XI1 measurements were made on a Perkin-Elmer single-beam, recording s p e c t r o p h o t o n i e t e r , Model 12B, equipped with a sodium chloride prism and 0.05-mm. sodium chloride cells. DISCUSSION
The determination of nitrocellulose in mixtures of other resins is difficult and time-consuming. Some methods are also hazardous. The infrared procedure represents a rapid, convenient method, sufficiently accurate for many practical applications. The method is not subject to interferences from any other resins normally compounded with nitrocellulose. This includes any resin which can be precipitated from the solvent solution with 17-ater. Traces of aldehydes which may be occluded in the film do not interfere. The method n-as
evaluated by measuring the nitrocellulose content of carefully prepared synthetic mixtures. Four separate determinations were made on each mixture (Table I). For rapid control work a single determination may be expected to show a maximum deviation of = k l % nitrocellulose. The over-all accuracy of the method is reflected in the averaged, estimated standard deviations of 0.31001, over a range of samples having a nitrocellulose content from 1 to 80%. ACKNOWLEDGMENT
The authors are grateful to the A. B. Dick Co. for permission to publish this work. LITERATURE CITED
(1) Am, Soc. Testing lfaterialq, .ISTlI
Standards, Part 4, “Standard Specifications and Tests for Soluble Xitrocellulose,” pp. 361-9, D 301-50, 1952. (2) Furman, N. H., ed., “Scott’s Standard Methods of Chemical Analysis,” 5th ed., Vol. I, pp. 650-2, Van Nostrand, Ken- York, 1939. (3) Genung, L. B., ANAL.CHEXI. 22, 401 (1950). (4) Newburger, S. H., J . ilssoc. Om. Aqr. Chemists 39, 259-60 (1956). (5) Pitman, J. R., J . Soc. Chem. Znd. (London) 19, 983 (1900). (6) Shacfer, W. E., Becker, IT. K., ANIL. CHEM2 5 , 1226 (1953). ( 7 ) Swann, M. H., Zbid., 29, 1504 (1957). (8) Swann, AI. H., -4dams, RI. L , Esposito G. G., Zbid., 27, 1426 (1955). (9) U. S. Dept. Commerce, “Infrared Spectra of Plastics and Resins,” PB 111338 (1964).
RECEIVED for review Xovember 13, 1058. ilccepted March 23, 1959.
Quantitative Determination of Platinum in Alumina Base Reforming Catalyst by X-Ray Spectroscopy ARNOLD J. LINCOLN Research and Development Division, Engelhard Industries, Inc., Newark, N. I .
ELWIN N. DAVIS Sinclair Research laborafories, Inc., Harvey,
111.
,X-ray spectrophotographic techniques are sufficiently fast and accurate for platinum assay in new refining techniques requiring a catalyst containing platinum metal. Sample preparation is relatively simple, requiring only grinding and calcining to reduce all samples to a uniform condition. Two methods of calibration are proposed depending upon the x-ray tube supplying the exciting radiation. The standard deviation of the method is about 0.0025% a t the 0.6% platinum level. The elapsed time between receipt of a sample and the result is about 3 hours, about 30 minutes of which requires the operator’s attention.
ing catalyst utilizing a fluorescent x-ray spectrograph. With minor modifications, they should be adaptable to the determination of most of the impurities of interest found in this type of catalyst. The catalyst described is usually manufactured in extruded form which can be reduced to a fine powder by grinding. This is desirable for x-ray spectrographic work to give a reproducible sample radiating surface. Excellent studies using x-ray spectrographic techniques in the analysis of powders have been reported. Equipment. This work was conducted in two laboratories. Identical equipment was used, except for t h e x-ray tubes supplying t h e exciting radiation. I n one laboratory a molybdenum target tube was used; in t h e other, a tungsten target. Although both tubes were satisfactory, a n alternative procedure was necessary with the tungsten tube.
