ACKNOWLEDGMENT
Table II.
Recovery of 4-Methylthio-m-cresol from Mixture Containing Sulfur
Material mixed Sulf ur-wettable powder Lime-sulf ur
Phenol added, jig.
50.0 50.0
Absorbance a t 458 mp 0.770 0.775 0.770 0.770
solution with Asozin, Monzet, phenylmercuric acetate-wettable powder or tablet, Dithen, and Bordeaux mixture. The values of these experiments are less than 3% of mean error and the standard deviation is 0.934. The recoveries from the mixed solution with sulfur-wettable powder or lime-sulfur were, however, below 95%. Though another experiment, shown in Table 11, demonstrated t h a t sulfur-wettable or sulfur-lime did not interfere with the color produced by the reaction between 4-aminoantipyrine and 4-methylthio-m-cresol, it is possible that the sulfur might interfere with the hydrolysis of Lebaycid. Up to now, many applied studies of Emerson’s reaction (3, 6, 9, 12) have been reported and these studies have
Phenol recovered, jig.
51.5 51.8 51.5 51.5
Av. recovery, 103.3
Mean error 3.3
103.0
3.0
%
been carried out by using potassium ferricyanide. However the use of this reagent generates a blue pigment with a maximum absorption between 550 and 650 mp. The procedure using the periodate results in a lower blank. Parathion does not interfere with this Apdetermination of Lebaycid. parently the p-nitrophenol does not undergo the oxidative condensation with 4-aminoantipyrine. But this method has a limitation that no Lebaycid can be differentially determined from its 0,O - dimethyl - S - [4 - (methy1thio)m - tolyl]phosphorothioate or its partial decomposition products. So the values obtained by this method should be estimated taking into account these limitations.
The authors thank Hiroo Awoki and Hiroshi Tanaka for their helpful criticisms of the manuscript. LITERATURE CITED
(1)Beilsteins “Handbuch der Organischen Chemie,” Julius Springer Verlsg, Berlin, 13,70(1930). (2) Emerson, E., J . Org. Chem. 8, 417 (1943). (3) Ettinger, M. B., Ruchhoft, C. C., Lishka, R. J., ANAL.CHEM.23, 1783 (1951). (4) Farbenfabriken Bayer, A.G., Leverkusen, Germany, “Analytical Method for S-1752,” February 13, 1959. (5) Gibbs, H. D.,J . Biol. Chem. 72, 649 (1927). (6) Gottlieb, S.,Marsh, P. B., IND.ENG. CHEM.,ANAL.ED. 18, 16 (1946). (7) Inouye, H., Kanaya, Y., hfurata, Y., Chem. Pharm. Bull. 7, 573 (1959). ( 8 ) Johnson, C. A., Savidge, R. A., J . Pharm. Pharmcol. 10, Suppl. 171 T (1958). :a&, R. W., ANAL. CHEM. 21, 9 (1949). bfurai, K.,J . Pharm. SOC.(Japan) 1961 1. p. 1013. (12) Shaw, J. .4., ANAL. CHEM.23, 1788 (1951).
RECEIVEDfor review May 14, 1963. Accepted October 21,1963.
Systematic Neutron Activation Technique for the Determination of Trace Metals in Petroleum UMBERTO P. COLOMBO and GIUSEPPE SlRONl Geochernisfry Departrnenf, G. Donegani Research Insfifufe, Montecatini Co., Novara, Italy
G. B. FASOLO and RENZO MALVANO SORIN Nuclear Research Cenfer, Saluggia (Vercelli), Italy
b A systematic neutron activation technique, based on a series of consecutive irradiations and analyses of individual activities has been applied to the determination of trace metals in crude oils, distillation fractions, asphalts, and related substances. Analytical procedures are described which allow the determination of traces of V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Mo, Sb, and Au in petroleum products, even when only samples of about 1 gram are available to the laboratory. A detailed description of the experimental method is given, and the data obtained in the analysis of several Italian oils and asphalts are reported. The method does not require ashing of samples under analysis: some metals, such as V, Cr, Mn, Cu, and Sb, are determined directly on irradiated samples, while the other metals are counted after low temperature coking and chemical separation of activities. Serious losses of volatile metal-organic 802
0
ANALYTICAL CHEMISTRY
complexes are thus prevented. The different sources of error are discussed. The accuracy of the method i s in the order of 10%.
