Table
d , A. 1/11 39. Tetrasilver tetrametaphosphimate A&(POINH)~ 7.22 100 4.82 9 6 3.75 28 3.60 3.24 55 3.16 8 3.00 33 2.96 14 2.83 32 2.79 36 2.55 13 2.51 11 2.41 40 2.26 6 2.23 6 2.16 4 2.11 4 2.04 5 1.983 5 1.933 16 1.866 4 1.807 13 1.771 4 1.713 5 1.681 9 1.616 6 1.583 11
II. Powder Diffraction Data (Continued)
d , A. 1/11 40. Bis(o-tolidine) tetrametaphosphimate octahydrate (C14Hie")nIHdP02")r I. 8Hz0 17.2 100 8.72 53 7.23 9 6.70 4 5.68 6 5.37 4 4.98 35 4.84 4 4.70 4 4.41 12 4.33 13 4.19 13 4.04 4 3.80 17 3.57 13 3.53 13 3.36 8 3.28 13 3.10 10 3.05 8 2.98 10 2.95 10 2.86 8
salts, were analyzed also for ammonium nitrogen by the method of Varner and his associates (21). This method is capable of distinguishing ammonium nitrogen from amido as well as imido nitrogen. Total nitrogen was determined by the micro-Dumas method, phosphorus by the microphosphomolybdate method, carbon and hydrogen by microcombustion, and sodium and potassium by flame photometry. The data agree generally with those derived from published lattice constants for the phosphonitrilic chlorides (1, 7 , 8 ) , monosodium phosphoramidate (6), and tetrametaphosphimic acid (tet-
d , A. 1/11 40. Bis(o-tolidine)
tetrametaphosphimate octahydate
( Continued) 2.80 7 2.69 3 2.60 5 2.49 4 2.40 3 2.33 1 n n, " L.Ll
a
2.14 2.10 2.06
3 2 4
41. o-Tolidine orthophosphate (CiJWV")(HaPOa) 13.8 100 9.75 15 6.92 12 6.46 3 6.19 13 5.99 8 5.72 30 5.13 20 4 80 11
d , A. 1/11 41. o-Tolidine orthophosphate (Continued) 4.60 7 4.46 34 4.37 55 4.25 36 6 3.79 3.62 21 3.47 44 3.39 20 3.22 10 3.14 12 2.98 7 2.77 6 2.70 3 2.66 3 2.59 5 2.55 5 2.46 1 2.42 2 2.34 6 2.25 5 2.18 3 2.15 3 2.10 3 2.01 2 1,966 6 2 1 ,870 2 1.807 2 1.765 4 1.713 1.542 1
raphosphonitrilic acid) and its dipotassium, dirubidium, and diammonium salts ( 2 ) . Substantial agreement is noted with published powder patterns for the phosphonitrilic chlorides and the trisodium trimetaphosphimate mono- and tetrahydrates (20)) as well as ammonium phosphorodiamidate (4). The pattern reported for phosphoramidic acid (mono-amidophosphoric acid) (5) is identical with that given here for phosphoramidic acid monohydrate. The ease with which water is lost probably accounts for the difficulty of others in Preparing this hYdrate ( 9 ) : The transition temperature
-
for HzP03NHz.HzO H2P0&H2 is approximately 31' C. ( I d ) . ACKNOWLEDGMENT
The authors wish t o thank Ralph
R. Ferguson for assistance in preparing the powder patterns and J. W. Edwards for helpful suggestions during the course of the work, LITERATURE CITED
(1) Audrieth, L. F., Steinman, R., Toy, A. D. F., Chem. Revs. 32,109 (1943j. (2) Corbridge, D. E. C., Acta Cryst. 6, 104 (1953). (3) Gbel, J. P., Eolas, J., Busch, N., Bull. SOC. chim. France 1955, 108793. Goehring, bI., Niedenzu, K., Chem Ber. 90, 151 (1957). Goehring, M Sambeth, J., Ibid., 90, 232 (19i7). Hobbs, E., Corbridge, D. E. C., Raistrick, B., Acta Cryst. 6, 62 (1953). Jaeger, E. XI., Beintema, J., Proc. Acad. Sci. Amsterdam 35, 156 (1952). Ketelaar, J. A. A., Vries, T. A. de, Rec. trav. chim. 58, 1081 (1939). Klement, R., Z. anorg. Chem. 260, 267 (1949). Xlement, R., Becht, K.-H., Ibid., 254. 217 (19471. (11) Klement, R., Biberacher, G., 2. anorg. u allgem. Chem. 283, 247 (1956). (12) Nielsen, &I. L., unpublished data. (13) Nielsen, M. L., Nielsen, W. W., Micro-Chemical J., to be published. (I4) Schenck, R.1 Rbmer, G., Ber. Chem. Ges. 57B, 1343 (1924). them. J . 15, (15) Stokes, H. N., 198 (1893). (16) Ibid., 16, 123 (1894).
f!:::; F'
(19) Ibid,, 20, 740 (1898). (20) Thilo, E., Ratz, R., 2. anorg. Chem. 258,33 (1949). (21) Varner, J*E., Bulen, A., Vanecko, S., Burrell, R. C., ANAL. CHEX 25, 1528 (1953).
w.
