X-Ray Spectrographic Method for the Determination of Vanadium and Nickel in Residual Fuels and Charging Stocks ELWIN N. DAVIS and BARBARA CROSS HOECK Sinclair Research Laboratories, Inc., Harvey, 111. An x-ray spectrographic method is described for the determination of nickel and vanadium in trace quantities. The method is fast, requires in most cases no more than 20 grams of sample, and is comparable with other methods in accuracy. A sample to be analyzed is first ashed and the ash put in solution. A portion of this solution is evaporated on a sample plate and rotated under the x-ray beam. The intensity of the fluorescent radiation from the sample, which is a function of the metal concentration, is determined. Procedures are described for handling interference problems which arise from other elements.
T
HE occurrence of trace quantities of vanadium and nickel in petroleum is of considerable significance to the refiner, Fluid cracking catalysts are easily poisoned by these elements, with a resultant loss in catalj st activity and poor product yields. The refiner, therefore, must know the concentration of nickel and vanadium in stocks being charged to fluid cracking units. The user of residual fuels finds that vanadium causes corrosion of turbine blades and decomposition of fire brick. As a result of these eflects, much effort is expended in analyzing charging stocks and residual fuels for nickel, vanadium, and other elements. A number of approaches to this analytical problem have been described in the literature. Gamble and Jones ( 6 ) , Key and Hoggan ( 9 ) , and Dyroff, Hansen, and Hodgkins (3) describe emission spectrographic procedures for a number of elements in various petroleum fractions. Wrightson (12), Dyroff, Hansen, and Hodgkins (S), and Glass and others (7) describe colorimetric methods for vanadium, nickel, copper, and iron in crudes and other petroleum oils. d polarographic method (unpublished) has been used in these laboratories for a number of years. These methods are all reported by their authors to yield satisfactory results. The purpose of this Lvork mas to investigate still another approach to the quantitative determination of nickel and vanadium by use of the x-ray spectrograph with the hope of reducing both the time and quantity of sample necessary for an analysis. X-ray Rpectrography, although a relatively new analytical technique, has been described rather thoroughly in the literature (1, 6 ) . Dvroff and Skiba (4)have applied the x-ray spectrograph to the determination of trace amounts of nickel and vanadium on fluid cracking catalysts. The method finally developed requires ashing of the oil sample, with the aid of sulfuric acid, and subsequent solution of the ash. An aliquot of the ash solution is evaporated on a glass disk and analyzed with the x-ray spectrograph for vanadium, nickel, and iron. If the iron concentration is more than 10 times that of the nickel, another aliquot is extracted to remove interfering iron and then evaporated on a second glass disk and analyzed for nickel. A sample rotator improves the precision and accuracy of the determination. The method is judged to be equal in speed and accuracy to emission spectrographic methods and somewhat faster than chemical methods. The elapsed time for one determination is about 5 hours, of which 1 hour is analyst working time. APPARATUS
The apparatus used in this work was a Norelco (North American Philips Co.) x-ray spectrograph. Several modifications of
the instrument used in this laboratory have been made to facilitate the analysis of the metallic components of lubricating oils and gasolines. These have been described in dn earlier paper ( 2 ) and have been found to be generally helpful in this procedure. Models of the Norelco x-ray spectrograph supplied before about March 1953 were equipped with a collimating device made by packing a bundle of nickel tubes in an aluminum holder. This has since been replaced with a collimator made of flat nickel foil plates 4 inches long, spaced 0.010 inch apart. This collimator is necessary to obtain sufficient x-ray intensities for this type of work. The half width of emission lines with this system is about 0.4". This is sufficient resolution to separate the nickel K alpha line from the tungsten L1, which is the closest line combination that will be encountered in this work, The wave lengths of the K x-ray emission lines of vanadium are 2.502 A. for the alpha and 2.280 A. for the beta. X-rays of this order are absorbed heavily in the air path between the sample and the Geiger tube. To circumvent this difficulty, a helium path arrangement has been used. The constructional details of such an assembly are described by Davis and Van Nordstrand ( 2 ) . A commercial helium path has recently been made available by the instrument manufacturer.
