EPA/520/5-83/027
United States
Environmental Protection
Agency
Office of Radiation Programs
Eastern Environmental
Radiation Facility
P O. Box 3009
Montgomery, AL 36193
EPA 520/5-83-027
December 1983
Radiation
Methods and Results of EPA's
Study of Radon in Drinking
Water
U S, £Mvij?aWMENWi RROTECfiON
AyENCV
1445 ROSS AVENUE
DALLAS, TEXAS ?5202
-------
EPA 520/5-83-027
METHODS AND RESULTS OF
EPA'S STUDY OF RADON IN
DRINKING WATER
Thomas R. Horton
December 1983
U.S. Environmental Protection Agency
Office of Radiation Programs
Eastern Environmental Radiation Facility
P.O. Box 3009
Montgomery, AL 36193
-------
TABLE OF CONTENTS
Page
List of Figures ill
List of Tables iv
Acknowledgements v
1.0 Introduction 1
2.0 Sampling and Analysis Methods 2
2.1 Sampling Method 2
2.2 Liquid Scintillation Counting 2
2.3 Ra-226 Calibration 3
2.4 Rn-222 Concentration Determination 3
2.5 Precision and Accuracy of Rn-222 Determinations ... 4
2.6 Other Radionuclide Determinations 4
3.0 Results 8
4.0 Summary 17
References 25
ii
-------
LIST OF FIGURES
Figure Page
1 A Plot of Least Squares Linear Regression Analysis
Measuring the Precision of Rn-222 Determinations 5
2 The Results of a Rn-222 Cross-check 7
3 Locations of Public Water Supply Samples Collected
1981-1982 *. 9
4 Average Rn-222 Concentrations in Drinking Water Samples
Collected 1981-1982 (three-dimensional)- 12
5 Average Rn-222 Concentrations in Drinking Water Samples
Collected 1981-1982 (contour) 13
6 Average Rn-222 Concentrations in Drinking Water in
Massachusetts, New Hampshire, Rhode Island, and
Vermont, 1981-1982 15
7 Average Gross Alpha Concentrations in Drinking Water
Samples Collected 1981-1982 18
8 Average Ra-226 Concentrations in Drinking Water Samples
Collected 1981-1982 (three-dimensional) 19
9 Average Ra-226 Concentrations in Drinking Water Samples
Collected 1981-1982 (contour) 20
10 Average Total Uranium Concentrations in Drinking Water
Samples Collected 1981-1982 (three-dimensional) 21
11 Average Total Uranium Concentrations in Drinking Water
Samples Collected 1981-1982 (contour) 22
12 U-234/U-238 Ratios in Drinking Water Samples (Samples
with Total Uranium > 3.5 pCi/1) Collected 1981-1982
(three-dimensional) 23
13 U-234/U-238 Ratios in Drinking Water Samples (Samples
with Total Uranium > 3.5 pCi/1) Collected 1981-1982
(contour) 24
iii
-------
LIST OF TABLES
Table Page
^^
1 Results of a Radon Intercomparison Study 6
2 Summary of Radon and Other Natural Radioactivity Results . 10
3 Arithmetic and Geometric Means for Missing States .... 14
IV
-------
ACKNOWLEDGEMENTS
Special thanks are extended to all the state health department
personnel who collected most of the samples reported in this paper.
Without their help this study would have been impossible to conduct.
Special appreciation is also extended to Kitty Newman for her patience
and steadfastness in logging in and preparing all the samples for
counting, making the radon calculations, maintaining the raw data
sheets, and digitizing all the locations by latitude and longitude.
Without the dedication and talent of Keith McCroan, the computer
generated plots would have been unavailable for this paper. Without
the plots much of the potential impact of the data would be lost. The
author would also like to recognize the efforts of Cody Partridge and
Ed Sensintaffar in developing the methods employed in sample collection
and radon analysis and calibration. Mr. Partridge was also responsible
for conducting the pilot study. The author especially appreciates the
sincere efforts of both individuals to impart their knowledge of this
project to the author. Special thanks also to the reviewers of this
paper, especially to Jon Broadway and Charles Porter for their helpful
suggestions, to Mardy Wilkes for typing the many drafts, and to Chuck
Petko for providing editorial support. Finally, the author thanks the
EPA's Office of Drinking Water for supplying computerized listings of
all public groundwater systems serving 1000 or more people.