RECEIVEDfor review November 4, 195i. Accepted May 5, 1958.
S pect rogra phic An a I ys is of Si Iico n- G e rma nium Alloys MARVIN C. GARDELS and HUBERT H. WHITAKER RCA Laboratories, Radio Corp. of America, Princeton,
b During a study of the optical properties of silicon-germanium alloys, a spectrographic method was developed for the binary system over a concentration range of 1 to 98 mole Yo silicon. Only graphite was added as a diluent; no internal standard was added. Germanium served as a variable internal standard for silicon determination up to 70 mole yo silicon, 1496 *
ANALYTICAL CHEMISTRY
N.J.
and silicon served as a variable internal standard for germanium up to 30 mole yo germanium. By a proper choice of line pairs and exposure conditions it was possible to cover this concentration range by using three silicon lines and one germanium line. The standard deviation was less than i 8 % over the entire range.
I
detailed study (6) of the intrinsic optical absorption spectra of the germanium-silicon alloy system i t became necessary to analyze these alloys quantitatively. Earlier work ( 9 ) had sh0n.n that up t o 15 mole % silicon could be determined spectrographically. However, the method as developed a t that time did not have the required accuracy to be extended to higher conN A
-
I
I
I I I1111
I
I
I
I I1111
I
I
-
I I I Ill
h
I
-
-I
3 CC
4
0.1
I 2
I
I
I I1111 5
Figure 1.
I
I
I
0
-
WT % S I MOLE % S I
I I IIII
10 20 50 S I L I C O N CONCENTRATION
100
I 200
I
I
I Ill1 500
IO00
Group of analytical curves for silicon
centrations. Consequently, a more accurate polarographic method for germanium was used to cover the rest of the concentration range. At the start of the spectrographic analysis in this study it was soon discovered that more accurate results could be obtained by mixing the sample with powdered graphite t o give more even burning characteristics. Thus, it was believed possible that the concentration range could be extended and still have the required accuracy. This was confirmed by Anthony, Chandler, Huckabay, and Kenner (4),who analyzed powdered rock samples for silica. I n their procedure, rocks containing up to 70% silica (SiOa), were analyzed by introducing a constant amount of germanium as the internal standard. They also used powdered carbon as a diluent. I n the present study, germanium was a mriable internal standard, as no additional germanium was added. After it was found that silicon could be determined up to 70 mole % by using germanium as an internal standard, it was of interest to see if the complete range could be covered spectrographically by determining germanium, using silicon as the variable internal standard. Up to 30 mole % germanium could be determined by using a single line pair. Two line pairs were required to cover the silicon range. No attempt was made t o overlap the range of 70 mole 70silicon, although such should be possible by choosing additional line pairs and adjusting the exposure conditions. EXPERIMENTAL
Apparatus. A Hilger Large-Littrow quartz spectrograph was used for obtaining the spectrograms. Excitation was obtained from a Bausch and Lomb direct-current arc unit set a t 10 amperes. Transmittance . d a t a were obtained on a Jarrell--4sh
nonrecording console microphotometer. A Spex Industries Wig-L-Bug was used for grinding and mixing samples. Spectrum analysis S o . 1 plates were used throughout and were developed in Eastman D-19 developer in a Jarrell-Bsh plate processor. The electrodes used were X'ational Carbon L-4006 3/16-inch diameter necked electrode and L-4038 1/8-inch diameter pointed electrode. Materials. Du Pont semiconductor-grade silicon was used. The germanium was obtained by reduction of very high purity Eagle Picher germanium dioxide. The synthetic standards were prepared by melting together weighed portions of germanium and silicon and then quenching. The graphite powder (Spex Industries) was spectroscopically pure. Procedure. PREPARATIONOF SAMPLES. The samples and standards n-ere ground in steel capsules, using the Wig-L-Bug. The entire sample r\-as ground and mixed to minimize errors due t o segregation effects. Ten milligrams of the sample mere then mixed with 90 mg. of graphite in a plastic capsule, using the Fig-LBug. I n the initial experiments, 10 nig. of the above mixture was weighed into the graphite electrode. However, in later experiments, it was found that satisfactory results could be obtained by measuring a constant volume in a small dipper made for that purpose. By using the dipper, which held a 12mg. sample, it was found that the amount of the sample measured was reproducible to better than &lo%. PREPARBTIOX OF SPECTROGRBMS. The exposures were made by opening the shutter before striking the arc. The sample was then burned to completion, which required about 2 minutes. The exposure was controlled by a rotating sector to pass about 3% of the light, and had to be individually determined for each range of concentration used. The analytical gap was maintained a t 4 mm., and the slit used was 25 microns.
