Determination of Arsenic in Hydrocarbon Reforming Cata Iysts by N eutr on Act ivat ion G. F. SHIPMAN and 0.1. MILNER Research and Development luboratory, Socony Mobil Oil
b Activation analysis has been applied to the determination of arsenic in platinum-alumina reforming catalysts a t concentration levels as low as 1 p.p.m. or less. The sample is irradiated in a reactor having a flux of the order of 10l2neutrons per square cm. per second. The catalyst is fused with alkali and the arsenic is separated b y distillation as the pentabromide along with carrier arsenic, and is isolated as the metal. The over-all error of the method is equivalent to 10% of the amount present. Interferences are negligible. The method has been used to follow the arsenic lay-down in commercial reforming units.
T
Co., Inc.,
Paulsboro, N. J.
this reactor is about 3.4 X 1OI2 neutrons per square em. per second. TRACER STUDIES
Separation of Arsenic. Because i t n a s impossible t o ascribe the observed activity to arsenic, in the presence of the other radioisotopes formed during the irradiation, it was necessary to separate the radioarsenic chemically by using weighable amounts of inactive arsenic as a carrier. The procedure chosen was similar to that of Bartlett ( I ) in which the arsenic is distilled as the pentabromide from a sulfuric acid solution. The arsenic was then precipitated as elementary arsenic by ammonium hypophosphite ( I O ) . The separation procedure was examined using microcurie amounts of arsenic-76 (in the form of sodium arsenate) as a tracer. Recoveries of arsenic were quantitative (Table I). The samples to which carrier arsenic mas added show greater activities than the one where the carrier was absent. Although this difference is within the limits of experimental error, it is believed that it actually reflects back-scatter from the inactive arsenic present. Self-absorption. I n measuring the beta activity of a radioisotope, consideration must be given to self-absorption effects-that is, the tendency of the upper layers of the mount to absorb the radiation from the lower layers (9, 6). This effect is large for low-energg betas, but becomes less important with increasing energy. The degree of self-absorption may be determined by activity measurements on a series of samples of increasing activity, under conditions of constant specific activity (the ratio of active arsenic to carrier arsenic is constant) If self-absorption is negligible, a plot of the activity versus the weight of the sample per unit area will yield a straight
line; if self-absorption is significant, an exponential saturation-type curve will be obtained. The self-absorption curve obtained for arsenic-76 is shown in Figure 1. The dashed line represents a typical self-absorption curve. The linear experimental curve indicates that selfabsorption is negligible. This substantiates the inference drawn from Table I-i.e., a higher count is found in the presence of carrier than in its absence. Repeatability of Mounting Technique. The repeatability of the counting mount was determined by precipitating identical aliquots of the active solution together with 25 mg. of carrier, mounting, and counting. The results were 10,130, 9,928, and 10,570 counts per minute, corresponding to a precision to 3.3%. This is poorer than the statistical counting error for 10,000 counts (about I%), but it also includes handling errors.
petroleum industry is vitally concerned with the presence of traces of elements such as nickel, vanadium, and arsenic in crude and distillate oils, because these elements poison catalysts. As arsenic is believed to be particularly harmful to platinumS E L F - A B S O R P T I O N FOR As" reforming catalysts, it has become ( S P E C I F I C A C T I V I T Y CONSTANT) 9.0, necessary t o determine the arsenic conI I I tent of used catalysts. It was believed that the amount of arsenic deposited on a catalyst that would cause a significant degree of poisoning was in the parts-per-million range. Therefore, a method was sought which would have a high sensitivity. , Neutron activation analysis appeared attractive because it has a high sensitivity for many elements and it is specific. Also, in contrast to standard methods of analysis, it is almost completely free of blank considerations. Complete discussions on the application of activation analysis for deterov I mining trace elements have been given 0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 by Boyd (S), Smales (8-II), Jenkins and DENSITY THICKNESS, MG. PER CM.' Smales ( 5 ) , and Leddicotte and Reynolds (7). Upon thermal neutron irFigure 1. Self-absorption for arsenicradiation, many elements undergo il 76 (specific activity constant) n,y reaction giving rise to radioactive isotopes. The radioisotopes produced are identified by their radiations ( 8 , ~ ) Table I. Recoveries of Carrier and Active Arsenic after Distillation and their half lives. The amount of and Precipitation the element is found by coniparing its Carrier, Mg. Activity, activity with that produced by a known -4dded Recovered C.P.M. amount of the same element. In the 25.4 25.0 Distillation f pptn. 5384 work reported here, the nuclear reactor 24.7 25.0 Pptn. only 5460 a t the Brookhaven National LaboraWeightless mount (no ... ... distillation or pptn.) 5129 tory constituted the source of thermal neutrons; the thermal neutron flux of HE
210
ANALYTICAL CHEMISTRY
placed between the sample and the counter. Although this latter procedure reduced the measured activity of arsenic by about 20%, it presented no particular disadvantage with the amounts of activity available. Also the gold interference could probably have been minimized by adding a "hold-back carrier" in the distillation.
