OCTOBER, 1966 THE ANALYST Vol. 91, No. 1087 The Application of the Oxygen-flask Combustion Technique to the Determination of Trace Amounts of Chlorine and Sulphur in Organic Compounds BY R. McGILLIVRAY AND S. C. WOODGER ( I C I Fibres Limited, Hookstone Road, Harrogate, Yorkshire) The Schoniger oxygen-flask technique has been adapted to trace analysis. An electrically fired apparatus has been designed that is capable of com- busting up to 100 mg of material. As little as 0-1 pg of chloride can be titrated potentiometrically with silver nitrate solution. Sulphate is titrated with barium perchlorate with thorin as indicator. Visual and spectrophotometric detection of the end-point has been investi- gated and the latter is preferred for amounts down to 0.5 pg of sulphate. Cations which interfere can be readily removed by ion exchange.IN recent years the oxygen-flask combustion technique has become established for deter- mining chlorine, sulphur, phosphorus and other elements in organic compounds,l but its use for the determination of minor and trace amounts has received scant attention. Lehner2 used the technique to determine chlorine and sulphur in organic compounds at the 0.1 per cent. level, and Haslam and Squirrel13 used it to determine sulphur at a similar level in poly- (methyl methacrylate). It was thought that the technique could be applied to the deter- mination of microgram amounts of these elements in organic compounds and polymers. COMBUSTION OF THE SAMPLE- In trace analysis the reagents used should be free from the element being determined.The paper in which the sample is enclosed is a potential source of contamination. It was found that Whatman No. 42 paper contains about 300 p.p.m. of chlorine and about 150 p.p.m. of sulphur, which limit the sensitivity of the determination. No Whatman papers were found to be free from chlorine. The impurities could be partially washed out with water, but not enough to make the paper a practical proposition. EXPERIMENTAL I-litre flask with 834 neck Securi to hold tightly Heating element and quartz-wool pads Absorption solution Fig. 1. Combustion assembly 61 1612 MCGILLIVRAY AND WOODGER: APPLICATION OF THE [A?Zd$%t, VOl. 91 Other materials were tried. Polythene was not suitable because it tended to melt and drop into the absorption solution, taking with it a little of the sample.“Melinex” film washed with alcohol contained a negligible amount of chlorine and sulphur, but occasionally, molten drops of polymer fell from the basket, as with polythene. Juvet and Chiu4 enclosed the sample in glass-wool and ignited it by means of an electrically heated nickel chrome wire. This firing principle was adopted for the present work with two main alterations. Platinum was used for the heating filament to eliminate possible loss of chlorine and sulphur by reaction with the nickel chrome wire. Quartz-wool pads were used because the heat of combustion melted the glass-wool, which then coated the platinum heating filament. The apparatus shown in Fig. 1 consists of a 1-litre round-bottomed flask fitted with a B34 joint, into which fits the electrically fired combustion head (Fig.2). The current to the 1 -Platinum heating filament atinum wire 18s 9c m v4.g. wire joint Fig. 2. Electrically fired combustion head platinum heating element is supplied from a 25-volt Variac, connected in series to an ammeter. A current of 4 to 5 amps heats the elements to a bright red glow, sufficient to fire the sample. This combustion head is similar in principle to that used by H e m ~ e l . ~ The sample (about 100 mg) is placed on the centre of a fairly tightly compressed quartz- wool pad, approximately 30 mm long, 10 mm wide and 5 mm deep. A loose pad of quartz-wool, thin enough for the heat from the platinum filament to ignite the sample, is placed on top of the sample. The pads are fitted round the heating element, the thinner pad being next to the heating element, and are held in position by a