Spectrographic Determination of Sodium in a Silica-Alumina Catalyst JOHN B. MARLIYG', Diakel Corporation, Cincinnati, Ohio
A spectrographic method for the determination of sodium in a silica-alumina cracking catalyst is described. The procedure involves the use of a copper counter electrode to provide reference lines and is especially suited for analyses in which an internal standard is not available. Determinations of the sodium content in the range 0.001 to 3.0% sodium oxide are obtained in approximately 40 minutes for routine industrial control.
T
finished plate, although, when time permitted, a longer period was used, since prolonged washing further reduces the red tint remaining from the emulsion-sensitizing dye.
HE most generally accepted procedure for the chemical determination of sodium in silica-alumina catalysts is the gravimetric method, which employs zinc uranyl acetate reagent (1, 7). The sodium is precipitated as sodium-zinc uranyl acetate, XaZn(U02)3(C2H302)s.6H20, over a 45-minute interval, washed with 95% ethyl alcohol saturated with sodium zinc uranyl acetate, dried, and weighed. This chemical procedure requires a great deal of sample preparation and considerable time, particularly for samples that contain interfering cations or only traces of sodium @),and it has been the author's experience that the results obtained by semiskilled analysts are frequently inaccurate. Several alternate procedures have the advantages of speed and a higher accuracy than is ordinarily obtained in routine chemical analysis. The increase in accuracy is derived from averaging large numbers of chemical analyses for sodium in order t o calibrate these procedures, thus averaging the deviation in the chemical results. Briefly, the procedures consist of the use of the flame photometer (d), the nephelometer (e), and the spectrograph. The first two methods are recommended for the analysis of solutions. The spectrograph seems to be the best available means of analysis when the sodium is contained in an insoluble and inert material, as the time and labor required to prepare solutions from the analytical samples would exclude the photometer and the nephelometer. The silica-alumina cracking catalysts, as used in the oil industry, belong to this latter category of inert and insoluble materials. For the specific catalyst analyzed, the silicon content was approximately 85c4 silica ~ i t ah maximum content of sodium of 0.02% as sodium oxide.
PRELIMINARY INVESTIGATIOV
The excitation source was a 2300-volt, 2.5-ampere, alternating current arc. In order to reduce the hazard t o the operator a safety circuit was designed, which consisted of a remote control switch that automatically remsined open unless the operator maintained pressure on it. This switch was so positioned that it was impossible to hold it closed and touch the electrodes, although the operator was able to view the focus of the arc relative to the slit. Three signal lamps were inserted in the safety circuit: one directly in front of the operator, one over the door, and the third conspicuously located in the spectrographic laboratory. A survey of the most sensitive available sodium lines is given in Table I (3, 5 ) .
Table I.
Sodium Lines Available for Analysis
Wave Length, .i,
Sensitivity
5895 5890
3303 3302 2852
Preliminary investigation indicated that for the range of sodium concentration in tbe finished product (0.027, or less) the D lines at 5890 and 5895 A. were the only ones showing in the exposures. This was unfortunate from the standpoint of the spectrograph employed, as the reduction in dispersion in the range 5000 to 7000 A. did not permit resolving the D lines. The preliminary investigation also disclosed that the internal standard available was silicon, an element emitting comparatively few lines. The only availabl: silicon lines in the vicinity of the sodium D lines were at 4100 A. and, thus, not homologous from the standpoint of wave length (4). In addition, the intensity of these lines was entirely too high. Instead of silicon, a substitute internal standard was utilized. Copper offered convenient homologous lines, as well as ease of preparation. Pure copper, obtained by precipitation from C.P. copper sulfate solution by addition of C.P. zinc, was mixed in 1 to 1 proportions with the material to be checked for sodium content. The mixture of copper and sample was placed in the crater of a high purity graphite electrode, obtained from the Dow Chemical Co., and volatilized in the alternating current arc, The main objection to the above method was sodium contamination. Blank determinations very frequently produced sodium percentages of the same order of magnitude as the routine analyses. The correlation between chemical and spectrographic results was very poor. In an effort to reduce contamination, a copper electrode was substituted for the graphite counter electrode and powdered cop-
APPARATUS
A Bausch & Lonib medium quartz spectrogmph was used. This instrument covered the region 2200 t o 7000 A. with one 25-cm. (10-inch) exposure, thereby making a large portion of the ultraviolet and all of the visible spectrum simultaneously available. However, with this type of spectrograph the reduction in dispersion in the range 5000 t o 7000 A. is very pronounced and a small grating instrument with high efficiency in the first-order spectrum might be more suitable. Auxiliary equipment consisted of a Leeds & Northrup recording microphotometer, an electrode cutter, a plate-developing machine, and a plate dryer. Dow high purity graphite electrodes were employed. PHOTOGRAPHY
Eastman panchromatic process plates were used t o photoqraph both the ultraviolet and visible spectrum. These plates were processed for 5 minutes in D-8 developer a t 70' F., washed for 3 minutes in water, rinsed in ethyl alcohol t o remove the red emulsion dye, and fixed in standard sodium thiosulfate solution for 2 minutes. An additional 5 minutes was utilized for washing the 1 Present address, Los Alamos Scientific Laboratory, Box 1663, Los Alamos, N. M.
299
300 Table 11.
ANALYTICAL CHEMISTRY Range of Sodium Concentrations, Line Pairs, and Exposure Times
Range Concentration, % KaJO Line pairs Pre-exposure time, sec. Exposure time, see.