-------
1.0 INTRODUCTION
In 1978, the Environmental Protection Agency (EPA), through its
Eastern Environmental Radiation Facility (EERF), began sampling radon
in drinking water. During the next two to three years, approximately
27 states were included in this pilot study, the purpose of which was
to determine the need for a nationwide study of radon in drinking
water; to demonstrate the feasibility of such a study; and to develop a
limited data base of radon in drinking water nationwide. The subject
of this paper is the nationwide study that developed from that pilot
study.
The nationwide study, which began in November of 1980, was
designed to systematically sample water supplies in all 48 contiguous
states. The results of the study will be used, in cooperation with
EPA's Office of Drinking Water (ODW), to estimate population exposures
nationwide and to support future standards for radon, uranium, and
other natural radioactivity in public water supplies.
The study design called for sampling only finished water; limited
sampling to once per water supply; targeted composite samples or system
samples instead of individual well supplies; encouraged sampling as
near the source of water as possible; and excluded surface water
supplies (no significant radon was detected in surface water in the
pilot study) and supplies that served less than 1000 people. Our
intent was to collect samples that represented what people actually
consume from a given public water supply.
Of the more than 2500 public water supplies that we sampled, more
than 95 percent met our criteria. Only about one percent were surface
systems (less than 30 supplies), and less than four percent of the
groundwater supplies served populations of less than 1000.
The scope of this study is also noteworthy. Although we sampled
only about five percent of the total number of groundwater supplies in
the 48 contiguous states of the U.S., those samples represent 45
percent of the water consumed by U.S. groundwater users.
-------
2.0 SAMPLING AND ANALYSIS METHODS
2.1 Sampling Method
The sampling procedure for radon in drinking water is described in
the EPA/EERF manual (Ref. 1). This method is reliable if the
instructions outlined in the manual are followed carefully.
Two samples per water supply are taken. Collecting and analyzing
two samples provides us a backup sample if one vial is broken or leaks
during shipment, gives us a measure of overall precision in sampling
and measuring radon, and permits the computation of an average value
instead of a single value.
After collection, the 10 ml aliquot of sampled water is added to a
20 ml glass scintillation vial containing 10 ml of mineral oil based
liquid scintillator. The two scintillation vials are carefully packed.
in a mailing tube to be shipped to the EERF. The vials are separated
from each other with newspaper, a paper towel, or other packing
material and must be well packed to withstand shipping impacts. A
completed sampling and analysis form, which identifies the samples and
provides information necessary for calculating radon concentration, is
returned with the vials. The vials must be mailed on the collection
day or the following day to avoid unnecessary radon decay. Radon in
samples received seven to ten days after collection has generally
decayed beyond detection, unless the sample has a relatively large
initial radon concentration (e.g., 1000 pCi/1 or greater).
2.2 Liquid Scintillation Counting
Radon samples are analyzed by liquid scintillation counting. Our
method varies from the Prichard and Gesell procedure (Ref. 2) in that
we use a mineral oil based scintillation mixture instead of a
toluene-based liquid scintillation fluid. We use the mineral oil
mixture because it has a higher flash point, which allows sample
shipment through regular mail without restrictions. We also use 10 ml
of mix instead of 5 ml. Typical instrument settings are 1.0 for the
gain with a wide open window for the energy discriminator (i.e., lower
level 0.1 and upper level 10.0). Background count rate varies from
about 35 counts per minute (cpm) to 45 cpm depending on the
instrument. A 50 minute count gives a minimum detectable level (MDL)
of about 15 pCi/1, while a 20 minute count raises the MDL to about 23
pCi/1. The MDL is a function of the counting efficiency, counting
time, and background count rate (Ref. 3 and Ref. 4).
