|
1. |
Proceedings of the Society for Analytical Chemistry |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 743-744
Preview
|
PDF (170KB)
|
|
摘要:
OCTOBER, I963 THE ANALYST Vol. 88, No. 1051 PROCEEDINGS OF THE SOCIETY FOR ANALYTICAL CHEMISTRY DEATH Edmund Frankel. WE record with regret the death of SCOTTISH SECTION A JOINT Meeting of the Scottish Section with the Institute of Chemistry of Ireland was held on Thursday and Friday, September 5th and 6th, 1963, in the Rupert Guinness Hall Dublin. The Company was welcomed by Mr. M. J. Cranley, M.Sc., F.R.I.C., F.I.C.I., President of the Institute of Chemistry of Ireland. The meeting was divided into four sessions, the chairmen being: Dr. A. K. Mills, F.I.C.I. ; Mr. A. F. Williams, B.Sc., F.R.I.C. ; Mr. J. K. McLellan, M.A., B.Sc., A.R.I.C., Vice-chairman of the Scottish Section; and Dr. L. Brealey, B.Sc., F.R.I.C. An apology for absence from Dr. D. C. Garratt, F.R.I.C., President of the Society for Analytical Chemistry, was conveyed to the meeting by Mr.McLellan. The subject of the meeting was “Modern Aspects of Chromatography” and the following papers were presented and discussed : “Thin-layer Chromatography of Long-chain Amines,” by E. S. Lane, B.Sc., Ph.D., F.R.I.C. ; “Thin-layer Chromatography and its Applications to Dyestuffs and Plasticisers,” by G. R. Jamieson, B.Sc., F.R.I.C. ; “Thin-layer Chromato- graphy-Applications to Steroid Chemistry,” by P. Oxley, M.A., B.Sc., A.R.I.C. ; “Some Applications of Gel Filtration and Ion-exchange Chromatography to the Fractionation of Polysaccharides,” by D. M. W. Anderson, B.Sc., Ph.D., F.R.I.C. ; “Ion-exchange in the Study of Metal Complexes in Solution,’’ by J. K. Foreman, BSc., A.R.I.C. ; “Chromatography in the Identification of Plant Products,’’ by D.M. Donnelly, B.Sc., Ph.D., A.R.I.C.; “Some Recent Applications of Paper Chromatography in the Petroleum Industry,” by R. B. Delves, A.R.I.C. ; “Recent Applications of Gas Chromatography,” by B. A. Rose, B.Sc., Ph.D., A.R.C.S., D.I.C., A.R.I.C. ; “A Statistical Evaluation of Gas - Liquid Partition Chromato- graphy as a Method of Quantitative Analysis,” by C. E. Roland Jones and D. Kinsler; “Determination of Nitro Constituents of Explosives by Gas Chromatography,” by A. F. Williams, B.Sc., F.R.I.C. ; “The Gas-chromatographic Analysis of Beer,’’ by G. A. F. Harrison, B.A., F.R.I.C., F.I.C.I. ; “The Chromatographic Nehaviour of Some Nitrophenols on Alumina Impregnated Paper,’’ by L. S. Bark, B.Sc., F.R.I.C., and R.J. T. Graham, M.Sc.; “The Chromatography of $-Substituted 9-Hydroxyazobenzenes on Alumina-impregnated Papers, by R. J. T. Graham, M.Sc., and C. W. Stone, M.Sc., Ph.D., A.R.I.C. Morning coffee, afternoon tea and lunch were provided each day by Messrs. A. Guinness, Son & Co. (Dublin) Ltd., to whose generous hospitality much of the success of the meeting is due. On the Thursday evening a bus tour to Glendalough was marred by rain, but nevertheless enjoyed by all who went. The informal dinner on the Friday evening was well attended: the Chair was taken by Professor C. L. Wilson. 7437 44 PROCEEDINGS [AIzalyst, Vol. 88 MIDLANDS SECTION AN Ordinary Meeting of the Section was held at 3.30 p.m. on Wednesday, September 18th, 1983, in the Chemistry Department, The Universitv, Edgbaston, Birmingham 15.The Chair was taken by the Chairman of the Section, Mr. W. H. Stephenson, F.P.S., D.B.A., F. Xi. I. C . The theme of the meeting was “Contributions to Analytical Chemistry by Younger Members of the Profession” and the following papers were presented and discussed: “Develop- ments in Periodxte Oxidation of Organic Compounds,” by G. Dryhurst; “Some Aspects of Cathode-raj? Polarography,” by M. L. Kichardson, A.R.I.C., A.C.T. ; “The Gas-chromato- graphic Determination of Impurities in Dichloran,” by J. R. Ellis; “Stereoisomerism of vic-Diosinies,” by S. Thompson ; “Phase-solubility Analysis,” by R. E. King, A.R.I.C. ; “The Friedel - Crafts Acylation of Cycloalltenes,” by E. J. Rudd, A.R.I.C. ; “Reduction Methods for the Determination of Trace Quantities of Sulphate,” by D.B. Adams, MA., B.Sc., A.R.I.C. ; “Proposals for Modifications to the British Standards Dealing with the Analysis of Raw Copper,” by K. H. Denmead. MICKOCHEMISTKY GROUP A SUMMER Meeting of the Group was held on Wednesday, Thursday and Friday, July 17th, 18th and 19th, 1963, at the School of Pharmacy, Rrunswick Square, London, W.C.1. The following papers were presented and discussed : “Future Possibilities in Micro- analysis,” by Dr. A. J. P. Martin, F.K.S. ; “Methods of Organic Microanalysis: a Comparative and Critical Study,” by Dr. K. L k y ; “Mass Spectrometry and Microanalysis,” by A. Quayle, M.Sc., F.R.I.C.; “Infraredand Microchemistry,” by D. M. W. Anderson, BSc., Ph.D., F.K.1.C; “Some Recent Developments in Functional-group Analysis on the Micro Scale,” by Professor S. Veibel; “The Status of Microgram Analysis,’’ by Professor R. Belcher, Ph.D., D.Sc., F.R.I.C., F.1nst.F. (For a fuller report of this meeting and summaries of the papers, see pages 745 to 750.) There were also two discussion meetings. ATOMIC ABSORPTION SPECTROSCOPY DISCUSSION PANEL THE fourth meeting of the Atomic Absorption Spectroscopy Discussion Panel of the Physical Methods Group was held a t 6.40 p.m. on Tuesday, September loth, 1963, in the Meeting Room of the Chemical Society, Kurlington House, London, W.1. The Chair was taken by the Chairman of the Panel, Mr. W. T. Elwell, F.R.I.C. The discussion was initiated by Dr. J. €3. Willis.
ISSN:0003-2654
DOI:10.1039/AN9638800743
出版商:RSC
年代:1963
数据来源: RSC
|
2. |
Joint A.B.C.M.-S.A.C. Committee on Methods for the Analysis of Trade Effluents |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 744-744
Preview
|
PDF (45KB)
|
|
摘要:
7 44 PROCEEDINGS [AIzalyst, Vol. 88 JOINT A.B.C.M.-S.A.C. COMMITTEE ON METHODS FOR THE ANALYSIS OF TRADE EFFLUENTS A JOINT Committee of the Association of British Chemical Manufacturers and the Society for Analytical Chemistry on Methods for the Analysis of Trade Effluents has been set up under the Chairmanship of Dr. S. G. Burgess. The other members of the Committee are: (representing the Association of British Chemical Manufacturers) Mr. F. G. Broughall, Mr. L. A. Keelan, Mr. J. G. Maltby, Dr. I. S. Wilson; (representing the Society for Analytical Chemistry) Mr. G. E. Eden, Dr. S. H. Jenkins, Mr. J. G. Sherratt and Mr. N. T. Wilkinson. The Committee has begun work on reviewing “Recommended Methods for the Analysis of Trade Effluents,” published in 1958, with the intention of revision of those methods in need of it. Suggestions for revision from any user of the methods would be welcome, and would assist the Committee by widening the experience on which it can draw; they should be sent to Mr. P. W. Shallis, Secretary of the Analytical Methods Committee, 14 Belgrave Square, London, S. W. 1.
ISSN:0003-2654
DOI:10.1039/AN9638800744
出版商:RSC
年代:1963
数据来源: RSC
|
3. |
Future possibilities in microanalysis |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 745-746
A. J. P. Martin,
Preview
|
PDF (101KB)
|
|
摘要:
October, 19631 PROCEEDINGS Future Possibilities in Microanalysis BY A. J. P. MARTIN DR. MARTIN suggested that future developments in analytical chemistry must be directed towards reducing the scale of the operations. The processes of chemistry were seldom influenced by the quantity of material, and 1000 molecules could react chemically with the same efficiency as Man’s ability to observe and measure much smaller quantities was limited by the size of his hands, and the apparatus he could manipulate. The advantages of working on a much smaller scale included speed and economy, the latter being essential in many fields of research work. In gas chromatography and ionisation methods the analyst already has tools sufficiently sensitive to work with 10-9g of material, but he needed first of all a balance capable of weighing such quantities.The present method of applying 1 pg to a capillary column was to weigh 1 mg and by a divider throw the rest away. There should be no difficulty in constructing a suitable balance employing similar principles to the ones used at present, but it must be much smaller. To do this it was essential to overcome the problem of the size of human hands. Dr. Martin believed he now saw how this should be tackled. It was necessary to make a micromanipulator to replace the hands. I t would be crude in comparison with a real hand, but given sufficient degrees of freedom, and fingers for gripping, should not be too difficult to operate. A mechanical system for movement and a pneumatic system for the sensory part would probably be the easiest to make and the most useful to operate.It was impossible completely to reproduce the senses of the hands, but sufficient could be reproduced to begin making one-tenth and one-hundredth scale apparatus. It was interesting to compare this concept of a micromanipulator with previous ones. Although man had used the microscope for 300 years, the only tools available for making manipulations under it were “stone-age” implements-a “spear” and a “battle-axe,” which had only three or four degrees of freedom. (The proposed manipulator would have many more than that.) The mass of the machine went down as the cube of the scale of reduction, and the increased acceleration varied as the reduc- tion in size. Wear and corrosion would be the same as for a larger machine, but the quantity molecules. The properties of this small world must be considered.746 PROCEEDINGS [AnaZyst, Vol. 88 of material used was so small that platinum and diamond could be used. Surface tension would be greater and it would be necessary to work in either a completely wet or a completely dry atmosphere. The operator’s ability to see would be diminished-at one-tenth he would become very short sighted, at one-hundredth scale almost blind, and at one-thousandth scale would need to use an electron microscope. If this reduction in our scale of operations could be achieved, it became possible to imagine a micro feeler catching a molecule and “feeling” its composition. Dr. Feynman, of the California Institute of Technology, who had put forward ideas similar to Dr. Martin’s own, had suggested that single molecules could be synthesised by pressing the atoms together.
ISSN:0003-2654
DOI:10.1039/AN963880745b
出版商:RSC
年代:1963
数据来源: RSC
|
4. |
Methods of organic microanalysis: a comparative and criticalstudy |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 746-747
R. Lévy,
Preview
|
PDF (142KB)
|
|
摘要:
746 PROCEEDINGS [AnaZyst, Vol. 88 Methods of Organic Microanalysis: a Comparative and Critical Study BY R. LBVY DR. LBVY reviewed and compared various methods for the micro-determination of elements in organic substances, and included some reference to the effects of a wide range of unusual (het ero-) elements. For the determination of fluorine, in which the fluoride was measured spectrophoto- metrically by its bleaching action on iron sulphosalicylate, the oxy-hydrogen flame was the only method he recommended for liquids and volatile compounds, but for other samples the oxygen- flask method could be used, with combustion initiated with a hot silica rod. Use of gelatine capsules always led to low results, but packing in filter-paper with about 15 mg of glucose worked well. The flask combustion had proved itself better for polymers such as Teflon. The spectrophotometric method had a limited precision, and taking 5- to 6-mg sample weights was therefore recommended, standard fluoride solutions being treated at the same time and in an identical manner with the unknowns.Results for phosphorus by the phosphovanado- molybdate colorimetric method were excellent if similar precautions were taken. For the determination of chlorine, bromine and iodine, five methods of decomposition had been compared. The oxy-hydrogePzJEame had few advantages and required the most skill; iodine could be lost as a solid deposit, and hetero-elements gave more trouble than in other methods. The Pregl-type bead tube with detachable absorber was much used without plati- num or silica filling, but with at least 100 mm of the combustion tube at over 1000" C.With a series of samples, chlorine or bromine could be determined in 25 minutes. Volatile samples had to be heated with care, but others could be heated automatically. Oxygen--ask combustion was cheaper and had little that could break down, such as electric heaters. Generally the bead-tube method was best for routine work and the flask combustion for occasional use, but Dr. Lkvy's laboratory had uses for both. Phosphorus, boron, zinc, sodium, man- ganese and cobalt did not interfere in the determination of chlorine by flask combustion, although some of them interfered in the iodine determination. For similar reasons the peroxide-fusion and sulphuric - chromic acid methods were still sometimes used.In the former, fluorine, phosphorus, nickel and zirconium did not interfere, manganese and cobalt could be removed by filtration and tin was the only important element that interfered. The sulphuric - chromic acid method (Zacherl and Krainick) was the most widely applicable when hetero-elements were present. It took the most time, but was the easiest to carry out. Chlorine and bromine were invariably determined by potentiometric argentometry to a pre-determined potential and the condition, e g . , absorbing solutions and volumes, were strictly defined. Iodine was oxidised to iodic acid as in Leipert's method, but by the use of bromine in absence of acetic acid after flask combustion. In presence of chlorine, iodine could be removed by boiling after the addition of more peroxide and sulphuric acid.\Then chlorine was present together with bromine, the method of correction already published was suitable. This was intended to follow a sodium peroxide fusion, and consequently, when it was used after other methods of decomposition, sodium peroxide had to be added to the solution. Addition of sodium hydroxide and sulphuric acid instead did not give the same results. For the determination of chloride an electrode of 5- to 6-mm diameter silver rod was used; it was polished daily with fine emery and chamois leather by spinning at 2500 r.p.m. The end-point potential was pre-determined with 5- or 10-ml portions of 0.01 N silver nitrate and 0.01 N potassium chloride (the chloride being added to the nitrate), repeated until results were consistent. Subsequsntly samples were titrated to this potential by adding excess of silver nitrate and back titrating with potassium chloride.October, 19631 PROCEEDINGS 747 For bromide the cleaned electrode was stabilised in a silver bromide suspension and its stability was checked by repeated titrations of potassium bromide with 0.002 N solution of silver nitrate; here the back titration was not used. In conclusion, Dr. Litvy considered the effect of many hetero-elements on the carbon and hydrogen determination using either the Zimmermann filling of copper oxide and silver or the VeEefa filling of cobalt oxide on pumice grains at 750" C. The latter method was shown to have certain advantages.
ISSN:0003-2654
DOI:10.1039/AN9638800746
出版商:RSC
年代:1963
数据来源: RSC
|
5. |
Mass spectrometry as a microanalytical technique |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 747-748
A. Quayle,
Preview
|
PDF (135KB)
|
|
摘要:
October, 19631 PROCEEDINGS 747 Mass Spectrometry as a Microanalytical Technique BY A. QUAYLE MR. QI'AYLE said that the first working mass spectrograph had been built by Aston in 1919, and this had set the pattern on which succeeding instruments had been based. The substance under analysis was excited by an electron beam into giving off streams of positively charged particles. These were focused by passing them through a pair of slits; they were then subjected to electrical and magnetic fields, and the resulting rays focused on some form of detector, either a photographic plate or an ion collector connected to a suitable amplifier. He showed that the amount of deflection of the beam by the magnetic field was proportional to the mass and charge of the particles, by the following mathematical treatment.Hence a particle experienced a centrifugal force equal to i14v2/v. This force was caused by the field H , which was exerting a force HeV. Hence- The fields deflected the charged particles in an arc. 1yI v 2 Mv Y eH . . .. .. H e v = - or Y = - . . But the kinetic energy of the particle after acceleration was the same as its potential energy, i.e.- -$iWv2 = el/ or v = (2eI'/M)+. Substituting for v in equation (1)- M22eV + 2wv Q = (S)' which, when rearranged, gave M/e = r2H2/2V, and if the fields H and I' were constant, (.W/e)& varied as the deflection (as measured by r ) . Thus in a beam of ionised particles of equal energy the different masses could be distinguished by the variation in deflection. In practice the magnetic or electric fields were varied so that ions of different masses were brought in turn to the collector.There were various methods of recording this deflection -focusing the ions on a photographic plate, or measuring the electrical discharge, either by means of a galvanometer spot on photographic paper or by other devices. For qualitative work plane plates could be used, but for accurate quantitative work, concave plates were preferable. When the readings were likely to be indistinct, they could be intensified by making a longer exposure. Gases could readily be handled, and hydrocarbons up to a five-carbon chain had been analysed at room temperature. The size of the sample used was 2 litres at a pressure of 0.1 mm of mercury (approximately 0.3 ml at S.T.P.). Even smaller samples could be used, and analyses had been carried out on 1 cu.mm of gas, e.g., the gas bubbles formed in plate glass during manufacture.Volatile liquids (having a vapour pressure of more than 30 p a t room temperature) could also be examined, the sample being introduced into the receiver through a tiny orifice. About 1 p1 of liquid was required. Samples having a low vapour pressure might have to be heated and so had to be thermally stable. Solids also had been examined, about 1 mg of sample being introduced into the apparatus, but nine-tenths of the sample was pumped away before analysis. One method of introduction was to cover a sintered plate with molten gallium and to introduce the sample as a liquid under the metal by a syringe. The sample was then pumped through the sintered plate into the discharge chamber. Thermally unstable or volatile samples could be introduced directly into the chamber through a pipe. Solid samples were introduced into the electron beam as pellets. Samples having molecular weights of up to 800 had been analysed by these techniques.748 PROCEEDINGS [Analyst, Vol. 88 He illustrated the analyses of monatomic and diatomic gases by taking fluorine, krypton and nitrogen as examples. Typical spectra such as those given by air and hydrocarbons were discussed in detail, the examples taken being the analyses of methane, ethane, propane and butane. Other materials whose analysis the speaker described were methyl esters, polychloronaphtha- lenes and alkyl phosphates. The instrument could be used semiquantitatively for the analysis of gas mixtures and was finding increasing application in conjunction with gas chromatography. Mr. Quayle described in detail some of the applications of the technique.