folded platinum wire.A suitable absorption solution is placed in the 1-litre flask, the flask is filled withmedicinal oxygen, purified by passing it through a tube filled with soda asbestos, and the combustion head carrying the sample is inserted into the flask. The joints are secured with metal springs and the apparatus connected as shown in Fig. 1, the flask being inverted so that the absorption solution forms a seal with the joint. A current of 4 to 5 amps is passed through the heating element and is allowed to flow until combustion is complete (usually in 10 to 20 seconds) in order to prevent the formation of soot on the quartz-wool pads. The maximum amounts of sample that have been combusted smoothly and cleanly in the electrically fired apparatus have been about 100 to 150 mg, and it is felt that the technique could be modified to handle larger amounts, probably by using a larger combustion flask and enclosing the sample in sufficient quartz-wool to limit the supply of oxygen and slow down the combustion, making it less violent than when paper is used to wrap the sample.Non-volatile liquids can be added by micrometer syringe to a quartz-wool pad underneath the heating element, the pad then being folded round the heating element and the sample combusted as described above. The use of medicinal oxygen purified by passing through a U-tube filled with soda asbestos is necessary to obtain low “blanks” with the chlorine determinations.The quartz- wool used (type A, supplied by the British Thermal Syndicate Ltd.) contains no chlorideOctober, 19661 OXYGEN-FLASK COMBUSTION TECHNIQUE 613 or sulphate which can be washed out under the conditions used, but it is essential that it should not be handled with bare hands. DETERMINATION OF CHLORINE- For the determination of 1 p.p.m. of chlorine in a 100-mg sample, a method of deter- mining 0.1 pg of chloride was needed. The best method seemed to be the potentiometric argentometric titration carried out in a medium of relatively low dielectric constant. The titration assembly, which can deal with volumes of titration liquid of from 0.2 to 5 ml, consists of a vessel that accommodates a micro air-stirrer, a silver electrode, the tip of a micrometer syringe and one end of a salt bridge.The other end of the salt bridge and a calomel electrode dip into saturated potassium sulphate solution in a beaker. The silver and calomel electrodes are connected to an E.I.L. automatic titrimeter or a suitable potentio- meter. The smallest titration vessel used is 2 ml in capacity. The apparatus was used to determine chloride in the range 0.4 to 4 pg. The chloride was dissolved in 0.5 ml of 50 per cent. v/v aqueous ethanol, and 0.01 N silver nitrate was added in increments of 0.0002 ml from the micrometer syringe. The end-point was obtained by taking the mid-point of the steepest part of the curve of potential plotted against titrant volume. The results summarised below show that the method is satisfactory for the deter- mination of chloride in the range 0.4 to 4 pg, the end-points even with 0.4 and 0.7 pg of chloride being still sharp- Chloride, pg of C1- added .. 4.38 2-73 1.83 0-73 0.36 Chloride, pg of C1- found . . 4-42 2.70 1.88 0.73 0-38 The sensitivity of the method might be further increased by carrying out the titration in a total volume of 0.2m1, but with the present apparatus this is about the minimum amount that can be handled. When the above method was used to determine chloride in a total volume of 5ml of 50 per cent. v/v aqueous ethanol, it was found that below 2.5 pg of chloride the accuracy of the results fell sharply, and it is therefore recommended that the concentration of chloride be at least 0.5 pg per ml in the solution being titrated.METHOD APPARATUS- Combustion apparatus. Micrometer syringe-This is of 0 6 m l capacity, delivering 0.00005 ml as described in British Standard 1428, Part 6: 1955. Ultra-micro potentiometric titration assembly. REAGENTS- Hydrogen peroxide, 1 vol. Sodium hydroxide, 0.1 M. Ethanol, 75” O.P. Nitric acid, 0.1 M. Silver nitrate, 0.01 N. Potassium sulphate - agar solutiorlz-Heat together in a beaker 12 g of potassium sulphate, Oxygefi-Medicinal oxygen purified by passage through a tube filled with soda asbestos. 