United States
Environmental Protection
Agency
Environmental Monitoring Systems —
Laboratory
Las Vegas NV 89114
Research and Development
EPA-600/S4-83-018 Aug. 1983
Project Summary
Test Procedure for Gamma
Emitters in Drinking Water:
Interlaboratory Collaborative
Study
V. R Casella, C. T. Bishop, and E. L Whittaker
This Intel-laboratory collaborative
study was conducted for the test pro-
cedure. Gamma Emitting Radionuclides
in Drinking Water (Method 901.1 in
Prescribed Test Procedures for Mea-
surement of Radioactivity in Drinking
Water, EPA-600/4-80-032, August
1980). The purpose of the study was to
determine the estimated precision and
accuracy of multilab test results when
this test procedure is used for analyz-
ing drinking water type samples con-
taining variously concentrated mixtures
of four gamma emitting radionuclides.
Drinking water samples would not be
expected to contain large numbers of
radionuclides.
Four reference water samples con-
taining three or all four of the gamma
emitters, cobalt-60, ruthenium-106,
cesium-134 and cesium-137, at con-
centrations ranging from 6 to 400
pCi/l, were prepared for the study. The
concentrations were selected to cor-
respond to 0.1 to 2 times the Maximum
Contaminant Level (MCL) indicated in
the National Interim Primary Drinking
Water Regulations (NIPDWR). Test re-
sults from 32 participating laboratories
were evaluated. The extent of report-
ing test results for the various sample
radionuclides ranged from as few as
six participants to all participants. For
the most part, the average of the con-
centration values reported by the par-
ticipating laboratories agreed quite well
with the known values for all four sam-
ples. The only striking exceptions were
the test results for ruthenium-106 at
the three lower concentrations. Those
results indicate that for ruthenium-
106 the method is not sufficiently
sensitive to meet the requirements of
the NIPDWR.
A statistical analysis of the test
results for cobalt-60 gave coefficients
of variation for repeatability (within-
laboratory precision) of 12.2, 3.5, and
4.9 percent respectively for concentra-
tion levels of 9.81, 98.8, and 201.9
picocuries per liter (pCi/l), for an aver-
age repeatability precision of 6.9 per-
cent The corresponding coefficients
of variation for reproducibility (com-
bined within and between laboratory)
were 23, 6.6, and 5.7 percent for an
average reproducibility precision of
11.8 percent. For ruthenium-106 at
the 60.5 pCi/l concentration (2 MCL),
the coefficient of variation for repeat-
ability was 23 percent and the coef-
ficient of variation for reproducibility
was 32 percent For cesium-134 at
concentration levels of 7.9, 79.6, and
161 pCi/l, the coefficients of variation
for repeatability were 30, 5.0, and 7.1
percent respectively, for an average
repeatability precision of 14.0 percent.
The corresponding coefficients of vari-
ation for reproducibility were 30,18.5,
and 8.7 for an average reproducibility
precision of 19.1 percent For cesium-
137 at concentrations of 19.9, 99.5,
202.8, and 393.5 pCi/l, the coeffi-
cients of variation for repeatability
were 10.2, 4.5, 4.0, and 2.0 percent
respectively for an average repeatabil-
ity precision of 5.2 percent The corre-
sponding coefficients of variation for
reproducibility were 14.4,7.2,6.6, and
-------
5.6 for an average reproducibility pre-
cision of 8.4 percent
The average accuracy indexes for
cobaft-60, cesium-134 and cesium-137
were 107, 94.9, and 101.6 percent
respectively. Forruthenium-106 at the
60.5 pCi/l concentration, the accuracy
index was 100.7 percent.
This Intel-laboratory collaborative
study was conducted by the Mound
Facility of the Monsanto Research Cor-
poration under an Interagency Agree-
ment, EPA-IAG-AD-89-F-1-395-0, for
the Quality Assurance Division of EMSL-
Las Vegas. The Mound Facility of
Monsanto Research Corporation oper-
ates under DOE Contract Number DE-
ACO4-76-DP00053. The work covered
the period from November 1, 1980 to
November 1, 1981.
