected to a CAMAC crate. The logic pulse strobes the

system and merely tells the computer to look at the light

28

Table 3. Data for Number of Photoelectrons per Kev for an
NE213 Scintillator Mounted on an RCA 8850 Photo-
multiplier.

 

 

Pulser Setting Centroid F.W.H.M. PeIEPhotoelectrons/KeV]

 

9 979.8 56.0 1.66
7 896.5 53.5 1.67

PelAvg. l.8i0.1
5 774.5 48.1 1.78

3 612.17 41.2 1.92

 

29

Table 4. Theoretical Figure of Merit (M) at a Given Energy
for NE213.

 

 

Energy [MeV] Figure of Merit

 

0.28 1.60
0.57 2.28
0.85 2.75
1.14 3.16
1.42 3.63
1.70 3.97
1.99 4.17
2.27 4.49
2.56 4.88
2.84 5.02
3.13 5.34
3.41 5.69

 

30

and PSD signals. The system is capable of collecting the
PSD spectra gated by light levels set by software.

After tuning the system, the spectra of the 60Co Compton
edge were collected to calibrate the channel numbers in
term of equivalent electron energy deposited in the scintil-
lator. Then a PuBe neutron and gamma-ray source was placed
in position and spectra of the PSD signal, gated by various
light levels, were taken. The results are shown in Figure 11
for various energies. The FWHM of the peaks and the cen-
troids were obtained using the program BKANAL on the PDP
11/45. The background was not subtracted. In practice this
cannot be done in real time; hence it is more meaningful to
leave it in. The figure of merits are tabulated in Table 5.

The figure of merit for an equivalent electron energy
range of 1.1 to 1.4 MeV was found to be 1.7. This is less
than the predicted value of 3.2 at 1.14 MeV, which however
does not include the contribution due to system noise. Thus
the actual value should be expected to be less than that
predicted. R. A. Winyard gt g1.2° found M to be 3.1 for
NE213 at an energy range of 0.91 to 1.1 MeV. Winyard's
value for M at this energy may be higher due to reduced
noise. However, it had been assumed, in this thesis, that
the values of Aes’ Aef’ Aps' Apf were independent of energy.
This is clearly not the case, since the ionization density
changes as a function of the energy of the ionizing particle.

The history of the NE213 was not known either. Although

care was taken to deoxygenate the scintillator, it may have

COUNTS/CHANNEL

 

11—
«r

l
l

 

ll 9

.LI

 

 

   

Aj‘__

4L JA 1 l ,
r l V I
l J}//\\L¥J

 

 

 

q-

 

 

CHANNEL NUMBER

NE213 P.S.D. spectra L+5 mm X 50 mm cell.

Figure 11.

Energg Range in MeV

«3 ‘”<0 ‘1 OICT O
o " o I I o I

0.13 Thru 0.53
0.53 Thru 1.12
1.12 Thru 1.65
1.85 Thru 2.15
1.12 Thru 1.23
0.13 Thru 2.15
2.15 Thru 3.13.

32

Figure of Merit (M) for an 45 mm x 50 mm NE213
Scintillator

Table 5.

 

 

Energy Range [MeV] Figure of Merit

 

0.13 thru .53 1.11
0.53 thru 1.12 1.47
1.12 thru 1.65 1.65
1.65 thru 2.15 1.71
1.12 thru 1.23 1.71
0.13 thru 2.15 1.62
2.15 thru 3.13 1.69

 

33

contained some impurities which could quench the desired

triplet states. Oxygen is just one such impurity, although

probably the most common.

3.2. Figure of Merit for NE224

Since there is little reliable data for NE224 concern-

ing its decay modes and the relative contributions of these

modes, its theoretical figure of merit cannot be calculated.

The figure of merit versus energy was measured using an

identical cell and electronics
tion 3.1. The PSD spectra are
figures of merit are listed in
equivalent energy range of .62

of 1.23 was obtained. This is

following the method of Sec-
shown in Figure 12 and the
Table 6. For an electron

to 1.24 MeV a figure of merit

less than that of NE213, but

NE224 still gives reasonable pulse shape discrimination.

Less attenuation of the light collected can be tolerated for

NE224 since the figure of merit is less to start with. This

will be seen in Section 4.

 

O
(D

 

 

CT
'I.‘

 

 

O
(.Q

 

l 1 L j11

 

COUNTS/CHANNEL

 

Energg Range in MeV
0.31 INTEGRAL
0.31 Thru 0.82
0.82 Thru 1.2‘+
1.24 Thru 1.87
1.87 Thru 2.49
2.49 Thru 3.11
311 INTEGRAL

.ofioaoao
O O...

 

 

l L. L

 

CHANNEL NUMBER

Figure 12. NE22'-l P.S.D. spectre 55 mm X 50 mm cell.

 

35

Table 6. Figure of Merit (M) for a 45 mm x 50 mm NE224

Scintillator.

 

 

Energy Range [MeV] Figure of Merit

 

0.31 Integral 1.08
0.31 thru .62 1.12
0.62 thru 1.24 1.23
1.24 thru 1.87 1.35
1.87 thru 2.49 1.46
2.49 thru 3.11 1.62
3.11 Integral 1.57

 

4. LIGHT COLLECTION IN THE SCINTILLATION COUNTER

4.1. Effects of Light Attenuation on Figure of Merit

The figure of merit M defined by Equation 19 is inversly
proportional to the sum of the FWHM of the distributions for
the electron and proton induced scintillations. The FWHM of
the distributions are in turn equal to e/2.36. The standard
deviation of the distribution is, by Equation 12, inversly
proportional to the square root of the total photoelectron

number P. Therefore

(24) Mar/$9 .

It is then clearly desirable to make as many photo-
electrons as possible by collecting as much of the light pro-

duced by the scintillations as is possible.

4.2. Light Collection Problems with a Large Volume Neutron
Time-of—Flight Detector
The large neutron time-of—flight detector at Michigan
State University Cyclotron Laboratory consists of an 84 cm
by 13 cm x 2.5 cm (inside dimensions) cell constructed of
plexiglass G, filled with NE224. A light pipe placed at each

end couples to an RCA 8850 photomultiplier tube (Figure 13).

36

.HOuooumo ucwwam-mo-oefiu couosmc owpmfi < .mH ouowflm

=5 m . N . XE Ems

F\\ .0 mmmawflxoam mo poooouomcoo
.HmcuoucH mum mCmecoEHQ

 

 

\\\
\
Illl'll'"'|Ilnl'|l.l'llll'llll.lIIIIIII-IIIIIIII‘III

omwm<om

 

 

 

 

 

 

 

 

 

 

 

 

 

38

To estimate the light loss as the position viewed down
the cell is increased, the program listing in Appendix D
was used. It assumes an extinction coefficient for the
plexiglass of 0.005 cm-l. The excitation coefficient W=1/l,

where l is the attenuation length defined by the relation

(25) §= exp [- é] .
0

where I and I0 are the transmitted and incident intensities,
and A is the path length. The loss at each plexiglass-air
interface was assumed to be 5%, and the attenuation length
of the liquid to be 3.3 meters. The program traces rays
that are emitted at less than the critical angle of 410 and
counts their losses as they travel down the length of the
cell.