Table I. Wave Lengths and Spectrometer Angles for Vanadium, Nickel, and Iron Lines Using Lithium Fluoride Crystal E 1em ent
Vanadium Ka Pr'ickel Ka Iron Ka
x,
A. 2.502
1,659 1.937
Spectrometer Angle Using LiF Crystal, Drgrees 76.92 48.65 57.49
One of the limitations to trace element work with the spectrograph is the x-ray tube. Most tubes currently available emit a spectrum which, in addition to the lines of the target material, shows appreciable amounts of other elements, principally nickel, copper, and iron. If this background intensity is too high, the measurement of the radiation from a very small amount of the element in the sample becomes rather difficult. There is no firm criterion as to when a tube is suitable for use in this type of work, but Friedman and Birks (6) state that i t is inadvisable to attempt to detect a line intensity of less than one tenth the background. Tf one is willing to spend longer periods of time counting, this ratio might be somewhat lower. X-ray tubes vary considerably in the amounts of contamination present and i t is possible to select or have selected a tube with a low level of contamination. The age of a tube is also a factor, with the general level of contamination increasing with tube life. One x-raj tube manufacturer has recently announced the availability of an x-ray tube of high spectral purity. One of these, an FA60 tungsten target tube, was made available to the authors for this work. A molybdenum tube should also be satisfactory. Crystals for x-ray spectrography are available in several different materials, Lithium fluoride gives slightly better intensities than any other currently available, and hence was used in this work. EXPERIMENTAL
I n considering the determination of nickel and vanadium by means of the x-ray Spectrograph, it seemed possible that sufficient intensities might be obtained from these elements to permit
1880
1881
V O L U M E 27, NO. 1 2 , D E C E M B E R 1 9 5 5 direct determination on the oil sample. The nickel K alpha line (Table I ) falls in the most sensitive region for x-ray spectrography. The vanadium line is somewhat less sensitive, because of its longer wave length, but it was thought that its intensity might be raised to a sufficient level by use of the helium path. The examination of charging stocks and residual fuels by this procedure failed to yield sensitivities of a high enough order to be useful in a quantitative determination. The possibility of making a direct determination has not been abandoned, as an improvement in x-ray tube characteristics, and the use of a thin-windowed Geiger tube, may still make such a determination possible. COLLIMATOR
4
i 0
I
13
1
1
‘.-+
--’
817
450
0
0
0
0
0
0
0
490
I
1
I
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Figure 1. Plot of intensity distribution over sample area
As the direct determination seemed impractical with the present equipment, work on a concentrated sample was next considered. The most convenient method of concentrating the nonhydrocarbon portion of the sample is through ashing. The ashing of petroleum oils is generally accomplished by one of the following three methods. Dry Ashing. The oil is ignited and burns freely until only a carbonaceous residue remains. Further ignition is done either over a burner or in a muffle furnace until only the ash remains. The possibility of loss of metallic components is greatest with this method. Wet Ashing. The oil is coked with sulfuric acid and an oxidising agent, such as nitric acid, is used to oxidize the organic material. When oxidation is complete, the acid is evaporated, leaving the ash residue. This method is time-consuming, but no loss of metallic constituents should occur. Coking-Ignition. The sample is heated with sulfuric acid to form a coke, excess acid is evaporated, and the carbon residue is oxidized by ignition. This method should also retain all of the ash and is faster than the wet oxidation procedure. The details and relative merits of these procedures are reported in the literature by various investigators (6-9, 1 2 ) . The cokingignition procedure has been shown by Gamble and Jones (6) to yield satisfactory results. Such a procedure has been used in these iaboratories for a number of years and has also been applied to this work.