ISSN:0003-2654
DOI:10.1039/AN9638800747
出版商:RSC
年代:1963
数据来源: RSC
|
6. |
Some recent developments in functional-group analysis on the micro scale |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 748-749
Stig Veibel,
Preview
|
PDF (145KB)
|
|
摘要:
748 PROCEEDINGS [Analyst, Vol. 88 Some Recent Developments in Functional-group Analysis on the Micro Scale BY STIG VEIBEL PROFESSOR VEIBEL stated that micro-elemental analysis permitted the establishment of the empirical, but usually not the structural, formula of an organic compound. For this, func- tional-group analysis was more important, being the equivalent of the ionic reactions of the elements in inorganic analysis, but it had to be remembered that mutual interference of the different functional groups was much .more serious than the mutual influence of the different elements. Chemical methods had in recent years been supplemented by physical methods such as chromatography (column, paper, gas, thin-layer) and spectroscopy (ultraviolet, infrared, nuclear magnetic resonance, micro-wave, mass), which facilitated the use of micro techniques for the identification of organic compounds.October, 19831 PROCEEDINGS 749 Three different steps in the identification had to be considered, viz.: the detection of the presence of a particular functional group, the characterisation of the compound through the preparation of a derivative or through determination of a physical constant (m.p., n,, etc.) and the determinatioiz of the equivalent weight of the compound with respect to the functional group considered. Commission I of the Division of Analytical Chemistry of I.U.P.A.C. had started work on a Report on Reactions and Reagents useful for these three purposes. DETECTION OF THE PRESENCE OF A PAKTICULAK FUNCTIONAL GROUP- Feigl’s spot tests had been extended to cover many functional groups, but as most of the reactions were colour reactions, they were not sufficiently specific to be full; trusted unless subsequently corroborated by more specific tests.For this the use of reagents such as 2,4-dinitrophenylhydrazine and other hydrazine derivatives, aralkylthiouronium halides, fluoro-2,4-dinitrobenzene, or other classical or new reagents, had been considerably developed in recent years. CHARACTERISATION OF FUNCTIONAL GROUPS THROUGH DERIVATIVES- The reagents described above would often permit the preparation of crystalline derivatives under the microscope. Physical constants, such as the melting-point, could then be deter- mined directly under the microscope, but it had to be remembered, e.g., that all hydrazine derivatives of carbonyl compounds were able to exist in sy” and anti forms, which in the neigh- bourhood of the melting-point easily underwent mutual interconversion, causing a depression of the melting-point.Other derivatives, such as the aralkylthiouronium salts, were easily decomposed a t elevated temperatures, and the rate of heating of the bath should therefore be recorded together with the melting-point found. XF values (paper or thin-layer chromatography) or retention times (gas chromatography) were, for some derivatives, more useful than melting- or boiling-points. Infrared spectra were also of great value for the characterisation of different compounds containing particular functional groups. Mr. A. Jart had determined the infrared spectra of some 300 benzyl- thiouronium salts and of some 160 9-bromophenyl esters of carboxylic acids. These spectra were to be published in Acta Polytechmia.The latest development was a combination of a gas chromatograph with a mass spectro- graph. g. The investigation of the specificity of the various reagents suggested has not always been sufficiently thorough. I t is to be hoped that the above-mentioned I.U.P.A.C. report will be critical, especially in this respect. DETERMINATION OF FUNCTIONAL GROUPS- Great progress had been made by the introduction of acid - base titrations in non-aqueous solvents, making possible the titration of substances with very weak protolytic activity, e.g., acids with pK, values up to 10 or 11 and bases with pKb values up to 11 or 12. The introduction of titration machines, automatically recording the titration curve, had been a great help. Polarography, which until recently had been used mainly in inorganic analysis, had found applications in organic analysis, too, oscillographic polarography being possibly the most useful variant, By a combination of chromatographic and spect rophot ome tric methods an automatically recorded investigation of, e.g., protein hydrolysates had been made possible. Other instrumental methods included chronopotentiometry, conductometric titrations and electrophoresis. References to detailed methods could be found in annual or bi-annual progress reports in most of the journals of analytical chemistry and in the “Organic Analysis’’ series edited by Mitchell, Kolthoff, Proskauer and Weissberger. It was claimed that this method was applicable to samples from 10 g down to Progress had also been made with redox titrations.
ISSN:0003-2654
DOI:10.1039/AN963880748b
出版商:RSC
年代:1963
数据来源: RSC
|
7. |
The status of microgram analysis |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 749-750
R. Belcher,
Preview
|
PDF (161KB)
|
|
摘要:
October, 19831 PROCEEDINGS 740 The Status of Microgram Analysis BY R. BELCHER PROFESSOR BELCHER explained that the submicro procedures developed a t the University of Birmingham for the elementary and functional-group analysis of organic compounds used a sample weight of about 30 to 50 pg, but achieved the same order of accuracy as comparable micro techniques, This order of sample weight was chosen because it represented about750 PROCEEDINGS [Analyst, Vol. 88 the lowest level with which it was possible to work without the use of magnification and manipulators. It was not envisaged that these sub-micro procedures would replace existing micro techniques for routine analysis, but rather that they would be employed for those determinations, particularly in biochemistry, where only very small amounts of material were available.The first method developed had been a sealed-tube digestion for nitrogen-containing compounds, sulphuric acid being used. Ammonia in the digest was eventually oxidised by an excess of hypobromite, the excess being determined iodimetrically. Sealed-tube decomposition with sodium metal was applied to the determination of chlorine, bromine and iodine. The last two elements were conveniently and easily evaluated by amplification titrations (Van der Meulen and Leipert, respectively). Chlorine had proved more trouble- some; an indirect amplification procedure based on the use of mercury1 iodate had been developed, but the technique was extremely exacting. Fluorine was determined by oxygen- flask combustion, followed by colorimetric determination of fluoride with alizarin complexone and ceriumII1. The barium sulphate formed was filtered, then dissolved in an excess of ammoniacal EDTA and the excess of EDTA back-titrated with magnesium solution with Solochrome black 6B as indi- cator.Although a successful combustion technique had been developed for carbon, it had not proved possible to determine hydrogen at the same time because of losses of water by absorption on the surface of the silica apparatus. The carbon dioxide formed by the com- bustion was frozen out and eventually determined manometrically. Some procedures had been developed for functional groups, particularly for alkoxyl, alkalimino and carboxyl groups and some non-aqueous titrations had been investigated.Considerable attention had been focused on the determination of hydrogen, almost every conceivable approach having been tried, even methods that would involve its deter- mination independently from carbon, e.g., the Feigl spot-test reaction for hydrogen by heating the sample with sulphur and evaluation of the resulting hydrogen sulphide. Recent work had shown promise with a combustion procedure in which the water formed was immediately trapped by a desiccant. The carbon dioxide was determined manometrically; then the desiccant was heated, the evolved water passed over heated carbon, the resulting carbon monoxide subjected to the Schutze reaction, and the carbon dioxide from this determined manometrically. Several successful procedures had been developed for application after a sub-micro oxygen-flask decomposition of the sample.Chloride was evaluated mercurimetrically in an ethanolic medium with diphenylcarbazone as indicator. Bromide and iodide were deter- mined, as before, by amplification titrations. Sulphur, as sulphate, was titrated with barium perchlorate in an ethanolic medium with Thorin as indicator. Previous studies of the deter- mination of phosphorus (and arsenic) by means of a sealed-tube decomposition with nitric acid had indicated substantial losses of phosphorus on the walls of the tube. This effect was also present in the flask combustion, but it had proved possible to eliminate the difficulty. After flask combustion, phosphate was determined via a titrimetric quinoline phospho- molybdate procedure or a molybdenum-blue spectrophotometric finish.For arsenic, only a molybdenum-blue finish had proved possible. The nitro group was being determined by reduction with excess of titanium111 sulphate by using a special sub-micro storage - delivery system for the reagent. Excess of iron111 was then added and the excess back-titrated with the titaniumII1. This procedure had been successfully extended to the determination of the nitroso group, but, as on the micro scale, results had been erratic for azo groups. It had not so far proved possible to differentiate between aldehydic and ketonic carbonyl, but a general procedure for the carbonyl group had been evolved based on an oximation reaction and eventual non-aqueous titration of hydroxylamine. The bromine chloride method was being investigated as a possible general determination of unsaturation. Non-aqueous titration of the thiol group with mercuryn: perchlorate had shown promise. Remarkable success had been achieved with the Malaprade reaction-periodate titration of glycol groups. On the sub-micro scale it had proved possible to titrate the formaldehyde, formic acid and excess of periodate in a single solution, whereas three aliquots were normally required. A method for determining molecular weight was based on depression of the freezing-point of camphor. A sub-micro Carius digestion was used to determine sulphur. Much work was currently being carried out on functional groups.
ISSN:0003-2654
DOI:10.1039/AN9638800749
出版商:RSC
年代:1963
数据来源: RSC
|
8. |
A review of the methods available for the detection and determination of small amounts of cyanide |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 751-760
L. S. Bark,
Preview
|
PDF (1269KB)
|
|
摘要:
October, 19631 BARK AND HIGSON 75 1 A Review of the Methods Available for the Detection and Determination of Small Amounts Of Cyanide* BY L. S. BARK AND H. G. HIGSON (Department of Chemistry and Applied Chenzistry, The Royal College of Advanced Technology, Salford, Lancashire) SUMMARY OF CONTENTS Introduction Kon-colorimetric methods Titrimetric methods involving visual end-point detection Titrimetric methods involving instrumental end-point determination I’olarographic methods The use of gas chromatography Methods involving formation of a metal complex Other colorimetric methods Colorimetric methods based on the Kijnig reaction Colorimetric methods Conclusion THE detection and determination of small amounts of cyanide ions are important because of the extreme toxicity of cyanide to living matter.Cyanides are used in electroplating, in precious metal refining, in case hardening of steels and in many other processes; in the gas industry, many atmospheres and effluents contain cyanides, e.g., the effluents from coke ovens and other gas-making plant. Hydrogen cyanide is used extensively as a rapid and con- veniently applied fumigant for food-storage plants, such as warehouses and the holds of cargo ships. Because of the extreme toxicity of cyanide, low maximum concentrations are allowed in water (which may be used for drinking purposes) and in atmospheres (where human beings may be working). The World Health Organisationl recommends that water containing more than 0.01 p.p.m. of cyanide (as CK-) should be rejected as unfit for public use (domestic supply); the toxic limit for fish is 0.03 p.p.m.The maximum allowable concentration in atmospheres in the United States of America2 is 10 p.p.m. (5 mg of HCN per cubic metre of air). These low limits mean that it is essential to use extremely sensitive tests for detecting and determining cyanide. For several years there has been a steady increase in the extent and variety of the methods available for detecting and determining cyanides, which are found not only as “free cyanides’’ (from HCN, KCN and NaCN) and unstable cyano-complexes, such as [Zn(CN),I2-, but also as relatively stable complex cyanides, such as ferrocyanides and cobalticyanides, which, although not showing typical cyanide properties, are still toxic and classified by most health authorities with cyanides.Most of the methods used for detecting and determining cyanide involve the formation of hydrogen cyanide. Of the suggested procedures for decomposing complex cyanides to give hydrogen cyanide (either for detection or determination), that proposed by William~,~ in which the solution of simple and complex cyanides is distilled after addition of an acid solution of cuprous chloride, is recommended by British authorities4 ; and Kruse and Mellon’s m e t h ~ d , ~ in which the complex cyanides are decomposed by distillation with phosphoric acid in the presence of citric acid and ethylenediaminetetra-acetic acid (EDTA) under partial vacuum, is used by many chemists in the U.S.A. The latter workers6 also suggested solvent extraction as an isolation method.Serfass, Freeman, Dodge and Zabban’ used a reflux method with an air current to transfer hydrogen cyanide from the sample to absorption tubes. Decomposi- tion of complex cyanides is hastened by the action of magnesium chloride, mercuric chloride and sulphuric acid, yielding uniform high recoveries of hydrogen cyanide from simple or complex cyanides; boiling under reflux for 1 hour is generally adequate. Complex cyanides of iron, copper or cobalt generally require longer treatment. Schwapowalenko* recommended that the hydrogen cyanide be liberated by heating the test solution in the presence of sodium hydrogen carbonate to prevent any interference from volatile oxidising or reducing compounds. *Reprints of this paper will be available shortly. For details please see p.822.752 sections, viz., non-colorimetric and colorimetric ; these can be further sub-divided, e,g.- BARK AND HIGSON: A REVIEW OF THE METHODS AVAILABLE FOR THE [A.tzaZyst, Vol. 88 The methods used for detecting and determining cyanide can be divided into two main Non-colorimetric- (a) Titrimetric methods involving visual end-point indicators. (b) Titrimetric methods involving instrumental methods for determining the end- point, (c) Polarographic methods. ( d ) Gas-chromatographic methods. (;) Methods involving formation of metal complexes. (ii) Other methods. Colorimetric methods- Some methods are applicable only to the detection and some-only to the determination of cyanide, but most of the available colorimetric methods are used for both.NON-COLORIMETRIC METHODS FOR DETECTING OR DETERMINING TRACE AMOUNTS OF CYANIDE The detection of hydrogen cyanide by smell can be one of the most sensitive tests (0.001 p.p.m. can be detected), but many people are relatively insensitive to the smell. TITRIMETRIC METHODS INVOLVIKG VISUAL END-POINT DETECTION- The earliest reported method for determining cyanide is Liebig’s titrimetric method,B based on the formation of turbidity due to silver cyanide after all the cyanide has reacted with standard silver nitrate solution. An argenticyano-complex is formed initially, and then a slight excess of silver nitrate produces turbidity due to the formation of silver cyanide. This titration is subject to error in alkalinelO~ll and a m m o n i a ~ a l l ~ , ~ ~ solution. A similar method proposed by D4nig&sl4 is based on the turbidity due to silver iodide in the presence of ammonium hydroxide with potassium iodide as indicator ; although this method gives high results in the presence of a large excess of ammonia,12 J3 J5 J6 accurate results are obtained if the concentration of ammonium hydroxide is carefully reg~1ated.l~ 9 1 3 9 1 5 917 Beerstecherls modified the Dhig&s silver nitrate titration ; he used a photo-electric colorimeter or turbidi- meter for end-point detection, and delivered 0.001 M silver nitrate solution from a piston microburette. Ryan and Culshaw19 reported the use of 9-dimethylaminobenzylidene rhodanine as an indicator in Liebig’s method, and this modification was recommended by the American Public Health Association20 for determining cyanide concentrations of 1 p.p.m.and more. For concentrations greater than 10 to 20 p.p.m., the British recommended method4 is also the Liebig method, which is therefore still widely used. Ricci2I published a paper on the inter-relations of the equilibrium constants in aqueous solutions saturated with silver cyanide, and considered mathematically the relation of the point of precipitation of silver cyanide or of silver iodide, as the titration end-point, to the equivalence-point in the argentimetric titration of cyanide. In a second publication22 he considered the shape of the titration curve, or the graph obtained by plotting a function of the concentration of silver’ ion present (p[Ag+]). against the amount of silver nitrate added, to determine the position of the inflexion-point in the curve and its relation both to the equivalence-point and to the point of first appearance of a precipitate.Two sets of conditions were examined mathematically- (i) the titration of pure aqueous potassium cyanide with silver nitrate, in which the hydrogen ion concentration is variable during the titration, and (ii) the titration of potassium cyanide in presence of excess of ammonia-in this condition the value of the hydrogen ion concentration is