摘要:

THE ANALYSTTHE ANALYTICAL JOURNAL OF THE CHEMICAL SOCIETYEDITORAL ADVISORY BOARD"Chairman: H. J. Cluley (Wembley)"L. S. Bark (Salford)R. Belcher (Birmingham)L. J. Bellamy, C.B.E. (Waltham Abbey)L. S. Birks (U.S.A.)E. Bishop (Exeter)L. R. P. Butler (South Africa)E. A. M. F. Dahmen (The Netherlands)A. C. Docherty (Billingham)D. Dyrssen (Sweden)J. Hoste (Belgium)"W. T. Elwell (Birmingham)"J. A. Hunter (Edinburgh)H.H.Mw"G.M. N. H. Irving (Leeds)Kaiser (Germany)T. Kelley (U.S.A.)Kemula (Poland)F. Kirkbright (London)G. W. C. Milner (Harwell)G. H. Morrison (U.S.A.)"J. M. Ottaway (Glasgow)"G. E. Penketh (Billingham)"T. B. Pierce (Harwell)E. Pungor (Hungary)D. I. Rees (London)"R. Sawyer (London)P. H. Scholes (Sheffield)"W. H.C. Shaw (Greenford)S. Siggia (U.S.A.)A. A. Smales, O.B.E. (Harwell)A. Walsh (Australia)T. S. West (Aberdeen)A. L. Wilson (Medmenham)P. Zuman (U.S.A.)*A. Townshend (Birmingham)"Members of the Board serving on The Analyst Publications CommitteeREGIONAL ADVISORY EDITORSDr. J . Aggett, Department of Chemistry, University of Auckland, Private Bag, Aucltland, NEWProfessor G. Ghersini, Laboratori CISE, Casella Postale 3986, 20100 Milano, ITALY.Professor L. Gierst, Universit6 Libre de Bruxelles, Facult6 des Sciences, Avenue F.-D. Roosevelt 50,Professor R. Herrmann, Abteilung fur Med. Physik., 63 Giessen, Schlangenzahl 29, GERMANY.Professor W. E. A. McBryde, Dean of Faculty of Science, University of Waterloo,Waterioo, Ontario,Dr.W. Wayne Meinke, KMS Fusion Inc., 3941 Research Park Drive, P.O. Box 1567, Ann Arbor,Dr. I. Rubeska, Geological Survey of Czechoslovakia, Kostelni 26, Praha 7, CZECHOSLOVAKIA.Dr. J . RGiiEka, Chemistry Department A, Technical University of Denmark, 2800 Lyngby, DEN MARK.Professor K. Saito, Department of Chemistry, Tohoku University, Sendai, JAPAN.Dr. A. Strasheim, National Physical Research Laboratory, P.O. Box 395, Pretoria, SOUTH AFRICA.ZEALAND.B ruxel les, B E LG I U M .CANADA.Mich. 481 06, U.S.A.Published by The Chemical SocietyEditorial: The Director of Publications, The Chemical Society, Burlington House,London, W1 V OBN. Telephone 01 -734 9864. Telex No. 268001.Advertisements: J. Arthur Cook, 9 Lloyd Square, London, WC1 X 9BA. Telephone 01 -837 631 5.Subscriptions (non-members): The Chemical Society Publications Sales Office, Blackhorse Road,Letchworth, Herts.. SG6 1 HN.Volume 101 No 1202@ The Chemical Society 1976May 197

ISSN:0003-2654    

DOI:10.1039/AN97601FX017

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

ANALAO 101 (1 202) 321-408 ( I 976)I SS N 0003-2654May 1976THE ANALYSTTHE ANALYTICAL JOURNAL OF THE CHEMICAL SOCIETYCONTENTSREVIEW PAPER321341348356367379381386391396404405Analysis and Assay of Polyene Antifungal Antibiotics-A. H. ThomasORIGINAL PAPERSThe Determination of Arsenic( 111) and Total Arsenic by Atomic-absorptionSpectroscopy-J. Aggett and A. C. AspellThe Determination of the Precious Metals by Flameless Atomic-absorptionSpectrophotometry-G. L. EverettThe Development of Remote Spectrographic Heads for Metallurgical Analysisand their Application t o Product Inspection Analysis-A. D. Ambrose andJ. D. HobsonDetection and Determination of Polynuclear Aromatic Hydrocarbons by Lumines-cence Spectrometry Utilising the Shpol'skii Effect at 77 K.Part II. AnEvaluation of Excitation Sources, Sample Cells and Detection Systems-9. S. Causey, G. F. Kirkbright and C. G. de LimaDetermination of Dicumenyl Peroxide by Gas Chromatography-P. Hudec,B. NovotnA and J. PetrijThe Determination of Uric Acid in Animal Feeding Stuffs Using High-performanceLiquid Chromatography-G. 9. Cox, C. R. Loscombe and J. A. UpfieldDetermination of Residues of Inorganic Bromide i n Grain-Report by The Panel onFumigant Residues in GrainImproved Extraction of Steroid Hormonesfrom the Blood Plasma o f the DomesticFowl ( Ga//us domesticus) Using Light Petroleum at Elevated Temperatures-J. CulbertThe Determination of Boron i n Magnesites Using Carminic Acid-N. F. C. Sheltonand R. A. ReedElimination of Interference from Aluminium in the Determination of Total IronComment in Soils and Plant Materials Using 1 ,lo-Phenanthroline Reagent.on the Paper by Jayman et a/.-R. J. JuliettiBook ReviewsSummaries o f Papers in this lssue-Pages iv. v, viii, xPrinted by Heffers Printers Ltd, Cambridge, EnglandEntered as Second Class a t New York, USA, Post Offic

ISSN:0003-2654    

DOI:10.1039/AN97601BX019

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

May, 1976 THE ANALYST...111Reprints of Review PapersReprints of the following Review Papers published in The Analyst since 1965 are available fromthe Publications Sales Officer, The Chemical Society, Blackhorse Road, Letchworth, Herts., SG61HN (not through Trade Agents).The price per reprint is k1; orders for six or more reprints of the same or different Reviewsare subject to a discount of 26%. The appropriate remittance, made out to The ChemicalSociety, should accompany any order.“Activation Analysis,” by R. F. Coleman and T. B. Pierce (January, 1967).“Techniques in Gas Chromatography. Part I. Choice of Solid Supports,” by F. J. Palframan“Heterocyclic Azo Dyestuffs in Analytical Chemistry,” by R. G. Anderson and G. Nickless“Determination of Residues of Organophosphorus Pesticides in Food,” by D.C. Abbott and“Radioactive Tracer Methods in Inorganic Trace Analysis : Recent Advances,” by J. W.“Gamma-activation Analysis,” by C. A. Baker (October, 1967).“Precipitation from Homogeneous Solution,” by P. F. S. Cartwright, E. J. Newman and“Industrial Gas Analysis,” by (the late) H. N. Wilson and G. M. S. Duff (December, 1967).“The Application of Atomic-absorption Spectrophotometry to the Analysis of Iron and“Inorganic Ion Exchange in Organic and Aqueous - Organic Solvents,” by G. J. Moody and“Radiometric Methods for the Determination of Fluorine,” by J . K. Foreman (June, 1969).“Techniques in Gas Chromatography. Part 11. Developments in the van Deemter RateTheory of Column Performance,” by E. A.Walker and J. F. Palframan (August, 1969).“Tcchniques in Gas Chromatography. Part 111. Choice of Detectors,” by T. A. Gough andE. A. Walker (January, 1970).“Laser Raman Spectroscopy,” by P. J. Hendra and C. J. Vear (April, 1970).“Ion-selective Membrane Electrodes,” by Ern6 Pungor and Kldra T6th (July, 1970).“X-ray Fluorescence Analysis,” by K. G. Carr-Brion and I<. W. Payne (December, 1970).“Mass Spectrometry for the Analysis of Organic Compounds,” by A. E. Williams and H. E“The Application of Non-flame Atom Cells in Atomic-absorption and Atomic-fluorescence“Liquid Scintillation Counting as an Analytical Tool,” by J. A. B. Gibson and A. E. Lally“The Determination of Some 1,4-Benzodiazepines and Their hletabolites in Body Fluids,”“Atomic-fluorescence Spectrometry as an Analytical Technique,” by R.F. Browner(October, 1974).“The Use of Precipitate Based Silicone Rubber Ion-selective Electrodes and Silicone RubberBased Graphite Voltammetric Electrodes in Continuous Analysis,” by 2s. Fkher,G. Nagy, K. T6th and E. Pungor (November, 1974).“The Examination of Meat Products with Special Reference to the Assessment of the MeatContent,” by D. Pearson (February, 1975).“Chemiluminescence in Gas Analysis and Flame-emission Spectrophotometry,” by J . H .Glover (July, 1975).“The Analytical Role of Ion-selective and Gas-sensing Electrodes in Enzymology,” byG. J. Moody and J. D. R. Thomas (September, 1975).“Thiazolylazo Dyes and Their Applications in Analytical Chemistry,” by H%vard R. Hovind(November, 1975).“Sample Preparation in the Micro-determination of Organic Compounds in Plasma or Urine,”by Eric Reid (January, 1976).“Recent Advances in the Ring Oven Technique,” by Herbert Weisz (March, 1976).“The Radioimmunoassay of Drugs,” by J .Landon and A. C. Moffat (April, 1976).“Analysis and Assay of Polyene Antifungal Antibiotics,” by A. H. Thomas (May, 1976).and E. A. Walker (February, 1967).(April, 1967).H. Egan (August, 1967).McMillan (September, 1967).D. W. Wilson (November, 1967).Steel,’’ by P. H. Scholes (April, 1968).J. D. R. Thomas (September, 1968).Stagg (January, 1971).Spectroscopy,” by G. F. Kirkbright (September, 1971).(October, 1971).by J. M. Clifford and W. Franklin Smyth (May, 1974)iv SUMMARIES OF PAPERS I N THIS ISSUE May, 1976Summaries of Papers in this IssueAnalysis and Assay of Polyene Antifungal AntibioticsA ReviewSummary of ContentsIntroductionChemical and Biological PropertiesProductionChemical Methods of Analysis and AssayBiological AssayStabilityConclusionsReprints of this Review paper can be obtained from The PublicationsSales Officer, The Chemical Society, Blackhorse Road, Letchworth, Herts.,SG6 lHN, a t L l per copy (with a 25% discount for six or more copies),post free.A remittance for the correct amount, made out to The Chemical Society,should accompany every order ; these reprints are not available throughTrade Agents.A.H. THOMASDivision of Antibiotics, National Institute for Biological Standards and Control,Hampstead, London, NW3 6RB.Analyst, 1976, 101, 321-340.The Determination of Arsenic(II1) and Total Arsenicby Atomic- absorption SpectroscopyIn the atomic-absorption determination of arsenic by the hydride evolutionmethod with sodium borohydride, maintenance of the pH between 4 and 5permits the selective determination of arsenic(II1) in mixtures of arsenic(II1)with arsenic(V) .Total arsenic can be determined separately by evolutionfrom 5 M hydrochloric acid. A dissolution technique has been developedfor herbage and the method applied to the analysis of water and orchardleaves.J. AGGETT and A. C. ASPELLChemistry Department, University of Auckland, Auckland, New Zealand.Analyst, 1976, 101, 341-347.The Determination of the Precious Metals by FlamelessAtomic-absorption SpectrophotometryThe optimum conditions for the determination of palladium, platinum,rhodium, ruthenium, iridium and osmium by carbon rod atomic-absorptionspectrophotometry are described and data for sensitivity, detection limit andreproducibility are given.With the exception of osmium the sensitivityand detection limits are such that the working range of the atomic-absorptiontechnique can be extended downwards by an order of magnitude or morewhen compared with that with an air - acetylene flame.The effect of mutual interference is investigated and methods of reducingit are proposed.G. L. EVERETTJohnson Matthey Chemicals Ltd., Royston, Hertfordshire.Analyst, 1976, 101, 348-355May, 2976 SUMMARIES OF PAPERS I N THIS ISSUEThe Development of Remote Spectrographic Heads for MetallurgicalAnalysis and Their Application to Product Inspection AnalysisDirect-reading optical-emission spectroscopy has been widely adopted in theiron and steel industry as a rapid analytical technique for process controland for other laboratory analyses.The conventional spark stand, however,severely restricts the size and shape of the sample that can be accommodatedon the instrument and thus limits the more general use of the technique foron-site analysis outside the laboratory. This disadvantage has now beenovercome by using a fibre-optic light-guide to link a mobile excitation headto a conventional spectrometer.Parallel but independent work has been carried out in the authors' labora-tories, first on photographic and later on direct-reading spectrometers.Atthe Corby laboratories, further work has resulted in the manufacture andapplication of an inspection analyser that incorporates a 5-m guide. Typicalcalibration graphs, reproducibilities and examples of material identificationare given, and possible applications are discussed.A. D. AMBROSEVResearch Centre, British Steel Corporation, Tubes Division, Corby, Northamptonshire,"17 1UA.and J. D. HOBSONDunford Hadfields Ltd., East Hecla Works, Sheffield, S9 1TZ.Analyst, 1976, 101, 356-366.Detection and Determination of of PolynuclearAromatic Hydrocarbons by Luminescence SpectrometryUtilising the Shpol'skii Effect at 77 KDetection SystemsA comparison has been made of the use of a xenon arc continuum source,a mercury-vapour lamp and a fixed-wavelength helium - cadmium laser sourcefor excitation in the production of quasi-linear luminescence emission spectraof some polynuclear aromatic hydrocarbons in n-alkane solvents at 77 K.The performance of a new design of cryostat sample cell for work at 77 K hasbeen evaluated and compared with that of a commercially available Dewar-flask sample cell.The use of d.c. integration, photon counting and of repetitiveoptical scanning in conjunction with a signal averager has been investigatedfor registration of the luminescence obtained from PAH compounds by use ofthe Shpol'skii effect.B. S. CAUSEY, G. F. KIRKBRIGHT and C. G. de LIMAPart 11. An Evaluation of Excitation Sources, Sample Cells andDepartment of Chemistry, Imperial College of Science and Technology, London,SW7 2AY.Analyst, 1976, 101, 367-378.Determination of Dicumenyl Peroxide by Gas ChromatographyA method is described for the determination of small amounts of dicumenylperoxide in liquid samples or powdered materials. Dicumenyl peroxide isisolated from the latter by extraction.P. HUDEC, B. NOVOTNA and J. PETRGJResearch Institute of Macromolecular Chemistry, TkalcovskA 2, Brno, Czechoslovakia.Analyst, 1976, 101, 379-380