Quantitative transfer of the ash from the ashing vessel to the x-ray sample holder is virtually impossible; hence, Bolution of the ash must be effected with subsequent evaporation on the sample holder. This, in early work, was nothing more than a flat aluminum plate. The evaporation of aqueous ~ t a n d a r d son a flat
plate showed that reproducibility was difficult to achieve. The distribution of the sample on the holder seemed to be overly critical, yet the holder was smaller than and completely contained in, the area supposedly being viewed by the collimating system. The assumption had been made that the x-ray photons detected by the Geiger tube were coming from an area illuminated by the x-ray tube and bounded by the dimensions of the collimating system. In view of this nonreproducibility, an investigation was made of the effective radiating area of the sample. To do this an aluminum plate, made to fit in a sample tray, was ruled in 0.25-inch squares. A count was then taken on this plate a t the wave length of the nickel K alpha line. A film of nickel was then evaporated on a piece of aluminum foil and from this a 0.25-inch square was cut. The foil square was moved to cover successively each square of the plate and a count was taken in each position. These data give an intensity pattern as shown in Figure 1. From this it can be seen that the radiation being detected by the Geiger tube does not come uniformly from the area being viewed by the collimator. Thus, it is evident that to work with small quantities of a solid sample and obtain reproducible results, a good distribution of the sample is necessary. Furthermore, to obtain the best intensity, the sample must be contained in an area not more than 0.5 inch square. A uniform distribution of the ash on the sample plate was not easily achieved. Evaporation on an untreated surface left the residue in a series of concentration rings which varied in size and spacing with the speed of evaporation and the salt concentration. -4 number of procedures and conditions were tried to improve this, such as evaporation under controlled conditions, the addition of gelatin, poly(viny1 alcohol), wetting agents, and nonwetting agents, and spraying of the sample onto a hot sample plate. Spraying the sample onto a hot disk was unsatisfactory because of loss of sample in the spray mist. Gelatin and poly(viny1 alcohol) gave good dispersions nhen used with the aqueous standard solutions with no acid present. The acid necessarily present in the ash solution caused a charring of the gelatin or alcohol and made evaporation to complete dryness difficult. The use of water-repellent coating (Desicote) eliminated the formation of concentration rinps and gave a Satisfactory distribution. The aluminum plates used in the earlier work were satisfactory when used with the aqueous solutions, but were easily attacked by the acid present in the ash solutions, yielding salts which interfered with the determinations. Platinum, although satisfactory in this respect, gave erratic background counts due to its crystalline structure. Annealing of the platinum improved, but did not eliminate, these diffraction effects. As glass gives a uniform background and is unaffected by the acid, it proved to be the most satisfactory material for the sample holder. The final work was done using a 15-mm. microscope cover glass which had been treated with Desicote. To minimize further the effect of the lack of uniformity in the sample radiating area, it was found desirable to rotate the sample during the counting period. B device was constructed (Figure 2) to hold and rotate the sample disk at about 30 r.p.m. The post supporting the sample vas located so that the sample disk rotates in the area shown by the dotted lines in Figure 1. Table I1 shows the result of sample rotation. The data shown under “Without
Table 11.
Effect of Sample Rotation
Without Rotation during Counting, C.P.S.
With Rotation during Counting, C.P.S.
137.3 143.1 140.4 135.8
124.8 124.7 125.0 124.8
ANALYTICAL CHEMISTRY
1882 Rotation" vere obtained by rotating the disk 0.25 turn following each count. The "With Rotation'' data were obtained by succe8sive counts on the same disk 8 8 it revolved on the rotator a t 30 r.p.m. The data. represent a fixed 12,800 count on the vanadium line.
plotted against quantities of nickel on cover glasses ranging from 0 to 15 y. Each curve of the family represented a. different level
b t o 50 7.-
Each 'ourve of the family represented a dit&& of nickel content from 0 to 10 y.
level
The various metals were deposited ou the covw glasses by evaporation of 0.5 ml. of aqueous standard containing the proper amount of each metal. Typical calibration cuivea me shown in Figures 3, 4, zud 5. Background intensities from sample to sample were found reproducible within the statisticd emor. The determination of background on each sample was, therefore, abandoned and all of the counts per second values shown include background.
Figure 2.
