decay of the Np2* tracer. A decay curve for the iron samples is shown in Elgure 2. The curve shows the rapid decrease in the activity of the Fes9 component, and the activity due to the longer lived component, Feu. The time required for the radiochemical decontamination of iron was about 8 hours, but if higher yields were mandatory, another hour was added to the plating time to increase the yield at that step. The neptunium procedure took 6 to 7 hours. The chemical yield for the iron samples averaged 63%, while the neptunium yields averaged 46%. In this study the analysis for iron was made by comparing the combined Fe66859 activities in the samples with those in a simultaneously irradiated iron standard. R.aulta from previous irradiations had shown that the ratios of Fe" to Fe60 in the samples and in the standards have been the same within a small experimental error. It seemed advisable, therefore, to count both isotopes together to increase the sensitivity for the detection of radio-iron. A 2r flow counter seemed well suited for this purpose. A counting gas mixture of 90% argon-IO% methane
was used to maximige the counting efficiency of the x-rays from Fe". Such an arrangement optimiees the counting efficiency for the FeS.69 mixture produced by the irradiation. With this counting gas a high voltage plateau was obtained at 3000 to 3200 volts, which had a slope of 1.7% per 100 volts with Feu x-rays. A high voltage plateau for Fe59was obtained at 2800 to 3400 volts with a slope of 0.8% per 100 volts. The counter was operated at 3150 volts and had a background of 58 c.p.m. Efficiencies of 43.2% for Fe59 and 7.4% for Fern were obtained with this counting technique. Lead has many nuclear properties which are desirable in a matrix element for neutron activation analysis. Because of the large atomic number of lead, the amount of contaminating radionuclides produced from (n,p) and (%,a) reactions is reduced. Because the samples activated in this analysis weighed less than 1 gram and lead has a small neutron-capture cross-section, the problems arising from self shadowing and flux perturbations are avoided. The radioactivity produced from lead is low level and requires little shielding. A possible source of error in the d e
termination of iron is the nuclear reaction on cobalt, Cob9(n,p)Fe6*. Several of the lead samples were analyzed for cobalt by the reaction CoSg(n,7)Corn. The analyses showed, however, that the error from this source was insignificant compared with the amount of iron in the samples. There are no nuclear reaotions which interfere with the determination of uranium. LITERATURE CITED
(1) Atchison, G. J., Beamer, N. H., ANAL. CHEM.24, 1812 (1952). (2) Gilmore, J. S., U. S. At. E w g y Comm. Rept. LA-1721, 64 (1954). (3) HuiTman, J. R., Zbid., AECD-3587, 2-24 (1953). (4) Iredale, P., U.K.At. En.Comm. Rept. AERE EL/M 96, 1-15 (1957). (5) Kant, A., Cali, J. P., Thompson, H. D., ANAL. CHEM.28, 1867 (1956). (6) Mahlman, H. A., Leddiootta, G. W., Zbid., 27,823 (1955). (7) Mitchell, R. F., Zbid., 32, 326 (1960). (8) Smales, A. A., Anal 8t77, 778 (1952). bentley, W. (9) stewart, D. Science 120, 50 (1954).
c.,
c.,
RECEIVED for review January 12, 1962. Accepted May 28, 1962. Preaented 1961 International Conference on Modem Trende in Activation Andyeia, College Station, Tex., December 1516, 1961.
Thallium-204 Radiometric Determination of Dissolved Oxygen in Water HAROLD G. RICHTER and ARTHUR S. GILLESPIE, Jr. The Research Triangle Institute, Durham, N. C.
b A technique for determining the dissolved oxygen concentration of both pure and natural waters is based on quantitative oxidation of thallium metal containing TIrn4 by aqueous dissolved oxygen. The analyzer consists of a column of thallium-204, deposited on copper turnings, below which is a flow-type Geiger-Miller tube. Water containing the dissolved oxygen is pdssed through the column, whereupon TI+ ions are formed. The T P 4 i s detected in the G-M tube, and the counting rate is a direct measure of the dissolved oxygen concentration. In pure water the T P counting rate is linear with dissolved oxygen concentration down to the lowest concentration investigated, 0.2 p. p.m. In sea water and salt water, it is not linear above about 3 p.p.m. and a calibration curve must be used. Linearity in sea water could be achieved b y depositing the thallium on a metal more inert than copper. The effects of pH, buffers, and various oxidizing agents were studied. Using TIS4 of 1 1 16 *
ANALYTICAL CH€MISTRY
the highest specific activity available, it should be possible to determine 1 part of dissolved oxygen in lo1*parts of water.
