Test Procedure for Uranium in Drinking Water: Interlaboratory Collaborative Study

PB83-247239
TEST PROCEDURE FOR URANIUM IN DRINKING WATER
INTERLABORATORY COLLABORATIVE STUDY
Monsanto Research Corp.
Miamisburg, OH
Aug 83
U.S. DEPARTMENT OF COMMERCE
National Technical Information Service

-------
P333-247239
EPA-600/4-83-038
August 1983
TEST PROCEDURE FOR URANIUM IN DRINKING WATER:
Interlaboratory Collaborative Study
by
C. A. Phillips and C. T. Bishop
Environmental Assessment and Planning Section
Mound Facility, Monsanto Research Corp.
Miamisburg, Ohio 45342
DOE Contract No. DE-AC04-76-DP00053
Project Officer

Earl Whittaker
Quality Assurance Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
Contract No. EPA-IAG-79-D-X0736
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.

EPA-600/4-83-038
2.
3. R£f I PINT'S ACCESSION NO
'$3 -.-.-•£
* '
4. TITLE AND SUBTITLE
TEST PROCEDURE FOR URANIUM IN DRINKING WATER:
Interlaboratory Collaborative Study
5. REPORT DATE
August 1983
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
C. A. Phillips and C. T. Bishop
8. PERFORMING ORGANIZATION REPORT NO.
3. PERFORMING ORGANIZATION NAME AND ADDRESS
Monsanto Research Corporation
Mound Facility
P.O. Box 32
Miamisburg, Ohio 45342
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA-IAG-79-D-X0736
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency—Las Vegas, NV
Office of Research and Development
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
13. TYPE OF REPORT AND PERIOD COVERED
Project. 4/80-9/80
14. SPONSORING AGENCY CODE
EPA/600/07
15. SUPPLEMENTARY NOTES
Mound Facility is operated by Monsanto Research Corporation for the U.S. Department of
Energy under Contract No. DE-AC04-76-DP00053.
16. ABSTRACT
An inter!aboratory collaborative study was conducted to test a procedure selectee
for measurement of uranium in drinking water. Drinking water samples containing 8.1,
17.4, and 75.3 pCi/1 were analyzed in triplicate by 18 participating laboratories.

The statistical analyses of the test results gave coefficients of variation for
repeatability (within-laboratory precision) of 14.6, 8.1, and 8.3 percent for the three
samples for an average repeatability precision of 10.3 percent over the uranium
concentration range of 8 to 75 pCi/1. The analyses also gave coefficients of variatior
for reproducibility (combined within- and between-laboratory precision) of 15.3, 14.9,
and 9.1 percent for an average reproducibility precision of 13.1 percent over the
uranium concentration range of 8 to 75 pCi/1.

The accuracy indexes of the test procedure for the three uranium concentrations
was 98.0, 102.6, and 101.9 percent for an average of 100.8 percent over the uranium
concentrations range of 8 to 75 pCi/1.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATl Field/Group
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)

UNCLASSIFIED
21. NO. OF PAGES
50
20. SECURITY CLASS (This page)

UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (R*v. 4-77) PREVIOUS EDITION is OBSOLETE j
-------
NOTICE

This report has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved
for presentation and publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
-------
CONTENTS
Tables iv
Abbreviations and Symbols ... v
Acknowledgements vii

Introduction 1
Conclusions 2
Recommendations 4
Experimental Procedures 6
Analytical test procedure 6
Collaborative test procedures 6
Uranium standard solution preparation 6
Uranium concentrate preparation 6
General procedures 8
Data processing procedures 8
Results and Discussion 13

References 26

Appendices

A. Laboratories Participating in the Uranium in
Drinking Water Collaborative Study 27
B. Analytical Test Procedure "Uranium in Drinking
Water-Radiochemical Method, Method 908.0" 29
C. Data Sheet 36
D. Collaborative Study Instructions 37
E. Questionnaire, Summary on Collaborative Study 39
F. Comments from Participants in Collaborative Study 40
i i i
-------
TABLES

Number Page

1 Uranium in Drinking Water, Replicate Analyses Data ...... 17

2 Uranium in Drinking Water, Precision Summary ......... 20

3 Uranium in Drinking Water, Cell Average (X-,-,-) and
Cell Standard Deviation (S^-) ...... . ......... 21

4 Uranium in Drinking Water, Deviation of Lab Average from
Grand Average (di-i, pCi/1) and Percent Deviation
(X di) ..... • ..................... 22
5 Uranium in Drinking Water, Reduced Deviations,
(e1Jf pCi/D ........................ 23

6 Uranium in Drinking Water, Ratio of Lab Standard
Deviation to Pooled Standard Deviation (k-j,-) ........ 24

7 Uranium in Water, Accuracy Index (A-,-) and
Bias (tj) ............ .... ........... 25
IV
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS

ASTM
CT
c/m
d/m
DOE
EPA
g/cm3
g/i
IAG
kg
MCL
mg/1
nCi/g
MIPDWR
pCi/1
ppm
yCi/g
2S %
The American Society for Testing and Materials
chloride
counts per minute
disintegrations per minute
The Department of Energy
Environmental Protection Agency
grams per cubic centimeter
gram/liter
Interagency Agreement
kilogram
maximum contaminant level
milligram/liter
nanocurie/gram
National Interim Primary Drinking Hater Regulations
picocuries per liter
parts per million
microcurie/gram
Two sigma percent (two standard deviations in percent)
SYMBOLS
%di

n
P
the accuracy index for the sample j level of uranium
concentration
the deviation of Lab i average from the grand average
for sample j
the percent deviation of Lab i average from the grand
average for sample j
the reduced deviation of Lab i average from the grand
average for sample j
the repeatability interval for sample j level
the reproducibility interval for sample j level
the ratio of Lab i average standard deviation (Sij)
to the pooled standard deviation (Sr.) for sample j
the number of replicate analyses '•'
the number of participants in the collaborative study
the standard deviation of the replicate test results
for sample j by Lab i
the standard deviation of between-laboratories
precision for sample j
-------
Sr. — the repeatability (within-laboratory) standard
J deviation for the sample j
$r. — the smoothed repeatability standard deviation for
^ sample j
SR • — the reproducibility (combined within- and between-
J laboratory) standard deviation for sample j
!>R. — the smoothed reproducibility standard deviation for
J sample j
Sx. — the standard deviation of the grand average for
J sample j
tc -- the critical value of the bias test for P
participants
tj — the bias value for sample j
Tn -- the T-test value of the nth replicate, outlier test
— the adjusted within-laboratory standard deviation
-- the coefficient of variation for between-laboratories
precision for sample j
Vr.% ~ the coefficient of variation for repeatability
J (within-laboratory) for sample j
-- the coefficient of variation for reproducibility
(combined within- and between-laboratory) for sample j
— the test result of the replicate h
-- the arithmetic average of all replicate test results
of sample j by Lab i
~ the test result of replicate h of sample j by Lab i
X~j°" — the grand average value for sample j
Yi -- the known value of the sample j uranium concentration
(pCi/1)
VLj%
VRj%
-------
ACKNOWLEDGEMENTS
The authors would like to thank all of those individuals from the
participating laboratories involved in the collaborative study (Appendix A).
Special thanks are due to Richard Velten of the U.S. Environmental Protection
Agency's Environmental Monitoring and Support Laboratory, Cincinnati, Ohio,
who developed the radiochemical method used in this study. Thanks are also
given to A. A. Glosby of Mound Facility for her efforts in sample preparation
and packaging, to V. R. Casella and W. H. Yanko for many helpful discussions,
and to K. N. McLennan for typing the final manuscript.
vii
-------
INTRODUCTION
The National Interim Primary Drinking Water Regulations (NIPDWR) imply
that when a gross alpha measurement of a drinking water sample exceeds 15
picocuries per liter (pCi/1), then an analysis for uranium should be done to
determine the uranium alpha contribution to the gross alpha concentration.
The method listed in the NIPDWR for the measurement of uranium in drinking
water is a fluorometric method which determines uranium in mass units. It is
now known (subsequent to the promulgation of the NIPDWR) that the ratio of
uranium alpha activity to uranium mass concentration in ground water can vary
significantly from that which is common to natural uranium ore deposits.
Therefore, a test procedure for measuring total uranium alpha activity
concentrations is needed that will more adequately relate to drinking water
gross alpha activity. Such a test procedure was selected and tested in an
inter!aboratory collaborative study. This is a report of the results of that
study.

