Determination of Carbon in Sodium by Isotope Dilution Mass Spectrometry KEN Y. ENG, RAYMOND A. MEYER,’ and CARLETON D. BINGHAM Atomics International, A Division of North American Aviation, lnc., Box 309, Canoga Park, Calif.
b A method utilizing isotope dilution mass spectrometry has been developed for the determination of carbon in sodium. The details of the method are described for the determination of “elemental” carbon employing Van Slyke oxidation, although, by choice of spike material, other forms of carbon may b e determined. Accuracy and precision in the ranges 50 f 10 pg. to 150 25 pg. of carbon have been demonstrated. The determination is not affected by loss of generated carbon dioxide or by contamination from any source except carbon.
*
T
CARBON-IN-SODIUM determination finds direct application in the high temperature sodium-cooled nuclear reactor system. Metallurgical experiments reveal that carbon concentrations in the 100-p.p.m. range are capable of carburizing and thus weakening stainless steel (1). This problem is crucial in the thin fuel element cladding where rupture causes serious release of fission products. To meet the carbon control requirements, the analytical technique must measure 10 p.p.m. of carbon with an accuracy of < + 5 p.p.m. Several analytical techniques for the determination of “elemental” carbon in sodium are described in the literature (4-6). The principle of the existing techniques is the oxidation of carbon to carbon dioxide and the subsequent measurement of the quantity of evolved gas. The gas measurement techniques reported are manometric, conductometric, or gas chromatographic (4). Several of the techniques were tested during the First Round Robin Analysis for Carbon in a program sponsored by the Atomic Energy Commission Sodium Component Development Program Working Group. The highly scattered data indicated in the report ( 3 ) did not permit comparison of the analytical methods. The scatter could, however, equally well be attributed to sampling, sample preparation techniques, or analytical technique. While the available analytical techniques are adequate HE
1 Present address, North American Aviation Science Center, Canoga Park, Calif.
1832
ANALYTICAL CHEMISTRY
from technical considerations, their accuracy depends on the complete conversion and recovery of all carbon. Although the isotope dilution technique eliminates the problems of complete recovery and contamination from most sources, it remains sensitive to carbon contamination. Analytical sensitivity may be optimized by adjusting the sizes of the added isotope (spike) and the sodium sample. THEORETICAL CONSIDERATION
The theory of the analysis (based on the material balance of tracer) has been well described by Calvin, Heidelberger, Reid, Tolbert, and Yankwich ( 2 ) . Stable C13 was chosen as a tracer because of its similarity to the elemental carbon in the sample and its lack of radioactivity. If z is the w i g h t of C1* in the sodium sample and y is the weight of C13 added to the sodium sample, the resulting ratio z = x/y may be measured. If z and y are known, z can be calculated. The simple ratio as given above must be expanded to include corrections for the occurrence of 1.1% C13 in natural carbon and the isotopic purity of the spike. Let
R,l3
=
Rf13
=
X
=
WB Ws Then
=
~
~
=
initial C13/C weight ratio in spike carbon; observed C13/C weight ratio in final gas; unknown weight of carbon in sample, pg.; weight of carbon blank, pg.; weight of carbon in spike, pg.