Q
determination of minute amounts of oxygen dissolved in water is usually one of the more difficult analytical problems. Dissolved oxygen is of particular interest in biological studies relating life processes to oxygen concentration as well as in the study and prevention of corrosion. Two instrumental techniques for determining dissolved oxygen are in use today, but neither is completely satisfactory. One uses the polarographic reduction of dissolved oxygen at a platinum electrode as a measure of oxygen concentration. Clark et d. (3) were apparently the first to solve some of the problems associated with irreversible processes occurring on stationary solid electrodes at the large negative potential required for oxygen UANTITATIVE
reduction. They s i m ~ l y surrounded the electrode wkh a $k& membrane permeable only to oxygen. Only recently, however, have other disadvantages of this electrode been overcome by Carritt and Kanwisher ( 8 ) . Commercial versions of this improved electrode are offered by the Jarrell-Ash Co., Newtonville, Mass., and the Beckman Instrument Co., Fullerton, Calif. The second method is based on rapid oxidation of thallium metal by aqueous oxygen (7). Oxidation to the thallous state takes place quantitatively at as low a concentration of diasolved oxygen as can be measured by available techniques. The increase in conductivity of the effluent stream from the thallium metal column, compared with the conductivity of the i d u e n t stream, is a measure of the oxygen concentration. A commercial instrument based on this technique is made by Industrial Instruments, Inc., Cedar Grove, N. J. One great disadvantage of this thallium technique is that ionized sub-
stances in the water diminish its sensitivity. Deioniaing columns before the inlet conductivity cell help, but waters containing large amounts of dissolved salts soon saturate the deionizing columns. It appeared that these difficulties could be eliminated by substituting radiometric techniques for conductivity techniques.
loo 80 60
401
EXPERIMENTAL
Apparatus. The analyzer consists of a column of radioactive thallium metal electrodeposited on copper turnings, and a flow-type Geiger-Muller tube. The aqueous sample is introduced at the top of the thallium column and allowed to flow slowly through it. Radioactive T P ions released by the reaction of dissolved oxygen with the thallium metal enter immediately into the G-M tube and are detected. The T P counting rate is a direct measure of the oxygen concentration. In practice, the radioactive thallium obtained from Oak Ridge is diluted with inactive thallium metal for determination of oxygen concentrations in the parts-per-million range. As the TP' has a half life of 3.6 years, decay over several months does not greatly reduce the sensitivity of the technique. Since T P is primarily a beta emitter (0.76m.e.v. @ particles, 2.6% 0.4-m.e.v. x-rays), the glass walls of the column provide sufficient shielding and little radiation hazard exists near the device. If necessary, the effluent from the column could be saved and worked over for the T P content, so that a single charge of radioactivity could be used many times. THALLKJM COLUMN.To prepare the column, 1.03 grams of T1+ as TlNOa and 0.47 mg. of 4480 mc. of T P 4 per gram in 100 ml. of water were placed in a beaker containing 20 grams of copper turnings, previously washed in dilute nitric acid. Ten milliliters of glacial acetic acid and 5 grams of sodium acetate were added and electroplating was carried out, using the copper turnings BB cathode and a platinum strip as anode. Nitrogen WBB rapidly bubbled through the solution during the plating process. Experience showed that the thallium deposited was loose and spongy unless rather high voltages were used for plating. Eight volts proved to be satisfactory, even though large amounts