This Project Summary was developed
by EPA's Environmental Monitoring
Systems Laboratory, Las Vegas, NV, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Many man-made radionuclides decay
by gamma photon emission. Because
gamma photons are discrete in energy,
they can be resolved and quantitated by
gamma spectrometry. The NIPDWR list
the MCL for all significant man-made
radionuclides. The required detection limits
of the test procedure used to measure
those radionuclides are 0.1 MCL, unless
otherwise stated in Table B of Section
141.25 of the Regulations. The test pro-
cedure in this study uses a gamma spec-
trometric analysis of unprocessed (no
chemical separations) portions of drinking
water samples. The purpose of the study
was to estimate the precision and ac-
curacy of the test procedure from multi-
laboratory test results of prepared sam-
ples that contain radionuclide concentra-
tions ranging from 0.1 MCL to 2 MCL The
radionuclides tested in this study were
cobalt-60, ruthenium-106, cesium-134,
and cesium-137.
The test procedure did not stipulate
which of the two commonly used detec-
tors to use. The two commonly used
gamma detectors are the thallium acti-
vated sodium iodide, IMal(TI), detector and
the lithium drifted germanium, Ge(Li),
detector. The sodium iodide detectors
generally have higher counting efficien-
cies than the germanium detectors, but
the germanium detectors have consider-
ably better resolution than the sodium
iodide detectors. Also, a specific data
reduction method was not stipulated in
the test procedure. Therefore, participants
were at liberty to use the counting sys-
tems and data reduction methods avail-
able to them.
Procedures
Analytical Test Procedure
The analytical test procedure used in
this collaborative study was Method 901.1,
"Gamma Emitting Radionuclides in Drink-
ing Water," which was published in EPA-
600/4-80-032, August 1980, "Prescribed
Procedures for Measurement of Radio-
activity in Drinking Water." Method 901.1
is also contained in Appendix B of the
Project Report
Collaborative Test Procedure
Four reference samples, a copy of the
EPA Method 901.1, "Gamma Emitting
Radionuclides in Drinking Water," stan-
dard solutions of the gamma emitters to be
measured, and a set of the collaborative
study instructions were sent to 38 labora-
tories. The four reference samples con-
tained different concentrations of the four
gamma emitters used in the study. Test
results from 32 laboratories were received
and evaluated.
Data Processing Procedures
A statistical eval uation of the test resu Its
was conducted with the use of procedures
described in E-691, E-1 77, and E-1 78 of
the ASTM Standard Part 41,1980, to de-
termine the repeatability precision (within-
laboratory variation), the reproducibility
precision (combined within- and between-
laboratory variation), and the accuracy of
the test procedure. The standard devia-
tions and equations for their calculations
are listed below.
Standard deviation of replicate test results
within Lab i, for sample j, (Sy)
5
|_h=1
Eq. 1
where: Xljh= the result reported for the
h replicate of the j sample
_ material by Lab i
Xy = the mean of the individual
results of sample j for Lab i
n|j= the number of replicates of
sample j reported by Lab i.
Repeatability (within- laboratory) standard
deviation for sample j, (Sr)
(n,J-1)Sij2+(n2j-1)S2j2
(riij + n2j +
-(npj-1)Spj2
"npj)-P < Eq.2
where: P = the number of participants
in the study.
Standard deviation of individual labora-
tory average from grand average for the j
sample material, (SxJ
p (XrXj)2/(P-1)
Eq. 3
where: X,j = the average of the test re-
sults for sample material j
by Lab i
Xj = the grand average for sam-
ple material j.
Standard deviation of between-labora-
tories for the j sample material, (SL.)
_2 _
V*
Eq4
Reproducibility (combined within- and
between- laboratory) standard deviation
for the j sample material, SR
The percent coefficient of variation for
repeatability (within- laboratory precision)
(also called repeatability index) for sample
_
Vrj%= 100SrVXJ
The percent coefficient of variation for
between-laboratory precision for sample j,
_
VL.%= 100SL/X
J ' Eq. 7
The percent coefficient of variation for
reproducibility (combined within- and
-------
between-laboratory precision) (also called
reproducibility index) for sample j, (VR%)
VR.%=100SRj/XJ
Eq. 8
Accuracy index, a percent relationship
of the grand average to the known value
for the j sample material, (Aj%)
Aj%=100-J
Eq. 9
where:
Yr
the known value for the j
sample material (pCi/l).
t-test to determine significant differ-
ences or systematic error for sample j, (tj)
(P-1) degrees of freedom
Eq. 10
where: P = number of participants
YJ = known value of the sample
j concentration
tc = critical values shown in
Tables 1 through 4, values
for tj greater than tc are
significantly different and
show a systematic error.
