LISA
72
DES
2
.7.
.
..
.
1
1
PA
11
PL
ACUM! ALTERA
.
.
:
.
TYP
Vitu
SA
5
M
.
::.
...
..
-
1
..
:
.."
LIN
21
.
YA
HEN
A4
2.23
7
.
.
,--.
;
'
H.,'
E-
LE
AI
UNCLASSIFIED
ORNL
P
1127

ir
.
!
!"
I
"
.
.
.
.
4

.
.
:
-
LY
ATLE
AT3
I
tre
RR
**
***r
ORNn-Pullah
CONF-650408's
APR 27 1966
THE EVALUATION OF COLPUTER PROGRAMS
FOR Y-RAY SPECTROMETRY IN ACTIVATION ANALYSIS*
J. F. Emery
F. F. Dyer
T. Alexander**
E. Schonfeld
Oak Ridge National Laboratory
Oak Ridge, Tennessee
ABSTRACT

The present trend in activation analysis is to non-destructive
analysis. Satisfactory performance of such analyses requires very
complex and expensive instruments. When one uses hand resolution
LEGAL NOTICE -
techno ya
o imated, wo roupact to do near
revert
bo No wployment of contract
uployee a mucha contractor proper
o
plosna nd contractor, to the one that
ad the combination
duke u proces declourd to the report may not to stay
die oond t
ctor.
This report was prepared as an account of Government uponsored work. Neither the Vatted
do
formation porn
3. Assumon any Habilities with respect to the une ac, or for damages rosulting from the
Momo, wr the Creaton, more porno schug on behalf of the Council
moy, oplote, as weten of the worsham contafond tout sport, or the
rrentadon, spre
d, or proor
t,
play
at,
As and we the stone, person why we h
long or connector of the Coundhon, or
, or pronto town i, a
o
tyd mets or
dan,
Images ar autoctra O Contatom,
months,
Colton, or to
A. Mat
tetor
2
Iwo
dent
much
no
-
of complex gamma-ray spectra to calculate the percentage of some
component, he soon finds himself spending most of his time doing
calculations with the result that much of his "expensive instru-
mentation" is in standby. The answer to this problem is, of course,
to use computers both for resolution of gamma-ray spectra and for
calculations.
The analyst wants accurate results with the least amount of
effort and, of course, the management wants to hold down the cost.
With these criteria in mind, three operational computer programs
for resolving gam-ray spectral data have been evaluated. Two of
these programs use linear least squares methods and the third, a non-
linear least squares method for resolving the spectra. Some problems
considered are: ability to make gain shift corrections, the effect of
zero activity nuclides on other nuclides present, and the significance
and reliability of computed "standard deviations." Results are pre-
* Research sponsored by the U. S. Atomic Energy Commission under contract
with the Union Carbide Corporation.
**Present address: Georgetown College, Georgetown, Ky.
PATENT CLEARANCE OBTAINED: REEEASE TO
THE PUBLIC IS APPROVED. PROGEDURES
ARS ON FHE HY THE RECOVING SECTION,
...
.
.
.
.
.
.
!
.
2.
sented for a series of samples with five replicates each analyzed
by the three programs.
Computer programs for resolving gamma-ray spectra require a
number of computer control data, such as peak channel, gamma-ray
energies, gain shift, et al. The number and complexity of this control
information depends largely upon the versatility of the program and
upon the purpose for which the program was designed. Each of these
programs was evaluated for its simplicity of use with a large number
of samples. The computer cost per analysis for each program was
also determined.
Introduction
The present trend in activation analysis is to nondestructive analysis.
This technique is ideal for processing a large number of samples by automation.
The resolution of the gamma-ray spectra obtained and the calculation used are
easily programmed for a computer.
Replacement of an obsolete analyzer system with a more advanced system
offered an opportunity to study a number of gama-ray resolution programs.
Three least squares analysis programs were selected for study. Program I
was developed for the CDC 2604A computer and written by E. Schonfela'). Program II
was written for an IBM 7090 computer by R. G. Helmer et alls) and compiled on
the CDC 1604A. Program III was written by L. Salmon 5) and adapted to the
CDC 1604A by E. Schonfeld.
The solution of activation analysis gamma-ray spectra and presentation of
the analysis require a very versatile program. The program should be able to
correct both the standards (library spectra) and the sample (composite spectrum)
for background, different live time count intervals (oth), and decay. Each
SPECrhum
speetra should also be corrected for any instrumental drifts such as threshold
(zero energy calibrations) and gain shifts.
.
. 3.
The results of the analysis should be presented in units and dimensions
that are useable. The time to prepare the spectra for computer analysis as
well as the time to compute the problem should be a minimum.
. Comparing these programs using the above criteria requires separate
examination of the library and composite spectra control parameters.
Control Parameters
1. Library
Background subtraction: Most multichannel analyzers are capable of subtracting
background, and analyzers with data reduction systems can select a fraction of
a background that has been accumulated over a long period of time and subtract
this fraction from a standard spectrum. However, it is more convenient if
these subtractions can be done by the computer. Programs I and II can make
these subtractions. Table I compares these and other features.
A th: Count rates of the various standards will vary considerably; therefore
it is not practical to count every standard for the same live time. Proe 'anas I
and III can normalize the standard spectra to some unit live time, for example,
c/m, a/m, etc.
Intensity: Usually activation analysis results are reported in ppm or mg/ml.
To compute these values each standard must have intensitys in units of micro-
grams or milligrams. The standard spectra of Programs I and III may have any
assigned value and units.
Decay: Standards are frequently short lived so that significant decay has
occurred from reactor discharge time to the counting time. This necessitates