Table I. Mole Per Cent Silicon in Silicon-Germanium Alloys Si-2435 Ge-2498 SampleSi-2506 Ge-2498 Si-2835 Xo, Ge-2498 1 6.4. 6 . 6 7.0. 6 . 5 2 6 . 7 ; 6.7 6.71 6 . 5 3 6.3, 5 . 8 6 . 2 , 5 . 6 4 10.1, 10.0 10.4, 8 . 2 5 13.0, 13.2 12.4, 11.8 6 12.5. 12.8 13.0. 13.7 7 13.0; 13.7 15.8; 13.2 8 13.0, 15.0 14.3, 14.1 26.0, 24.5 9 26.7, 24.1 10 26.8, 24.6 11 20.3, 28.2 12 28.7. 33.6 13 38.7, 34.8 14 50.0, 47.5 15 16 66.0, 64.0 17 68.5, 72.0 76.5, 77.5 18 96. 7, 96.6 19 97. 1, 97.3 20 96. 9. 96.6 21 96.5; 96.9 22 95.8, 95.7 23 94.2, 94.5 24 91.4, 90.9 25 95.6, 95.1 26 RESULTS
The emulsion was calibrated by a tIvo-step method (5). Frequent calibrations were made fo minimize errors due to change in strength of developer. A step-sectored iron spectrum was taken for the calibration, and the third and fourth steps of a seven-step spectrum were used for the calibration. The two-step method was chosen over the stepped sector method because of the difficulty of obtaining uniform illumination over the entire face of the sector. The iron lines chosen were in the wave length region t o be used in the analysis. The lines \\-ere chosen so as to give an even spread over the entire range of transmittance. The working curves were prepared from synthetic standards. The line pair Si-2506/Ge-2498 was used for 1 to 15 mole % silicon, and the pair Si-2435/ Ge-2498 was used for the 5 to 70 mole 70 range. I n each case a straight line working curre was obtained ivhen relative intensity ratio was plotted against mole per cent silicon. The working curve for the 5 t o 70 mole 70 range is shown as curve 4 in Figure 1. To go to higher silicon content ranges, germanium was used for the analysis line and silicon was used as the internal standard. The line pair Ge-2498/Si2535 was satisfactory for 2 to 30 mole yo germanium or 70 to 98 mole Yo eilicon. This working curve was also a straight line when the germanium concentration wis expressed in mole per cent. It is therefore possible to obtain an analysis varying from 1 to 98 mole Yo silicon by using only three silicon lines and one germanium line. The results of a number of analyses in duplicate are shown in Table I. VOL. 30, NO. 9, SEPTEMBER 1958
1497
The figures in the last column are obtained by subtracting the mole per cent germanium from 100. No independent chemical analyses were made to check the accuracy of the spectrographic results. However, four samples were chosen from Table I to check the precision of the analyses. The samples Tvere chosen to corer a range of concentrations and several runs were made on each sample. The standard deviation was calculated from the formula
where d is the per cent deviation and n is the number of determinations. Sixty-seven per cent of all determinations should be within one standard deviation of the mean value. A summary of these results is given in Table 11. The composition is given in terms of the element used for the analysis line and the standard deviation refers to this same element. DISCUSSION
The fundamental assumption in quantitative spectrographic analysis is that the intensity of an emitted line within a source of excitation is directly proportional to the concentration of emitting atoms. If an internal standard is used t o compensate for various factors which are difficult to control, then the equation relating intensity of concentration is I K = log I, K,
log 2 7
n C +2 log 2 ne C,
where I , and I , are the intensities of the internal standard line and analysis line, respectively; TZ, and n, are their respective emission factors, K . and K , their respective constants, and C, and C, their respective concentrations ( 1 ) . I n ordinary spectrochemical analysis the concentration of the internal standard is held constant so that a plot of log I,J8 us. log C, produces a straight line. I n a twocomponent alloy such as silicon-germanium alloy, it would seem that such a plot Fould produce a curved working curve because a variable internal standard is used. Herman (3,10) did find a curved working curve for the two component mixtures niobium pentoxide plus tantalum pentoxide \Then using one component as the internal standard with respect t o the other. Raikhbaum (11, 1%’) studied the lead-tin system spectrographically and obtained his results by picking out various line pairs of equal intensity and relating these to concentrations.