Results and Discussion
A summary of the statistical evaluation
of the test results for the four radionu-
clides is given in Tables 1 through 4. Table
5 shows the radionuclide concentrations
for the four samples.
Measurement parameters varied greatly
for the study (see Table 11 of the full
report). Three sodium iodide and 31
germanium detectors were used in the
study. All germanium detectors were of
the lithium drifted type. One laboratory
used only 50 ml sample portions, whereas
all other laboratories used 400 ml or
greater sample portions (average volume
used was 1,850 ml). Counting times varied
from 50 minutes to over 4,000 minutes
(average counting time being about 1,240
minutes). Spectral stripping was achieved
by computer calculation for 22 sets of
results, by hand calculation for eight sets
of results, and by a combination of com-
puter and hand calculation for four sets of
results.
Although the test results obtained with
sodium iodide detectors were less than 10
percent of the total, a brief study of these
results indicates that they did not con-
tribute to the limitations or biases of the
test procedure to a significantly greater
degree than the germanium detector test
results.
Table 1 shows the tj value to be greater
than the critical value (tj for the lowest
concentration of cobalt-60 (0.1 MCL). The
grand average of the test results, then, was
significantly different from the known value
and therefore was biased on the high side.
Test results for the other two concen-
trations (MCL and 2 MCL) were in good
agreement with the known values (accu-
racy indexes of 102.2 and 99.9 percent).
The table shows an average repeatability
precision of 6.9 percent and an average
reproducibility precision of 11.8 percent
Table 2 shows a general failure of the
participating laboratories to determine the
lowest concentration (0.2 MCL). The table
also shows significant high biases for the
next two concentrations (0.5 MCL and
MCL). The grand average of the test results
for the highest concentration (2 MCL) was
in good agreement with the known value
(accuracy index of 100.7 percent). The
table shows an average repeatability pre-
cision of 23 percent and an average repro-
ducibility precision of 36 percent.
Table 3 shows significantly low biases
for the two highest concentrations (MCL
and 2 MCL). The lowest concentration
grand average (0.1 MCL) was in good
agreement with the known value (accu-
racy index of 98.7 percent). The average
repeatability precision was 14.0 percent
and the average reproducibility precision
was 19.1 percent
Table 4 shows a high bias for the lowest
concentration (0.1 MCL) and good agree-
ment between the test results of the three
higher concentrations (0.5 MCL, MCL, and
2 MCL) and the known values (respective
accuracy indexes of 100.0, 99.7, and
Table 1. Cobalt-60 (Co-60) in Water, Precision and Accuracy Summary
Parameter3
V, (PCi/l)
^(pCi/l)
Sy (pCi/l)
Sr,'(pCi/l)
SLl (pCi/l)
SRl (pCi/l)
vr (%)
VL, (%)
\/ /o/ J
r pi f /Of
A, (%)
t
tc (P)b
Co-60
(0. 1 MCL)
9.81
11.7
2.5
1.43
2.3
2.7
12.2
20
23
119.3
3.63
2.62 (23)
Co-60
(MCL)
98.8
101.0
6.1
3.5
5.6
6.6
3.5
5.6
6.6
102.2
2.07
2.79
Co-60
(2 MCL) Average
201.9
201.6
9.1
9.9
5.8
11.5
4.9
2.9
5.7
6.9
9.5
11.8
99.9 107.
-0.19
(33) 2.80 (34)
"Terms defined in text
b Number of laboratories that reported data.
Table 2.