I
N BUSINESS transactions involving the
use of platinum or the platinum group metals, the accurate determination of platinum content is important. A relatively small error in an assay can result in a large monetary gain or loss. The x-ray spectrograph suggested a convenient, reliable, and accurate procedure for the quantitative determination of platinum on alumina catalyst. The methods described were designed for a platinum-on-alumina base reform-
North American Philips Co., Inc., x-ray spectrographs Lvere used. Machlett OEG 50 tungsten and molybdenum target tubes n-ere operated a t 60 kv. and 50 ma. Jlilliampere stabilizers maintained the current through the x-ray tube constant to within 0.1%. Lithium fluoride crystals \\-ere used as
the dispersing elements. The collimating system employed open tube primary and parallel plate secondary collimators. A Geiger counter n-as used as the detector. For ease and accuracy in time measurements, automatic electric stop clocks, reliable to 0.1 second, were substituted for the standard timer. The sample holder, specifically designed to fit the standard specimen tray, permitted a backloaded specimen with a smooth flat surface with an area of 1.19 square inches and containing approsimately 4.5 grams of catalyst. Platinum Analytical Line. The platinum spectrum of a typical platinum-on-alumina reforming catalyst in Figure 1 s h o n s t h a t the L , and Lo lines possess intensities well above background and of a suitable magnitude for comparative measurements. With t h e molybdenum tube t h e background region in t h e vicinity of the Lp line is smooth and nearly horizontal, while the L , region is irregular because of the molybdenum lines of the tube and some tungsten contaminant lines. Therefore, the Lo1 line at 32.29’ 29 was chosen as the analytical line with the molybdenum tube. Figure 2 shows a manual count over the Lpl line and the accompanying VOL 31, NO. 8, AUGUST 1959
1317
I
42
40
Figure 1 . x-ray tube
l
I
38 36 34 32 30 ANGLE " 2 8 Platinum spectrum with molybdenum
background measurements of a typical platinum-on-alumina catalyst using a molybdenum tube. Background measurements with the molybdenum tube were made at 30.0' 28 and 35" 28. Figure 3 shows a manual count over the spectrum in the vicinity of the platinum L a and La lines using a tungsten target tube. IYith this tube, the spectrum in this region is crowded with tungsten lines contributed by the x-ray tube and leaves little choice as to which platinum line n-ill be used as the analytical line. The La line gave more intensity and also gave a better slope to the calibration curve and was used as the analytical line with this tube. Platinum Peak Intensity Counting. .4n analysis of thc platinum lines has shonn that with a lithium fluoride analyzing crystal, the sharp platinum peaks do not have plateaus covering a n-ide enough range for simple peak counting trchniques. Maximum precision could be obtained by taking an average of the three counts as the indicated peak. This procedure also serves as a constant check for the peak shift that occurs. These shifts have brei1 traced to temperature changes and slight changes in crystal position due to building vibratioii and to changes that occur in t h r goniometrr setting through constant uie of the machine. Figure 4 Shows a plot of the Lp, peak of a typical platiiinni-alumina catalyst. Preparation of Standards. Five platinum-on-alumina standards were preparrd in concentrations of platinum from 0.550 to 0.65070 in increments of 0.025c7,. These standards
13 18 *
l
ANALYTICAL CHEMISTRY
32 33 34 35 3E ANGLE 0 2 8 Figure 2. Manual count over platinum la, line with molybdenum x-ray tube
30
31
were prepared t h e same as t h e production catalyst, except for t h e platinum concentrations. The extruded catalyst standards were stored in tightly sealed bottles until ready for the final specimen preparation. The platinum content of each standard was accurately determined independently by spectrophotometric techniques (1) in two laboratories. Sample Preparation. Both standards and unknowns were received in pelleted form and had to be reduced to a unifoim condition t o give a reproducible specimen. The pore structure was destroyed and the usual water and carbon components of a spent catalyst were removed by a n airsteam calcination a t 800' C. as follows: Extruded catalyst and standards were ground to a fine powder, passing 170 mesh, and placed in a Coors Porcelain Co. KO. 13 boat. The boat was kept in a furnace, preheated to 800" C. and presaturated with steam, for 2 hours a t 800" C. It n a s then removed to a drying oven a t 120' C. for 1 hour. After removal from the oven the specimen was immediately bottled and sealed until ready for use. Because of the critical geometry of the system, the sample is packed into the specimen holder (Figure 5 ) so as to ensure a flat, smooth, reproducible radiating area. This can be done by filling the specimen holder from the back, firmly packing the sample into the holder with a flat blade or block. When properly packed, the top surface of the sample is smooth, flat, and selfsupporting in the holdrr. Because of