The method of analysis used in this study is a simplified version of a
method that measures the uranium isotopic concentrations in the sample by
alpha spectrometry. This method measures total uranium alpha activity, the
measurement needed for a gross alpha assessment of a drinking water sample.
This method is included in the updated EPA procedures manual, EPA-600/4-
80-032, August 1980, "Prescribed Procedures for Measurement of Radioactivity
in Drinking Water".

The purpose of the collaborative study of the selected test procedure is
to determine what precision and accuracy can be expected by the use of the
procedure by any competent laboratory in the analysis of drinking water
samples for alpha activity contributed by the uranium in the water samples.
-------
CONCLUSIONS

The repeatability precision (within-laboratory precision), reproduci-
bility precision (the combined within- and between-laboratory precision), and
the accuracy have been determined for the test procedure by a multilaboratory
test in this study, and they hereby become criteria by which to evaluate an
alternate test procedure for equivalency.

The estimated repeatability (within-laboratory), single-operator, single
machine, same-day, relative precision of the test procedure for the
determination of total uranium alpha concentrations at the 8.1 pCi/1 level
(averaged over the 18 participants) is ±2.3 pCi/1 (28.4%, 2S%); at the 17.4
pCi/1 level (averaged over the 17 participants) is ±2.9 pCi/1 (16.5%, 2S%);
and at the 75.3 pCi/1 level (averaged over the 18 participants) is ±12.8 pCi/1
(17.0%, 2S%), for an average 2S percent of 20.6 percent for the range of
uranium alpha concentrations of 8 to 75 pCi/1.

The estimated reproducibility (combined within- and between-laboratory),
multi-operator (multilaboratory), single-machine, same-day, relative precision
of the test procedure in the determination of total uranium alpha
concentrations at the 8.1 pCi/1 level (averaged over the 18 participants) is
±2.4 pCi/1 (29.6%, 2S%); at the 17.4 pCi/1 level (averaged over the 17
participants) is ±5.3 pCi/1 (30.4%, 2S%); and at the 75.3 pCi/1 level
(averaged over the 18 participants) is ±14.0 pCi/1 (18.5%, 2S%); for an
average 2S% of 26.2 percent for the range of uranium alpha concentrations of
8 to 75 pCi/1.

In terms of the 95 percent repeatability interval (on the basis of test
error alone) the difference in absolute value, of two test results obtained in
the same laboratory at the 8.1 pCi/1 level of uranium alpha concentration can
be expected to exceed 2.3 pCi/1 only 5 percent of the time; at the 17.4 pCi/1
level, 5.2 pCi/1 only 5 percent of the time; and at the 75.3 pCi/1 level, 22.4
pCi/1 only 5 percent of the time. These average to be 29.3 percent of the
uranium alpha concentration over the range of 8 to 75 pCi/1. When differences
exceed 29.3 percent of the concentration, then one or both of the test results
are suspect.

In terms of the 95 percent reproducibility interval (on the basis of test
error alone), the difference, in absolute value, of two test results obtained
in different laboratories at the 8.1 pCi/1 level of uranium alpha
concentration can be expected to exceed 2.9 pCi/1 only 5 percent of the time;
at the 17.4 pCi/1 level, 6.6 pCi/1 only 5 percent of the time; and at the 75.3
pCi/1 level, 28.5 pCi/1 only 5 percent of the time. These average to be 37.2
-------
percent of the uranium alpha concentration over the range 8 to 75 pCi/1. When
differences exceed the 37.2 percent of the uranium alpha concentration, then
one or both test results are suspect.

The accuracy of the test procedure expressed as the accuracy index in
percent for the three uranium alpha concentration levels are 98.0, 102.6, and
101.9 percent respectively, for an average accuracy index of 100.8 percent
over the range of concentrations of 8 to 75 pCi/1.

The sensitivity of the test procedure is not limited as much by the
chemistry of the procedure as it is by the background and the counting
efficiency of the alpha counting system used. With a 1-liter sample, an alpha
background of 0.1 cpm, a counting time of 100 minutes, and a counting
efficiency of 15 percent (0.15 cpm/dpm), the sensitivity of the test procedure
for uranium alpha would be about 0.3 pCi/1 at the 95 percent confidence level
(using sensitivity as defined in the NIPDWR).

The lack of an adequate carrier with which to determine chemical recovery
in each sample analyzed, and having to use a predetermined recovery factor,
causes some uncertainty in the adjusted test results (adjusted by recovery
factor). However, the predetermined recoveries averaged for the 18
laboratories was 91. ± 15. percent, indicating a relatively high chemical
recovery for the test procedure and therefore minimal uncertainty in the
adjusted test results.

The t-test to show significant differences and method bias showed that
the test procedure did not contain significant systematic errors and therefore
no bias was indicated for uranium alpha concentrations up to 75 pCi/1.
-------
RECOMMENDATIONS

It Is recommended that the test procedure used in this study (described
in detail in Appendix B) be used to monitor drinking water samples for uranium
alpha contribution to the gross alpha activity when the gross alpha activity
exceeds the amount specified in the NIPDWR (15 pCi/1, Section 141.15).

It is recommended that a change be made in the preparation of the
separated uranium fraction for counting the alpha activity. It is also
recommended that the change be verified experimentally to demonstrate the
benefit to the Test Procedure before the change is incorporated into the Test
Procedure. To make that change in the preparation of the separated uranium
fraction for alpha counting, the following changes and additions to the Test
Procedure (Method 908.0) are proposed. Replace existing steps 8.2.5 through
8.2.9 with the following 8.2.5 through 8.2.10' steps.

8.2.5 Elute the uranium with six column volumes of 0.1N_HC1, collecting
the eluate in a 150 ml beaker.

8.2.6 Evaporate the eluate to near dryness, then add 1 ml 12N HC1
(cone.), 10 ml of water, 0.2 ml 20% TiCIs, and 1 ml of lanthanum
carrier solution, stir.

8.2.7 Add 0.5 ml HF (cone.), stir well and allow to stand for 30
minutes.

8.2.8 Filter through 47 mm, 0.2 \im pore membrane filter, collecting the
coprecipitated U/LaF3.

8.2.9 Wash the U/LaF3 with 10 ml of water followed by 10 ml of
ethanol.

8.2.10 Air dry the filter for at least 1 hour before counting for alpha
activity.

Also add the following item to the "Apparatus" section of the procedure.

5.7 0.2 pm pore, 47 mm diameter membrane filter that withstand the
acid treatment in the Test Procedure and will lay flat after
drying (such as Gelman AN-200).

And add the following items to the "Reagents" section of the procedure.

6.15 Lanthanum nitrate, (1.0 mg La+^/ml). Disolve 3.11 g
La(N03)3'6H20 in one liter of 0.1N_ HN03.
-------
6.16 Hydrofluoric acid, 29N : HF (cone.) sp. gr. 1.18, 49%.

6.17 Titanium trichloride, TiCl3 : 20%

The term "column volume" in the test procedure should be clarified. This
can be done by simply giving the milliliters of the resin bed volume specified
in the procedure. A resin bed 1.3 cm in diameter by 8.0 cm high will have a
volume of 10.6 ml. Then in Section 8.2.3 of the procedure, after the
expression "— with 6 column volumes", put in parentheses (6 x 10.6 ml = 63.6
ml, or 65. ml, rounded to the nearest 5 ml).

Since no carrier or tracer is used in the procedure with each sample to
determine chemical recovery, it is recommended that with each set of samples
to be analyzed by this test procedure, a spiked sample (to determine recovery)
and a sample duplicate (to verify precision) be analyzed. This recommendation
should be incorporated into the procedure.
-------
EXPERIMENTAL PROCEDURES
ANALYTICAL TEST PROCEDURE
The analytical test procedure used in this study is described in detail
in Appendix B. The procedure is capable of quantifying all uranium alpha
emitting isotopes in drinking water.

COLLABORATIVE TEST PROCEDURES

Uranium Standard Solution Preparation

An ampoule of uranium standard solution, prepared by the National Bureau
of Standards, was sent to each participant for recovery and counting
efficiency determinations. That solution had a total uranium (U-238, U-235,
and U-234) alpha specific activity of 5,171 dpm/g.

Uranium Concentrate Preparation

Three uranium concentrates were prepared for the collaborative study.
The concentrates were prepared by dilution of a standard uranium solution of
natural isotopic composition prepared by the National Bureau of Standards
(NBS) for the Quality Assurance Division, U.S. Environmental Protection
Agency, Las Vegas, Nevada. The standard solution was one molar in nitric
acid. The calculated isotopic uranium radioactivity concentrations on
November 1978, as specified by the National Bureau of Standards, were
uranium-234, 2,478 ± 129 disintegrations per minute per gram (dpm/g);
uranium-235, 119 ± 0.5 dpm/g, and uranium-238, 2,574 ± 8 dpm/g.

