512 KAKABADSE AND MAXOHIN RAPID MICRO-DETERMINATION Rapid Micro-determination of Nitrogen [Vol. 86 in Fluorine-containing Compounds BY G. KAKABADSE AND B. MANOHIN (Department of Chemistry, College of Science and Technology, Mamhester 1) A modified Dumas combustion train for the rapid determination of nitrogen in fluorine-containing compounds and those difficult to combust is described. The tube does not contain a temporary filling, and hydrogen peroxide is used as an intermittent source of oxygen. THE conventional Dumas method for determining nitrogen in organic compounds1 requires considerable modification when analysing compounds that contain fluorine2 or are difficult to burn, leaving nitrogenous chars,, or when attempting to combust rapidly.4 With fluorine- containing compounds, tetraflu~romethane~ often appears to cause the difficulty, and it is possible that CF, radicals formed by initial C-C cleavage have time to react with each other instead of with copper oxide, so forming hexafluoroethane. This suggests that the normal pyrolysis temperature of the Dumas train, 600" to 700" C, is too low.The formation of fluorocarbons, which are not absorbed by the potassium hydroxide solution in the nitrometer, will cause high results for nitrogen. According to Sidgwick,g the decomposition of hexa- fluoroethane vapour begins at temperatures in excess of 800" C, so that the first requirement when dealing with fluorine-containing compounds would be appreciably to raise the tem- perature of the ordinary Dumas train.' A satisfactory technique has been developed in the Microanalytical Laboratory at Cambridge (personal communication from Professor R.N. Haszeldine) in which are used a gas-heated combustion train and a silica tube about 70 cm long containing a permanent filling of copper oxide and copper and, in the "beak" end, a layer of sodium fluoride for absorbing silicon tetrafluoride; the sample is contained in a platinum boat. The replacement of the copper oxide - copper mixture with nickel oxide permitted Kirsten8 to use a much higher pyrolysis temperature, 1050" C; although this method and itsAugust, 19611 OF NITROGEN I N FLUORINE-CONTAINING COMPOUNDS 513 modification by Belcher and Macdonald9 proved to be successful for determining nitrogen in fluorine-containing compounds and those difficult to combust, its disadvantage is high wear of the silica tube, nickel oxide apparently catalysing the crystallisation of silica.For substances that are difficult to burn and leave nitrogenous chars, so causing low results far nitrogen, many remedies have been suggested, based on a combined pyrolytic and oxidative attackl0?l1 as well as on Kirsten's method. Special reference must be made to techniques devised by Unterzaucher,12 involving oxygen generated by the catalytic decom- position of hydrogen peroxide, Swift and M ~ r t o n , ~ who used oxygen from a cylinder, and Cropper, Reed and Rothwell,13 who generated oxygen electrolytically in small amounts at a known rate. Finally, rapid methods for determining nitrogen based on modifications of the Dumas method have been advocated by many investigators, these modifications involving, for example, increase in temperature14 and in length of high temperature combined with oxygen injection16y17 and the use of a pre-combustion techniquels and a specially designed nitrometer.19 By increasing the length of the tube to 100 cm and using two high- and one medium- temperature electrically heated furnaces for that part of the tube filled with copper oxide and copper and by inserting hydrogen peroxide as an intermittent source of oxygen in the path of the carbon dioxide, we have been able to combust the classes of compounds mentioned above quantitatively and rapidly.The procedure described in this paper is recommended after much investigation. COMBUSTION TUBE AND HEATING UNITS- is equipped with a side-arm a t the rear end (mouth) for admission of carbon dioxide.DESCRIPTION OF APPARATUS The quartz tube (see Fig. 1) is approximately 100 cm long and 11 mm in diameter and The Silver wool -- -Carbon dioxide in Fig. 1. Combustion tube and heating units rear portion of the tube (25 cm) is empty and serves for the insertion of the sample in a platinum boat; it is heated by a movable furnace of the split type. The remaining portion of the tube is filled permanently and is heated by furnaces A, B, C and D, the first three being close to each other. The lengths of the furnaces and the sequences of the layers surrounded by them are: A, 16 cm long, filled with equal lengths of M.A.R., powdered cupric oxide, copper wire prepared by reducing M.A.R.cupric oxide wire with hydrogen, and cupric oxide powder; B, 28 cm long, filled with 18 cm of cupric oxide wire and 8 cm of copper wire; C, 10 cm long, filled with cupric oxide wire; D, 9 cm long, filled with 10- to 14-mesh granules of sodium fluoride puri~s.