slowly toward the thermowell temperature as the reaction slowed. The thermowell temperature is thus a totally inadequate measure of sample temperature, errors of 30” C. or more resulting in some cases. The rate of increase of sample temperature changes significantly during chemical reactions (Figure 3). Any treatment of data to obtain kinetic information which involves this rate, therefore, cannot rely simply on the rate a t which the furnace temperature is increasing. For kinetic studies with the thermobalance to have significance, one should measure sample temperature and sample weight simultaneously.
R.
A&NO
w LEDGMENT
The authors are indebted to E. T. Domboski and W. H. Chappell, Jr., for setting up the thermobalance equipment and performing the runs, and L. R. Ocone, who suggested using the sleeve to prevent condensation on the crucible support and who prepared the chromium and cobalt compounds. They are also indebted to a referee for the information that Newkirk’s work was done in air. LITERATURE CITED
(1) Doyle, C. D., WADC Tech. Rept. 59-136, Contract AF 33(616)-5576,
Project No. 7340, April 1959. (Ale0 subsequent reports on this project.) (2) Duvd, ClBment, “Inorganic Therraogravimetric Analysis,” Elsevier, New York, 1953. (3) Mielenz, R. C., Schieltz, N. C., King, E., in L‘Claya and Clay Minerals,” NAS-NRC Publ. 327, pp. 285-314,
ruary 4, ‘1960. RECEIVED for review March 15, 1961. Accepted August 28, 1961. Work supported in part by the Ofice of Naval Research.
w. KLlPP
Research and Development Department, American Oil Co., Whiting, Ind.
b Sulfur below 380 p.p.m. in leaded gasoline cannot b e determined accurately by conventional methods. The ASTM lamp-turbidimetric method for trace sulfur in petroleum products gives low results because water-insoluble sulfates are deposited on the chimney during cornbustion. By a modified procedure, these deposits are recovered by washing with hydrochloric acid and ammonium acetate. In the subsequent determination of sulfate in the absorber solution, provision is made for the effects of the lead ion present. With the new procedure, accuracy and precision of the lampturbidimetric method are within 5% relative on leaded samples containing 30 to 350 p.p.m. of sulfur. compounds in gasoline reduce the effectiveness of such antiknock agents as tetraalkyllead in raising octane number. Modern petroleum processes-alkylation, reforming, and isomerization-produce high-octane gasoline of low aulfur content. Most premium gasolines today contain only traces-less than 300 p.p.m.-of sulfur, and the trend is toward still lower concentrations. For controlling quality and €or evaluating the relationship between sulfur content and antiknock characteristics, a reliable method for determining trace sulfur in the presence of lead is needed. Sulfur in gasoline is usually determined by the ASTM method for sulfur in petroleum products by lamp combustion (1). The liquid sample is fed ULFUR
through a cotton wick and burned in an atmosphere of 70% carbon dioxide and 30% oxygen. Sulfur is burned to oxides that are adsorbed and oxidized to sulfate by 1,5% hydrogen peroxide. Sulfate in the absorber solution is determined either acidimetrically as sulfuric acid or turbidimetrically as barium sulfate. For traces of sulfur, the acidimetric method lacks sensitivity, so the turbidimetric method is usually employed. An alternative is the spectrophotometric barium chloranilate method (2). With unleaded gasolines, the turbidimetric or the chloranilate method gives reliable results for sulfur concentrations as low as 1 p.p.m. But, on samples containing lead alkyls, both methods give low results-for example, with samples containing 3 ml. of tetraethyllead fluid per gallon and 35 to 350 pap.m. of sulfur, turbidimetric results are 20 to 4Q7, low and chloranilate results are more than 50% low. Loss of sulfur in water-insoluble deposits on the chimney and burner tip of the combustion apparatus is the major cause of the error., In addition, lead ion in the absorber solution interferes with both methods for determining sulfate. DEVELOPMENT OF METHOD
Principal problems in the development of a reliable method were identification of chimney deposits, selection of a suitable solvent for these deposits, and elimination of interference by lead in the sulfate determination. The deposits left on the chimney by the combustion of a leaded gasoline
are not completely dissolved by the hot-water wash used in the ASTM method (1). The partial solubility of these deposits produces “straggling” of the sulfur, low results on samples containing lead, and high results on unleaded samples analyzed in apparatus previously used for samples containing lead. Chemical analysis of deposits remaining after washing with hot water showed a lead-to-sulfur ratio of 1.6 to 1, which could be accounted for by a mixture of PbS04 and PbO . PbS04. The presence of PbSOa was confirmed by x-ray diffraction. Further studies by x-ray diffraction showed lead SUIfate in all chimney deposits but lead halides only in the deposits from gasolines of very iow sulfur content. Deposits from lamp combustion are similar to those found on the hot central electrode of spark plugs, which have been identified as mostly PbO and PbO . PbS04with some PbO . PbClBr (3). A commonly used solvent for lead salts is ammonium acetate solution. This solvent alone, however, did not dissolve the chimney deposits completely, probably because PbO is only slightly soluble in it. Washing the deposits with 3 ml. of 1M NC1 followed by 5 ml. of 5% ammonium acetate dissolved them rapidly. Lead recovery averaged 95%, as compared with 60% by the normal hot-water wash. Retention of lead and sulfur on the cotton lamp wick appears to be negligible; neither element was detected in the hydrochloric acid-ammonium acetate extracts from 20 wicks used to burn leaded samples.