The fraction of the light transmitted versus the source
position viewed is listed in Table 7. It is evident that
the transmission is independent of X and Y but depends
strongly on Z. Over the whole length of the detector one
could expect to collect only 34% of the light emmitted within
the solid angle in which there is total internal reflection.
Since, in order to increase the timing resolution, it is nec-
essary to view the cell with a photomultiplier at each end,
the PSD signal can be obtained from the photomultiplier that
looks at a distance of one half the detector or less. In
this case there is a predicted attenuation of 50%. The
figure of merit M should then be decreased by a factor of

/T5. Thus, for NE224 the best figure of merit of 1.23 for

39
an equivalent electron energy of .62 to 1.24 MeV is now

reduced to 0.87. This is already a poor figure of merit.
The attenuation through the light pipe has to be taken into
account also.

To measure the attenuation through the light pipe a
1.1 cm x 2.5 cm x 12.7 cm scintillation cell was constructed
of plexiglass Gand filled with NE224. The scintillator was
then coupled to an Amperex 58AVP photomultiplier21 placed as
shown in Figure 14. The block diagram of the electronics
is shown in Figure 15. The photomultiplier was coupled to
an Ortec 269 tube base. The linear dynode signal was ampli-
fied by an Ortec 113 preamplifier and further amplified and
shaped by an Ortec 460 delay line amplifier. The singly
differentiated output pulse was fed to a Canberra Model 8100
Multichannel analyzer. ,

The high voltage supply for the photomultiplier was set
at 2240 volts. The coarse gain on the Ortec 460 was set at
50 with a fine gain of 1. Shaping time was set at 0.04 usecs.

228Th

A spectrum was collected for the Compton edge of a
source collimated to irradiate the center 5/8 inches of the
scintillator. The spectrum is shown in Figure 16a. The
scintillator was then coupled to the light pipe and the light
pipe coupled to the center of the 58AVP photomultiplier. It
was necessary to readjust the gain of the Ortec 460 delay
line amplifier to a coarse gain of 200. All other settings
were kept unchanged. A new spectrum was collected of the

228Th Compton edge while irradiating the entire scintillator.

40

Table 7. Theoretical Light Transmission, in percent, through
a Rectangular Scintillator of Dimension 2.7 cm x
83.8 cm x 12.8 cm. The source positions x, Y and
Z are Defined in Figure 13.

 

 

 

 

 

 

 

 

 

 

Z = 83.8 cm
Y = 0.0 cm .33 cm .67 cm 1.0 cm
X = 0.0 cm 34.0 34.0 34.0 34.0
1.6 cm 34.0 34.0 34.0 34.0
3.2 cm 33.9 34.0 34.0 34.0
4.8 cm 33.9 33.9 33.9 33.9
Z = 62.9 cm
Y = 0.0 cm .33 cm .67 cm 1.0 cm
X = 0.0 cm 40.7 40.7 40.7 40.7
1.6 cm 40.7 40.7 40.7 40.7
3.2 cm 40.7 40.7 40.6 40.7
4.8 cm 40.6 40.6 40.6 4.06
z = 41.9 cm a_
Y = 0.0 cm .33 cm .67 cm 1.0 cm
X = 0.0 cm 49.3 49.2 49.2 49.3
1.6 cm 49.2 49.2 49.2 49.2
3.2 cm 49.2 49.2 49.2 49.2
4.8 cm 49.1 49.1 49.1 49.1
Z = 21.0 cm
Y = 0.0 cm .33 cm .67 cm 1.0 cm
X = 0.0 cm 60.4 60.4 60.4 60.4
1.6 cm 60.4 60.4 60.4 60.3
3.2 cm 60.3 60.3 60.3 60.3
4.8 cm 60.2 60.2 60.2 ' 60.1

 

41

.Haoo cowumaawucwum mo ucoEmomHm .qH muswwm

 

 

 

m><mm
xmpoda<

 

 

 

 

sNNmz nuns emLHHa
Lame coaumhhuucaum

 

 

 

 

 

42

coaumscouum mead unwfla wcflusmmme pom oHconooon

 

oag
.A.Q.Q
cos omuuo

 

 

 

 

 

 

 

 

oedema
mHH oouuo

AD

 

<02
ooam Ho

 

 

 

 

 

 

ococ>m

 

comm mosh
mom omuuo

 

 

.mL muawua

m><wm
xmuoda<

 

 

 

CDU°HCZU°HHHCULJOH

43

“by mafia ufiwfia Lows can .Amv mafia unwwa unonuHB muuoodm owwo coudEoo

 

.GH.mpsta

 

44

This was done so that the entire light pipe would be utilized.
However, the active area of the photomultiplier's photo-
cathode utilized remained the same due to the reduction in
area by the light pipe. This ensured that any change in the
channel number of the Compton edge would not be due to the
differing sensitivity of the center and the outer edge of the
photocathode. The spectrum is shown in Figure 16b. Notice
that the Compton edge has deteriorated due to the reduction
in light collected.

Using the gain settings and channel numbers of the 228Th
Compton edge, an attenuation through the light pipe of approx-
imately a factor of 6.3 is obtained. This would reduce the
figure of merit by a factor of 2.5. This, coupled with the
attenuation through the original neutron time-of-flight
detector, would give a figure of merit of .35 for an equiv-
alent electron energy range of .62 to 1.24 MeV for scintil-
lations at the center of the detector. It is then clear that

the light pipe, in this particular case, is the major source

of degradation of the figure of merit.

5. DEOXYGENATION OF SCINTILLATION COUNTERS

5.1. Method of Deoxygenation

It is necessary to remove as much dissolved oxygen from
the liquid scintillator as is possible, since the presence
of oxygen leads to quenching of the desired triplet states.
The method employed by this author was to place a long, thin
teflon tube down to the base of the scintillation cell and
bubble helium gas up through the liquid scintillator. The
bubble stream mixes the liquid and allows for greater effi-
ciency in removing the dissolved oxygen. It is also necessary
to provide a means for the helium gas to escape. The excess
helium was allowed to pass through a narrow constriction.
This was done to prevent backstreaming of oxygen where it
could interact with the surface of the scintillator liquid.
Helium was chosen since this is an inert gas that does not

quench the triplet states.

5.2. Diffusion Model
If the helium was not bubbled through the scintillator,
but only allowed to interact with the open surface area, the
removal of oxygen would require a much longer time period.
To show this, assume we have an L=1 meter long, cyclin-

drical cell with one end open. If the scintillator has

45

46

an initial concentration of oxygen of K, and one end is
then placed in contact with an oxygen free environment, the
time for 95% of the oxygen to diffuse out of solution can
be estimated by the following analysis: For diffusion in
one direction only, ensured by the glass walls, the rate of
flow of oxygen dQ/dt, through an area dydz, will depend on
the rate of change of concentration dC/dx, times the area
normal to x dydz, times a constant coefficient A, which has
2t-l

units of cm . Since dQ/dt is in the direction of decreas-

ing concentration, we have

dQ _ _ g9
(26) dt - A[dx]dydz ,

or for a rate of change of concentration

2
3x

With one end of the cell closed and the other exposed to a

zero concentration of oxygen we have the initial conditions

3C(L,t)

3x = 0, C(x,0) = K.

(28) C(O.t) = 0,
Separation of variables leads to

(29) X"(x)+lX(x)=0, X'(L)=0, X(x)=K, X(0)=0 and

(30) T'(t)= -lAT(t), where C(x,t)=X(x)T(t).