ISSN:0003-2654    

DOI:10.1039/AN97601FP033

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

...Vlll SUMMARIES OF PAPERS IN THIS ISSUE May, 1976The Determination of Uric Acid in Animal Feeding Stuffs UsingHigh - performance Liquid ChromatographyA method is given for the extraction of uric acid from animal feeding stuffsthat contain dried poultry waste. The extract obtained is analysed by high-performance liquid chromatography using a strong anion-exchange bondedsilica column and an ultraviolet method of detection. Results of a seriesof recovery checks together with those for the analysis of actual samples aregiven. The application of the method to ammonium urate additions isinvestigated.G. B. COX, C. R. LOSCOMBE and J. A. UPFIELDDepartment of Industry, Laboratory of the Government Chemist, Cornwall House,Stamford Street, London, SE1 9NQ.Analyst, 1976, 101, 381-385.Determination of Residues of Inorganic Bromide in GrainReport by the Panel on Fumigant Residues in GrainCOMMITTEE FOR ANALYTICAL METHODS FOR RESIDUES OF PESTI-CIDES AND VETERINARY PRODUCTS IN FOODSTUFFS (DR.N. A.SMART, SECRETARY)Ministry of Agriculture, Fisheries and Food, Plant Pathology Laboratory, HatchingGreen, Harpenden, Hertfordshire, AL5 2BD.Analyst, 1976, 101, 386-390.Improved Extraction of Steroid Hormones from the Blood Plasmaof the Domestic Fowl (Gallus domesticus) Using Light Petroleumat Elevated TemperaturesA study has been made of the extraction of steroid hormones from hen bloodplasma into light petroleum (boiling range 60-80 "C) at 50 "C. This techniqueresults in a high yield of steroid hormones such as testosterone, androst-4-ene-3,17-dioneJ progesterone, corticosterone and oestrone, which partitioninto the epiphase with a low level of lipid contaminant.17&Oestradiol waspoorly extracted, giving only a 52.5% yield. The relatively high temperatureused did not cause steroid degradation. The low level of lipid contaminantin the light petroleum fraction contrasted favourably with that found withuse of conventional methods when using solvents such as diethyl ether ordichloromethane.J. CULBERTThe Agricultural Research Council's Poultry Research Centre, King's Buildings,West Mains Road, Edinburgh, EH9 3JS.Analyst, 1976, 101, 391-395X SUMMARIES OF PAPERS IN THIS ISSUEThe Determination of Boron in Magnesites Using Carminic AcidThe development of a method for the determination of boron in magnesitematerials as used in the refractories industry is described. The method isbased on the spectrophotometric measurement of the boron - carminic acidcomplex, and the most probable sources of error have been investigated andeither eliminated or compensated for by appropriate experimental techniques.N. F. C. SHELTONG.R.-Stein Refractories Ltd., Central Research Laboratories, Sandy Lane, Worksop,Nottinghamshire.and R. A. REEDBritish Ceramic Research Association, Queens Road, Penkhull, Stoke-on-Trent,ST4 7LQ.Analyst, 1976, 101, 396-403.May, 1976Elimination of Interference from Aluminium in the Determinationof Total Iron in Soils and Plant Materials Using1,lO- Phenanthroline ReagentComment on the Paper by Jayman et at.R. JULIETTIMorgan Thermic Technological Centre, Bewdley Road, Stourport-on-Severn,Worcestershire, DY 13 8QR.Analyst, 1976, 101, 404