~o The gap between C and D, which is at least 4 cm (although there was no detrimental effect on the results when it was increased up to 15 cm), is filled with cupric oxide wire. The "beak" of the combustion tube is packed with M.A.R. silver wool. All layers are held in place by plugs of silica wool, the sodium fluoride layer being held rather loosely. The temperatures of furnaces A, B, C and D are adjusted to 750" (or between 750" and S O O O ) , 850", 600" and 180" C, respectively. When analysing compounds not containing fluorine and those easy to combust, furnace B can be used at 700" C.Furnace A is of the split type; B, C and D are tubular. OXYGEN GENERATOR- A conical 75-ml suction flask (see Fig. 2) is placed in an inclined position between the bubbler (adjoining the carbon dioxide generator) and the combustion tube, and carbon dioxide passes over the surface of 50-volume hydrogen peroxide (M.A.R.). A piece of 60-mesh silver gauze suspended from a platinum hook at the end of a bent glass rod can be made to dip into or merely touch the surface of the hydrogen peroxide by turning the rod. Interruption514 KAKABADSE AND MANOHIN RAPID MICRO-DETERMINATION [Vol. 86 of the contact causes almost instant stoppage of the flow of oxygen. The supply of oxygen is regulated visually, i.e., by observing the disappearance of the nitrogenous char in the combustion tube.Oxygen is produced intermittently, rather than continuously, to avoid excess of oxygen in the combustion train. NITROMETER- We found that the introduction of a few milligrams of black selenium powder into the 50 per cent. solution of potassium hydroxide prevents bubbles of gas from adhering to the surface of the mercury and breaks down foam at the potassium hydroxide meniscus; selenium partly dissolves in potassium hydroxide, forming a dark-reddish solution. Also effective for this purpose is Pornatti’s method, which involves agitation of a steel needle inside the nitrometer by means of a magnet.21 The addition of mercuric oxide also prevents bubbles of gas from adhering to the surface of the mercury.An ordinary semi-micro nitrometer has been used throughout. CARBON DIOXIDE GENERATOR- The source of carbon dioxide is a Tucker generator,22 which is connected to a bubbler containing a saturated solution of potassium carbonate to retain any drops of liquid carried over from the generator. The flow of gas is adjusted with a precision screw-clamp between tube Carbon dioxide in \ \ Hydrogen peroxide Fig. 2. Oxygen generator the bubbler and the oxygen generator. The bubbler also serves as a useful detector of leakage ; when the nitrometer is isolated by rotating the tail stopcock and the precision screw-clamp is fully opened, the appearance of bubbles would indicate leakage. METHOD PROCEDUFE- The mouth of the combustion tube is closed by a rubber stopper, and air is swept out with a fast stream of carbon dioxide, the gas escaping through the tail stopcock into the atmosphere, passing first through a small volume of water in a conical flask; a short piece of rubber tubing attached to the tail stopcock dips into the water. This simple arrangement acts as a hydraulic valve and prevents air from diffusing into the combustion tube.(After a determination, it is advisable to remove the conical flask and to close the stopcock, so as to prevent any water from being sucked into the tube as it cools.) The furnaces are switched on and allowed to attain the required temperatures. When small bubbles can be observed in the nitrometer, the platinum boat, containing between 10 and 20 mg of sample, is inserted in the combustion tube 5 to 10 cm from furnace A, with the stream of carbon dioxide still passing through the apparatus.The mouth of the combustion tube is closed, and carbon dioxide is passed until small bubbles again appear in the nitrometer. The flow of gas is then adjusted so that one or two bubbles rise in the nitrometer per second. The movable furnace is then pulled into position round the combustion tube. In the presence of nitrogenous chars, the rate of combustion when the oxygen-injection technique is used is governed by the rate of disappearance of the char; as oxygen is generated,August, 19611 OF NITROGEN I N FLUORINE-CONTAINING COMPOUNDS 515 the flow of carbon dioxide is appreciably slowed down (one or two bubbles per 2 seconds). When all the char has disappeared, generation of oxygen is stopped at once, and the slow stream of carbon dioxide is maintained for 3 to 5 minutes to ensure complete absorption of the excess of oxygen by metallic copper.At the same time, the