Uranium concentrate #1 was prepared by delivering 0.6910 g of the NBS
stock solution to a tared 1,000-milliliter volumetric flask and diluting to 1
liter with 0.5 M^ HC1. The final weight of the solution was 1,005.4 grams and
the uranium concentration was calculated to be 3.55 ± 0.19 dpm/g. The
specific gravity of the solution was 1.0067 g/ml at 22.2°C based on the
average of two determinations using a calibrated pycnometer. The concentra-
tion of the solution was then calculated to be 3.58 ± 0.19 dpm/ml. A
5-milliliter sample of this solution diluted to 1 liter with uranium-free
water is equivalent to a total uranium concentration of 8.1 ± 0.4 pCi/1.
After complete mixing of the solution, approximately 20 ml portions were
sealed in 25 ml glass ampoules labeled uranium concentrate #1 and sent to the
participants for analysis. Uranium concentrates #2 and #3 were prepared in
the same manner as uranium concentrate #1 and data relative to all three
concentrates are shown as follows:
-------
Cone.
Weight of Total Weight Specific When
Uranium NBS Standard of 0.5 M HC1 Cone. Gravity Cone. Diluted*
Concentrate Standard (g) Solution (g) (dpm/g) (g/ml) (dpm/ml) (pCi/1)
#1

#3
0.6910

1.4924

6.4559
1005.4

1005.1

1005.3
3.55
± 0.19

7.68
± 0.40

33.2
± 1.7
1.0067
1.0068
1.0072
3.58
± 0.19

7.73
± 0.40

33.4
± 1.7
8.1
± 0.4

17.4
± 0.9

75.3
± 3.9
*5 ml of the concentrate diluted to 1 liter.
The concentrations of the three uranium concentrates were verified by
alpha pulse-height analysis comparisons to a uranium-232 standard which had
been previously calibrated with NBS standardized uranium-236. A 1-milliliter
aliquot of uranium-232 standard solution was added to a 1-milliliter sample of
each concentrate. These solutions were mixed, made 8 molar in hydrochloric
acid, and passed through AG1-X4 anion exchange resin to remove daughter
activities which were present. The uranium was eluted with 0.1 M^ HC1 and
electrodeposited on a 0.75-inch diameter stainless steel disk for alpha
pulse-height analysis (APHA) spectrometry. By comparing the net count rate of
the uranium-232 to the net count rates of uranium-234, -235, and -238, the
concentrations of the uranium concentrates were determined. The values of the
concentrates fell within the error limits as derived from the errors specified
by the NBS for the standard solution. The following data summarizes the
results of this verification.
Uranium
Concentrate
#3
Known Concentration
Calculated Fran
Dilution of NBS
Standard
(dpm/ml)

3.58 ± 0.19

7.73 ± 0.40

33.4 ± 1.7
Concentration
Measured by APHA
(dpm/ml)*

3.57 ± 0.12

7.62 ± 0.19

34.4 ± 3.4
% Difference From
Known Value

0.3

1.4

3.0
*Errors are one sigma counting errors only.

7
-------
General Procedures

A total of 25 laboratories agreed to participate in the collaborative
study and 19 laboratories submitted their results. All laboratories who
submitted results are listed in Appendix A. All laboratories were supplied
with the proposed EPA procedure for the analysis of total uranium alpha
activity in drinking water (Appendix B), a Data Sheet (Appendix C), a
Collaborative Study Instruction Sheet (Appendix D), a Questionnaire (Appendix
E), and three unknown reference samples labeled uranium concentrate 1, 2, and
3.

By diluting a 5-milliliter aliquot of each concentrate to 1 liter with
the participant's drinking water, the concentrations of the reference samples
were calculated to be 8.1, 17.4 and 75.3 pCi/1 total uranium alpha activity.

Each participating laboratory was requested to perform triplicate
analyses of a 1-liter sample of their drinking water to determine a blank
value. The authors believed that some of the participants might have had a
high drinking water blank and therefore bias the reference values. A blank
value for this study was therefore defined as the counts per minute of 1 liter
of the participant's drinking water including counter background.

The results of each laboratory were reported on a data sheet which
specified sample identification, volume, gross counts, counting time, counter
background, total uranium alpha activity in pCi/1, average recovery factor
(R), and average counting efficiency (E). All calculated data were checked at
Mound Facility, and any significant differences between the calculated and
reported values were resolved with the individual participants.

DATA PROCESSING PROCEDURES

The data from the 19 participants were tested for outliers by the ASTM
recommended criterion for rejection (ASTM 1980). For this rejection
criterion, with n observations listed in order of increasing value by xj _<
X2 1 X3 .1 ••• .1 xn» ^ tne outlying values xn or xj are in question,
then Tn or Tj were calculated as follows:

Tn = (xn - x)/s

- x)/s
where: Tn; Tj = test criterion

x = arithmetic average of all n values

s = the estimate of the population standard deviation
based on the sample data
-------
If Tn or TI values exceeded the critical value, the measurement in
question was rejected. Critical values of T for a 5 percent two-sided level
of significance were used for the rejection criterion.

A statistical evaluation of the test results was carried out by the
procedures described in E-691, E-177, and E-178 of the ASTM Standard Part 41,
1980. The standard deviations and other statistical parameters, and equations
for their calculations are listed below. The standard deviation of individual
participant (or laboratory) test results, S-jj, was determined by Equation 2.
sij
1/2
(2)
h=l
where: X-jj^ = the result reported for the h replicate of the j
sample material by Lab i

R-jj = the mean of the individual results for sample j for Lab i

nn-j = the number of replicates reported for sample j by Lab i

The repeatability (within-laboratory) standard deviation, Sr., for each
sample was determined by Equation 3 and 3A. J

Where the number of replicates was the same (3) for all participants,
Equation 3 was used.
(3)
where: P = the number of participants in the study.

Where the number of replicates were not the same for all participants,
Equation 3A was used.
(3A)
-------
The standard deviation of the grand average for each sample, S%.,
was determined by Equation 4. J
p

i=]
1/2
(4)
where: R-JJ = the average of the test results for sample
material j by Lab i

Xj = the grand average for sample material j

The standard deviation of between-laboratories precision for each sample
material, SL., was determined by Equation 5.
J
2
- /n
\l/2
(5)
The reproducibility (combined within- and between-laboratory) standard
deviation for each sample, SR., was determined by Equation 6.
S
R = (Srj2 + SLj2)l/2
(6)
The deviation of laboratory averages from the grand average for each sample,
dij, was determined by Equation 7.

The percent deviation of laboratory averages from the grand average for each
sample, % dij, was determined by Equation 8.
= 100
(8)
The estimated standard error of a cell average, calculated on the basis of
replication error only, U j , for sample #1, is determined by Equation 9.
= S
(9)
10
-------
For samples #2 and #3, duplicate analyses were made by one lab. Thus for
samples #2 and #3 the nP value (total replicates) in Equation 9 was replaced
by P
Reduced deviations (e-jj) for each participant for each sample were
determined by Equation 10.
(10)
The ratio (kij) of the standard deviation of participant replicate analyses
(Sjj) to the pooled standard deviation (Sr.) for each sample was
determined by Equation 11. J

kij = Sij/Srj (11)

The coefficient of variation for repeatability (within-laboratory) for each
sample, Vr.%, was determined by Equation 12.

Vrj% = 100 Srj./Xj (12)

The coefficient of variation for between-laboratories precision for each
sample, Vi-%, was determined by Equation 13.
J

vLj% = 100 SLJ/XJ (13)

The coefficient of variation for reproducibility (combined within- and between
laboratory) for each sample, VR.%, was determined by Equation 14.
J

VRj% = 100 SRj/Xj (14)

Smoothed values of the repeatability and reproducibility standard deviations
for each sample, S"r. and SR. were determined by Equations 15 and 16,
respectively. J J

§ri = (vV%)(Kj)/100 (15)
J
SR.J = (Wj/ioo (16)
11
-------
Where 7r% = average of Vr.% for the 3 samples
J
V~R% = average of VR.% for the 3 samples
J
Assuming that the test errors are normally distributed and using a 95 percent
probability level, the 95 percent repeatability interval (Irj) and the 95
percent reproducibility interval (iRj) for each sample were determined using
the smoothed standard deviations in Equations 17 and 18, respectively.