-------
2.3 Ra-226 Calibration
A traceable National Bureau of Standards Ra-226 standard solution
is used in calibration. A known quantity of Ra-226 is added to a known
volume of distilled water; 10 ml aliquot of the Ra-226 solution is
combined with 10 ml of a mineral oil based scintillator mix and sealed
in a 20 ml glass scintillation vial; and radon, the daughter of Ra-226,
is allowed to build up for approximately 21 days. At this point,
radon, for all practical purposes, has reached secular equilibrium with
Ra-226. By shaking the vial, nearly all the radon is transferred to
the scintillator phase (radon is highly soluble in the scintillator).
By waiting three hours before counting, the radon short-lived daughters
are allowed to build up to secular equilibrium with radon. The Ra-226
remains in the aqueous phase and, therefore, does not contribute
significantly to the count rate. This was verified by counting the
standard before significant buildup of radon occurred. A near
background count rate was observed. The slight increase in count rate
would be due to the Ra-226 at the aqueous/scintillator interface. The
standard and the background samples are counted for 50 minutes or
longer. To obtain the cpm/pCi conversion factor, the background cpm is
subtracted from the gross cpm for the standard and the difference is
divided by the known radon activity in pCi. The radon activity equals
the Ra-226 activity at secular equilibrium. A typical cpm/pCi
conversion factor is about 10.2 cpm/pCi of radon. This relates to a
counting efficiency of about 90%.
2.4 Rn-222 Concentration Determination
The radon concentration in pCi/1 in the sample is given by—
pCi/1 = (net cpm/c.f./decay) x (1000 ml/liter/10 ml),
where
net cpm = gross cpm - background cpm,
c.f. = cpm/pCi conversion factor,
decay = exp(-7.56E-3 x time), and
time = time lapse from time of collection
to time of counting in hours.
Two sigma
counting error
in percent = 2 x (gross cpm/length of count + background/
length of count)(1/2 power)/net cpm x 100.
-------
2.5 Precision and Accuracy of Rn-222 Determinations
Using 246 duplicate pairs of data (where concentrations ranged
from 100-500,000 pCi/1) collected from November 1978 through February
1981, a plot of the average range between paired data versus the
average concentration using least squares linear regression analysis is
shown in Figure 1. The slope of the line indicates an approximate five
percent degree of precision over the entire range of concentrations
(100-500,000 pCi/1). Note that the highest concentration interval is
not plotted in Figure 1, but it was used to calculate the least squares
fit.
The EERF participated in an intercomparison study with the
University of Texas School of Public Health at Houston where several
different methods were employed (Ref. 5). Our data compared favorably
with the overall mean values obtained by the study and control sample
results as shown in Table 1.
An informal cross-check with the University of South Carolina
(USC) Geology Department was also conducted. A set of samples was
collected from ten different water supplies (0-7000 pCi/1) which had a
wide range of radon concentrations. The results of this
intercomparison are shown in Figure 2. A very high correlation is seen
between the two sets of data. The USC data were obtained by an
entirely different method of collection and analysis. In the USC
method, a large volume sample is collected followed by radon
de-emanation and alpha counting.
2.6 Other Radionuclide Determinations
In addition to the radon samples, a one gallon cubitainer water
sample was collected for each water supply included in the study.
These samples allowed us to obtain other data on natural radioactivity
in public drinking water for very little extra collection effort. Our
analyses of these samples were guided, generally, by the requirements
of the Safe Drinking Water Act (SDWA) (Ref. 6).
All samples were analyzed for gross alpha and gross beta. If the
gross alpha was equal to or greater than 5 pCi/1, a Ra-226 analysis was
performed. Shortly after the study began, the cutoff for Ra-226
analyses was dropped to 3 pCi/1 to provide more data. Ra-228 analyses
were performed for samples where the Ra-226 was equal to or greater
than 3 pCi/1. During the second half of the study, samples whose gross
beta exceeded 15 pCi/1 were also analyzed for Ra-228. Where the gross
-------
o
o
o
o
o
o
o
o
co
o
o
o
o
o
o
to
o
o
o
in
o
o
o
o
o
-------
time. The vast majority of the samples were analyzed under the lowest
cutoff criterion. After performing more than 100 thorium analyses, it
was decided that thorium analyses were unnecessary because the
concentrations found in groundwater were very low (typically less than
0.1 pCi/1 for Th-227, Th-228, Th-230, and Th-232). At these levels,
the measurements have an inherent uncertainty. One Ra-226, uranium,
and thorium analysis was performed for each state involved in the
study, regardless of whether any sample from a given state met our
cutoff criteria. This provided us baseline data for each state.