Low
Medium
High
0.001 to 0 . 0 3 Na 5890-95
0 . 0 3 to 1 . 0 N a 5890-95 c u 5218 5 10
1 . 0 to 3 . 0 N a 3302-3 Cu 2961
Cu 5700
5 20
0
30
per was abandoned as a standard. The copper electrodes were cut from 0.6-cm. (0.25-inch) pure copper bar stock. After use, these electrodes were cleaned 15 ith emery cloth, dipped in concentrated nitric atid, and washed with distilled water. The use of copper electrodes had several advantages. The sodium contamination, which originated in the plant dust that seeped into the laboratory was greatly reduced. The sensitivity of the analysis was doubled, owing to the discarding of the practice of diluting the sample 1 to 1 with the pondered copper. The elimination of time-consuming procedures was an added advantage. The only apparent objection to the use of copper electrodes as a source of reference lines is the possibility of noncorrelation between the intensities of the copper lines and the sodium lines. The probability of this noncorrelation also exists to a lesser extent when powdered copper is used as an internal standard. To overcome this objectionable feature each sample was analyzed three or four times and the final result was obtained as the average. ANALYTICAL PROCEDURE
In setting up spectrographic calibration curves for sodium the lack of correlation between chemical and spectrographic results provided a problem, which was solved by assuming spectrographic correlation and by securing a large number of chemical determinations. Approximately twenty to thirty results of chemical analyses for each range of concentration required were plotted against the corresponding averaged spectrographic line intensity ratios, using log log graph paper. Linear curves were drawn, averaging the points for each range. The average correlation was found to be positive: this indicated the validity of the method. The slopes of the calibration curves increased in going from a lower to a higher concentration. This increase seems in agreement with theory and previous observation. Some spectroscopists are of the opinion that the slopes of calibration curves should approximate 45 ’, which is the optimum sensitivity compromise. This is a desirable condition, but it has been the author’s experience that a 45 slope is obtainable only in a limited range of concentration. The spectrographic calibration curves for sodium were set up in three ranges, low, medium, and high, as given in Table 11. The calibration curves obtained are linear. They are based on plant samples secured a t various stages of the purification process in which sodium is removed. The low range was specifically designed to cover the finished production of the catalyst, while the medium and high ranges covered the intermediate plant process. In order to check the calibration curves, an independent outside source of analyses for sodium was obtained (Table 111). Samples of catalyst were secured that were analyzed in the laboratory of the M, IT-. Kellogg Co. by trained chemists highly experienced in the determination of the sodium content of silica alumina catalysts. The average deviation in the range of specification (0.02% K a 2 0 or less) is 0.0027, sodium oxide. Improvements in the technique discussed in the summary in all probability would improve the accuracy. However, the spectrographic results, as is, are satisfactory for routine control, particularly as analyses by the chemical laboratory on the Bellogg samples produced errors that were uniformly greater than those produced in spectrographic analysis.
The good agreement noted between the Kellogg chemical analyses and the spectrographic analyses indicates that the technique of averaging large numbers of chemical determinations in order to obtain calibration curves is valid, that the method of chemical analysis contains no systematic errors, and that the large deviations found are the result of inexperience and the difficulty of the method. Undoubtedly in the hands of an experienced chemist the zinc uranyl acetate method produces very accurate results. However, the spectrographic technique offers a far more rapid and fairly accurate determination of sodium content and does not require a high degree of training or long practice. SUMMARY AIID DISCUSSION
An easy and rapid method foi the spectrographic determination of sodium involves the use of a copper counter electrode as a standard to furnish reference lines. The material to be analyzed is placed in the crater of a graphite electrode and volatilized in an alternating current arc. This procedure is repeated three or four times and the‘analytical results obtained from the processed plate are averaged. The accuracy is adequate for industrial control and superior to the average chemical methods in the low percentage range. Analyses are obtained in about 30 to 50 minutes, much quicker than by chemical means. Several modifications in the technique may improve the accuracy of the method. The use of a graphite electrode with a center post undoubtedly would localize the arc, provide steadier volatilization of the sample, and lead to greater reproducibility. It may be possible to correct the experimental data by providing a curve based on a carbon-to-copper line pair. The magnitude of the ratio of this line pair would indicate relative volatilization of the sample and graphite electrode and a positive or negative correction would be applied to the percentage of sodium oxide, depending on whether the carbon-to-copper ratio was greater or lest than the average for the specific percentage of sodium oxide. However, this correction has been considered only from a theoretical standpoint and remains to be checked by esperiment.
Table 111. Analyses of Kellogg Samples Chemical Analysis, % 0,005
Sample 1
Spectrographic Analysis, % 0,009 0,005
0.005
0.005 0.007 0.006 0,018
2
O
3
0.04
4
0.33
0,008 0.021 0.026 0.019 0.017 0.016 0.018 0.03 0 04 0.31
LITERATURE CITED
(1) Barber, H. H., and Kolthoff, I. M., J . Am. Chem. SOC.,50, 1625 (1928). \----I
(2) Barnes, R. B., Richardson, D., Berry, J. W., and Hood, R. L., IND. ENG.CHEM.,ANAL.ED.,17, 606 (1945). (3) Brode, W. R., “Chemical Spectroscopy,” p. 555, New York, John Wile)- & Sons, 1943. (4) Gerlach, W.,and Schweitser, E., “Foundations and Methods of Chemical Analvsis by the Emission Spectrum,” Vol. 1, Leipsig, L. Voss, London, Adam Hilger, 1929.(5) Harrison, G. R., “Wave Length Tables,” New York, John Wiley &Sons, 1939. (6) Lindsay, F. K., Braithwaite, D. G., and Diamico, J. S., IND. E s c . CHEM.,- 4 w . k ~ .ED., 18, 101 (1946). (7) Scott, W. W., “Standard Methods of Chemical A4nalysis,”ed. by N. H. Furman, 5th ed., pp. 878-9, S e w York, D. Van Nostrand Co., 1937. RECEITED February 5 , 1947