& decay correction to some reference time. Only program III can correct the
standard spectra for decay.
Gain and Threshold: Shifts: Standard spectra should always be corrected for
gain and possibly threshold shifts. Gain shifts may arise from small fluctu-
ations in temperature and voltage at the detector. High intensity count rates
may cause gain shifts of several percent in the multiplier phototube. Only
Program II can correct the individual standard spectra. Program III uses the
first spectrum in the library as a reference for making gain corrections to
the composite spectrum.
Component Selection: The selection of different library spectra for each
composite spectrum or group of composite spectra should be accomplished in
a straight forward and easy manner.
Program I's component spectra are called from the library for each
composite spectrum while in Program II the entire library is used for the
composite spectrum. If a composite spectrum requires a component that is not
present in the original library, then a new library must be used. The component
spectra in Program II are designated by indicating the spectra that should not
be used. This is accomplished by setting the intensities for these spectra
equal to zero.
2. Composite Spectra
Background Subtraction: All three programs can subtract background counted
for any time interval.
Gain and Threshold Shifts: Composite spectra should always be corrected for
gain and threshold shifts. A correction of 0.1% may be very critical when two
of the components are very close in gamma-ray energy. A different approach
is used by each program to show this problem. Table II summarizes these features
of the programs.
Summ
Program I requires a library very carefully prepared. All of the component
spectra should be on the same energy scale. With this condition met or assumed
the program includes the gain and threshold shifts as varibles along with the
components in the least squares analysis of the composite spectrum.
Program II requires an exactly determined pulse height (gamma-ray energy)
vs channel scale. This requires that the gamma-ray energy of the nuclides be
precisely known. All of the components in the library are gain shifted to
correspond to the previously determined pulse height scale. A well defined
photopeak of known garma-ray energy is used to determine the gain shift and
the program then corrects the composite spectrum accordingly.
Frogram III also requires a library that has been very carefully prepared.
A component of the composite spectrum that has a well defined photopeak is
selected and placed first in the library. This component spectrum is now used
as a reference to determine the gain shift in the composite spectrum.
A th: Programs I and III normalize the composite spectra to some unit time.
Both programs require the stl to have the same units i.e., seconds, minutes
etc.
Decay: A number of composite spectra will require decay corrections and only
Programs I and III can unake these corrections.
Sample Fraction: Some means must be provided to account for sample weight and
concentration or dilution factors. Only Program I will make these calculations.
Problem Setup
1. Input
The problem presented to the computer for analysis consists of cards
containing spectral data and control parameters. The number and complexity
of these parameters vary considerably in these programs. The number of control
cards and some of the control parameters are discussed below for a hypothetical
problem that assumes one composite spectrum to be gain shifted and a library with
five standard spectra for each program.
Program I: The first card. specifies the data format. Cards two through six
The second card lists the number of standards in the library, the number
of channels of data, the type of input, the region of fit, and the count
interval of the background spectrum. The third card names the various stand-
ard spectra in the library. The fourth lists the half: lives of the standard
- BA
spectra, the fifth, the Atz's of each of the standard spectra, and the sixth
specifies the intensity of the standard spectra.
The remaining three cards are used only for the composite spectrum.
The seventh lists the sti, the decay time, and the weight of sample or
dilution fraction. The eighth is the identification card for the composite
spectrum, and the last specifies gain shift options and identifies the
various components used in the analysis of the composite spectrum.
Program II: The first card is a problem identification card.
The second.
lists the region of fit, the number of standard spectra, the number of back-
ground spectra, and input-output options. The third identifies the composite
spectrum, the fourth identifies the background spectra and the fifth specifies
the intensity of the background spectra. The sixth lists the standard spectra.
Normally there are only two cards required for each spectrum, however,
a third card is required when a spectrum needs to be gain shifted. The first
card identifies the composite spectrum and lists the upper and lower channels
of the photopeak that is used for gain shift corrections. The second lists the
new position for the photopeak.
The last card inciicates the end of the spectrum;
the last spectrum of the problem requires two cards.
Program III: The general directions for this program are contained on two cards.
The first lists the number of standard spectra, the region of fit, and the
reference time for decay corrections; the second specifies the data format.
The background, standards and composite spectrum each requires a card
listing the str, the time of count, the upper and lower channel of the photopeak
used for gain shift corrections, and the title of the spectrum. The standard
spectra require an additional card specifying the intensity, the units of the
intensity, and the half life.
2. Output
The computer output of the three programs vary considerably. The CDC 1604A
costs #30.00/hour and 41.10/thousand lines; therefore, the computer time and the