*
M
TRACE metals have been found in crude oils. Ball and his collaborators ( 1 ) have established the presence of as many as 28 trace metals in the ashes of 24 American crude oils. Their concentration varies from many parts per million to parts per billion, according to the metal and to the nature of the organic constituents of petroleum. There have been several endeavours by petroleum geochemists to use the data of quantitative analyses of trace metals for geological correlations among the crude oils bearing these trace metals. Attempts were made to detect relationships between trace metal concentrations and the ages of the crude oils, the stratigraphic positions of their traps, and AKY
the paleogeographic trends in the basin (2). More recently, interest in trace metals arose in connection with the progress made by geochemists in the understanding of the fundamental principles of the origin, migration, and alteration of petroleum. The work by Erdman and Harju ( 7 ) on the capacity of petroleum asphaltenes to complex heavy metals has shown that i t might be possible, from a measure of the degree of trace metal unsaturation in the asphaltenes, to reach an understanding of the history of migration of a crude oil, both as a micellar dispersion in the waters of the basin and a s a discrete phase. The systematic determination of the variations of trace metal content in the oils of a basin may provide a key to the history of the accumulation of oils in the basin, as shown by the work of Hodgson and Baker (11 ) . The importance of trace metals in
petroleum is obvioudy not only of geochemical value. I n catalytic petroleum processes trace metals from feedstocks can be deposited on the catalyst surface and slowly poison it. Heavy stocks often contain over 1 p.p.m. of the trace metals, while lighter stocks, which are at least as important to the petroleum industry, normally contain less than 1 p.p.m. of metal conlaminants. I n the range below 1 p.p.m., the available analytical methods usuelly involve dry or acid wet-ashing to concentrate the metals for subsequent analysis. The dry ashing spectrographic procedure is cumbersome and time-consuming. Furthermore, there is ,t serious problem from loss of volatilr~ metal complexes with most ashing procedures. I t is therefore important ,o develop a rapid and sensitive analytical method, not involving ashing procedures, to be applied to both residual and lighter streams. I n geochemical studies an analytical method requiring only a few grams of oil or asphali is highly desirable for correlation purposes during exploration work. Several activation methods for the determination of trace elements in petroleum products have been reported in the literature (3, 9, IO). I n general, however, such methods do not contemplate the analysis of many elements on a single sample. The authors have developed a systematic neutron activation technique for detrhrmination of trace metals in oils and related substances. This procedure is currently employed in their laboratories in connection with geochemical work, but it is believed that it could be employed successfully in the analysis of refinery FCC feedstocks and, more in general, of petroleum products.
tilled water passed through a deionizing column was used. Sodium tartrate solution: dissolve 10 grams of sodium d-tartrate in 100 rnl. of water. Ammonium oxalate solution: dissolve 10 grams of ammonium oxalate in 100 ml. of water. Hydroxylamine hydrochloride solution: dissolve 10 grams of hydroxylamine hydrochloride in 100 ml. of water. EDTA solution : dissolve 75 grams of disodium salt of (ethylenedinitrilo) tetraacetic acid in 1000 ml. of distilled water. Solution is ea. 0.02.11. Tribenzylamiiie solution: dissolve 1 gram of tribenzylamine in 100 ml. of chloroform. 8Quinolinol solution: dissolve 1 gram of 8-quinolinol in 100 ml. of chloroform. Dimethylglyoxime solution: dissolve 1 gram of dimethylglyoxime in 250 ml. of concentrated NHaOH. Add 250 ml. of water. Xeocuproine solution: dissolve 0.1 gram of neocuproine in 100 ml. of ethanol. 1-Nitroso-2-naphtho1 solution: dissolve 1 gram of 1nitroso-Bnaphthol in 50 ml. of acetic acid. Add 50 ml. of water. 1,lOPhenanthroline and hydrosylamine stripping solution: dissolve 0.25 gram of 1,lO-phenanthroline in 100 ml. of a 10% hydroxylamine-in-water solution adjusted at p H 3 t o 5 with acetic acidsodium acetate buffer. Potassium chloride stripping solution: dissolve 225 grams of potassium chloride in 1000 ml. of 0 . 5 S hydrochloric acid solution. Procedures. The metals which have been considered in the present