47

For Equation 29 we have a series of solution
(31) xn=An31n(x/I) + Bncos(x/I).
For the boundry condition X'(L)=0 we have
(32) x'(L)=An/I cos(L/I) - Bn/I sin(L/I)=o,

and for X(0)=0

2 2
_ _ . _ 2n-l 1

Thus,

—- 5132.1.
(34) Xn-An51n [ L 2 ]] .

For Equation 30 we then get the solution

2n-l 2 n 2
(35) Tn(t)=Cnexp - [2 ] A [f] t ,

 

which leads to a solution of Equation 27 of the form

 

 

m 2 2
(36) C(x,t)= Z Anexp[- 1T gt [23-1] ] sin i—flfg'lfl .
N=l L

where

 

(37) C(x,0)=K= 2 A sin £11- [Zn-l] .
n=l n

48

Solving for An we get

 

 

 

 

F
x #514?" 23"1 dx
(38) An- L 2xn. n-l °
f sin [£— 2 ]dX

 

Integrating yields

_ 4K
‘39) An“ FTEEZIT °

The final solution is then

2 2
m . x1 2n-l _ K At 2n-1
3.13 811') [L [—2 ]] exp 2 [2 ]
n

 

 

 

 

 

n=l (Zn-1)
nzAt 2n-l
[LC(x,t)dx 8 w exp ' 2 [if‘]
-; o = _§ L
. N n=1 (2n—1Y'

Solving for C(t=?)=.05K we can neglect n=2,3... terms since
they are much less than the n=l term for the above condition.

Thus,

 

2

4L 3 l
(42) t= ——— log .
An2 9 [n2(o.05)J

The constant of diffusivity A for NE213 is assumed to be on

the order of that for cyclohexane which has a diffusivity for

oxygen at 20°C of 4 x 10-5 cmzsec-l.22 This gives us the

time t, at which 95% of the oxygen is removed, of 2.8 x 108

seconds, or approximately 9 years.

49
It is therefore imperative that the inert gas be bubbled
through the liquid to provide thorough mixing. In this way
a large gradient of concentration exists at the gas-liquid

interface speeding the removal of oxygen.

6. A NEUTRON TIME-OF-FLIGHT DETECTOR WITH IMPROVED PULSE

SHAPE DISCRIMINATION

6.1. The Detector

A 45 mm diameter pyrex glass cylinder 1.24 meters long
was constructed as shown in Figure E1. The windows on the
end are flat pyrex. An RCA 8850 photomultiplier was coupled
to one end. The 45 mm diameter of the cell made it unneces-
sary to use a light pipe to couple the photomultiplier to
'the cell. This improves the light collection efficiency.
The attenuation of light down the detector while it contained
NE213 was measured by the methods described in Appendix E.
The attenuation length for the detector filled with NE213
was equal to 0.95 r .01 meters. Thus, the attenuation A for
the detector as a function of the position x of the scintil-

lation from the photomultiplier can be expressed
(43) A=1-exp[- -—’£—] .
0.95

If one were willing to accept a reduction in the figure of
merit of a factor of 1.4, then the farthest a scintillation
could occur from the photomultiplier would be x=.66 meters.
This is over half the length of the detector. Since, for

neutron time-of—flight studies, one uses a photomultiplier

50

51

at each end for reasons of time resolution, a photomultiplier

is always close enough for good pulse shape discrimination.

6.2. Performance
The pulse shape discrimination characteristics of the
detector were measured for both NE213 and NE224 as the active

scintillator.

6.2.1. Detector Filled with NE213

A test of the figure of merit obtainable from the scin-
tillator for NE213 was made for a source position near the
photomultiplier and for a position midway down the detector.
This was accomplished by placing the appropriate source at
the two positions along the detector. The figure of merit
at these two positions were obtained as a function of electron
equivalent energy deposited in the scintillator.

The electronics used are those in Figure 3. The data
were collected on the PDP 11/45. The same procedures were
used to tune the electronics as earlier. First a spectrum

60

of the 1.33 MeV Co Compton edge was taken for a position

60Co source was replaced

near the photomultiplier. Then the
with a PuBe source and a spectrum taken of the pulse shape
discrimination signal, gated as a function of energy. The
results are tabulated in Table 8. Representative plots of
the data are in Figure 17. The procedure was then repeated

for a position at the center of the detector. The results

are given in Table 9 and spectra shown in Figure 18.

52

Table 8. Figure of Merit (M) for a 1.24 m x 45 mm NE213
Scintillator with Source at Near End of Scintil-

 

 

 

lator.
Energy Range [MeV] Figure of Merit
1.1 thru 5.1 1.2
3.6 thru 5.1 1.3
2.4 thru 5.1 1.0
1.7 thru 2.4 1.3
1.1 thru 1.7 1.2
.62 thru 1.1 1.0

.25 thru .38 .85

 

COUNTS/CHANNEL

53

 

 

 

 
        

 

   

l I l l
CHANNEL NUMBER

 

;

    

1 l I

Energg Range in MeV

(D'NCDQOU'O
0.0....

1.12 In+egrol

3.80 Integral

2.38 Thru 3.80
1.77 Thru 2.38
1.12 Thru 1.73
0.82 Thru 1.12
0.25 Thru 0.38

Figure l7. P.S.D. for NE2|3 cell I.24rn, near end.

 

54

Table 9. Figure of Merit (M) for a 1.24 m x 45 m NE213
Scintillator with Source at Center of Scintil-

 

 

 

lator.
Energy Range [MeV] Figure of Merit
1.02 Integral .98
3.29 Integral 1.00
2.16 thru 3.29 .99
1.58 thru 2.16 1.0
1.02 thru 1.58 .95
.56 thru 1.02 .90

.45 thru .67 .80

 

COUNTS/CHANNEL

55

 

 

 

 

1 l 1

 

 

 

1 I 1 l l l
CHANNEL NUMBER

Figure l8. P.S.D. for NE2I3 cell

 

Energg Range in MeV

""(DQOO'O
'0..-

1.02 Imegral

3. 28 Imegral

2.18 Thru 3.29
1.58 Thru 2.18
1.02 Thru 1.58
0.58 Thru 1.02
0.45 Thru 0.87

l.24 m , middle.

 

56

When the Compton edge channel numbers for the two posi-
tions were used to evaluate the attenuation one finds that
the figure of merit should deteriorate in the central posi-
tion by a factor of 1.37. For an electron equivalent energy
range of 3.6 to 4.4 MeV the observed degradation of the
figure of merit was a factor of 1.35. It therefore appears
that the reduction in the figure of merit is due mainly to
light loss and not time spread due to multiple paths of
transit of the light. If this were the case one would expect
the figure of merit to be worse than that predicted by light

attenuation alone.

6.2.2. Detector refilled with NE224

The detector was emptied of the NE213, refilled NE224,
which was then deoxygenated. The pulse shape discrimination
performance of the detector was evaluated using the same
procedure. The results are tabulated in Tables 10 and 11.
Typical spectra are shown in Figures 19 and 20. As before,
the PSD quality of NE224 is inferior to that of NE213.
However, the ability of NE224 to be contained in a plexiglass
container makes it useful in some instances where the pulse
shape discrimination characteristics are not of prime impor-

tance.