time per problem. should be minimized.
Program I: A problem consisting of one composite spectrum ard five library
standards each having 200 data points will require 12.5 seconds of computer
time and 45 lines of output. The cost of this problem is 0.15.
Program II: This program, using the same problem as above, requires 23seconds
of computer time and 274 lines of output. The cost is $0.50/probien.
Program III: This program requires only 7.1 seconds of computer time but the
output has 154 lines. The cost of $0.23/problem is due to the large number
of lines in the output. Table III summarizes the input and output of these
programs.
Accuracy and Precision
Table IV compares the results from the three programs on five replicates
from each of several samples. The standard deviations listed are the standard
deviation of the mean.
The computed standard deviation of an individual
analysis was usually a factor of 2 or 3 smaller than the standard diviation
of the mean. The computed standard deviation is only a measure of the internal
agreement of the data, but the standard diviations of the mean includes external
errors such as sample inhomogenity, geometrically positioning of the sample
for counting, variations in irradiation time and the neutron flux, and
inaccuracies in the decay time.
Another test to demonstrate the effects of count rate consisted of an
experiment in which five radionuclide 884, 1370s, 8oco, 244ce and 203Hg were
counted individually at a given geometry and used as standards for the library.
A composite spectrum was obtained by.counting all of the nuclides at the given
geometry at the same time. The plotted points in Figure 1 represent the composite :
137Cs nuclide in the composite spectrum had a count rate of 10,000c/s and
accounted for 83% of the total activity. The composite spectrum was analyzed
by each of the three programs (Table V) using a 1370s standard with a count
rate of 200 c/s. A count rate of 10,000 c/s woula be expected to produce
a random sum count rate of 100 c/s. The random sum count rate results in
a photopeak at 2 x 0.662 Mev or 1.3 Mev, which is in the region of the two
60C0 photopeaks. Random summing from 157Cs woula be reflected in erroneously
high results for 60co analyse.
The count rate for boco was 520 c/s and an error of 100 c/s amounts to
an error of 20% in the value of 60Co. The observed error was 25%.
Analysis of the above problem by use of a 137Cs standard spectrum with
TYO
a count rate similiar to that of the sample eliminates the random summing
error and the answer shows the expected agreement for BoCo (Table VI).
Conclusions
The above results demonstrate that all three programs resolve complex
gamma-ray spectra equally well. Programs I and III have the added advantage
of performing a number of additional calculations necessary in neutron acti-
vation analysis. The higher cost of Programs II and III might be reduced by
changing the output format to provide fewer lines/problem.
One source of error in resolution of complex gamma-ray spectra is the high
count rate of one or more of the components of the composite spectrum. The
proper selection of standard spectra for a composite spectrum with a high count
rate will certainly improve the results.
References
E. Schonfeld, A. I
r and W. Davis, Jr., "Determination of
Nuclide Concentrations 'in Solutions Containing Low Level of
Radioactivity by Least Squares Resolution of the Gamma-Ray
Spectra,"
ORNL-3744
2.
R. G. Helmer, D. D. Metcalf, R. L. Heath and G. A. Cazier, "A
.Linear Least Squares Fitting Program For The Analysis of Gamma-
Ray Spectra Including a Gain Shift Routine," IDO-17015,
111.
Sept. 1964.
3.
L. Salmon, "Computer Analysis of Gamma-Ray Spectra from Mixtures
of Known Nuclides by the Method of Least Squares," in G. D. O'Kelley
Editor, Proceedings of the Symposium on Applications of Computers
to Nuclear and Radiochemistry, Gatlinburg, 1962, NAS-NS-3107.
Table I
Program Capabilities of Library
Program
Parameter
III
Sub Background
Yes
Yes
Yes
Yes
Ath
Intensity
Yes
Yes
Decay
No
Yes
Gain Shift
No
NO
Component Selection
Yes
NO
Table II
Program Capabilities of Composite
Program
Paraneter
II
III
Sub Background
Yes
Yes
Yes
Gain Shirts
Yes
Yes
Yes
sti
Yes
NO
Yes
Decay
Yes
Yes
Sample Fraction
Yes
No
NO
Table III
Problem Set Up for One Composite
with Five Library Spectra
Program
Input: 1. Control Cards
2. Control Items
r w woh
Additional Compositers
1. Control Cards
2. Control Items
- 20
Output:
E
274
1. Lines/problem
2. Time/problem
3. Cost/problem
12.5 sec.
7.1 sec
23 sec.
+0.50
50.15
0.23
u
i
...
.
.
.
..
.
.
.
.
.
-
-
- .
Table IV
Determination of Al and Ti Using Three Least Square
Computer Programs for Data Resolution
II
III
Chemical
Sample
pom
ppm
שסט
ppm
AL
3029+20
15467+97
302120
15486+127
3030_21
15449190
3150
16800
Al
3039+17
25516+81
3033+15
15514+91
3040416
15495795
3270
16900
Al
2025+21
16998+145
2022+20
16952-143
2026+12
169334140
11
2657+60
10451310
2652:60
204430314
i
70624295
8339+425
2657757
104313335
7054266
8287+420
6487+226
7712+278
6560
8300
Al
Ti
64904220
7763+269
6430
8000
Table V
Gamma-Ray Spectral Resolution by Least Squares
Computer Programs
Composite 914578*
% Error (T-F) x 102
Nuclide
III
Counts
884
9.0
137CS
30
144ce
10.8
-2.9
25.3
16.5
11.1
-3.2
25.1
13.0
59.1
12.7
-3.1
30.2
-10.9
56.1
2.8
72.0
0.7
69.7
*composite and standards corrected for gain shifts.
.
Table VI
Gamma-Ray Spectral Resolution by Least Squares
Computer Programs
Composite 914578*
(T-F) x 102
ror
Nuclide
I
II
Present
Counts
III
884
137CS
9.0
3.6
83.0
4.5
2.8
-0.65
0.45
10.6
71.2
3.1
-0.90
0.22
11.).
71.2
144ce
203Hg
-0.65
0.45
68.
golo
2.8
0.7
* Composite and standards co
Composite and standards corrected for gain shifts. Hot: 137Cs used
as standard
Fig. 1. Composite Spectrum Showing High Count Rate in Channel 120-140
From137Cs Suraming
VERA
SEW
SLATOR
..
-
-
:
.
.
.
WYTW2R2
1
4
#
!
7
-
.
..
.
-
"
!
"
"
. 1 "
.
.
-
'
'
'
4
6
:
14:
*.
-
,
1
2
>
TVDIN10 - JIY - INDO
obedienti
ORNI - AEC - OFFICIAL
1E+05
914578.
ORNL-DWG. 65-947