practically constant, being fixed mainly by the ammonia. Under the first conditions the precipitation of silver cyanide would precede the inflexion- point, which would occur only in the supersaturated solution, but the precipitation of silver cyanide, if it should become visible when the solution first becomes theoretically saturated, would give only a negligible (negative) error when taken as the end-point. The inflexion- point occurs before the equivalence-point, but the titration error with the inflexion-pointOctober, 19631 DETECTION AND DETERMINATION OF SMALL AMOUNTS OF CYANIDE 753 as the end-point would be extremely small. In the presence of excess of ammonia, the inflexion-point occurs at the equivalence-point. Ricci worked out mathematically that the precipitation of silver cyanide (unless the concentration of potassium cyanide was large) occurred after the equivalence-point, leading to a small (positive) error.The precipitation of silver iodide occurred before the inflexion-point. The calculations would suggest that the inflexion-point, without potassium iodide, would be a much better end-point.The importance of this theoretical mathematical treatment is that the best conditions may be selected for titrating different concentrations of cyanide with silver nitrate solution. Other titrimetric methods involving use of different indicators and titrants have been reported : diphenyl~arbazide~~ 324 and ~ a l c e i n ~ ~ have been used as indicators in argentimetric titrations; titrations involving the formation of a mercury cyanide have been described by Tanaka and Yamamoto,26 Gregorowisz and Ruh12' and Kraljic,2s who used cupric diet hyldithio- carbarnate, Variamine Blue and nitrobenzene (with ferrocyanide) , respectively, as indicators. It is often necessary before titration to separate the cyanide as hydrogen cyanide by dis- tillation to avoid interference from organic substances and oxidising and reducing agents.W r b n ~ k i ~ ~ reported the independent titration of cyanide and sulphide ; he used o-hydroxy- mercuribenzoic acid to complex the sulphide and titrated with standard nickel sulphate solution, with murexide as indicator. He also suggested the use of thiofluorescein as an indi- cator in the argentimetric determination of cyanide; the colour change is, however, not very sharp, and the indicator is not widely used. The use of standard nickel solution in the titra- tion of cyanide was also reported by de Sou~a,~O who determined cyanide in the presence of thiocyanate and chloride by adding a known excess of ammoniacal nickel sulphate solution to the sample and titrating the non-complexed nickel with EDTA, with murexide as indicator.TITRIMETRIC METHODS INVOLVING INSTRUMENTAL END-POINT DETECTION- Instrumental methods of analysis are frequently used; the potentiometric titration of cyanide with silver nitrate has been described by Treadwell, Waller and La~terbach,~~ and Clark.32 Read and Read33 suggested bimetallic electrode titration and Gregory and Hughan,34 using the null-point equivalence potential method proposed by C a ~ a n a g h , ~ ~ reported that this method is superior to Liebig's method for determining cyanide in plating solutions.36 Wick13 reported that potentiometric titration with mercuric chloride as titrant was an accurate method; Thompson,l7 however, stated that the silver nitrate titration of cyanide to Ag(CN),- was more accurate than the titration with mercuric chloride.Laitinen, Jennings and Parks37 claimed that the amperometric titration of cyanide with silver nitrate is equal in accuracy and precision to the visual DGnig6s method, and is applicable at much higher dilution. Direct amperometry with a rotating silver anode and a stationary platinum cathode has been used by McCl~skey,~~ who claims a lower limit of linear response at about 0.01 pg of cyanide in a 7-ml sample and 0.0005 pg of cyanide for a 0.5-pl sample introduced into 10 ml of sup- porting electrolyte. The electrode response and the sensitivity levels change daily and hence must be frequently checked. A similar system has been operated and patented39 in this country for electrically determining the concentration of hydrogen cyanide in air (in coal- mines, etc.).Shinozuka and Stock40 have described the amperometric titration of low concentrations of cyanide with silver nitrate; they used a rotating platinum electrode and a sodium sulphite medium. A potentiometric method, in which an auxiliary current of 5 to 20 p A per sq. cm is used, has been patented*l for determining cyanide concentrations in solution (particularly solutions obtained in the refining of gold and silver). Electrometric titrations of cyaiiide by the dead-~top~~ end-point system with 0.1 N silver nitrate43 and 0.01 N iodine solutiong4 have been reported. Baker and Morrison45 measured the current from the electrochemical cell Ag/NaOH(O.l M)/Pt, containing cyanide ; the cell operates spontaneously as a measure of one of the reactants in the cell (CN-)-for 2.6 pg of cyanide, a response current of 17 to 18 pA was reported. The coulometric titrations of cyanide with mercury and silver have been compared by Przybylowicz and Rogers,46 who showed that mercury was slightly superior t o silver for the determination of small amounts of cyanide.POLAROGRAPHIC METHODS- Anodic reactions of cyanide at a dropping-mercury electrode have been reported. Jura47 described the determination of cyanide in flowing alkaline solutions with a shielded dropping- mercury electrode ; empirically derived correction factors for the polarographic determination754 [Analyst, Vol. 88 of free cyanide in the presence of sulphide were used by Karchmer and Walke1-,~8 and Hetman49 reported a method, depending on a pure reduction wave, in which a base electrolyte consisting of pyridine and potassium nitrate with a mercuric salt was used.The last-named method is uncomplicated and permits the determination of cyanide in concentrations as low as 2 p.p.m. with a comparatively simple polarograph. BARK AND HIGSON: A REVIEW OF THE METHODS AVAILABLE FOR THE THE USE OF GAS CHROMATOGRAPHY- Gas - liquid chromatography has been used for determining hydrogen cyanide. Wool- mington50 used a method in which water vapour and hydrogen cyanide are separated from oxygen, nitrogen, methane and carbon monoxide by passing the sample (with hydrogen as carrier gas) at 100" to 107" C through a column packed with acid-washed Celite (30 to 60 mesh) impregnated with 20 per cent.w/w of polyoxyethylene glycol (molecular weight 1500). Hydrogen cyanide emerges from the column after 4 to 6 minutes, and its concentration is calculated from peak heights. Schneider and Freund51 attained sensitivity in their method by a 2000-fold concentration step involving a short column cooled in a mixture of solid carbon dioxide and acetone, and amplification of the thermistor-bridge detector output before recording it on a strip chart. COLORIMETRIC METHODS METHODS INVOLVING FORMATION OF A METAL COMPLEX- Some of these methods, although specific for cyanide and of ample sensitivity, produce unstable colours. The formation of t h i o ~ y a n a t e ~ ~ ~ ~ ~ @ ~ ~ ~ from cyanide (by reaction with ammonium polysulphide) and then ferric ferrithiocyanate (by addition of ferric chloride ; producing a blood red colour) is an excellent test for cyanides; however, it is extremely unreliable as a method for determining small amounts of cyanide owing to the instability of the ferric thiocyanate colour.The test is applicable in the presence of sulphide or sulphite; if thiocyanate is originally present, the cyanide must first be isolated before the test is applied (sensitivity, 1 pg of CN-; concentration limit, 1 in 50,000). Another extremely sensitive test for cyanide depends on the formation of Prussian blue53 @ 9 5 7 9 5 8 959 (Fe,[Fe(CN),],). Gettler and G o l d b a ~ r n ~ ~ found that the sensitivity of the Prussian-blue method could be enhanced by conducting the gas, obtained by aeration of the solution containing hydrogen cyanide (at 90" C), through filter-paper impregnated with ferrous sulphate and dilute alkali (sensitivity, 0.2 pg of CN-).The cyanide (originally present in the sample) could then be determined by comparing the blue stains produced by immersion of the paper in dilute sul- phuric acid with standard stains. HubachG0 modified this method for detecting cyanides and ferrocyanides in wines. For the detection of cyanide, the aeration of the sample was carried out a t room temperature (instead of 90" C). For ferrocyanides, use is made of the fact that hydrogen cyanide is liberated quantitatively from soluble and insoluble ferrocyanides by sulphuric acid in the presence of cuprous chloride at temperatures near 100" C.British authorities6I recommend Gettler and Goldbaum's method for detecting hydrogen cyanide vapour in air. The copper acetate - benzidine test,62,63,64 sensitive to 0.25 pg of cyanide (CN-) in a limit of dilution of 1 in 200,000,65 is one of the most sensitive and easily applied methods for detecting cyanide. The reaction involves the formation of benzidene blue and is due to the oxidation of benzidine by the removal of copper' from the copper" - copper1 system, which raises the oxidation potential of the system and the benzidine is oxidised. It is also probable that copper - benzidine complexes are formed. This test is extremely sensitive for cyanide, provided that oxidising or reducing substances are absent ; hence Schwapowalenko's* recommendation that hydrogen cyanide be liberated by heating the test solution in the presence of sodium hydrogen carbonate.Cullinane and Chard6, claimed that a four-fold increase in sensitivity resulted if a substituted benzidine, 2,7-diaminodiphenylene oxide, is used in place of benzidine, but we have not been able to substantiate this claim. The Weehuizen method,67 y68 969 y 7 0 which involves the oxidation of phenolphthalein in alkaline solution to the corresponding red phthalein by copper'' ions in the presence of cyanide is a recommended method4 for determining cyanide. It is an extremely sensitive method, but requires careful control ; it is not specific for cyanides because erratic results may be caused by minute amounts of foreign oxidising materials. A test-tube for detecting hydrogen cyanide in air or other gases has been ~atented.7~ The tube contains a carrier substance, e.g., silica gel, impregnated with silver chloride, mer- curous chloride, lead chloride or a chloride of a metal forming water-insoluble or sparinglyOctober, 19631 DETECTION AND DETERMINATION OF SMALL AMOUNTS OF CYANIDE 755 soluble cyanides (such as those of copper or palladium), and with an indicator for hydrogen chloride, e.g., bromothymol blue or bromocresol green.The prepared tubes are claimed to be insensitive to oxygen (oxidation) and stable for several years. The ability of the cyanide ion to form stable complexes and cause de-masking of inner complex-bonded transition metals has also been used by various workers for detecting and determining cyanide. Feigl and Feig172 reported a colorimetric procedure in which the de-masking of dimethylglyoxime by the action of cyanide on the palladium dimethylgloximate complex permits nickel ions present to form a red dimethylgloximate.An advantage of this test is that it can be made in alkaline solution, and, in contrast to other methods, does not require the previous liberation and evolution of hydrogen cyanide. Other palladium complexes gave analogous results, and, compared with palladium dimethylgloximate, the palladium salt of 1 ,2-cyclohexanedione-dioxime73 gives increased sensitivity. Feigl and Hei~ig'~ have also reported the detection of 2-5 pg of cyanide by the de-masking of copper from copper" oxinate, which permits aluminium ions present to form the fluorescent aluminiumIII oxinate.Brooke75 selected palladium cc-furildioxime as the most sensitive reagent for determining cyanides in "refinery waste water" in the range 0.5 to 3 p.p.m. of cyanide, and Hanker, Gelberg and Witten76 used the reaction of cyanide with potassium di-(7-iodo-5-sulpho- oxino) palladium'I in alkaline solution. The addition of iron"' in acid solution produces a blue colour due to the iron complex with an absorption maximum at 650 mp. These workers also developed a fluorimetric method depending on the de-masking of 8-hydroxy-5-quin- olinesulphonic acid by cyanide from the non-fluorescent potassium di- (5-sulpho-oxino) pal- ladium". The liberated quinoline derivative then co-ordinates with magnesium ions present to form a fluorescent chelate, which is used for measuring the cyanide concentration.It is claimed that 0.02 pg of cyanide per ml of solution can be determined. Hanker, Gamson and Klapper77 had earlier developed a method involving the formation of a fluorescent com- pound by reaction between nicotinamide and cyanogen chloride (formed by the reaction of cyanide and chloramine-T). Although it is rapid, the method is not widely used. Musha, Ito, Yamamoto and Inam0ri7~ examined the inhibiting action of cyanide ions on thechemi- luminescence of luminol (3-aminophthalhydrazide), and determined the cyanide concentration by measurement of the induction period. Yamasaki and Tto79 indirectly determined cyanide by converting it into a stable nickel complex [Ni(CN),I2- at pH 9 to 11 (in presence of an am- monium buffer) ; the excess of nickel was determined colorimetrically with furil a-dioxime in a suspension at pH 4 containing gelatin, the optical density at 480 mp being measured. Several methods involving the de-masking effects of mercuryII have been used for detecting and determining cyanide.In one methodso cyanide is used to liberate 2-hydroxy- ethyldithiocarbamic acid from its mercury" complex ; the liberated acid reacts with copper to form a soluble yellow complex having an absorption maximum at 383 mp. It is claimed that between 0.2 and 4 p.p.m. of cyanide can be rapidly determined by this method. Some ions interfere, but up to a 60-fold excess of thiocyanate can be tolerated. Tanaka and Y amamotosl recently developed a method involving the use of mercuric diphenylcarbazide ; paper impregnated with this reagent changes from blue-violet to red in the presence of cyanide (0.25 pg per 0.05 ml) in a neutral or weakly basic solution.With gaseous hydrogen cyanide flowing in a thin glass tube containing a strip of paper, the length of the colour change on the strip of paper is proportional to the amount of cyanide. The sample (approximately 1 ml containing less than 10 pg of cyanide) is heated with sodium hydrogen carbonate in a glass tube. The error is less than 5 per cent., and interference from ferrocyanides is avoided by using cadmium nitrate mixed with the sodium hydrogen carbonate. Ohlweiler and Meditschs2 used the colour reaction of mercuric ions with p-dimethyl- aminobenzylidene rhodanine as the basis of an indirect absorptiometric method for deter- mining cyanide, based on the masking of the mercury by reaction with cyanide and measure- ment of the excess of mercury'I by reaction with the organic reagent. These workers also used diphenylcarbazones3 to react with the mercury not masked by reaction with cyanide, and state that greater sensitivity is thus obtained.Both these methods require the use of a Conway microdiff usion apparatus for preliminary concentration of the cyanide and hence require fairly lengthy test times to obtain results. The methods are not applicable below 0-5 p.p.m, without the use of a differential technique, Hoffmans4 described a similar reaction756 BARK AND HIGSON: A REVIEW OF THE METHODS AVAILABLE FOR THE [Analyst, Vol. 88 with diphenylcarbazone ; many other ions (iodides, sulphides, etc.) commonly present in effluents interfere.The ready formation of an argenticyanide complex has been used in a de-masking reaction.85986 In the first method,B5 paper impregnated with silver dithizonate is treated with a drop of test solution containing cyanide, and the chromatogram is developed with a mixed water - 0.1 N potassium hydroxide (40 + 1) solution. The area of the white spot (on a violet- pink background) is proportional to the concentration of cyanide. This method is rapid and gives reasonably accurate results ; its main advantage is that thiocyanate, chloride and bromide do not interfere. The second method86 is based on the same principle, but the weaken- ing of the colour of the silver dithizonate in carbon tetrachloride solution is taken as a measure of the amount of cyanide present.When copper, zinc and lead are present in the original solution, they are sequestered with EDTA. For samples containing sulphides or free chloride, the solution is treated with lead nitrate, sodium sulphite or sodium thiosulphate before the addition of EDTA. Schilts7 described the spectrophotometric determination of cyanide based on the forma- tion and extraction of the neutral dicyano bis-( 1,lO-phenanthroline) - iron" complex produced by the exchange reaction between the reagent ferroin and cyanide ions. The method is convenient, relatively free from troublesome interferences and applicable to microgram amounts of cyanide ion and concentrations of the order of microgram per ml (p.p.m.).An indirect colorimetric method for determining cyanide (and sulphide) with thio- fluorescein has been reported by Wr6nski.88 An alkaline solution of thiofluorescein, de- colorised with silver nitrate, becomes blue on the addition of hydrogen cyanide or hydrogen sulphide. The phenomenon is used to determine hydrogen cyanide (1 to 7 pg) and hydrogen sulphide (0.5 to 2 pg). For simultaneous determination, the optical density is measured, then a drop of formaldehyde is added to discharge the colour caused by hydrogen cyanide, and the optical density due to hydrogen sulphide alone is measured after 10 minutes. Gregorowicz, Buhl and Sliwasg determined traces of cyanide in alkali sulphides by precipitation of the sulphide ions with zinc ions, and then steam-distillation of hydrogen cyanide from the diluted solution into 0.01 N sodium hydroxide and reaction of the acidified distillate with Variamine Blue solution in the presence of copper sulphate solution.In a second paperg0 the colour reaction of copper" and Variamine Blue with cyanide (0.03 pg per