ISSN:0003-2654    

DOI:10.1039/AN97601BP037

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

MAY 1976 The Analyst Vol. 101 No. 1202 Analysis and Assay of Polyene Antifungal Antibiotics A Review* A. H. Thomas Division of Antibiotics, National Institute for Biological Standards and Control, Hampstead, London, N W3 6 R B Summary of Contents Introduction Chemical and Biological Properties Production Chemical Methods of Analysis and Assay Biological Assay Stability Conclusions Introduction In this review, the chemical and biological properties of the polyene antifungal antibiotics are examined in order to show how these properties can be used for the analysis and assay of the antibiotics. The biosynthesis, production and stability of the antibiotics have also been included as this information is useful in anticipating the occurrence of likely impurities that would affect the analysis and assay.The use of the antibiotics has been considered, as this should determine the method of assay used. Previous reviews have dealt with the chemi~try,l-~ production,4 biology5s6 and therapeutic use7 of the polyene antifungal antibiotics. Chemical and Biological Properties Numerous antifungal antibiotics have been isolated from soil actinomycetes, predominantly of the genus Streptomyces. The antibiotics exhibited a characteristic ultraviolet absorption spectrum, which suggested that the chromophores were unsaturated straight-chain p~lyenes.~,~ The number of consecutive conjugated double bonds in the chromophore formed the basis of the classification of these antibiotics as tetraenes, pentaenes, hexaenes and heptaenes2 The stereochemistry of the chromophore can be deduced from the infrared and ultraviolet absorp- tion spectra. A chromophore that has all-trans bonds exhibits a spectrum in which the peak at the longest wavelength, which is also the narrowest, is either the strongest in the group or only slightly weaker than the second peak.On the evidence of their ultraviolet absorption spectra, stability to stereoisomerising reagents and presence of trans-ethylenic peaks in the 10.1-10.6 pm region of the infrared spectrum, Oroshnik and Mebane2 reasoned that the chromophores of the tetraenes, pentaenes, methylpentaenes and hexaenes had all-trans configurations. The spectra of the heptaenes were more complicated to interpret, as they were degraded in both wavelength maxima and fine structure, indicating the presence of stereoisomers that possibly contain more than one cis bond.The peak due to cis bonds is found at a wavelength of 1 / 4 2 of the maximum wavelength, but it was not prominent in the cis-heptaenes, which indicated that the cis bond was adjacent to the terminal double bonds. The cis peak is a measure of the strain in the chromophore due to the cis configuration, which is greatest when the cis bond is centrally placed. During the isolation of some heptaenes, wavelength maxima were found to vary from 401 to 405 nm. The original cis-heptaenes readily underwent light-induced stereoisomerisation to give products with an increased wave- length maximum where the peak was more prominent. The hindrance-free spectra and stability of the maximum wavelength of candidin and amphotericin B indicated that they were produced and isolated in the all-trans configuration. The spectra of three representative polyene antifungal antibiotics are shown in Fig.1. * Reprints of this paper will be available shortly. For details, see summaries in advertisement pages. 321322 THOMAS ANALYSIS AND ASSAY OF POLYENE Analyst, Vol. 101 500 Wavelength/nm Fig. 1. Typical spectra of polyene antifungal antibiotics. Solution in dimethylformamide (1-cm cell). A, nystatin (tetra- ene), 20 pg ml-l; B, trichomycin (heptaene), 15 pg ml-l; and C, amphotericin B (heptaene), 7 pg ml-l. A sharp peak in the infrared spectrum at about 5.83 pm, indicative of an unconjugated lactone ring, is a feature common to all polyene antifungal antibiotics. The size varies from the 26-atom ring of natarnycinl0s1l to the 38-atom ring of nystatinl29l3 and amphotericin B14J5; in the last two molecules, the lactone bond is between carbon atoms 1 and 37.Most of the polyenes contain carbon, hydrogen, oxygen and nitrogen. The nitrogen atom is present as a primary amino group and acid hydrolysis results in the liberation of an amino- sugar, mycosamine (3-amino-3,6-dideoxy-~-mannose). An isomer of mycosamine, peros- amine (4-amino-4,6-dideoxy-~-mannose), was isolated from perimycin. Mycosamine in the pyranose form is glycosidically linked to the hydroxyl at carbon atom 19 of the lactone ring of amphotericin B. In addition, most polyenes, with the exception of some of the methyl- pentaenes, possess a carboxyl group that is present as a zwitterion, indicated by a strong peak at 6.35-6.50 pm.Most of the polyenes behave as amphoteric compounds with a characteristic isoelectric point (q., natamycin, pH 6.5).16 They are soluble in acids and bases but they are not usually stable at the pH required to dissolve them. Another feature of the polyenes is the large number of free hydroxyl groups in the lactone ring: candidinl7 has seven, while amphotericin B14915 and n y ~ t a t i n l ~ , ~ ~ have eight each. The presence of the hydroxyl groups and zwitterion is responsible for the comparative insolubility of these compounds in organic solvents. Another reason put forward for the exceptional insolubility and relatively high stability of amphotericin B is the existence of a six-membered ketal ring structure,15 joining carbon atoms 13 and 17, and such a structure is also thought to be present in crystalline nystatin.12 The heptaenes are sub-divided on the basis of the amino-sugar released by acid hydrolysis and the aromatic amine released by alkaline retroaldol cleavagels (Table I).Complete mass spectra of the pertrimethylsilylated derivatives of natamycin, nystatin and amphotericin B have been reported26 and the structures are shown in Fig. 2. The stereo- chemistry of the 14 asymmetric centres and the configuration of the glycosidic link of ampho- tericin B have been determined by X-ray single-crystal analysis of N-iodoacetylamphotericin The solubility and the colour of the pure antibiotic is determined by the chromophore. Colour is increased from buff to dark yellow and the limited water solubility is decreased with increasing hydrophobic nature of the chromophore as the number of the double bonds is increased.The antibiotics are soluble in dimethylformamide, dimethyl sulphoxide and aqueous solutions of alcohols. Infrared studies showed the presence of strong intermolecular hydrogen bonds, involving not only the hydroxyl groups but also the carboxyl and the amino groups of amphotericin B in dimethyl sulphoxide solution. The zwitterionic character of the antibiotic is also disrupted by the solvent molecules, which produces exceptional solubility in B -14 927May, 1976 ANTIFUNGAL ANTIBIOTICS. A REVIEW TABLE I CLASSIFICATION OF THE HEPTAENE ANTIBIOTICS Group Amino-sugar Aromatic amine 1 Mycosamine Absent 2 Mycosamine 9- Aminoacetophenone 323 Antibiotic Amphotericin Candidin17 Mycoheptinlg Candicidin18 HamycinZ0 Heptafungin A21 Levorin A and Be2 Trichomycin A23 3 Mycosamine N-Methyl-9-aminoacetophenone A~reofungin~~ 4 Perosamine N-Meth yl-paminoacetophenone Perimycin26 dimethylformamide and dimethyl sulphoxide.28 Antibiotic solutions can be carefully diluted with water, as they are not visibly precipitated at concentrations of less than 50 pg ml-l, although they exist as micellar suspensions in aqueous media.29s30 The ultraviolet absorption spectra of these micelles or colloidal dispersions exhibit a peculiar type of spectral “degrada- ,O NH* OH OH OH 0 OH COOH A B C CH3 Fig. 2.Structure of polyene antifungal antibiotics : A, natamycin (tetraene), relative molecular mass 665.7, references 10 and 11; B, nystatin A1 (tetraene), relative molecular mass 926.1, references 12 and 13; and C, amphotericin B (heptaene), relative molecular mass 924.1, references 14 and 15.324 THOMAS; ANALYSIS AND ASSAY O F POLYENE Analyst, Vol.101 tion” similar to the effects of severe steric hindrance, which has caused more than one investiga- tor some perplexity.2 Both the flexible polyhydroxyl group and the rigid double-bond system are necessary for biological activity. Amphotericin B can be considered to have two chains. The chain that contains the polyene chromophore is completely hydrophobic, whereas the chain that con- tains the large number of hydroxyl groups has a hydrophilic and hydrophobic face, rendering the chain amphipathic. One end of the molecule is very polar, containing both mycosamine and the carboxyl group; the other end is completely non-polar except for one very conspicuous hydroxyl group.Amphotericin B molecules can be packed together to form a cylindrical structure with the interior lined by the hydroxyl groups of the amphipathic face. The exterior of the cylinder is completely non-polar. Thus, aqueous pores can be made in thin lipid membranes, which may account for the biological action of amphotericin and other polyene antibiotics that results in the leakage of potassium ions from susceptible ce11~.~~,~2 The polyene antibiotics possess both fungicidal and fungistatic properties against a wide range of fungi, but there are great differences between the sensitivities of different species of fungi to the different polyenes.The incidence of resistance to these antibiotics is rare both in clinical practice and in the laboratory. In addition to their antifungal activities, many polyenes also inhibit algae and some protozoa. Nystatin and natamycin (pimaricin) are administered orally; as they are poorly absorbed, relatively large doses can be tolerated although they are ineffective in the treatment of systemic mycosis. Natamycin is given by inhalation for respiratory infections. Nystatin and natamycin are used topically but both are too toxic for parenteral use. Amphotericin B is used orally for systemic therapy, although it is poorly absorbed, and is also used topically. Parenteral administrationof arnphotericin B is effected by using a bile salt complex (Fungizone), which forms a colloidal suspension when mixed with water. Owing to the toxicity of this preparation, intravenous therapy should be used only in life-threatening infection^.^,^^ The bile salt complex is considered to be more toxic than the parent antibiotic,6J* although it is not clear whether the increased toxicity is due to the bile salt or the increased availability of the “solubilised” amphotericin B.Toxicity tests in animals did not confirm that the bile salt complex was more toxic than amphotericin B.29 Less toxic derivatives of amphotericin B that still retained antifungal activities have been prepared. Substitution of a proton on the amino group of mycosamine by acetylation resulted in a loss of antifungal activity,28 whereas glycosylation did not .35 Esterification of the carboxyl group gave more soluble compounds, the methyl ester retaining the antifungal activity of the parent antibioti~.~6,~7 Animal tests showed that the toxicity of the methyl ester of amphotericin B was less than that of the parent c o m p ~ u n d .~ ~ - ~ ~ The increased solubility of the more finely dispersed colloidal suspension of the methyl ester did not increase its toxicity and therefore the role of the bile salt in the toxicity of the amphotericin B complex should be re-investigated, as an increase in the availability of amphotericin B per se does not result in an increase in toxicity. Candicidin has been found to be less effective than nystatin for eliminating fungi from the intestinal tract; as with other polyenes, no absorption could be detected.40 A method for detecting the very low level absorption of the polyene antibiotics is required; experiments with carbon-14 labelled polyenes would be helpful.Following the observation that the oral administration of candicidin resulted in a reduction in the size of the prostate gland of dogs,41 candicidin has been used with moderate success for the treatment of human benign hypertrophied prostate glands.42 Also, some polyenes produced a hypocholesterolemic effect in small laboratory animals,43944 although this property has not yet been exploited in man. The polyenes can be used to prevent fungal contamination in tissue culture or as food preservatives (e.g., natamycin on the rind of hard cheese and nystatin on the surface of bananas),45 or nystatin can be used as a growth promoter in animal feeds.Some biological and chemical properties of selected polyenes, derivatives and common impurities are shown in Tables 11-IV. The wide range of figures quoted reflect the different methods used in the measurement of biological activity and toxicity and the variation in the purity and homogeneity of polyenes used. Production There are many examples of complex mixtures of antifungal substances being obtained from micro-organisms that produce the polyene antifungal antibiotics. Non-polyene antifungalMay, 1976 ANTIFUNGAL ANTIBIOTICS. A REVIEW 325 TABLE I1 CHEMICAL PROPERTIES OF POLYENE ANTIFUNGAL ANTIBIOTICS AND CO-PRODUCED NON-POLYENE ANTIFUNGAL ANTIBIOTICS Chromophore group Antibiotic Tetraene Amphotericin A aD(so1vent) * Reference +136 (DMF) 2 +258 (DMF) 2,16 + 180 (DMSO) + 12 (DMF) 2 + 21 (CsH,X) +238 (DMF) 2 +363 (DMF) 2 19 2 +I63 (CSHSN) 291.0 289.5 302.5 317.6 Natam ycin (pimaricin) N ys t atin 710 1100 1020 570 850 780 291.0 304.0 318.5 Heptaene Amphotericin B Candidin LMycoheptin Candicidin Ham ycin Heptafungin Levorin A Levorin B Trichom ycin Trichomycin A Fungimycin (perim ycin) Aureofungin 363.0 364.0 362.0 360.0 363.0 360.0 356.0 362.0 361.5 363.0 361.0 382.0 406.0 383.5 407.5 382.0 406.0 379.0 401.0 383.0 406.0 380.0 402.0 379.0 398.0 381.0 404.0 382.0 404.0 384.5 407.0 383.0 406.0 980 1670 1890 985 1730 1910 940 1600 1800 - 962 918 702 1000 798 620 1000 947 625 863 750 - - - 660 - __ - - - f216 (CsH,N) 2 21 22 22 2 23 2 359.0 379.0 402.0 +83.7 (DMF) 46 47 48 Non-polyene Antimycin A 225.0 321.0 - 625 116 - 130 - - antibiotics Cycloheximide 287.0 - - * DMF = dimethylformamide; DMSO = dimethyl sulphoxide; C5H,N = pyridine. antibiotics may also be co-produced with the polyene antibiotic.Strains of Streptomyces noursei produced cycloheximide and n y ~ t a t i n . ~ ~ Antimycin has been demonstrated as a contaminant of the aromatic heptaene a s ~ o s i n ~ ~ and levoristatin, which is co-produced with levorin by strains of Actinomyces Z e v o ~ i s , ~ ~ has been identified as an antimycin A complex.gs Antimycin A and cycloheximide have been isolated from strains of Streptomyces g ~ i s e u s , ~ ~ @ which are used for the production of candicidin, so there is a possibility that either of these non-polyene antibiotics could contaminate candicidin.TABLE I11 TOXICITY RANGE OF POLYENE ANTIFUNGAL ANTIBIOTICS, THEIR PREPARATIONS AND CO-PRODUCED NON-POLYENE ANTIFUNGAL ANTIBIOTICS TO MICE LD,,/mg kg-l in mice Antibiotic Oral Intraperitoneal Subcutaneous Amphotericin A - 450 - Natam ycin 1 500 250* 5 ooo* Nystatin 2 300-8 000 8-1 60 - Amphotericin B >8 000 280-1 640 - Candidin > 100 7-3 6 30 Candicidin 98-400 2.1-7 - Candicidin A - 47-65 277 Candicidin B - 53 159 Hamycin 100-300 8-1 8 - Levorin - 9-40 - Trichomycin 300 2.2 17 Fungim ycin > 500 Amphotericin B + sodium deoxycholate (Fungizone) - 88 Amphotericin B methyl ester - 1320 - Antimycin A 55 1.7-7.6 25 - - - Cycloheximide 375 - 150-1 60 * Rats. Intravenous - 5-10 3 4-11.3 1.5 - Reference 49 5, 60 7, 37, 61 20, 29, 37, 49 62, 63 7, 37 53, 64 63 6, 7 5 55 56 - 6-9 2.2 43 250 4-4.5 29, 39 75-106 0.9 150-160 28, 37, 39 57 58326 THOMAS: ANALYSIS AND ASSAY OF POLYENE TABLE IV ANTIFUNGAL ACTIVITY RANGE OF POLYENE ANTIBIOTICS AND CO-PRODUCED Analyst, Vol.101 NON-POLYENE ANTIBIOTICS Minimum inhibitory concentration/ pg ml-l 7- A \ Candida Antibiotic albicans Aniphotericin A 4.7 Natamycin 3.0-6.0 Nystatin 1.2-3.1 Amphotericin B 0.05-3.7 Candidin 0.6 Mycoheptin 0.62 Candicidin 0.02-0.5 Ham ycin 0.01 Heptaf ungin 0.5-1.0 Levorin A 0.2 Levorin B 0.2 Trichom ycin 0.03-0.7 Fungimycin 0.06-0.1 Aureof ungin 0.01-0.4 Amphotericin B methyl ester 0.05-0.5 Antimycin A - Cycloheximide > 1 000.0 cryptococcus neoformans 3.1 3.0-10.0 1.6 0.1-0.6 0.62 0.005 0.05 0.025 0.17 0.05 - - - 0.01-0.05 0.1-0.5 >0.24 - Saccharomyces Trichophyton cerevisiae rztbrum 2.4 0.9-1 5.0 0.8-3.1 0.05-0.6 0.5 0.55 0.01-0.03 0.012-0.01 5 - 0.05 0.05 0.07 0.03-0.25 0.1-0.4 4.7 3.0-15.0 6.3-14.0 0.5-30.0 - - 500 20.0 5.0 - - - I 0.4-1.0 0.0 5-0.5 0.5-1.0 40.0 - 10.0 > 1 000.0 Reference 59 5 5 5, 28, 29, 60 5 19 3, 5, 21 5 21 22 22 5 5 5 28, 60 61 62 Mixtures of closely related polyenes with the same chromophore have been reported in nystatin (A1 and A2),69 candidin (candidin, candidinin, candidoin) ,17 candihexin (I and 11),70 candicidin (A, B and C),53 heptafungin (A and B),21 levorin (A and B)22 and trichomycin (A and 13).71 The formation of amphotericin A and amphotericin B from Streptomyces nodosus72 is an example of the co-production of a tetraene and a heptaene.Similarly, a pentaene antifungal antibiotic is produced with mycoheptin.ls The hexaene complex candi- hexin produced by a mutant strain of the candidin-producing Streptomyces viridojavzts contained two heptaenes identified as candidin component^.^^ Numerous non-polyene anti- bacterial antibiotics have been isolated along with the polyene antibiotics, but these are less likely to affect the analysis and assay of the polyene antibiotics although they may affect toxicity. The same polyene has been recovered from micro-organisms isolated from widely separated areas. In 1955, a Streptomyces strain taken from a soil sample in Natal, named Stre@to.wcyces rzatalenis, produced the antibiotic n a t a m y ~ i n . ~ ~ Four years later, the antibiotic tennecetin was isolated from a Streptomyces strain collected in Chattanooga, Tennes~ee,'~ which was shown to be identical with n a t a m y ~ i n .~ ~ The antibiotics candicidin, hamycin, levorin and trichomycin have very similar properties and the evidence to date suggests that they consist of the same major component with variations in their minor component^.^^-^^ Tetramycin, claimed to be a new tetraene antibiotic, has been isolated from a strain of Streptomyces nozt~sei,~8 a micro-organism usually associated with the production of the tetraene n yst atin. 63 The proposal to classify the streptomycetes species based on the use of morphological and physiological criteria, including the identity of antibiotics produced by different strains,'O cannot be completely realised until the definitive differentiation of many of the polyene antibiotics has been completed.Because of the co-production of antibiotics by cultures producing the polyenes, strain selection and special extraction procedures have to be employed in order to obtain high titres of the required antibiotic. Strains have been selected for the production of nystatin that have lost the ability to produce cycloheximide and related antibiotics.80 Most polyene antibiotic- producing cultures are grown under submerged aerobic conditions at temperatures ranging from 26 to 34 "C in a nutrient broth containing one or more nitrogen sources, a carbon- containing metabolisable energy source and inorganic Nitrogen sources included yeast products, meat products, soyabean meal and casein hydrolysates.Among the amino- acids examined experimentally, asparagine had a stimulating effect on the synthesis of amphotericin BS3 and candi~idin.~~ Energy sources used include carbohydrates, sugars,May, 1976 ANTIFUNGAL ANTIBIOTICS. A REVIEW 327 alcohols and animal and vegetable oils. Alteration of the carbohydrate composition of the medium in Streptomyces nodusus fermentations resulted in a significant change in the ratio of amphotericin A to amphotericin B.85 The addition of glucose during the latter stages of fermentation extended the period of synthesis of candicidin, provided that the medium was maintained at pH 8, resulting in yields of 4 g 1-1.82 Optimum yields of amphotericin B (4-5 g 1-l) were obtained with glucose as the sole carbon source.83 Increased titres of amphotericin B, candicidin and nystatin have been recorded by the addition to the media of nickel, zinc, cobalt, iron and m a g n e s i ~ m .