6.3. Comments
As shown in Appendix E, the light transmission down a
detector decreases as the diameter of the detector decreases.

Therefore, if a narrow detector is made to achieve high time

57
resolution, the figure of merit deteriorates due to the

increased light loss.

58

Table 10. Figure of Merit for a 1.24 m x 45 mm NE224
Scintillator with Source at Near End of

 

 

 

Scintillator.
Energy Range (M) Figure of Merit
1.14 thru 5.07 .98
3.65 thru 5.07 1.15
2.39 thru 3.65 1.10
1.76 thru 2.39 1.08
1.14 thru 1.76 1.01
.63 thru .94 .97

.25 thru .38 .85

 

COUNTS/CHANNEL

59

 

-

b

b

T

[CUM __ [e]

I r I r j F

 

[bl __ if]
-L

 

“F

.1-

 

 

 

[c] A

-1:—

Energg Range in MeV
1.1‘+ Integral

3. 85 Integral
2.38 Thru 3.85
1.78 Thru 2.38
1.1'-l Thru 1.78
0.83 Thru 0.84
0.25 Thru 0.38

«D ;“ (D 51 C) ST C)

 

 

CHANNELl NUMBER

Figure 19. P.S.D. for NE22Ll cell 12% m. near end.

 

60

Table 11. Figure of Merit for 1.24 m x 45 mm Scintillator
filled with NE224 with Source at Center of

 

 

 

Scintillator.
Energy Range (M) Figure of Merit
1.12 Integral .85
3.57 Integral .93
2.34 thru 3.57 .93
1.73 thru 2.34 .92
1.12 thru 1.73 .88
.61 thru .91 .86

.46 thru .69 .82

 

61

 

 

I

COUNTS/CHANNEL

 

 

1 l

CHANNEL

P.8.D. for NE22L+ cell 1.2% m. middle.

Figure 20.

-—

(0"‘(0000'0
o'ooolo

 

 

 

1 l 1 l

 

Energg Range in MeV
1.12 Integral
3.57 Integral
2.3‘+ Thru 3.53
1.73 Thru 2.3”:
1.12 Thru 1.73
0.81 Thru 0.81
0.'-+8 Thru 0.88

7. Photomultiplier Problems

7.1. Gas Ionization

A photomultiplier, after having produced photoelectrons
in response to incoming photons, may produce an after-pulse.23
The photomultiplier contains some residual gas and as the
first group of electrons is emitted from the photocathode
and accelerated towards the first dynode the electrons ionize
the residual gas. These positive ions are then accelerated
back towards the photocathode, whereupon they liberate more
electrons. This after-pulse occurs, of course, after the
main pulse. The time at which it occurs depends on the kind
of residual gas and the accelerating potential. If the after

pulses occur soon enough they can interfere with the desired

signal.

7.2. After-Pulses in the Present System

Afterpulses were observed in taking data with the small
detector, shown in Figure 8, while it was filled with NE213.
These after-pulses occurred typically 700 nsecs later. The
anode pulse with its associated after-pulse is shown in
Figure 21a. Also shown in Figures 21b and 21c are the dynode
pulses after delay line shaping, and double delay line
shaping. Since the zero-crossover method of pulse shape

discrimination depends on the time the double delay line

62

700
nsec

0. ~— a

1,—
5—After Pulse

 

\ Main Pulse

 

l usec |‘_
700

b. m »

 

FJT

C. r

 

 

 

Figure 21. After-pulses in anode signal (a), singly differ-
entiated dynode signal (b), and doubly differentiated dynode
signal (c) with l usec delay line shaping.

64

shaped pulse crosses zero, it is clear that the after-pulse
would affect the pulse shape discrimination ability of the
system. One method of eliminating afterpulses is to age
the tube as suggested by Kuchnir and Lynch.2“

The method employed here was to process the signal
before the after pulse occurred. This was accomplished by
replacing the 1 usec delay lines in the Canberra delay line
amplifier with 400 nsecs delay lines. The effect of this can
be seen by noting the delay line and double delay line shaped
signals in Figures 22a and b.

The effect on the figure of merit was measured by first
accumulating the PSD spectra for NE213 in the small cell of
Figure 8, using the electronics in Figure 3. The delay line
amplifier was unmodified. This was done for a discrimination
level of .5 to 5.5 MeV equivalent electron energy. The delay
line amplifier was then replaced with a modified delay line
amplifier and a new spectrum taken. The results are shown
in Table 12. The observed pulse shape spectra are shown in
Figures 23a and b respectively. The increase of figure of
merit from .95 to 1.28 represents a significant increase.

To achieve the same goal by increased light collection would
necessitate an increase of photons collected by a factor

of 1.8.

65

 

 

 

 

V 700
nsecs ->

400
nsecs *—-

b’ GUI/i 7‘)

 

 

 

 

 

Figure 22. After-pulses in singly differentiated dynode
signal (a) and doubly differentiated dynode signal (b)
with 400 nsecs delay line shaping.

66

muuomdm amm

.wcfidmnm mafia mmaoo cannon moomc.ooq LDMB
Amv wcfldmzm mafia >mamo cannon com: H SDH3 muuomdm 9mm .mN gunman

 

Table 12.

67

Figure of Merit for 1.0 usec Double Delay Line
Shaping and 400 nsecs Double Delay Line Shaping
with PuBe Source.

 

 

Discrimination Level at 500 Kev

 

1.0 nsec Shaping 400 nsecs Shaping

M = 0.95 M = 1.28

 

8. CONCLUSION

Building a neutron time-of-flight detector having good
pulse shape discrimination requires that attention be paid
to optimizing the collection of the light from the radiation
induced scintillation. This is done by maintaining the
largest diameter one can tolerate in the detector while
still maintaining good timing resolution. Also, elimination
of light pipes, or minimizing the reduction in area through
the light pipe, is of major benefit in eliminating light
loss. Light loss through attenuation in the scintillator
itself is of minor importance except for very large detectors.

Deoxygenation of the scintillator must be done with care
to prevent recontamination with atmospheric oxygen. It has
been shown that in bubbling the inert gas through the scintil-
lator it is necessary to provide a bubble stream so as to
mix the scintillator and thus reduce the time necessary to
deoxygenate the scintillator.

After-pulses can be a major source of degradation in
the figure of merit and can easily be eliminated as a source
of problems by simply changing the double delay line shaping
time.

Amplifier noise should be reduced so as to provide good
pulse shape discrimination at the low energy end of the
range detected.

68

69

By considering these factors a neutron time-of-flight
detector can be built with acceptable neutron/gamma-ray
pulse shape discrimination.. NE213 appears to be the optimal
choice for detector geometries not requiring construction

with plexiglass.

