Analyst, October, 1968, Vol. 93, pp. 663-668 663 A Combus tion Method with a Radiometric Finish for the Determination of Microgram Amounts of Sulphur in Light Petroleum BY E. V. GOODE (Chemistry Department, Derby and District College of Technology, Derby) A radiochemical technique has been combined with a modification of the surface combustion method of Schoberl to give a rapid method for the determination of sulphur in the range 1 to 10 p.p.m. in organic substances such as purified light petroleum. The sulphate formed is precipitated with barium chloride labelled with barium-133, and the amount of barium sulphate obtained determined by measuring its activity. A radiometric finish has the advantage of potential sensitivity, and barium-133 can be obtained with a specific activity of 1 Ci per g.Thus, a scintillation counter with an efficiency of 30 per cent. and a background of 20 counts per second (shielded) can detect 0.4 x 10-3pg of sulphur, as barium sulphate, assuming no solubility losses. The lowest detectable concentration of sulphur is dependent on the magnitude of the blank value, but because sulphate appears to be an almost universal contaminant it was found, under normal laboratory conditions, that it was impossible to reduce the blank value to much below the equivalent of 1 p.p,m. in the sample. THE various methods described in the literature for determining sulphur in organic substances involve either reduction of the sulphur compound to hydrogen sulphide or combustion to sulphur dioxide and absorption in hydrogen peroxide to give sulphuric acid.Both types of method must be combined with a suitable finish to determine the resulting compound. The most popular reduction procedure appears to be Granatelli'sl method with Raney nickel, which was modified by Reed.2 The resulting hydrogen sulphide is converted into methylene blue by the method of Kriege and Wolfe3 and measured spectrophotometrically. Detection down to 0.02 p.p.m. is claimed but the method is lengthy (2 to 3 hours), and there is evidence that some sulphur compounds are resistant to the Raney nickel reduction. In the Wickbold4 method 50-g samples can be burnt in 15 minutes in an oxy-hydrogen flame pro- duced with a special burner; with such high gas flow-rates (50 cu. ft. per hour) gas purification is difficult, and high blank values are obtained.The oxygen-flask method5 appears ideal for handling small amounts when the sample has a high sulphur content but would not be suitable in the p.p.m. range. The Schoberl surface combustion method6 falls between the two extremes. While primarily intended for milligram amounts, it can, as described later, be modified to enable 2-g samples to be burnt in about 15 minutes. Methods for determining the resulting sulphuric acid include nephelometric titration with barium chloride (40.0 p.p.m.), a colorimetric method in which barium chloranilate' is used (0.06 p.p.m.) and flame molecular-emission spectroscopy8 (6.4 p.p.m.). Because of the widespread occurrence of sulphate, high blank values are to be expected. There do not appear to be reports of a radiochemical finish, yet such a method would be sensitive, providing that the radiochemical reagent could be obtained with a sufficiently high specific activity.Barium-133, in the form of barium chloride (Radiochemical Centre, Amersham), is an ideal tracer; it can be obtained with a specific activity of 1 Ci per g. It is a y-emitter, giving a cascade of photons with an average emission of 1.36 per nuclear transformation, and this factor, together with an average energy of 0.3 MeV, makes detection with a scintillation counter easy and efficient; its half-life of 10 years results in small decay losses and a long shelf-life and, finally, it is comparatively cheap. The combustion processes differ in the amount of sample that can be handled. 