COMPONENTS PRESENT
88Y
@ 137CS
© 60 Co
© 144 Ce
© 203 Hg
1E+04
COUNTS/CHANNEL
-COMPOSITE
:
1E+03!
-
Hist
o
ria
.
.'
. r
.
.
...
.
1E+02 40
20
40
60
80
100
120
140
160
180
200
220
240
260
CHANNELS
.
-
---
!...
* ОINI - AC - OFFICIAL
ORNL - ABC - OFFICIAL
A
1.
'
.
1
V
Y
.
r
1
.
.
.
-
HL
DATE FILMED
5 / 26 /65







.
TRW
-LEGAL NOTICE

This report was prepared as an account of Government sponsored work. Neither the United
States, nor the Commission, nor any porson acting on behalf of the Commission:
A. Makes any warranty or representation, expressed or implied, with respect to the accu-
racy, completeness, or usefulness of the information contained in this report, or that the use
of any information, apparatus, method, or process disclosed in this report may not infringe
privately owned rights; or
B. Assumes any liabilities with respect to the use of, or for damages resulting from the
use of any information, apparatus, method, or process disclosed in this report.
As used in the above, "person acting on behalf of the Commission" includes any em-
ployee or contractor of the Commission, or employee of such contractor, to the extent that
such employee or contractor of the Commission, or employee of such contractor prepares,
disseminates, or provides access to, any information pursuant to his employment or contract
with the Commission, or his employment with such contractor.
END



3