variations in reproducibility which might result from several observers' packing the sample, the difference in intensities introduced was studied. The scatter of the results of several observers was no greater than those of any one observer. Instrumental Standards. I n a n y instrumental method where results are based on a predetermined calibration curve, t h e variation in general instrument sensitivity during t h e course of a d a y and from d a y to d a y must be considered. These variations can be compensated for by using a n internal or a n external standard, or by establishing a new calibration curve n-ith each day's operation. I n this work, the addition of a n internal standard was discarded because of the difficulty in obtaining homogeneouc mixtures. For use nith the molj-bdenum tube, the external standard method \vas satisfactory. This standard was 99CG aluminum-1 yo platinum alloy formed into a block suitable for insertion in the sample tray of the x-ray spectrograph. Intensity ratios were obtained by running the external standard immediately before and after each catalyst standard or unknoivn. Thr data reported in later sections show that the method is self-compensating. The peak intensity fluctuates during the analysis of a single sample run over a period of days with no accompanying change in the intensity ratio. Because the external standard con-
ANGLE
-ze
Figure 4. Manual count over platinum I p , peak with molybdenum x-ray tube
Figure 3. Manual count over pl lines with tungsten x-ray tube sisted of a metal block and exactly the same area was exposed to the x-ray beam each time it was used, no correction was made for background. When a tungsten target tube was used, the intensity ratio with the external standard did not adequately compensate for the daily variations in instrument sensitivity. The multiplicity of tungsten lines contributed by the tube is thought to complicate the spectrum in the region of the platinum lines and cause the slight changes in slope. The most satisfactory method of handling variations in instrument sensitivity was to establish a new calibration curve each day that platinum determinations were made. The instrument stahility after complete warm-up (about 2 hours) held very well, and generally no further calibration checks were needed throughout that day, if the voltage and current controls were not changed. Preparation of External Standard for Intensity Ratio Method. The external standard was prepared as a n alloy of 1% platinum and 99% aluminum because in this form t h e standard could he repeatedly used with no change in composition due to changes in temperature or humidity. Both the platinum sponge and aluminum used t o prepare this standard were chemically pure. To ensure homogeneity and freedom from oxides and occluded gases, the mixture was vacuum melted at a pressure of less than 1 micron. After a second melt under the same conditions, the
b",,dlblj
"l
ULlO
cam"
'"U
I"YU
"UIIIU.y
off in a lathe to remove possible contamination. The resulting rod .was rolled and cross-rolled to a specified thickness, surface planed, and machined to fit in the sample tray in place of the specimen bolder (Figure 5 ) . The finished external standard was given a standard metallographic polish of the face to be exposed t o radiation.
Procedure for Intensity Ratio Method (Molybdenum Tube). The external standard is run immediately before and after each calibrated catalyst standard, T h e average peak intensities of the before and after runs of the external standard, w-ith the peak intensity of the catalyst standard corrected for background, are used to form the intensity ratio. The average intensity ratio obtained from 20 determinations made on each of the five standards over a period of 20 days when plotted against the corresponding platinum content gave a working curve that was linear for the region covered by these standards. NO corrections have been made for coincidence losses because of the limited range of concentration covered by this working curve. Routine samples, prepared as described, are handled in exactly the same manner as the standards, and the intensity ratio is used directly on the working curve to determine the platinum concentration. Periodic checks are made by running the five standards to ensure the validity
Figure 5 Iwert dimensions X 1 inch inride
curve established for this study was used for 24 months, with no observed shift. Procedure for Calibration Method (Tungsten or Molybdenum Tube). The working curve for this method is rstahlished daily from the intensity values obtained from the calibration standards as described and prepared earlier. Counts-per-second values are calculated and plotted against the corresponding platinum conccntration and the resulting cufve is linear. With this method, all components of the instrument must come to a temperature equilibrium before any calibration is attempted. This requires approximately 2 hours a t operating conditions. The sample is prepared as described and careful intensity measurements are made on the platinum peak. Intensity values are calculated to counts per second and the platinum concentration is read directly from the working curve. The background is not determined in this procedure because of the large number of other lines present in this area of the spectrum. The hackground intensity at the platinum wave length remains constant as long as the matrix VOL. 31,
NO. 8, AUGUST 1959
1319
Table I.
Precision of Methods
Method
Average Pt, 70
Detns.
Intensity ratio
0.597
35
No. of
Range Pt, % 0.594 to
c, %
Pt 0,
%
0.35
0 0021
0.39
0.0023
0 602
Calibration
Table
II.