Devise for holding and rotating sample
The fluorescent or secondary radiation emitted from the sought elements in the sample is subject to absorption effects from other elements which may be present. An examination of a table of mass absorption coefficients (if) shows that, of those elements normally found .in chargiug stocks or residual fuels, iron is the only element which should give any appreciable interference due to ahaorption. This was found t o be the case and in some instances the iron was present in a high enough concentration t o absorb all of the nickel radiation from the sample. A generally useful method for handling interference problems such as this is through the construction of a family of calibration curves (9)varying the iron concentration through a range of nickel concentrations. This procedure is used when the iron concentration is not more than 10 times that of the nickel. Such a procedure is not practical when the iron concentration is much greater than this ratio because of the extreme reduction in nickel radiation by the iron. When the iron concentration is high, the most convenient method of handling the problem is by the removal of the iron from the nickel and vanadium. Any reagent used in the separation must not remain in the final mnple, a9 any salt deposit on the disk will interfere with the determination. A number of methods were evaluated, but none completely met the above requirements. The most satisfactory was the thiocyanate method (fO),which gave a good separation of the iron from the nickel, but also removed part of the vanadium. The excess thiocyanate was easily destroyed with acid. When it was necessary t o use the separation procedure, it was found best to determine the vanadium directly on part of the sample ash solution and to remove the iron from another aliquot before determining the nickel
MICROGFI*LMS
c€ MtCUEC
Figure 3. Nickel calibration
CALIBRATION CURVES
Calibration curves were constructed for the determination of vanadium, nickel, and iron. I n order to take into account the reduction of nickel radiation by the iron present and the enhaneement of the iron radiation by the nickel, it was necessary to construct families of cnrves for these two elements. The following curves or groups of cnrves were made.
A single curve for vanadium, in which counts per second were plotted against quantities of vanadinm on cover glasses ranging from 0 to 15 y. A family of curves for nickel, in which counts per second were
MICROGRAMS O F VANADIUM
Figure 4.
V a n a d i u m calibration
V O L U M E 27, NO. 12, D E C E M B E R 1 9 5 5
1883 Table IV.
Table 111. X-Ray Operating Conditions Tungsten
Tube Kilovolts Milliamperes Crystal Helium flow Sample rotator speed Total counts Nickel Vanadium Iron
Comparative Results
Vanadium, P.P.M. Sample Nickel, P.P.M. Description Polarograph X-Ray Polarograph X-Ray Synthetic blend, 2 6.10- 6 . 0 5.50- 5.70 4 . 8 -4.9 5.30- 5 . 4 0 p.p.m. V and hi Syntheticblend, 10 11.2 -10.8 1 0 . 6 -10.6 9.50-9.40 1 0 . 1 -10.1 p.p.m. V and Ni Feed stock
50 45
LiF
1 liter/minute 30 r.p.m. 25,600 12.800 25,600
1 2
1.61 0.59 0.43 1.10 0.90 0.90 0.59 0.59
3 4 5 6
PROCEDURE
7
A sample of the appropriate size is weighed into a 250-ml.
Vycor beaker. (Sample size varies with the type of material to be analyzed and the volume to which the solution of the ash is made. The 15-mm. cover glass, when treated with Desicote, will hold about 0.5 ml. of solution. So that only one evaporation will be required, it is desirable to have a minimum metal concentration of 0.5 y each of vanadium and nickel per milliliter of ash solution. A 20-gram sample is adequate for most charging stocks and a 5- or 10-gram sample is generally sufficient for residual fuels.) An equal volume of concentrated sulfuric acid is added and carefully heated on a hot plate until a solid coke is formed. Additional heat is applied to the sample until the excess acid has been evaporated and no further fumes are evolved. The resulting coke is placed in a muffle furnace a t 500' C. until all carbon is removed. After cooling, a drop of 1 to 1 nitric acid and 2 to 3 ml. of water are added and solution of the ash is completed with gentle heating. The solution is transferred to a 5- or IO-ml. volumetric flask and made to volume with distilled water. A 0.5-ml. portion of the solution of the ash is pipetted onto a 15-mm. cover glass which has been treated with Desicote. The cover glass must be level and should be supported above the table by a disk or a washer which is slightly smaller in diameter than the glass. A 250-watt infrared lamp is mounted about 6 inches above the cover glass and the sample is evaporated to dryness. The glass disk is then placed on the sample rotator and a count is taken on the vanadium, nickel, and iron K alpha lines. If in the above determination the indicated iron-nickel ratio is greater than 10, the following procedure should be used to deter-
8
Table V.