ml), thiocyanate (40 to 200 pg) and iodide (1 to 10 mg) was reported. In the determination of cyanide, a 100-fold excess of sodium, potassium, ammonium, chloride, nitrate and sulphate ions may be present without any ill-effects. OTHER COLORIMETRIC METHODS- The use of picric acidQ1 to 98 involves simple procedures, easily prepared reagents and a stable developed colour. Although it is not a very sensitive method it has recently been usedg9 for determining the residual cyanide in and on apples after fumigation with hydrogen cyanide.Alkali cyanides and complex cyanides can be detected to a limit of 0-05 pg of cyanideloo by their catalysis of the benzoin condensation in alkaline solution. The benzoin formed is readily detected by the violet colour obtained on addition of o-dinitrobenzene. COLORIMETRIC METHODS BASED ON THE KONIG REACTION- For the detection and determination of small amounts of cyanide in trade effluents, the method recommended by the Joint Committee of the Association of British Chemical Manufacturers and the Society for Analytical Chernistrylol is based on the method developed by Aldridge,lo2 which is an example of the KOnigl03 synthesis for pyridine dyestuffs, in which cyanogen bromide or chloride reacts with pyridine and an aromatic amine to form a dyestuff.Aldridge's method consists in allowing the cyanide to react with excess of bromine, removing the excess of bromine with arsenious acid solution, and then allowing the cyanogen bromide formed to react with a mixed pyridine - benzidine reagent. An intense orange colour is formed, which changes to red during 6 minutes, according to Aldridge, and is stable for 6 to 24 minutes at 20" C when measured with a Spekker spectrophotometer in 1-cm cells and a 2-303 Ilford blue filter. In his second paper improvements are described, and, by using an Ilford 604 green filter, he claims that an increase in sensitivity of 75 per cent. is obtained with a colour stability of 30 minutes; 0.2 pg of hydrogen cyanide at a concentration of 0.1 pg per ml couldOctober, 19631 DETECTION AND DETERMINATION OF SMALL AMOUNTS OF CYANIDE 757 be determined with an error of Krawczyklo4 reported that the method gives a perceptible orange colour with as little as 0.1 pg of hydrogen cyanide per ml on 100 ml of sample.In the method, cyanide and thiocyanate are determined together, and the thiocyanate is then determined in a separate portion of the acidified and aerated sample; the cyanide is obtained by difference. Bruce, Howard and Hanzallo5 modified Aldridge’s method to determine cyanide and thiocyanate separately, rather than the cyanide by difference. They used three aeration tubes in series; A (air-washing tube) contained 20 per cent. sodium hydroxide solution, B contained 20 per cent. trichloroacetic acid solution and C contained 0-1 N sodium hydroxide solution.The sample containing cyanide and thiocyanate ions was placed by pipette in tube B, and aeration was begun. After 16 minutes’ aeration, the cyanide collected in tube C was determined by Aldridge’s method. The thiocyanate left in tube B was also determined after aeration. Saltzman106 used Aldridge’s original procedure and reported that the final colour was unstable, changing in both tint and optical density; the instability of the colour increased with temperature. Good results were obtained by controlling the temperature and by allowing for the age of the sample colour by means of a set of readings made at various ages of the standards used. Whereas Aldridge used filters (initially 2.303 Ilford blue, then Ilford 604 green) that transmit light over a relatively broad spectrum region, other workers have used a definite wavelength.These wavelengths have differed rather widely-480, 510, 520 and 532 mp, which, to some extent, is an indication of the many divergent view-points about this method. For coloured or turbid samples or samples containing low concentrations of cyanide, Aldridge’s method has been modified by Nusbaum and Skupekolo7 ; the determination is carried out in the presence of n-butanol, the dye formed is thus extracted with the n-butanol and the sensitivity of the method is extended (0.02 to 0.5 p.p.m,). For samples containing a large excess of sulphide ions, Aldridge’s procedure was modified by Baker et al. lo* The sulphide ions are oxidised t o sulphuric acid with bromine at the same time as cyanide ions are brominated to form cyanogen bromide.This avoids the necessity for prior removal of the sulphide ions from the sample. The method is sensitive to less than 0.5 pg of cyanide in the presence of 2500 pg of sulphide (error t-5 per cent. at a concentration of 10 pg of cyanide). Russell and Wilkinsonlog have modified Aldridge’s method to make possible the deter- mination of cyanide in the presence of thiocyanate, ferrocyanide and ferricyanide, and also in colour solutions. The complex cyanides are “fixed” with zinc acetate’ solution and the cyanide (as hydrogen cyanide) is distilled from the acidified solution into 0.1 N sodium hydroxide and then determined with a pyridine - benzidine reagent after formation of cyano- gen bromide.(The pyridine - benzidine reagent is not the same as that recommended by Aldridge.) Up to 3 p.p.m. of CN-, present as cyanide ions, can be measured in the presence of up to 25 p.p.m. of CN-, present as thiocyanate, and up to 5 p.p.m. of CN-, present as complex ferro- and ferricyanides, in aqueous solutions containing all these substances and also in the presence of sodium and calcium chlorides. The method is accurate to within 0.05 p.p.m. of CN- at the maximum concentrations of these substances and to within 0.01 p.p.m. of CN- at lower concentrations. Aldridge’s method has been used for a wide range of materials, including the determination of cyanide in biological materialsllO and foodstuff s.ll1 9 l l 2 For biological materials a preliminary distillation step is carried out. For foodstuffs, a modified Conway cell is used to effect a preliminary separation of the cyanide by microdiffusion.They used 25 ml of sample (0.01 to 1.5 pg of cyanide per ml), formed cyanogen bromide and incubated the solution at 40” C for 40 minutes to allow colour formation after addition of a mixed pyridine - barbituric acid reagent. This method was claimed to be more satisfactory, since the colour is stable for approximately 3 hours. KratochviP14 oxidised cyanide to cyanogen chloride with chloramine-T solution, and allowed the cyanogen chloride to react with pyridine to form glutaconaldehyde, which reacts with dimedone to produce a violet colour that can be measured, after 30 to 40 minutes, at 580 to 585 mp. Desmukh and Tatwawadi115 made use of the fact that cyanides and thiocyanates develop a yellow colour when treated with chloramine-T solution in the presence of pyridine.Schulek, Burger and Feh4r116 reported a method for determining ammonia, cyanide, nitrite and nitrate when present together. 2 per cent. by this modified method. Colour measurement was made at 510 mp. Murty and Vi~wanathanll~ used barbituric acid in place of benzidine.758 BARK AND HIGSON: A REVIEW OF THE METHODS AVAILABLE FOR THE [Analyst, Vol. 88 Ammonia and cyanide are distilled off, ammonia is determined by titration of the distillate and the cyanide is then determined by the cyanogen bromide - iodide reaction. A method similar to Aldridge’s method and also based on the Konig reaction is that proposed by Epstein.l17 Cyanide is oxidised to cyanogen chloride with chloramine-T, the cyanogen chloride is then allowed to react with pyridine containing 3-met hyl-l-phenyl- 5-pyrazolone and a small amount of bis-(3-methyl-l-phenyl-5-pyrazolone), and the blue colour produced is measured at 630mp.The excess of chloramine-T is reduced by the pyrazolone, but since this will also reduce the cyanogen chloride, the pyridine must be added to avoid this. If the “bis-pyrazolone” is not used, a blue colour is still developed, but it is not stable. Epstein’s method is sensitive and may be used in acid, neutral or slightly alkaline media; it is recommended by the American Public Health Association.20 It has been used for determining cyanide in fish tissuells and in wines treated with ferr0~yanide.l~~ Dodge and Zabban120 and Balla and Bene44 also described the use of Epstein’s method for deter- mining cyanide and a copper - pyridine reagent for thiocyanate.The cyanide was separated from thiocyanate by extraction as hydrogen cyanide from acidic (glacial acetic acid) solution with four 40-ml portions of isopropyl ether. The hydrogen cyanide was re-extracted from the isopropyl ether with three 30-ml portions of 3 per cent. sodium hydroxide solution. The extraction was reported to be over 95 per cent. efficient. Ludzack, Moore and Ruchhoft121 considered that three published methods offered the most promise for the determination of cyanide in water and waste samples: (a) Aldridge’s method as modified by Nusbaum and Skupekolo7; (b) Epstein’s method; (c) the modified Liebig titration with the use of Ryan and Culshaw’s 9-dimethylaminobenzylidene rhodanine indi- cator.19 They concluded that all samples should be treated by acid distillation except when experience had shown that no difference occurs in the results obtained with or without distilla- tion.Certain samples containing interfering substances that could be steam distilled may require solvent extraction before analysis can be attempted. When this is so, 2,2,4-trimethyl- pentane (iso-octane), hexane or chloroform will remove the soluble interfering substances without extracting more than traces of cyanide. Epstein’s method was preferable for cyanide below 1 p.p.m. (Sensitivity could be extended to 5 parts in lo9 by use of a colour extraction procedure with n-butanol.) When more than 1 p.p.m.of cyanide was to be determined, the silver nitrate titration was recommended. Other workers who favour Epstein’s method are Whiston and Cherry.122 They dis- continued using Aldridge’s method owing to the carcinogenic properties of benzidine, and obtained a straight-line calibration graph over the range 0 to 2 p.p.m. with Epstein’s method. CONCLUSION From the review of methods available for determining small amounts of cyanide ions it is apparent that colorimetric methods are the best for amounts such as 0.1 to 1.0 p.p.m. Sensitivity in this range is evident in the methods involving use of phenolphthalein (or o-cresolphthalein) or Prussian blue, or the methods based on the Konig synthesis, namely Aldridge’s method and Epstein’s method.Of these methods the phenolphthalein and o-cresolphthalein methods are non-specific for cyanides and must usually be applied to samples that have been concentrated by distillation. The Prussian blue method requires the volatil- isation of the cyanide and its absorption on treated filter-paper; it is thus subject to similar errors of separation. Small amounts of ferro- or ferricyanides, if present in the sample, tend to break down on distillation, releasing cyanide and causing appreciable errors in the micro-determination of cyanide in the original sample. The two methods based on the Konig synthesis permit the determination of cyanide directly on the original sample. These two methods are generally considered to be the best for small amounts of cyanide, and in fact both have been selected for use as standard or recommended20J01 methods.The pyridine - pyrazolone colour development (Epstein) depends on the presence of a small amount of bis- pyrazolone and requires the preparation of a relatively unstable reagent. The pyridine - benzidine reaction with cyanogen halide (Aldridge’s procedure) has the necessary sensitivity and stability, but has aadistinct disadvantage in that the amine used is a well known active carcinogen. Also, there is some confusion as to the correct wavelength to use when measuring the developed colour in Aldridge’s method. The problem is to improve on the existing methods by- (i) the development of a one- or two-stage process that will give a colour with cyanide ions, hence reducing the amount of time necessary for determining cyanide ion,October, 19631 DETECTION AND DETERMINATION OF SMALL AMOUNTS OF CYANIDE 759 and probably permitting a suitable field method for determining microgram amounts of cyanide to be developed; or (ii) use of a non-carcinogens in the Konig synthesis, preferably compounds that are more stable and have a greater molecular extinction coefficient than either pyrazolone or benzidine.1. 2. 3. 4. 5. 6. 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. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. REFERENCES World Health Organisation, “International Standards for Drinking Water,’’ Geneva, ,+958. Jacobs, M. B., “The Analytical Chemistry of Industrial Poisons, Hazards and Solvents, Second Edition, Interscience Publishers Inc., New York, 1949.Williams, H. E., “Cyanogen Compounds, their Chemistry, Detection and Estimation,” Second Edition, Edward Arnold (Publishers) Ltd., London, 1948, p. 168. Ministry of Housing and Local Government, “Methods of Chemical Analysis as Applied to Sewage and Sewage Effluents,” Second Edition, H.M. Stationery Office, London, 1956, p. 72. Kruse, J. M., and Mellon, N. G., Sewage Ind. Wastes, 1951, 23, 1402. -- - , Anal. Chem., 1953, 25, 446. Serf& E. J., Freeman, R. B., Dodge, B. F., and Zabban, W., Plating, 1952, 39, 267. Schwapowalenko, A. M., Chem. Zentr., 1930, 588. Liebig, J. von, Annalen, 1851, 77, 102; J. Chem. SOC., 1852, 4, 219. Clennell, J .E., “Chemistry of Cyanide Solutions,” McGraw-Hill Book Co. Inc., New York, 1910. Sharwood, W. J., J . Amer. Chem. SOC., 1897, 19, 400. Kolthoff, I. M., and Sandell, E. B., “Textbook of Quantitative Inorganic Analysis,” The Macmillan Wick, R. M., Bur. Stand. J . Res., 1931, 7, 913. Dthighs, G., Ann. Chim. Phys., 1895, 6, 381. Kolthoff, I. M., and Stenger, V. A., “Volumetric Analysis,” Interscience Publishers Inc., New York, 1947, Volume 11, p. 227. Treadwell, W. D., 2. anorg. Chem., 1911, 71, 223. Thompson, M. K., Bur. Stand. J . Res., 1931,6, 1051. Beerstecher, E., jun., Analyst, 1950, 75, 280. Ryan, J . A., and Culshaw, G. W., Ibid., 1944, 69, 370. American Public Health Association, “Standard Methods for the Examination of Water, Sewage and Industrial Wastes,” Tenth Edition, New York, 1955. Ricci, J .E., J . Phys. Colloid Chem., 1947, 51, 1375. -, Anal. Chem., 1953, 25, 1650. Vogel, A. I., “Quantitative Inorganic Analysis,” Second Edition, Longmans, Green & Co. Ltd., Salka, A., Metal Finishing, 1960, 58 ( 5 ) , 59. Vydra, F. MarkovA, V., and P?ibil, R., Coll. Czech. Chem. Commun., 1961, 26, 2449; Anal. Abstr., Tanaka, Y . , and Yamamoto, S., Japan Analyst, 1960, 9, 6; Anal. Abstr., 1961, 8, 4547. Gregorowicz, Z., and Buhl, F., 2. anal. Chem., 1960, 173, 115. Kraljic, I., Acta Pharm. Jugosl., 1960, 10, 37; Anal. Abstr., 1962, 9, 2253. Wrbnski, M., Analyst, 1959, 84, 668. de Sousa, A., Talanta, 1961, 8, 782; I n f . Quim. Anal., 1961, 15, 61; Anal. Abstr., 1962, 9, 127. Treadwell, W. D., Muller, E., and Lauterbach, H., 2.anorg. Chem., 1922, 121, 178. Clark, W., J. Chem. Soc., 1925, 749. Read, N. J., and Read, C. P., Metal Finishing, 1941, 39, 612. Gregory, J. M., and Hughan, R. R., Ind. Eng. Chem., Anal. Ed., 1945, 17, 109. Cavanagh, B., J . Chem. Soc., 1927, 2207. Gregory, N. N., J . Coun. Sci. I n d . Res., 1943, 16, 185. Laitinen, H. A., Jennings, W. P., and Parks, T. D., I n d . Eng. Chem., Anal. Ed., 1946, 18, 574. McCloskey, J . A., Anal. Chem., 1961, 33, 1842. Roth, H. H., Mine Safety Appliances Co., British Patent 841,548, 1958. Shinozuka, P., and Stock, J. T., Anal. Chem., 1962, 34, 926. British Patent 843,028, 1957; Anal. Abstr., 1960, 7, 5186. Foulk, C. W., and Bawden, A. T., J . Amer. Chem. Soc., 1926, 48, 2045. Clippinger, D. Ii., and Foulk, C. W., I n d .Eng. Chem., Anal. Ed., 1939, 11, 216. Balla, B., and Bene, T., Nehe‘zvegyipari Kutatd Inte’zet Kozleme‘nyei, 1959, 1, 199; Anal. Abstr., Baker, B. B., and Morrison, J . D., Anal. Chem., 1955, 27, 1306. Przybylowicz, E. P., and Rogers, L. B., Ibid., 1958, 30, 65. Jura, W. H., Ibid., 1954, 26, 1121. Karchmer, J. H., and Walker, M. T., Ibid., 1955, 27, 37. Hetman, J., J. Appl. Chem., 1960, 10, 16; Lab. Practice, 1961, 10, 155. Woolmington, K. G., J . Appl. Chem., 1961, 11, 114. Schneider, C. R., and Freund, H., Anal. Chem., 1962, 34, 69. Johnson, M. O., J . Amer. Chem. Soc., 1916, 38, 1230. Vichoever, A., and Johns, C. O., Ibid., 1915, 37, 601. Francis, C. K., and Connell, W. B., Ibid., 1913, 35, 1624. Fasken, J. E., J . Amer. Wat. Wks Ass., 1940, 32, 487. Co., New York, 1943, pp.497 and 574. London, 1951. 1962, 9, 1801. 1960, 7, 1590.760 56. 57. 58. 59. 60. 61. 62. 63, 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 118. 117. 118. 119. 120. 121. 122. [Analyst, Vol. 88 BARK AND HIGSON Fulton, R. A., and Van Dyke, JI. J., Anal. Chem., 1947, 19, 922. Gettler, A. O., and Goldbaum, L., Ibid., 1947, 19, 270. Rathenasinkam, E., J . Proc. Chem. India, 1946, 18, 151. Friel, F. S., and Wiest, G. J., Waterworks & Sewerage, 1945, 92, 97. Hubach, C. E., Anal. Chem., 1948, 20, 1115. Ministry of Labour, H.M. Factory Inspectorate Booklet, No. 2, H.M. Stationery Office, 1961.Moir, J., Chem. News, 1910, 102, 17. Pertusi, C., and Gastaldi, I;., Chem.