Because the absorber solution and washings contain lead, interference by lead must be considered. In the turbidimetric method, lead increases the turbidity of the barium sulfate suspension and results in a positive error of about 15Yc relative; either the lead must be removed, or its concentration must be controlled and a special calibration curve must be used. In the chloranilate method, lead precipitates the colored chloranilate ion released by sulfate ion and causes a negative error as high as 100% relative; the lead must be completely removed to obtain correct results. Although ion exchange is a convenient technique for removing lead from aqueous solution, only about 95% of the lead in synthetic absorber solutions was removed by passage through a cation exchange resin. This treatment eliminated the lead interference in turbidimetric determination of sulfate, but a negative error of about 0.1 mg. of sulfur was still obtained with the chloranilate method. Failure to remove lead completely is probably due to the presence of ammonium acetate, which may prevent complete ionization of lead. Ion exchange can be used with the turbidimetric method if the lead content of the sample is unknown, or if a large sample is needed for maximum precision. However, for gasolines of known lead content containing more than 50 p.p.m, of sulfur, ion exchange can be eliminated by controlling lead content. In the method ultimately developed, lead concentration is controlled by adjusting sample size according to lead alkyl content. Hydrochloric acid and ammonium acetate are used to dissolve chimney and burner deposits. Sulfate in the absorber solution is determined turbidimetrically by using a special calibration curve. METHOD
In applying the method, one must first develop a turbidimetric calibration curve for sulfate in the presence of the same quantity of lead and ammonium acetate encountered in the final absorber solutions. Place in 50-ml. volumetric flasks, 0.25, 0.50, 0.75, 1.00, 1.50, 2.00, 3.00, 4.00, and 5.00 ml. of 0.00624N sulfuric acid. To each of these flasks and to a flask reserved for a reagent blank, add 5 ml. of make-up solution-a 0.6M HC1 solution containing 0.9 gram of PbC12, 0.58 gram of PbBrz, and 50 grams of ammonium acetate per liter. Dilute to the mark with water. For each solution, determine the turbidity of a barium sulfate suspension a t 425 mp, according to Appendix I of ASTM D 1266 ( I ) . Obtain the net absorbance for each standard by subtracting the initial reading and the net reagentblank reading from the observed read-
Table 1.
Precision and Accuracy of
Knowna
36
Av. 36.1
&hod on Leaded Gasoline Sulfur Content, P.P.M. 73 135 338
73.7 2.8 31 40 56
Std. dev. 1 . 9 Lamp-turbidimetric(ASTM D 1266) 31 20
...
134 4.9
326 9.4 26 1 280 263
90
74
...