0 SAC and the author.664 GOODE: A COMBUSTION METHOD WITH A RADIOMETRIC FINISH FOR [Analyst, Vol.93 EXPERIMENTAL Providing the combustion is completely clean, as shown by absence of partly burnt products, it can be assumed that all of the sulphur compounds present have been converted into sulphur dioxide. Further, the hydrogen peroxide absorption appears to be quantitative down to very low concentrations and at high gas flow-rates. The sulphate in the hydrogen peroxide absorbent is then precipitated with the radioactive barium chloride reagent, and the amount thus obtained is determined by measuring the activity. Because of anticipated high blank values, the specific activity of the reagent, as purchased, was reduced by a factor of 350 by dilution with the inactive form to bring the sensitivity to the same order of magnitude as the antici- pated blank values.Nevertheless, the method was still sensitive because it was found that each microgram of sulphur, in the form of sulphate, will precipitate sufficient barium chloride to give a count-rate of about 2250 counts per 10 seconds, with a counting efficiency of 30 per cent. The recovery of the barium sulphate must be quantitative and the precipitate com- pletely free from excess of active barium chloride. A limitation is imposed on this method by the solubility of barium sulphate in water. At room temperature, it varies from 3 mg per litre in neutral solution to 100 mg per litre in 2 N acid solution. Acid must be present to prevent the interference from carbonate, phos- phate and silicate ions.Assuming a solubility of 10 mg per litre in weakly acidic solution, and a volume of 10 ml, the lowest limit for the detection of sulphur will be 13.0 pg, corre- sponding to 6.5 p.p.m. for a 2-ml sample. In this work the solubility limitation was overcome by evaporating to dryness after decomposing the carbonate with acid, and the excess of barium chloride removed with anhydrous methanol ; barium chloride is fairly soluble in this solvent, whereas barium sulphate is extremely insoluble, being about 0.3 mg per litre at room temperature. This would still permit interference by phosphate and silicate ions, but in some preliminary work it was found that both are easily removed with a solution of 1 per cent. hydrogen chloride gas in methanol, although this increases the solubility of barium sulphate to 0.5 mg per litre; these values were determined by a radiochemical m e t h ~ d .~ Originally, it was intended to wash the barium sulphate with methanol in a centrifuge tube, spinning after each washing, and then measuring the activity by placing the tube on top of the sodium iodide crystal of a scintillation counter, but it was discovered that the barium sulphate precipitate adheres tenaciously to the bottom of the tube so that centrifuging was unnecessary between washings. Consequently, the thick-walled centrifuge tubes were replaced by thinner-walled Pyrex glass test-tubes, cut down to a convenient length. In this method, the sample is volatilised in a stream of air and the mixture passed through heated sintered-silica discs, where combustion occurs ; the sulphur dioxide formed is absorbed in hydrogen peroxide solution.As this is a semi-micro method it required to be modified to enable 2-ml samples to be used. In this work, a combustion process was selected because it is absolute. The Schoberl surface combustion method was used for burning the sample. DESCRIPTION OF THE APPARATUS, REAGENTS AND TECHNIQUE COMBUSTION FURNACE- A tube furnace, fitted with a thermo- couple pyrometer and thermostat, was used. The tube consisted of a piece of transparent quartz tubing 60 cm in length, external diameter 1.4 cm and wall thickness 0.1 cm. The section running through the furnace was loosely packed with fused quartz-wool, with the fibres, as far as possible, running longitudinally.This plug, 20 cm in length, provided a large surface area for the combustion, which was necessary for 2-ml samples, as opposed to the Schoberl method in which only 20 to 30-mg samples are used. The plug was held in position by means of a platinum wire, which had a loop at one end to retain the quartz plug, the other end being secured by bending it over the inlet end of the furnace tube. Other furnace packings were tested, including narrow-bore thin-walled translucent silica tubing, but the combustion was not so efficient, and large variable blanks were obtained. The quartz-wool plug initially gave considerable blank values that were reduced by flushing out with air for 6 hours at 850" C. These furnace blanks were assumed to be caused by traces of sulphate volatilising from the silica.The possibility of silica volatilising from the Details of the apparatus are shown in Fig. 1.October, 19681 THE DETERMINATION OF MICROGRAM AMOUNTS OF SULPHUR 665 furnace was considered but barium silicate, unlike barium sulphate, does not adhere ten- aciously to glass surfaces. A diffusion plug of quartz-wool was inserted into the inlet end of the furnace tube to homogenise the vapour - air