-Ztg., 1913, 37, 609. Sieverts, A., and Hermsdorf, A., Z . Angew. Chem., 1921, 34, 3. Feigl, F., “Spot Tests in Inorganic Analysis,” Fifth Edition, Elsevier Publishing Co., Amsterdam, Cullinane, N. H., and Chard, S. J., Analyst, 1948, 73, 95. Weehuizen, F., Pharm. Weekblad., 1905, 42, 271. Kolthoff, I. M., 2. anal. Chem., 1918, 57, 11. Nicholson, R. I., Analyst, 1941, 66, 189. Kobbie, W. A., Arch. Riccham, 1944, 5, 49. Heidrich, H., German Patent No. 1,105,199, 1961. Feigl, F., and Feigl, H. E., Anal. Chim. A d a , 1949, 3, 300. Voter, R. C., Banks, C. V., and Diehl, H., Anal. Chem., 1948, 20, 458 and 652. Feigl, F., and Heisig, G. B., Anal. Chim. Acta, 1949, 3, 561.Brooke, M., Anal. Chem., 1952, 24, 583. Hanker, J . S., Gelberg, A., and Witten, B., Ibid., 1958, 30, 93. Hanker, J. S., Gamson, R. M., and Klapper, H., Ibid., 1957, 29, 879. Musha, S., Ito, M., Yamamoto, Y . , and Inamori, Y., J . Chem. SOC. Japan, Pure Chem. Sect., 1959, Yamasaki, J., and Ito, R., J . Chem. Soc. Japan, Pure Chem. Sect., 1959, 80, 271; Anal. Abstr., Hikime, S., and Yoshida, H., Bunseki Kagaku, 1961, 10, 832; Chem. Abstr., 1962, 56, 6659. Tanaka, Y., and Yamamoto, S., Japan Analyst, 1960, 9, 8 ; Anal. Abstr., 1961, 8, 4546. Ohlweiler, 0. A., and Meditsch, J . O., Anal. Chem., 1958, 30, 481. Hoffmann, E., 2. anal. Chem., 1959, 169, 258. Kiekzewski, W., and Tomkowiak, J . , Chem. Anal., Warsaw, 1960, 5, 889. Miller, A. D., and Aranovich, M. I., Zavod. Lab., 1960, 26, 426.Schilt, A. A., Anal. Chem., 1958, 30, 1409. Wrhski, M., Chem. Anal., Warsaw, 1960, 5, 457; Anal. Abstr., 1961, 8, 1496. Gregorowicz, Z., Buhl, F., and Sliwa, E., Z . anal. Chem., 1962, 186, 407. Gregorowicz, Z., and Buhl, F., Ibid., 1962, 187, 1. Waller, A. D., J . Amer. Chem. SOL., 1910, 35, 406. Gutzeit, G., Helv. Chim. A d a , 1929, 12, 713. Chapman, A. C., Analyst, 1910, 35, 469. Finkelshtein, D. N., Zhur. Anal. Khim., 1948, 3, 188. Sinclair, W. B., and Ramsey, R. C., Hilgardia, 1944, 16, 291. Smith, R. G., J . Amer. Chem. SOL., 1929, 51, 1171. Sullivan, J . T., J . Ass. Og. Agric. Chem., 1939, 22, 781. Fisher, F. B., and Brown, J. S., Anal. Chem., 1952, 24, 1440. Feuersenger, M., Dtsch. LebensmittRdsch., 1959, 55, 277. Feigl, F., and Caldas, A., Mikrochim. Acta, 1955, 992. Joint Committee of the Association of British Chemical Manufacturers and the Society for Analyti- cal Chemistry, “Recommended Methods for the Analysis of Trade Effluents,” IV, Heffer & Sons Ltd., Cambridge, 1958. Aldridge, W. N., Analyst, 1944, 69, 262; 1945, 70, 474. Konig, W., Z. angew. Chem., 1905, 115; J . prakt. Chem., 1904, 69, 105. Krawczyk, D. F., Sewage I n d . Wastes, 1954, 26, 388. Bruce, R. B., Howard, J . W., and Hanzal, R. F., Ibid., 1955, 27, 1346. Saltzman, B. E., Anal. Chem., 1961, 33, 1100. Nusbaum, I., and Skupeko, P., Ibid., 1951, 23, 875. Baker, M. O., Foster, R. A., Post, B. G., and Hiett, T. A., Anal. Chem., 1955, 27, 448. Russell, F. R., and Wilkinson, N. T., Analyst, 1959, 84, 761. Tompsett, S. L., Clin. Chem., 1959, 5, 587. Piekacz, H., and Mazur, H., Roczn. ZakE. Hig., Warsaw, 1961, 12, 481; Anal. Abstv., 1962, 9, 2919. Mazur, H., and Piekacz, H., Roczn. Zakl. Hig., Warsaw, 1961, 12, 523; Anal. Abstr., 1962, 9, 3432. Murty, G. V. L. N., and Viswanathan, T. S., Anal. Chim. Acta, 1961, 25, 293. Kratochvil, V., Coll. Czech. Chem. Commun., 1960, 25, 299. Deshmukh, G. S., and Tatwawadi, S. V., J . Sci. I n d . Res. India, 1960, 19, 195. Schulek, E., Burger, K., and Fehdr, M., 2. anal. Chem., 1959, 167, 423. Epstein, J., Anal. Chem., 1947, 19, 272. Marsden, K., Analyst, 1959, 84, 746. Jaulmes, P., and Mestres, I<., A n n . Falsif., 1960, 53, 455. Dodge, B. F., and Zabban, W., Plating, 1953, 42, 71. Ludzack, F. J., Moore, W. A., and Ruchhoft, C. C., Anal. Chem., 1954, 26, 1784. Whiston, T. G., and Cherry, G. W., Analyst, 1962, 87, 819. New York and London, 1958, p. 276. 80, 1285; Anal. Abstr., 1960, 7, 4186. 1960, 7, 920. 3 , Zhur. Anal. Khim., 1959, 14, 303. -- Received March 7th, 1963
ISSN:0003-2654
DOI:10.1039/AN9638800751
出版商:RSC
年代:1963
数据来源: RSC
|
9. |
The analysis of14CO-14CO2mixtures by gas-chromatographic separation and aqueous solution counting |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 761-770
H. J. Cluley,
Preview
|
PDF (935KB)
|
|
摘要:
October, 19631 CLULEY AND KONRATH 761 The Analysis of 14C0 - 14C0, Mixtures by Gas-chromatographic Separation and Aqueous Solution Coun tings BY H. J. CLULEY AND J, H. KONRATH (The General Electric Company Limited, Central Research Laboratovies, Hirst Reseavch Centre, Wembley, Middlesex) A method is described for analysing carbon-14 monoxide - carbon-14 dioxide mixtures. Gas-chromatographic separation on a silica-gel column is used to determine both the percentage composition and the activity of each constituent. The carbon monoxide emerging from the column is oxidised to carbon dioxide and then frozen out in a cold trap t o which sodium hydroxide solution has been added ; the carbon dioxide component is isolated in a second cold trap. The activity of the solution from each trap is directly deter- mined by scintillation counting in a cell containing a layer of plastic phosphor cemented to the inside of its base.The rapid separation achieved by gas chromatography permits an analysis to be completed in 40 to 60 minutes, depending on counting time. Results are presented for carbon-14 monoxide - carbon-14 dioxide mixtures covering a range of compositions. IN connection with tracer studies involving the use of carbon-14, it has been necessary to analyse numerous mixtures of carbon monoxide and carbon dioxide in which both constituents were labelled. For each mixture it has been required to determine the percentage composition and also the amount of activity present in each chemical form, so that the specific activity of each constituent could be calculated.In our laboratories it has been customary for some years to analyse inactive gas mixtures containing carbon monoxide and dioxide by gas chromatography. For determining minor constituents, including carbon monoxide, in gases containing mainly carbon dioxide, the carbon dioxide has been removed by passage through soda-lime and the minor constituents separated on a molecular-sieve co1umn.l When a determination of carbon dioxide has been required, its separation from carbon monoxide has been achieved with a silica-gel column. Such analyses can be completed in a few minutes, a matter of some importance if many samples are to be handled. On the grounds of familiarity with the technique and the speed of analysis afforded, gas chromatography was adopted as the means of determining the percentage composition of the labelled gas mixtures. In addition, as determination of the individual activities of the carbon monoxide and carbon dioxide also necessitated their separation, it seemed obvious to take advantage of the separation of these constituents achieved in the gas-chromatographic column.For measuring the activity of the chromatographically separated components, various procedures were available. The activity of the gas from the column could be continuously monitored by a suitable radiation detector, operated in conjunction with a ratemeter and recorder. Alternatively, the active components could be separately trapped out from the gas stream; counting in gas, liquid or solid forms could then be applied as desired.The different techniques employed in the gas-chromatographic separation of radioactive compounds have recently been reviewed by Adloff .2 For the work described here, continuous monitoring of the effluent gas would have required additional specialised counting equipment, and this approach was rejected in favour of isolation of the active components. The carbon monoxide, which was the first to emerge from the column, was oxidised to carbon dioxide with heated copper oxide and then frozen out in a cold trap, to which sodium hydroxide solution had initially been added. The carbon dioxide component was later isolated in a similar trap. After the traps had been warmed to room temperature, the activity of the solution in each trap was determined by a direct counting method.to 17th, 1963. * Presented at the XIXth International Congress of Pure and Applied Chemistry, London, July 10th762 CLULEY AND KONRATH: ANALYSIS OF l4c0 - l4c0, MIXTURES BY GAS- [A,4%a&St, VOl. 88 This procedure has proved satisfactory in use over 2 years. The speed of the gas- chromatographic separation is of particular advantage, and an analysis can be completed in 40 to 60 minutes, depending on the time required for counting. The development of this procedure involved investigations into the gas chromatography of active mixtures and into the counting of aqueous solutions containing carbon-14. These investigations are described below, preceded by an account of the apparatus and of the analytical procedure finally evolved. APPARATUS GAS-CHROMATOGRAPHIC APPARATUS- The gas-chromatographic apparatus used was made in our laboratories and was similar to that previously described.l The analytical column was a 6-foot length of *-inch internal diameter stainless-steel tubing of 20 s.w.g.wall thickness, packed with 36- to 52-mesh silica gel that had been heated in vacuo before being packed into the column. A second packed column was used to obtain a parallel flow of carrier gas to serve as a reference gas for the detector. The columns were maintained at 100” C by an electrically heated steam-jacket. The carrier gas was argon, supplied from a cylinder and dried by passage over pellets of molecular sieve (Linde type 5A); the flow rate through each column was 30 ml per minute. The detector was a katharometer, made as previously described,l in which the sensitive elements were tungsten filaments.The signal from the katharometer, after attentuation by known factors, if required, was fed to a potentiometric recorder so that the heights of recorded chromatographic peaks could be measured. In addition, an attachment was made to the recorder so that the mechani- cal movement of the recorder pen produced, from a slide wire, a voltage proportional to the pen deflection ; application of the voltage produced to an indicating integrating motor (Electro- methods Ltd. , Stevenage) permitted automatic reading of the “afeas” of chromatographic peaks. The gas-sampling system, incorporating a valve of the type previously described,l is shown in Fig. 1. With the valve in the position shown, the carrier gas passed directly to the B A = Sampling valve body B = Groove in rotating valve head C = Detachable loop D = Manometer Fig.1. Gas-sampling system, showing valve in sampling position column; at the same time the sample inlet was connected to a detachable sample lo-op of known volume and thence to a mercury manometer and to a vacuum pump. The sample loop was evacuated, and then the sample gas was admitted to the desired pressure. Rotation of the valve through 60” then allowed carrier gas to pass through the loop and sweep the sample on to the column, with minimal interruption in the flow of carrier gas.October, 19631 CHROMATOGRAPHIC SEPARATION AND AQUEOUS SOLUTION COUNTING 763 The gas samples to be analysed were usually in silica bulbs provided with a break-seal.To permit portions of the gas to be withdrawn without admitting air, the break-seal end of a bulb was connected by pressure tubing to a sampling head of the type shown in Fig. 2. The sampling head had a side-arm connected via a tap to the gas-sampling system described above. The end of the sampling head was closed with a rubber serum cap through which passed a steel wire with a shaped end-piece. In operation, the sampling head was evacuated simul- taneously with the sample loop, and the tap between the sampling head and the gas-sampling valve was then closed. The steel wire was twisted to break the break-seal, and gas was admitted to the sample loop as required. TRAPPING SYSTEM- A two-way tap was fitted at the point where the gas stream from the analytical column emerged from the katharometer block.With the tap in one position, the gas passed directly to a trap cooled in liquid oxygen, to freeze out the carbon dioxide component. (Liquid nitrogen, if used, would condense the argon used as carrier gas.) With the tap in the other position, the gas passed over copper oxide a t 500” C, to oxidise carbon monoxide to carbon dioxide, To tap and sampling valve t C A = Bulb B = Pressure tubing C = Brass fitting D = Rubber serum cap E = Steel wire with shaped end-piece Fig. 2. Sampling head used with sealed bulbs and thence to a second cold trap to isolate the carbon monoxide component. The copper oxide was in the form of a roll of copper gauze, oxidised by prior heating in oxygen and con- tained in a length of silica tubing of 3 mm internal diameter that was heated over a length of 4 inches by an external winding of Nichrome wire.To facilitate control of the apparatus, the volumes of tubing (glass and poly(viny1 chloride)) between the katharometer block and the traps were kept small, so that only a few seconds would elapse between emergence of a con- stituent from the column (as detected by the katharometer) and its arrival in the appropriate trap. The traps, of borosilicate glass, were of the type shown in Fig. 3; their design was based on that of traps used to freeze out small amounts of carbon dioxide in the determination of carbon in steel by the “low pressure” r n e t h ~ d . ~ Sodium hydroxide solution was added to each trap before connection to the apparatus, so that the trapped active component was finally obtained as an alkaline aqueous solution.In use the traps were immersed in liquid oxygen up to the bottom of the joint. A vacuum grease was used on the joint of each trap, and, in addition, the joint was held in place with springs (not shown in Fig. 3). There was thus no risk of loosening the joint when the traps were held by the inlet and exit tubes for the purpose of manipulating the traps in and out of the liquid oxygen. COUNTING CELL AND COUNTING EQUIPMENT- The active aqueous solutions obtained in the traps on re-warming to room temperature were counted in a specially constructed cell, which is shown in Fig. 4. The main body of the cell was turned from 2-inch diameter Perspex rod. The base consisted of a disc of NE102 plastic phosphor, 0.005 inch thick (Nuclear Enterprises Ltd.), cemented to a &-inch thick Perspex disc; amyl acetate was used as the cementing solvent.The base was in turn cemented to the main body of the cell. The dimensions of the cell were chosen to permit use in an Ekco764 CLULEY AND KONRATH: ANALYSIS OF l4c0 - l4Co2 MIXTURES BY GAS- [Analyst, VOl. 88 N664 scintillation counter, operated in conjunction with an Ekco N530 scaler. It was found appropriate to use the same instrumental settings as used for scintillation counting of carbon-14 in toluene-2,5-diphenyloxazole solutions, namely : E.H.T., 1100 volts ; gain, 1000 x ; discriminator setting, 17 volts. METHOD FOR ANALYSING 14co - 1 4 ~ 0 , MIXTURES PROCEDURE- The bulb containing the gas sample to be analysed was connected to the sampling head (see Fig.2), and the sample loop and sampling head were evacuated with the vacuum pump. The tap between the pump and the sampling system was then turned off and the system allowed to remain evacuated for several minutes to check for possible leaks. Meanwhile, 5 ml of 0.1 N sodium hydroxide were placed by pipette into each of two dry traps, the traps were re-assembled and their taps closed. Each trap was dipped for a few seconds into liquid oxygen, removed and swirled, and the process was repeated until the sodium P-l I inch Fig. 3. Trap for freezing out carbon dioxide i I: -1: Ii i nc h es -{ inch inch Phosphor disc, b m’O.005 inch thick 1- inchtJ :> Perspex disc’ inch thick Fig. 4. Details of construction of counting cell hydroxide solution had frozen in a thin layer on the wall of the trap tube.(Freezing in this manner avoided cracking the tubes.) The two traps were connected to the gas-chromato- graphic apparatus, immersed up to the bottom of the joint in liquid oxygen, and in turn purged of air with the flow of carrier gas from the column. The taps of the carbon dioxideOctober, 19631 CHROMATOGRAPHIC SEPARATION AND AQUEOUS SOLUTION COUNTING 765 trap were then closed, and carrier gas was allowed to continue flowing through the carbon monoxide trap. The break-seal of the sample bulb was then broken, and gas was admitted to the sample loop to the required pressure as shown on the manometer. (With a sample loop of volume about 1.5 c.c., sample pressures of about 200 mm of mercury were generally used, if sufficient sample was available).The sample gas in the loop was then injected into the column by operating the gas-sampling valve. The air temperature near the sample loop, assumed to be the same as that of the sample, u7as noted at this point. Any carbon monoxide present in the sample was shown by a peak on the recorder chart occurring l a minutes after injection of the sample. After a further 2 minutes, i e . , 3 to 3; minutes after injection, the taps were operated to divert the effluent gas from the column to the carbon dioxide trap, and the taps of the carbon monoxide trap were closed. The carbon dioxide peak was recorded about 4; minutes after injection, and the gas stream was allowed to flow through the carbon dioxide trap for a further 2 minutes after the peak had been recorded.While the carbon dioxide was being eluted, the exit tube of the carbon monoxide trap was connected to the vacuum pump via a capillary “leak.” The exit tap of this trap was opened for a brief period, then closed and the vacuum connection removed. (It was necessary to remove most of the argon present in the trap, so that undue pressure would not result on subsequent re-warming of the trap to room temperature; the capillary leak restricted the flow of gas, so that small particles of solid carbon dioxide would not be blown out of the trap.) The trap was removed from the liquid oxygen and re-warmed to room temperature by im- mersion for a few minutes in warm water. Finally, the trap was shaken mechanically for 5 minutes to dissolve any carbon dioxide that was now in the gaseous form.The same series of operations was applied to the carbon dioxide trap. For counting, most of the solution from a trap was transferred to the counting cell by means of a dry pipette (quantitative transfer is unnecessary, but dilution must be avoided). Fifty milligrams of a dye, Naphthalene Leather Carbon G.S. (Imperial Chemical Industries Ltd.), were added to the cell, and the solution was swirled to dissolve the dye. The cell was then transferred to the scintillation counter and the solution counted. Each count was corrected for the background count of the cell, determined by counting 5 ml of 0.1 N sodium hydroxide with 50mg of dye in the cell. The background count was measured for each series of determinations, usually during the preliminary preparations of the chromatographic apparatus. By having traps available, replicate analyses could be performed in rapid succession, the solutions from one analysis being counted during the chromatographic stage of the subsequent analysis.After use, the traps were well washed with water, then with acetone, and dried by drawing air through them. After each count the counting cell was washed well with water and dried with tissues; the cell was then ready for re-use. CALCULATION OF RESULTS AND CALIBRATION- The percentage of the minor constituent (usually carbon monoxide) was obtained by correcting the observed peak heights, peak areas, or both, to a standard sample pressure and temperature, then reading off the percentage from a calibration graph prepared by chromato- graphic separation of inactive gas mixtures of known composition.The percentage of the major constituent was usually obtained by difference. (Traces of nitrogen and hydrogen were sometimes present, but invariably in insignificant amounts.) As discussed later, the counting technique used was such that the observed count rate, corrected for background, was proportional to the concentration of carbon-14 in the solution, regardless of the volume of solution added to the cell. The cell factor f (the net count rate given by a carbon-14 concentration of 1 pC per ml) was determined by counting a solution of sodium l*C-carbonate of certified activity (nominally 1 pC per g) obtained from the Radio- chemical Centre, Amersham.If the solution from a trap gave a count rate r (corrected for background), then the total activity (in pC) in the trap was 5r/f, the total volume of the trap solution being 5 ml. The specific activity of a constituent, in pC per standard c.c., was then given by the expression- At least two, usually three, analyses were made on each gas sample. 