3.0 RESULTS
Locations of the more than 2500 public water supplies sampled in
this study are shown in Figure 3. Nationwide, the public groundwater
systems represent about 45 percent of the total groundwater usage or
about five percent of the total number of public groundwater systems.
Thirteen states were not included in the study, primarily because of a
shortage of manpower and money. Even though the state health
departments were reimbursed for collecting the samples, the
reimbursement did not cover actual expenses incurred in the collection
effort.
Results for radon, gross alpha, gross beta, Ra-226, Ra-228, total
radium, and total uranium are summarized in Table 2. Buildup of radon
and its short-lived daughters during typical household activities
involving water usage can cause a significant indoor working level (WL).
Based on the model used by Partridge et al. (Ref. 8), a 0.01 WL is
possible for a radon in water concentration of 10,000 pCi/1 and a
relatively slow turnover rate of air (e.g., 0.25 air changes per
hour). Energy efficient homes can have a ventilation rate somewhat
less than 0.25 air changes per hour resulting in a higher WL.
Figures 4 and 5 present nationwide radon concentrations in public
water supplies. Elevated radon levels are seen in the New England
states, North and South Carolina, Georgia, Virginia, and western states
such as Arizona, Colorado, Nevada, Montana, and Wyoming. Actual
individual sample radon concentrations range from essentially zero to
greater than 16,000 pCi/1. Some localized averaging is used to
generate the plots. The purpose of the plots is to show general trends
nationwide for inclusion in this paper. Actual individual radon
concentrations for a given water supply are available and will be used
in the final analysis of the data. These comments also apply to other
natural radioactivity results presented in this paper.
-------
CM
00
00
ON
01
4-1
o
fl)
o
o
CO
>
t-1
I
CO
0)
4J
I
co
§
8
o
M
•H
-------
TABLE 2
SUMMARY OF RADON AND OTHER NATURAL RADIOACTIVITY RESULTS
Concentration Range, pCi/1
Approximate Number of
Public Water Supplies
Radon
> 10,000
5,000-10,000
1,000- 5,000
500- 1,000
< 500
Arithmetic mean
Geometric mean
340 pCi/1
100 pCi/1
Gross Alpha
> 15
10 -15
5 -10
< 5
Arithmetic mean
Geometric mean
1.8 pCi/1
0.6 pCi/1
Gross Beta
Ra-226
> 15
10 -15
5 -10
< 5
Arithmetic mean
Geometric mean
> 5
2-5
1-2
< 1
Arithmetic mean
Geometric mean
3
7
160
250
2100
51
56
128
2300
60
75
340
2040
3.4 PCi/l
2.1 pCi/1
34
85
35
232
1.6 pCi/1
0.6 pCi/1
(0.1%)*
(0.3%)
(6 %)
(10 %)
(83 %)
(2 %)
(2 %)
(5 %)
(91 %)
(2 %)
(3 %)
(14 %)
(81 %)
(9 %)
(22 %)
(9 %)
(60 %)
10
-------
(TABLE 2)-Continued
Concentration Range, pCi/1
Approximate Number of
Public Water Supplies
Ra-228
2
1
> 5
- 5
- 2
< 1
25
41
12
22
(25 %)
(41 %)
(12 %)
(22 %)
Total Ra
Total U
Arithmetic mean
Geometric mean
> 10
5 -10
2-5
1-2
< 1
Arithmetic mean
Geometric mean
> 20
10 -20
5 -10
2-5
< 2
Arithmetic mean
Geometric mean
3.5 pCi/1
2.3 pCi/1
29
43
18
1
0
8.5 pCi/1
7.6 pCi/1
18
31
51
96
148
5.1 pCi/1
1.9 pCi/1
(32 %)
(47 %)
(20 %)
(1 %)
(0 %)
(5 %)
(9 %)
(15 %)
(28 %)
(43 %)
* Percent of total
11
-------
CM
00
0.