5 g of agar and 200 ml of water. PROCEDURE- Place 5 ml of hydrogen peroxide and 0-05 ml of sodium hydroxide solution in a 1-litre combustion flask. Accurately weigh 60 to 100mg of sample (see Note 1) on to the centre of a fairly tightly compressed quartz-wool pad.Assemble the combustion apparatus and combust the sample as described in the sub-section on “Combustion of Sample” (see Note 2). Liquids are delivered on to a pad from the micrometer syringe. After the combustion, turn the flask the correct way up, taking care that the liquid flows down the side of the flask and that it does not splash on to the quartz-wool pads. Leave it for 10 minutes to allow the combustion products to be absorbed. Remove the combustion head and wash the joint, adjoining tubes and quartz-wool with “metal-free” water, adding the washings to the 1-litre flask. Transfer the solution to a 50-ml separating funnel and wash614 MCGILLIVRAY AND WOODGER: APPLICATION OF THE [A%@@St, VOl. 91 the flask with water, adding the washings to the separating funnel.Place a suitably sized titration vessel (see Note 3) on a heated steam-bath, and half fill the vessel with liquid from the separating funnel. Allow the solution to evaporate, and, as it does so, add more solution from the funnel at a rate about equivalent to that of the evaporation. When the funnel has been drained of liquid, wash it with a few millilitres of water and add this to the titration flask. Evaporate the solution in the titration flask to the appropriate volume to give a final chloride concentration of at least 0.5 pg per ml and add an equal volume of ethanol. Add 0.1 ml of nitric acid. Assemble the potentiometric titration apparatus, switch on the stirrer and titrate with silver nitrate solution added from a micrometer syringe in increments of 0.0002 ml (see Note 4).Record the potential after each addition of silver nitrate and plot a graph of the silver nitrate added against potential. The mid-point of the steeply rising portion of the graph is the end-point. Carry out a blank combustion and titration by using the same apparatus and reagents as used in the determination, keeping the heating element at bright red heat for 5 minutes. NOTES- 1. If the sample contains more than 100 p.p.m. of chlorine, then a 60 to 70-mg sample is sufficient. If the sample contains less than 100 p.p.m., then it is desirable to take as large a sample as will combust smoothly, but not much more than 100 mg. If the chlorine content is between 1 and 5 p.p.m., a 70 to 100-mg sample can be combusted, the combustion products absorbed, the flask re-filled with oxygen and a further 70 to 100-mg sample combusted. The quartz-wool pad and the wire clip should not be handled with bare hands.Fingerstalls or thin rubber gloves should be worn. 2. The apparatus should be placed in a fume cupboard with the Variac located outside, and the cup- board kept closed during the combustion. 3. The size of the titration vessel and the necessary concentration of the liquid are dependent on the amount of chlorine present in the sample, Chlorine expected, Size of titration Volume of evaporated p.p.m. vessel, ml solution, ml More than 100 30 5 50 to 100 10 2.5 20 to 50 5 1 5 to 20 2 0.25 4. With a silver and a calomel electrode, the end-point is usually between 290 and 310mV. If a t the start the potential reading is below 250 mV, silver nitrate can safely be added until this reading is reached; thereafter, it should be added in increments of 0-0002 ml.Fig. 3. Titration cell for Unicam SP600. All measurements are in inchesOctober, 19661 OXYGEN-FLASK COMBUSTION TECHNIQUE . 