5r 760 (273 + t) 100 - x - f I r x p ~ 2 7 3 ~ P ’’66 CLULEY AND KONRATH: ANALYSIS OF l4c0 - 14C02 MIXTURES BY GAS- [AnaZYSt, VOl. 88 vhere P was the percentage of the constituent, and V in c.c., 9 in mm of mercury and t in legrees C were, respectively, the volume, pressure and temperature of the gas sample. Of hese quantities, V was kept constant by using the same sample loop throughout the work.Isolation of the active constituents as aqueous solutions was adopted because direct :ounting of the solutions appeared to offer the minimum amount of manipulation compared vith counting in gas or solid form. The counting technique used was essentially that proposed ~y Jenkinson4 and which appeared particularly appropriate to the work described here. In the counting cell made by Jenkinson and in the cell used in this work, the scintillations letected by the counter occur in the thin layer of plastic phosphor that forms the inner side if the base of the cell. Because of the short range in water (< 1 mm) of the soft beta radia- :ion of carbon-14, only the activity present in a thin layer of solution adjacent to the phosphor :ontributes to the observed count rate.This resembles the condition of “infinite thickness’’ n solid counting, from which it might be expected that the count rate would be proportional :o the concentration of activity in the solution and be independent of the volume. However, Jenkinson observed some variation of count rate with different volumes of the same solution. He attributed this to variations in the extent to which light emitted upwards by the phosphor ayer was re-directed to the counter by internal reflection in the solution. Jenkinson there- ‘ore added a dye to eliminate transmission of light through the solution; the count rate was :hen proportional to the specific activity of the solution and independent of the volume :ounted. The shape of the inner member of each trap, lesigned to give a large cold surface, was such that quantitative transfer of a solution from a :rap would inevitably result in considerable dilution due to the addition of washings.By ising Jenkinson’s counting technique, any indefinite volume of the trap solution could be :ransferred to the cell for counting, provided, of course, that sufficient solution were added :o cover the base of the cell. Some experiments on this counting technique, relevant to the analysis of carbon-14 nonoxide - carbon-14 dioxide mixtures, are described below. PRELIMINARY EXPERIMENTS- The first experiments were directed toward establishing the conditions under which :omistent count rates could be obtained for a solution, independent of the volume counted. ?or this purpose, different volumes of sodium 14C-carbonate solution, with and without dye, vere counted.As recommended by Jenkinson,* the dye used was Naphthalene Leather Zarbon G.S. The results obtained (see Table I) confirmed Jenkinson’s observations* that count rate Jaried with volume in the absence of dye, and that count rates independent of volume were COUNTING OF CARBON-14 IN AQUEOUS SOLUTION This property is an advantage in our work. TABLE I PRELIMINARY EXPERIMENTS ON COUNTING OF Na214C03 SOLUTION Volume of solution counted, ml 3 5 10 3 3 3 5 7.5 10 Concentration of dye, Yo w/v 0 0 0 0-05 0.1 0.15 0.1 0.1 0.1 Count rate, C.P.S. 187.2 183.9 180.1 176.0 178.0 177.7 177.2 177.2 176.7 Ibtained with 0.1 per cent. w/v of dye, this concentration not being critical. :hat use of the dye only slightly reduced the counting efficiency.ZFFECT OF CONCENTRATION OF ALKALI- :ion in count rate, presumably by a self-absorption effect. It was apparent Jenkinson4 reported that increased concentrations of inactive solute caused some reduc- In the work described in thisOctober, 19631 CHROMATOGRAPHIC SEPARATION AND AQUEOUS SOLUTION COUNTING 767 paper, the effect on count rate of different concentrations of alkali was investigated. A dilute solution of sodium 14C-carbonate was counted after various amounts of inactive sodium hydroxide or carbonate solution had been added, the observed count rates being corrected for the resulting dilutions; the concentration of dye was 0.1 per cent. w/v in all solutions counted. The results in Table I1 showed a progressive reduction in corrected count rate with increasing concentration of alkali.It was however apparent that if 0.1 N solutions were used, appreciable relative variations in concentration would have negligible effect on count rate. For absorption of the carbon dioxide from the size of gas sample customarily used (1.5 C.C. at 200 mm pressure), 5 ml of 0-1 N sodium hydroxide would provide a substantial excess, and 0.1 N sodium hydroxide was adopted as the medium for counting. TABLE I1 EFFECT OF INACTIVE ALKALI CONCENTRATION ON COUNT KATE OF AN Na,14C0, SOLUTION Concentration of added alkali Count rate, yo of count rate Alkali added in solution counted, N without added alkali 0.10 99.8 0.17 99.2 0.27 98.8 0.50 98.8 1.0 97.8 0.10 99.8 0.18 99.4 0.33 98.9 0.67 97.5 -1 1.0 96.9 -1 Sodium hydroxide Sodium carbonate CALIBRATION AND COUNTING EFFICIENCY- Calibration of the cell has been effected, as described on p.765, by counting a sodium 14C-carbonate solution of certified activity. The first cell used in this work gave, on first calibration, a cell factor of 330 C.P.S. (the count rate for 1 pC of carbon-14 per ml). A second cell gave a similar factor. As the same count rate is obtained with different volumes of the same solution, the counting efficiency achieved depends on the volume of solution counted. With 5ml of solution, the cell factor quoted above corresponds to a counting efficiency of 0.2 per cent. This relatively low efficiency was no disadvantage in the work described here, the main virtue being the simplicity of the counting technique.The gas samples to be analysed contained several pC of carbon-14 per standard c.c., so that sizable count rates, e.g., of the order of 100 C.P.S. for the main constituent, were usually obtained. With a cell in almost daily use, it has been found that the base tends in time to become slightly opaque, possibly owirrg to weakening of the adhesion between the plastic phosphor layer and the underlying Perspex. Possibly for this reason the counting efficiency of a cell decreases in time, and it is advisable to re-calibrate a cell every few months. Jenkinson stated that high background count rates with his cell could arise either from undue exposure of the cell to light or from use with neutral or acid solutions, hence his recom- mendation that all solutions be made alkaline with sodium hydroxide before- ~ounting.~ In one cell used by us, the latter effect was observed in attempting to count a solution containing carbon-14 labelled sucrose.The former effect has not been observed; handling of the cell in daylight or normal lighting conditions has invariably given background counts indistinguish- able from that of the scintillation counter with no sample present (0-5 to 0.6 c.P.s.) The precaution has, however, been taken of storing the cell in the dark when not in use. GAS CHROMATOGRAPHY OF THE LABELLED GAS MIXTURES In developing the gas-chromatographic part of the procedure described, particular attention was directed to assessing the efficiency of the trapping system, to examining the possibility of “trailing” of active constituents from the column and to determining the precise volume of gas sample analysed.EFFICIENCY OF THE TRAPPING SYSTEM- The efficiency of the cold traps for freezing out carbon dioxide was tested by putting samples of carbon-14 dioxide through the gas-chromatographic procedure, two traps in series being used. In two runs on slightly different amounts of sample, count rates of 48-3 and 45-5 C.P.S. were obtained for the solutions from the first trap, corresponding, respectively, to These points are discussed below.768 CLULEY AND KONRATH: ANALYSIS OF l4c0 - 14c02 MIXTURES BY GAS- {Analyst, VOl. 88 2.56 and 2.57 pC per standard C.C. for the specific activity of the gas; in each test the second trap yielded no detectable activity (< 0.1 c.P.s.).These and other results showed that a single cold trap gave quantitative removal of carbon dioxide from the steam of carrier gas. However, to ensure complete removal of the carbon monoxide component, it was also necessary to achieve its complete oxidation to the dioxide on passage through the heated copper oxide. The efficiency of the oxidation process was checked by putting samples of inactive carbon monoxide through the gas-chromatographic process and collecting the carrier gas emerging from the copper oxide tube during the period when the carbon monoxide was being eluted from the column. Analysis of the collected gas showed that quantitative oxidation of the carbon monoxide was being obtained. EXAMINATION FOR TRAILING EFFECTS- Under the conditions employed in the method, the peaks of carbon monoxide and dioxide occurred, respectively, at lt and 44 minutes after injection of the sample.For the first 3 or 3; minutes after injection, the gas from the column was passing into the carbon monoxide trap; thereafter, until 2 minutes after the end of the carbon dioxide peak (as observed on the recorder chart) the gas was passing into the carbon dioxide trap. With this system, any trailing, L e . , the slow release of a constituent from the column after the main peak had passed, could result in incomplete recovery of the activity. In addition, trailing of the carbon dioxide could result in activity from this constituent being included in the carbon monoxide fraction of a succeeding analysis. It was therefore important to ascertain whether or not trailing occurred to any significant extent. With a comparatively new column being used, a sample of carbon-14 dioxide was passed through the gas-chromatographic apparatus, the carbon dioxide trap being replaced at intervals after the peak had been recorded.Only an extremely small amount of activity, negligible when compared with the total amount present, was recovered after the normal period of collection (see Table 111). A similar test with carbon-14 monoxide, though less sensitive because of the smaller amount of activity used, showed no evidence of trailing (see Table 111). I t was apparent from these results that trailing was unlikely to be a significant source of error when a fairly new column was used. It is, however, well known that trailing ma>- become more evident with continued use of a column.In this and related work it has been found that trailing and other adverse effects may necessitate renewal of the columns after about 6 months. (The life of a column appears to be largely a function of time, rather than of frequency of use, possibly because of our practice of maintaining a slow flow of carrier gas through a column when it is not in use.) It may be of interest to note that chromatography of active gases can be made an extremely sensitive means of detecting trailing. TABLE I11 EXAMINATION FOR TRAILING EFFECTS Period of collection Count rate of trap solution Sample used of active constituent corrected for background, C.P.S. Carbon-14 monoxidc . . Up t o 2 minutes after CO peak 13.4 2 to 5 minutes after CO peak <o-1 5 t o 10 minutes after CO peak (0.1 Carbon- 14 dioxide .. Up t o 2 minutes after CO, peak 2 to 5 minutes after CO, peak 5 t o 10 minutes after CO, peak 159 0.2 <0.1 DETERMINATION OF SAMPLE VOLUME- The same sample loop, of measured volume 1.42 c.c., was used for all determinations. However, this volume did not represent the total volume of gas sample analysed, a small additional amount of gas being present in the connections inside the gas sampling valve (see Fig. 1). This additional volume was known, from the construction of the valve, to be about 0.15 C.C. For determining the percentage composition of gases, the presence of this additional volume of sample was irrelevant, as the same total volume of gas would be used both in analysis and in calibration with gases of known composition.However, to permit c&uhtion of the specific activity of constituents, in pC per standard c.c., it was essential to know the precise volume of gas sample analysed.October, 19631 CHROMATOGRAPHIC SEPARATION AND AQUEOUS SOLUTION COUNTING 769 Samples of active and of inactive carbon dioxide were submitted to the chromatographic process, with the use of both the normal sample loop and a small loop of volume (0.253 c.c.) similar to that in the valve. At a given sample pressure, the ratios of peak areas and of amounts of activity found were assumed to be proportional to the ratio of sample volumes used, i.e. (1.42 + x)/(0-253 + x), from which x, the volume of gas in the valve, could be calculated.Four such determinations gave values of 0-15, 0.15, 0.18 and 0.19, the mean value of 0.17 C.C. being taken. This value was consistent with the expected volume of about 0.15 C.C. With the normal loop, of volume 1-42 c.c., the total sample volume was therefore taken to be 1.59 C.C. This value was thought to be accurate to within about 1 per cent. Because of the expected accuracy of the activity measurements, a much greater accuracy in measuring the precise sample volume would have been superfluous. The additional volume of gas in the valve was determined indirectly. The results obtained on a 117, in which sample pressures RESULTS sample of carbon-14 dioxide are shown in some detail in Table and temperatures and also the observed count rates are listed. TABLE IV RESULTS ON A SAMPLE OF 14C02 Count rate Specific activitv Sample Count rate pressure, Temperature, of trap solution, mm of Hg "C C.P.S.151 30 48.9 142 30 46.1 92 31 28.8 135 26 44.2 101.5 31.5 32.6 corrected for background, C.P.S. 48-3 45-5 28-2 43-6 32.0 of co,, PC Per standard C.C. 2.56 2.57 2-46 2.55 2.54 This particular sample was of lower specific activity than normally used, giving rise to lower count rates than normally observed. The sample, in which no carbon monoxide was detected, was subjected to the analytical process primarily to assess the precision obtainable in that part of the process involving the trapping of the carbon dioxide and the measurement of activity . To illustrate the precision obtained in the analysis of mixtures, in which an additional possible source of error arises from the determination of percentage composition, the results of analyses of a typical series of samples are shown in Table V.Each figure quoted for the TABLE V RESULTS FOR A TYPICAL SERIES OF 14co - 1 4 ~ 0 , MIXTURES Sample number 1555 1552 1553 1547 1549 1546 1548 Carbon monoxide present, yo 30.5 31.3) 30'9 it::] 13.8 13.8J 10.1 10.0 J 9.9)- 10.0 7-8 7.1 7.15) 6.65 g:!:} 5-6 Specific activity of CO (A), pC per standard C.C. 5'2 } 5-2 5.2 5.75 5.75 6.3 6-0 ::? 6.2 }6*05 6*o } 6.0 6.0 Specific activity pC per standard C.C. of co, (B), 5-29) 5.3 5.3 1 5.96 6.10 6.38 6.16 6.41 6*50} 6.45 Ratio of specific activities 1 so2 B/A 1-04 1.05 1.02 1.04 1.03 1.07770 CLULEY AND KONRATH [AnaZyst, Vol. 88 percentage carbon monoxide was the mean of the two values deduced from the height and from the area of the peak; for calculating the specific activity of the carbon dioxide, the percentage of carbon dioxide was taken by difference. The gas samples referred to in Table V were from reactor-irradiated bulbs that had originally contained only graphite and labelled carbon dioxide, the carbon monoxide found being a product of the radiation-induced reaction between graphite and carbon dioxide. The carbon monoxide, being partly derived from the carbon of the unlabelled graphite, would be produced at a specific activity lower than that of the carbon dioxide; in addition, interchange of activity between the two gaseous constituents would arise from the radiation- induced exchange reaction-14C0, + CO + 14C0 + CO,. At equilibrium, the ratio of specific activities of the two constituents is equal to the equilibrium constant for the exchange reaction. The results in Table IT show essentially constant values for the ratio of specific activities of carbon dioxide and carbon monoxide, the mean ratio being 1-04. This figure is in excellent agreement with the value of 1-03 -+ 0.008 for the equilibrium constant of the exchange reaction, reported by S t r a n k ~ . ~ The support of the U.K.A.E.A. is gratefully acknowledged. REFERENCES 1. 2. 3. 4. 5. Timms, D. G., Konrath, J. H., and Chirnside, R. C., Analyst, 1958, 83, 600. Adloff, J. P., J . Chromatography, 1961, 6, 373. Wells, J. E., J . Iron 15 Steel Inst., 1950, 166, 113. Jenkinson, D. S., Nature, 1960, 186, 613. Stranks, D. R., in Hurst, R., Lyon, R. N., and Nicholls, C. M., Editors, “Progress in Nuclear Energy,” Series IV, Pergamon Press, Oxford, London, New York and Paris, Volume 2, 1960, Received March 4th, 1963 p. 54.