APPENDICES

APPENDIX A

‘ COMPUTER PROGRAM FOR CALCULATING E. D and e

APPENDIX A
COMPUTER PROGRAM FOR CALCULATING E, D and 8

DO 2 R=500.00,6000.0,500.0

N=R/2

AF=.8

AS=.2

TF=4.0

TS=25.0

CALL PSD(AF,AS,TF,TS,R,N,DVE,TME)
AF=.5

AS=.5

TS=47.0

CALL PSD(AF,AS,TF,TS,R,N,DVP,TMP)
D=(TMP-TME)/SQRT(DVP**2+DVE**2)
PRINT 1,D

FORMAT (' D=',F6.2)

CONTINUE

END

70

71

SUBROUTINE PSD(AF,AS,TF,TS,R,N,DV,TM)

WN=0

T=0

TM=0

TM2=0

T=T+.l

EXF=0

IF((T/TF).LE.170.0) EXF=EXP(-T/TF)

EXS=0 —

IF((T/TS).LE.170.0) EXS=EXP(-T/TS)
F=AF*(l-EXF)+AS*(1-EXS
FDT=AF/TF*EXF+AS/TS*EXS

AL=0
IF((F.NE.0.0).AND.(F.NE.l.0).AND.(FDT.NE.0.0))
lAL=(RrN)*ALOG(1.0-F)+(N-l)*ALOG(F)+ALOG(FDT)+R*.69
PT=0

IF((AL.GE.-170.0).AND.(AL.NE.0.0)) PT=EXP(AL)
WN=WN+PT*0.5

TM=TM+T*PT*0.5

TM2=TM2+T**2*PT*0.5

IF(T.LE.1000) GO TO 2

TM=TM/WN

TM2=TM2/WN

DV=SQRT(TM2-TM**2)

PRINT 3,DV,TM

FORMAT (' STD DEV=‘,F6.2,' T=',F6.2)

RETURN

END

APPENDIX B

COMPUTER PROGRAM FOR GENERATING PULSE SHAPE SIGNATURES

APPENDIX B

COMPUTER PROGRAM FOR GENERATING PULSE SHAPE SIGNATURES
DIMENSION DY(256),DX(256)

DO 4 K=l,7
R=25.*l.9307**K
N:R/2

YMAX=O.

DO 1 I=1,256
T=2.+6./255.*(I-l)
DX(I)=T
DYlI):PT(.8,.2,4.,25.,R,N,T)+PT(.5,.S,4.,47.,H,N,T)
1 IF(DY(I).GE.YMAX) YMAX=DI(I)
DO 2 1:1,256
DY(I)=DY(I)/YMAX
UNIT 106 IS A FILE FOR TEMPORARY STORAGE OF PSD SIGNATURE
UNIT 106 IS THEN CALLED BY DRAFTKESIST TO BE PLOTTED
HRITE(106,3) DX(I),DY(I)
FORMAT(1OX,2F10.3)
HRITE(106,5)
FORMAT('END')
END

0'1ch

72

73

APPENDIX B

FUNCTION PT(AF,AS,TF,TS,R,N,T)
no 315:0.0
so IF((T/TF).LE.170.0) EXF:EXP(-TITF)
60 313:0
70 IF ((T/TS).LE.170.0) EXS=EXP(-T/TS)
80 F=AF*(1-EXF) +AS‘(1-EXS)
90 FDT:AF/TF*EXF+AS/IS*EXS
100 AL=O
110 IF((F.NE.0.0).AND.(F.NE.1.0).AND.(FDT.NE.0.0))
120 1AL:(R-N)*ALOG(1.0-F)+(N-1)‘ALOG(F)+ALOG(FDT)+R*0.69

130 PT=O
tuo IF((AL.GE.-170.0).AND.(AL.NE.0.0)) PT=EXP(AL)
RETURN

END

APPENDIX C

COMPUTER PROGRAM FOR CALCULATING FIGURE OF MERIT D AS A
FUNCTION OF PULSE HEIGHT FRACTION K

APPENDIX C

COMPUTER PROGRAM FOR CALCULATING FIGURE OF MERIT D AS A
FUNTION OF PULSE HEIGHT FRACTION K

—.

3:300

DO 2 N=20,250,20

AF=.8

AS=.2

TF:4.0

TS:25.0

CALL PSD(AF,AS,TE,TS,R,N,DVE,TME)
AF=.5

AS=.S

TS:N7.0

CALL PSD(AF,AS,TF,TS,R,N,DVP,TMP)
D:(IMP-TME)/SQRT(DVP**2+DVE‘*2)
PRINT 1,D

FORMAT (' =',F6.2)

CONTINUE

END

74

75

APPENDIX C

SUBROUTINE PSD(AF,AS,TF,TS,R,N,DV,TM)
NN=0
T=0
TM=0
TM2=0
T=T+0.1
EXF=O
1F ((T/TF).LE.170.0? EXF=EXP(-T/TF)
EX$=O
IF ((T/TS).LE.170.0) EXS=EXP(-T/TS)
F=AF*(i-EXF)+AS*(1-EXS)
FDT:AF/TF'EXF+AS/TS'EXS
AL=0
IF ((E.NE.0.0).AND.(F.NE.1.0).AND.(FDT.NE.0.0))
1AL=(R-N)*ALOG(1.0-F)+(N-1)‘ALOG(F)+ALOU(FDT)+R'0.69
PT=O
IF((AL.GE.-l70.0).AND.(AL.NE.0.0)) PT=EXPiAL)
HN=NN+PT'0.5
TM=TM+T*PT*O.5
TM2=TM2+T**2'PT*O.5
IF (T.LE.1000) GO TO 2
TMzTfi/NN
TM2=TM2/HN
DV:SQRT(TM2-TM*'2)
PRINT 3,DV,TM
FORMAT (' STD DEV=',Fb.2,’ T=',F6.2)
RETURN
END

APPENDIX D

COMPUTER PROGRAM FOR CALCULATING LIFHT ATTENUATION IN
A RECTANGULAR DETECTOR

APPENDIX D

COMPUTER PROGRAM FOR CALCULATING LIGHT ATTENUATION IN A
RECTANGULAR DETECTOR

COMMON X0,IO,ZO,THETA,PHI,L,N,H,FRAC,D,EXTNK,SCAT
REAL‘H X0,YO,ZO,THETA,PHI,L,N,h,PRAC,D,EXTNK,TRNS(10)
DATA PI/3.1Q159265N/
N=2.67
h=12.83
L=63.83
0:0.1555
SCAT=.95
EATNK=0.005
THI=PI/25.0
PHII:H9.0/50.0/150.0'PI
D0 7 J3=1tu
PRINT 5
5 FORMAT (' ')
DO 4 J1=1,4
DO 6 J2=1,4
X=H/d.0‘(Jl-l)
I:h/U.O'(J2-l)
Z=L/4.0'(J3-l)
PHI=41.0/100.0'PI
TD=0.0
D0 2 1:1,50
TH:0.0
DO 1 J=1,50
X0=X
10:!
20:2
THETAzTH
CALL LFRAC
TD:TD+FRAC'5.61/4.0/P1'0031Ph1)‘Thl'Phll'EXP(-0.003'L/51N1Ph1)l
TH:TR+THI
PHI=PHI+PHII
TD=TD*100.0
TRNS(J2)=TD
PRINT 3,(TRNS(Ii),11=1,u)
FORMAT (' ', “57.2)
PRINT 6
FORMAT(' ')
STOP
END

N—O

“'44.” 3:0"

76

U‘lgLUN

\DO‘

11

77

APPENDIX D

SUBROUTINE LFRAC

COMMON X0,YO,ZO,TRETA,PHI,L,h,H,FRAC,D,EXTNK,SCAT
REAL‘H X0,10,ZO,ThETA,PhI,L,H,H,bRAC,D,EXTNK,PI,h1,h1,Ri,82

DATA PI/3.lu1592054/
PHI=PI/2.0-PHI

FRAC:1.0

hl=H/2.0

N1=N/2.0

ST=SIN(THETA)