615 ANALYSIS OF SAMPLES- The method was tested on samples containing known amounts of chlorine prepared by adding l-chloro-2,4-dinitrobenzene (M.A.S.) to purified benzoic acid that had been prepared from AnalaR material by two zone refinings followed by two crystallisations from water. Appropriate amounts of a solution of l-chloro-2,4-dinitrobenzene were added from a micro- meter syringe to approximately 50-mg samples of purified benzoic acid. The ethanol was allowed to evaporate before the combustion was carried out.Determination of the chlorine in the “purified” benzoic acid gave a titre of silver nitrate not significantly different from that of a blank combustion, viz., 0-0006ml of 0.01 N silver nitrate. Collected below are the results obtained with several samples containing from 5 to 500 p.p.m. of chlorine. The experiments with up to 50 p.p.m. were titrated in a total volume of 0.5 ml of 50 per cent. ethanol, while that with 500 p.p.m. was titrated in a total volume of 5 ml of 50 per cent. ethanol- Chlorine, p.p.m. of C1 addcd . . . . 510 51 12 6 Chlorine, p.p.m. of C1 found . . . . 506 50 11 8 of organic compounds and polymers, and some results are given in Table I. The method has been used to determine the chlorine content of a number of samples TABLE I DETERMINATION O F CHLORINE I N ORGANIC COMPOUNDS Sample Chlorine, p.p.m.of C1 Biphenyl and diphenyl ether . . . . . . 12, 13 Biphenyl and diphenyl ether . . . . . . 71, 71 Biphenyl and diphenyl ether . . .. . . 74, 82 Cyclohexanol and cyclohexanone . . . . 99, 108, 98, 103 Cyclohexanol and cyclohexanone . . . . 313, 326, 327, 340 Poly(ethy1cne terephthalate) . . . . . . 9, 10 DETERMINATION OF SULPHATE- To determine down to 1 p.p.m. of sulphur in a 100-mg sample of organic material, it is necessary to determine down to 0.1 pg of this element. The method of Fritz and Yamamura,6 consisting of the titration of sulphate with barium perchlorate in the presence of thorin - methylene blue mixed indicator, was scaled down. To the sulphate dissolved in 1 ml of water and 4 ml of ethanol, was added 0.05 ml of 0.05 per cent.ethanolic solution of thorin and 0-05 ml of 0.003 per cent. aqueous solution of methylene blue. The sulphate was then titrated with 0.01 N barium perchlorate with the micrometer syringe, the end-point being detected visually by the colour change from pale green to pale pink. The solution was stirred during the titration and the tip of the burette was kept just below the surface of the solution. Typical results of determining 20 to 200 pg of sulphate by this method were- Sulphate, pg of SO,2- added. . . . 216 43-5 23 Sulphate, pg of found . . . . 218, 216 42.0, 41.6 21, 22 These results are satisfactory, but with less than 10 pg of sulphate the end-point was less distinct, and it was necessary to reduce the volume of the titration medium. By reducing the volume of the titration to 0.5 ml, it was possible to titrate 1 to 10 pg of sulphate.A blank determination on the reagents used corresponded to about 0.5 pg of sulphate, which appeared to come from the methylene blue solution. Some results were- Sulphate, p g of SO,2- added. . . . 5.26 2.63 1.3 0.65 Sulphate, pg of found . . . . 5.2 2.7 1.4 0.7 These results show that the method is satisfactory for determining 1 to 500 pg of sulphate. A disadvantage is that the titrations have to be carried out in good daylight or in front of an illuminated screen approximating to daylight. SPECTROPHOTOMETRIC TITRATIONS- Menis, Manning and Ball7 overcame the difficulty of the poor end-point of the visual titration by carrying out the titration spectrophotometrically, with pentanol as solvent.They found methanol, butanol and isopropanol unsuitable as solvents. By titrating in a total volume of about 35 ml of pentanol they were able to determine down to 6 pg of sulphate.61 6 MCGILLIVRAY AND WOODGER: APPLICATION OF THE [A%@&St, VOl. 91 Pentanol is unsuitable for use in conjunction with the oxygen-flask combustion technique because it is preferable for the titration solvent to accommodate