IP- = 2(2)1/2 Sr. = 2.83 Sr. (17)
J J J

IR. = 2(2)1/2 L. = 2.83 Sr.
J J J

The accuracy index (Aj%) a percent relationship of the grand average
(Xj) to the known value (Yj) for each sample material was determined
by Equation 19.

Aj% = 100 Xj/Yj

A comparison of the grand average value (Xj) with the known value
(Yj) for each sample in a bias test was determined by Equation 20 (Youden
and Steiner, 1975).

?. Y-
J J f?ni
tj = ,,y , (P-l) degrees of freedom { u>

where tj = the bias test result for the j sample

P = the number of participants
12
-------
RESULTS AND DISCUSSION

For this collaborative study of the selected test procedure (Appendix B),
uranium in water concentrate solution batches were prepared and a portion was
sent to each participant. Three separate batches were prepared with three
different uranium concentrations. Concentrates were used so that participants
would make specified dilutions with their own drinking water, thereby
subjecting the test procedure to 19 different drinking water sources.

The test results from the 19 participants for the three uranium
concentrations in drinking water are tabulated in Table 1. The participants
are identified by randomly assigned laboratory numbers. Table 1 also lists
the counting error (1 sigma, Su) associated with each test result. The
recovery data in Table 1 is a listing of the three (and sometimes four)
replicate uranium analyses of spiked (with standard uranium) samples by the
test procedure by each laboratory for the purpose of recovery determination.
The recovery factor (average of the replicate tests) was used to correct
analytical results of the three test samples.

The test results of Lab 12 for Sample 2 and Lab 20 for samples 1 and 3
failed the laboratory outlier test at the 5 percent significance level (ASTM,
E-178, 1980) and were therefore rejected (not used in the precision and
accuracy calculations). The test results of Lab 20 for Sample 2 did not fail
the outlier test but was found to be significantly different from the grand
average by the t-test for significant differences (Youden and Steiner, 1975)
and was therefore rejected (not used in the precision and accuracy
calculations).

Table 2 gives the known value (Yj) and grand average (Rj) for
each sample and the following precision parameters.

Sj»., the standard deviation of the grand average for the j sample
J
Sr., the pooled standard deviation (within-laboratory) for the j sample
J
St., the square root of the component of variance between-!aboratories,
CO
Lj 5 for the j sample

SR., an estimate of the between-1aboratories precision, a combined
J within- and between-laboratory precision, or reproducibility precision
for the j sample
13
-------
Vr.%, the coefficient of variation for repeatability, or the estimated
J relative standard deviation for within-laboratory precision, or
repeatability index for the j sample

V|_.%, the coefficient of variation for between-laboratories precision for
J the j sample

Vp.%, the coefficient of variation for reproducibility, the percent
J coefficient of variation for the total variability of a single test
result including both within- and between-laboratory variability, or
reproducibility index for the j sample

S"r., the smoothed value of Sr.

So., the smoothed value of SR.
J J
Ir., the repeatability interval—the maximum permissible difference due to
J test error between two test results obtained in a single laboratory (or
by a single participant) at the j sample level

ID., the reproducibility interval—the maximum permissible difference due
J to test error and systematic error between two test results obtained in
two different laboratories at the j sample level

Table 3 lists, by participant, the average and the standard deviation of
the participant test results for each of the three sample materials (cells).
Table 3 also lists the Sr. and an adjusted Sr. value designated by Uj for
for each sample material.J J

Table 4 lists, by participant, the deviation and the percent deviation of
the participant average from the grand average for each sample material and
the standard deviation of the grand average for each sample material,

S*j'
Table 5 lists, by participant, the reduced deviations (e-j) for each
sample material.

Table 6 lists, by participant, the ratio (k-j) of the participant test
results standard deviation (Sjj) to the pooled standard deviation (Sr-)
for each sample material. This table points out potential outliers. JSample 1
for Participant 8, Sample 2 for Participant 2, and Sample 3 for Participant
14, show ratio values higher than the critical value for 17 or 18 participants
and 3 replicates (2.06). Since the average of those participant and sample
test results were not significantly different from the respective grand
average, those test results were not rejected.

For each of the three sample materials, Table 7 lists the known value
(Yj, pCi/1); the grand average, (Kj, pCi/1); the accuracy index, (Aj, %);
the standard deviation of the grand average, (S^, pCi/1); the bias
value, (tj); the critical value for significance for 17 and 18 participants,
(tc); and the number of participants in the study, (Pj). The table also
14
-------
shows the average accuracy index for the test procedure (100.8%). Table 7
shows no bias in the test procedure for the range of uranium concentrations
tested (8 to 75 pCi/1).

The average of the recoveries for the 18 participants is 91 ± 15 percent,
indicating that the chemistry of the test procedure is quite accurate and
reliable.

From Table 2 it can be seen that the standard deviations (Srj, SM,
and SRJ) depend on the level of uranium concentration. However, the percent
coefficients of variation as calculated from those standard deviations (Vr.
%, VM %, and VRJ %) show a rather fixed percent relationship to all threej
levels of uranium concentration.

The repeatability (within-laboratory) and reproducibility (total
between-laboratory) intervals (Irjand IRJ) for each sample material were
calculated with smoothed Srj and SRJ values (Sri and SRJ, respectively). The
values listed in Table 2 were calculated at the 95 percent probability level
and are thereby designated as the 95 percent repeatability and 95 percent
reproducibility intervals. The 95 percent repeatability interval is that
difference value based on test error alone between two test results obtained
by the same participant (or laboratory) that would be exceeded only about 5
percent of the time. The 95 percent reproducibility interval is that
difference value between two test results obtained by two different partici-
pants (or laboratories) that would be exceeded only about 5 percent of the
time. When difference values exceed the interval values, one or both test
results are suspect. The study shows the estimated average 95 percent
repeatability interval to be 29.3 percent of the uranium concentration level
over the range of 8 to 75 pCi/1. The estimated average 95 percent reproduci-
bility interval is shown to be 37.2 percent over the same uranium concentra-
tion range. It should be realized that those interval values are based on a
limited 18 participant average.

In the questionnaire sent to all of the participants in the study,
Question 5 asked, "Do you believe that this would be a good reference method
for uranium in drinking water?" Fifteen participants answered "yes", two said
"no", and two were undecided. Appendix E is a summary of responses to the
questionnaire.

Appendix F is a listing of the comments made by the participants about
the test procedure. Responses to Questions 2, 5, 6, and 7 are addressed in
the Recommendations section of this report. In response to Question 3 one
participant commented that the method of transfer of the uranium eluant from
the beaker to the counting planchet was questionable. Such transfers are made
in many radiochemical methods and should not cause serious problems. In
response to Question 4 one participant suggested that new ion exchange resin
should be specified for each analysis rather than regenerating the resin.
This can and should be done if the analyst does not have confidence in the
regenerating procedure. The regenerating procedure should completely
regenerate the resin from drinking water levels of uranium concentrations.
Re-use of the resin is not recommended if high levels of uranium concentration
(>500 pCi/1) have been put through the resin bed. In response to Question 10

