ANALYTICAL CHEMISTRY
1304
indicator end point slightly, but it does not interfere with the determination of tungsten. Of the other elements that may be preJent in the filtrate from the carbonate fusion, attention need be given only to molybdenum. All the others either are not reduced in hydrochloric acid or their half-wave potentials are more negative than that of tungsten. Molybdenum, however, would necessitate a separation if its molar concentration greatly exceeded that of tungsten. The acid concentration must be carefully controlled so that the final acidity of samples and standard solutions is the same. The half-wave potential and particularly the height of the tungsten wave are sensitive to changes in the molarity.
Table 111. Comparison of Polarographic with Gravimetric Results Description of Sample 1. From edge of tailings pond 2 . Grab sample of mill heads, Idaho 3. From a tungsten mine, Alaska 4. From a tungsten mine, Alaska 5. Dredge concentrate of placer gravel Alaska 6. Dredge honcentrate of placer gravel, Alaska 7 . Dredge concentrate of placer gravel, Alaska
Tungsten, % Polarographic 1 2 Bv. 1.23 1.27 1.25 0.32 0.43 0.37
G ra vi. metric 1.25 0.40
3.25 5.97 3.20
3.06 5 . 80 3.37
3.15 5.88 3.28
3.11 5.30 3.40
0.32
0.32
0.32
0.30
1.00
0.96
0.98
0.96
ACKNOWLEDGMENT
The author wishes to acknowledge the helpful advice and suggestions of J. L. Hague, H. $. Bright, J. I. Hoffman, and J. K. Taylor, of the National Bureau of Standards, and D. R. Norton and J. J. Fahey, of the U.S. Geological Survey.
tending from zero applied electromotive force to -0.80 volt, which makes the measurement of tungsten impossible. Addition of cinnamic acid to the solution eliminates the current due to vanadium. Cinnamic acid has no effect on the wave of the tungsten and does not contribute any current to the system. Cinnamic acid is only slightly soluble in acid solution and precipitates copiously immediately when added to the acid, but this causes no difficulty. Several soluble complexing agents (cupferron, quinoline, and diphenylamine) for vanadium were tried, but the complexing agents themselves produced interfering waves. Most of the iron is removed by the filtration. However in many samples, some of the iron remains in the ferrous state after the carbonate fusion. Potassium nitrate added to the flux to oxidize all the iron to ferric later interfered with the tungsten current-voltage curve. Ferrous iron in the solution may obscure the
LITERATURE CITED
(1) Hillebrand, W.F., Lundell, G. E. F., Bright, H. A., and Hoffman, J. I., “Applied Inorganic Analysis,” pp. 683-93, New York, John Wiley & Sons, 1953. (2) Kolthoff, I. lf.,and Parry, E. P., J . Am. Chem. SOC.,73, 5315 (1951). (3) Lingane, J. J., and Small, L. A., Ihid.,71, 943 (1949). (4) Stackelberg, .\I. v., Klinger, P., Koch, W.,and Krath, E., Tech. Mitt. K r u p p , A . Forschungsber., 2, 59 (1939). (5) Ward. F. X., U . S. Geol. Surcey, Circ., 119, (1951).
RECEIVED for review October 31, 1953. Accepted May 13, 1954. Publication authorized b y the Director, U. S . Geological Survey.
Radioactivity Assay of Water and Industrial Wastes with Internal Proportional Counter LLOYD R. SETTER, ABRAHAM S. GOLDIN, and JOHN S. NADER Department of Health, Education, and Welfare, Robert
A method for determining low levels of nonvolatile radioactive contamination in water is proposed. The suspended and dissolved radioactivities are separated by filtration and evaporation. This permits counting both the alpha and beta radiation at levels less than the maximum permissible concentration of unknown isotopes in drinking water. When a 250-ml. sample of water containing about 50 ppc. per liter is prepared and counted for 30 minutes, the activity may be assayed with an accuracy (at the 95% confidence limit) of 10% for alpha and 20% for beta radiation. Levels as low as 10 ppc. per liter (beta) and 2 ppc. per liter (alpha) are detectable. Such sensitivity and accuracy are made possible by counting the dry solids spread over a large area and by using instruments with efficient counting characteristics.