1.38 0.82 0.50 1.18 0.93 1.11 0.50 0.72
1.54 0.50 0.11 1.76 0.34 0.34 0.64 0.64
1.38 0.52 0.10 1.52 0.37 0.43 0.80 0.75
Determination of Precision Ni, P. P .vi.
Average
v,
P.P.RI
.
1.30 1.28
mine nickel. A 0.5-ml. portion of the solution of the ash is pipetted into a 20-ml. separatory funnel and diluted to about 5 ml. Then 0.5 ml. of 10% ammonium thiocyanate solution is added and shaken well. The iron thiocyanate complexes are extracted with successive isoamyl alcohol washes until the red color is no longer present in the raffinate. The extract is washed with water containing a few drops of the ammonium thiocyanate solution and the wash is added to the raffinate. The raffinate is transferred to a beaker and evaporated to dryness. A drop of concentrated sulfuric acid and 2 ml. of 1 to 1 nitric acid are added to destroy the remaining ammonium thiocyanate, and the solution is again evaporated to dryness. The residue is redissolved in a drop of 1 to 5 nitric acid and 0.5 ml. of water with the aid of gentle heat. After solution is completed, the sample is quantitatively transferred to a cover glass and the evaporation on the disk is carried out as for the vanadium determination. The disk is placed on the rotator and a count is taken on the nickel K alpha line. X-ray operating conditions are shown in Table 111. Each day a count is taken, on the nickel and vanadium lines, on one of the disks used in the preparation of the calibration curves. If the values so obtained differ from the original, a factor is derived to adjust back to the calibration curves. This factor is applied to all values obtained during the day. CALCULATIONS
The elapsed time required to obtain the fixed counts is converted t o counts per second by: Counts/second =
I 2
0 MICROGRAM
total fixed count elapsed time in seconds
and the quantity of nickel and vanadium on the cover glasses is obtained from the appropriate calibration curve. Total of micrograms of metal in the total sample is obtained by:
OF N I C K E L
I M I C R O G R A M OF NICKEL
Total micrograms of V or Ni = micrograms in aliquot x ml. of ash solution ml. of aliquot
3 2.5 MICROGRAMS OF NICKEL 4
5 M I C R O G R A M S OF NICKEL
5
1 0 M I C R O G U A M 5 @F
Parts per million of metal in the oil sample are obtained by: P.p.m. in sample =
0
10
20
30
MICROGRAMS
Figure 5.
-40 O F (RON
Iron calibration
50
J
60
total micrograms of metal in sample wt. of sample in grams
Table IV s h o m a comparative study made on a number of charging stocks and two synthetic blends between the x-ray method and a polarographic method. In the analysis of feed stocks, it is felt that the agreement is good. To obtain an estimation of the precision of the method, replicate analyses were run on one sample a t intervals covering a week's time (Table V).
ANALYTICAL CHEMISTRY
1884 The indicated precision is & l o % for the vanadium and +2.3% for the nickel. ACKNOWLEDGMENT
The authors express their thanks to Ralph GrifRth and E. I. Bradshaw for their aid and suggestions in the preparation of this paper and to ?IT. J. Schlesser for sample preparations. LITERATURE CITED
(1) Birks, L. S.,Brooks, E. J.. and Friedman, H.. ANAL.CHEM..25, 692-7 (1953). (2) Davis, E. N., and Van Kordstrand, R. d., Ibid., 26, 973-7 (1954) . (2) Dsroff, G. V., Hansen, J., and Hodgkins, C. R., Ibid., 25, 1898905 (1953).
(4) Drroff, G. V., and Skiba. P., Ibid., 26, 1774-8 (1954). ( 5 ) Friedman, H., and Birks, L. S..Rev. Sci. Insfr., 19, 323-30 (1 948). (6) Gamble. L. W., and Jones, I\-. H . , ASAL. C H E h r . . in press. (7) Glass, J. R., Kirchner, J . P.. l l i l n e r , 0. I., and Yurick, -1.S . , Ax.4~.CHEM.,24, 1728-32 (1952). (8) Karchmer, J. H., and Gunn, E. L., Ibid., 24, 1733-41 (1952). (9) Kev. C. W.. and Honean. -- G. D . . Ibid.. 25. 1Gi3-6 (1953). , , (10) Sandell, E. B., “Colorimetric Dktermiiiatkii of Trace Metals,” pp. 363-75, Interscience, S e w York. 1950. (11) Sproull, W.T., “X-Rays in Practice,” pp. 668--71, XfcGraxHill, New York, 1946. (12) Wrightson, F. 31.: ANAL. CHEM., 21, 1543-5 (1949).