~ ~ ~ ~ ~ S ~ ~ Fermentation is usually stopped after 90-168 h, depending on the polyene being produced. Most of the polyene antibiotic is found associated with the mycelium, an exception being candihexin, for which about 80% of the antibiotic was found in the supernatant broth.70 The first step in the recovery process is the removal of the mycelium from the fermented medium and the antibiotic is then extracted from the mycelium with a suitable solvent. Amphotericin A and B are extracted with propan-2-01 at pH 10.5; the extract is neutralised and concentrated in vacuo, forming a precipitate which is washed with water and acetone before drying.This crude material is slurried with a 2% m/V solution of calcium chloride in methanol and the amphotericin A dissolves. The insoluble residue is extracted with acidic dimethylformamide, the extract is diluted with methanol, the pH is adjusted to 5 and the subsequent addition of water results in the precipitation of amphotericin B.49 On a pilot scale, amphotericin B has been purified by countercurrent distribution with a system composed of chloroform - methanol containing 3% m/V of calcium chloride - water - butan-1-01 Levorin is recovered by adjusting the whole culture fluid to about pH 5.6, adding diatomite and filtering. The residue is washed several times with methylene chloride to remove levoristatin (antimycin A) and other impurities and then extracted several times with solutions of propan- 2-01 and acetone. The use of solvents of low boiling-point allows the levorin to be concentrated at 3540 "C, reducing thermal inactivation to a minimum.The levorin precipitates on concentration and the precipitate is washed with methylene chloride prior to drying.85 Candicidin was originally purified by chromatography of a chloroform suspension of candi- cidin on a cellulose powder column.86 Purification of the crude eluate from the column, containing about 20% m/V of candicidin, is achieved by solution and precipitation from a pyridine - acetic acid - water solvent system followed by thorough washing with acetone to give a preparation of 70% purity. Countercurrent distribution (pyridine - ethyl acetate - water, 3.5 + 6.5 + 8.3) was used to obtain preparations of 90% purity, final purification being achieved by repeated precipitation from pyridine - acetic acid - water mixtures.Pure candicidin has been crystallised as needles or rosettes from aqueous tetrah~drofuran.8~ Fraction A was eluted with Sorensen's phosphate buffer (pH 8), fraction €3 with a mixture of 20% W/V pyridine and 5% V/V hydrochloric acid (95 + 5) and fraction C with pyridine - butan-1-01 - water (3 + 4 + 7).88 Extraction of trichornycin with a specially developed solvent mixture (pyridine - dioxan - water - butyl acetate, 30 + 40 + 52 + 18) resulted in samples of high potency but uncertain p~rity.8~ The main fraction of trichomycin has been purified by countercurrent distribution [chloroform - methanol - borate buffer (pH 8.4), 2 + 2 + 11 and shown to consist of two similar components, A and B.'l The use of substrates labelled with carbon-14 has shown that propionate and acetate are the precursors of the macrolide ring of amphotericin B83 and n y ~ t a t i n .~ ~ The aromatic moiety of candicidin is synthesised from glucose through the shikimic acid pathway to p- aminobenzoic acid, which is then incorporated into ~andicidin.~~ Studies with candihexin suggest that the biologically active candihexins A and B are formed from the biologically inactive candihexins E and I?, which lack mycosamine. Glycosylation was thought to occur during cell secretion of the antibiotic, and it is likely that the formation of candicidin is similar .g The purity of most polyene antibiotics will be dependent on the strain of antibiotic- producing micro-organism, the fermentation conditions and the extraction procedures used.The antibiotic could contain a non-polyene antibiotic, an unwanted polyene antibiotic, inactive precursors and degradation products. The active impurities could interfere with a biological assay and influence toxicity, while inactive impurities could interfere with a spectrophoto- metric assay. (20 + 20 + 10 + 1).83 Chromatography on an alumina column was used to fractionate trichomycin.328 THOMAS : ANALYSIS AND ASSAY OF POLYENE Analyst, Vol. 101 Chemical Methods of Analysis and Assay A polyene can be assigned to a specific chromophore group, e.g., tetraene or heptaene, by its ultraviolet spectrum.The heptaenes are differentiated by the presence of either mycosamine or perosamine as the amino-sugar and often by the presence of an aminophenyl moiety (either @-aminoacetophenone or N-methyl-9-aminoacetophenone) . The amino-sugar can be identified by chromatography after acid hydrolysis of the p ~ l y e n e . ~ ~ ? ~ ~ The aminophenyl moieties are characterised after alkaline hydrolysis of the polyene by chromatography and ultraviolet absorption (p-aminoacetophenone, Amax. 318 nm ; N-methyl-P-aminoacetophenone, Amax. 330 nm).20s22 A limit test for p-aminoacetophenone is included in the monograph for candi- cidin in the British Pharmac~poeia.~~ Numerous paper and thin-layer chromatographic systems have been used to check the identity and homogeneity of the polyene antibiotics; some are given in Tables V and VI.The considerable variation between the results of different groups is a reflection of the variation in the purity of the antibiotic and the experimental conditions used. Tailing is frequently encountered with both methods, possibly owing to the heterogenous nature of many of the polyenes, to degradation of the antibiotic during development , or to the unsuitability of the chromatographic methods for such insoluble compounds, and it is possible that the polyene may complex with divalent metals present in the stationary phase. Chromatographic methods are rarely described in detail, and only one literature reference mentions the temperature at which the R, values were determined.g1 No mention is ever made of precautions taken against photo-inactivation or oxidation, yet both may accelerate decomposition of the polyenes.Even the solvent used for preparing the polyene solution for chromatography can influence the result. Nystatin dissolved in either methanol or dimethylformamide was degraded to give two spots on a thin-layer chromatogram but only one spot when dissolved in a mixture of methanol and dimethylformamide (1 + 3).94 For identification purposes, the methods used for detecting the polyenes do not have to be very sensitive, but sensitive methods are needed in order to assess purity. Spray reagents described for use in thin-layer chromato- graphy are not very satisfactory; a load of 10-100 pg was necessary to show the main compon- ent on a thin-layer c h r ~ m a t o g r a m , ~ ~ ~ ~ ~ whereas a load of 2.5-10 pg gave good resolution of components when detected by ultraviolet absorption96 or biological activity.Detection of active components is usually made with a strain of Saccharomyces cerevisiae. Care should be taken that residues of the developing solvents do not inhibit the growth of the yeast. A direct spectrodensitometric method has been developed to quantify the components of candihexin and candidin after separation by thin-layer chr~matography.~~ Chromatographic systems are satisfactory for distinguishing polyenes with different chromo- phore groups, e.g., amphotericins A and B59s83 or a co-produced pentaene in mycoheptin.19 The tetraene natamycin was differentiated from the tetraenes amphotericin A and nystatin by the use of two separate chromatographic systems.50 However, the aromatic heptaenes ascosin , candicidin , hamycin, levorin and trichomycin have never been convincingly separated from each other by chromatographic methods.75s76~100s101 Thin-layer chromatography has been used to check the purity of polyene antibiotic deri~atives,~8p~~p~~ and has also been used to show that de-esterification of the methyl ester of amphotericin B does not occur in mice.104 Inactive components in candihexin complex lacking mycosamine, being less polar than the active candihexins, were demonstrated by thin-layer chromatography.92~gs Contamination of ascosin with antimycin A was detected by developing an acetone extract of ascosin on a silica- gel plate with chloroform - acetone (9 + l).64 The absence of levoristatin (antimycin A) in levorin is confirmed by thin-layer chromatography on an alumina plate at pH 4 using water- saturated butanol as solvent.Levoristatin, if present, is revealed as a bluish lilac spot at R, 0.6 when examined under an ultraviolet lamp.lo5 An attempt to identify polyenes by paper electrophoresis was unsuccessful; no single system differentiated all of the polyenes examined and no indication was given of the stability of the polyenes under the conditions of the experiment.lo6 Column chromatography has been used for the purification of polyene antibiotics but as a batch elution procedure it is unsuitable for use as an analytical method. High-performance liquid chromatography using a bonded non-polar stationary phase and a polar mobile phase (water - methanol - tetrahydrofuran, 420 + 90 + 45-90) allowed the separation of the com- ponents in different classes of polyeneslo7 by varying the proportion of tetrahydrofuran in the Chemical analysis is used to establish the identity and purity of polyene antibiotics.TABLE V PAPER-CHROMATOGRAPHIC SYSTEMS FOR 10 ANTIFUNGAL ANTIBIOTICS RP values of antibiotics L Solvent system (V/V) Ascosiii Candicidin Tricliomycin Water-saturated butan-1-ol* 0.33 Butan-1-01- aceticacid - water ( 2 + 1 + 1)* 0.88 Hutan-1-01 - pyridine - water (1 + 0.6 + 1)* 0.79 Acetone -water (1 + 1)* 0.43, 0.7; Butan-1-01 - McIlvaines buffer (pH 6.5) 0.00, 0.55 Methanol -ammonia - water (20 f- 1 + 4) 0.00, 0.61, 0.93 Butan-1-01 - acetic acid - water (20 + 1 + 25) 0.00,0.42,1.0 Butan-1-01 - pyridine -water (4 4- 1 + 5 ) Butan-1-01 - pyridine - water (6 + 1 + 7) Butan-1-01 - ethanol -water (4 + 2 + 1) Butan-1-01 -ethanol - water (3 + 3 + 2) Butan-1-01 - pyridine - water (6 + 4 + 5)* Methanol - 25% ammonia - water (20 + 1 + 4)* - Ethyl acetate - pyridine -water (4 + 3 + 2)* Water-saturated butan-1-ol* - Butan-1-01 - pyridine - water (I + 0.6 + l ) * Acetone - water (1 + 1)* Methanol - ammonia -water (20 $- 1 + 4) Butan-2-01 - ethanol - water (2 f 2 + 1) Amy1 alcohol -methanol -water (10 + 5 + 4) Ethyl acetate - pyridine -water (12 + 5 + 14)* Butan-1-01 - acetic acid -water (20 + 1 + 33)* Propan-1-01 -water (4 + 1)* Propan-2-01 - water ( 3 + 1)* Butan-1-01 - acetic acid -water (20 + 1 + 25)* Propan-1-01 - water (4 + 1)* Ethyl acetate - pyridine - water (12 + 5 + 14) Butan-1-01 -methanol - water (2 + 2 + :)* Butan-1-01 -methanol - water (4 + 1 + 2)* Butan-1-01 - pyridine - water* - - - - - - - - - - _- - - __ - - - - - - - Ethyl acetate - pyridine - 25% acetic acid ( 5 + 5 + 4)* - Whatman No.1 paper specified. t R, varied with purity of samplc. 0.44 (J.35 0.S6 0.55 0.50 0.76 0.38, 0.69 0.69 o.no,0.52 o . o o , o . ~ ~ 0.00, 0.14, 0.66 0.00, 0.43, 0.65 0.00, 0.44 0.00, 0.40 0.39 0.54 0.28 0.32 0.53 0.48 0.71 - 0.55 - 0.66 - __ 0.26 - 0.75 0.58 - 0.42,0.67 0.60 0.59 - 0.49 - 0.17 - 0.34 - 0.41 - 0.88 ._ 0.60 - 0.58 - 0.14, 0.17 - 0.94 - 0.32 - 0.28 - 0.72 0.89 0.84 - Nystatin Antimycin A Cycloheximide Levorin A Aureofungin Hamycin - - - - - - - - - - - - - - - - - - 0.69 0.66 - - - - 0.63 0.59 0.18, 0.42 0.30, 0.67 0.26, 0.64 0.71 0.96 - __ - - - - - - - - - - - - 0.26 0.55 0.77 0.44, 0.6T - - 0.08-0.22t 0.22-0.44t 0.38 0.8 7-0.88t 0.59 0.56 0.17 0.95 0.31, 0.64 0.27, 0.61 0.69 Reference 9i !I 7 9i 97 95 98 98 99 99 99 99 100 100 100 101 101 101 101 46 46 20 20 20 20 24 24 24 24 24 24 24330 THOMAS: ANALYSIS AND ASSAY OF POLYENE TABLE VI THIN-LAYER CHROMATOGRAPHIC SYSTEMS FOR 13 ANTIFUNGAL ANTIBIOTICS Analyst, VoL.101 In all systems silica gel G was used as the stationary phase. R, values of antibiotics Ampho- Ampho- Trichp- Solvent system (V/V) Detection* Natamycin Nystatin tericin A tericin B mycm Ethanol - ammonia - water Butan-1-01 -acetic acid - water (8 + 1 + 1)t A A f3 + 1 + lkt Me'thanol 1 p;dpan-Z-ol -acetic acid Methanol - acetone - acetic acid (90 + 10 + I)$ (8 + 1 + 1)t Ethanol - ammonia - dioxan - water Butan-1-01 - pyridine -water Butan-1-01 - ammonia -methanol - water (20 + 1 + 2 + 4) Chloroform - methanol -borate buffer Butan-1-01 - ethanol - acetone - 32% Chloroform - 95% ethanol - water (8 f 1 + 1+ !If (3 + 2 + I)$ (PH 8.3) (7 + 6 + 1) ammonia (2 + 5 + 1 + 3) (50 + 50 + 8) B B C and D C and D B B E E Butan-1-01 - ammonia -methanol - Chloroform - methanol -borate buffer Butan-1-01 - ethanol - acetone - 32% water(20+1+2+4) B (pH8.3) (7 + 5 + 1) ammonia (2 + 5 + 1 + 3) Lower phase of chloroform - (2 + 2 + 1) B E E E methanol - 2O%ammonia 0.34 0.18 0.33 - 0.45 0.34 0.18 0.33 - 0.17 0.40 0.54 - 0.18 - 0.54 0.66 - 0.45 - 0.18 0.28 0.19 0.19 0.30 0.55 0.65 0.56 0.32 0.68 - - - 0.07 0.13 - - - 0.60 0.49,0.67 - - - 0.41 - - - - I - R, values of antibiotics Anti- Candi- mycin A cidin Reference 0.72 - 102 102 - 103 I 103 95 95 - 0.13 21 - 0.49,0.67 21 - 0.55 37 - 0.34, 0.74, 0.95 91 - - - - - - - - I I \ Myco- Hepta- Candi- heptin fungin Hamycin Levorin hexin Candidin 21 21 - 2s 0.06 0.10 0.13 0.13 - - 0.18 0.55 0.49,0.67 0.49,0.67 - - - - - - 0.42 * Method of detection- A.loo/ KMnO and 0 20/ bromophenol blue. B I 0.2% p-dim&hylakin:benzaldehyde in HISO, containing PeCl,. C. 5'7' KMnO D': 0rothophos;horic acid, heat for 5 min at 100 "C. E : Spectrodensitometric. t Buffered with phosphate (pH 8). 3 Activated for 1 h at 110 "C. mobile phase. Nystatin was shown to consist of three components, two tetraenes and one heptaene.Candidin was resolved into at least five components, two trace components were detected in amphotericin B and candicidin was separated into five components. Candicidin was differentiated from hamycin and trichomycin and there was an indication that hamycin and trichomycin differed although there were similarities in their minor components. Pyrolysis gas ~hromatography~~ can differentiate nystatin and amphotericin B from each other and from candicidin, levorin and trichomycin. Small differences in the pyrograms of samples of levorin were explained as showing quantitative differences in the composition of the samples. Gas chromatography of the products of chemical degradation of candicidin, levorin and trichomycin indicated that differences do exist between these antibiotics.108 The successful purification of trichomycin by countercurrent distrib~tion~~J0~ led to the use of this technique as an analytical method for identifying and examining the polyene antibiotics.In an attempt to compare the results of different workers, the partition coefficient, K , has been calculated using the equation A list of the systems used and results obtained is given in Table VII. where Nmax. is the tube number containing the maximum concentration of antibiotic and NT is the total number of transfers made. Many polyenes have been shown to be complex mixtures when subjected to countercurrent distribution. Levorin was separated into two main components, levorin A and B, and levorin A was further fractionated into three components. Three components have beenMay, 1976 ANTIFUNGAL ANTIBIOTICS.A REVIEW TABLE VII PARTITION COEFFICIENTS OF POLYENE ANTIFUNGAL ANTIBIOTICS OBTAINED IN COUNTERCURRENT DISTRIBUTION SYSTEMS Solvent system (V/V) NT* Chloroform - methanol - borate 240 200 300 49 200 buffer (pH 8.2-5.4) (2 + 2 + 1) 200 600 80 3 60 240 200 200 300 560 Chloroform - methanol - sodium acetate (pH 8.4) (2 + 2 + 1) Chloroform - methanol - water Chloroform - methanol - citrate + (2 + 2 + 1) (2 + 2 + 1) phosphate buffer (pH 6.26) citrate + phosphate buffer (pH 5) (12 + 17 + 29) (3.5 + 6.6 + 8.3) n-Amy1 alcohol - isoamyl alcohol - 200 Pyridine - ethyl acetate - water 49 Partition coefficient, Kt Trichomycin A 0.41 1 ; trichomycin B 0.739 Amphotericin B 2.70; candidin 3.44; mycoheptin 4.26 Candicidin 0.78 ; levorin A 0.74 Levorin A 0.90; levorin B 5.1 Trichomycin (major fraction) 0.61 ; trichomycin Candicidin 0.6 1 ; levorin 0.6 1 ; hamycin 0.61 Candicidin 1.02 ; levorin A2 1.07 Levorin A1 0.60; levorin A2 1.07; levorin A3 1.37 Candidoin 1.81 ; candididin 2.65; candidin 4.16 Trichomycin A 0.47 Hamycin 1.10 Aureofungin 0.72 Candicidin 0.38; heptafungin A 0.77 Candihexin 1.19; 1.37; 1.85; 2.18; 2.60; 3.37 (minor fraction) 1.31 Nystatin A1 4.6; nystatin -42 16.S Levorin A 1.00; levorin B 0.46 331 Reference 71 19 22 22 76 76 100 100 17 109 20 24 21 70 69 32 * NT = total number of transfers; Nmsx.= number of tube containing maximum concentration. f Calculated from Nm&x./(NT - Nmax.). demonstrated in both candidin17 and in heptafungin complex; the major component hepta- fungin A accounted for about 80% of the total and heptafungin B had thesame partition coefficient as trichomycin B.21 The separation of candihexin complex was markedly improved by reducing the pH of the partitioning system from 8.3 to 6.25, which revealed four hexaene, two heptaene and two non-polyene antifungal component^.