A
METHOD for separately assaying the gross alpha and beta radioactivity of waters and industrial wastes is presented. Filtered or evaporated samples are counted with an internal proportional counter which is particularly applicable for assaying low levels in the order of 1 to 10 ppc. of alpha activity and 10 to 100 ppc. of beta activity per liter of water. This counter has the properties desirable for high efficiency counting. Graf et al. reported ( 7 ) that i t gives a counting rate as much as 7 to 11 times greater than that from an endwindow counter. Window and air absorption losses are eliminated because the sample is within the counting volume, and the
A.
r a f t Sanitary Engineering Center, Cincinnati,
Ohio
geometry of counting is 50%. Because some internal counters aill accept large ( 5 em. in diameter) dishes, the losses from selfabsorption for equal sample volumes are minimized. This is important for samples which have an appieciable (>200 p.p.m.) dry solids content. The counter aill count alpha activity in the presence of up to at least 500,000 counts per minute of beta activity without interference. Beta activity may be counted a t the higher beta operating voltage, provided the alpha activity is negligible or alpha interference is eliminated with an absorber. A detailed evaluation of the counting efficiency of the counter has been made (11). IN STRUM ENTATION
A commercially available internal proportional gas-floM- counting set was used, This counting set consisted of a Nuclear RIeasurements Corp., Indianapolis, Ind., PCC-10 converter plus a Suclear Instrument and Chemical Corp., Chicago, Ill. scaler, Model 161. A cross section of the counting chamber is shown in Figure 1. I t consists of a center wire assembly in a hemisphere-and-piston chamber. The chamber will accept dishes, 5 cm. in diameter by 1 cm. in depth, and may be readily taken apart for maintenance. One of the criteria of reproducibility and accuracy in counting is that a curve of count versus voltage (6) has a broad and flat plateau. The slope, length, and position of the plateau depend on the area of the source, and on the size and position of the center wire loop. To count samples of extended area, i t is
1305
V O L U M E 2 6 , NO. 8, A U G U S T 1 9 5 4 Table I. Backscattered Radiation Maximum Energy, M.e.v. 0.167
Backscatter Factor0 LMF A1 Cub A1 cue Isotope Sulfur-35 1.09 1.15 MFPd 1.25 1.29 Thallium-204 0.762 1.28 1.31 Phosphorus-32 1.704 1.37 1.39 0 Ratio of activity w,ith backscatter t o activity without backscatter. b Membrane filter (4.6 mg./sq. cm.) aluminum (0.0035 inch) copper ( 2 0.040 inch) C Aluminum (0.0035 inch) i copper (0.040 inch). d Mixed fisson products, 2 . 5 years old.
++
+
+
+
CENTER
WIRE L E 4 0
TEFLON
BRASS'
I
F z
3 0 0 W
0.6
-
0.4
-
o.2 0.0
t
I
0
2
2 c
I
Figure 2.