1
-
Irradiation in rabbit 30 60
EXPERIMENTAL
Reagents. All rezgents were prepared from reagent grade chemicals (British Drug Houfjes, Ltd.). Dis-
work, and their related radionuclides, are listed in Table I. The considerable differences of half lives among the various radionuclides measured allow the selection of suitable irradiation and cooling times, so as to limit to a few species the activities present in the sample under analysis. Furthermore, the absence of appreciable radioactivity in the organic matrix makes it possible to determine several trace metals with nondestructive neutron activation techniques. Therefore, a systematic procedure can be followed on the same sample, according to the analytical program outlined schematically in Figure 1. It was assumed that the analysis should be carried out with samples of about 1 gram, in order to allow the geochemical study of oils obtained from accumulations of purely geochemical importance. The irradiation time required for the analysis depends on the metals to be determined, and varies from about 30 seconds in the case of vanadium, to approsirnately 100 to 200 hours for such elements as chromium, antimony, iron, zinc, and cobalt. Nickel may be determined in either of two irradiation groups. If the rapidity of analysis is an important factor, the radionuclide considered is 2.56-hours Ni65 produced from Ni64 by ( n , ~reaction. ) If cobalt and iron also have to be determined, which requires a longer irradiation time, the radionuclide considered may be '71.3-days C058, produced from Ni58 by (n,p) reaction with fast neutrons. Irradiation. Oil samples (each of about 1 gram) and comparative standards (aluminum-base alloys or solutions of pure products, AnalaR B D H ) were irradiated in the Avogadro RS-1 swimming pool reactor of SORI;\', a t Saluggia. For short-time irradiations, as requested in the determination of vanadium, the pneumatic device (rabbit) supplied with the
hours ( decay 2 hours 1 111
0 Cu 6 M n
Irradiation in pool 100-200
Q Cr 6 S b
J3-L Coking and related
Figure 1.
olwsations
Coking,addition of carriers and separation of activities
Ni L C o
Coking and related
operations
&T
Scheme of determination of trace metals in petroleum by systematic neutron activation technique VOL. 36, NO. 4, APRIL 1964
803
reactor was used with a thermal neutron flux of about 2 X 1OI2 n/sq. cm. sec. All other irradiations were performed in high flux positions inside the core of the reactor (the highest thermal neutron flux is 4.2 X l O I 3 n/sq. cm. see., while the highest fast neutron flux is 1.3 X lO13n/sq. cm. sec.). For the irradiations u p to a few hours, polyethylene sample containers were used, while for longer irradiation periods, sealed quartz vials were necessary. The transfer of samples from one container to another after irradiation was performed using specially purified benzene as the solvent (4). The same benzene was employed as a diluent of viscous samples. Nondestructive Analyses. By selecting proper irradiation and cooling times (see Figure l ) , i t is possible to resolve, in the y-spectra of irradiated samples, the activities of different radionuclides. Therefore, the direct
Table
I.
Metal
Mn, and Zn, for which a n amount equivalent t o 15 mg. of metal was found to be convenient. The chemical yields, determined spectrophotometrically or gravimetrically, ranged between 65 and 100%. The radiochemical purification is generally satisfactory, and, in all cases, sufficient for gamma spectrometric measurements. The time involved in the separation of activities may be estimated in the order of 2 to 3 hours, since a difficult and lengthy ashing step after coking is not required. Counting. The counting of the activities was performed by means of a y-spectrometer equipped with a 200channel LABEN C-31 pulse analyzer and a Harshaw NaI (Tl) 3- x 3-inch high-resolution scintillation detector. Comparative standards were dissolved in hydrochloric acid (to which a few drops of H202 were previously added) and counted each time under the same conditions as the samples being analyzed. I n the analysis of molybdenum it was found convenient to perform the counting about 30 hours after chemical separation in order to reach the equilibrium value of Tc9gm formed from the decay of hlo99. Thus the sensitivity of the measure was substantially improved.