about 5 ml of water. For the visual end-point, a medium containing 80 per cent. of ethanol was used and the spectro- photometric titration of sulphate in this medium was investigated. A titration cell was made of glass with two clear glass windows about 4 to 4.5 cm apart. The rest of the vessel was coated with black paint.A vessel suitable for use with a Unicam SP600 is illustrated in Fig. 3. The sulphate was dissolved in 5 ml of water and placed in the titration cell, and 30 ml of ethanol and 1 ml of 0-01 per cent. ethanolic thorin solution were added. The solution was stirred by a stream of nitrogen, 0.001 M barium perchlorate was added in increments and the change in optical density at 520mp after each addition was noted. Optical density was plotted against volume of titrant added, and the end-point obtained from the intersection in the usual way. Some results obtained with 4 to 300 pg of sulphate were- Sulphate, pg of SO,2- added. . .. 298 149 37.3 5.3 3.7 149 37 5.2 3-8 148 36 4-8 3.8 . . . . Sulphate, p g of S0,Z- found { 22:; The minimum amount of solution that can be used in this titration cell is about 25 ml and in order to titrate amounts of sulphate smaller than 4 pg it is necessary to reduce the volume.A second titration cell was constructed 4 cm long and 1 cm wide, so that with 4 ml of titration solution the depth of the solution in the cell was 1 cm. The efficient stirring of this long narrow column of liquid presented some difficulties. Stirring by gas bubbles and a con- ventional rotary stirrer was unsatisfactory. A vacuum-operated vibratory stirrer such as that described by Stock and Fill8 with a paddle about 3-5 cm long and an amplitude of about 0.5 cm was found to be satisfactory. By using this cell it was found possible to titrate as little as 0.4 pg of sulphate with 0.0005 N barium perchlorate in a titration volume of 4 ml.INTERFERENCE DUE TO CATIONS- Cations which give a coloration with thorin interfere. They include the alkali metals, the alkaline earths, iron and cobalt. When 30 pg of calcium as a chloride solution was added to 144 pg of sulphate as sulphuric acid and the mixture titrated with 0-001 N barium perchlorate with a thorin - methylene blue indicator, only 88 per cent. of the sulphate was recovered. The cations could be removed by passing the solution through an ion-exchange resin column, but it is simpler to add resin and remove it when exchange has been completed. Amberlite IR-l20(H) resin was thoroughly washed with water until the washings gave no titre for sulphate when titrated with 0.001 N barium perchlorate. One scoopful (about 1 ml) of resin was added to the solution containing the sulphate and cation, the mixture was shaken well and allowed to stand for 10 minutes. The resin was removed by filtration, and the sulphate in the filtrate was titrated visually with barium perchlorate.TABLE I1 REMOVAL OF CATIONS BY AMBERLITE IR-l20(H) RESIN IN SULPHATE DETERMINATION Amount of cation present, Cation added as chloride mg Aluminium C a 1 c i u m Calcj uni Calcium Iron . . Sodium Strontium Zinc . , .. * . .. 1.1 .. .. .. 0.03 . . .. .. 0.12 .. .. . . 1.2 .. .. .. 0.96 .. .. .. 2.2 . . .. .. 1.0 .. .. .. 1-1 Sulphate, p g of SO,2- * added found 288 283 144 143 144 144 144 142 288 284 288 287 147 147 288 283 The results given in Table I1 show that the resin treatment is effective in removing up to about 1 mg of calcium, zinc, aluminium, iron and sodium.The treatment is probably effective for other cations, provided they are present in amounts of less than 1 mg.October, 19661 OXYGEN-FLASK COMBUSTION TECHNIQUE 617 The technique has proved generally useful for micro work. Samples received for sulphur determinations often contain small amounts of cations, the presence of which is indicated by the formation of a pink colour with the thorin indicator. Addition of Amberlite resin will remove the interfering cation, provided that not too much is present. The resin usually removes most