0.597
32
Accuracy of Methods
Pt, Intensity Ratio Method, % PreEent Found 0 593 0.595 0.594 0.597 0.599 0.602 0.594 0.593 0 593 0 595 0 599 0 599 0 601 0 603 0 601 0 599 0 602 0.603 0.602 0,599 0.591 0.586 0,600 0.603 0 599 0.599 0 601 0.602 0 600 0.602 0 596 0.598 0.600 0.602 0,599 0.603 0.599 0.601 0.599 0.601 Std. dev. 0,43% or 0 0026% Pt
Pt, Calibration Method, % Present Found0.592 0.594 0,599 0.598 0,596 0.596 0.602 0.605 0 605 0.605 0 607 0.609 0 603 0.603 0 600 0.605 0.601 0.601 0.598 0.603 0,595 0.600 0.597 0.600 0.597 0.601 0.597 0.599 0.593 0,597 0.594 0.594 0.606 0.602 0.604 0.602 0 594 0.598 0.600 0,603 Std. dev. 0.52% or 0.0031% Pt
retains approximately the same composition as the standards. With this method the current and voltage settings
0.595 to 0.611
must not be disturbed after the calibration has been established. Precision and Accuracy of Methods. The precision of both methods is summarized in Table I. The d a t a shown for t h e intensity ratio method were obtained by making 35 analyses over 15 months from one batch of material selected from a production lot. The d a t a were collectively obtained by three analysts at different times. The d a t a shown for the calibration method were obtained over 1 year on a single sample analyzed 32 times by three different analysts. The data in Table I indicate comparable precision for both methods: the standard deviation being 0.35% or 0.002170 platinum for the intensity ratio method, and 0.39% or 0.0023% platinum for the calibration method. The accuracy of the methods, using 20 production samples, as compared to a spectrophotometric method ( 1 ) is shown in Table 11. The same serira of samples were not used in each case because the work on the two methods was done in separate laboratories. A
standard deviation of 0.43% or 0.0026% platinum with the intensity ratio method and 0.52% or 0.0031% platinum with the calibration method, showed that the accuracy of both methods was comparable. Analysis Time. The preparation of the sample as received requires 15 minutes, which includes all grinding and preparation for t h e steam calcining. The prepared specimen can be run on the x-ray spectrograph and all the necessary calculations made in about 30 minutes for the intensity ratio method. The calibration method requires about 10 to 15 minutes of x-ray and calculating time, plus the time each day for calibration. Because the air-steam calcination requires 2 hours, the total time lapse betxeen receipt of sample and final platinum result is about 2l/, to 3 hours depending on the method. Further Study. The use of both scintillating and proportional counters as detectors in place of the Geiger counter is under investigation and preliminary data indicate t h a t both counters will tend to improve the precision of the method with a n accompanying decrease in analysis time. LITERATURE
CITED
(1) Ayres, G. H., hleyer, A. S.,Jr., ANAL. CHEM.23, 299 (1951).
RECEIVEDfor review June 20, 1958. Accepted April 6, 1959. American Petroleum Institute Meeting, Los rlngeles, Calif., LIay 1958.
Spectrochemical Determination of Vanadium in Alkali Brines W. L. BAGGETT and H. P. HUYCK Pennsalf Chemicals Corp., Calverf City, Ky. ,Vanadium in alkali chloride brines has been determined in concentrations from 2 to 25 y per 1000 grams. Samples are adjusted to a p H value of 5.0, treated with 8-quinolino1, and quinolates extracted with chloroform. The chloroform extract is concentrated by evaporation to a known volume and placed on a graphite electrode previously coated with 20% sodium hydroxide and dried in an oven with a carbon dioxide atmosphere. Excitation is carried out by a 2300-volt alternating current arc with photographic recording of the spectra. Molybdenum is used as the internal standard.
1320
ANALYTICAL CHEMISTRY
M
INUTE quantities of vanadium (10 y per 1000 grams) in alkali
chloride brines cause increased hydrogen evolution in electrolytic mercury cells, producing chlorine and caustic soda. Colorimetric wet chemical methods are subject t o significant errors in low concentrations, primarily as a result of necessary analytical operations and large blank corrections. JIitchell (2) and Mitchell and Scott (3) havedescribed a means of concentrating trace quantities of metals in soil by collecting with organic prrcipitants, ashing the precipitate, and spectrochemically analyzing the ash. Other workers (1, 4, 6) have applied this
technique to different metals and materials using organic solvent extraction as a means of collection. I n all these procedures, however, the final determination is either by colorimetric determination or by spectrochemical analysis of the ashed precipitate. This laboratory adapted this concentration technique to the determination of minute traces of vanadium in alkali brines, using 8-quinolinol as the precipitant at a p H of 5.0. The resulting insoluble quinolates are extracted with chloroform from the brine solution. However, after evaporation of most of the solvent, a measured aliquot of the resultant solution is transferred