RECEIVED for review Yay 26. 1955. Accepted August 22, 1955. Presented before the American Petroleum Institute, 8t. Louis, Mo.. 1955.
Com bined Radiometric and Fluorescent X-Ray Spectrographic Method of Analyzing for Uranium and Thorium WILLIAM J. CAMPBELL and HOWARD F. CARL Eastern Experiment Station,
U. S. Bureau o f Mines,
College Park,
Employing radioactivity measurements and fluorescent x-ray spectroscopy, a rapid method of analyzing ores for uranium and thorium has been developed. The total radioactivity of uranium and thorium is determined and expressed as per cent equivalent uranium. From the line intensity ratio ULalThLcu measured by fluorescent x-ray spectroscopy the weight ratio of uranium to thorium is calculated. As the relative radioactivity of uranium to thorium is Imown, the weight per cent of uranium and thorium can be calculated. The time required is approximately 20 minutes per analysis for both elements. The accuracy is +lo% of the amount present in samples containing more than 0.5% of the elements and the lower limit of detection is 0.01 to 0.03% of either element.
A
Md.
The sample is positioned in a holder at a constant distance from the Geiger tube window. The radioactivity is counted for a fixed time, usually 8 minutes, and the intensity is recorded as counts per minute above background. These values are compared directly with calibration curves prepared from radioactivity standards. A series of calibration curves is prepared for the various distances from the sample to the detector window. By choosing a suitable distance, it is possible to remain within the linear response range of the Geiger tube. Regardless of the uranium-thorium ratio, the radioactivity is expressed as “per cent equivalent uranium,” that amount of uranium in pitchblende in radioactive equilibrium necessary to give an equal activity. DETERMINATIOY OF THORIUM-URANIUM RATIO
After the radioactivity measurements are completed, the samples are placed in the fluorescent x-ray spectrograph to determine the thorium-uranium ratio. These elements have atomic numbers 90 and 92, so that a s a first approximation the line-intensity ratio of ThLa to U L a equals the weight-per cent
S A public service, Ihe Bureau of Mines, College Park, Md.,
examines a large number of mineralogical samples submitted for identification and for determining possible commercial value. This service includes approximately 2000 tests for radioactivity per year. Samples with measurable activity are studied further to determine the approximate concentration of uranium and thorium. For various reasons, principally the time and cost of analyses, chemical methods are not satisfactory. Optical spectroscopic techniques have only limited application. Preliminary investigations indicated that a combined technique for determining total radioactivity, with a subsequent uranium--thorium ratio measurement by fluorescent x-ray spectrography, offered excellent possibilities.
ThL9,
,
XENOTIME
l 1
U.125 WT Yo Th = 2 75 W T Yo
~
NaCI 24.564 A 5Ohv.llm0
1
OUARTZ
26 = 3 6 4 P 50 k v . 4 5 m o
1 ;
,
u Ld,aZ ’
,
I f
‘
, ’
,
8
u LU,
’,
~
,
IIVSTRUMENTATION
Radioactivity is measured with standard commercial Geigercounter equipment, with provisions for accurately positioning powdered samples. The x-ray spectrograph used is a modified Sorelco 90’ spectrometer employing commercially available parts. Any one of the commercial units of the North American Philips Co., the General Electric Co.. X-Rav Deuartment. or the hudied Research Laboratories, should be saiisfactory. ’ * A
RADIOACTIVITY MEASUREMENTS
The samples, ground to -325 mesh, are placed in suitable holders and packed to a constant level with the aid of a spatula.
I
1
’
uhap
4 -4
1’6
18
9
2b -22 1 DEGREES
Figure 1.
k
I
’v,d
A 29 30 31
28
, $$
i*vp,v~,ic.‘. 32
>:
28
Uranium La and thorium La spectra from xenotime
Showing relationship between line intensities and concentrations of respective elements