movable heater is brought back to its starting position, and the empty part of the tube is re-traversed with it. After this, the tube is swept out with a fast stream (approximately 9 ml per minute) of carbon dioxide for 1 to 2 minutes, and the rate of flow is then decreased. When small bubbles appear once more, the movable heater is switched off and pushed clear of the combustion tube. With a 10-mg sample of p-nitroaniline (calculated nitrogen content 20.28 per cent.) the entire operation took 15 minutes (nitrogen content found 20.1 and 20.2 per cent.).Some typical results are shown in Table I. The nitrometer is isolated, and, after 5 minutes, the volume of gas in it is read. TABLE I RESULTS AFTER COMBUSTION BY PROPOSED PROCEDURE Unless otherwise stated under "Remarks," the temperatures of furnaces A, C and D were maintained at 750", 600" and 180" C, respectively Experi- No. ment Sample Trifluoroacetanilide (M.A. R.) Approximately (1 + 1) mixture of Teflon and Fn-dinitrobenzene (M.A.R.) 10 $-Nitroaniline (M.A.S.) 11 Sulphanilic acid (A.R.) 12 8-Hydroxyquinoline (M.A.S.) Tempera- Nitrogen Nitrogen ture of content content furnace B, found, calculated, Remarks "C 700 860 600 850 700 800 850 700 10.50 7.41 20.29 20.28 9-59 9.65 { 8.11 8-09) Slow combustion Rapid combustion Furnace C at 800" C Oxygen injected; rapid combustion No oxygen injected; slow combustion No oxygen injected; boat placed in close contact with permanent CuO filling Oxygen injected ; rapid combustion Oxygen injected ; very slow combustion I BLANK TEST- A check was carried out under rather artificial conditions by passing a very rapid stream of carbon dioxide through the combustion train for 30 minutes and keeping the temperature of furnace B at 850" C.With the normal Dumas temperature setting (furnace B at 700" C), the blank was hardly perceptible, even with a magnifying lens, i.e., it was practically zero. DISCUSSION OF THE METHOD A blank of less than 0.02 ml was obtained in the nitrometer. According to Clark and R e e ~ , ~ , ~ ~ the ordinary Dumas micro-determination of nitrogen can be used without modification for fluorine-containing compounds.In our view, an increase in the temperature of combustion over that ordinarily used is imperative. Obviously, the temperature requirements will vary somewhat with the nature of the compound, e.g., whether it is partly or completely fluorinated' ; for example, we found that heptafluorobutyramide was more resistant to pyrolytic attack than was trifluoroacetanilide. As a general rule, we established that the lower limit of temperature for complete combustion was about 800°C; the upper limit was 900" C or less, being determined by the extent to which the copper oxide attacked the silica tube.l* Not infrequently, this produced cracks in the tube, and we therefore recommend that the temperature of furnace B be maintained at 850°C.First, the pre-combustion zone (heated by furnace A), containing powdered copper oxide to assist combustion. Here, most substances bum quantitatively when combusted slowly, but fluorine-containing compounds There are three combustion zones in our apparatus.516 KAKABADSE AND MANOHIN : RAPID MICRO-DETERMINATION [Vol. 86 tend to behave differently (see experiment No. 3, in Table I). The temperature setting of A is not altered throughout. Second, the “finishing” zone (heated by furnace B), the function of which is two-fold; it completes the combustion of “awkward” compounds (with B at the high-temperature setting) and helps to carry out combustions rapidly for all compounds (with B at the high- or middle-temperature setting). Third, the “eliminating” zone (heated by furnace C), which helps to restore the carbon dioxide balance prevailing in the normal Dumas method,l6J7 so eliminating carbon monoxide24 to 29 produced in the hot “finishing” zone.30 Experiment 6 (see Table I) emphasises the well known fact that under high- temperature conditions the Dumas method is not very satisfactory.6 As furnaces A and B differ relatively little from each other functionally they could conceivably be amalgamated into one large high-temperature heater.We found it more practicable to have them separate, especially as A is of the split type and hence easy to inspect visually. The introduction of oxygen into the Dumas train has, in our experience, no effect on C-F cleavage.31 The presence of hydrogen, however, according to Milton and Waters:e seems to be necessary for the satisfactory analysis of fluorine-containing compounds. Pre- sumably, fluorine formed by initial C-F cleavage combines with hydrogen to form hydrogen fluoride, which, with silica, gives silicon tetrafluoride and water vapour.Only traces of moisture, always present in the Dumas system, would