I
I
1
6 8 SAMPLE THICKNESS 4
I
IO IN
I
I
I
12
14
16
rng/crnz
Self-Absorption in an Internal Counter
INSULITOR
PISTONA
Figure 1. Internal Counter Chamber necessary to establish a plateau using an extended source. The position and size of the center wire loop selected for the optimum plateau characteristics are shown in Figure 1. For a radioactive source 5 em. in diameter, the plateau is greater than 150 volts in length with a slope of less than 3% per 100 volts. The use of metal counting dishes causes backscattering ( 5 ) which improves the counting efficiency. With aluminum dishes (milk bottle caps) on the piston head, the backscatterer consists of 0.0035 inch of aluminum and more than 0.040 inch of brass. When suspended solids are counted, the backscattering is changed by the addition of the filter. The amount of backscattered radiation from thip combination with and without a membrane filter Mas determined for a range of beta energies. The results, presented in Table I, show that backscatter and the efficiency of counting are energy dependent. Self-absorption ( 4 ) , the absorption of radiations by the sample solids, reduces the efficiency of counting and is independent of the counter. The ektent of self-absorption by sample salts in aluminum dishes counted in the internal cbounter is shoiin in Figure 2. The salts used were magnesium ammonium phosphate hexahydrate for phosphorus-32, cerous carbonate pentahydrate plus strontium carbonate for mixed fission products, NFP, and lead iodide for iodine-131. Reduction in efficiency of counting becomes apparent at thicknesses of sample solids of 1 mg. per sq. cm. or more. At a thickness of 10 mg. per sq. em. the count is reduced by 27, 36 and 45% for phosphorus-32, mixed fission products and iodine-131, respectively. The counting dish may reduce the efficiency of counting by afiecting the electric field. Dielectric or ungrounded metallic sample dishes tend to reduce the counting rate. For an ungrounded metallic dish the counting rate has been observed to decrease with time by as much as 35% in about 15 minutes. I t is important, therefore, that the sample dish be properly grounded. When lightweight aluminum dishes are used, a positive grounding of the dish may be obtained by impaling the
side of the dish on a grounded pin within the counting chamber as shown in Figure 1. Heavier dishes or planchets are grounded by their own weight. Dielectrics may be made conducting by spraying them with a very thin conducting film. Efficiency. The alpha counting efficiency of the counter for samples having negligible self-absorption is 51 =!= 1% because the geometry is 50% and the backscattered radiation is between 0 and 2% ( 1 4 ) . The beta counting efficiency of the counter for samples having negligible self-absorption will vary from more than 50% to over 75% because the backscattered radiation varies from a few per cent to over 25% of the total radiation, depending on the energy of the radiation, the atomic number, and thickness of the sample backing. The phosphorus-32 counting efficiency was determined from backscatter experiments and checked by counting standard solutions of phosphorus-32 from the Sational Bureau of Standards. For an unknown beta emitter, whose backscatter factor cannot be determined, the counting efficiency may be estimated by assuming an intermediate backscatter value. PROCEDURE