mixture, and to contain any flash-back if the sample was introduced too rapidly. The furnace tube was connected to the other parts of the apparatus by using polythene bungs fitted around the tube at about 3 cm from the ends; B 19 ground silica cones fused to the ends of the furnace tube would have been preferred. The 25-cm length of tube emerging from the furnace provided some surface cooling of the effluent gases, but not sufficient for condensation of moisture to occur, which would carry sulphur dioxide down with it.The furnace temperature was controlled at between 800" and 850" C; at lower temperatures there was evidence of incomplete combustion. Suction 4. Vandermic Furnace 850°C minute Fig. 1. Modified Schoberl surface combustion apparatus INJECTION AND VOLATILISATION OF THE SAMPLE- At about 3 cm from the end of one of the horizontal arms a B 19 socket was fused by using an internal seal; a 2-ml nylon syringe fitted with a hypodermic needle was connected to the vertical arm, and the sample was added, dropwise, from the syringe so that it fell on to a small quartz-wool plug placed at the junction of the T-piece.The sample was added at a rate such that the platinum wire adjacent to the mouth of the furnace glowed red hot, but not so fast that flash-back occurred. With an air flow-rate of 2 litres per minute the combustion time was 17 to 18 minutes. To ensure complete volatilisation of any traces of high-boiling fractions left on the plug, the T-piece was heated for a further 2 to 3 minutes to bring the total time to 20 minutes. To avoid this manual method of injection the T-piece could be replaced by a small Drechsel gas washing-bottle, the 2-ml sample being placed in the bottom and the stream of air bubbled through it, but with this method care must be taken because any flash-back passing beyond the diffusion plug would lead to an explosion in the bottle. PURIFICATION OF AIR USED FOR THE COMBUSTION- Air was drawn from outside the laboratory and purified by bubbling through 10 vol hydrogen peroxide contained in a 125-ml Drechsel gas washing-bottle ; it was then passed through a similar bottle containing distilled water, PVC tubing being used to make the connections.For example, the air was passed through sintered-glass discs supporting columns of reagents, such as alkaline permanganate as well as hydrogen peroxide, but no distinct improvements were obtained either in the magnitude or the constancy of the blanks. A possible improvement would have been to use cylinders of compressed air, which is of constant composition and would have placed the whole of the apparatus under a slight pressure instead of a slight vacuum.This would have pre- vented possible contamination from the laboratory atmosphere at any joints not completely air-tight. ABSORPTION OF SULPHUR DIOXIDE- To keep the blank from the absorbent as low as possible, only 10-ml portions of hydrogen peroxide were used at the lowest concentration possible. The vessel used for containing the absorbent consisted of the head from a Drechsel bottle fitted to a 2.5-cm diameter Pyrex test-tube that had a B 24 ground-glass socket; the bubbling tube of the Drechsel head was A 4-mm bore T-piece was used to inject the sample into the furnace. Other methods of purification were tested.666 GOODE : A COMBUSTION METHOD WITH A RADIOMETRIC FINISH FOR [A~zalyst, Vol. 93 extended to reach the bottom of the test-tube. A B 19 ground-glass socket was fused to the inlet tube of the head for making a connection to the polythene bung of the furnace tube.Considerable condensation occurred in the cylindrical part of this B 19 socket during the combustion, and it was always rinsed out with a portion of the hydrogen peroxide absorbent before being transferred for the radiochemical determination. It was found that 1 vol hydrogen peroxide was sufficient to absorb the sulphur dioxide completely, but there was about a 5 per cent. loss, which appeared to arise from carry-over by spray. Before use, the absorption vessel was well rinsed but not dried; after the combustion there was always a considerable increase in volume caused by condensation of water from the combustion. The final volume was found by weighing the absorption vessel, which avoided the rinsing out with distilled water that would have led to dilution. The 1 vol hydrogen peroxide was prepared by diluting a specially purified non-stabilised grade of 100 vol hydrogen peroxide* with water obtained