therefore be required to initiate this chain reaction. If it reaches the nitro- meter undecomposed, it will be quantitatively absorbed by the potassium hydroxide solution, but we found that it might be partly hydrolysed by traces of moisture, resulting in deposition of silica near or in the stopcock and so blocking the flow of gas. This is effectively prevented by a layer of sodium fluoride% in the “beak” end of the combustion tube (maintained at 180” C by furnace D).After having carried out determinations of nitrogen both with and without this sodium fluoride, we strongly recommend its use in the routine analysis of fluorine- containing compounds. For substances difficult to combust, giving nitrogenous chars, the addition of controlled amounts of oxygen to the cupric oxide - copper-filled tube is, in our experience, satisfactory. Our oxygen generator differs from that proposed by Unterzaucher12 in that carbon dioxide does not bubble through a solution of hydrogen peroxide; the flow of oxygen is therefore independent of the flow of carbon dioxide, and oxygen is produced intermittently instead of continuously. Further, we use silver instead of platinum, the 60-mesh silver gauze being relatively cheap and readily available. The catalytic activity of silver is similar to that of platinum; the rate of decomposition of hydrogen peroxide on silver is lo7 times that on an inert material such as polythene.a Experiment No.9 in Table I shows a rather interesting observation: when no external oxygen is available, errors due to the formation of nitrogenous chars can often be decreased by placing the platinum boat containing the sample close to the cupric oxide powder of the permanent filling. This technique, although not a complete remedy, is significantly simple. The achieved increase in the rate of combustion we attribute mainly to increases in length of tube and temperature, both of which favour more thorough combustion ; the greatest danger in any rapid method is the occurrence of incompletely burnt gaseous products of decomposition in the nitr0meter.l’ The active life of the combustion tube in our apparatus is about sixty determinations, with an average weight of sample of 15mg.This raises the problem of the disposal of silicon tetrafluoride. We thank Professor R. N. Haszeldine for the communication referred to on p. 512 and Mr. B. Woodbridge for carrying out various timing operations. REFERENCES 1. 2. 3. 4. 5. Macdonald, A. M. G., I n d . Chem., 1957, 33, 360. 6. Belcher, R., and Godbert, A. L., “Semi-micro Quantitative Organic Analysis,” Longmans, Green Clark, S. J., “Quantitative Methods of Organic Microanalysis,” Butterworths Scientific Publica- Swift, H., and Morton, E. S., Analyst, 1952, 77, 392.Colson, A. F., Ibid., 1950, 75, 264. Sidgwick, N. V., “The Chemical Elements and Their Compounds,” The Clarendon Press, Oxford, and Co. Ltd., London, 1946, p. 72. tions, London, 1956, p, 87. 1950, p. 1128.August, 196 11 OF NITROGEN I N FLUORINE-CONTAIKING COMPOUNDS 517 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. Rush, C. A., Cruikshank, S. S., and Rhodes, E. J. H., Mikrochim. Acta, 1956, 858. Kirsten, W., Anal. Chem., 1947, 19, 925. Belcher, R., and Macdonald, A. M. G., Mikrochim. Acta, 1956, 1111. Belcher, R., and Godbert, A. L., 09. cit., p. 84. Spies, J. R., and Harries, T. H., Ind. Eng. Chem., Anal. Ed., 1937, 9, 304. Unterzaucher, J., Mikrochem. Mikrochim. Ada, 1951, 36/37, 706. Cropper, F. R., Reed, R. H., and Rothwell, R., Mikrochim. Acta, 1954, 223. Levy, R., and Cousin, B., Ibid., 1960, 854. Eder, K., Ibid., 1959, 631. Ingram, G., Ibid., 1953, 131. Trutnovsky, H., Ibid., 1960, 157. Schliniger, W., Mikrochem. Mikrochim. Acta, 1952, 39, 229. Gustin, G. M., Mikrochim. Acta, 1958, 581. Milton, R. F., and Waters, W. A., “Methods of Quantitative Micro-analysis,” Second Edition Pomatti, R., Ind. Eng. Chem., Anal. Ed., 1946, 18, 63. Clark, S. J., op. cit., p. 80. Clark, H. S., and Rees, 0. W., Illinois State Geological Survey, Report of Investigation No. 109, Pagel, H. A., and Oita, I. J., Anal. Chem., 1952, 24, 756. Kao, S., and Woodland, W. C., Mikrochern. Mikrochim. Acta, 1951, 36/37, 309. Kirsten, W., Ibid., 1953, 40, 121. Hozumi, K., and Amako, S., Mikrochim. Ada, 1959, 230. Gore, T. S., and Kulkarni, A. S., Ibid., 1960, 558. Belcher, R., and Godbert, A. L., op. cit., p. 82. Charlton, F. E., Analyst, 1957, 82, 643. Sidgwick, N. V., op. cit., p. 1126. Milton, R. F., and Waters, W. A., op. cit., p. 72. Belcher, R., and Goulden, R., Mikrochem. Mikrochim. Acta, 1951, 36/37, 679. “Hydrogen Peroxide Data Manual,” Laporte Chemicals Ltd., Luton, Beds., 1960, p. 5. Edward Arnold (Publishers) Ltd., London, 1955, p. 71. 1954. Received Novenzbev 15th, 1960