=1
o.o11(X f 3
x+
+ J?,131vS +
W E ) TvS
m’B
Solving for X, one obtains X =
can be seen from the formula, X can be calculated after the R i l 3 determination because R,’3, TI’,s, and 1T-B are known. In practice, a weighed amount of a known mixture of CI2 and Cl3 is added to the sample. The carbon already in the sample is assumed present in the natural isotopic
ratio and, when this carbon mixes wit’h the enriched C13 addition, the resulting mixture will have a new C13:’C ratio. If the carbon is converted to a gaseous form, such as carbon dioxide, the isotopic ratio is easily determined with a mass spectrometer. The sensitivity of the isotope dilution method depends significantly on the size of the spike used. For example, if a small amount of carbon in the sample, approximately 0.1 mg., should be diluted with a relatively large spike of C13, 100 mg., the ratio of C13/C will change by only 0.1% from that of the spike. However, a spike the same size as the sample would produce a change of 50%. Thus, sensitivity may be greatly improved by optimizing spike size. The method becomes extremely sensitive if the spike size can be accurately predetermined. Ideally, the spike material should be in the same chemical form as the element in the sample. The spike material used for this development was amorphous carbon with an enrichment to 50.4y0 of C13. d large sodium sample is beneficial since a greater amount of carbon is available for analysis; but, a t the same time, t’he larger quantity of sodium is awkward in the analytical procedure because of hazards in converting to the sodium sulfate form. A reasonable sodium sample size is in the range of 1 to 5 grams. EXPERIMENTAL
Preparation of Reagents and Spike Mixture. Of the several reported
carbon oxidation techniques, the Pepkowitz method was select’ed for this investigation. The oxidizer used was Van Slyke reagent consisting of 165 ml. of fuming sulfuric acid (15%), 13 grams of chromic acid, and 85 ml. of phosphoric acid heated to 140’ C. and cooled t’o room temperature under vacuum. The enriched spike mixture was prepared by diluting amorphous carbon, enriched to approximately 50 wt. yo in CI3 (Isotope Specialties Co., Rurbank, Calif.), 1 : 1000 with reagent grade sodium sulfate by slurrying under ether and grinding with a mortar and pestle. Determination of Reagent and Diluent Blanks. During blank run$i.e.! without sodium-it was adcertained that CO, is generated from the Van 81yke reagent or is obtained from
-TO MECH. PUMP
Ar IN-
r
4OO'C-
ASCARITE-J
v,'
A FLOW METER
13 COLD TRAP--50DC
B GREASELESS STOPCOCK FOR VAN SLYKE INTRO c COLD TRaP--50°C
II REACTION FLASK
Figure 1 . apparatus
lr LIQUID Np COLLECTION TRAP
Schematic
drawing
G EXPANSION BULB H Ll9UlD N2 MASS SPEC COLLECTION TRAP J Hp RING SEAL
of
carbon
analysis 50
the sodium sulfate diluent and/or leakage during the analytical process. The CO, blank causes analytical error and must either be eliminated or accurately determined. Both approaches have been pursued t o minimize the problem of t h e existence of the blank. At the present time, the reagent blank has not been completely eliminated but has been reduced. After repeated observations of the dilution of a spike in the absence of a sodium sample, the reagent and diluent (Sa2S04) blank were determined to be 22 & 3 pg. of C per 20 ml. of Van Slyke reagent and 13 .+ 1 pg. of C per 100 mg. of Na2S04,respectively. Sampling and Samyple Preparation. Inasmuch as t h e samples received for analysis differ in size configuration and container, only general statements with regard to sampling can be made. Sample conta,i.ners are opened in a n inert atmosphere dry box. Core samples are removed and placed in a reaction vessel. When the sample container is a small diameter tube, the core is a complete radial cross section. The sodium is dissolved by bubbling argon through hot water, then over the sample. The resulting solution is acidified with 5M H2S04,heated to drive off CO,, and evaporated to near dryness before the spike is added. The reaction flask is connected to the gas handling system, shown in Figure 1, and purged with dry, Con-free argon for 5 minutes. Van Slyke reagent (20 ml.) is added and the contents of the reaction flask are heated to a nominal temperature of 150" C. for 15 minutes. C02is condensed by liquid nitrogen while water vapor, SO2, etc., are trapped by a dry iceacetone bath. If manometric measurements are made for comparison purposes, the COz is collected first in trap F , then recollected in trap H . The trap is closed off and transflerred to the inlet of a Consolidated Electrodynamics 21-620 mass spectrometer for measurement of the isotopic ra,ltio. Isotopic Analysis. An assumption, based on calibration d a t a for other materials, was made t h a t t h e sensitivities for C1z02and C 1 3 0 2 were equal. Graham's law calculations showed the leak rate inequality caused by the difference in molecular weights to be 0,2y0and thus negligible in the overall analytical procedure. The glass U-tube was immersed in liquid nitrogen and evacuated to remove noncondensjble gases. .in acetonedry ice bath was substituted for liquid nitrogen and the carbon dioxide allowed to expand into