from alkaline permanganate solution by distillation with a fractionating column.All of the solutions were stored in plastic bottles. Blank determinations were repeatedly made on both the distilled water and the 1 vol hydrogen peroxide. RADIOCHEMICAL DETERMINATION OF THE SULPHATE IN THE ABSORBENT- The volume of hydrogen peroxide absorbent used for the radiochemical determination will depend on the sulphur content of the sample. Ideally, the volume taken should be such that its sulphate content is equivalent to about 2 pg of sulphur, e.g., if the petroleum sample contains about 10 p.p.m.of sulphur, then 1 ml of absorbent would be sufficient, but if the anticipated sulphur concentration is only 1 p.p.m., then at least 5 ml of the absorbent should be taken. The absorbent was introduced, by pipette, into a 3-cm diameter Pyrex glass test- tube, cut down to 6 cm in length and kept upright by standing in a small beaker. The appro- priate amounts of radioactive barium chloride, to precipitate the sulphate, and hydrochloric acid, to decompose the carbonate, were added by using auto-zero pipettes. The solution was evaporated to dryness under a heat lamp, the time taken being about 15 minutes, but it could be reduced by using larger diameter tubes. The use of test-tubes has the advantage that the barium sulphate precipitate will always be located at the bottom of the radius of the tube, whereas with flat-bottomed containers the precipitate may be located anywhere on the base, and this would lead to inconsistent activity measurements.The active precipitate was washed free from excess of barium chloride reagent with small portions of anhydrous methanol until constant activity was obtained. The total blank activity was subtracted from this value and the resulting net activity expressed as micrograms of sulphur. The scintillation counter used for the activity measurements was fitted with a 3-cm diameter sodium iodide crystal, and the test-tube containing the activity mounted directly above it. The y-photons from the barium-133 are sufficiently energetic to penetrate both the wall of the test-tube and the aluminium can enclosing the crystal.In all of the activity measurements the time taken to record 10,000 counts was noted and results expressed as counts per 10 seconds. In this way the counting error caused by the random nature of decay was reduced to 1 per cent. The background of the counter was about 400 counts per 10 seconds and the crystal was not shielded. RADIOACTIVE BARIUM CHLORIDE REAGENT- Barium-133 was selected for labelling because of both its long half-life and energetic y-radiation; 1OOpCi of barium-133 were purchased in the form of barium chloride, in a volume of 1 ml, the weight of barium being 70 pg. This solution was diluted to 250 ml with inactive 0.002 N barium chloride. For each determination, 0.1 ml of this solution was used, the amount of sulphate that it will precipitate being equivalent to 3.2 pg of sulphur. The reagent was calibrated periodically against 0-1 ml of 0.001 N sulphuric acid; in this way a check was kept on the specific activity of the reagent. Dilution of the reagent does not affect its calibration, and for very low sulphur contents the reagent was diluted with twice its volume of water.With a half-life of 10 years, the decay correction is very small and, in any event, it is avoided by the calibration with sulphuric acid. * Mallinckrodt Transistar Chemicals. Supplier Camlab (Glass) Ltd., Cambridge.October, 19681 THE DETERMINATION OF MICROGRAM AMOUNTS OF SULPHUR 667 Health hazard and waste disposal are not serious problems as 0.1 ml of reagent contains only 0.04 pCi of activity; however, the usual precautions recommended when handling radio- active materials should be observed.The isotope is comparatively inexpensive and the cost of each determination is negligible. METHOD Flush the apparatus for 30 minutes with an air flow of 2 litres per minute, and a furnace temperature of 800" to 850" C. Rinse the absorption tube and head with distilled water and drain by shaking them. Add 10 ml of 1 vol hydrogen peroxide and connect the vessel to the furnace tube, moistening the polythene bung. Draw 2 ml of sample into the nylon syringe and fit it to the vertical arm of the T-piece injector. Initially, draw air through the furnace slowly and note that the bubbling rates in both the absorber and wash-bottles are the same, thus checking that the apparatus is air-tight.Set