Determination of Oxygen, Hydrogen, and Nitrogen in Refractory Metals SIR: A vacuum extraction technique for determining ox) gen, hydrogen, and nitrogen in refractory metals has been developed in this laboratory, and has been used since late 1957 in the routine analysis of molybdenum, 0.5% titaniummolybdenum alloy, tantalum, and tungsten. The vacuum extraction of gaseous elements from solid metals is not new. Holm and Thompson ( 3 ) in 1941 described a n extraction method for the determination of hydrogen in ferrous metals. I n November 1952 several vacuum fusion analysts, including the senior author of this communication, unofficially reported quantitative recovery of oxygen from molybdenum samples that failed to dissolve in the iron bath during attempted vacuum fusion analysis (6). Mallett and Hansen have described the deterinination of nitrogen in molybdenum (6) and of oxygen in niobium (2) by vacuum extraction. I n general, however, this relatively simple method seems to have received little attention. The purpose of this Communication is to point out the reliability of the results obtained by vacuum extraction and the applicability of the technique to the refractory metals in general. As it is not likely that necessary additional FT-ork on this method can be completed in this laboratory in the near future, i t !vas decided to publish the data thus far accumulated, with the hope that others might participate in the further evaluation and development of the technique. The advantages of vacuum extraction relative to conventional iron bath vacuum fusion analysis are severalfold: Many more samples can be analyzed per crucible, smaller blanks are obtained in normal outgassing time, and initial preparation for analysis is simplified. I n addition, gettering and fogging of the sight TvindoIy by condensed bath metal are eliminated samples can be recovered intact after analysis, and crucibles and funnels can be re-used several times. PROCEDURE
Conventional vacuum fusion equipment is used. Assembly and outgassing are accomplished in the usual manner. The outgassing temperature is held to 2300' C. maximum t o avoid excessive carbon deposition on the furnace walls. A11 determinations are made a t 2000' C. A blank rate of 12 to 15 micron-liters
Table 1.
Typical Results
Vacuum Extraction, P.P.M. Iron-Tin Bath, P.P.M. Sample 02 Sz Hz 02 h'z H2 b l o rod, 170-mil (sintered and swaged) Av . 7.6 3 . 0 2.6 7 . 3 2 . 5 0.58 S.D. 2.7 0.69 0.24 1 . 5 1 . 9 0.25 No. 11 11 11 4 4 4 W rod, 125-mil (sintered and swaged) Av . 2 . 9 0.36 0.37 2 . 8 1 . 4 0.16 S.D. 0.60 0.39 0.30 1 . 3 1 . 1 0.054 No. 10 10 10 5 5 5 W powder Av. 1340 52 8.8 1194 45 15 ... S.D. 13 3.6 3.7 ... NO. 6 6 6 2 2 ' 2 Ta poxder
AV.
S.D. YO.
1890 ... 2
10 ... 2
... 2
1867 82 7
84 34 7
1.6
77
6
3
7 4 7
1.8
Av. rlverage S.D. Standard deviation No, Number of replicates a Individual samples, single determinations.
per hour is readily achieved in normal outgassing time. The composition of the blank averages about 75y0 carbon monoxide, 20% hydrogen, and the balance noncondensable (presumably nitrogen). Samples (about 0.2 to 2.0 grams) are dropped from the loading arm into the bathless, outgassed crucible a t 2000' C. The crucible temperature is held constant throughout the determination. Gas evolution starts immediately after the sample falls into the crucible. A thermocouple vacuum gage and strip chart recorder are used to monitor the evolution of gases from the sample (4). Collection time is normal: 20 to 30 minutes. The collected gases are analyzed manometrically in the usual manncr. RESULTS
Typical results obtained by the routine application of the vacuum extraction technique for a variety of metals are compared with iron-tin bath vacuum fusion results in Table I. The extraction technique leaves little to be desired relative to iron bath vacuum fusion analysis. The advantages claimed are realized in practice with no special precautions required. The net saving in time for the analysis of refractory metals is of the order of a factor of 4, compared to the iron bath tech-
nique (twice as many samples are analyzed in half the time). I n fact, the only sample limit seems to be the physical capacity of the crucible (approximately 24 samples). The most surprising result of this m-ork is the observed agreement between methods for the titanium-molybdenum alloy samples. Satisfactory results for titanium-bearing metals Tvere not expected by vacuum extraction. In general, the nitrogen results are probably of the same questionable validity a s most vacuum fusion nitrogen results-see ( 5 ) , however. Unfortunately, results by Kjeldahl analysis are not available for comparison. Until the absolute reliability of the nitrogen determinations is established, it seems appropriate to view these results with the usual suspicion. The repeatability of the nitrogen results is good, however, so that they are of analytical value in the relative sense. The large disparity lietm-een nitrogen results for the tantalum sample is unexplained. -4pparently the reduction of oxygen in the metal samples is substantially a surface phenomenon-that is, oxygen diffuses through the metal to the surface and then reacts with carbon (vapor) and is evolved as carbon monoxide. A sample of pressed molybdenum powder (1600 p.p.m. of oxygen, 38 p.p.m. of VOL. 31, NO. 6, JUNE 1959
1 1 15
nitrogen and 15 p.p.ni. of hydrogen) was analyzed for carbon before and after vacuum extraction analysis. Within the error of the determination, the molybdenum, after its several-hour stay in the graphite crucible, did not pick up a significant amount of carbon (90 p.p.ni. before, 70 p.p.m. after, =t15 p.p.m.). The solubility of carbon in molybdenum a t 2000" C . i5 of the order of 150 p.p.m. (1). TVhatever the details of the mechanism are, it seems clear that the rate of diffusion of the gaseous elements (except, perhaps, nitrogen) through the refractory metals is relatively great. Samples successfully analyzed by vacuum extraction h a r e
included both sintered and arc-melted stock, and have been as thick as 5 mni. ACKNOWLEDGMENT
The carbon determinations were provided by Agnes MclIichael. LITERATURE CITED
(1) Few,
W. E., llanning, G. K., J .
Metals 4, 271 (1952).