CT=COS(THETA)

SP=SIN(PH1)

IF (CT‘SP) 29593

hl=-h/2.0

IP (ST'SP) u,1l,5

hlz-hl

R1:(Hl-XO)/(SP'CT)
R2:(N1-YO)/(SP'ST)
IP(R1.GT.R2) GO TO 6
XO=H1
IO=R1'SP*ST+IO
ZO:ZO+R1*COS(PHI)
THETA:PI-THETA
IF (THETA.LT.0.0) ThETA=THETA+2*P1
IP(ZO.GE.L) GO TO 12
FRAC=PRAC'EXP(-2.0'D/ABS(SP'CT)'EXTNK)'SCAT
GO TO 7
XO=R2*SP*CT+XO
IO=N1
ZO=ZO+R2*COS(PHI)
THETA=2.0*PI-THETA
IF(ZO.GE.L) GO TO 12
PRAC=FRAC*EXP(-2.0‘D/ABS(SP*ST)'EXTNK)'SCAT
IF (ZO.GE.L) GO TO 12
GO TO 1
IF (ST'SP) 9.10.10
le-Nl
Ri=(N1-YO)/(ST*SP)
XO=0.0
IO=N1
ZO:ZO+R1*COS(PHI)
THETA=2.0'PI-THETA
IP(ZO.GE.L) GO TO 12
FRAC=FRAC*EXP(-2.0*D/ABS(SP'ST)‘EXTNK)'SCAT
GO TO 7
R1=(hl-XO)/(CT'SP)
10:0.0
XO:H1
ZO=ZO+R1'COS(PHI)
THETAzPI-THETA
IF (THETA.LT.0.0) ThETA=ThETA+2.0*PI
IF(ZO.GE.L) GO TO 12

12

78
APPENDIX D

FRAC:FRAC*EXP(-2.0*D/AES(SP*CT)'EXTNKl'SCAT
GO TO?

PHI=PI/2.0-Ph1

RETURN

END

APPENDIX E

.\

LIGHT ATTENUATION IN LIQUID ORGANIC SCINTILLATORS

LIGHT ATTENUATION IN LIQUID ORGANIC SCINTILLATORS

John E. Yurkon and Aaron Galonsky

Cyclotron Laboratory, Michigan State University
East Lansing, Michigan, USA

Abstract

Measurements of the attenuation length of monochromatic
light in NE 213 and NE 224 and of the wavelength-averaged
attenuation length in NE 213 are given in order to provide
the designer Of large-volume detectors with information
necessary in deciding how large a detector can be construc-
cted without excessive loss Of light. In particular, the
attenuation length of scintillation light in NE 213 is found
to be 2.16 i 0.24 m.

*
This material is based upon work supported by the National

Science Foundation under Grant No. Phy-7822696.

El. INTRODUCTION

The timing and neutron/gamma-ray pulse shape discrim-
ination qualities of time-Of-flight detectors depend in
part on the collection efficiency Of the light produced
in a scintillation. Therefore, in building a large-volume
detector it is important to know the light attenuation
length ligiven by

(l) I/Io = exp(%%)
where I0 is the incident light intensity, I the transmitted
light intensity, and d the length of the path traveled.

Two different methods of measuring the light attenuation

lengths of the liquid scintillatorsza NE 213 and NE 224

will be discussed.

79

E2. METHODS OF MEASUREMENT
E2.1. Direct

A collimated beam of light was directed down a 45 mm
0.0. cylindrical Pyrex cell (Fig.El) filled with the liquid
scintillator. Placed inside Of the cell was a mirror at-
tached to a teflon cylinder. The teflon cylinder contained
a piece of iron so that it could be moved along the length
of the cell with an external magnet. The reflected light
intensity was measured with a light meter. A photograph
of the apparatus is shown in Fig. E2.

The reflected light intensities were measured for
various positions Of the mirror. Since the number Of inter-
faces remained constant, the resultant plot Of intensity
versus path length in the liquid should reflect only the
attenuation by the liquid scintillator.

The wavelength of maximum emission for NE 213 and
NE 224 is 425Q,A.2“ Therefore, it is desirable to measure
the light attenuation length near this wavelength. The
4416 2 line of a helium-cadmium laser was used as the closest
available wavelength. TO see if there is a wavelength
dependence on the attenuation length, the 4689 A line of
a krypton laser was also used.

The data for NE 213 and NE 224 are shown in Figs. E3
and E4, respectively. The measured attenuation lengths ob-

tained from a least-squares fit are listed in Table El.

80

81

mmemw x®u>m mo Umuosuumcoo

St m as N.o H m.H 35 w. + o.mm

SE N CE N.o H o.N :E_H w me

w m m

mc0wrcohflc sud; HHCO MODOHHHDCHOM mo waflmumo .HN Ousoflm

 

 

 

E UN;

 

 

88

O

 

 

 

TLr

\.
8
'1

e "mmwzxoie .34;

.msuwm “5an can HHUU Houmaafiucwom mo nmmuwouonm .NW 925%;

E

 

 

83

 

 

 

 

U)
F- i i I I r I i '
i—l . _
Z ‘10” A MEASURED AT 4889 ANGSTROMS
:3 x MEASURED AT 4H18 ANGSTROMS
>.
m "l
<< 30‘
a:
f.—
H
03
a:
< d
\—I 20...
>-
P—
H
(D
2
L1J
F—
2:
H
t- ._
(D .. ..
H
_l ,. ..
O b u:
LLJ
I..—
l— _ -
H
5)
2 5'
< l 1 l L I 1 l 1
0C 0 80 120 180
'— PATH LENGTH IN CM.
Figure E3. Transmitted light intensity vs. path length

for NE213.

TRANSMITTED LIGHT INTENSITY (ARBITRARY UNITS]

84

 

Lio-

 

] 1

 

l l l I T l

A MEASURED AT H888 ANGSTROMS ‘
x MEASURED AT 4413 ANGSTROMS

 

l l

 

Figure E4.
for NE224.

l l l
80 120 180

PATH LENGTH IN CM.

Transmitted light intensity vs. path length

85

Table I. Attenuation lengths in meters as measured with
laser light.

 

 

4416 A 4689 X

 

NE 213 1.85 i 0.07 3.02 i 0.10

NE 224 1.22 t 0.03 1.40 0.03

H-

 

86
As can be seen, the attenuation length is shorter for the
o
wavelength closer to 4250 A. Furthermore, NE 213 exhibits

a stronger wavelength dependence than NE 224.

E2.2. Measurement with Radiation-Induced Scintillations

A measurement of the wavelength-averaged attenuation
length for NE 213 was made as a comparison to the laser
method. The spectrum of radiation-induced scintillations
in NE 213 does not consist Of a single line. It has a
finite width, probably on the order of a few hundred Ang-
stroms. The attenuation length Of NE 213 is wavelength
dependent, as shown in Table El. Thus, a filtering Of the
induced scintillations takes place with the result that
the rate of attenuation decreases as the more rapidly at-
tenuated wavelengths are removed. The wavelength-averaged
attenuation length is obtained by picking a region of the
data where the attenuation appears to be exponential.