1498
ANALYTICAL CHEMISTRY
To obtain a straight-line plot using a variable internal standard, log I J I 8 should be plotted against log C./C,. This method of plotting was used by Coulliette (8) for the determination of nickel and chromium in stainless steel. He plotted log IPil/ICr us. log Ch-,/Cc, and log I F e / I C r us. log CF,/Ccr and obtained straight lines for what was essentially a three-component system. The concentration ratios could then be used to calculate the percentage of each of the three components. The same method was used by Churchill and Russell (7) for the determination of silicon, iron, and titanium in xluminum ores in which aluminum served as the variable internal standard. They referred to this method as the “mutual” standard method.
Table II. Summary of Standard Deviation Determination Sample S o . of Composition, Std. So.
7 10 13 22
Detns. 10 9 10 10
%
13.6 Si 23.9 Si 32.2 Si 3.42Ge
Dev., % &7.8 1.7.5 1.5.2 1.4.6
The method of plotting the working curves was chosen so as to give the answer in mole per cent directly, rather than through a series of calculations. However, it was somewhat surprising to find that these working curves were straight lines over such a large concentration range. To determine a reason for this, a plot of log I s l / I G ,us. log wt. % Si/wt. % Ge was made. This should be a straight line. However, curve 1 of Figure 1 shows that the curve bends ton-ards the concentration axis a t higher concentrations. This is indicative of self-absorption in the analysis line ( 2 ) . If a mole per cent ratio is plotted, the curve would be parallel to the previous curve because Tt. % Si - mole % Si wt. % Ge mole % Ge
mol. wt. of Si mol.wt. of Ge
Curve 2 of Figure 1 s h o w such a plot. Curve 3 shows weight per cent silicon as the abscissa while curve 4 shows mole per cent silicon, These latter two curves would not be parallel as weight per cent silicon is not directly proportional to mole per cent silicon. It would seem, therefore, that the straightline plots obtained when using mole per cent as the abscissa are due to a compensation for self-absorption of the analysis line rvhen a ratio plot is used.
Although any of the four curves could have been used, curve 4 is most convenient because it is a straight line and gives the required answer directly. Although no statistical data were obtained for standard samples, i t seems that the accuracy of the determination would be comparable to the precision obtained on the unknown samples tested. As there are only two elements present, the standards and samples would be very similar, and there would be no interference due to other elements. SUMMARY
Silicon-germanium alloys containing 1 to 98 mole Yc silicon have been analyzed spectrographically, using a total of three silicon lines and one germanium line. The line pair Si-2506/Ge-2498 is used for the 1 to 15 mole 7csilicon range; the line pair Si-2435/Ge-2498 is used for 5 to 70Y0 silicon range; and the line pair Ge-2498/Si-2535 is used for the 2 to 30 mole % germanium range, The standard deviation is less than =kS% of the amount of the element being determined over a range of concentrations. ACKNOWLEDGMENT
The authors wish to thank S. &I. Christian of these laboratories for supplying the silicon-germanium standards. LITERATURE CITED
(1) Ahrens,
L. H., “Spectrochemical Analysis,” p. 80, Addison-Wesley, Cambridge, Mass., 1954. (2) Ibid., p. 95. (3) I b i d , p. 203. (4) Anthony, J. W., Chandler, R. J., Huckabay, W. B., Kenner, C. T., ANAL.CHEM.28, 470 (1956). (5) Boltz, D. F., ed., “Selected Topics in Modern Instrumental Analysis,” pp. 202-9, Prentice-Hall, Xew York, 1952. (6) Braustein, R., Moore, A. R., Herman, F , Phys. Rev. 109, 695-710 (1958). ( 7 ) Churchill, J. R., Russell, R. G., ANAL.CHEM.17,24-8 (1945). (8) Coulliette, J. H., Ibid., 15, 732-4 (1943). (9) Gardels, 11.C., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Feb. 28-March 4, 1955; ANAL. CHEM.27, 322 (1955) (abstract). (10) Herman, H., Spectrochim. Acta 3, 389 (1948). (11) Raikhbaum, Ya. D., Zavodskaya Lab. 8, 1101-5 (1939). (12) Smith, G. S., Spectrochim. Acta 3, 238 (1948).
RECEIVED for review h’ovember 1, 1957. Accepted Bpril 18, 1958. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 1957.