INTERFERENCES
I n activation analysis there are two possible sources of interferences: elements that also become activated and follow arsenic in the separation, and elements that upon irradiation undergo nuclear reactions to give rise to radioisotopes of arsenic. The presence of elements that may cause interference is detected by decay and energy measurements. Fortunately, arsenic-76 has a very energetic beta particle (E,,, = 3.15 m.e.v.), and few elements will yield equally energetic decay particles. In addition , arsenic can be readily separated from most of the elements present by distillation as the pentabroniide. Of the elements that might accompany arsenic in the separation (germanium. molybdenum, rhenium, selenium, tellurium, antimony, tin , mercury, and gold) none have half lives and energies similar to arsenic. Thus, the activity of arsenic can be found by analysis of decay curves and determination of energy of the beta particles. As discussed by Smales (IO), isotopPs of arsenic may be produced from other elements by (n, y), (n, p ) , and (n, a reactions. Consideration of sample composition indicates that the production of spurious arsenic by any of the above reactions is negligible in this case. When applied to catalyst samples the separation procedure did nor yield radio-pure arsenic (Figure 2 , curve A ) . Therefore, it was necessary to follow the decay until only the longerlived contaminant was present; two weeks m-as usually required. The activities of the individual components were then found by a peeling-off procedure (6). Curve B is the decay curve of the contaminant extrapolated to zero time, and curve C is the reconstructed decay curve of arsenic-76. The longer-lived radio contamination was identified by its half lift, the maximum energy of its associated beta particles, and consideration of the sample composition and nuclear reactions involved. The half life of the radio contamination was found to be 75 hours (Figure 2, curve B), and the energy of its beta particle determined from an aluminum absorption curve n a s 0.3 m.e.v. (Figure 3). The catalyst samples are mainly alumina with a small amount of platinum. Upon activation, aluminum forms a very short-lived radioisotope (half life = 2.3 minutes) and so does not interfere. Platinum, on the other hand, produces four fairly long-lived radioisotopes. Three of these, platinum-191, platinum-193, and platinuni197 decay to stable isotopes of iridium and gold. Platinum-199, however, decays by beta emission with a 29-niinUte half life to gold-I99 which (also un-
C
E3
-
DECAY C J R V E
Au-199
C - RECONSTRUCTED D E C A Y C U R V E OF A 5 - 7 6
10,000 W
3
E APPARATUS
K
a
The apparatus used for activation analysis is the usual equipment for handling radioisotopes, including survey meters, film badges, remote handling tongs, and counting equipment (6, 12). In this work, a Tracerlab "1000" scaler and a Geiger-Muller counter with a Victoreen end-n-indow were used to count the samples.
v1
g
1,000
3 U ).
k
z U
PROCEDURE
IO
50
0
100
150
200
250
300
HOURS
Figure 2. Decay curve of mixture of radioisotopes
'A
+ 3
U
> BETA P A R T C L E
I I \
1
10 0
I
400 600 800 1000 1200 DENSITY T H I C K N E S S , MG. PER CM!