ISSN:0003-2654
DOI:10.1039/AN9638800761
出版商:RSC
年代:1963
数据来源: RSC
|
10. |
The determination of oxygen in vacuum-melted steels, molybdenum and single-crystal silicon by vacuum fusion |
|
Analyst,
Volume 88,
Issue 1051,
1963,
Page 771-781
P. D. Donovan,
Preview
|
PDF (1280KB)
|
|
摘要:
October, 1963; DONOVAN, EVANS AND BUSH 771 Vacuum-melted The Determination of in Oxygen Molybdenum and Single-crystal Silicon by Steels, Vacuum Fusion BY P. D. DONOVAN, J. L. EVANS AND G. H. BUSH (The War Ojice, Royal Armament Research and Development Establishment, Fort Halstead, Sevenoaks, Kent) The equipment and techniques used for determining oxygen in vacuum- melted steels, mol$bdenum and single-crystal silicon by vacuum fusion, at levels down to less than 1 p.p.m., are described. The oxygen content of steels at levels of 115 p.p.m. and 6 p.p.m. has been determined with coefficients of variation of 2.3 and 5.2 per cent., respectively. Results on sintered and arc-cast molybdenum with oxygen levels of 42 p.p.m. and 2 p.p.m. showed coefficients of variation of 2 and 8 per cent., respectively.Experiments with zone-refined molybdenum indicated that the oxygen content consisted of a residual surface contamination of 0-8 pg per sq. cm with an internal oxygen content of less than 0.2 p.p.m. With vacuum grown single-crystal silicon a coefficient of variation of 10 per cent. at a level of 17 p.p.m. was obtained on a limited sample weight of 0.5 g. A COMPREHENSIVE account of the vacuum fusion method of analysis and its applications has been given by S1oman.l to In recent years developments of the technique have been concentrated chiefly along two lines of research, viz., analyses of the rarer metals (beryllium, molybdenum, tantalum, titanium, tungsten, etc.) and greater accuracy and precision in determining low levels of oxygen content.Progress in the latter field, however, still lags behind metallurgical requirements, and dissatisfaction has been expressed8 9 9 at the lack of precision and accuracy obtainable at low levels of oxygen content (0 to 20 p.p.m.) in the vacuum-fusion analysis of steel and molybdenum. Little evidence is available in the literature for reproducibilities of determinations of oxygen of below 100 p.p.m. Lassner and Wolfello have obtained a coefficient of variation of 6.1 per cent. on six samples of sintered molybdenum with a mean value of 30 p.p.m., and Mallet and Griffithll have demonstrated the internal consistency of the method below 40 p.p.m. to within 1 p.p.m. on samples of molybdenum spiked with molybdenum powder. McDonald and Fagel12 have determined oxygen in molybdenum by vacuum extraction at 2000" C and by vacuum fusion in a tin - iron bath.By vacuum extraction they obtained a mean of 7.6 p.p.m. on eleven samples with a coefficient of variation of 36 per cent. ; by vacuum fusion the coefficient of variation was 21 per cent. on a mean of 7.3 p.p.m. on four samples of the same material. Many of the results quoted indicated that the method was reliable for steel, although, recently, Pearce and Masson13 working on the method of isotopic dilution have produced evidence suggesting that the results obtained by vacuum fusion analysis of steels were seriously in error below 200 p.p.m. owing to adsorption of the evolved gases on the cooler parts of the apparatus. However, their experiments do not appear to be conclusive, and their conclusions conflict with the main body of e v i d e n ~ e ~ $ ~ * ~ * J ~ and our experience. Recent work by Coleman16 on fast neutron-activation analysis has also confirmed the reliability of vacuum-fusion analysis on two of the samples quoted by Pearce and Masson.The object of the work described here was to develop the technique for determining oxygen in vacuum-melted steels, sintered, arc-cast and zone-refined molybdenum and single- crystal silicon, all with an oxygen content below 100 p.p.m. It was considered that, to meet the needs of the metallurgist, a method having a coefficient of variation of not more than 10 per cent. was required at levels down to 1 p.p.m.772 GENERAL- The vacuum fusion apparatus (see Fig. 1) was similar in its basic design to that described by Gray and Davis.17 Use was made of Sloman’s design of crucible and furnace assembly and the principles of Ransley’s low pressure analysis system.ls~lg The measuring system has been described previously by Baconz0 and was designed to measure volumes of gas from 0.005 to 2.0 ml with a precision of 1 per cent.by using a McLeod gauge and a series of expansion volumes that permitted the pressure to be kept within that of the maximum precision of the McLeod gauge by varying the capacity of the collecting volume. DONOVAN, EVANS AND BUSH: DETERMINATION OF OXYGEN [Analyst, Vol. 88 APPARATUS MEASURING AND ANALYTICAL SYSTEMS- The minimum collecting volume normally used was 917 ml, although the smaller volume of 692 ml could be used by excluding the McLeod gauge and using the Pirani for pressure measurements.This was done for determining the blank rate or for analysing small samples (below 2 g) with extremely low contents of gas (below 2 p.p.m.). The collecting volume could be increased to 21 litres in steps, increasing each time by a factor of approximately 2, by means of expansion volumes. The collecting volumes were calibrated volumetrically by adding water and checked by means of expansion ratios. The results agreed to within 1 per cent. Two pressure gauges were used, a McLeod for making accurate measurements and a Pirani for making rapid measurements, determining precise pressure ratios and registering the gas evolution and so following the course of the reaction. The McLeod gauge had two pressure ranges, 0 to 70 x loh3 to r and 0 to 300 x The fine scale had both linear and square-law calibration.The linear range, which was based on a compression ratio of 2000 to 1, had 1-mm graduations corresponding to 0-5 x torr pressure increments. Repeated observations of the pressure with this scale showed the reading error t o be no more than 0.5 mm (0-25 x torr the gauge had, therefore, a maximum reading error of 1 per cent. The Pirani gauge had a normal G.E.C. gauge head and was operated with a potential of 4 volts across the bridge. The out-of-balance current was recorded on a milliammeter with 1 mA full-scale deflection and an eye sensitivity of 1 PA. Three pressure ranges were provided by means of shunts. The first two were linear over their full ranges, which were 0 to 7 x 10-3 torr and 0 to 35 x torr, and the third scale, which was non-linear, was used merely as a guide for pressure in excess of 35 x torr.Pressure ratios, obtained by readings of near full-scale deflection on the linear scales, had a theoretical reproducibility of 0.1 per cent., and results close to this were obtained in practice. For readings less than full scale, correspondingly larger percentage variations were of course obtained. In practice, the coarser scale (0 to 35 x torr) was preferred, as slight pressure changes due to adsorp- tion and mechanical disturbances had less effect at the higher pressures. torr. torr). Used in the range 25 to 70 x N E W FEATURES AND IMPROVEMENT OF TECHNIQUE AND DESIGN- A summary is given below of the new features introduced and modifications in technique and design.( a ) The furnace head was re-designed in Pyrex glass and a fast mercury diffusion pump (rated at 30 litres per second) joined into the head with wide bore tubing (greater than 45mm internal diameter) to allow rapid removal of the gases from the furnace tube. A positive means of guiding samples into the crucible was provided, and the optical flat was largely screened from the crucible to avoid blackening by carbon and metallic films. (b) An ultra-pure grade of graphite was used (208 spectrographic, obtainable from Le Carbone (Gt. Britain) Ltd., 64 Finsbury Pavement, London, E.C.2). This allowed more rapid de-gassing and lower ultimate blank rates than could be obtained with normal “pure” graphite. (c) A vacuum lock was incorporated to replace the mercury lift used in earlier work.This provided a rapid and convenient means of introducing all types of samples directly into the high-vacuum system. The lock was a scaled-up version of that des- cribed by Parker,z1 with the addition of vacuum connections to the gaps between theAgqalyst, Vol. 58: faciag page 7721October, 19632 IN STEELS, MOLYBDENUM AND SILICON BY VACUUM FUSION 773 O-rings. Little air entered the system with the samples, and no change was detected in the furnace blank rate after the lock had been used. (d) A silica furnace tube in which samples could be heated to 1000” C was attached to the sample storage arm. Pre-heating removed much of the surface contamination and also greatly reduced the hydrogen content of samples.This allowed a more precise and accurate determination of the oxygen content, particularly on samples containing relatively large amounts of hydrogen or having an extremely low oxygen content. (e) A gas “doser” was connected to the system to introduce different amounts of a gas of known composition into the apparatus to study gettering problems. This apparatus consisted of a 180-ml glass bulb connected to two smaller bulbs of 0.71- and 0-18-ml capacity, by good quality vacuum taps. The secondary volumes were so arranged that either 1/1000 or 4,/1000 of the content of the main bulb could be frac- tionated into the vacuum-fusion system. (f) ,4 series of eight storage bulbs, each of about 110-ml capacity, was connected to the gas-collection system through a mercury diffusion pump.Gas samples could thus be stored during a run and analysed at a more convenient time. A silica furnace tube was also connected to this diffusion pump so that the equipment could be used for determining hydrogen by vacuum extraction. This system replaced the water cooling that had been used previously. Little difference was noticed between the two methods, but it seemed preferable to allow the walls of the furnace tube to reach the higher temperatures attained by air cooling as they would be less likely to absorb gases. (h) Two types of induction heater were used; one of 8 kW operating at a nominal frequency of 750 kc/s and the other of 17 kW at a nominal frequency of 5 kc/s. The latter unit was preferred as the lower frequency induced a stirring action in the melt giving a more rapid evolution of gas.Low blank rates were obtained more consistently with this lower frequency as there was no tendency for the graphite powder to pick up induced currents as occurred with the higher frequency, which at times gave rise to “hot spots” and consequent higher blank rates. (i) High-vacuum taps of good quality were used throughout the system in preference to mercury cut-offs. A liquid nitrogen trap was kept on the analytical system while the apparatus was in use, and this served to remove grease vapours, water, carbon dioxide and other condensable substances from the system. ( j ) A series of four expansion volumes as from 0.7 to 11 litres was incorporated into the collecting system so that the volume could be adjusted to bring the pressure measurements within the region of linearity of the Pirani gauge and close to the pressure of maximum precision of the RlcLeod gauge.(g) The vacuum-fusion furnace tube was air cooled. EXPERIMENTAL DE-GASSIXG- The method of packing the graphite powder and the initial application of heat to the crucible were found to be critical operations, on which the successful de-gassing and the final blank rate were dependent. Graphite powder (less than 325 mesh) was first compacted in the narrow extension of the furnace tube, to a depth of 0.5 inch. The crucible assembly was then placed in position and held centrally in the tube by means of a jig that fitted into the throat of the crucible. Graphite powder was packed around the crucible as loosely as possible.The furnace tube was then sealed into the head with vacuum wax, the induction heater and air cooling system were switched on, and the crucible was heated to approximately 1400” C. (All temperatures were measured with a disappearing filament optical pyrometer and are uncorrected.) The pressure in the furnace was then reduced by means of the “rough” vacuum pump at a rate of 3 cm per minute to a pressure of 10-1 torr, after which the furnace diffusion pump was switched on and the pumping routed through the analytical system. The temperature of the crucible was then raised to 2200” C and the apparatus allowed to de-gas for 3 hours.774 [Analyst, Vol. 88 BLANK RATES- The procedures outlined above gave a final blank rate of approximately 1 pl per minute at 2200” C.When the temperature was dropped to 1750” C the blank rate fell to 0.3 pl per minute. Lower blank rates could be obtained by a more prolonged de-gassing period, but were difficult to maintain during a “run.” A blank rate of 0.3 to 1 p1 per minute was there- fore accepted as a working level and was usually maintained if the samples added were of low gas content. The “blank” gas was fairly constant in quantity and composition, and the results were corrected for this additional gas if it amounted to more than 1 per cent. of the total gas collected from the sample. A typical analysis of “blank” gas gave 75 per cent. of carbon monoxide, 7 per cent. of hydrogen and 18 per cent, of nitrogen. DONOVAN, EVANS AND BUSH: DETERMINATION OF OXYGEN GAS AXALYSIS- A sample of the gas to be analysed was isolated in the smallest volume (the analytical section) at a pressure of 25 to 30 x torr (k., nearly full-scale deflection on the coarse Pirani scale).To obtain this reading the pressure of the gas sample could be either reduced by appropriate expansion or increased by pumping the gas from the volumes in which it was contained into the gas storage bulbs and then back into the analytical section. Small adjustments were made by varying the level of the mercury in the McLeod gauge. The first stage of the analysis was carried out by exposing the gas to Hopcalite reagent contained in the small bulb attached to the analytical system. This oxidised the carbon monoxide to carbon dioxide, which was then removed from the system by condensation in the cold finger (-196” C).The palladium thimble was then heated to 450” C and the hydrogen allowed to diffuse out. The percentages of the three gases present were calculated from the pressure changes on the Pirani gauge. For this calculation the recorded decrease in pressure due to hydrogen was multiplied by a factor of 0.65, since the Pirani has a higher sensitivity for hydrogen than for carbon monoxide and nitrogen, which have similar sensitivities. A volume of 0-010ml at S.T.P. was taken for each analysis, 2 minutes were allowed for removal of carbon monoxide and 1.5 minutes for hydrogen diffusion. The time taken for a complete analysis was 4 minutes. The precision of the analytical system was tested in practice on gas mixtures of known composition, and the over-all coefficient of variation for oxygen was found to be less than 1 per cent.when carbon monoxide formed more than 50 per cent. of the gas collected. The gas remaining was assumed to be nitrogen. USE OF TIN AS AN “ANTI-GETTER”- The use of tin as a flux or to overcome gettering is well known, and our experience has confirmed the efficacy of this technique. The addition of tin (0.1 g) at intervals has been found to overcome the effect of gettering during vacuum fusion analysis, even with steels containing relatively large amounts of manganese, copper, vanadium and other alloy additions, possibly by blanketing the deposited films. Other theories have been advanced,22 but the mechanism is still somewhat obscure. ASSESSMENT OF ACCURACY Direct tests of the accuracy of the method at low levels of oxygen, which were of interest, by preparing and analysing samples of known oxygen content were not considered feasible.Spiking samples with known weights of oxide, although not simulating exactly the conditions encountered in the analysis of a sample, gave an assessment of other stages of the process and was some indication of the accuracy to be expected. In this method the choice of a suitably prepared metal envelope having a low oxygen content, to contain the additions, minimised the objections to this type of evaluation. EXPERIMENTS ON SAMPLES WITH KNOWN AMOUNTS OF OXIDE ADDED- Two series of experiments were carried out, one with a steel capsule as a container for silica and the other with nickel foil to contain additions of tantalum oxide.The mean oxygen content of the capsules and the nickel foil was determined by carrying out ten determinations on each material.October, 19631 IN STEELS, MOLYBDENUM AND SILICON BY VACUUM FUSION 775 Small fragments of drawn silica filament were used for the first series of experiments, and tantalum oxide, prepared from pure tantalum wire ignited at 1000" C in air under condi- tions in which the form of the wire was retained, was employed for the second series. The increase in weight of the tantalum was found to be that required by the reaction Ta --f Ta,O,. The results of these experiments are shown in Table I. TABLE I Oxide added Silica . . Tantalum oxide RESULTS Amount of Oxide content determined Equivalent to oxygen in oxide added, by vacuum fusion, metal on 10-g sample, Pg Pg p.p.m.02 99 4.9 156 8.5 153 7.8 180 10.4 260 294 163 173 325 316 280 254 296 315 190 215 220 197 315 33 1 243 263 4.7 2.95 5.9 5.1 5-35 3-45 4-0 5.7 4.4 Error, + 7.6 - 1.9 + 4.8 - 7.2 - 13.1 + 6.2 - 2.7 - 9.3 + 6.4 + 13.1 - 10.4 +5*1 + 8.3 % With the exception of two results the recovery of oxygen from the samples was within about The gas evolved from samples of approximately 27 pg of tantalum oxide and 94 pg of silica would be equivalent to that evolved by a 10-g sample of metal containing approximately 5 p.p.m. of oxygen. OTHER INDICATIONS OF ACCURACY- serve to give added confidence in the validity of the results obtained. 10 per cent. of the amount added. The considerations listed below, although not constituting any proof of accuracy, do 1.A t low oxygen levels samples were analysed at two sample weights, usually 10 g and 1.5 g. Agreement between the results indicated satisfactory sample preparation and absence of interference by surface oxide. 2. Similar results were obtained under widely different conditions, for example, molybdenum fused in an iron bath at 1750" C and without a bath at 2200" C (see Table V). 