Parameter*
Y (pCi/l)
Xt (PCi/l)
S% (pCi/l)
Srl (pCi/l)
SLl (pCi/l)
SR, (pCi/l)
vr (%)
VL, (%)
VR (%)
A (%)
t'
tc (P)b
Ruthenium- 106 (Ru-106) in Water, Precision and Accuracy Summary
Ru-106
(0.2 MCL)
5.9
<47
-
-
-
-
-
-
-
-
-
-
Ru-106
(0.5 MCL)
15.4
28.0
12.8
7.5
11.6
13.8
27
41
49
181.8
2.41
1.82 (6)
Ru-106 Ru-106
(MCL) (2 MCL)
29.9 60.5
38.3 60.9
9.6 16.6
7.2 13.7
8. 1 13.5
10.8 19.2
19 23
21 22
28 32
128.1 100.7
3.03 0.09
2.29(12) 2.41(15)
Average
23
28
36
137
"Terms defined in text
b Number of laboratories that reported data.
-------
Table 3. Cesium- 134 (Cs-134) in Water, Precision and Accuracy Summary
Cs-134 Cs-134 Cs-134
Parameter3 (0. 1 MCL) (MCL) (2 MCL) Average
Y^fpd/l) 7.9 79.6 161.0
Xjfpd/l) 7.8 73.3 151.2
S-x,(pCi/l) 1.0 9.6 10.6
Sr/(pCi/l) 2.3 3.7 10.8
S^fpd/l) 0° 9.2 7.4
Sfufpd/l) 2.3 9.9 13.1
Vrl(%) 30 5.0 7.1 14.0
VL4(%) Oc 12.6 4.9 8.7
VR/(%) 30 18.5 8.7 19.1
A,(%) 98.7 92.1 93.9 94.9
tf -0.40 -3.83 -5.39
tc(P)b 2.44(161 2.80(34) 2.80(34)
"Terms defined in text.
* Number of laboratories that reported data
c Following the ASTM £69 1 procedures, the value is set = 0.
Table 4. Cesium- 137 fCs- 137) in Water, Precision and Accuracy Summary
Cs-137 Cs-137 Cs-137 Cs-137
Parameter3 (0. 1 MCL) (0.5 MCL) (MCL) (2 MCL} Average
Yjlpd/l) 19.9 99.5 202.8 393.5
X~j(pCi/l) 21.5 99.5 202.2 389.3
SxjfpCi/l) 2.7 6.4 12.1 20.9
Srl(pd/l) 2.2 4.5 8.0 7.8
SLl(pCi/l) 2.2 5.6 10.7 20.2
SRl(pCi/l) 3.1 7.2 13.4 21.7
Vri(%) 10.2 4.5 4.0 2.0 5.2
V^(%) 10.2 5.6 5.3 5.2 6.6
VRl(%) 14.4 7.2 6.6 5.6 8.4
AJ(%) 108.0 100.0 99.7 98.9 101.6
tf 3.25 0.0 -0.29 -1.17
tc(P)b 2.75(30) 2.77 f 32) 2.80(34) 2.80(34)
"Terms defined in text.
b Number of laboratories that reported data.
Table 5. Reference Sample Radionuclide Concentrations
Radionuclide Concentrations, pCi/l
Sample cobalt-60 ruthenium- 106 cesium- 134 cesium- 137
1 201.9 60.5 161.0 202.8
2 98.8 15.4 7.9 393.5
3 9.81 5.9 - 19.9
4 - 29.9 79.6 99.5
98.9 percent). The average repeatability there was generally good agreement be-
precision was 5.2 percent and the average tween the test result grand averages and
reproducibility precision was 8.4 percent the known values for all concentrations
(0.1 MCL to 2 MCL) for cobalt-60, cesium-
134, cesium-137, and for the highest
Conclusions concentration (2 MCL) of ruthenium-1 06.
Although many variations existed in the This study demonstrates that the test
measurement parameters in the study, procedure used can produce acceptable
results over a rather wide range of varia-
tion in the measurement parameters. It is
essential to the success of the procedure
that the counting system be calibrated at
the same sample-to-detector geometry
that will be used for counting samples.
The high biases shown for the lowest
concentration (0.1 MCL) for cobalt-60
(Table 1) and cesium-137 (Table 4) are
not serious.
The low bias for the cesium- 1 34 mea-
surements is likely due to summing effects
which can cause lower counting efficien-
cies. However, the bias is not serious,
since it is only an average of 5. 1 percent for
the concentration range of 0.1 MCL to 2
MCL.