of the thorin, probably as the cation - thorin complex, but more indicator can be added after removal of the resin. INTERFERENCE DUE TO ANIONS- Fritz and Yamamuras examined the effect of chloride, nitrate, fluoride, phosphate and sulphite ions on the barium perchlorate titration of sulphate and found that only phosphate and sulphite seriously interfere.The interference of sulphite can be neglected because the combustion products are absorbed in hydrogen peroxide which will convert any sulphite to sulphate. There is some confusion in the literature as to the extent of the phosphate inter- ference and consequently the effect of milligram amounts of phosphoric acid on the deter- mination of milligram amounts of sulphate was examined. The results shown in Table I11 are high in the presence of phosphate, the error increasing when more phosphate is present. The spectrophotometric titration gave slightly better results than the visual titration. TABLE I11 EFFECT OF PHOSPHATE ON THE DETERMINATION OF SULPHATE Visual titration with 0.01 N barium perchlorate- Sulphate added, mg .. . . 2.06 2-06 2-06 2-06 2.06 Phosphate added, mg . . 0 0.32 0-80 1.6 3-2 Sulphate found, mg . . . . 2.06 2.12 2-16 2.22 2-27 Percentage error . . . . + 3 +5 +8 + 10 Spectrophotometric titration with 0.01 N barium perchlorate- Sulphate added, mg . . . . 1.92 1-92 1.92 1.92 1.92 Phosphate added, mg . . 0 0.32 0.80 1-6 3.2 Sulphate found, mg . . . . 1-92 1.96 1-99 2-05 2.06 Percentage error . , . . +2 +4 +7 +7 For trace analysis, it is the effect of the phosphate on the determination of microgram amounts of sulphate that is important; 20, 52 and 98 pg of sulphate were determined in the presence of 100, 500 and 1000 pg of phosphate, the titrations being carried out spectro- photometrically with 0.001 N barium perchlorate, with thorin as indicator.TABLE I V EFFECT OF PHOSPHATE ON THE DETERMINATION OF MICROGRAM AMOUNTS OF SULPHATE Sulphateadded, pg 98 98 98 98 52 52 52 52 20 20 20 20 0 0 0 Phosphateadded, pg 0 100 500 1000 0 100 500 1000 0 100 500 1000 100 500 1000 Sulphatefound, pg 98 98 101 106 52 52 52 55 20 20 20 20 (1 <1 <1 Percentageerror 0 0 +3 + 8 0 0 0 +6 0 0 0 0 The results in Table IV show that up to 100 pg of sulphate can be determined in the presence of up to 500 pg of phosphate with reasonable accuracy. This means that if a 100-mg sample is combusted, then up to 300 p.p.m. of sulphur can be determined accurately in the presence of up to 0.15 per cent.of phosphorus. If the sample contains more phosphorus than this, then it is better to remove the bulk of it by precipitating it as the magnesium salt and removing it by filtration. Any magnesium that dissolves to form magnesium sulphate is then removed with an ion-exchange resin. This procedure has been described by Fritz and Yamamura,6 who used magnesium carbonate. The use of purified magnesium oxide is preferred because lower blanks can be obtained. The magnesium oxide (AnalaR) is purified by washing it well with hot water and drying. When 5 ml of solution containing 98 pg of sulphate (as sulphuric acid) and 30mg of phosphoric acid were treated with 100mg of magnesium oxide, the sulphate found in four separate spectrophotometric determinations was 93, 97, 101 and 98 pg, The precision is not as good as usual, but the results are satis- factory in view of the number of manipulations.618 MCGILLIVRAY AND WOODGER: APPLICATION OF THE [AfzaZyst, Vol.91 METHOD APPARAT u S- Combustion apparatus. Titration cells-(a) A cell for titrations in the range 25 to 35ml consisting of a black- painted glass vessel with two clear windows 4 to 4-5 cm apart, suitable for use with a Unicam SP600 is shown in Fig. 3. I t is provided with a top cover made from black Perspex, with holes for the burette and for a capillary tube, through which nitrogen is passed for stirring the solution. ( b ) A cell for titrations in the range 4 to 8 ml, similar to that described above but having a smaller volume between the