determination of vanadium, manganese, copper, chromium, and antimony is made in three different groups of irradiations. Whenever convenient, the decay of the activity of characteristic photopeaks can be measured. This step is necessary in the direct determinations of vanadium, because in most cases interfering Cl38 is present in relatively high amounts. The direct determination of such metals as Mo and Au is not possible because of the strong interference given by the activity of NaZ4, which prevents the resolution of the y-spectra of irradiated samples. Destructive Analyses. For each irradiation group, t h e chemical separation of activities was performed according to the procedures outlined i n Figures 2 t o 4. As carriers, salts of the metals under study were used. A weighed amount of each carrier was added, corresponding to 0.5 mg. of the metal, with t h e exception of Au, Cr,
Trace Metals Determined in Oils and Asphalts, and Characteristics of Related Radionuclides (6) Radionuclide Energy of ?-radiation
produced
V
Half life 3.77m. 27.8d. 2.58h.
v 5 2
Cr
Cr51
hln
Fe co
4.5 __
Ni Xi cu Zn hlo Sb Au
a. -.
~
5.3y. 71.3 d. 2.56 h. 12.8h. 245 d. 66 .O h. 60.9d. 2.7d.
Zn66
+
Moss Tc99" Sb1Z4
Au198
measured (rn.e.v.) 1.43 0.32 0.84 1 .lo-1.29 ~. ~. 1.17-1.33 0.511-0.80 0.37-1.11-1.48 0.511
RESULTS
Eight samples of oils and asphalts from different Italian sources were analyzed with the method illustrated above. A description of the samples studied has been published elsewhere ( 5 ) . The data are presented in Table I1 along with the respective experimental standard deviations, which in most cases were found to be better than +lOyo.
~
1.11 0.141-0.181 0.60-1.69-2.09 0.411
( 1 gram)
Addition of c a r r i e r s , c o k i n g w i t h HZSO;, fusion with Na2CO3+K2CO~,leaching with dilute H C I , oxidation with drops of H 2 0 2 ,precipitationwi~NH4OH,(iltration
I
cake
Filter
L;_l
Solubilization in dilute H2SOI+dropr of H 2 0 2
I
Additionof sodium t a r t r a t e solution
I
Additmn of NaOHup t o p H 6 - 1 2
I
Extraction w i t h 8-quinolinol solution I
r-l
j : i Organic phase
Aqueous phase
I
#activity counting lor Mn56
I
Spectrophotometric determination of c h i
I
Waste
I Addition of H C I
I sactivitycounting for N i 6 5
to p H 4 - 5
I
Extraction w i t h neotuproine solution
mica1 y i e l d ( 3 9 5 m ) r ) o n ~ l i q u o t
I
Spectrophotometric determination of chc mica1 yield 0 6 6 m p ) on aliquot
Organic phase
z r c t i v i t y counting for ~ ~ 6 4 .
Waste
I Sprclrophohmetric dekrmination d c k mica1 yield ( 4 5 8 m p 1 on aliquot
Figure 2.
804
Scheme of y-activity separation for Mn, Ni, Cu after irradiation of petroleum (irradiation time: 1-2.5 hr.)
ANALYTICAL CHEMISTRY
Table 11. Results of Trace Metal Determinations on Italian Oils and Asphalts Sample Vanadium Chromium Manganese Iron type Locality Cobalt Xckel Crude oil Alanno 7.2 f0.1 0.068 f 0.003 0.017 f 0.003 0 . 9 f 0 . 0 1 0.003 f 0.00005 3.0 f0.2 Crude oil Chieuti 0.042 f 0.002 0.023 f 0.003 0.023 f 0.003 0.6 f0.1 0.0015 f 0.0005 0.2 5 0.0 Crude oil Cercemaggiore 0.025 f 0.001 0.045 f 0 . 0 1 0.013 f 0.001 0 . 7 f0.05