( 5 ) Rlallett, RI. W.,Hansen, PV. R. (HarmoEd, J., ed.), "The Metal Molyb-
denum, Chap. 16, pp. 391-2, -4m. Soc. Metals, Cleveland, Ohio, 1958. (6) Subcommittee on Analysis of blolybdenum for Small Traces of Oxygen, Nitrogen, and Other Gases, ONR Advisory Committee on Rlolybdenum, Schenectady, N. Y., Nov. 17, 1932. J. E. FAGEL R. F. KITBECK H. A. SMITH
12'1 Hansen. K. R.. Mallett. lf. .~ W.. ' ANAL.C I ~ E X29,1868 (i957): (3) Holm, V. C. F., Thompson, J. G., J . Research Natl. Bur. Standards 26,
Refractory Metals Laboratory General Electric Co. Cleveland 17, Ohio
(4) llcDonald, R. S., Fagel, J. E., Balis,
RECEIVEDfor review March 1, 1959. Accepted hpril 10, 1959.
245 (1941).
E. W., ANAL.CHEM.27, 1632 (1955).
Quantitative Determination of Traces of Pyrophosphate in Orthophosphates SIR: Colorimetric analysis of small quantities (less than 1%) of pyrophosphate in the presence of orthophosphate has been reported (1). Although this technique is rapid, we find that i t is useless for quantities of pyrophosphate below ca. 0.1 weight % ' as NazHzPzOi Furthermore, the method is subject to considerable error in analysis of commercial orthophosphates containing small amounts of divalent and trivalent metals. Aluminum, which is found in some industrial grades of orthophosphates, is a n important source of error in the Chess and Bernhart procedure, causing the colorimetric values to be as much as two or three times larger than the true pyrophosphate content. It has been found inadvisable to employ a correction for this error in routine determinations, as aluminum (or iron) can be present in various forms which do not respond the same way in the scheme of competing reactions between complex ions that form the basis of the Chess and Bernhart procedure. Related to this, n e found that their method of eliminating iron interference through use of a blank does not work. Attempts to improve the colorimetric procedure by adding coniplexing agents, such as (ethylenedinitri1o)tetraacetate and triethanolamine, were not successful. Attempts to improve the procedure by addition of complexing agent(s) are inadvisable, as the chemical system is already very complicated. A t present, i t seems t h a t the best answer to the problem of determining small amounts of pyrophosphate in orthophosphates is found in paper chromatography (S), using a large num1 1 16
ANALYTICAL CHEMISTRY
ber of overloaded sample spots. Aluminum interferes and must be removed, as through precipitation by 8quinolinol and extraction of the aluminum quinolinolate, which was accomplished with chloroform in a manner similar to that used by Gentry and Sherrington (2). The chromatographic separation was carried out by placing 150 pl, of solution containing 2.5 nig. of solid sample in 30 spots along a starting line 16 em. long. Weak pyrophosphate bands were located by the aid of reference spots on both sides. The technique used for quantitative evaluation of the resulting patterns was the same as described ( S ) , except that a calibration curve n-as established for the pyrophosphate (4). Because it is sometimes difficult to obtain pyrophosphate free orthophosphate, calibration with pyrophosphate alone is identical to calibration with pyrophosphate mixed into a truly pyrophosphate free orthophosphate (4). LITERATURE CITED
(1) Chess, W. B., Bernhart, D. N., AXAL. CHEM.30, 111 (1958). (2) Gentrv. C. H. R.. Sherrington, L. G.,
. Analvst-71. 432 (19k6). (3) Karl-Kroupa, 'E., h A L . CHEM.28, 1091 (1956).
(4).Karl-Kroupa, E. , private communica-
tion.
c. D . SCHhlULBACH Pennsylvania State University University Park, Pa.
SIR: We appreciate the work that Dr. Schmulbach has done and feel that he has some very valid points. I n our original v a r k we lvere mainly
concerned with the commercial grades of phosphate of high purity. The aluminum content of these materials was lovi and our blank method for iron correction gave good recoveries in spite of considerable variation in iron content. We found that for quantities of pyrophosphate below 0.2 weight yo NazH2P207 the higher temperature of 38" C. was desirable. As far as paper chromatography is concerned, Tve have used the method for the purpose extensively and agree it is good. However, the use of the small sample was always such that our precision was not good in the micro range. At the present time we are using ion exchange chromatography, which vie believe is by far the best method for this purpose. A 20-mg. sample is added to a column using Dowex 1-X8 anion exchange resin (100 to 200 mesh). All the orthophosphate is eluted with a solution which is 0.05iM in hydrochloric acid and 0.1M in potassium chloride; 300 ml. is required, a t a downflow rate of 15 to 18 ml. per minute. The polyphosphates are then eluted with 100 ml. of 1-If potassium chloride (same rate), hydrolyzed with acid and read colorimetrically. The large sample size increases the accuracy of n-eighing and the sensitivity of the amount of polyphosphate which may be detected. This method (not published) is based on work done by R. H. Kolloff a t Rfonsanto and H. J. Tt'eiser at Procter and Gamble, along with the other people who have recently done work on ion exchange resins.
W.B. CHESS T-ictor Chemical Works D. PI'. BERXHART Chicago Heights, Ill.