of Ht and 00 formed a t the cathode and anode, respectively. The thallium plate formed was grayish white and adherent. Other electrodeposition procedures (1, 6) could be used. The specific activity for this column waa 2.04 mc. per gram of T1. GEIGER-M~LLERCOUNTER. The counter was a Model 18510 liquid flow counter made by Amperex Elkctronics CorD.. Hicksville. L. I., N. Y. It is unu$ual in that 'the anode is a tube through which the solution to be counted flows. Plateau length is over 500 volts, starting at 375 volts. The effective counting volume is 1.02 ml. and wall thickness is 30 mg. per sq. cm. The counter h d B long dead time of 500
0
11
L
OXYGEN
IN
PARTS
PER
MILUON
Figure 1 . Response of oxygen analyzer to waters of varying salt content
microseconds as determined by counting losses and verified by oscilloscope observations. For accurate counting, dead time corrections are necessary; however, in practical application in the oxygen analyzer, only a calibration of observed count with oxygen content is required. Such a plot will show a d e parture from linearity a t high count rates due to the dead time. This particular counting tube was chosen because only a small volume of solution was necessary to sweep out the column and counter. Counting efficiency with this counter was about 1.4%. For applications where larger solution flows are useful, two flow counters, Models 10306 and 10315 (available from the Rsdiation Counter Laboratories, Skokie, Ill.), have a solution volume of 35 ml. and wall thicknesses of 30 mg. per sq. cm. Radiation Recording Equipment. Counting was done with a Model 186 scaler made by the Nuclear Chicago Corp., Des Plaines, Ill. A separate power supply was used to supply voltage at an anode voltage divider assembly at the counting tube. With a minor modification of the scaler, a separate high voltage supply would be unnecessary. For oceanographic or other field work, a portable scaler or rate meter could be used to record the radiation. A complete oxygen analyzer could be compact and portable. RESULTS
Response of the oxygen analyzer
to waters of markedly different salt contents with varying oxygen concentrations are shown in Figure 1, where count rate corrected for dead time losses and for backgrounds is plotted as a function of oxygen concentration. Oxygen concentrations were
determined by the Winkler method ( 4 ) . The linearity for oxygen solutions in distilled water is excellent. A marked departure from linearity was found for sea water and mock sea water containing only sodium chloride. Pure Water-Oxygen Solutions. Thallium determinations on the effluent solutions from the column indicate that, oxidation of thallium proceeds stoichiometrically, as shown by the equation: 4T1(S) Oe(Aq) 2HaO(liq.) -* 4Tl+(Aq) 40H-(Aq) (1) One gram of oxygen liberates 25.6 grams of thallium. One gram of thallium is therefore sufEcient to assay approximately 100 samples of airsaturated water (total volume about 5 liters). The column retained ita calibration from beginning to end and the thallium was completely inert to oxygenfree water: Water through which nitrogen was bubbled was allowed to stand over the column overnight. Negligible thallous ion concentration was found the following morning. Sea Water-Oxygen Solutions. Count rate us. oxygen concentration in sea water shows a marked deviation from linearity. The sea water used was assayed and found to contain 17.7 grams of chlorine per liter. Thallium analyses of the effluent solution from the column indicate that count rate corresponds to liberated thallium, but agree with the quantity expected from the influent oxygen solution only at low oxygen concentrations. Mock sea water consisting of 29.6 grams of pure NaCl per liter of distilled water solution
+
+
+
VOL 34, NO. 9, AUGUST 1962
11 17
Table I.
EfFed of pH on Oxidation of
Thallium Equiv.