The study shows that the test proced u re
is not sufficiently sensitive for the mea-
surement of ruthenium-1 06 with the
sample volumes and counting times used in
the study. The NIPDWR require a measure-
ment sensitivity of 0.1 MCL (ruthenium-
1 06 MCL is 30 pCi/l). The test procedure
has adequate sensitivity for cobalt-60,
cesium-134, and cesium-137.
Recommendations
The test procedure in this study should
be used to analyze drinking water samples
for gamma emitting radionuclides for
compliance underthe Safe Drinking Water
Act. An applicability test should be in-
cluded in the procedure to determine
whether the test procedure for a given set
of conditions of counting time, counting
efficiency, and sample volume can be used
to analyze drinking water samples for a
selected radionuclide.
Section 1 .4 of the "Scope and Applica-
tion" should be changed to read:
The method is applicable for analyz-
ing water samples that contain radio-
nuclides emitting gamma photons
with energies ranging from about 60
to 200 keV. The required sensitivity
of measurement for the more hazard-
ous gamma emitters is listed in the
National Interim Drinking Water Reg-
ulations (NIPDWR), Section 1 41 .25.
For a method to be in compliance, the
detection limits for photon emitters
must be 1/10 of the applicable limit
(MCL).
Section 1 .5 of the "Scope and Applica-
tions" should be amended to read:
The criterion for the application of th is
test procedure to analyze drinking
water samples for a selected radio-
nuclide to comply with the NIPDWR
is derived from Equation 4 of Appendix
C, using the factor 2 for the 95%
-------
confidence factor instead of 1.96 and
is given by the following expression.
4+ (16 +32tB)'/z <
2t x E x b x 2.22 x V~
0.1 MCL (pCi/l)
where:
t =countmg time in minutes
for the sample and the
background
B =background in cpm of pho-
ton energy peak in question
E=photon counting efficien-
cy, cpm/dpm
b =abundance of photon ques-
tion, ydpm/dpm
2.22 =conversion constant dpm/
pCi
V =volume of sample analyzed,
liters (I)
Analysts using this test procedure for
the measurement of gamma emitters in
drinking water should determine counting
efficiencies for each radionuclide to be
measured, with standard solutions at
exactly the same geometry (sample-to-
detector geometry) as that at which sam-
ples are to be counted. This will provide
accurate counting efficiencies and cancel
summing effects. If counting efficiency is
read from a general efficiency curve, then
care should be taken to make a separate
correction for summing effects when that
is known to occur with particular radio-
nuclides analyzed.
The present study showed that ruthenium-
106 at the 2 MCL concentration can be
measured by the test procedure, and the
average lower limit of detection was esti-
mated to be 44 pCi/l (1.5 MCL). There-
fore, an increased sensitivity by a factor of
1 5 to 20 is needed to meet the required
sensitivity (lower detection limit of 3 pCi/l,
required by NIPDWR). The low gamma
abundance per decay and the relatively
low MCL are largely the causes for the
inadequate sensitivity of the test proce-
dure for ruthenium-106. However, ruthen-
ium-106 also decays with beta particle
emission at a decay abundance of 100
percent. The high beta abundance com-
bined with the considerably higher beta
counting efficiency (than for gamma-ray)
would provide adequate sensitivity for
ruthenium-106. Therefore, a method that
provides for the separation of ruthenium-
106 from water samples followed by a
beta count is recommended.
V. R. Casella and C. T. Bishop are with Monsanto Research Corporation-Mound,
Miamisburg, OH 45342; the EPA author E. L. Whittaker (also the EPA Project
Officer, see below) is with the Environmental Monitoring Systems Laboratory,
Las Vegas, NV 89114.
The complete report, entitled "Test Procedure for Gamma Emitters in Drinking
Water: Interlaboratory Collaborative Study," (Order No. PB 83-207 381; Cost:
$10.00, subject to change} will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
U. S. GOVERNMENT PRINTING OFFICE: 1983/659-095/0723
-------
United States
Environmental Protection
Agency
Official Business
Penalty for Private Use $300
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
S8S
AGENCY
CHICAGO It- 60b04
------- |