The mirror was removed, and an RCA 8850 photomultiplier
was attached to one end of the 45 mm O.D. cell. The cell
was wrapped loosely with black paper so that the light was
collected only by total internal reflection. A collimated
Th-228 source was placed along the side Of the cell as shown
in Fig. ES. A block diagram of the electronics is shown
in Fig. E6. The channel number of the Compton edge of the
2.62-MeV y-ray was measured for various positions Of the

source .

.ennmeme
cofluosccuum oumouu>oinumcwam>o3= ecu mo mucwaousmama now moumm OCH .mm musmflh

QNWMY.
O

a ..m_x..mx..m ea lie—ET) en. _

m ..v_ - _T 1—

 

 

 

 

 

Ommm

 

 

 

 

. .qul Aunv
." ma m2 i 1
la .

 

 

88

.ezumCUH :oflumscmuuc Ummmum>mi£pmcmao>n3=
on» mo mucofieudmmmE How OUflcouuomac Una mo Emumoao xOOHm .mm musmflm

595424 5.13.24 ”12:
Elem: mmhae Eda ”1.538

 

 

 

EESEEAE

 

 

 

 

 

00¢
1H0 omhmo

 

 

 

m:

 

Duhmo

 

mm<m mmDH

 

m;m mz

 

 

00mm
40m

 

mmN

 

 

owhmo mooz>o

 

 

89

The resulting plot of intensity versus position re-
flects not only the attentuation in the liquid scintillator,
but also attentuation due to scattering at the liquid/glass
interface and absorption within the glass. We wish to know
the attenuation in the liquid scintillator alone. By reducing
the diameter of the cell, the path traveled through the
liquid stays the same, but the number Of reflections is
increased. Therefore, by the following analysis we can
determine the attenuation length Of the liquid scintillator:

The transmission within the scintillator itself may

be written as

.. :91
(2) I/Io - exp(Az)
where 11 is the attenuation length in the liquid scintillator.

Then the transmission of light down the cell, including
losses at reflection, is
- :91 n
(3) I/Io - exp(A£)R ,
where n is the number Of reflections and R is the fraction

of the light reflected at each reflection. For a ray travel-

ing at a given angle 6 from the z axis of the cylinder

._Ji_
cosG

(5) n =[z Sane]

where D is the diameter of the tube and [x] is the greatest-

(4) d

integer function. For n large the approximation of equation

5 by

(6) n _ z Bane

90

is sufficient. Then using equations 4 and 6, Rn can be

expressed as

(7) Rn _ exp( CdDSlne)’
where C = -ln R. By letting
_ D

(8) Ag 7 C sine

equation 3 can be written in the more tractable form

A

d d
(9) I/I = exp(- -— - -)
o g kg

where 19 represents losses other than those in the liquid,
e.g., losses in the glass. We can define a total attenuation
length A by A W L

(10) i = (l_.+ l

__) .
A 11 Ag

For different diameters (D1,D2) of cells, equation 8 gives

the relation

 

 

D A (D )
= l g 2
(ll) 19(D1) D2 .
Of course, Al is the same in each cell. Then using equations
10 and 11 to eliminate 19 we finally obtain for 12:
D2
AiDZ) (5- - l)
(12) A2, = 1 I
D2 1(027
(-— - -—-——0
D1 1(Dl)

where 1(92) and 1(Dl) are the total attenuation lengths
for cells Of the two diameters, D2 and 01'

A 22 mm O.D. cell, otherwise comparable to the 45 mm
O.D. cell (Fig. BIL was used to collect data following
the same procedure as for the 45 mm O.D. cell. The data

for both cells are shown in Fig. E7.

Figure E7. Compton edge channel vs. path length for NE213 in
the unpainted 45 mm and 22 mm cell (darkened line indicated.
region over which the least squared fit was performed).

 

 

91

 

H00“ “

300'

200

I

CHANNEL 0F COMPTON EDGE

100 r-

 

  

1

LiSMM. O.D..

 

 

 

50

Figure E7.

1
30

l
80

SOURCE POSITION IN CM.

92

Applying a least-squares fit to the data in Fig. E7
gives us the total attenuation lengths )(D1) and 1(D2) as
listed in Table E2. These are measured with respect to
the z axis. However, not all of the light travels parallel
to the z axis. Averaging the path length over the solid
angle for which there is total internal reflection and
allowing for exponential attenuation of the light gives

the expression for the mean path length:

 

Ti’
2" 2" c .
f0 0 d 6X? (-d/A)81neded¢ ,
(13) d =
2n g-ec
f0 f0 exp(-d/1)sineded¢

where 6C is the critical angle. Substituting for d from

equation 4 and integrating over a yields

 

(14) <d>
-8

c ' - z
Sineexp( coseide

 

1T
-'6
c z
Zr: taneexp( x33§§)de
IL
2
o

f

A corrected attenuation length, l.(corrected), can be

defined by the expression:

<d>i )
(corrected)

 

I

(15) exp (-% = expf

<d>
z

1 .

 

(16) that is, 1(corrected =

1(corrected) is now the apprOpriate attentuation length
for expressions 12 and 14. However, since the expression
for <d>/z requires 1(corrected) we must use the uncorrected
value of attenuation length as a first approximation and
iterate until the function converges. The expression for

<d>/z does not depend strongly on the value of <z>, so a

93

 

 

 

Table II. Wavelength-averaged attenuation length in meters
for unpainted cells containing NE 213.

D(mm) A(D) 1(corrected)

45 O.D. 0.954 f 0.009 1.16 t 0.01 2.53 0.07

22 O.D. 0.645 I 0.015 0.78 i 0.02 1.23 0.09

A - 2.16 0.24

 

94

typical value Of z for the measurements taken was used.
The integration was done numerically on an XDS Sigma-7
computer, as was the iteration.

The critical angle so was obtained by measuring the
index Of refraction of the liquid scintillator using a prism

spectrometer and fitting the data to the equation25

(17) n2 = 1 + A

where n is the index of refraction, w is the frequency of
the light, and A andmo are constants. .The index of re-
fraction was measured over the range of 4047 A through
5791 A. At 4250 A the index of refraction 1.52792 f 0.00021,
which in turn gives a critical angle of 40.881o : 0.0070.
With this value of 0c, <d>/z was found to be 1.214 t
0.001. Multiplying this value by 0.954 t 0.009 m (from
Table E2) for (D) of the 45 mm cell gives us a A(corrected)
of 1.16 i .01 m. Similarly, for the 22 mm cell the A(D)
value of 0.645 t 0.015 m gives us a A(corrected) of 0.78 i
0.02 m. Substituting these values of A(Corrected) into
equation 12 gives us an attenuation length in the liquid
of 2.16 r 0.24 m.

The cells were then painted black to eliminate total
internal reflection. It was thought that the only attenuation
would be that in the liquid plus a i/R2 attenuation which
could be easily corrected for. It turns out, however, that
there is still total internal reflection, now at the liquid/
glass interface. Since the critical angle is much larger

than at the glass/air interface, the number of reflections

95

 

 

 

Table III. Practical attenuation lengths in meters for
blackpainted cells containing NE 213.