^^ On the basis of countercurrent distribution, the main active fractions of candicidin, hamycin, levorin and trichomycin were considered to be the same, but there were differences between minor, biologically inactive, components.76 On re-running the main fractions of the above antibiotics, the distribution curves plotted on the basis of antifungal activity and absorption at 380 nm did not match. The tubes of maximum antifungal activity and maximum absorption varied by six and nine tubes, depending on the antibiotic, which suggests that more than one component may have been present.Reproducible differences between the partition coefficients of candicidin and levorin A have been reported, e.g., candicidin 0.785, levorin A 0.744,22 and candicidin 1.02, levorin A (complex) 0.98, levorin A2 1.O7.lo0 It has been suggested that the heptaenes can be differentiated by the partition coefficients in two solvent systems22 (see Table VIII). TABLE VIII PARTITION COEFFICIENTS OF HEPTAENE ANTIBIOTICS Partition coefficient of heptaene A I \ Solvent system Group 1* Group 2* Group 4* Pyridine - ethyl acetate - water (3.6 + 6.6 + 8.3) 0.5 1.0 2.0 Chloroform - methanol - borate buffer (pH 8.3) (2 + 2 f 1) 4.0 1.0 0.1 * See Table I.The group 2 heptaenes could be further divided into three sub-groups on the basis of their distribution in chloroform - methanol - borate buffer (pH 8.3) (2 + 2 + l), giving K values of about 1 .O for candicidin, trichomycin A and levorin A, 1.5 for trichomycin B and 5.1 for levorin B. K values obtained in other laboratories do not accord with this classification, possiblyAnalyst, Vol. I01 because the experimental conditions varied, e.g., differences in phase ratio [upper : lower, 1 : 1 (ref. 71) or 1:2 (ref. 76)], temperatures where quoted (11 OC;lo9 17, 19 and 25 "C71) and the purity of the samples, all variables which can affect the partition coefficient. In addition, the antibiotic may have degraded during the separation, which can take between 10 h (ref. 71) and 15 h (ref.76) for 220 and 200 transfers, respectively. Countercurrent distribution has yielded useful information on the composition of the polyene antibiotics, but it would be even more useful if workers reported the experimental conditions and tried to standardise these conditions and the calculation of the partition coefficients. Monitoring the separation on the basis of absorption and biological activity would reveal the presence of both active and inactive components, while increasing the efficiency of countercurrent distributions would provide a powerful tool for the analysis of these complex substances. An apparatus for increasing the number of transfers without increasing the running time, under controlled temperature, has been built on the concept of the coil planet centrifuge.l1° Chemical assays for the polyene antibiotics include titrimetric, colorimetric and spectro- photometric methods and a non-aqueous titration for nystatin has been described.Optimum conditions were obtained for both potentiometric and visual end-point titrations by using a solution of 5-50 mg of nystatin in 15 ml of acetic acid - dioxan (1 + 14), titrated with 0.01 N perchloric acid.lll The yellow colour produced when nystatin is hydrolysed with sodium hydroxide112 has formed the basis of a colorimetric assay for nystatinll3 and amphotericin B,114 interference by extraneous matter being minimised by extracting the pigment with chloroform. This procedure resulted in an assay which correlated well with the microbiological assay and which was considered suitable for monitoring the stability of the antibiotics, including ampho- tericin B in intravenous infusions115 provided that the infusion was kept in darkness.A quantitative method for the determination of natamycin, nystatin and amphotericin B, based on the formation of a blue colour in a strongly acidic medium, has been described in detail.103 Samples were evaluated by comparison with a standard graph; the method was shown to be precise when tested on the intact antibiotic. Spectrophotometric assay of the polyene antibiotics involves determining the net absorbance of the central peak by subtracting the average absorbance of the minima on each side of the peak itself. The calculated specific absorption of the sample is compared against that of a standard of known purity.The antibiotic concentration can also be calculated from the E i g value of the pure antibiotic. Spectrophotometric methods are useful as they give a presumptive identification of the antibiotic. They are less sensitive than biological assays but more specific and sensitive than colorimetric methods, especially when measuring anti- biotic concentrations in body fluids and tissues. A correction for non-specific absorption in the near-ultraviolet region is required and for this reason the method is unsuitable for the determination of tetraene antibiotics in biological specimens. A spectrophotometric method has been developed to measure amphotericin B in whole blood and plasma,ll8 in which ampho- tericin B was extracted with butanol in order to reduce the high background absorbance that was due partly to trace amounts of haemoglobin and to recover any antibiotic that may be associated with cholesterol in the cellular fraction of whole blood.Spectrophotometric assays usually give higher values than do biological assays.l17 These differences may be due to degradation, which results in loss of activity without loss of the characteristic light absorption, or to the presence of an inactive aglycone as found in candihexin c0mplex.~2 A good correlation is apparent between the specific absorption at 380 nm and potency for samples of levorin.77 During stability studies on nystatin solutions, the correlation between biological activity and absorption at 360 and 320 nm disappeared for solutions over 24 h old.118 A spectrophotometric method for the measurement of nystatin in the presence of its degrada- tion products, in which irrelevant absorption was corrected for by the application of orthogonal functions, has been describedllg ; unfortunately, there was no concurrent biological evaluation of the nystatin samples.The requirements of the FDA120 include a limit test for the photo- degradation of nystatin, which is determined spectrophotometrically. A qualitative assay based on the enhanced fluorescence of amphotericin R in acidic solution and claimed to be sensitive and linear in the range 0.1-10 pmol 1-1 has been developed to measure the binding of amphotericin B to yeast cells.121 Chemical assays are convenient methods for evaluating the polyene antibiotics ; however, they do not supply information about antifungal activity and are best used when the composi- tion of the sample is known.332 THOMAS : ANALYSIS AND ASSAY OF POLYENEMay, 1976 ANTIFUNGAL ANTIBIOTICS. A REVIEW 333 Biological Assay Biological assay is the accepted criterion by which the polyene antibiotics and their prepara- tions are assessed. The usual methods are diffusion and turbidimetric assays and details of some of the established methods are given in Table IX. The basis of the biological assay is to determine the potency of the sample under test by comparison with a standard of defined activity. International standards exist for amphotericin B131 and n y ~ t a t i n , ~ ~ ~ and national standards for candicidin are available in the USA and UK and for trichomycin in Japan.There is a need for international reference preparations of candicidin and trichomycin, but as there is only limited information on the relative compositions of these and other related antibiotics such as hamycin and levorin, more information is required before reference samples of these anti- biotics can be e~tab1ished.l~~ Ideally, the composition of the sample and the standard should be identical but, because of the heterogeneity of most polyenes, especially the aromatic heptaenes, this is unlikely to be achieved, so it may be difficult to obtain statistically valid assays with parallel dose-responses. Similarly, batches of these antibiotics currently available may have relative compositions that are completely different from that of the material used to establish the standard. The minimum potency requirement of the FDA and USP for nystatin has been increased from 2 000 U mg-l (ref.120) to 4 400U mg-l (ref. 134) and the present international standard for nystatin established in 1963 may no longer be representative of material in current use. An indication of the interference that may be encountered as a result of heterogeneity can be seen by reference to the minimum inhibitory concentrations (Table IV). The presence of a heptaene in a sample of nystatin may inflate the assay result and even a less active component will have an additive effect, e.g., amphotericin A in amphotericin B. By using a strain of Candida tropicalis that was more sensitive to amphotericin B than to amphotericin A, a turbidimetric assay was developed for the assay of amphotericin B in the presence of amphotericin A.135 Similarly, a strain of Candida aZbicans that was not susceptible to cycloheximide was chosen for the assay of nystatin fermentation broth likely to contain cycloheximide.lZz For other samples of nystatin, including products that contain tetracycline (Mysteclin), Saccharomyces cerevisiae is used, as tetracycline reduces the activity of nystatin against Candida albicans.122 The alkaline diluent that is used to enhance diffusion in the diffusion assay for amphotericin B also inactiv- ated amphotericin A.126 Amphotericin A can be assayed in the presence of amphotericin B by using a mixed standard that approximates to the expected composition of the sample and by the selectivity of the diffusion process and the test organism Rhodotorula gZutinis.lzZ An assay based on the inhibition of the formation of blastospores of Candida tropicalis has been described for ascosin, candicidin and n y ~ t a t i n .l ~ ~ Inhibition of respiration of sensitive yeast cultures has been used to assay nystatin or amphotericin B; the amount of carbon dioxide produced by the yeast was inversely proportional to the antibiotic concentration.ls7 This assay was adopted for use with an AutoAnalyzer,l26 the amount of carbon dioxide being determined by the degree of decolorisation of an alkaline phenolphthalein solution. The incorporation of a continuous dilution system and a sample module of large capacity has further automated the method.13* Conductimetric measurements showed that the leakage of intracellular constituents from yeast cells treated with amphotericin B or nystatin was proportional to the concentration of the antibiotic139 and the use of potassium-sensitive e l e ~ t r o d e s l 4 ~ ~ ~ ~ ~ to monitor the loss of potassium ions from susceptible cells treated with polyene antibiotics should allow this effect to be used for the rapid assay of polyene antibiotics.A modified agar diffusion assay has been used for trichomycin, in which the antibiotic solution was placed on the surface of seeded agar in a test-tube. The concentration of the antibiotic was proportional to the distance between the liquid - agar interface and the depth where growth was ~isib1e.l~~ It is usual to use freshly prepared suspensions of test organisms for the assay of polyene antibiotics.Candida tropicalis grown in liquid inoculum medium can be used for up to 2 weeks when stored as a suspension at 4 O C . 1 2 6 Saccharomyces cerevisiae (ATCC 9763) suspen- sions in a 0.5% m/V solution of peptone in water could be kept frozen at -70 "C for up to 12 months with little change in the dose-response to amphotericin B, and unfrozen suspensions could be used for 2 weeks if kept a t 4 OC.143 Suspensions of Saccharomyces cerevisiae (ATCC 2601) were not recommended for storage at -70 "C, though a saline suspension was usable during 4 weeks when stored at 4 "C. When suspended in 0.012 M phosphate buffer and stored334 Test organism Rhohtorula glutinis (Squibb 2358) Saccharomyces ccrm-siae (A.T.C.C. 9763) Saccharomyccs ccrevisiae (N.C.Y.C.87) Saccharomyces ccrevisiac (A.T.C.C. 2601) Candida albicans (Squibb 1539) Candida trofiicalis (A.T.C.C. 13803) S a C c h a ~ ~ m y ~ s certvisiae (N.C.Y.C. 87) Saccharmyces ccrevisiac (A.T.C.C.9763) Saccharomyces cneviskc (A.T.C.C.9763) Paccilomyces uarwti (MSSC 5605 NIAID) Candidu tropicalis (A.T.C.C. 13803) S ~ C C ~ ~ T ~ Y C C S ctrcvisiac (N.C.Y.C. 87) Pascilomyccs v w w t i (MSSC 6605 NIAID) Candida albicans Yu 1200 THOMAS: ANALYSIS AND ASSAY OF POLYENE TABLE IX ASSAY METHODS FOR THE POLYENE ANTIBIOTICS Analyst, Vol. 101 Antibiotic solution$ Amount Incubation L I range§/ - Solution mi--’ Tempera- Time/h Reference Method* Medium7 ‘ Antibiotic D D D D D T D D D D T D T D D D D 3 5 1 1 4 10 1 1 1 1 10 2 9 1 8 6 I Amphotcricin A Natamyciii Nystatin Amphotericin B Candicidin Hamycin Levorin Trichom ycin ture/oc 1 mg ml-l DMSO, dilute with 10-40 pg 24 SO% V/VDMSO All dilutions with 50% V/V 14.6-50 pg 30 methanol dilute 10 ml to 200 ml with asolution containing 9.56% m/V of KH,PO, and 11.5% V/V of 1 N KOH 1 000 U ml-l, DMF, dilute to 12.8-31.2 U 30 256 U ml-l DMF, dilute with 10% m/V potassium phosphate buffer (pH 6.0) 1 000 U ml-l DMSO, dilute 26-200 U 37 with SO% V/V DMSO 1 000 U r$-l DMSO, dilute 3.25-8 U 30 with mdum 10 75 mg in 50 ml of DMF, 25-1oou 35-37 60 mg per 100 ml DMF, dilute 1 4 U 36-37 10 ml to 100 ml with DMF, dilute with potassium phosphate buffer (pH 10.6) containing 8% V/V DMF 1 mg ml-l DMSO, dilute to 31.2 vg ml-.l DMSO, dilute with potassium phosphate buffer (pH 10.5) 0.64-1.66 “vg” 50 20 wg ml-l DMSO, dilute 0.6-1.96 “yg” 37 with potassium phosphate buffer (pH 10.5), keeping the concentration of DMSO a t 5% v/v 10 “mg”ml-1 DMSO, dilute 0.01-0.15 “vg” 30 1 to 10 with 60% V/V DMSO, then dilute 1 to 40 with 60% V/V propan-2-01, dilute with Dotassium phosphate h e r (pH 10.5) DMSO, dilute to 30 p ml-’ 0.03-0.1 “pg” 30 with 25% V/V DMS8, dilute to 1 u ~ m l - ~ with alkaline m’dium 10 containing 0.1% m/V butylated hydroxyanisole, leave for 20 min, dilute with potassium phosphate buffer (pH 6) 26 mg in 30 ml DMSO 0.1-2.0 U 30-32 1 mg ml-l DMSO, dilute with distilled water 0.03-0.12 “pg” 2 500 U ml-1 in alkalinised 1-10 U 80 60% V/V propan-2-01, dilute with potassium phosphate buffer (pH 10.5) 60% V/V ethanol, dilute 1-6 pg 97 with an aqueous solution of 0.02% m/V Tween 80 1 mg ml-* DMSO, dilute with 30-80 U - phosphate buffer (pH 7) 2 mg ml-’solution containing 25-100 U 37 50% V/V acetone and 0.008% m/V NaOH, dilute with 25% V/V propan-2-01 D E diffusion assay; T = turbidimetric assay.f Number refers to medium given in Table X. § Amount/ml-I stated by mass (pg), units of activity (U) or micrograms of activity (“vg”). 11 Now called CochlioboZw lunarus. 7 Now called Candida utilis. Dilute with solvent to assay levels; DMSO = dimethyl sulphoxide; DMF = dimethylformamide. 24-36 24 16 16-18 isao 3-4 16 16-18 16-18 24 3-4 16 24 18-20 - - 122 16,123 124 126 126 126 124 la6 126 127 126 ias l2b 127 It9 108 130May, 1976 ANTIFUNGAL ANTIBIOTICS.A REVIEW 335 at 4 "C the culture proved satisfactory for the assay of nystatin for 1 ~ e a r . 1 ~ ~ Paecilomyces varioti has been used for the assay of amphotericin B and hamycin in biological f l ~ i d s . 1 ~ 6 3 ~ Inoculated agar slants have to be incubated for 1 week at 30 "C in order to obtain the mature spores which are necessary as inoculum to obtain the high sensitivity required for the assay. Media used for the assay of polyene antifungal antibiotics are listed in Table X. Most assay methods require the polyene antibiotic initially to be dissolved in a polar solvent (dimethylformamide or dimethyl sulphoxide). This solution is then further diluted in an aqueous buffer, which results in the formation of a micellar suspension which does not diffuse freely.This effect probably accounts for the poor diffusion of amphotericin B in agar and the poor dose-response observed with many polyene diffusion assay methods. In a search for a suitable diluent for amphotericin A and B, methanol was rejected as being too volatile to pipette accurately, propan-l- and -2-01 were toxic to the test organism, a colloidal suspension prepared in aqueous sodium deoxycholate foamed and dimethyl sulphoxide was viscous. A mixture of 30% V/V of dimethyl sulphoxide plus 40% V/V of methanol in water maintained amphotericin A and B in a true solution at a concentration of 0.1 mg ml-l but not at 1.0 mg ml-l.l35 Aqueous diluents containing 80% V/V of dimethyl sulphoxide are satisfactory for diffusion assays, while higher concentrations are inhibitory. The activity of nystatin in an agar diffusion assay was dependent on the solvent used, the activity decreasing in the order propylene glycol, methanol, 40% V/V aqueous dimethyl sulphoxide and water.145 TABLE X MEDIA USED FOR THE ASSAY OF THE POLYENE ANTIFUNGAL ANTIBIOTICS Concentrations of ingredients are given in grams per litre.Number of medium f 1 - Ingredient Peptone Yeast extract Ma1 t extract Beef extract Dextrose NaCl KC1 KH,POI K,HPO, Na,HPO, NaOH Glycerol Beef infusion Agar ::3, PH 1 2 9.4 9.4 4.7 4.7 2.4 2.4 10.0 12.0 10.0 10.0 - - - - - - 23.6 23.6 6.0-6.2 6.0-6.2 3 6.0 3.0 3.0 133.0 10.0 - I - - - - - - I - 18.0 Not stated 4 6.0 3.0 6.0 1.6 21.0 10.0 - - - - - - - - - 16.0 5.0-6.6 6 2.5 - - - 10.0 - - 8.5 - - - - 1.5 - - 16.0 7.0 6 7 10.0 10.0 1.6 - - - 10.0 5.0 10.0 - 30.0 0.6 - - - - 3.0 - - 0.5 I 2.0 - - - - 600ml - 16.0 15.0 6.2-6.4 7.8-8.0 8 5.0 - - - 10.0 10.0 0.1 - - - 0.05 - - 10 ml Not stated 7.0 - 10 5.0- 1.6; 1.5 11.0 3.6 3.68 1.32 - - - - - - - - - 6.0 The polyenes react with susceptible sterol-containing cell membranes, causing the leakage of intracellular constituents; compounds that compete with the membrane sterol for the antibiotic, such as most sterols,l41 digitonin126 and serum, will reduce its activity.l45 Chole- sterol and protein are reported to be involved in the binding of levorin to serum lipoproteins, but only cholesterol is associated with the binding of amphotericin B.146 The inhibition of glycolysis due to the loss of potassium ions from yeast cells treated with nystatin is prevented at neutral pH by the presence of ammonium or potassium ions,147 and care should be taken in the respirometric assay that the pH is low enough to abolish this interference.In nystatin diffusion assays, an increase in the concentration of monovalent cations re- sulted in an increase in zone size.l17 Increasing the sodium chloride content of nystatin assay agar from 1 to 4% m/V produced an increase in the size of zones of inhibition, but at the expense of decreased growth. In the absence of added sodium chloride there were no zones336 THOMAS: ANALYSIS AND ASSAY OF POLYENE Analyst, Vol. 101 of inhibition even at a concentration of nystatin of 200U ml-1.126 The monovalent cations were thought to reduce the non-specific absorption of nystatin in the medium.Special extraction procedures have to be used for the biological assay of preparations that contain polyene antibiotics. Amphotericin B and candicidin products are extracted with dimethyl sulphoxide148 and nystatin products are blended with dimethylformamide to extract the active ingredient.120 Diethyl ether is used to dissolve the ointment base of amphotericin B ointment, and hexane is used for the same purpose with candicidin ointment. The assay of nystatin animal feeds is complicated by the very low concentration of antibiotic present. Modifications to earlier m e t h o d ~ ~ ~ ~ ~ l ~ ~ have included the replacement of Candida alicans by Saccharomyces cerevisiae as the test organism and dilution of the methanolic extract with 10% phosphate buffer instead of 50% aqueous dimethyl sulphoxide.The animal feed was first extracted with methanol, a portion of which was inactivated by heating and used as the blank for the preparation of the standard. A collaborative study showed the procedure to be satisfactory and it was recommended as the official method.150 The assay of antibiotics in biological specimens involves the measurement of low concentra- tions in small samples and requires methods of high sensitivity. An allowance has to be made for the natural antifungal activity of the biological specimen,126 which can be effected by using, as a diluent for the standard, blood or serum of the same type and species as the samples to be assayed. A more accurate method is to determine the pre-dose antifungal activity of the individual specimen, but such an approach is not always practicable.When samples from patients infected with pathogenic fungi are to be assayed, the sample and stock standard solution can be pasteurised by heating them at 56 "C for 30 min.126 In a microturbidimetric assay126 for amphotericin B, the samples are diluted with 95% ethanol in order to precipitate the proteins and extract the antibiotic. The ethanolic solutions of standard and sample are dispensed into the compartments of a polystyrene tray and evaporated to dryness, Inoculated assay broth is then added to each compartment, the tray incubated and the opacity of each mixture read. The method for nystatin is similar except that the ethanol extraction is omitted so as to prevent loss of sensitivity because of dilution and drying.Corrections had to be made for the intrinsic colour and turbidity of most body fluids. The sensitivity of the method was 0.01 pg ml-l for amphotericin B and 1 U ml-l for nystatin. A sensitive diffusion assay for hamycin and amphotericin B in serum at the 0.01-0.02 pg ml-1 level has been described.12' The severest limitation to the method resulted from the presence of natural antifungal agents in various sera, which invalidated the assay. Stability There are general statements in the literature describing the polyenes as a group of unstable compounds which in solution are decomposed by exposure to air, heat and light.2J In the dry state and in the absence of heat and light, the polyenes are stable.l61151 Most polyenes in aqueous conditions exist as micellar suspensions, which may confer some protection by shielding the labile sites of the molecule.The methyl ester of amphotericin B formed a finer colloidal suspension than amphotericin B in water and this suspension was less stable than a suspension of the parent material.151 Similarly, an aqueous solution of natamycin was less stable than an aqueous suspension.16 Nystatin and amphotericin B solutions in phosphate - citrate buffer were stable between pH 5 and 7,152 while natamycin was stable between pH 5 and 9.16 Nystatin lost its activity more readily at extremes of pH than did amphotericin B or natamycin,lo3 and the loss of nystatin activity proceeded at a greater rate than did the loss of absorption at the wavelength maxima.152 The hydrochloride salt of amphotericin B methyl ester is a solid and in aqueous solution was found to be