200
Figure 3.
Aluminum absorption curve
stable) decays with a 79-hour half life by beta emission (Emsx= 0.38 m.e.v.) to stable mercury-199. From the agreement in half life and energy of the beta particle, and the fact that gold mill volatilize %hen heated to fumes of sulfuric acid ( 4 ) , the radio contamination was identified as gold-199. To eliminate the need for resolving decay curves, the gold interference was avoided in two ways. The distillation was stopped before the point a t which sulfuric acid fumes (minimizing the amount of gold carried over), and an aluminum absorber of sufficient thickness to filter out the rreaker beta particles completely from gold-199 was
Activation. Prepare a standard sample by impregnating 1 gram of fresh reforming catalyst with 50 y of arsenic. Simultaneously irradiate for 72 hours in a nuclear reactor the used catalysts (unknowns), the standard, and the base from which the standard was prepared. It is necessary to irradiate the base material only once to establish the arsenic content. Sample Dissolution. Transfer the samples from the irradiation containers to' previously weighed nickel crucibles and reweigh. For each gram of sample, add 10 grams of sodium peroxide, mix intimately, and sprinkle a layer of sodium peroxide on top of the mixture. (These operations are best performed in a shielded glove box.) Place in a muffle furnace maintained a t 500" =k 25" C. After 1 hour, remove, cool, and leach in about 30 ml. of water. .4dd 1 t o 1 sulfuric acid cautiously until the precipitate of aluminum hydroxide redissolves and then add a 5-ml. excess. Transfer to a 50-ml. volumetric flask. cool, and dilute to volume. Assemble the distillation apparatus as shown in Figure 4, using concentrated sulfuric acid to lubricate the groundglass surfaces. Place 20 ml. of concentrated sulfuric acid in the distilling flask. Add to the flask, through the dropping funnel, 2.00 ml. of carrier arsenic solution (5.00 mg. arsenic per ml.) and an aliquot of the samples estimated to yield an activity of 10,000 c.p.m. Rinse with two 2-ml. washes. Finally, add 2 ml. of 4770 hydrobromic acid. Turn on the nitrogen gas and adjust to a rate of 1 bubble per second. Heat the solution gradually until the thermometer reads about 120' C. Then heat strongly. When the thermometer reading begins to fall, stop the distillation. Adjust the distillate so that it is 7 or 8 N with respect t o hydrochloric acid and add 2 to 3 grams of ammonium hypophosphite. Digest on a steam bath for 45 minutes. Filter the precipitated arsenic through a previously weighed filter disk ('/*-inch Whatman No. 42) by means of the apparatus shown in Figure 5. Wash the precipiVOL. 30, NO. 2, FEBRUARY 1958
21 1
tate thoroughly with water and finally with alcohol. Dry the precipitate by drawing air through it for 5 minutes and then heating it for 15 minutes in an oven maintained at 105’ C. Cool, and weigh to determine the carrier recovery. Place the precipitate on a counting card. Count the sample, the base, and the standard through a 66-mg.-per-sq.cm. aluminum absorber and under identical conditions. Correct the count rate in each case for background count, for any decay of significance between counting of the sample and standard, for carrier recovery, and, where count rates differ appreciably, for resolvingtime losses. Normalize to equal sample weights. Correct the count rate of the standard for any arsenic in the base material. Compute the arsenic in the unknown from the ratio of its count rate to that of the standard.
NITROGEN GAS INLET
RESULTS
Several standard samples, prepared by impregnating fresh catalyst with arsenic gave the results shown in Table 11. Table
11.