15
-------
one participant commented that the uranium yields would vary with the
dissolved solids in the water sample. High carbonate content is usually
associated with high dissolved solids and carbonate does complex uranium and
tends to prevent its hydroxide precipitation. The test procedure addresses
this problem adequately. The 19 different water supplies used in this
collaborative study suggest that uranium yield is not tied to the dissolved
solids content when this test procedure is followed.
16
-------
TABLE 1. URANIUM IN DRINKING WATER, REPLICATE ANALYSES DATA
Lab
1
2
3
4
5
7
8
9
10
Sample Uranium
1
8.67 CO. 335)
9.42 (0.349)
9.18 (0.345)
10.46 (0.641)
7.85 (0.558)
7.69 (0.553)
7.02 (0.432)
7.69 (0.450)
7.97 (0.458)
9.12 (0.809)
8.23 (0.771)
8.02 (0.761)
8.17 (0.298)
7.89 (0.293)
7.62 (0.288)
6.13 (0.367)
6.40 (0.373)
7.11 (0.388)
3.97 (0.888)
11.01 (1.341)
8.28 (1.186)
8.31 (0.742)
7.05 (0.689)
7.17 (0.694)
9.22 (0.602)
8.87 (0.591)
8.95 (0.594)
Concentration, (counting
2
22.76 (0.537)
22.51 (0.534)
19.57 (0.499)
25.01 (0.981)
19.94 (0.877)
18.72 (0.851)
13.66 (0.587)
13.33 (0.580)
12.05 (0.553)
. 14.62 (1.016)
15.44 (1.043)
(15.0)
17.61 (0.432)
15.59 (0.407)
17.18 (0.426)
17.71 (0.681)
18.52 (0.690)
18.86 (0.693)
21.05 (1.800)
21.92 (1.834)
22.92 (1.873)
17.22 (1.043)
18.73 (1.086)
16.68 (1.027)
17.88 (0.833)
17.49 (0.824)
17.53 (0.825)
error), a pCi/1
3
93.03 (1.081)
84.07 (1.027)
75.21 (0.972)
92.06 (1.872)
74.13 (1.680)
86.83 (1.818)
83.88 (1.418)
75.37 (1.345)
76.86 (1.358)
73.38 (2.252)
76.47 (2.298)
(74.9)
74.39 (1.242)
79.29 (1.282)
75.48 (1.251)
74.84 (1.133)
82.85 (1.183)
73.72 (1.126)
64.98 (3.090)
75.32 (3.321)
80.48 (3.431)
76.60 (2.160)
69.25 (2.055)
71.24 (2.083)
70.16 (1.642)
72.65 (1.671)
66.52 (1.599)
Recovery &
Percent
81.43
76.23
91.08
78.2
69.1
80.0
101.0
97.3
106.0
82.5
79.5
102.5
97.6
96.1
83.0
83.0
89.9
45.0
36.0
41.0
115.5
104.9
108.3
99.0
96.0
99.0
(continued)
See footnotes at end of table, page 18.

17
-------
TABLE 1. (Continued)
Sample Uranium Concentration, (counting error), a pCi/1
Lab
11
12
14
18
20
21
22
23
1
8.16 (0.525)
6.76 (0.482)
7.41 (0.503)
8.38 (0.239)
7.70 (0.230)
6.78 (0.217)
7.54 (0.721)
8.46 (0.755)
8.84 (0.769)
6.97 (0.662)
9.11 (1.141)
7.12 (0.668)
5.29 (0.348) e
4.33 (0.316) e
5.47 (0.354) e
7.10 (0.367)
6.18 (0.344)
6.65 (0.356)
8.08 (0.436)
8.44 (0.446) .
7.44 (0.270)
7.98 (0.309)
7.98 (0.309)
8.01 (0.309)
2
14.67 (0.691)
17.05 (0.743)
15.72 (0.715)
2.79 (0.148) d
3.23 (0.157) d
3.14 (0.156) d
17.57 (1.032)
16.21 (0.995)
19.41 (1.079)
18.83 (1.018)
17.82 (1.550)
20.09 (1.640)
11.96 (0.518) e
12.20 (0.524) e
10.33 (0.483) e
15.95 (0.536)
17.25 (0.874)
17.60 (0.883)
16.73 (0.620)
18.10 (0.645)
17.02 (0.626)
16.24 (0.438)
17.30 (0.451)
17.24 (0.451)
Recovery b
3 Percent
83.22 (1.614)
88.80 (1.666)
65.67 (1.435)
76.84 (0.696)
77.85 (0.701)
77.35 (0.699)
81.61 (2.130)
92.51 (2.265)
64.37 (1.898)
73.28 (1.946)
80.15 (3.206)
78.64 (3.176)
56.64 (1.122) d
68.99 (1.731)
77.57 (1.835)
70.54 (1.750)
77.83 (1.073)
85.21 (1.120)
76.72 (1.065)
69.18 (0.898)
70.74 (0.908)
71.50 (0.791)
96.9
92.6
102.9
91.6
90.3
89.9
77.6
82.9
109.0
129.0
95.3
81.3
97.2
109.6
106.0
99.9
91.8
87.9
105.3
90.1
76.8 c
93.1
92.6
96.9
95.6
(continued)
See footnotes at end of table, page 18.
1R
-------
TABLE 1. (Continued)
Sample Uranium Concentration, (counting error), a pCi/1
Lab
24
25
1
8.18 (0.602)
6.23 (0.528)
7.12 (0.563)
9.17 (0.623)
7.47 (0.567)
10.83 (0.672)
2
21.72 (0.921)
21.99 (0.927)
17.88 (0.837)
18.42 (0.863)
16.52 (0.820)
14.70 (0.776)
3
70.50 (1.761)
69.43 (1.748)
71.14 (1.769)
81.44 (1.784)
83.91 (1.811)
85.31 (1.826)
Recovery D
Percent
101.7
91.6
104.3
93.7
' 93.0
111.1
a Su, counting uncertainty, pCi/1, one sigma.

b R, chemical recovery using a spiked (standard) sample, the average of the
three replicates for each lab was used as factor (R) in calculating the
uranium concentrations in samples 1,2, and 3. The recoveries given here do
not correspond to the sample data listed, but were determined independently
with spiked samples by the test procedure.

c This chemical recovery factor was used only in calculating the uranium
concentration for sample 3. The average of the other three values was used
to calculate the uranium concentration of samples 1 and 2. '

d Rejected by the ASTM Outlier Test.

e Rejected by t-test for significant differences (Youden and Steiner, 1975).
19
-------
TABLE 2. URANIUM IN DRINKING WATER, PRECISION SUMMARY
Parameter a
Yj (pCi/1)
Xj (pCi/1)
SXj (PCi/D
Srj (pCi/1)
SLj (PCi/D
SRj (pCi/1)
vrjx
V
VRj%
5rj (pCi/1)
sRj (pci/i)
Irj (pCi/l)
iRj (PCI/D
p

1
8.1 ± .4
7.9
0.76
1.15
0.37
1.21
14.6
4.7
15.3
0.82
1.04
2.31
2.93
18
Sample
2
17.4 ± .9
17.9
2.39
1.44
2.24
2.66
8.1
12.5
14.9
1.84
2.34
5.22
6.63
17

3
75.3 ± 3.9
76.8
4.6
6.39
2.83
6.98
8.3
3.7
9.1
7.92
10.06
22.43
28.47
18
Average

10.3
7.0
13.1

a Terms defined in text and the List of Abbreviations and Symbols.
20
-------
TABLE 3. URANIUM IN DRINKING WATER, CELL AVERAGE (Xjj)
AND CELL STANDARD DEVIATION (S)
Sample
1

*„ S,j *„ Sll *,-J S,j
1 9.1
2 8.7
3 7.6
4 8.4
5 7.9
7 6.5
8 7.7
9 7.5
10 9.0
11 7.4
12 7.6
14 8.3
18 7.7
21 6.6
22 8.0
23 8.0
24 7.2
25 9.2
Xj 7.9
Srj 1.15
Uj 0.65
0.38
1.56
0.49
0.58
0.28
0.51
3.55
0.70
0.18
0.70
0.80
0.67
1.19
0.46
0.51
0.02
0.98
1.68

21.6
21.2
13.0
15.0
16.8
18.4
22.0
17.5
17.6
15.8
0
17.7
18.9
16.9
17.3
16.9
20.5
16.5
17.9
1.44
0.82
1.77
3.34
0.85
0.58
1.06
0.59
0.94
1.06
0.21
1.19
0
1.60
1.14
0.87
0.72
0.60
2.30
1.86

84.1
84.3
78.7
74.9
76.4
77.1
73.6
72.4
69.8
79.2
77.3
79.5
77.4
72.4
79.9
70.5
70.4
83.6
76.8
6.39
3.62
8.91
9.22
4.54
2.18
2.57
4.98
7.89
3.80
3.08
12.07
0.50
14.19
3.61
4.57
4.61
1.18
0.86
1.96

0.76 2.39 4.65
9.1
-------
TABLE 4. URANIUM IN DRINKING WATER, DEVIATION OF LAB AVERAGE FROM
GRAND AVERAGE (djj, pCi/1) AND PERCENT DEVIATION (% d^)
Sample
Lab
1
2
3
4
5
7
8
9
10
11
12
14
18
21
22
23
24
25

dij
1.2
0.8
-0.3
0.5
0.0
-1.4
-0.2
-0.4
1.1
-0.5
-0.3
0.4
-0.2
-1.3
0.1
0.1
0.7
1.3
1
* dij
15.2
10.1
3.8
6.3
0.0
17.7
2.5
5.1
13.9
6.3
3.8
5.1
2.5
16.4
1.3
1.3
8.9
16.4