two windows. Vibratory stirrer-This is vacuum-operated with a paddle about 3.5 cm long and a stroke of about 0-5cm.REAGEXTS- Water-All water must be “metal-free.” Hydrogen peroxide, I vol. Ethanol, 75” O.P. TJtorin solution, ethafzolic-Dissolve 25 mg of thorin in 5 ml of water and dilute to 50 ml Barium perchlorate, 0.001 and 0.0005 N-Prepare by dilution of 0.01 N barium perchlorate iwedicinal oxygen-Purify by passing through a U-tube filled with soda asbestos. Amberlite IR-l20(H) resin-Wash well with “metal-free” water before use and store with ethanol. and store in a polythene bottle. under water in a flask. PROCEDURE- Place 5 ml of hydrogen peroxide in the 1-litre combustion flask and weigh out accurately a 60 to 70-mg sample (see Note 5). Enclose the sample in quartz-wool and combust it as described in the chlorine determination, If the sample contains more than 50 p.p.m.proceed as follows- After the combustion, turn the flask the correct way up, taking care that the liquid flows down the side of the flask and that it does not splash on to the quartz-wool pads. Leave it for 10 minutes to allow the combustion products to be absorbed. Carefully remove the combustion head and wash the joint, the adjoining tubes and the quartz-wool with ethanol, adding the washings to the 1-litre flask (see Note 6). Carefully transfer this solution to the titration cell (a) and wash the flask carefully with ethanol, adding the washings to the titration cell; a total volume of 30 ml of ethanol is used for washing. Add 2 ml of thorin solution. Place the titration cell in a holder and set it in the Unicam so that the light passes through the solution between the two clear windows.Bubble nitrogen through the capillary tube to stir the solution. Stop the nitrogen flow and set the wavelength to 520 mp. Set the optical density to 0.300 and adjust the slit control so that the galvanometer pointer is at zero. Add 0.001 N barium perchlorate in increments of 0-04 ml. Stir after each addition by bubbling nitrogen through, and with the nitrogen supply off, read the optical density. Titrate until the optical density rises sharply and then continue for at least a further 0.2 ml. Plot optical density against volume of titrant added. The intersection of the base-line with the projection of the line forming the steeply rising portion of the curve gives the end-point.For this, fill the combustion flask with oxygen and keep the heating element at bright red heat for 5 minutes; use the same volumes of reagents as in the test and add the titrant in 0.02-ml increments. NOTES- Larger samples should be sub-divided, the flask re-filled with oxygen and the combustion repeated. The interference of calcium, aluminium, copper, iron, sodium and zinc can be overcome by removing the cations with hmberlite IK-l20(H) resin. To the liquor in the flask, add about 1 ml of Amberlite IR-lSO(H) resin, shake well and allow to stand for 10 minutes. Filter through a sintered-glass crucible (porosity 3), collecting the filtrale in the titration cell. Wash the combustion flask and crucible with 30 ml of ethanol, adding the washings to the filtrate.Carry out a blank combustion and titration. 5. The weight of sample taken should not appreciably exceed 100 mg. 6. Certain cations interfere and cause poor end-points and low results. Then add 2 ml of thorin solution and proceed with the titration.October, 19661 OXYGEN-FLASK COMBUSTION TECHNIQUE 619 If the sample contains less than 50 p.p.m. proceed as above to the stage where the combustion products are absorbed, then carefully remove the combustion head and wash the joint, the adjoining tubes and the quartz-wool with the minimum volume of ethanol, adding the washings to the l-litre flask. Carefully transfer this solution to a clean beaker and evaporate on a steam-bath until the volume is 1 ml. Transfer this solution to the smaller titration cell ( b ) and wash the beaker with a total of 4 ml of ethanol, adding the washings