Determined by multiple lamp-turbidimetricanalyses before adding 3 ml. TEL/gallon.
ing. Plot the net absorbance of each standard against milligrams of sulfur to obtain a nearly linear calibration curve. Burn the gasoline sample according to the ASTM method (1). Because the method involves dissolving lampchimney deposits, all glassware must be chemically clean to avoid sulfur contamination that would be significant at the trace level. Use a full-carburetion burner for all samples (4). This burner is suitable for the combustion of highly aromatic samples and has an elongated single tip from which deposits are dissolved easily. Maintain a total lead content of 5 mg. by burning a volume of sample equivalent in milliliters to 19 divided by the grams of lead per gallon in the sample. Draw combustion atmosphere through one absorber of a set to serve as a combustion-atmosphere and reagent blank. Rinse the spray trap with about 10 ml. of water and add the rinsings to the solution in the absorber. Add 3 ml. of 1N HCl to the inverted chimney and place the tip of the burner, with wick removed, in the exit tube of the chimney Wet the deposits with the solution by appropriate manipulation of the chimney until the deposits are loosened. Remove the burner tip, wash it with water, and transfer the solution in the chimney to the absorber. Repeat these washing manipulations with 5 ml. of 5% ammonium acetate, adding the washings to the absorber. Swirl the solution t o dissolve any deposits on the absorber walls and transfer to a 250-ml. beaker. Evaporate on a hot plate to a volume of about 20 ml. Transfer to a 50-ml. volumetric flask and make up to volume with water. For each absorber solution, determine the turbidity of a barium sulfate suspension at 425 mp, according to Appendix I of the ASTM method (1). If the total sulfur content of the absorber solution is known to be less than 0.5 mg., determine the turbidity of the entire contents. If the sulfur content is known t o be more than 0.5 mg., select an aliquot t o yield about 0.2 mg. of sulfur. If the sulfur content is un175 176
153 155
184 199
111 125
125 138
known, use a 10-ml. aliquot. Place the aliquot in a 50-ml. volumetric flask, and add the milliliters of make-up solution determined by subtracting the aliquot fraction taken for analysis from unity and multiplying the result by 5 . Make up to volume with water. Determine the turbidity of these aliquots. If the absorbance of a 10-ml. aliquot is less than 0.05, bring the remaining 40 ml. in the original flask to the full mark by adding 1 ml. of make-up solution and water. Repeat the turbidity determination. From the absorbance reading, subtract the initial absorbance and the net absorbance of the combustionatmosphere and reagent blank diluted like the sample. Convert the net absorbance of the sample to milligrams of sulfur by using the calibration curve. Calculate the sulfur concentration from the equation: P.p.m. S
=
A WF X 1000
where A is milligrams of sulfur in the absorber solution or aliquot, W is the grams of sample burned, and F is the aliquot fraction taken for turbidity measurement. RESULTS
The new method was used to analyze four gasoline samples containing 3 ml. of motor tetraethyllead fluid per gallon. The sulfur content of these samples was established by multiple lamp-turbidimetric determinations before adding TEL fluid. Results for the leaded samples are shown in Table I. Precision is similar to that of turbidimetric analyses in the absence of tetraethyllead (9). Accuracy of the method is good, the averages deviating less than 2% relative from the established values. By comparison, results obtained by the normal turbidimetric procedure show poor precision and are about 30% low. Duplicate sulfur determinations of ten commercial premium gasolines gave the values (in p.p.m.) : 256 249
252 255
342 350
398 404
VOL 33, NO. 13, DECEMBER 1961
418 426
1913
The standard deviation was 4.8 p.p.m. TEL content ranged from 2.3 to 3.1 ml. per gallon. Precision is equivalent t o that observed on the four previous samples. The inclusion of ignition-control additives containing phosphorus or subatitution of a lead alkyl other than tetraethyllead not affect The similarity between deposits from
spark plugs and from lamp chimneys suggests the possibility of using the analysis of chimney deposits t o Predict what types of inorganic deposits will be formed in engines by various fuel ComPOnents. LITERATURE CITED
(1) Am. SOC. Testing Materials, Phil-
adelphia, Pa., "ASTM Standards 0: Petroleum Products and Lubricants,
Method
126659T, 1960.
((3)2Lauer, k ~ ~ ~ ~Friel, ~ ~H P.&9 .~Miller, ~ E. ~ ~ & B. D., "1
J. L.,
J.,
SAE Meeting, Tulsa, Okla., Nov. 5, 1958.