-11
3.8
3.4
-4.8
-2.8
-1.0
0.6
4.2
-0.3
-0.2
-2.0
0
-0.1
1.1
-0.9
-0.5
-0.9
2.7
-1.3
2
% dij
21.3
19.1
27.0
15.7
5.6
3.4
23.6
1.7
1.1
11.2
0
0.6
6.2
5.0
2.8
5.0
15.2
7.3

dij
7.3
7.5
1.9
-1.8
-0.4
0.3
-3.2
-4.4
-7.0
2.4
0.5
2.7
0.6
-4.4
3.1
-6.3
-6.4
6.8
3
%dij
9.5
9.8
2.5
2.4
0.5
0.4
4.2
5.7
9.1
3.1
0.6
3.5
0.8
5.7
4.0
8.2
8.3
8.8
0.58
5.72
21.59
0.76
2.39
4.65
22
-------
TABLE 5. URANIUM IN DRINKING WATER, REDUCED
DEVIATIONS, (e-jj a, pCi/1)
Sample
Lab 1
1 1.9
2 1.2
3 -0.5
4 0.9
5 0.0
7 -2.1
8 -0.2
9 -0.6
10 1.7
11 -0.7
12 -0.4
14 0.6 '
18 -0.2
21 -1.9
22 0.1
23 0.1
24 1.1
25 1.9
Uj 0.65
Sr. 1.15
2
4.6
4.1
-6.0
-3.5
-1.3
0.6
5.0
-0.4
-0.3
-2.5
0
-0.2
1.3
-1.2
-0.7
-1.2
3.3
-1.6
0.82
1.44
3
2.0
2.1
0.5
-0.5
-0.1
0.1
-0.9
-1.2
-1.9
0.7
0.2
0.8
0.2
-1.2
0.9
-1.7
-1.8
1.9
3.62
6.38
a Term defined in text and the List of Abbreviations and Symbols.
-------
TABLE 6. URANIUM IN DRINKING WATER, RATIO OF LAB STANDARD
DEVIATION TO POOLED STANDARD DEVIATION, kj a
Lab
1
2
3
4
5
7
8
9
10
11
12
14
18
21
22
23
24
25

1
0.33
1.35
0.42
0.51
0.24
0.44
3.08
0.60
0.16
0.61
0.70
0.58
1.04
0.40
0.44
0.01
0.85
1.46
Sample
2
1.23
2.31
0.59
0.40
0.74
0.41
0.65
0.74
0.15
0.83
-
1.11
0.79
0.60
0.50
0.41
1.59
1.29

3
1.40
1.44
0.71
0.34
0.40
0.78
1.24
0.60
0.48
1.89
0.08
2.22
0.56
0.72
0.72
0.18
0.14
0.31
a Term defined in text and the List of Abbreviations and Symbols.

kc = 2.06 for 18 labs and 3 replicates
-------
TABLE 7. URANIUM IN WATER, ACCURACY INDEX (Aj) AND BIAS (tj)

Sample
Parameter a 1 2 3 Average

Yj (pCi/1) 8.1 ± .4 17.4 ± .9 75.3 ± 3.9

Xj (pCi/1) 7.9 17.9 76.8

Aj % 98.0 102.6 101.9 100.8

S*- (pCi/1) 0.76 2.39 4.65

tj -0.90 0.79 1.30

tc 2.10 2.11 2.10

P 18 17 18

a Terms defined in text and the List of Abbreviations and Symbols.
-------
REFERENCES
American Society for Testing and Materials, 1980 Annual Book of Standards,
Part 41, Designation: E177-71 (Reapproved 1980) p. 195; E178-80, p. 206;
and E691, p. 959.

Youden, W. J. and E. A. Steiner, "Statistical Manual of the Association of
Official Analytical Chemists", Association of Official Analytical
Chemists, Washington, DC, 1975.

Krieger, H. L., "Interim Radiochemical Methodology for Drinking Water",
EPA-600/4-75-008 (Revised) U.S. EPA Cincinnati, Ohio, 1976.

Krieger, H. L. and E. L. Whittaker, "Prescribed Procedures for Measurement of
Radioactivity in Drinking Water", EPA-600/4-80-032 August 1980, U.S. EPA
Cincinnati, Ohio, 1980.

"National Interim Primary Drinking Water Regulations". U.S. EPA, Office of
Water Supply, 401 M Street S.W., Washington DC, EPA-570/9-76-003.

Bishop, C. T., V. R. Case!la and A. A. Glosby, "Radiometric Method for
Determination of Uranium Water: Single-Laboratory Evaluation and
Interlaboratory Collaborative Study", EPA-600/7-79-093, U.S.
Environmental Protection Agency, Las Vegas, NV, 1979.
-------
APPENDIX A
LABORATORIES PARTICIPATING IN THE
URANIUM IN DRINKING WATER COLLABORATIVE STUDY
Alabama State Department of Public Health
Montgomery, Alabama

Arkansas Department of Health
Little Rock, Arkansas

California Department of Health Services
Sanitation and Radiation Laboratory
Berkeley, California

Florida Department of Health and Rehabilitative Services
Radiological Health Services
Orlando, Florida

Hazen Research, Inc.
Golden, Colorado

Massachusetts Department of Environmental Quality Engineering
Lawrence Experiment Station
Lawrence, Massachusetts

Michigan Department of Public Health
Division of Radiological Health
Lansing, Michigan

New Hampshire Water Supply and Pollution Control Commission
Concord, New Hampshire

Oregon Department of Human Resources
Health Division Radiation Section
Portland, Oregon

Orlando Laboratories, Inc.
Orlando, Florida

Rockwell Hanford Operations
Richland, Washington
27
-------
U. S. Army Environmental Hygiene Agency
Aberdeen Proving Ground, Maryland

U. S. Environmental Protection Agency
Environmental Monitoring and Support Laboratory
Cincinnati, Ohio

U. S. Environmental Protection Agency
Eastern Environmental Radiation Facility
Montgomery, Alabama

U. S. Food and Drug Administration
Winchester Engineering and Analytical Center
Winchester, Massachusetts

University of Pittsburgh
Graduate School of Public Health
Pittsburgh, Pennsylvania

Union Carbide Corporation
Nuclear Division
Oak Ridge, Tennessee

Union Carbide Corporation
Nuclear Division, Analytical Dept.
Paducah, Kentucky

Washington State Department of Social and Health Services
Seattle,.Washington
28
-------
APPENDIX B

URANIUM IN DRINKING WATER-RADIOCHEMICAL METHOD
METHOD 908.0
MAY 8, 1980

1. Scope and Application

1.1 This method covers the measurement of total uranium alpha particle
activity in drinking water. Most drinking water sources, especially
ground water sources, contain soluble carbonates and bicarbonates
which complex and keep uranium in the water in solution.

1.2 Uranium isotopic abundances in drinking water sources are apt to be
present in ratios different from the ratios in the deposits from
which the uranium entered the water sources. The two predominant
natural alpha emitting isotopes of uranium are uranium-234 and •
uranium-238. Uranium-238 is the predominant mass abundant isotope;
greater than 99% compared to about 0.006% for uranium-234. However,
uranium-234 has a specific activity for alpha particle emission that
is 1.8 x 101* times greater than that of uranium-238. For an equili-
brium condition, the activity of the uranium-234 is equal to that of
the uranium-238. Therefore, the uranium mass concentration in water
is not related to the alpha particle activity of the water.

1.3 The Drinking Water Regulations under the Safe Drinking Water Act, PL
93-523, require a measurement of uranium for drinking water samples
that have a gross alpha activity greater than 15 pCi/1. A mass
uranium concentration measurement of a water sample cannot be
converted to uranium alpha activity without first analyzing for
isotopic abundances. Therefore, a method such as this one is needed
to determine the total uranium alpha activity of the sample, without
doing an isotopic uranium analysis.

2. Summary of Method

2.1 The water sample is acidified by adding HC1 and the sample is boiled
to eliminate carbonate and bicarbonate ions. Uranium is coprecipi-
tated with ferric hydroxide and separated from the sample. The
uranium is then separated from other radionuclides which were
carried down with the ferric hydroxide by dissolving the hydroxide
precipitate in 8N^ HC1; putting the solution through an anion
exchange column; washing the column with 8j^ HC1; and finally,
eluting the uranium with 0.1N HC1. The uranium eluate is evaporated
and the uranium chemical form is converted to nitrate. The residue
is transferred to a stainless steel planchet, dried, flamed, and
counted for alpha particle activity.
-------
2.2 Uranium recovery is determined with blank samples spiked with known
amounts of uranium and taken through the procedure as a regular
sample.

2.3 Counting efficiency is determined by transferring measured aliquots
of a uranium standard to a planchet, diluting with 6-8 ml of a 1
mg/ml HI03 solution in 4N^HN03, evaporating to dryness, flaming the
planchet, and counting in an alpha counter.