00
ON
O
(U
O
O
CO
01
(0
CO
00
a
•H
CD CD CD
CD CD CD
CD CD CD
(Ti (XI -H
CO
c
O
2
4-1
C
01
c
O
U
CM
CM
CM
CU
60
2
-*
•
00
12
-------
esi
oo
msnnn
en en en en #
en en CT'i CTi #
ij'i Ch 0"i 0"i ¥;
** \r iji #
*
u
CL-
C3
CD Si
CD CD
•a
ai
u
1
h
01
00
<0
CD CD
CD CD
to in
60
•H
13
-------
Table 3 lists the means for seven states where radon results were
not available from our nationwide study. These means are based on
results from our pilot study and Prichard's study (Ref. 7). By
comparing the geometric means of Table 3 for individual states with
Figures 4 and 5, it is evident that the radon results used to generate
these figures, which do not include the results of Table 3, do show a
reasonably good approximation for those seven states. Having radon
results from all 48 contiguous states would be desirable, but in
practice may not be necessary.
TABLE 3
ARITHMETIC AND GEOMETRIC MEANS FOR MISSING STATES
Radon Concentration, pCi/1 Approximate Number of
Arithmetic Mean (Geometric Mean) Public Water Supplies
AR
CA
IA
LA*
MO
NE*
NJ
120 (12)
1200(470)
1500(220)
180 (55)
300 (24)
320(200)
690(300)
22
9
40
15
69
16
19
US 340(100) 2500
* Ref. 7; results for all other states are from our pilot study.
A regional map of the New England states is shown in Figure 6. An
area of elevated radon concentration is seen stretching from northern
Vermont and New Hampshire through Massachusetts into Rhode Island. All
locations were used in producing this map, i.e., no localized averaging
was employed. As can be seen, more detail is preserved in this
regional map versus the U.S. map.
14
-------
CM
00
9\
r-t
H
00
s
CO
p
p
o
_J
.
2 «
15
-------
Nationwide gross alpha concentrations are displayed in Figure 7.
Most of the U.S. is below 3 pCi/1. In general, only those locations
with elevated uranium show elevated levels.
Ra-226 concentrations are shown in Figures 8 and 9. North and
South Carolina, Georgia, Florida, and the midwest states of Illinois,
Wisconsin, Minnesota, and Kansas show elevated levels. Almost no
Ra-226 is seen in the western states.
Total uranium (U-234 plus U-238) concentrations are presented in
Figures 10 and 11. The western states of New Mexico, Colorado,
Wyoming, and Montana and a few eastern states such as Maine and
Pennsylvania exhibit elevated levels.
The ratio of U-234 to U-238 is of interest to those areas where
elevated total uranium concentrations exist. This interest stems from
the fact that a lower initial cost and simpler method of analyzing for
total uranium (fluorometric method versus alpha spectroscopy used at
the EERF) assumes secular equilibrium between U-234 and U-238 while the
fluorometric method only measures the U-238 content. In certain cases
the total uranium activity may be severely underestimated using the
fluorometric method.
Using uranium results whose total uranium exceeds 3.5 pCi/1,
Figures 12 and 13 were generated. Nearly all the ratios are between
one and two with an arithmetic mean of 1.8 and a geometric mean of
1.7. Low activity samples with their inherent uncertainty are not
included.
Some of the water supplies sampled during the nationwide study did
not meet our sampling criteria. These include surface water supplies
and groundwater supplies serving less than 1000 people. A decision was
made to include these supplies in the analysis of results, since they
are public water systems.
The natural radioactivity associated with the surface supplies is
very low. This includes radon, gross alpha, and gross beta. The
groundwater supplies serving less than 1000 people did not
significantly alter the overall results. In general, the same
concentration pattern is observed in the less than 1000 people
groundwater supplies as is observed in the greater than 1000 people
systems.