I
I
50
100
I
2W
1%
CARBON OBSERVED (pg)
Figure 2. added
Correlation of carbon observed with carbon
the mass spectrometer. This effectively prevented any water vapor in the sample from entering the mass spectrometer. The mass 44 and 45 peaks (caused by C1202 and C1302, respectively) were scanned at least five times and the calculated areas averaged to determine the C13/C ratio. Several determinations indicated a relative standard deviation of 0.5%. This precision can be improved to +0,2y0 of the ratio by special techniques should the rest of the analysis warrant. RESULTS AND DISCUSSION
Precision and accuracy of the technique were investigated by analyzing N a and Na2S04-carbon samples. Data from the carbon-in-sodium sulfate runs establish the accuracy of the method, as shown in Figure 2, where comparison of observed values and theoretical values is presented. Data from analyses of sodium with an unknown (but assumed homogeneous with each set) amount of carbon are presented in Table I. These data demonstrate the precision of the method. A correlation and regression analysis (7') on the C-in-NazS04 data were performed. A least squares fit to all the data points yielded a regression curve whose equation is y = 0.814X 16.2. A correlation coefficient of 0.9868 showed an extremely high degree of correlation between the observed and added values. Analysis of variance showed the correlation is significant at greater than the 99% confidence level. Because the success of the isotope dilution technique depends only on the measurement of the C13/C ratio and the absolute quantity of spike material used, the main advantage of this technique over others is in the elimination of the requirement for complete recovery of the COz. Since a knowledge of the absolute quantity of COz
+
generated is not necessary, the analytical procedure becomes simpler. The volume of gas is not the measured variable; therefore, the determination is not affected by air leakage or extraneous gaseous oxidation products. I n addition, the COz need not be separated completely from the carrier gases-e.g., argon, helium, oxygen, or nitrogen. The technique for introducing the COP into the mass spectrometer will remove all these gases except argon. A further advantage is that the sensitivity of the technique can be greatly increased by optimizing the spike and sample size, as has been described. The problem associated with the CO, blank, discussed earlier, is the only major technical disadvantage to this technique. As for the data in Figure 2, the bias observed a t the lower concentration range is probably related to the inappropriate spike size used relative to the quantity of carbon in the sample coupled with the effect of propagation of uncertainties. Because blank uncertainties contributed to the final uncertainty, it was decided to use a larger spike (100 pg. of C) than optimum to eliminate most of the contribution from any weighing error associated with the spike. Such a contribution would have
Table I. Experimental DataDemonstration of Precision of Method
Kominal 3-gram sample Concentration, p.p.m. Set 1 14 Set 2
22 17
Set 3
5 8 9 11
10
VOL. 36, NO. 9, AUGUST 1964
1833
increased an already large final uncertainty in the low concentration range. The data in Table I clearly show that precision of the method is well within the uncertainty limits quoted above. When problems of reagent blank, diluent blank, and spike form are better resolved, it is predicted that accuracy and precision values will converge on the value reported herein for precision. The source of and means for reducing the reagent blank are under investigation. Although only a technique for the measurement of elemental carbon has been described in detail, the use of a spike of carbide prior to sample dissolution and a spike of BaC1303 prior to acidification, followed in each case by collection of the gas and measurement of isotopic composition, would enable measurement of total carbon in a sample. The method as described demonstrates improved sensitivity and precision over previous chemical techniques. I n the range of critical in-
terest from the carburization of stain10 to 30 less steel standpoint-i.e., p.p.m., 50 to 150 wg. in a 5-gram *2 to 5 p.p.m. accuracy sample-a uncertainty (95% confidence limit) has been obtained. This is considerably better than the 10 to 20 i 10 p.p.m. essentially given as a lower limit for chemical procedures, and is within the initially established goals for the method. The technique has been tested for accuracy over the range of 10 to 150 pg. of elemental carbon which represents 2 to 30 p.p.m. in a 5-gram sodium sample. Extension to higher concentrations facilitates the analysis and should introduce no problems. Precision in the 10-p.p.m. range has been demonstrated to be below the deviations reported for accuracy.