PH 2.0 3.9 5.5
02,P.P.M. 13.7 3.6 0.0
gave results similar to those found for sea water. REASONS FOR NONLINEARITY. The deviations from linearity could be caused by exceeding the solubility product of TlCl, oxidation of thallium' to states higher than monovalent, or removal of oxygen by reaction with something other than thallium. Each of these was checked. The solubility product for thallous chloride a t 25' C. is 2.65 X lo-' This theoretically limits the solubility of the T1+ ion in sea water to 5 X 10-4 mole per liter, which corresponds to 3.8 p.p.m. of oxygen, roughly where the instrument response departs from linearity. Precipitated TlCl might be held on the column by the filtering action of the metal turnings and glass wool plug in the bottom, which would reduce the counting rate. However, in an experiment to demonstrate the precipitation of TIC1 from sea water, 300 p.p.m. of T1+ (as TINOs) caused no visible precipitation. This corresponds to 11.3 p.p.m. of oxygen, which is well beyond the maximum expected for air-saturated sea water a t any temperature above 0' C. (6). On the basis of this observation, TIC1 precipitation should not be a problem. That the thallium is oxidized to the Tl+3 state a t high oxygen concentrations seems unlikely, since thallic ion is unstable and should be easily reduced to the thallous state by metallic thallium or copper. The half-cell potentials indicate that this is true. A third possibility is that oxygen may be removed by the copper turnings used as support for the thallium. To study this, freshly cleaned copper turnings were placed in a flask with sea water and air was passed slowly through the solution. Within half an hour a precipitate of cuprous chloride was clearly visible, and letting the reaction proceed overnight produced many milligrams of the salt.
It must be concluded that copper is not a good supporting medium for thallium, if chloride ion is in the solution in which oxygen is to be determined. A more inert element (to oxygen) should be chosen. Therefore, as a check of this idea, thallium was electrodeposited on molybdenum metal, a column prepared, and oxygen-saturated sea water passed through. Winkler oxygen analvses of the influent solutions and thaliium analyses of the effluent yielded stoichiometrically equivalent results (Equation 1). Thus, the nonlinear 1 1 18
ANALYTICAL CHEMISTRY
results of the previous experiments are clearly due to the oxidation of copper by oxygen in the presence of chloride ion. Effect of pH and Oxidizing Agents. The effect of p H was determined by saturating a solution of known p H with nitrogen, passing it through the thallium column, and determining thallium ion in the effluent solution. The equivalent oxygen concentrations, calculated from these experiments. are tabulated in Table I. Since oxygen in basic solutions may react with copper metal, air-saturated solutions of pH 10 were passed through a copper turnings column. Oxygen in the effluent was determined by the Winkler method. No difference was found between these oxygen concentrations and those from air-saturated water (pH 5.5). Therefore, it can be concluded that the technique is suitable for solutions of p H between 5 and 10. A solution made pH 3 with sulfuric acid was brought to p H 7.8 with a NaOH-H8BOs buffer, saturated with air and passed through a thallium column. The effluent thallium concentration was identical with that of a sample of air-saturated distilled water. It can be concluded that acidic solutions can be buffered to a slightly alkaline pH and the dissolved oxygen concentration measured accurately by this technique. Solutions 0.01M in Mg(NO& were saturated with air and passed through a thallium column. The resulting thallium concentrations were variable; the solution was therefore saturated with nitrogen. Thallium concentration in the effluent corresponded to 1.5 to 2.0 p.p.m. of oxygen in the influent solution. Since the pH of the solution was about 6, it must be concluded that nitrate ion can oxidize the thallium metal and give false readings of dissolved oxygen concentrations. Chromate ion solutions (0.01M K2Cr04, pH 8) were likewise passed through the thallium column. No thallium was oxidized in nitrogensaturated solutions, even though the half-cell potential would permit it. Whether chromate solutions under other pH conditions would yield false readings is not known. Sensitivity. Serisitivity of our column with a specific activity of 2.04 mc. per gram of thallium is about 0.2 p.p.m.-that is, an aqueous solution of oxygen of about 0.2 p.p.m. concentration [about 6 X 10-6M, or 1.4 x 10-4 ml. of O2 (STP) per ml. of HzO] produces a Tlm4 counting rate equal to the background counting rate of, the detector. Catalog listing of thallium-204 by the Oak Ridge National Laboratory is a minimum of 1