3. Sample Handling and Preservation

3.1 Although carbonate ions in a water sample will help to keep uranium
in solution, the addition of extra carbonate or bicarbonate ions to
the sample as a preservative is not recommended because an increased
carbonate concentration in the sample may cause some precipitation.
Therefore, it is recommended that samples be preserved with HC1 to
pH 2 at the time of collection.

3.2 A sample size of, at least, 1 liter should be collected for uranium
analysis.

4. Interferences

4.1 The only alpha-emitting radionuclide that may come through the
chemistry and cause interference would be protactinium-231.
However, protactinium-231 results from the decay of uranium-235, a
low abundance natural isotope of uranium, and would therefore cause
only a very small interference.

4.2 Since uranium is a naturally occuring radionuclide, reagents must be
checked for uranium contamination by analyzing a complete reagent
blank by the same procedure as used for the samples.

5. Apparatus

5.1 Gas-flow proportional counting system or

5.2 Scintillation detection system.

5.3 Electric hot plate.

5.4 Ion exchange column: approximately 13 mm (i.d.) x 150 mm long with
a 100 ml reservoir.

5.5 Stainless steel counting planchets, 2 inch diameter by 1/4 inch
deep.

5.6 Millipore filter apparatus, 47 mm.
30
-------
6. Reagents

6.1 All chemicals should be of reagent grade, or equivalent, whenever
they are commercially available.

6.2 Ammonium hydroxide, 6N; Mix 2 volumes 15f[ NH^OH (cone.) with 3
volumes of water (carbonate-free).

6.3 Anion exchange resin - Strongly basic, styrene, quaternary ammonium
salt, 4% crossl inked, 100-200 mesh, chloride form (such as Dowex
1x4, or equivalent).

6.4 Ferric chloride carrier, 20 mg Fe+3/ml: Dissolve 9.6 g of FeCl3.6
H20 in 100 ml of 0.5N. HC1.

6.5 Hydriodic acid: HI (cone.), sp. gr. 1.5, 47%.

6.6 Hydrochloric acid, 12N; HC1 (cone.), sp. gr. 1.19, 37.2%.

6.7 Hydrochloric acid, 8Nk Mix 2 volumes 12f[ HC1 (cone.) with 1 volume
of water.
6.8 Hydrochloric acid, 6H_: Mix 1 volume 12f^ HC1 (cone.) with 1 volume
of water.

6.9 Hydrochloric acid, 0.1N_: Mix 1 volume 0.5f[ HC1 with 4 volumes of
water.

6.10 lodic acid, 1 mg/ml : Dissolve 100 mg HIO in 100 ml 4N. HN03.

6.11 Nitric acid, 16^: HN03 (cone.), sp. gr. 1.42, 70.4%

6.12 Nitric acid, 4N_: Mix 1 volume \6H_ HN03 (cone.) with 3 volumes of
water.

6.13 Sodium hydrogen sulfite, 1% in HC1: Dissolve 1 g NaHSO, in 100 ml
6N. HC1 .

Calibrations

7.1 Determine a counting efficiency (E), for known amount of standard
uranium (about 1000 dpm) evaporated from a 6-8 ml volume of a 1
mg/ml HI03 solution in a 2-inch planchet. If the standard solution
is an HC1 solution, then aliquot portions of that solution must be
converted to nitrate/HN03 solutions, eliminating all chloride ions
from the solutions. This can be done by three successive
evaporations after adding 5 ml portions of 16H_ HN03 to aliquot
portions of the standard in small beakers (avoiding dry baking of
the evaporated residue). The final solutions of the standard
aliquots are made by adding 2 ml 4IN HN03 solution to the third
evaporated residues. Transfer the uranium standard aliquot
solutions to 2 inch diameter stainless steel planchets.

31
-------
Complete the transfer by rinsing the beakers two times with 2 ml
portions of 4N[ HN03 and evaporate to dryness. Flame the planchets
and count for, at least, 50 minutes for alpha particle activity. A
reagent blank should be run along with the standard aliquots and
should be alpha counted.
A B'
Efficiency, cpm/dpm, (E) = c Eq- 1

where:
A = gross cpm for standard
B1 = cpm for instrument background
7.2 C = dpm of standard used

A uranium recovery factor R, is determined by the following
procedure: Spike one liter tap water samples with aliquots of
uranium standard solution (500-1000 dpm per sample). Take these
spiked samples and a tap water blank through the entire procedure
and count the separated and evaporated uranium for alpha particle
activity.
(F B}
Recovery factor, (R) = * r °' Eq. 2
C E
where:
C = dpm of uranium standard added
F = gross cpm of spiked sample
B = cpm of reagent blank
E = efficiency factor, cpm/dpm
8. Procedure
8.1 Measure the volume of approximately one liter of the water sample to
be analyzed.

8.1.2 If the sample has not been acidified, add 5 ml 12N. HC1 and 1
ml ferric chloride carrier.

8.1.3 Mix the sample completely and use pH paper to check the
hydrogen-ion concentration. If the pH is >1, add 121^ HC1
until it reaches this value.

8.1.4 Cover with a watch glass and heat the water sample to boiling
for 20 minutes.

8.1.5 The pH must be checked again after boiling and if it is >1,
12H_ HC1 must be added to bring the pH back to 1.

8.1.6 While the sample is still boil ing,gently add 6N. NH^OH to the
sample from a polyethylene squeeze bottle with the bottle
delivery tube inserted between the watch glass and the
pouring lip of the beaker. The boiling action of the sample
provides sufficient stirring action. Add 6f^ NHi+OH until
turbidity persists while boiling continues; then add an

32
-------
additional 10 ml, (estimated addition from the squeeze
bottle).

8.1.7 Continue to boil the sample for 10 minutes more; then set it
aside for 30 minutes to cool and settle.

8.1.8 After the sample has settled sufficiently, decant and filter
the supernate through a 47 mm 0.45 micron membrane filter,
using the larger millipore filtering apparatus.

8.1.9 Slurry the remaining precipitate, transfer to the filtering
apparatus and filter with suction.

8.1.10 Place the filtering apparatus over a clean 250 ml filtering
flask, add 25 ml 8J\[ HC1 to dissolve the precipitate, and
filter the solution.

8.1.11 Add another 25 ml 8N^ HC1 to wash the filter, and then filter.

8.1.12 Transfer solution to the 100 ml reservoir of the ion exchange
column.

8.1.13 Rinse the side arm filtering flask twice with 25 ml portions
of SN^HCl. Combine in the ion exchange reservoir.

8.2 Anion Exchange Separation

8.2.1 Prepare the column by slurrying the anion exchange resin with
8N^ HC1 and pouring it onto a column about 13 mm inside
diameter. The height of the resin bed should be about 80
mm.

8.2.2 Pass the sample solution through the anion exchange resin
column at a flow rate not to exceed 5 ml/minute.

8.2.3 After the sample has passed through the column, elute the
iron, (and plutonium if present), with 6 column volumes of BN^
HC1 containing 1 ml 47% HI per 9 ml of 8N. HC1 (freshly
prepared).

8.2.4 Wash the column with an additional two column volumes of QH_
HC1.

8.2.5 Elute the uranium with six column volumes of 0.11^ HC1.

8.2.6 Evaporate the acid eluate to near dryness and convert the
residue salts to nitrates by three successive treatments with
5 ml portions of 16f[HN03, evaporating to near dryness each
time.
8.2.7 Dissolve the residue (may be very little visible residue) in

33
2 ml 4N_HN03.
-------
8.2.8 Transfer the residue solution, using a Pasteur pipet, to a
marked planchet, and complete the transfer by rinsing the
sample beaker three times with 2 ml portions of 4N^ HN03.

8.2.9 Evaporate the contents in the planchet to dryness, flame to
remove traces of HI03, cool, and count for alpha particle
activity.

8.3 Column Regeneration

8.3.1 Pass three column volumes of 1% NaHS03 in 6N. HC1 through the
column.

8.3.2 Pass six column volumes of 61^ HC1 through the column.

8.3.3 Pass three column volumes of water through the column.

8.3.4 Pass six column volumes of 8N^ HC1 through the column to
equilibrate and ready the resin for the next set of samples.