Analysis of Samples
Standard
DeviaArsenic, P.P.M. tion, Sample Added Found 9% 1 0 0.5 2 20.0 2O.ga 4.5 3 28.9 30.4a 5.2 Corrected for arsenic present in base (Sample 1). Table 111. Arsenic Analysis of API Cooperative Samples Added, Found, Sample P.P.M. P.P.M. 1 2 2.5, 2.3 2 10 11.1, 10.7
Figure 4.
neutron flux, irradiation interval, and the time interval between removal from the pile and counting. The coefficient of variation of K is 9.7%. This represents the error that might be expected under the worst conditions-Le., where a standard is not simultaneously irradiated but an activity is assumed on the basis of previous standards. Table IV also gives an indication as to the ultimate sensitivity of the method. Activities tenfold greater than background can be readily determined. Assuming a background of 20 c.p.ni., the minimum practical count rate 11-ould be 200 c.p.m. Therefore, the smallest amount of arsenic that can be determined is 200 c.p.m. divided by 6 X 104 c.p.m. per y , or 0.003 y . APPLICATION
Table IV. Standard
Sample 1 2 3 4 5
Comparison of Standards Date Irradiated 2-7-56 5-21-56 12-16-56 1- 17-5 7 1-23-57
Kl
C.P.Prl./r As 6.2 x 104 6.2 x 104 7.2 X 104 5.6 x 104 5.0 x 104
The data given in Table I11 are duplicate tests made on two American Petroleum Institute cooperative samples, prepared to contain 2 and 10 p.p.m. of arsenic; the samples consisted of pure alumina impregnated with platinum and arsenic. The variation is of the order of 5% of the amount present while the error is about 10% of the amount present. Further indication of the reliability of the method was obtained by comparing the activities (counts per minute per microgram of arsenic) of standards irradiated on different dates, Table IV. The data have been corrected for such determinate factors as differences in 212
ANALYTICAL CHEMISTRY
Distillation apparatus
OF METHOD
The method has been used for folloming the arsenic lay-don-n in commercial catalytic reforming units. Initially, the arsenic concentration is less than 1 p.p.m. Depending on the charge stock, age of the catalyst, and position in the catalyst bed, samples may contain as high as 0.10% of arsenic. When high concentrations of arsenic are encountered, the activity may be allowed to decay before it is counted. The activity is then extrapolated back to the time a t which the standard was counted. However, it is preferable to repeat the distillation, reduction, etc., on a smaller aliquot of the original sample solution. It is most convenient to process samples in groups of three or four. The working time, after completion of the irradiation, is about I5 hours for a batch of four samples. LITERATURE CITED
(1) Bartlett, J. C.,Wood, M., Chapman, R. A., ANAL. CHEM. 24, 1821 (1952).
- - 1 9 M M .
c
I
GLASS HOOKS
NO 42 WHATMAN FILTER PAPER DISC, 7 / 8 * D I A
-
pw
SINTERED GLASS DISC
+
TO VACUUM
500 ML. SUCTION
i Figure 5. ratus
Filtration appa-
Borkowski, G. J., Ibid., 21, 348 (1949). Boyd, G.E.,Ibid., 335 (1949). Hillebrand, G. W.,Lundell, G. F., Bright, H. A., Hoffman, J. I., “Applied Inorganic Analysis,” 2nd ed., p. 343, Wiley, New York, 1953. Jenkins, E. N., Smales, A. A.,
Quart. Revs. (London) , 10., 83 (1956). (6) Lapp, R. E.,Andrews, H. L., “Nu-
clear Radiation Physics,” 2nd ed., pp. 445-71, Prentice-Hall, New York, 1964. ( 7 ) Leddicotte, G. W.,Reynolds, S. A., Xucleonics 8,62(1951). (8) Smales, A. il.,Ann. Repts. on Progr.
Chem. (Chem. SOC. London) 46, 285 (1949). Smalei, A. ’ A,, Atomics (London) 4,55 (1953). Smales, A. A,, Pate, B. D., BNAL. CHEM.24, 717 (1952). Smales, A. .4., Salmon, L., Analyst 80, 37 (1955). U. S. Dept. Commerce, Handbook No. 42, “Safe Handling of Radioactive Isotopes.” RECEIVEDfor review June 20, 1957. ilccepted October 22, 1957. First Delaware Regional Meeting, ACS, Philadelphia. Pa., February 16, 1956.