less stable at pH 44.5 than at pH 6-6.5.151 At an alkaline pH, the resistance of natamycin solution to oxidation was decreased.16 The methylpentaene filipin in methanolic solution underwent autoxidation, the decrease in activity being correlated with the decay of the ultraviolet spectra due to the loss of a double bond, leading to conversion of the chromophore into a tetraene epoxide; extended autoxida- tion resulted in the formation of polymeric materials.153 Very dilute methanolic solutions (O.OO1~o m/V) were stable for prolonged periods but concentrated solutions in the absence of a nitrogen atmosphere were unstable even in the dark at 4 "C.The addition of small amounts of the antioxidant butylated hydroxyanisole had a stabilising effect .153 Similarly, the oxid- ative inactivation of natamycin was accompanied by marked changes in the ultravioletMay, 1976 ANTIFUNGAL ANTIBIOTICS. A REVIEW 337 spectrum and disappearance of the maxima, which was not the case when photo-inactivation occurred.l6 Nystatin solution was oxidised in the presence of air and loss of biological activity proceeded at twice the rate of loss of absorption maxirna.l5* The inactivation of nystatin solution was increased at increased temperature and by ultraviolet irradiation, while the presence of an antioxidant reduced the rate of ina~tivati0n.l~~ The oxidation of nystatin powder was increased in the presence of water, while iron(II1) ions accelerated the degradation, which could be prevented by the addition of a chelating agent.l5‘j The products of nystatin oxidation were organic acids (mainly succinic acid) ; acetone and acetaldehyde were present if the oxidation occurred in the presence of water.157 Acid hydrolysis of nystatin resulted in the elimination of mycosamine and the formation of a pentaene chromophore.13 Reducing agents reacted irreversibly with amphotericin B, destroying its biological activity.158 Exposure of aqueous solutions of natamycin to ultraviolet irradiation resulted in the loss of biological activity, but the absorption spectrum did not disappear.Inactivation was thought to be due to a change in the configuration of the chromophore from trans to cis; chlorophyll protected natamycin from i r r a d i a t i ~ n l ~ ~ and oxidation of natamycin was prevented by chlorophyll or ascorbic acid.159 Ultraviolet irradiation of the methylpentaenes caused oxidative degradation, with the formation of an intermediate tetraene and finally a triene.Nystatin, when irradiated, yielded a mixture of two acidic compounds and a neutral triene.160 The heptaenes trichomycin, hamycin and aureofungin lost their polyenic character and bio- logical activity when exposed to ultraviolet irradiation, the reaction resembling a retroaldol cleavage accompanied by polymerisation. The aromatic compound $-aminoacetophenone, found as a degradation product of the irradiation of hamycin, was shown to accelerate the decomposition.161 The change in absorption spectrum caused by the ultraviolet irradiation of the heptaene antibiotic D J 400 was thought to be the result of isomerisation of the chromo- phore with one cis double bond to one with an all-trans configuration.Biological assay of intravenous infusions of amphotericin B in dextrose solution showed that a loss of activity of 26yo occurred when the infusions were exposed to light at room tempera- ture over a 3-d period, but a colorimetric assay did not detect any loss of activity.l15 When the infusions were protected from light, there was no reduction in activity. No loss in activity was noted when infusions were exposed to normal lighting conditions over a period of 4-8 h, the time required to administer the dr~g.162J~~ The addition of sodium and chloride ions resulted in a turbid solution which lost 25% of its activity within 4 h owing to the colloidal aggregation of amphotericin B163; this was not commented upon when infusions were prepared with dextrose solution that contained 0.2% m/V of sodium chloride.162 Conclusions Satisfactory qualitative and quantitative analytical methods have still to be developed that will permit the definition of the composition of the polyene antibiotic complexes and hence a comparison of the relative compositions of different antibiotics and different batches of the same antibiotic.This will be especially important if the heptaene antibiotics are used for their hypocholesterolaemic effect, which may be due to a particular component in the complex. The most promising methods available are high-performance liquid chromatography and countercurrent distribution.A satisfactory thin-layer chromatographic separation for the polyenes is still not available as a simple procedure for detecting active and inactive com- pounds. A separate assay to measure hypocholesterolaemic activity may be necessary if it is found not to be related to antifungal activity. The choice of diluent used to prepare the antibiotic solutions for assay is critical, as it affects the assay, and the choice of more suitable diluents could improve the dose-response of the bioassay. A rapid and accurate method of measuring the antibiotic concentration in blood is still required in order to monitor patients receiving the drug parenterally. The results of the chemical assessment of antibiotic stability should be treated with caution unless they are correlated with biological activity, as there is evidence that biological activity and chemical measurement do not correlate for degraded polyene antibiotics.References 1. 2. 3. Vining, L. C., Hindustan Antibiot. Bull., 1960, 3, 37. Oroshnik, W., and Mebane, A. D., Fortschr. Chern. Org. NatStofe, 1963, 21, 17. Dutcher, J. D., “Kirk-Othmer Encyclopedia of Chemical Technology,” Second Edition, Volume 16, John Wiley and Sons, New York, p. 133.338 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. 49. 59. 48. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. THOMAS: ANALYSIS AND ASSAY OF POLYENE Analyst, Vol. 101 Perlman, D., Prog.I n d . Microbiol., 1967, 6, 3. Mechlinski, W., Laskin, A. I., and Lechevalier, 11. A., Editors, “Handbook of Microbiology,’’ Volume Hamilton-Miller, J. M. T., Bact. Rev., 1973, 37, 166. Drouhet, E., “Systemic Mycoses,” Ciba Foundation Symposium, J. and A. Churchill Ltd., London, Oroshnik, W., Vining, L. C., Mebane, A. D., and Taber, W. A., Science, N.Y., 1965, 121, 147. Pledger, R. A., and Lechevalier, H., Antibiot. A., 1965-56, 249. Golding, B. T., Rickards, R. W., Meyer, W. E., Patrick, J. B., and Barber, M., Tetrahedron Lett., Ceder, O., and Hansson, R. W., Tetrahedron, 1967, 23, 3753. Chong, C. N., and Rickards, R. W., Tetrahedron Lett., 1970, 59, 5145. Borowski, E., Zielinski, J., Falkowski, L., Ziminski, T., Galik, J., Kolodziejczyk, P., Jereczek, E., Mechlinski, W., Schaffner, C.P., Ganis. P., and Avitabile, G., Tetrahedron Lett., 1970, 44 3873. Borowski. E., Zielinski, J., Ziminski, T., Falkowski, L., Kolodziejczyk, P., Golik, J., Jereczek, E., and Adlercreutz. H., Tetrahedron Lett., 1970, 45, 3909. Raab, W. P., “Natamycin (Pimaricin), Its Properties and Possibilities in Medicine,” Georg Thieme Verlag, Stuttgart, 1972, p. 12. Borowski, E., Falkowski, L., Golik, J., Zielinski. J., Ziminski, T., Mechlinski W., Jereczek, E., Kolodziejczyk, P., Adlercreutz, H., Schaffner, C. P., and Neelakanton, S., Tetrahedron Lett., 1971, 22, 1987. Borowski, E., and Schaffner, C. P., “Abstracts of Communications, Vth International Congress on Biochemistry, Moscow, Aug. 10-16, 1961,” Pergamon Press, Oxford and London, 1961, p.3. Borowski, E., Malyshkina, M., Kotienko, T., and Soloviev, S.. Chemotherapia 1964-66, 9, 359. Deshapande, G. R., Kerur, D. R., and Narasimhachari, N., Hindustan Antibiot. Bull., 1966, 8, 185. Kalasz, H., Szell, V., Gyimesi, J., Magyar, K.. Horvath, I., and Szabo, I., Acta Microbiol. Acad. Sci. Borowski, E., Malyshkina, M., Soloviev, S., and Ziminski, T., Chenzotherapia, 1965-66, 10, 176. Nakano, H., J . Antibiot., 1961, 14A, 72. Deshapande, G. R., and Narasimhachari, N., Hindustan Antibiof. Bull., 1966, 9, 76. Lee, C. H., and Schaffner, C. P., Tetrahedron, 1969, 25, 2229. Haegele, I(. D., and Desiderio, D. M , Biomed. Mass Spect., 1974, 1, 20. Ganis. P., Avitabile, G., Mechlinski, W., and Schaffner, C. P., J . Am. Chem. SOC., 1971, 93, 4560. Mechlinski, W., and Schaffner, C.P., J . Antibiot., 1972, 25, 256. Bartner, E., Zines, H., Moe, R. A., and Kulesza, J. S., Antibiot. A., 1957-68, 63. Lampen, J . O., Morgan, E. R., Slocum, A., and Arnow, P., J . Bact., 1959, 78, 282. Finkelstein, A., and Holz, R., “Membranes,” Volume 2, Marcel Dekker, New York, 1973, p. 377. De Kruijff, B., and Demel, R. A., Biochim. Biophys. Acta, 1974, 339, 57. Bennett, J . E., New En@. J . Med., 1974, 290, 30. Garrod, L. P., and O’Grady, F., “Antibiotics and Chemotherapy,” Third Edition, E. and S. Livingstone, Falkowski, L., Golik, J., Kolodziejczyk, P., Pawlak, J., Zielinski, J., Ziminski, T., and Borowski, E., Schaffner, C. P., and Mechlinski, W., J . Antibiot., 1972, 25, 259. Bruzzese, T., Cambieri, M., and Recusani, F., J .Pharm. Sci., 1975, 64, 462. Bonner, D. P., Mechlinski. W., and Schaffner, C. P., J . Antibiot., 1972, 25, 261. Keim, G. R., Poutsiaka, J. W., Kirpan, J., and Keysser, C. H., Science, N.Y., 1973, 179, 584. Bodey, G. P., Adachi, L , and Jones, V., Curr. Ther. Res., 1974, 16, 207. Gordon, H. W., and Schaffner, C. P., Pvoc. Nut. Acad. Sci., 1968, 60, 1201. Keshin, J. G., I n t . Surg., 1973, 58, 116. Schaffner, C . P., and Gordon, H. W., Proc. Nut. Acad. Sci. U.S.A., 1968, 61, 36. Parekh, A. C., Creno, R. J.. and Dave, C. V., Res. Commun. Chenz. Path. Pharmac., 1974, 9, 307. Tech. Rep. Sev. Wld Hlth Org., 1969, No. 430. Bhate, D. S., and Acharya, S. P., Hindustan Antibiot. Bull., 1964, 6, 170. “The Merck Index,” Eighth Edition, Merck & Co. Inc., Rahway, N. J., 1968, p.62. Kornfeld, E. C., Jones, R. G., and Parke, T. V., J . Am. Chem. Soc., 1949, 71, 160. Sternberg, T. H., Wright, E. T., and Oura, M., Antibiot. A., 1955-56, 566. Struyk, A. P., Hoette, I., Drost, G., Waisvisz, J. M., van Eek, T., and Hoogerheide, J. C., Antibiot. A ., Seneca, H., Antibiot. A., 1956-56, 697. Taber, W. A., Vining, L. C., and Waksman, S. A., Antibiot. Chemother., 1954, 4, 455. Lechevalier, H. A., Acker, R. F., Corke, C. T., Haenseler, C. M., and Waksman, S. A., Mycologia, Kligman, A. M., and Lewis, F. S., Proc. SOC. Exp. Biol. Med., 1953, 82, 399. Waksman, S. A., and Lechevalier, H. A., “The Actinomycetes,” Volume 111, Baillihre, Tindall and Oswald, E. J., Reedy, R. J., and Randall, W. A., Antibiot. A., 1955-56, 236. Vezina, C., Antimicrob. A g .Chemother., 1966, 757. Waksman, S. A., and Lechevalier, H. A., “The Actinomycetes,” Volume 111, Baillihre, Tindall and Gold, W., Stout, H. A., Pagano, J. F., and Donovick, R., Antibiot. A., 1955-56, 579. Howarth, W. R., Tewari, R. P., and Solotorovsky, M., Antimicrob. A g . Chemother., 1975,7, 58. 111, CRC Press, Cleveland, Ohio, 1973, p. 93. 1968, p. 206. 1966, 30, 3551. Gdulewicz, M., Shenin, Y., and Kotienko, T., Tetrahedron Lett., 1971, 8, 686. Hung., 1972, 19, 111. London, 1971, p. 235. J . Antibiot.. 1976, 28, 244. 1957-58, 878. 1953, 45, 155. Cox Ltd., London, 1962, p. 397. Cox Ltd., London, 1962, p. 248.May, 1976 ANTIFUNGAL ANTIBIOTICS. A REVIEW 339 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. 116. 116. 117. 118. 119. 120. 121. 122. Ahmad, K., Schneider, H. G., and Strong, F. M., Archs Biochem., 1950, 28, 381. Whiffen, A. J., J . Bact., 1948, 56, 283. Hazen, E. R., and Brown, R., Proc. Soc. Exp. Biol. Med., 1951, 76, 93. Kinsky, S. C., Gronau, G. R., and Weber, M. M., Molec. Pharmac., 1965, 1, 190. Kuljbakh, V. 0.. Sokolov, L. B., Sveshnikov, J. F., Kuznetsova, 0. A., Malinovskaja, G. N., Raigorod- Shenin, Y. D., Venediktov, V. V., Onoprienko, V. V., and Egorenkova, A. N., Antibiotiki, 1973, 18, Karasawa, K., Tanaka, N., Yonehara, H., and Umezawa, H., J . Gen. APPZ. Microbiol., 1969, 5, 13. Whiffen, A. J., Bohonos, N., and Emerson, R.L., J . Bact., 1946, 52, 610. Shenin, Y. D., Kotenko, T. V., and Exemplyarov, 0. N., Antibiotiki, 1968, 13, 387. Martin, J. F., and McDaniel, L. E., J . Antibiot., 1974, 27, 610. Hattori, K., Nakano, H., Seki, M., and Hirata, Y., J . Antibiot., 1956, 9A, 176. Vandeputte, J., Wachtel, J. L., and Stiller, E. T., Antibiot. A . , 1955-66, 587. Burns, J., and Holtman, D. F., Antibiot. Chemother., 1959, 9, 398. Divekar, P. V., Bloomer, J. L., Eastham, J. F., Holtman, D. F., and Shirley, D. A., Antibiot. Chemothev., Khokhlova, Y . M., Korenyako, A. I., Nikitina, N. I., Puchnina, A. V., and Blinovno, Z., AZZg. Mikro- Kalasz, H., Gyimesi, J., Uskert, A., Magyar, K., and Horvath, I., Acta Chim. Acad. Sci. Hung., 1967, Burrows, H. J., and Calam, D. H., J .Chromat., 1970, 53, 566. Dornbcrger, K., Fugner, R., Bradler, G., and Thrum, H., J . Antibiot., 1971, 24, 172. Pridham, T. G., Antimicvob. Ag. Chemother., 1963, 104. Vanek, Z., Spizek J., Puza M., Vondracek, M., and Dolesilova, L.. Abstr. Int. Congr. Biochem., 1964, Zurabova, E. R., and Romankova, A. G., BY. Pat., 1 254 212,1971. Liu, C.-M., McDaniel, L. E., and Schaffner, C. P., Antimicrob. Ag. Chemother., 1975, 7, 196. Linke, H. A. B., Mechlinski, W., and Schaffner, C. P., J . Antibiot., 1974, 27, 165. Brewer, G. A., and Frazier, W. R., Antimicrob. Ag. Chemother., 1961, 212. Musilikova, M., Folia Microbiol., 1961, 6, 175. Siminoff, P., U.S. Pat., 2 872 373, 1959. Waksman, S. A., Lechevalier, H. A., and Schaffner, C. P., Bull. Wld Hlth Org., 1966,33, 219. Hosoya, S., Hamamura, N., and Ogata, S., J .Antibiot., 1965, 8A, 48. Hosoya, S., J. Antibiot., 1956, 9A, 129. Birch, A. J., Holzapfel, C. W., Rickards, R. W., Djerassi, C., Suzuki, M., Westley, J., Dutcher, J. D., Liu, C.-M., McDaniel, L. E., and Schaffner, C. P., J . Antibiot., 1972, 25, 116. Martin, J. F., and McDaniel, L. E., Antimicrob. Ag. Chemother., 1976, 7 , 208. “British Pharmacopoeia 1973, Addendum 1975,” HM Stationery Office, London, 1975, p. 10. Nussbaumer, P. A., Pharm. Acta Helv., 1968, 43, 462. Ochab, S., Dissnes Pharm. Pharmac., 1970, 22, 351. Martin, J. F., and McDaniel, L. E., J . Chromat., 1975, 104, 151. Ammann, A., and Gottlieb, D., AppZ. Microbiol., 1955, 3, 181. Vining L. C., Taber, W. A., and Gregory, F. J., Antibiot. A . , 1954-55, 980. Hattori, K., Ohashi, M., Aoyama, K., and Mizuno, T., J .Antibiot., 1956, 8B, 89. Bosshardt, R., and Bickel, H., Experientia, 1968, 24, 422. Divekar, P. V., Vora, V. C., and Khan, A. W., J . Antibiot., 1966, l l A , 63. Ikekawa, T., Iwami, F., Akita, E., and Umezama, H., J . Antibiot., 1963, 16A, 66. Dryon, L., J. Pharm. Belg., 1966, 21, 433. Monji, N., Bonner, D. P., Hashimoto, Y., and Schaffner, C. P., J . Antibiot., 1975, 28, 317. Inter-Republic Specifications, Levorin, IRS-42, No. 3595-68, Ministry of Health of the USSR, Ochab, S., Dissnes Pharm. Pharmac., 1971, 23, 205. Mechlinski, W., and Schaffner, C. P., J . Chromat., 1974, 99, 619. Calam, D. H., J . Chromat. Sci., 1974, 12, 613. Nakano, H., J . Antibiot., 1961, 14A, 68. Ito, Y., and Bowman, R. L., J . Chromat.Sci., 1973, 11, 284. Mazor, L., and Papay, I<. M., 2. Analyt. Chem., 1961, 184, 272. Unterman, W. H., Revue Chim., Bucharest, 1961,12,415; Analyt. Abstr., 1962, 9, 847. Chang, J. C., Honig, A. B., Warren, A. T., and Levine, S., J. Pharm. Sci., 1963,52,636. Chang, J. C., Honig, A. B., Warren, A. T., and Levine, S., J . Pharm. Sci., 1963, 52, 603. Gallelli, J . F., Drug Intell., 1967, 1, 102. Herd, P. A., and Martin, H. F., Clin. Biochem., 1974, 7, 369. Horvath, I., and Koczka, I., Nature, Lond., 1964, 203, 1305. Venables, P., and Russell, A. D., Antimicrob. Ag. Chemother., 1975, 7 , 121. Amer, M. M., Wahbi, A. M., and Habeeb, A. A., J . Pharm. Pharmac., 1976, 27, 377. “Code of Federal Regulations,” Title 21 (1973), Part 148K, US Government Printing Office, Washing- Kotler-Brajtburg, J ., Medoff, G., Schessinger, D., and Kobayashi, G.S., Antimicvob. Ag. Chemother., Gerke, J . R., Levin, J. D., and Pagano. J. F., in Kavanagh, F., Editor, “Analytical Microbiology.” skaja, V. J., and Kholodova, G. V., Br. Pat., 1070 799, 1967. 483. 1961, 11, 377. biol., 1963, 3, 195. 51, 431. 445. and Thomas, R., Tetrahedron Lett., 1964, 23, 1485. Moscow, 1968. ton, D.C., 1973, p. 486. 1974, 6, 770. Academic Press, New York and London, 1963, p. 387.THOMAS Platt, T. B., and Isaacson, D. M., in Laskin, A. I., and Lechevalier, H. A., Editors, “Handbook of Microbiology,” Volume 111, CRC Press, Cleveland, Ohio, 1973, p. 1017. “British Pharmacopoeia 1973,” HM Stationery Office, London, 1973, p. A102. “Code of Federal Regulations,” Title 21 (1973), Part 141.101, US Government Printing Office, Washington, D.C., 1973, p.14. Platt, T. B., Levin, J. D., Gentile, J., and Leitz, M. A., in Kavanagh, F., Editor, “Analytical Micro- biology,” Volume 11, Academic Press, New York and London, 1972, p. 147. Shadomy, S., McCoy, J . A., and Schwartz, S. I., A p p l . Microbiol., 1969, 17, 497. “British Pharmacopoeia 1973, Addendum 1975,” HM Stationery Office, London, 1975, p. A8. Maniar, A. C., and Mavdikar, S., Hindustan Antibiot. Bull., 1962, 4, 168. “Minimum Requirements of Antibiotic Products,” Ministry of Health and Welfare, Japanese Govern- Lightbown, J . W., de Rossi, P., and Isaacson, P., Bull. Wld Hlth Org., 1972, 47, 343. Lightbown, J . W., Kogut, M., and Uemura, K., Bull. Wld Hlth Org., 1963, 29, 87. Tech. Rep. Ser. Wld Hlth Org., 1968, NO. 384, 12. “United States Pharmacopeia,” Nineteenth Revision, United States Pharmacopeial Convention Gerke, J. R., and Madigan, M. E., Antibiot. Chemother., 1961, 11, 227. Tarbert, J. E., and Sternberg, T. H., Mycologza, 1954, 46, 263. Gerke, J. R., Gentile, J.. Stander, H., and Pagano, J. F., Antibiot. A., 1960, p. 563. Platt, T. B., Weisblatt, H., and Guevrekian, L., Ann. N.Y. Acad. Sci., 1968, 153, 571. Isaacson, D. M., and Platt, T. B., Bact. Proc., 1968, 68, 1. Hammond, S. M., Lambert, P. A., and Kliger, B. N., J . Gen. Microbiol., 1974, 81, 325. Gale, E. F., J . Gen. Microbiol., 1974, 80, 451. Tsubura, E., J. Antibiot., 1954, 7B, 201. Salt, D. T., Arret, B., and Wilner, J., J . Pharm. Sci., 1973, 62, 2040. Tanguay, A. E., and Bogert, A. B., A@l. Microbiol., 1974, 27, 1175. Wawrzkiewicz, K., Polskie Archwm Wet., 1973, 16, 85. Klirnov, A. N., and Nikiforova, A. A., Experientia, 1973, 29, 661. Marini, F., Arnow, P., andLampen, J. O., J . Gen. Microbiol., 1961,24,51. “Code of Federal Regulations,” Title 21 (1973), Part 148 b and 148 v, US Government Printing Office, Platt, T. B., Levin, J. D., andMassey, M. A., J . A s s . 08. Analyt. Chem., 1969, 46, 675. Platt, T. B., and Itkin, A. G., J . A s s . Off. Analyt. Chem., 1974, 57, 536. Bonner, D. P., Mechlinski, W., and Schaffner, C. P., J . Antibiot., 1975, 28, 132. Hamilton-Miller, J. M. T., J . Pharm. Pharmac., 1973, 25, 401. Rickards, R. W., Smith, R. M., and Golding, B. T., J . Antibiot., 1970, 23, 603. Lolrshin, G. B., Zhdanovich, Y. V., and Kuzovkov, A. D., Antibiotiki, 1966, 11, 590. Belenkii, B. G., Solovev, S. N., Filippova, A. I., Nevinskii, A. A., Mamtsis, A. M., and Kogan, S. B., Antibiotiki, 1966, 11, 112. Bashkovich, A. P., Shenin, Y . I)., Kotenko, T. V., Raigorodskaya, V. Y., Karpenko, M. P., Krasno- baeva, N. A., Vasileva, N. G., Bogdanova, N. P., and Ekzemoplyarov, 0. N., Khim.-Farm. Zh., 1967, 1, 31. Zhdanovich, Y. V., Lokshin, G. B., and Kuzovkov, A. D., Khim.-Farm. Zh., 1967,1,42. Weiss, M. R., and Levine, H. B., Sabouraudia, 1972, 10, 132. Dekker, J., and Ark, P. A,, Antibiot. Chenzother., 1959, 9, 327. Narasimhachari, N., Dhawale, D. A., Joshi, V. B., and Gopalkrishnan, K. S., Hindustan Amtzbiot. Narasimhachari, N., Deshpande, G. R., and Patil, P. L., Hindustan Antibiot. Bull., 1967, 10, 1. Shadomy, S., Brummer, D. L., and Ingroff, A. V., Am. Rev. Resp. Dis., 1973, 107, 303. Block, E. R., and Bennett, J. E., Antimicrob. Ag. Chemother., 1973,4, 648. ment, Tokyo, 1959, p. 409. Inc., Rockville, Md., 1974, p. 348. Washington, D.C., 1973, pp. 415 and 540. Bull., 1967, 10, 290. Received November 1 Ith, 1976 Accepted December 16th, 1976 340 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163.