9. Calculations

Uranium alpha activity, pCi/1 = ^ J ? o°S° Eq. 3
£.•£.£. X t K V

where:
S = gross cpm for sample
B = cpm of reagent blank
V = volume of sample used, ml
E = efficiency, cpm/dpm
R = recovery factor
2.22 = conversion factor for dpm/pCi

10. Precision and Accuracy

In a single laboratory test of this method, a stock uranium solution was
prepared using tap water and spiked with an NBS uranium standard. The
calculated concentration was 26.7 pCi/1. This stock solution was
acidified with HC1 as a preservative. Nine 1-liter aliquots were
withdrawn and the procedure tested. Individual results were 22.4, 22.5,
24.0, 25.9, 26.9, 26.5, 24.6, 25.7 and 23.9 pCi/1. The average
concentration was 24.7 pCi/1 with a standard deviation of 1.7 pd'/l«
From these data, the method shows a negative 7.4% bias and a precision of
±6.7% without the correction of the recovery factor.
34
-------
REFERENCES

1. Bishop, C. T., et al. "Radiometric Method for the Determination of
Uranium in Water", EPA 600/7-79-093, EMSL-LV, April 1979.

2. Edwards, K. W. "Isotopic Analysis of Uranium in Natural Waters by Alpha
Spectrometry", Radiochemical Analysis of Water, Geological Survey Water -
Supply Paper 1696-F, U. S. Government Printing Office, Washington, D.C.,
1968.
35
-------
APPENDIX C

URANIUM IN DRINKING WATER
COLLABORATIVE TEST - MAY, 1980
DATA SHEET
LABORATORY:

CONTACT PERSON:

DATE:
Sample
ID
Vol
(ml)
Gross3
Counts
Counting
Time
(min.)
Counter
Bkg
(c/m)
Bkg
Counting
Time
(min.)
Uranium Alpha Activity15
pC1/l
(In sample or blank)
1 a

1 b

1 c
2 a

2 b

2 c
3 a

3 b

3 c
Blank a

Blank b

Blank c
Recovery Factor (R), Determinations in %

Average of (R) %

Counting Efficiency (E), Determinations in %

Average of (E) %
aTotal counts including background.

bAs calculated from the equation in section 9 of the procedure. For the purpose of this
collaborative study, the reagent blank B is defined as the cpm observed in one liter of your
drinking water including the counter background.

PLEASE SEND RESULTS TO C. A. PHILLIPS, MONSANTO RESEARCH CORPORATION, MOUND FACILITY, MIAMISBURG,
OH 45342, BY JULY 18, 1980. FTS 774-3228 (or 3927) or 513-865-3228 (or 3927)
36
-------
APPENDIX D

COLLABORATIVE STUDY INSTRUCTIONS - URANIUM IN DRINKING WATER
(May 1980)

1. Use the procedure "Uranium in Drinking Water - Radiochemical Method,
Method 908.0" dated May 1980.

2. If you have not used a procedure similar to the one enclosed, it would be
advisable to analyze a few known samples, before analyzing the test
samples.

3. If you do not have the ion exchange resin, or other reagents needed for
the analysis of the test samples, Mound Facility will furnish them if you
desire.

4. Use care in opening the glass ampoules which contain the sample
concentrates and are labeled uranium concentrate 1, 2 and 3. Each
ampoule contains a solution of known concentration of uranium-238 +
uranium-234 in 0.5M^ HC1. The analyst should pipet exactly 5 ml of the
concentrate and dilute this volume to one liter with drinking water in a
volumetric flask. Each sample should be prepared in triplicate from each
uranium concentrate.

5. The diluted one-liter samples just prepared are the samples to be
analyzed. For example, the 5 ml aliquot of uranium concentrate #1
diluted to one liter with drinking water is sample #1. The analyst
should therefore calculate the concentration of this diluted sample.

6. The above samples are listed as 1, 2, 3 on the enclosed Uranium In
Drinking Water Collaborative Test May 1980 Data Sheet.

7. A uranium standard of known concentration has been sent to you from the
Environmental Protection Agency in Las Vegas. This standard should be
used to determine your counting efficiency and uranium recovery factor.

8. Additional amounts of any uranium concentrates are available if a sample
is spilled, or an obviously incorrect uranium concentration is obtained,
etc.

9. Sample concentrations may be below 15 pCi/1. Reagent blanks of your
drinking water should be made in your laboratory to be certain that they
are not a problem at this uranium concentration.
37
-------
10. The reagent blank (B) for the purpose of this collaborative study is
defined as the cpm observed in one liter of your drinking water including
the counter background. (Ordinarily, the reagent blank is determined
using distilled, deionized water.)

11. We are requesting that uranium concentrations in all samples, counter
efficiency determinations (E), recovery factors (R), and drinking water
blanks be determined in triplicate.

12. All samples should be counted for at least 50 minutes. Longer counting
times are advisable for blanks and any low-level sample.

13. Individual counting data is requested so that the counting statistics
error can be resolved from other errors.
38
-------
APPENDIX E

QUESTIONNAIRE SUMMARY
QUESTIONNAIRE ON COLLABORATIVE STUDY
OF URANIUM IN DRINKING WATER, MAY, 1980
Laboratory Contact

1. Did you make a blank correction?
Yes 19 No

2. When you pipeted the tracer, was the volume
of the pi pet calibrated?

Yes 16 No _3_

3. Did you regenerate the anion exchange resin?
4. Did you deviate considerably from any part
of the procedure?
Yes 16 No 3
Yes No 19
If so, what were these deviations? (answer on other side)

5. Do you believe that this would be a good
reference method for uranium in drinking water?
Yes 15 No 2*
6. Any other comments that you might have would be appreciated.
*Two more were undecided.

39
-------
APPENDIX F

COMMENTS FROM PARTICIPANTS IN COLLABORATIVE STUDY
1. The general procedure appears to be a quick method for total uranium
determination. The sample loss during handling appears to be minimal,
and the analytical procedure appears much simpler than the fluorometric
procedure.

2. The procedure for calibration assumes that the addition of iodic acid to
the standard duplicates the final composition of the ion exchange eluate.
The residue on a planchet after the ion exchange procedure was visibly
less than the amount of iodic acid visible on a planchet prepared
according to the calibration procedure. Thus, the addition of iodic acid
is not felt to be an appropriate method of correcting for self-absorption
and counting efficiency simultaneously. The counting efficiency
determined by the calibration procedure is possibly too low and does not
allow for any variability in the amount of residue.

3. The quantitative transfer of the uranium eluant from the beaker to the
planchet prior to counting is questionable. In addition, "creeping" of
the sample on the side of the planchet was observed thereby introducing
some uncertainty in counting efficiency.

4. The procedure should specify new anion exchange resin rather than
regenerating resin.

5. For every set of samples, analyze a spiked sample and one of the samples
in duplicate to check the recovery factor.

6. The use of the term "column volume" would possibly be confusing to a
technician. In the entire procedure, column volume should be expressed
as milliliters.

7. A better estimate of the counting efficiency and a self-absorption factor
can be made by determining the "weightless" counting efficiency and
developing a self-absorption curve to use with the procedure. The
self-absorption curve can be made using iodic acid to increase the mass
per unit area on the planchet as the same amount of uranium is added to
each planchet. The flaming procedure is not used as this weight must be
accurately determined, and the flaming procedure could create variability
in the planchet weight.

The recovery factor would require weighing the planchet before and after
the residue has been dried on it and using the net mass per unit area to
find the self-absorption factor.

40
-------
The formula would be:

«-w
where: F = gross count per minute of spiked sample
B = counts per minute of reagent blank
C = disintegrations per minute of uranium standard added
E = counting efficiency = cpm weightless std./dpm
weightless std.
A = self-absorption factor = cpm of std. of this mass/
unit/area/cpm of weightless std.

Uranium alpha activity, pCi/1 = 2.2^'E^l A x°V°x R Eq* 5

where: S = gross cpm for sample
B = cpm of reagent
V = volume of sample used, ml
E = efficiency, cpm/dpm
A = self-absorption factor based on mass/unit area
(unitless)
2.22 = conversion factor for dpm/pCi
R = recovery factor

8. In step 8.2.6 by adding 16M HN03 to a small volume of the evaporated
uranium eluant, the iodate would be reduced to the free iodine; thereby,

eliminating the iodate on the counting planchet.

9. The procedure should call for the fresh'preparation of NH[tOH for each
analysis which would preclude the formation of the C03 ion.

10. Uranium yields will vary considerably depending on the amount of
dissolved solids in the water sample.

11. Some participants believed the method was good and easy to follow, while
others believed the method was time consuming, complex, and would be too
expensive for a routine method.

12. This method should be an alternative method of a fluorometric or alpha
spectrometric method.
41
-------