Preparation of Samples. The suspended matter in a sample (200 to 1000 ml.) was usually removed by filtration through a porous cellulose acetate membrane ( 3 ) or, a t some sacrifice of efficiency (1), through a smooth hard-surfaced filter paper. The suspended matter on a membrane was oven-dried a t 103' C., cooled, and counted. The suspended solids were determined by the Gooch method (2) to obtain the sample thickness for self-absorption correction. A sample of filtrate (200 to 1000 ml.), containing preferably less than 100 mg. of solids, was evaporated directly in a tared metal dish to avoid loss of radioactive material (16), for which an automatic evaporator ( 1 2 ) was proved to be convenient for direct evaporation of water samples. By this means a sample rould be evaporated unattended a t a rate of 20 ml. per hour. The temperature of evaporation did not exceed 90' C. until all but a few milliliters of water remained. The sample was then dried a t 103' C., cooled, weighed, and counted. Khen the sample was sufficiently radioactive, a volume containing a t least 1000 counts per minute was pipetted directly into the counting dish for drying and counting. Counting. PLATEBU A N D BACKGROUND. The alpha and beta operating voltages ( 5 )were chosen a t the middle of the alpha and beta plateaus, respectively, using a uranium oxide source, 5 cm. in diameter, for the alpha plateau and a similar source covered with aluminum foil (10 mg. per sq. cm.) for the beta plateau. h check on proper instrument functioning was made by counting these sources a t their respective operating voltages three times a dav. Each of the alpha and beta backgrounds was determined three times a day by counting clean sample dishes a t the proper voltages. SAMPLE ACTIVITY.The sample was counted a t both the alpha and the beta operating voltages for two 16-minute periods each, separated by a short flushing time. If the two counting rates in a test did not agree within counting statistics ( 8 ) ,a third count was taken immediately. When the alpha activity was less than 5 counts per minute, the observed beta count was corrected by subtracting the alpha
A N A L Y T I C A L CHEMISTRY
1306 Table 11. Reproducibility of Duplicate Radioactivity Assavs “~ Radioactivity, Counts per Minute Sample Assay Test 1 Test 2 Rain Beta Soluble 46.4 42.6 Insoluble 242.9 240.7 Snow Alpha Soluble 0.41 0.45 Insoluble 1.23 1.17 Beta Soluble 20.3 22.2 Insoluble 8.2 5.0 River Alpha Soluble 5.1 6.5 Insoluble 4.4 6.1 Beta Soluble 20.5 18.0 Tt-aste Alpha Soluble 4.1 4.5 Insoluble 14.5 15,5 Beta Soluble 31.0 31.5 Insoluble 221.9 227.9 Sewage Beta Soluble 2004 1977 Total 2553 2579 n Counting error a t the 95% confidence level. ~
~
Difference, Counts per Minute Observed Error= 2.8 4.0 2.2 8.7 0.04 0.33 0.06 0.45 1.9 3.3 3.2 3.2 1.4 I.! 1.7 I., 2.5 3.4 0.4 1.0 1.0 2.3 0 5 4.2 6.0 7.3 27 35 26 40
count. \Then the alpha activity wa8 5 counts per minute or more the sample was covered with an aluminum foil ( 7 mg. per sq. cm. thick) and the observed beta count was corrected for aluminum foil absorption by extrapolating an absorption curve ( 5 ) to zero thickness. CALCULATIOKS. The counting rate, because the counter efficiency and sample volume are known, can be converted to niicromicrocuries per liter. The counting rate error, E95 (at the 95% confidence level) was obtained ( 6 ) , and converted to micromicrocuries per liter. REPRODUCIBILITY AND ACCURACY
The reproducibility achieved with the assay method on a variety of typical samples is shonn in Table 11. A comparison of the last two columns of Table I1 indicates that the difference in duplicate assays is conqistent with the counting error E9;.
Table 111. Accuracy of Radioactive Assays Radioactivity, pfic /I. - countingb Source
Activity
Uranium
Alpha
Added
Observed
1.34
2.12c
27 .5 4 Thallium T120’
Beta
9
0.56C 0.56C
17.9C 23.4c 39 1%:
Corrected= 2.12 0.5a 0.56 17.9 23.4 49
Error,
%
73 270 270 17
1.i
10 69 185
225
208 239 6 182 209 6 204 234 0 Self-absorption correction for Tl204 taken from hIFP curve in Figure 2. Alpha (uranium) self-absorption was calculated. b Statistical error of counting a t 95%; confidence level. c Distilled water sample, all other samples in Cincinnati t a p water.