to the titration cell.Add 0.5 ml of thorin solution. Titrate as above with 0.0005 N barium perchlorate. ANALYSIS OF SAMPLES- The method was tested by preparing samples containing known amounts of sulphur by adding known volumes of an ethanolic solution of sulphonal (M.A.S.) to purified benzoic acid. The purified benzoic acid was found to contain less than 5 p.p.m. of sulphur. The following results by the spectrophotometric method agree well with the amounts added. By the visual titration method, a sample containing 10 p.p.m. of added sulphur gave a value of 9 p.p.m. of sulphur- Sulphur, p.p.m. of S addcd . . 999 350 56 Sulphur, p.p.m. of S found . . 996 346 58 The methods have been used to determine sulphur in a variety of compounds and poly- mers, and the results obtained with a selection of the products analysed are collected in Table V.In some of the samples the sulphur is present as an additive and in others as an impurity. TABLE V DETERMINATION OF SULPHUR IN ORGANIC COMPOUNDS AND POLYMERS Sample Sulphur found, p.p.m of S Terephthalic acid . . .. . . . . 20,lS Dimethyl terephthalate . . .. . . 47, 53 Dimethyl terephthalate . . .. .. 6, 8 Dimethyl terephthalate . . .. .. 3, 3 . . . . 5, 6 p-Xylene . . . . . . Tcrephthalic acid . . . . . . .. 78, 76 Dimethyl terephthalate . . .. . . 85, 80 Polystyrene . . . . . . .. .. 273, 275 Polystyrene . . .. . . . . . . 242, 233 Polypropylene . . . . . . . . 55, 62 Polypropylene . .. . .. . . 690, 700 Polypropylcne . . . . . . . . 320, 310 Poly(ethy1ene tcrephthalatc) . . .. 10,12 DISCUSSION The methods described have been in use for several years and have proved satisfactory. The electrically fired apparatus has been used to combust such widely different materials as 9-xylene, a fairly volatile liquid; dimethyl terephthalate, a fairly easily sublimable sub- stance ; polypropylene, an easily combustible polymer ; and poly(ethy1ene terephthalate) , a more difficultly combustible material. Doubts have been expressed about the efficiency of the combustion of materials that have been fired with a paper wick or by means of an electrically heated ~0il.1~9 The main objections are that the sample either melts, volatilises or sublimes before the combustion becomes established, or, with the heated coil, that a carbonaceous residue is left on the quartz-wool enclosing the sample. In the apparatus used, the heating element reaches its operating temperature almost immediately the current is switched on, and then the sample is fired within a second or two. The amount of volatilisation or sublimation of a 50 to 100-mg sample that takes place in this time must be small, and, if any does occur, it is more than likely that it condenses on the quartz-wool enclosing the sample. If the heating coil is kept at bright red heat during the combustion, no soot forms on the outside of the quartz-wool pad. Obviously not all compounds combust in the same way, but some measure of control can be gained by varying the thickness of the quartz-wool surrounding the sample. In this way the supply of oxygen to the sample can be varied.620 MCGILLIVRAY AND WOODGER These titration methods have been applied to the solutions obtained after combusting samples in oxygen to determine trace and minor amounts of chlorine and sulphur in semi- micro amounts of organic material. They have also been used for the ultra-micro deter- mination of major amounts in ultra-micro amounts of sample. 1. 2. 3. 4. 5. 6. 7. 8. 9. REFERENCES McDonald, A. M. G., Analyst, 1961, 86, 3. Lehner, H., Chimia, 1959, 13, 248. Haslam, J., and Squirrell, D. C. M., J . Appl. Chem., 1961, 11, 244. Juvet, R. S., and Chiu, J., Analyt. Chem., 1960, 32, 130. Hempel, W., Angew. Chem., 1892, 5, 393. Fritz, J. S., and Yamamura, S. S., Analyt. Chem., 1955, 27, 1461. Menis, O., Manning, D. L., and Ball, R. G., Ibid., 1958, 30, 1772. Stock, J. T., and Fill, M. A., Analyst, 1944, 69, 212. Tomlinson, C., Talanta, 1962, 9, 1065. Received July 22nd, 1965