ISSN:0003-2654    

DOI:10.1039/AN9760100321

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

Analyst, May, 1976, Vol. 101, $9.341-347 34 1 The Determination of Arsenic(ll1) and Total Arsenic by Atomic-absorption Spectroscopy J. Aggett and A. C. Aspell Chemistry Department, University of Auckland, Auckland, New Zealand In the atomic-absorption determination of arsenic by the hydride evolution method with sodium borohydride, maintenance of the pH between 4 and 5 permits the selective determination of arsenic(II1) in mixtures of arsenic(II1) with arsenic(V). Total arsenic can be determined separately by evolution from 5 M hydrochloric acid. A dissolution technique has been developed for herbage and the method applied to the analysis of water and orchard leaves. As there is considerable evidence to suggest that the toxicity and physiological behaviour of arsenic are dependent on its valence state and in particular that arsenic(II1) compounds are more toxic than arsenic(V) compounds,l it seems prudent to develop sensitive methods for the determination of arsenic in its different valence states. At present, the most satis- factory methods for the determination of low levels of arsenic are based on the evolution of arsine, e.g., the Gutzeit, silver diethyldithiocarbamate and atomic-absorption spectro- photometric methods.Standard procedures for the former two methods, and earlier proposed atomic-absorption methods, are based on the use of zinc - hydrochloric acid as hydride generator. In the first instance it appeared that these methods should be capable of differen- tiating between arsenic(II1) and arsenic(V), as recipes invariably seem to state that arsenic(V) must be reduced to arsenic(II1) with tin(I1) - hydrochloric acid prior to arsine generation.However, it has been our experience with the silver diethyldithiocarbamate method that zinc - hydrochloric acid generates arsine from arsenic(V) just as readily as from arsenic(III), which is not surprising as it is a stronger reducing agent than tin(I1). In 1973, Schmidt and Royer2 reported the use of solid sodium borohydride - hydrochloric acid for the generation of arsine in the atomic-absorption spectrophotometric procedure. Among the claimed advantages of sodium borohydride was the fact that preliminary reduction of arsenic(V) to arsenic(II1) was not necessary, which implies that this procedure also measures the total arsenic concentration of the sample.Now, a breakdown of the process of formation of arsine from arsenic(V) suggests that there are two steps in the reaction : Reduction .. ' . (1) As(V) ---+ As(II1) . . . . As(II1) ---+ ASH, .. .. .. (2) Hydride transfer Also, the fact that the use of both zinc and sodium borohydride in hydrochloric acid media does not differentiate between arsenic(V) and arsenic(II1) suggests that under these rather acidic conditions reaction (1) proceeds at a rate similar to, or faster than, reaction (2). However, the redox potential of the arsenic(V) - arsenic(II1) couple is a function of pH (Fig. 1) and, as the rates of redox reactions that involve multi-electron transfer are often slow, it appeared that it might be possible to differentiate between arsenic(II1) and arsenic(V) provided that reaction (1) is slower than reaction (2) at higher pH.This paper reports a study on the effect of pH on the evolution of arsine from aqueous solutions of both arsenic(V) and arsenic(II1) and its application to the determination of specific forms of arsenic at trace levels. Sodium borohydride was selected as generating agent because it is effective as a hydride transfer reagent over a wider range of acidity than zinc - hydrochloric acid. It was used as an aqueous solution rather than as a solid in order to avoid problems that might have arisen from the employment of a fast heterogeneous reaction. Atomic-absorption spectro- scopy was chosen as the measurement technique as it appears to be the most satisfactory342 AGGETT AND ASPELL : THE DETERMINATION OF ARSENIC(III) Analyst, VoZ.101 in terms of sensitivity and precision at low levels and is more compatible with a rapid arsine generation technique than are the Gutzeit and silver diethyldithiocarbamate methods. 0.6 I 1 Fig. 1. Redox potential of arsenic(V) - arsenic(II1) couple as a function of pH. Experimental and Results Apparatus Atomic-absorption measurements were made with a Unicam SP 90A atonlic-absorption spectrophotometer fitted with an EM1 9663B photomultiplier tube using an air-entrained hydrogen - argon flame. The flow-rates were hydrogen 1.5 1 min-l and argon 7.5 1 min-l. The arsenic resonance line at 197.2 nm was used in preference to the more sensitive 193.7-nm line because at the latter wavelength the noise level was greater by a factor of about 2.5.The slit width was 0.1 mm for all measurements. Signals were recorded on a digital integrator with a count rate of 100 digits mV-1 s-l. This integrator was constructed by the Electronics Division of the Chemistry Department, University of Auckland. It was adapted to integrate absorbance with respect to time by connecting a logarithmic converter between the integrator and spectrophotometer. It consisted of a 250-ml separating funnel The arsine generator is illustrated in Fig. 2. Fig. 2. Arsine generator.May, 1976 AND TOTAL ARSENIC BY ATOMIC-ABSORPTION SPECTROSCOPY 343 fitted with two extra inlets, one (A) fitted with a glass stopper for introduction of samples, the other (B) fitted with a rubber septum through which the sodium borohydride solution was injected. The gas-bubbling tube was fitted with a sintered-glass frit, which was ground to enable it to be placed as close to the tap as possible in order to maximise the efficiency of removal of arsine from the generating solution.The 250-ml gas bottle was used as a ballast jar, which made it possible to operate the system with the aid of the single two-way tap. Measurements were made in the following manner. The sample to be analysed and the appropriate acid solution were pipetted through inlet A, the stopper was replaced and argon passed through the reaction vessel to remove air. The two-way tap was then turned back to its original position so that the argon passed directly to the flame. Sodium borohydride solution (2 ml) was then injected through the septum (B) and the argon immediately diverted to pass through the generating vessel and transfer the liberated arsine to the flame.The absorption signal was integrated over a period of 25 s. Chemicals With the exception of sodium borohydride, all chemicals were of analytical-reagent grade and the preparation and dilution of standard solutions were carried out with water that had been deionised and double-distilled. The sodium borohydride (BDH Chemicals, laboratory-reagent grade) contained relatively large amounts of arsenic, which were removed by dissolving the sodium borohydride in aqueous sodium hydroxide and then recrystallising it by the addition of purified dioxan. The purified crystals were cooled in ice and washed with methanol (yield 80-85%).For arsine generation, sodium borohydride was used as a 50 g 1-1 solution in 0.1 M sodium hydroxide. This solution was sufficiently stable to be used for several days. Determination of Arsenic( 111) The performance of the apparatus was evaluated by preparing calibration graphs using arsenic(II1) standards and arsine generation from 5 M hydrochloric acid solution. These calibration graphs were linear for arsenic levels below 0.3 pg ml-1 and gave a detection limit of about 25 ng of arsenic. At the 0.5-pg level (5 ml of 0.1 pg ml-l solution), the relative standard deviation was 2.5-3.0%. While this detection limit is not as impressive as those reported by Schmidt and Royer2 and Knudson and Christian,s it should be borne in mind that the purpose of this work was to examine the chemistry of the system.The use of a more sophisticated spectrophotometer will undoubtedly lead to some improvement in the detection limit. Preliminary experiments also included investigations of the variation in absorption signal with the concentration of sodium borohydride (Table I) and with sample volume (Table 11). The fact that it was physically easier to inject a smaller volume more rapidly formed the basis for the selection of 2ml of 5% sodium borohydride in the method. The results in Table I1 indicate that an eight-fold increase in sample volume results in only about a 5% loss in atomic-absorption signal for the same amount of arsenic. This is no doubt attributable to the careful design of the flushing system. TABLE I ATOMIC-ABSORPTION SIGNAL AS A FUNCTION OF SODIUM BOROHYDRIDE CONCENTRATION Sodium borohydride I A > Absorption Concentration/g 1-1 Volume/ml signal*/counts 10 10 1 604 20 5 1 823 50 2 1 830 100 1 1777 150 0.7 1786 * 5-ml samples of 0.1 p g ml-l arsenic(II1).The effect of pH on the generation of arsine from both arsenic(V) and arsenic(II1) was determined by measuring the atomic-absorption signals obtained from 0.5-yg samples (6 m344 AGGETT AND ASPELL : THE DETERMINATION OF ARSENIC(III) Analyst, VoZ. 101 TABLE I1 ATOMIC-ABSORPTION SIGNAL AS A FUNCTION OF SAMPLE VOLUME Arsenic (111) r A \ Absorption Concentration/pg ml-l Volume/ml signal/counts 0.2 0.1 0.05 0.025 5 10 20 40 3 626 3 578 3 581 3 418 of 0.10 pg ml-1 solution) in a series of buffers. It can be seen (Fig. 3) that while generation of arsine from arsenic(II1) is essentially independent of pH in the region studied, generation from arsenic(V) is never quantitative and falls continuously with increasing pH until above pH 3.5 it becomes negligible with respect to that generated from arsenic(II1).The explanation for this behaviour appears to be that as the pH is raised the rate of reaction (1) decreases until it eventually becomes much slower than the rate of hydrolysis of sodium borohyride under the same conditions. e - - - - 2000 - 0 1 2 3 4 5 PH Fig. 3. Absorption signals from (A) arsenic- (V) and (B) arsenic(II1) as a function of pH. Addition of alkaline sodium borohydride and its subsequent hydrolysis also affects the system by consuming protons. In order to ensure quantitative conversion of arsenic(II1) into arsine, it is necessary to maintain the pH a t a value no greater than 5.5.Citrate and acetate buffers (0.5 M) were found to have sufficient buffering capacity and maintained the pH to within 0.25 pH unit of their initial value during arsine evolution. However, when buffer solutions of lower concentration were used the solutions became alkaline during the reaction with sodium borohydride and lower atomic-absorption signals were recorded. The application of the method to mixtures of arsenic(V) and arsenic(II1) was examined by measuring the atomic-absorption signals obtained (a) for a series of samples containing different concentrations of arsenic(II1) in the presence of a constant concentration of arsenic(V), and (b) for a series of samples containing a constant concentration of arsenic(II1) in the presence of increasing concentrations of arsenic(V).In the first set of experiments, it was found that the presence of 1 pg ml-1 of arsenic(V) made no difference to the atomic- absorption signals obtained from arsenic( 111) solutions (concentration less than 0.3 pg ml-l) when arsine was generated from an acetic acid buffer a t pH 5 . When the arsine was generated from a buffer at pH 3.5, an increase of 3 4 % was observed in the atomic- absorption signals. The results for the second set of experiments (Fig. 4) show that at pH 5 it is possible to measure arsenic(II1) in the presence of arsenic(V) at ratios approaching 50: 1 ; at pH 3.5, the critical ratio is about 40: 1.May, 1976 AND TOTAL ARSENIC BY ATOMIC-ABSORPTION SPECTROSCOPY 345 0 2 4 6 8 10 12 Arsenic (V) concentration/pg ml-' Fig.4. Absorption signal from arsenic(II1) (0.1 pg ml-l) as a function of arsenic(V) concentration: A, citrate buffer, pH 3.5; and B, acetate buffer, pH 5.0. Interference by a number of inorganic ions was examined in three different generating media, viz., 5 M hydrochloric acid, 0.5 M acetate buffer (pH 4) and 1 M citrate buffer (pH 4). Results for the first two systems are summarised in Tables I11 and IV. Of the species examined, only antimony(II1) interfered in the citrate buffer and then only when its con- centration was above 100 pg ml-l. This interference by antimony has been shown to be caused by non-atomic absorption, although the species has not been positively identified. At the time these studies were undertaken, very little information on inteferences in arsine generation methods was available.Voge14 had mentioned that copper, nickel and cobalt slow the evolution of arsine by zinc - hydrochloric acid. Braman et aZ.,5 who used sodium borohydride generation from neutral solution coupled with emission detection, reported that both silver(1) and copper(I1) interfered when present in 20-fold excess. However, while this manuscript was in preparation, Smith6 published more detailed data on interferences in the sodium borohydride - hydrochloric acid method for arsine generation. TABLE I11 INTERFERENCES IN ARSINE GENERATION FROM 5 M HYDROCHLORIC ACID* Results are percentages of original signal. Concentration/pg ml-l A r I Interferent t 10 100 1 000 Co(I1) 100 100 73 Fe (I1 I) 100 100 110 Mn(VI1) 89 78 45 Ni(I1) 108 51 2 Sb(II1) 130 145 405 * Arsenic(II1) concentration 0.1 pg ml-l.t The following species did not interfere at the 1 000 pg ml-1 level: Na(I), K(I), Ca(II), Mg(II), Cr(VI), Cu(I1) and Zn(I1). The most significant discrepancy between the two sets of results appears to be the difference between the observed behaviour of copper(I1). Smith reported that copper(I1) suppressed the atomic-absorption signal by 10-50~0 when present at 1000-fold excess, whereas we found no interference at this level. This discrepancy may be related t o the difference in hydro- chloric acid concentrations in the reaction solutions. Smith generated arsine from solutions that were approximately 1 M in hydrochloric acid, while in the present study, the concen- tration was almost 5 M.At the latter concentration, copper(I1) is known to form chloro complexes, which would result in free copper(I1) concentrations well below the formal concentration. The ability of the acid anions to complex with inorganic cations also appears to explain the observed differences between interferences in the acetate and citrate buffers. In the346 AGGETT AND ASPELL: THE DETERMINATION OF ARSENIC(III) Analyst, VoZ. 101 TABLE IV INTERFERENCES IN ARSINE GENERATION FROM ACETATE BUFFER (pH 4)" Results are percentages of original signal. Inter f erent t 16- Co(I1) 46 Cu(I1) 29 Fe( 111) 44 Ni(I1) 100 S b (I 11) 110 Zn(I1) 100 Concentrationlpg ml-1 - 100 1000 2 <1 20 <1 37 34 100 28 122 410 81 28 * Arsenic(II1) concentration 0.1 pg ml-l.t The following species did not interfere a t the 1 000 pg ml-l level: Na(I), K(I), Ca(II), Mg(II), Cr(V1) and Mn(I1). latter buffer, which is more strongly co-ordinating, the degree of interference is minimal even with a 10000-fold excess of interferent. As a preliminary to the determination of arsenic(II1) and arsenic(V) in solid biological samples, it is necessary to dissolve the solid without loss of arsenic or oxidation of arsenic(II1). It was found that when the wet digestion procedure with concentrated nitric and sulphuric acids (Official Method lA7) was used, arsenic(II1) was oxidised to arsenic(V). However, when the acids were diluted 1 + 2 with water there appeared to be no significant loss of arsenic or oxidation of arsenic(II1) (Fig.5). Belcher et aL8 have recently reported that the addition of EDTA is effective in reducing interferences in the determination of arsenic by arsine generation. The action of EDTA is probably similar to that of citric acid. Both of these acids were investigated as potential generating media in the present study but the former was discarded because its lower solubility made it ineffective as a buffer on addition of alkaline sodium borohydride. 2 1600- 7 1200- g 3 9 en C .- v) .? 800 - +, 2 a 400- I I I I Nodilution 1:l 1:2 1:3 1:4 Dilution ratio, "Official Methods"'acid mixture Fig. 5. Influence of digestion procedure on oxidation and recovery of arsenic : A, arsenic( 111) with arsine generation a t pH 3.5 ; and B, arsenic(V) with arsine generation from hydrochloric acid.Suggested Procedure for Plant Materials Transfer 2 g of the dried and powdered sample into a 100-ml flask and add 10 ml of 1 + 2 nitric acid. Place a watch-glass on the mouth of the flask and heat the mixture at 70 "C for 5 min. Cool, add gradually 10 ml of 1 + 2 sulphuric acid, then heat at 70 "C for 15 min. Allow the solution to cool, add a further 5 ml of water and boil the mixture gently for 5 min. Some flocculence remains if the sample is not thoroughly wetted with nitric acidMay, 1976 AND TOTAL ARSENIC BY ATOMIC-ABSORPTION SPECTROSCOPY 347 in the first stage. This flocculence can be removed by filtration or centrifuging without affecting the recovery of arsenic. The acidity of the sample is then adjusted to permit the determination of total arsenic from 5 M hydrochloric acid, or arsenic(II1) from citrate buffer at pH 4.5-5.0, and the volume is made up to 50 ml.Aliquots of 5 ml are normally used for analysis, although larger volumes can be used to obtain greater precision at low concentrations. These procedures have been used to determine the total arsenic and arsenic(II1) contents of orchard leaves and apples in an orchard adjacent to a timber treatment plant and to analyse standard orchard leaves (NBS Standard Reference Material 1571 containing 14 pg ml-l of arsenic). The arsine generation method has been used to determine the total arsenic and arsenic(II1) content of water from geothermal areas. Typical results are given in Table V. The results for water were obtained by analysing 5-ml aliquots directly with sodium boro- hydride reagent. The results of these applications will be presented in detail when the studies are complete. TABLE V RESULTS OF ANALYSIS OF WATER AND LEAF SAMPLES Sample Arsenic (total)/pg g-’ Arsenic(II1) / p g g-I Water: Waikato River . . .. 0.164 0.036 Orchard leaves (local) .. 19.0 2.30 Orchard leaves (NBS 157i) . . 13.5 4.90 arsenic content of 6.62 pg ml-1. Water: Geyser . . . . .. 6.76* 3.85 * Analysis of this sample by the silver diethyldithiocarbamate method gave a total References 1. 2. 3. 4. 6. 6. 7. 8. Shroeder, H. A., and Balassa, J. S., J . Chron. Dis., 1966, 19, 85. Schmidt, F. J., and Royer, I. L., AnaZyt. Lett., 1973, 6, 17. Knudson, E. J., and Christian, G. D., Analyt. Lett., 1973, 6, 1039. Vogel, A. I., “A Text Book of Quantitative Inorganic Analysis,” Third Edition, Longmans, Green Braman, R. S., Justen, L. L., and Foreback, C. C., Analyt. Chem., 1972, 44, 2195. Smith, A. E., Analyst, 1976, 100, 300. Jolly, S. C., Editor, “Official, Standardised and Recommended Methods of Analysis,” W. HeRer and Sons, Cambridge, 1963, p. 6. Belcher, R., Bogdanski, S. L., Hendon, E., and Townshend, A., Analyst, 1975, 100, 522. and Co., London, 1961, p. 797. Received October 13th, 1975 Accepted December 31st, 1976