50 mc. per gram. That supplied to us was 4479 mc. per gram. If this material had been used without dilution, the sensitivity of our column would have been 0.1 part per billion. We did not attempt to investigate oxygen concentrations in this range because of both time limitations and the difficulty of obtaining solutions of known parts per billion oxygen concentrations. Sensitivity could be increased by at least a factor of 10 by using the larger counting tubes mentioned earlier. Still another factor of 10 in sensitivity should be attainable by a reduction in background counting rate, which in our column was about 100 counts per minute. Thus, a sensitivity of about one part of oxygen per trillion parts of water should be possible. This speculation is based on an assumption of total insolubility of metallic thallium in oxygen-free water. Applicability of Technique. Two factors must be considered in the use of this radiometric technique for dissolved oxygen in water: sensitivity, accuracy, and reproducibility of the technique; and the use of radioactivity. The sensitivity of this technique is potentially greater than that of any other method, while its accuracy must depend upon the accuracy of a calibrating method. Whereas the Winkler procedure for dissolved oxygen determination is not without criticisms, its general accuracy and applicability are accepted by most workers. In this study we have found good agreement between published dissolved oxygen concentrations (6),those we determined by Winkler titrations, and the amount of thallous ion released by the dissolved oxygen (by both weighing TI1 and thiosulfate titration of iodine). Since, in distilled water, the relation between Tla4 counting rate and dissolved oxygen concentration was linear to as low an oxygen concentration as we were able to obtain, we predict that it will continue linear to much lower concentrations. Dissolved salts do not interfere, if they are not oxidizing agents. Sea water, the best example of a complex natural water available to us, can be analyzed satisfactorily if the thallium is electrodeposited on a supporting metal such as molybdenum. In such a column, the relation between T1204 counting rate and oxygen concentration will be linear. Because of the randomness of radioactive disintegrations, the reproducibility of any counting rate determination is a function of the number of disintegrations observed. In radioactivity measurements it is not customary to attempt answers to better than &l%,because of too long counting times. Therefore, errors in determination of oxygen concentrations by
this radiometric method cannot be less than 1 t o 2%, unless long counting times are acceptable. The second factor to be considered is the use of radioactivity per se. T P is not difficult to work with. It was not necessary to place, shielding around our short column containing 2 mc. of TlZO4, and no health hazard was associated with working with the analyzer. On the other hand, we were not careless with drippings and effluent solutions. Housekeeping with radioactive materials need not be more difficult than in other good analytical laboratory technique. All effluents were pooled, and T1I was precipitated
when convenient, and converted to TINOs. The Tlm4was thus ready for electrodepositing again. Tl204 can be obtained license-free from the AEC in 50-b~. quantities. By using this quantity of activity and a counter with larger active surface, a satisfactory oxygen analyzer could be constructed. It would be suitable only for oxygen concentrations above about 1 p.p.m. LITERATURE CITED
(1) Brown, 0. W., McGlynn, S. A., Trans. Am. Electrochem. SOC.53, 351 (1928). (2) Carritt, D. E., Kanwisher, J. W., ANAL.CHEM.31, 5 (1959).
(3) Clark, L. C., Jr., Wold, R., Granger,
D., Taylor, E., J. A p p l . Physiol. 6 , 189 (1953). (4) Furman, N. H., ed., “Scott’s Standard Methods of Chemical Analysis,” Vol. 11, p. 2079, Van Nostrand, New York, 1939. (5) Richards, T. W., Smyth, C . P., J. Am. Chem. SOC.44, 524 (1922). (6) Seidell, A., “Solubilities of Inorganic and Organic Compounds,” p. 471, Van Nostrand, New York, 1919. (7) Wright, J. M., Lindsay, W. T., Jr., Proceedings of Am. Power Conf., 21st Annual Meeting, 1959. RECEIVED for review January 29, 1962. Accepted May 23, 1962. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 1962. Work carried out under Atomic Energy Commission, Division of Isotopes DeveIopment, Contract No. AT-(40-1)-2513.