D(mm) 1(0) A(corrected) Ag

45 O.D. 1.82 i 0.16 1.85 i 0.16 4.81 i 0.12

22 O.D. 1.26 1 0.08 1.28 t 0.08 2.23 i 0.12

A 3.00 t 0.94

 

96

is reduced, and attenuation at the interface contributes
less. However, the solid angle of light detected is so
much smaller that the Compton edge is smeared out. The
smearing of the Compton edge makes it difficult to obtain
accurate values for the attenuation.

The critical angle was determined by measuring the
critical angles for the various lines (6764 %.-4689 3)
of the krypton laser and the 4416 3.1ine of the he-cd laser.
The index of refraction of the glass was found by using
the fitted equation for the index of refraction of the
NE 213 and measuring the relative index of refraction of
the NE 213 to glass. The data for the index of refraction
of the glass were then fitted to equation 17. The resulting
relative index of refraction for NE 213 to glass was 1.0307 i
0.0002, giving a critical angle of 75.980 : 0.04°.

With this value of 6c, <d>/z was found to be 1.015 t
0.001.

The data for the blackened cells were taken in the
same manner as for the clear cells. The data are shown
in Fig. E8. The attenuation lengths were computed as before,
with the exception that the appropriate diameters are the
inside diameters, 41 and 19 mm. The results for the black-
ened cells, which are listed in Table E3, give A2 = 3.00 i

0.94 m.

Figure E8. Compton edge channel vs. path length for NE213 in
the black painted 45 mm and 22 mm cells (darkened line indicatmi
region over which the least-sqares fit was performed).

 

100-

70-

50

30

20

I

CHANNEL OF COMPTON EDGE

 

97

 

 
   

 
   
     

l L

LPSMM. O.D..

22MM. O.D..

PAINTED ‘

 

 

Figure

E8.

1
30 6B 9b
SOURCE POSITION IN CM.

E3. CONCLUSIONS

The values of l.85,3.02, and l.22,l.40 meters for the
attenuation length of 4416 i and 4689 A light in NE 213
and NE 224, respectively, show that NE 213 has the advantage
of having a longer attenuation length than NE 224. Also,
both scintillators show more absorbtion at the shorter
wavelength.

The attenuation lengths measured by the laser method
are precise, but the wavelength-averaged attenuation lengths
are more appropriate for actual design applications. Due
to poor Compton-edge resolution, the value of the attenuation
length in NE 213 obtained with the painted cells, 3.00 t
0.94 m, may be suspect, although it is not inconsistent
with A = 2.16 t 0.24 meters obtained by not painting the
cells. The latter value is then what we believe accurately
represents the practical attenuation length in NE 213.

A 46% end-to-end attenuation measured in a detector con-
taining NE 224 at Michigan State University implies a value
of 1.36 meters for the attenuation length.26 It should be
noted, however, that these values include attenuation in
the walls of the detector itself, whereas the values in
Table I do not. Kuijper, Tiesinga, and Jonker27 by using
a similar method, obtained an attenuation length of 1.5_

meters measured over the first 10 cm of travel. This is,

98

99

of course, where the most rapid attenuation occurs for
reasons stated earlier. For large detectors our value for
the attenuation length may be more appropriate, since it
was measured over a 1.24-meter path.

In designing a detector it is clear that the shape
of the vessel is of importance. The 22 mm O.D. cell had
a higher rate of attenuation than the 45 mm O.D. cell due
to the greater number of reflections per unit length trav-
eled. Complex geometries may pose problems for that reason.
Evers et al.2'a report an attenuation of a factor of four
over the full length of an 80 mm x l m glass cell filled
with NE 213. This implies an attenuation length of 0.72
meters. This is not consistent with the increasing attenua-
tion length versus cell diameter we observed. However,
the condition of the surface of their cell may differ from
ours and the composition of the glass is unknown to us and
may account for the shorter-than-expected attenuation length
for a large-diameter cell.

Much of the light lost at each reflection is thought
to be due to scattering, since the attenuation length of
4250‘: light in Pyrex29 is large compared to the actual
attenuation observed. The glass used in our cells was
extruded and left with its original surface. The annealing
process may allow further imperfections to form on the
surface. By polishing the glass, better light transmission
might be achieved. Also, perhaps there may be better mate-

rial than glass for this application.

LIST OF REFERENCES

10.

ll.

12.

l3.

14.
15.
16.
17.

18.

LIST OF REFERENCES

Manufactured by Nuclear Enterprises, Inc., San Carlos,
CA 94070, USA.

Donald L. Horrocks, Application 9; Liquid Scintillation
Counting, New York: Academic Press, 1974, p. 95.

 

 

Donald L. Horrocks, J. Chem. Phys. 88, 1567 (1970).

 

 

Donald L. Horrocks, Application 28 Liquid Scintillation
Counting, New York: Academic Press, 1972, p. 45.

V.A. Doolin, and V.M. Litjaev, Nucl. Instr. and Meth.
88, 179 (1970).

Ortec Catalog 1002, Ortec Incorporated, 100 Midland
Road, Oak Ridge, TN 37830, USA, p. 70.

Doolin and Litjaev, 178.
Ibid 0
R.F. Post and L.I. Schiff, Phys. Rev. 88, 1113 (1950).

F.T. Kuchnir and F.J. Lynch, IEEE Trans. Nucl. Sci.
NS-lS, no. 3. (June 1969).

R.A. Winyard and G.W. McBeth, Nucl. Inst. and Meth.
88, 525 (1972).

Ibid.

B. Sabbah and A. Suhami, Nucl. Instr. and Meth. 88, 104

Ortec Model 460 Delay Line Amplifier Manual, p. 10.
Manufactured by Nuclear Enterprises, Inc.
Manufactured by RCA, Harrison, NJ

G.G. Kelley, P.R. Bell, K.C. Davis, and H.J. Lazar,
I.R.E. Trans. Nucl. Sci. NS-3, 56 (1956).

Kuchnir and Lynch, 108.
100

19.

20.

21.

22.

23.

24.
25.

26.

27.

28.

29.

101

E.C. Hagen and P.C. Eklund, Rev. Sci. Instr. 81,
no. 9, 1144 (Sept. 1976).

R.A. Winyard, J.E. Lutkin and G.W. McBeth, Nucl.
Instr. and Meth. 88, 143 (1971).

Manufactured by Amperex Electronic Corporation, Hicks-
ville, Long Island, NY 11802, USA.

Handbook 9: Chemistry and Physics, Chemical Rubber Co.,
54th ed.

 

Manufactured by Nuclear Enterprises, Inc., San Carlos,
CA 94070, USA.

Obtained from Nuclear Enterprises, Inc.

E. Hecht and A. Zajac, Optics (Addison-Wesley, 1974)
p. 40.

Ranjan K. Bhowmik, Robert R. Doering, Lawrence E. Young,
Sam M. Austin, Aaron Galonsky and Steve D. Schery,
Nucl. Instr. and Meth. 143 (1977) 63.

P. Kuijper, C.J. Tiesinga, and C.C. Jonker, Nucl.
Instr. and Meth. 88 (1966) 56.

D. Evers, E. Spindler, P. Konrad, K. Rudolf, W. Ass-
mann, and P. Speer, Nucl. Instr. and Meth. 124 (1975)
24.

G.W.C. Kaye and T.H. Laby, Tables of Physical and
Chemical Constants, (John Wiley & Sons, Inc., 1966)
p. 86.