The accuracy of the assay was tested by adding known quantities of uranium salts and of thallium-204 to 250 ml. of distilled or Cincinnati tap water. The activity of the uranium is knonn (9, IO). The activity of the thallium is based on the assay of weightless samples. The results, presented in Table 111, indicate that rather poor accuracy is obtained a t the low levels of 1 to 2 ppc. per liter of alpha activity and about 10 ppc. per liter of beta activity. At higher levels, equal to about 50 ppc per liter or half of the maximum permissible level of unknown isotope in water, the error is 10% for the alpha and 20% for the beta assays. DISCUSSION
The value of this assay method is limited by the errors introduced by the random nature of the radioactive disintegration
proceqi, the factors used in converting the observed counting rate to absolute activity, and the variations in sample preparation. I n theory, any radioactivity, however slight, could be detected if counted long enough. In practice, a counting period of 32 minutes was selected as a convenient time for loTv-level counting. The conversion factors of geometry, self-absorption, and backscatter may introduce errors in the determination of radioactivity. For counters having good plateau characteristics, the geometry factor is reliable. The self-absorption factor may introduce large errors, so it is desirable to minimize the error by counting thin samples. For radioactivity of known isotopes or unknown sources a t levels of 1000 ppc. per liter or more, the self-absorption factor may be determined from curves such as those in Figure 2. .4t lower levels, self-absorption may be estimated on the basis of an assumed energy. The backscatter factor for beta emitters varies with energy as shonn in Table I. When possible, the backscatter factor should be determined (15),or a backscatter factor may be assumed on thc ha& of an estimated energy. SUMMARY AND CONCLUSIONS
A method for separately assaying the alpha and hetn radioactivity in n-ater and industrial wastes is presented. The suspended solids in a sample are removed by filtration or centrifuging for an assa? of insoluble raiiioactivity. The supernatant or filtrate of a relatively large sample is evaporated in a large counting dish to deposit the dissolved solids in a thin layer 90 as to minimize self-absorption losses. The radioactivity is counted in an internal proportional counter, the counting efficiency of which varies from 50 to 52% for alpha activity and from 50 to 75% for beta activity, depending on the amount of backscattered radiation. At low levels of radioactivity, the statistical error of counting has a predominant effect on the accuracy of the assay. At a radioactivity level of 50 ppc. per liter (one half of the maximum permissible level (IS) of unknown isotopes in drinking water), this error is =k 10% for alpha activity and f 20% for beta activity at the 95% confidence level. Lower levels of activity are detected but with leas accuracy. LITER.4TURE C I T E D
(1) Alercio, J. S., and Harley, J. H., Sucleonics, 10, S o . 11, 87 (1952). (2) rlm. Public Health ..lssoc., Xew York, ”Standard Methods for the Examination of Water, Sewage and Industrial Wastes,” 9th ed., p. 21, 1946. (3) Clark, H. F., Geldreich, E. E., Jeter, H. L.. and Kahler, P. W., Public Health Repts.. 66, 951 (1951). (4) Cohen, B., and Hull, D. E., E. 9. ;Itomic Energy Commission, MDDC-387 (Oct. l S , 1946). (5) Friedlander, G., and Kennedy, J . W.,“Introduction to Radiochemistry,” New York, John Wiley & Sons, 1919. (6) Goldin, A. S.,Nader, J. S., and Setter, L. R., J . -4m,Water Works Assoc., 45, 73 (1953). ( i )Graf, W. L., Comar, C. L., and Whitney, J. B., .l’ucZeonics, 9, s o . 4, 22 (1951). (8) Jarrett, A. A,, U. S. .Itomic Energy Commission, AECU-262 (June 1946). (9) Kienberger, C. A, Phya. Rea., 76, 1561 (1949). (10) Kovarik, A. F., and ..ldams, K. I., Jr., J . -4ppl. Phys., 12, 296 (1941). (11) Sader, J. S.,Hagee, G. K., and Setter, L. R., Nucleonics, 12, No. 6, 29 (1954). (12) Kader, J. S., and Setter, L. R., Activity Rept., Environmental Health Center, Public Health Service, 14, 11 (OctoberNovember-December 1952). (13) National Bureau of Standards, Handbook 52 (1953). (14) Rossi, B. B., and Staub, H. H., “Ionization Chambers and Counters,” p. 128, National Nuclear Energy Series T’-2, Sew York, McGraw-Hill Book Co., 1949. (15) Seliger, H. H., Phys. Rew., 88, 408 (1952). (16) Wahl, A. C., and Bonner, S . A., “Radioactivity Applied to Chemistry,” Chap. 6, Sew York, John Wdey & Sons, 1951. RECEIVED for review December 17. 1953
4ccepted June 1, 1954