Determination of Micro Amounts of Carbon in Uranium Tetrafluoride R. E. SIMMONS and M. H. RANDOLPH Paducah Planf, Union Carbide Nuclear Co.,Division of Union Carbide Corp., Paducah, Ky. A method for the determination of microgram amounts of carbon in uranium tetrafluoride is presented which offers considerable improvement over conventional methods. Standard methods of carbon analysis, whether gravimetric, volumetric, or by gas volume are handicapped by interference of fluorides and consequently are unreliable for determining carbon in uranium tetrafluoride. The proposed method depends upon the conversion of carbon to carbon dioxide b y igniting the sample, which is fluxed with silicon dioxide, in a high temperature combustion furnace with a flow of oxygen passing over the sample. The gases generated by igniting the sample are collected in an evacuated bulb and analyzed for carbon dioxide on an infrared spectrophotometer with no interference from the other gases present. The precision of the method for a single determination at the 95% confidence level is + 13% for uranium fetrafluoride containing 100 pap.m. carbon and =t3Oy0 for uranium tetrafluoride containing 50 p.p.m. carbon.
S
for determining carbon in organic compounds containing fluorine have been reported in the literature (8-6). Carbon was determined gravimetrically by absorption of carbon dioxide in Ascarite following ignition of the sample and separation from water, fluorine, and in some instances chlorine. I n these methods the samples analyzed contained from EVERAL METHODS
11 to 83% carbon with the weight ratio of fluorine t o carbon never greater than 6 to 1. Determination of micro amounts of carbon in uranium tetrafluoride presents a somewhat more difficult problem as the ratio of fluorine to carbon may be greater than 5000 to 1. A few methods for determining carbon in uranium tetrafluoride are reported in the Atomic Energy Commission project literature (1, 7, 8). These methods are basically the same in that the sample is ignited in a stream of oxygen and the carbon is determined gravimetrically by absorption of the carbon dioxide in Ascarite. In the method reported by Warf (8)magnesium oxide powder was blended with the sample to retain most of the fluorine, and lead dioxide was used for absorbing fluorine not reacted with the magnesium oxide. Bernhardt et d. (1) report two methods for determining carbon in uranium tetrafluoride. In one of the methods the sample is mixed with zinc oxide to retain most of the fluorine with the escaping fluorine compounds trapped by sodium fluoride pellets. This method requires that the pellets be conditioned after each determination by baking a t 200’ to 300” C. while under vacuum. Combustion of uranium tetrafluoride in the presence of water vapor was used in the second met,hod to convert the fluorine to hydrogen fluoride. The water and hydrogen fluoride were separated from the carbon dioxide by absorption in Anhydrone and sodium fluoride pellets, respectively. Van Kooten and Gardner (7) used an electrically fused, 40 to 80 mesh,
magnesium oxide sand to mix with uranium tetrafluoride for removing fluorine from the combustion gases. About 40 grams of the magnesium oxide was used to mix with and cover a 3-gram sample for the determination. Because of the necessity for determining carbon in uranium tetrafluoride a t a very low level, a method in which fluorine compounds did not interfere with the determination would be more desirable than those described where fluorine compounds must be removed. Of the methods described the size of the uranium tetrafluoride samples ignited varied from 0.5 to 3.0 grams. If the sample contained 50 p.p.m. carbon, the weight of carbon dioxide obtained would amount to only 0.1 to 0.5 mg. Problems of weighing in this order of magnitude, where essentially complete recovery and separation of carbon dioxide from fluorine compounds is necessary, are quite obvious. To overcome these problems, an investigation was made of the feasibility of determining carbon in uranium tetrafluoride by igniting the sample a t 1100” C. in a stream of oxygen, collecting the combustion gases in an evacuated cylinder, and subsequently analyzing the gases by infrared spectroscopy. Neither oxygen nor fluorine combustion products interfere with the infrared determination of carbon dioxide. EXPERIMENTAL
Apparatus. A schematic of the ignition apparatus is given in Figure 1. The combustion gases were analyzed VOL 34, NO. 9, A U G U S T 1962
11 19