the air flow to 2 litres per minute and add the sample, dropwise, at a rate such that the platinum wire glows red hot but flash-back does not occur. When all of the sample has been added, heat the T-joint with a blast of hot air to volatilise any high-boiling fractions, the total time taken being 20 minutes. Weigh the absorption tube and head (previously weighed dry and empty) to the nearest decigram to give the total volume of absorbent, then rinse the inlet socket with some of the used absorbent and transfer the entire contents to a plastic sample bottle. Transfer, by pipette, a volume of the absorbent, depending on the sulphur content of the sample, into the special test-tube, together with 0.1 ml of barium chloride reagent and 0.1 ml of 4 N hydrochloric acid, and evaporate to dryness under a heat lamp.Wash the residue with 1-ml portions of anhydrous methanol until constant activity is obtained. Make a combined blank determination on the hydrogen peroxide absorbent, the air and the apparatus itself under exactly the same conditions. Determine the background activity of the counter. RESULTS Heptane, to which sufficient $-toluene sulphonyl chloride was added to give 10 p.p.m. (w/v) of sulphur, was used. A 2-ml portion was burnt off, and a known volume of the hydrogen peroxide absorbent reacted with 0.1 ml of the radioactive barium chloride reagent, as previously described. All activities have been expressed as counts per 10 seconds. Volume of absorbent from the combustion .. * . .. . . 13ml Net activity of barium sulphate from 1 ml of absorbent . . . . 4000 counts Volume of absorbent from the blank . . .. . . . . .. 11ml Net activity of blank from 2 ml of absorbent . . .. . . . . 700 counts From the calibration of the barium chloride reagent with 0.001 N sul- .. .. . . 2250 counts phuric acid, 1 pg of sulphur . . .. 2250 Sulphur in 2 ml of heptane = = 21.4 pg -- 10.7 p.p.m. w/v. The discrepancy of 0.7 is caused by the heptane containing about 1 p.p.m. of sulphur. DISCUSSION The lowest detectable amount of sulphur will be dependent on the magnitude of the blank, to which the quartz tube and packing make a considerable contribution. The grade of hydrogen peroxide used was the purest obtainable, but traces of sulphate and phosphate impurities were present.The distilled water also had a slight sulphate content, which possibly resulted from the evaporation in glass test-tubes. The purified air used for the combustion also had a blank value. The average blank value corresponding to a determination on a 2-ml sample was equiva- lent to 1.8 to 2.0pg of sulphur. The approximate contributions from each were: furnace tube and packing 2/5ths, the hydrogen peroxide, air and water l/5th each of the total. This668 GOODE value for the total blank gives the smallest amount of sulphur that can be determined with any degree of accuracy, which means that for a 2-ml sample the lowest detectable concentra- tion is 1 p.p.m., but an approximate value could be obtained down to 0.25 p.p.m. Analyses performed on a large number of samples of heptane, containing known amounts of sulphur in the form of either carbon disulphide or $-toluene sulphonyl chloride, disclosed that the accuracy of the method varied with the sulphur concentration. For those samples containing 10 p.p.m., the accuracy was better than +5 per cent., but when the concentration was down to 1 p.p.m. the accuracy was no better than +15 per cent. Similarly, the blank values had a spread of 15 per cent. of the average value. This wide range for low-sulphur samples, and for the blanks, was attributed mainly to the random contamination from the laboratory atmosphere, and to a lesser extent to the furnace tube and packing releasing sulphur trioxide impurities in an irregular manner. The author thanks the East Midlands Gas Board for assistance in the preliminary work and for providing the barium-133. 1. 2. 3. 4. 5. 6. 7. 8. 9. REFERENCES Granatelli, L., Analyt. Chem., 1959, 31, 434. Reed, R. H., Analyst, 1963,88, 736. Kriege, 0. H., and Wolfe, A. L., Talanta, 1962, 9, 673. Wickbold, R., Angew.Chem., 1957, 69, 530. Macdonald, A. M. G., Analyst, 1961, 86, 3. Schoberl, Angew. Chem., 1937, 50, 334. Bertolacini, R. J., and Barney, J. E., Analyt. Chem., 1957, 29, 281. Dagnall, R. M., Thompson, K. C., and West, T. S., Analyst, 1967.92, 506. Davis, C. L., L.R.I.C. Dissertation, Derby and District College of Technology, 1966. Received February 6th, 1968