LITERATURE CITED
(1) Anderson, W. J., Sneesby, G. V., “Carburization of Austenitic Stainless Steel in Liquid Sodium,” NAA-SR5282, September 1 1960. (2) Calvin, M., Heidelberger, C., Reid, J. C., Tolbert, B. M., Yankwich, P. F.,
~
“Isotopic Carbon: Techniques in Its Measurements and Chemical Manipulation,” Wiley, New York, 1949. (3) Lockhart, R. W., Sabol, W. W., ‘%esults of the First Round Robin Analysis for Carbon in Sodium,” GE-APED letter document, February
23, 1963. ( 4 ) Mungall, T. G., Mitchen, J. H., Johnson, D. E., ANAL.CHEM.36, 70 (1964). (5) Pepkowitz, L. D., Porter, J. T., 11, “The Determination of Carbon m Sodium,” KAPL-1444, November 1955. (6) Stoffer, K. G., Phillips, J. H., ANAL. CHEM.27, 773 (1955). (7) Volk, W., “Applied Statistics for
Engineers,” McGraw-Hill, New York, 1958.
ACKNOWLEDGMENT
The authors acknowledge the assistance of Rachel Seite Dunn and Robert Wilbourn, who performed the chemical manipulation of the samples prior to mass spectrographic analyses.
RECEIVEDfor review March 12, 1964. Accepted May 18, 1964. Presented at the Winter Meeting of the American Nuclear Society in New York, November 18, 1963. This work was sponsored by the Atomic Energy Commiesion, contract NO. AT-( 11-l)-GEN-8.
Spectrophotometric Determination of p -Chloroaceta nilide in Phenaceti n-Acid Hy d rolysis Method W. B. CRUMMETT, J. SIMEK, and V. A. STENGER Special Services laboratory, The Dow Chemical Co., Midland, Mich.
A procedure has been developed which is suitable for the determination of as little as 10 p.p.m. p-chloroacetanilide in phenacetin. Upon refluxing the sample with 48% hydrobromic acid, phenacetin is converted to p-hydroxyaniline hydrobromide whereas p-chloroacetanilide yields pchloroaniline hydrobromide. If the mixture is made alkaline, p-chloroaniline can be extracted selectively with cyclohexane and determined by ultraviolet spectrophotometry. No significant interferences have been encountered in the analysis of commercial samples. b
A
to Harvald and coworkers (S), undesirable side effects sometimes observed in patients who have taken considerable phenacetin may be ascribed to p-chloroacetanilide which was present as an impurity. Because the amount of p-chloroacetanilide present in commercial phenacetin may vary from 0 to 2500 p.p.m. depending on the process of manuCCORDING
1834
ANALYTICAL CHEMISTRY
facture and the regulations of the country in which it is sold, it becomes important to have an analytical method which will determine precisely and accurately the p-chloroacetanilide actually present. The polarographic method of the Raney-nickel Jones and Page (4, hydrogenolysis method of Hald (a), and the paper chromatography method of Ritter and coworkers (5) have detection limits of about 500 p.p.m. The paper chromatographic method has been modified by a USP committee (9) to determine the presence of 300 p.p.m. A second spot on the paper chromatogram is, sometimes observed and may be mistaken for p-chloroacetanilide since the R, values are very close. This spot has been shown by N. E. Skelly of this laboratory, to be caused by innocuous N,N-diacetyl phenetedine (6). The authors have sought for a method which would be more sensitive while retaining suitable specificity. Upon acid hydrolysis, phenacetin should yield p - aminophenol ( p - hydroxyaniline)
while p-chloroacetanilide should yield p-chloroaniline. Both compounds are soluble in aqueous acid solutions, but from an alkaline solution only the chloroaniline can be extracted by an organic solvent; the other compound remains in the water layer as a phenoxide. Ultraviolet spectrophotometric determination of p-chloroaniline in an organic solvent is sufficiently sensitive for the purpose. Hydriodic acid is usually used for the cleavage of alkoxy compounds, but there are difficulties with formation of free iodine. One of the authors has pointed out (7) that 48% hydrobromic acid is practically as effective and he has utilized it in an unpublished method for the determination of bis(p-chlorophenoxy)methane by hydrolysis to p-chlorophenol. Much earlier, a mixture of 48% hydrobromic acid with acetic acid had been employed by Stoermer (8) for the dealkylation of phenyl ethers; in the case of anisole he reported 85% conversion. More recently, Anderson and coworkers , ( I )