(4) Wear, G.E.C., Quiram, E. R., ANAL. CHEM.21,721-5 (1949). RECEIVED for review May 5, 1961. Accepted Au st 28, 1961. Division of
Petroleum &ernistry, 140th Meeting, ACS, Chicago, Ill., September 1961.
ire Assay Meth G. H. FAYE and W. R. INMAN Mineral Sciences Division, Mines Branch, Department of Mines and Technical Surveys, Ottawa, Ontario, Canada
b Gold is shown to be quantitatively collected in tin in the proposed fire assay method. The assay buttons are arted in hydrochloric acid and the old remains in the insoluble residue of intermetallic compounds. The procedure for treating the residue and selectively extracting the gold into diethyl ether is described. The proposed method has been applied to the analysis of an ore concentrate, a copper-nickel matte, and a number of rock samples. The results agree favorably with those obtained by conventional methods on the same materials. new fire assay method for the determination of platinum and palladium in copper-nickel matte and in ore concentrates was recently reported by Faye and Inman ( 3 ) . In this method, the precious metals are collected in molten tin during the crucible fusion process and the resultant tin alloy is then treated by wet chemical methods to isolate and determine the indjvidual platinum metals. In subsequent work, an attempt has been made to extend the tin collection technique and devise an analytical scheme in which gold and the platinum metals can be determined in a single sample. The behavior of gold in the scheme is studied in the present work. This metal should be isolated a t an early stage in the procedure so that it will not interfere later in the determination of the platinum group metals. In the reported method for platinum and palladium (3), the samples were given a preliminary acid leach to remove most of the copper, nickel, and iron. Since the publication of that work it has been found that, with certain minor modifications in the procedure, samples of widely varying composition can be
fused directly, even when they contain substantial quantities of copper, nickel, and iron. The present paper describes the successful use of the modified tin collection technique in the fire assay determination of gold in samples of rock, ore, and copper-nickel matte. For comparison purposes, the gold values obtained by the lead collection technique on these samples are also given, as this method is the single acceptable method so far recorded for these kinds of material. APPARATUS AND REAGENTS
Assay furnace and Vycor melting tube, described previously ( 3 ) . Jelras Handy-Melt electric furnace, ,Model B. Spectrophotometer, neckman Model
B.
Standard gold solution. This was prepared by dissolving 100 mg. of Johnson and Matthey Specpure gold sponge in aqua regia. The resulting solution was diluted to 1 liter so that the final volume contained approsimately 60 ml. of concentrated hydrochloric acid and 40 ml. of concentrated nitric acid. The gold content of the solution was determined by a gravimetric method involving hydroquinone ( I ) and was 0.100 mg. per ml. More dilute solutions were prepared from the stock solution by twentyfold dilution. Flus for crucible fusion: Grams 35 50 10-20 10
6-8
SiOl adjusted according to silica content of sample. 11
b
The larger quantity of coke taken
for samples high in iron.
Diethyl ether, analytical reagent, obtained from 3lallinekrodt Chemical Works.
o-Tolidine, analytical reagent, obtained from British Drug Houses Ltd. EXPERIMENTAL PROCEDURE
Preparation of Tin Assay Buttons. The crucible fusion process used in the present work was a modification of that described previously (3), in that the fusions were conducted a t approximately 1250" C. for 1 hour and the flux was of the composition given above. With the exception of certain tests involving leaching of copper-nickel matte, all samples analyzed by the proposed method were roasted a t 750" C. for approximately 1 hour before being mised with the flus and taken through the crucible fusion process. When it was known that the sample taken for fusion containpd more than 3 grams of combined copper and nickel, 16 to 25 grams of stick tin were added t o the charge to lower the melting point of the resultant alloy. Analysis of Tin Assay Buttons. For convenience, most assay buttons were melted under nitrogen in a Vycor melting tube, and the molten tin alloy was poured into several liters of water in an enamel pail t o produce a spongy mass as in the previous work ( 3 ) .. However, buttons containing appreciable quantities of copper and nickpl had a comparatively high melting point (600" to 650" C.) and were more easily handled in the Jelrus HandyMelt portable furnace. As before, these buttons were melted under nitrogen and the melts were poured into water to qranulate the alloy. Any large lumps produced in this operation were easily reduced in size with snips. Each sample was then transferred to a 400-ml. beaker and treated with approximately 150 ml. of Concentrated hydrochloric acid. The beaker and contents were heated until the excess tin had dissolved and vigorous evolution of bubbles from the black insoluble residue had ceased. After diluting to approsimately 360 mi. with water and
g
f
'