3. Single-crystal silicon gives an absorption band in the infrared at 9 p due to the oxygen content. The ratio of the mean oxygen contents of the vacuum fusion results for the two silicon samples (see Table VI) was 1.39. This compared reasonably well with the ratio of 1-47 obtained from infrared measurements of the 9-p band in the two samples. The results of the experiments detailed above constituted reasonable grounds for con- sidering that the vacuum-fusion process gives results for low levels of oxygen in metals which are accurate within the limits of the standard deviations obtained.SOURCES OF ERROR IN VACUUM-FUSION ANALYSIS Consideration was given to possible sources of error in vacuum fusion analysis, and for this purpose the procedure was broken down and examined in four categories. SAMPLE PREPARATION- Two possible sources of error were apparent in this procedure, namely errors due to oxide film remaining on the samples, and errors due to removal of internal oxygen by outward diffusion as carbon monoxide during the pre-heating. Although the possibility of outward diffusion of oxygen had to be considered, it was not thought that this was likely to occur to any great extent under the pre-heating conditions used, and this was shown to be true.776 [Analyst, Vol.88 A collection of the gas evolved during the pre-heating period showed that approximately 1.5 pg of oxygen per sq. cm was given off from the samples. Most of this gas was collected during the first 5 to 10 minutes, and heating for more prolonged periods (up to 1 hour) did not produce any appreciable increase in the amount of carbon monoxide collected. Some results obtained on steel and molybdenum are shown in Table 11. DONOVAN, EVANS AND BUSH: DETERMINATION OF OXYGEN TABLE I1 Material Steel . . . . Molybdenum . . EFFECT OF DIFFERENT PRE-HEATING TIMES Pre- Number heating of deter- time, minations minutes 0 6 10 6 0 5 10 6 Mean oxygen content, p.p.m.6.7 5.4 4.6 2.7 Stan- dard deviation, p.p.m. 0.25 0.17 1-2 0.6 Weight of sample taken, g 2 2 1.5 1.5 Surf ace area of sample, sq. cm 2.2 2.2 1.5 1-5 Approximate amount of oxygen removed from surface, pg per sq. cm 1.2 1.2 1.9 1.9 The extent of any remaining oxide should be shown up by differences in results obtained on large and small samples of the same material, but the simplicity of results obtained on vacuum-melted steel at the lowest level available (see Table IV), showed that this was not significant. For molybdenum, however, there was evidence (see Table V) of a remaining oxide contamination of about 0.8 pg per sq. cm. FUSION AND GAS EVOLUTION- Thermodynamic consideration^^,^^ have shown that the solubility of oxygen in an iron bath saturated with carbon at temperatures above 1650" C and in contact with a pressure of carbon monoxide below torr, is negligible.The similarity of results obtained under different conditions (see Table V) would seem to confirm this conclusion, which has been accepted as the basis of all vacuum-fusion work. GAS COLLECTION AND ANALYSIS- carbon monoxide to a pressure of 15 cm. as shown in Fig. 1. The completeness of the gas recovery was checked by using the gas "doser" filled with This was attached to the sample entry system The results are shown in Table 111. TABLE I11 RECOVERY OF CARBON MONOXIDE UNDER DIFFERENT CONDITIONS The gas was collected for 1 minute Amount of iron as Carbo t i monoxide Tin Carbon monoxide charge in crucible, Temperature, added, added, recovered, g "C ml g ml 0 1950 0-067, 0.0 0.067, 20 1950" 0.067, 0.0 0.049, 20 1950t 0.067, 0.1 0.066, * Temperature maintained for 2 hours.t Temperature maintained for 2 hours 20 minutes. These results gave a strong indication that, provided the crucible was not allowed to run for prolonged periods without additions of tin being made, carbon monoxide was recovered quantitatively from the region of the crucible. So far it has been tacitly assumed that any oxygen present in the samples would be converted to carbon monoxide. This was a reasonable assumption, as the monoxide is by far the most stable oxide of carbon under the conditions of fusion. To confirm that the dioxide was absent, the gas evolved from samples of steel was collected with a solid carbon dioxide - acetone cold trap in position in place of liquid nitrogen.After the gas had beenOctober, 19631 I N STEELS, MOLYBDENUM AND SILICON BY VACUUM FUSION 777 collected this trap was replaced by liquid nitrogen, when in all instances a pressure drop not exceeding 1 per cent. of the total pressure was recorded. Part of this decrease was due to the temperature effect of the liquid nitrogen, but the experiment indicated that, if carbon dioxide was present, it consisted of less than 1 per cent. of the total gas collected (80 per cent. of carbon monoxide). who have found that carbon monoxide, nitrogen and hydrogen form at least 99 per cent. of the gas evolved. This agrees with the conclusion of other HOMOGENEITY OF SAMPLE- To determine the effect of possible variations in the method it was necessary to have homogeneous samples.Several secondary steel standards were available from the British Iron and Steel Research Association and covered the higher part of the range 40 to 150 p.p.m.; the results obtained on these agreed closely with those of other analysts. However, an investigation of the oxygen distribution in one such sample (B.B.L.) showed a considerable concentration gradient of oxygen from the centre to the outside of the bar, and for this reason sections taken from ingots were preferred for determining the standard deviation of the method. Sections 2 inches x 2 inches x 1 inch were taken from near the centre of the ingot, the 1-inch dimension being along the longitudinal axis. Non-uniformity in the sample would of course increase the standard deviation obtained, but the similarity of the coefficient of variation for the different sample weights of A437 suggested that this was not a factor, for this sample at least.An independent check of homogeneity was possible only with the samples of silicon, when the coefficient of variation determined by infrared methods was less than 3 per cent. Arc-cast molybdenum samples were taken from near the centre of the ingots, but for sintered and zone-refined material 9-inch bar was used. METHOD PROCEDURE FOR STEEL- Various types of steels were analysed successfully, including alloys containing 20 per cent. of manganese. The steels for whibh the apparatus was developed, however, were vacuum-melted R.A.R.D.E.high tensile steels, usually having an oxygen content between 5 and 15 p.p.m. and containing various alloying elements such as copper, manganese and vanadium in amounts up to 3 per cent. of each element. Samples were usually received in cubes of side 1 inch taken from near the centre of the ingot. For analysis these were fabricated into right cylinders with hemispherical ends 1 cm in diameter, 2 cm in length and weighing 10 g. An alternative type of sample was a cube of side 6 to 7 mm weighing 1-5 to 2 g. The cylindrical samples were polished with carborundum cloth; the surfaces of the cubes were cleaned with a fine file. When clean, the samples were weighed immediately and introduced into the high vacuum by means of the vacuum lock. Usually the smaller type of sample was used; they were easier to prepare and clean, and more samples could be analysed in one run.When they had been placed in the sample loading compartment, the samples were moved magnetically into the silica pre-heating furnace tube which held 5 of the large or 15 of the small samples. When the samples were in position, the resistance furnace was moved over the silica tube and the temperature of the samples raised to 1000" C. Fifteen minutes were allowed for the 10-g samples to reach this temperature and 10 minutes for the small samples. Before the samples were fused, a 10-g slug of vacuum-melted steel was placed, together with 0.1 g of tin, in the previously de-gassed crucible. The steel added with the tin prevented spluttering of the latter and also assisted the rapid melting of the subsequent samples.Analysis was then begun. This was found to be sufficient time, since the main evolution of gas, as shown on the Pirani gauge, occurred within the first 30 seconds, and at the end of 2 minutes the evolution of gas had usually dropped to the level of the blank rate. After the analysis of every third sample, 0.1 g of tin was added to prevent gettering. Samples were analysed at the rate of one every 10 minutes, which included the time required for blank determinations. The gas evolved from each sample was collected for 2 minutes. The gas collected was measured and analysed.778 DONOVAN, EVANS AND BUSH: DETERMINATION OF OXYGEN [Analyst, Vol. 88 Samples of molybdenum were received as sections of vacuum arc-melted ingots or as sintered or zone-refined rods of approximately 1 cm diameter.These were usually machined into right cylinders of length 2 cm and diameter 1 cm with hemispherical ends (weight 13 g). The dimensions were varied occasionally to give samples of lower weight (10 g). The samples were cleaned by anodic etching for 3 minutes in a (1 + 7) mixture of sul- phuric acid, sp.gr. 1.84, and methanol, a nickel cathode and a current of 1.5 amps being used. The samples were washed with three successive portions of methanol, dried with paper tissue, weighed and loaded into the apparatus as rapidly as possible. The samples were pre-heated in batches of five for 15 minutes at 1000" C. Molybdenum samples were fused under two different sets of conditions; in an iron - tin bath at 1750" to 1800" C or directly in the graphite crucible at 2200" C.The iron - tin bath was prepared initially by placing 20 g of vacuum-melted steel and 0.1 g of tin in the crucible; when this had been de-gassed and the furnace blank rate had fallen to 1 p1 per minute, usually after 5 minutes, a sample of molybdenum was added. Before each subsequent addition of molybdenum, a further 10 g of steel and 0.1 g of tin were added to the contents of the bath. It was found essential to add the samples of molybdenum within 15 minutes of adding the steel. If the interval was more prolonged the bath became viscous and the evolution of gas from the molybdenum was slow and often incomplete. By using this technique a maximum of nine 10-g samples could be analysed per run.Samples of molybdenum were added to the contents of the crucible at 2200" C, and the gas was collected for 1 minute, measured and analysed. The samples melted rapidly under these conditions owing, possibly, to the formation of a carbon - molybdenum eutectic. By this technique a maximum of twenty 10-g samples could be analysed during one run. This method was preferred to the iron-bath technique since it was simpler, more rapid and more reliable, particularly on samples having a low gas content. After additions of sample the furnace blank rate rapidly returned to its previous level. During the run the blank rate tended to decrease slightly whereas with the iron-bath technique it usually tended to increase slightly. PROCEDURE FOR MOLYBDENUM- The direct-melting technique was comparatively simple.PROCEDURE FOR SILICON- The samples of silicon for analysis were taken from vacuum- and argon-grown single crystals of approximate diameter 1 cm. These had been cut into slices 3 mm thick and optically polished for infrared examination ; the slices had then been cut diametrically, and one half of each (weight 0-4 to 0.5 g) was used for the analysis. The samples were weighed and then cleaned by etching for 15 seconds in a 10-ml poly- thene beaker with 5 ml of a (1 + 1) mixture of nitric acid, sp.gr. 1.42, and hydrofluoric acid, sp.gr. 1.15. The beaker rested in a large $-inch diameter evaporating basin, and when etching was complete the beaker and contents were swamped with 200 ml of de-ionised water. The silicon was removed with platinum-tipped forceps, washed in fresh de-ionised water and then transferred to the vacuum lock as quickly as possible.Most of the adhering water was shaken off and the rest was removed in the first vacuum stage of the lock. It was found by experiment that 35 mg of sample were rernoved during the etching process, and, therefore, this amount was subtracted from the recorded weight. When in the vacuum system, the samples were pre-heated at 1000" C for 5 minutes before analysis. An iron bath was used for the analysis since molten silicon attacks carbon rather vigorously. The crucible was maintained at 1700" C, which gave a furnace blank rate 0.3 to 0.5 p1 per minute. Vacuum-melted steel (10 g) and 0.1 g of tin were used for the initial bath, and after every three analyses a further 10 g of steel and 0.1 g of tin were added.PROCEDURE FOR OTHER METALS AND POWDERS- The apparatus was also used successfully for samples of thorium in an iron bath at 1950" C and for chromium and copper at 1700" C. Samples of iron, ferro-silicon, ferro-molybdenum and ferro-aluminium powders were also analysed by wrapping them in nickel foil and fusing in an iron bath.October, 19631 IN STEELS, MOLYBDENUM AND SILICON BY VACUUM FUSION 779 RESULTS The mean results and standard deviations obtained on a series of typical vacuum- melted steels are shown in Table IV; the corresponding results obtained on sintered, arc-cast and zone-refined molybdenum are shown in Table V. Results obtained on two samples of single-crystal silicon grown under a partial pressure of argon (sample A) and in vacuum (sample B) are shown in Table VI.TABLE IV RESULTS OBTAINED ON VACUUM-MELTED STEELS Sample Approximate Number of Standard Coefficient number weight of sample, determinations Mean, deviation, of variation, g p.p.m. p.p.m. % A329t .. .. 1.5 12 12.0 0.59 4.9 A379* .. .. 1.5 11 115 2.7 2.3 12 6.37 0.40 6.3 - * { &.: 6 6-35 0.33 5.2 A437(b)$. . * Vacuum-melted Swedish iron, containing less than 0.1 per cent. of total impurity. t Vacuum-melted R.A.R.D.E. steel containing 1 per cent. of carbon, 1.8 per cent. of silicon, 1.4 per cent. of manganese, 1-3 per cent. of molybdenum, 0.6 per cent. of vanadium and 1.5 per cent. of copper. $ Vacuum-melted steel containing 0.6 per cent. of carbon with less than 0.1 per cent. of minor constituents.TABLE V RESULTS OBTAINED ON DIFFERENT TYPES OF MOLYBDENUM Approxi- mate Type of weight of molybdenum sample, g 14 Sintered bar X . . 14 2 Sintered bar Y . . 13 2 10 Arc cast. . . . 10 Zone refined . . 10 1.5 1.5 Number of Conditions deter- of fusion minations Iron - tin bath No bath No bath No bath No bath Iron - tin bath No bath No bath No bath No bath 8 7 13 11 11 20 12 6 4 4 Tem- perature, "C 1750 2200 2200 2200 2200 1750 2200 2200 2200 2200 Mean oxygen content, p.p.m. 57.0 57.4 54.9 42.5 42.8 1.32 1.97 2.66 0.35 0-85 Standard deivation p.p.m. 1.3 1 4 1-7 0.8 1.1 0.40 0.16 0.63 0.07 0.16 ( 1 4 9 Coefficient of variation, 2.3 2.4 3.1 1.9 2-5 30.0 7.9 24.0 20.0 19.0 % There was a significant difference between the means of the results on the large and small samples for the first sintered molybdenum bar (X) in which the samples were taken from different positions in the bar. A second bar (Y) was therefore analysed at different weights, and this time the large and small samples were taken from alternate positions along the bar; there was little difference in the means of these results.With arc-cast ingots the mean value of the results obtained on the 10-g samples by using an iron bath was lower than that obtained by direct melting. The spread of the results obtained by the former method was 0-76 to 2-40 p.p.m. compared with 1-73 to 2.31 p.p.m. obtained by direct melting. In fact, the higher results on the iron - tin bath series corre- sponded to the results obtained by the direct-melting technique, which suggests that, for some samples, alloying with the bath metal was slow and evolution of gas incomplete.More consistent results might have been obtained with a higher ratio of iron to molybdenum in the bath, but this would have limited the number of determinations per run still further. The advantages of the direct-melting technique were thus apparent as twenty samples could be analysed per run with greater precision. With zone-refined samples, there was a difference of 0-5 p.p.m. between results at different sample weights, which can be explained as being due to residual surface contamination. The results agree at both weights if a surface contamination of 0-8 pg per sq. cm is assumed.780 [AnaZyst, Vol. 88 The small samples were discs 8 mm in diameter and 2 mm thick with a total surface area of 1-5 sq.cm. A correction of 0-8 p.p.m. coulcl be applied to the results for the small samples and 0-4 p.p.m. to those for the large samples (surface area 5-0 sq. cm). When these corrections were applied, the figure suggested that the oxygen content of the zone-refined molybdenum was less than 0.2 p.p.m. DONOVAN, EVANS AND BUSH : DETERMINATION OF OXYGEN TABLE VI RESULTS OBTAINED ON 2 TYPES OF SINGLE-CRYSTAL SILICON Type of Number of Mean oxygen Standard Coefficient of single-crystal silicon determinations content, deviation ( 1 u), variation, 01 p.p.m. p.p.m. / O Argon grown . . .. 6 12.3 1.6 13 Vacuum grown . . . . 8 17-1 1.4 8.3 The ratio of 1-39 obtained on the results of the samples of silicon analysed by vacuum fusion compares favourably with the 1.47 from infrared measurements.Closer agreement would probably be obtained if the surface contamination of the silicon was determined and the results were corrected accordingly. CONCLUSIONS The vacuum-fusion technique has been developed successfully and applied to the analysis of steel, molybdenum and silicon at low levels of oxygen (less than 1 p.p.m. in molybdenum). With samples of steel the coefficients of variation obtained were much less than 10 per cent., and no bias due to remaining oxide films was observed. There was good indication that the method was accurate down to 6 p.p.m., the lowest level encountered. Coefficients of variation of less than 10 per cent. were obtained on the determination of oxygen in molybdenum down to 2 p.p.m., but at this lowest level there was a positive bias due to residual surface contamination.Analysis of zone-refined material suggested that most of the oxygen evolved came from the residual surface contamination and that the oxygen content of the material was in the range 0 to 0.2 p.p.m, Analysis of 0.5-g samples of single-crystal silicon gave a coefficient of variation of 10 per cent. at the level of 17 p.p.m. of oxygen. The ratio between the levels of oxygen obtained on the two different series of samples agrees fairly well with the same ratio obtained by infrared examination at 9 p. We acknowledge the work done by the late Mr. A. Bacon who made a significant contri- bution to the progress of this work and thank Mr. R. J. Loneragan who carried out the infrared analyses and was responsible for some early work in the field.25 Acknowledgment is also made to Mr.B. R. Watson-Adams, who carried out many of the analyses, for helpful suggestions and practical assistance. Acknowledgment is made to the Controller of H.M. Stationery Office for permission to publish this paper. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. REFERENCES Sloman, H. A., “The Sixth Report on the Heterogeneity of Steel Ingots Committee,” Special -, “The Seventh Report on the Heterogeneity of Steel Ingots Committee,” Special Report -, “The Eighth Report on the Heterogeneity of Steel Ingots Committee,” Special Report -, J . Inst. Metals, 1945, 71, 391. -, Ibid., 1946, 72, 441. Sloman, H. A., and Harvey, C. A., Ibid., 1951/52, 80, 391 (Appendix by Ilubaschewski, 0.). Sloman, H. A., J , Iron & Steel Inst., 1941, 143, 298. Olds, L. E., and Rengstorff, G. W. P., J Metals, 8 ; A.I.M.E. Gokeen, N. A., i n Elliott, J. F., Editor, “The Physical Chemistry of Steelmaking,” Technology Lassner, E., and Wolfel, E., Mikrochim. Acta, 1960, 394. Mallett, M. W., and Griffiths, C. B., Trans. Amer. Soc. Metals, 1954, 46, 375. McDonald, R. S., and Fagel, J. E., Anal. Chem., 1955, 27, 1632. Report No. 9, The Iron and Steel Institute, London, 1935, p. 71. No. 16, The Iron and Steel Institute, London, 1937, p. 82. No. 25, The Iron and Steel Institute, London, 1939, p. 43. Trans., 1956, 206, 150. Press of M.I.T. and John Wiley and Sons Inc., New York, 1958, p. 41.October, 19631 IN STEELS, MOLYBDENUM AND SILICON BY VACUUM FUSION 781 13. 14. 15. 16. 17. 18. 19. 20. 21. 2.2. 23. 24. 25. Pearce, ill. L., and Masson, C. R., Special Report No. 68, The Iron and Steel Institute, London, Grossfuss, E., Neue Hiitte, 1958, 10, 608; (B.I.S.I. Translation No. 1487). Stevenson, W. W., and Speight, G. E., J . Iron & Steel Inst., 1941, 143, 312. Coleman, R. F., Analyst, 1962, 87, 590. Davis, H. E., and Gray, J. A., Royal Aircraft Establishment Report Met., 86, 1955. Ransley, C. E., G.E.C. Journal, 1940, 11 (2). __ , Analyst, 1947, 72, 504. Bacon, A., Special Report ,Vo. 68, Thc Iron and Steel Institute, London, 1960, p. 133. Parker, A , , Analyst, 1961, 86, 550. Everett, AT. R., and Thompson, G. E., Ibid., 1962, 87, 515. Elliott, J. F., Editor, op. cit., p. 37. Martin, J. F., Friedline, J . E., Melnick, L. M., and Pellissier, G. E., Trans. Met. Soc. A . I . M . E . , Loneragan, R. J., Unpublished Ministry of Supply Memorandum, 1958. 1960, p. 121. 1958, 212, 514. Received January 9th, 1963
ISSN:0003-2654
DOI:10.1039/AN9638800771
出版商:RSC
年代:1963
数据来源: RSC
|
|