16
-------
4.0 SUMMARY
Samples from more than 2500 public water supplies representing 35
states were collected. For the most part, samples were from public
groundwater supplies serving 1000 or more people. Although we sampled
only about five percent of the total number of groundwater supplies in
the 48 contiguous states of the U.S., those samples represent nearly 45
percent of the water consumed by the U.S. groundwater users in the 48
contiguous states. Our intent was to collect samples that represented
what people actually consume from a given public groundwater supply.
The arithmetic means for radon, Ra-226, and total uranium were
calculated to be 340 pCi/1, 1.6 pCi/1, and 5.1 pCi/1, respectively.
The corresponding geometric means for radon, Ra-226, and total uranium
were found to be 100 pCi/1, 0.6 pCi/1, and 1.9 pCi/1, respectively.
The arithmetic mean for the U-234/U-238 ratio was determined to be
1.8 for higher activity samples (total uranium exceeded 3.5 pCi/1),
while the corresponding geometric mean was 1.7. In most cases, total
uranium activity determined by the fluorometric method will not be
significantly underestimated if a correction factor is applied based on
the mean U-234/U-238 ratio for a given geographic region or possibly
the entire U.S.
The results of this nationwide study will be used, in cooperation
with EPA's Office of Drinking Water, to estimate population exposures
nationwide and to support future standards for radon, uranium, and
other natural radioactivity in public water supplies.
17
-------
CM
00
ON
T-t
I
rH
00
O
(U
O
O
CO
0)
c
•H
X)
c
CO
o
(0
R
(U
O
a
o
o
a
iH
C0
CO
CO
O
00
01
SP
CD CO LT) Q I/)
•H -H
00
•H
18
-------
\
CD
CD
10
CM
oo
ON
00
o>
•o
0)
u
0)
o
u
co
01
C0
CO
a;
u
co
60
a
•H
^
a
•H
(0
a
o
u
Q
O
O
CN
CM
k
2
i!
oo
60
•H
19
-------
CM
00
ON
(XI IT)
00
Ov
U
(1)
O
U
CO
(U
CO
I*
01
oo
g
c
•H
M
CO
c
O
n
o>
o
a
o
o
CN
CN
a»
oo
2
a\
00
•H
20
-------
CM
00
I
iH
oo
-o
0)
4J
U
(U
o
U
co
CU
H
a
-------
ana
en en gi
ffi ffi S ffi
u
Q_
CO
i-H
00
-o
(U
u
0)
o
o
to
(U
(0
CO
(U
60
O
Q
•H
00
C
o
2
4->
O
-------
CJ
o.
B
«
M
4J
O
ta
0)
iH
a.
§
CO
co
I
CO
l-l
-------
X!
4J
00
a)
H
I
CO
V
CN
CO
S
00
a
•H
JX -0
Q 01
•H U
ll O
a i-i
•H O
O
CO
O '-N
a
oo
CO
2s
1=) iH
CD •H (\| CO
24
-------
REFERENCES
1. U.S. Environmental Protection Agency, Radon in Water Sampling
Program. EPA/EERF-MANUAL-78-1.
2. Prichard, H.M. and Gesell T.F., "Rapid Measurements of Rn-222
Concentrations in Water with a Commercial Liquid Scintillation
Counter," Health Physics, 33:577-581 (December 1977).
3. Currie, L.A., "Limits for Qualitative Detection and Quantitative
Determination - Application to Radiochemistry," Analytical
Chemistry, 40:586-593 (March 1968).
4. Altshuler, B. and Pasternack, B., "Statistical Measures of the
Lower Limit of Detection of a Radioactivity Counter," Health
Physics, 9:293-298 (1963).
5. Prichard, H.M., Radon in Water Intercomparison. Unpublished
Report - The University of Texas, School of Public Health,
(January 26, 1979) .
6. U.S. Environmental Protection Agency. National Interim Primary
Drinking Water Regulations. EPA-570/9-76-003.
7. Prichard, H.M. Unpublished radon results. The University of
Texas, School of Public Health (1979).
8. Partridge, J.E., Horton, T.R., and Sensintaffar, E.L., A Study of
Radon-222 Released from Water During Typical Household
Activities. U.S. Environmental Protection Agency, Technical Note
ORP/EERF-79-1 (March 1979).
25
------- |