ISSN:0003-2654    

DOI:10.1039/AN9760100341

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

348 Analyst, May, 1976, Vol. 101, pp. 348-355 The Determination of the Precious Metals by Flameless Atomic-absorption Spectrophotometry G. L. Everett Johnson Mutthey Chemicals Ltd., Royston, Hertfordshire The optimum conditions for the determination of palladium, platinum, rhodium, ruthenium, iridium and osmium by carbon rod atomic-absorption spectrophotometry are described and data for sensitivity, detection limit and reproducibility are given. With the exception of osmium the sensitivity and detection limits are such that the working range of the atomic-absorption technique can be extended downwards by an order of magnitude or more when compared with that with an air-acetylene flame. The effect of mutual interference is investigated and methods of reducing it are proposed. The atomic-absorption spectrophotometric determination of the precious metals was first performed in air - propane flames1y2 but severe interference from other precious metals was generally found to occur.The use of the air -acetylene flame3-6 did not remove these interferences but the introduction of various releasing agents such as uranium7 and lanthanum8 reduced them to acceptable levels such that concentrations of the metals down to 1 pg ml-1 could be measured. The nitrous oxide - acetylene flame was found to give relatively interfer- ence-free results for the determination of rhodium9 and platinurnlo but in both instances the sensitivity was below that obtained with the air - acetylene flame. Although the precious metals are relatively involatile , the use of non-flame methods of atomisation , with their high atomisation efficiencies and absolute sensitivities, could improve this limit of determination to well below the 1 pg ml-l level.There is, however, only a limited number of publications on this subject and the results reported differ to some extent. The first report was by L’vov,ll who obtained 1% absorption sensitivities for palladium, platinum and rhodium of 4, 10 and 6 pg, respectively. Guerin12 investigated the use of the NIM rod, a carbon filament containing a slot 20 x 1.9 mm and 1.5 mm deep, the light beam being collimated parallel to the filament and just above the slot. By using 50-pl samples this technique gave aqueous detection limits that were about ten times better than those obtained in an air - acetylene flame.Adriaenssens and Knoop13 used a graphite tube furnace to investigate the determination of platinum, rhodium and iridium but the sensitivities achieved were only of the order of 1 ng and, owing to the high atomisation temperatures used, trouble was encountered from carbon particle scatter and emission signals from the glowing tube. Pearton and Mallett14 compared the NIM rod with a Mini-Massmann, the latter being a carbon rod with a transverse hole acting as an atom reservoir, and an atomiser consisting of a small graphite tube held between two support electrodes. They found the tube atomiser to be the most satisfactory in all respects, but it was subject to interferences from the other precious metals, although only a limited study of this aspect was undertaken. Donega and Burgess15 have described a low-pressure non-flame cell that utilised atomisation from a tant- alum ribbon but this cell was insensitive to platinum, a figure of 300 ng for 1% sensitivity being found.This paper describes the optimum conditions for the atomic-absorption spectrophotometric determination of palladium, platinum, rhodium, ruthenium, iridium and osmium, using a carbon rod atomiser, and discusses some inter-element effects and their removal and also the scope of the technique for determining the precious metals. Experimental Apparatus A standard Varian Techtron AA-5 atomic-absorption spectrophotometer was operated on minimum damping in conjunction with a Model 63 carbon rod atomiser. Except where mentioned, the carbon rod atomiser was used in the “tube” mode with pyrolytically coated graphite tubes (Ringsdorff-Werke GmbH, Bonn-Bad Godesburg, West Germany).TheEVERETT 349 spectral sources used were Varian Techtron hollow-cathode lamps and read-out was in the absorbance mode, peak height being measured via a Model 3000 fast-response chart recorder (Oxford Instruments, Oxford). An Eppendorf microlitre pipette (5-pl capacity) with disposable tips was used for sample injections. Oxygen-free nitrogen and, when appropriate, argon containing 10% of methane were used as sheathing gases. The flow-rate of the cooling water was 0.8 1 min-1. Reagents and Standard Solutions All reagents used were of AnalaR grade and standard solutions were prepared from Johnson Matthey Specpure materials, namely gold, palladium and platinum sponges, ammonium chloroiridate, ruthenium( 111) chloride and ammonium chlororhodite, dissolved in the minimum volume of a 20% solution of aqua regia. These solutions were then made up so as to give a concentration of 10 g 1-1 of the appropriate metal in a 5% solution of aqua regia.Osmium standard solution was prepared by dissolving 2.01 g of osmium tetroxide in 100 ml of de-ionised water and was then diluted so as to give 1 1 of a solution that was 2 M in sodium hydroxide. This solution contained 1500 pg ml-l of osmium. Working solutions were prepared by diluting these stock standards with de-ionised water, those at concentrations below 100 pg ml-l being prepared daily and those below 1 pg ml-l immediately prior to use. Optimisation of Experimental Parameters The optimisation of experimental parameters was undertaken with a view to maximising both the sensitivity of the method and the signal to noise ratios.However, owing to the low volatility of the analyte metals high temperatures of about 2 500 "C are required for atomisa- tion. Sensitivity is at a maximum at the highest temperatures attainable (about 2 700 "C), but at this level a considerable signal due to carbon particle scatter is observed and some sensitivity must therefore be sacrificed by reduction of the atomisation temperature in order to produce a signal that is free from this non-atomic absorption. However, this reduction in atomisation temperature leads to a greater reproducibility and an increased lifetime of the carbon tubes.Sensitivity is at a maximum at a low monochromator band pass but for a given lamp current the loss in sensitivity with increasing band pass is balanced by a decrease in noise that results from an increase in radiation reaching the detector and this causes the signal to noise ratio to pass through a maximum. In a similar manner the fall in sensitivity caused by an increase in lamp current at a given band pass is also balanced by a decrease in noise, which gives a maximum signal to noise ratio at a certain lamp current. The lamp currents and mono- chromator band passes were optimised so as to give these maximum values. In each instance the sensitivity at the optimum values chosen (Table I) was greater than SOY0 of the maximum attainable. The nitrogen flow-rates gave a maximum sensitivity at the values tabulated and fluctuations of -&l 1 min-l had little effect on the signal, changing it by less than 5%.The ashing parameters given in Table I are the maximum settings that afford no loss of analyte from an aqueous sample and give an absorbance of about 0.4. TABLE I OPTIMUM EXPERIMENTAL CONDITIONS Samples dried a t 120 "C for 15 s. P d Pt Rh Wavelength/nm . . .. .. .. Lamp current/mA . . * . .. .. Monochromator band passlnm . . .. Nitrogen flow-rate/l min-1 . . .. .. Argon - methane flow-rate/l min-1 . . .. Ashing time/s .. .. .. .. Atomisation ternperaturel'c . . .. .. Atomisation time/s . . .. .. .. Maximum ashing temperature tolerable/"C . . 244.8 7 0.33 4.4 1 450 30 2 200 5 - 266.0 8 0.66 6.0 1730 60 2 200 5 - 343.5 10 0.33 4.4 1 450 60 2 200 5 I Ru 349.9 20 0.17 4.4 0.9 1730 60 2 400 5 Ir 208.9 15 0.17 4.4 0.9 1730 60 2 400 5 0 s 290.9 16 0.33 4.0 0.9 1730 60 2 400 7350 Calibration Graphs, Sensitivities and the Origin of the Signal The sensitivity values obtained under the optimum conditions chosen are listed in Table I1 and they compare very well with those obtained by other workers.Adriaenssens and KnooplS obtained 1 yo sensitivities of 1 .O, 0.18 and 2.0 ng for platinum, rhodium and iridium, respectively, while Pearton and Mallettl* compared three non-flame cells and found that the Model 63 CRA in the “tube” mode gave the best results, with a 1% sensitivity for platinum of 0.22 ng. Calibration graphs are reproduced in Fig. 1 and, as can be seen, the working ranges are fairly extensive except for osmium, because of incomplete atomisation of osmium at the higher levels.This effect was monitored by subjecting the carbon tubes to X-ray fluorescence analysis prior to and after atomisation, when it was found that approximately 60% of the osmium remains unatomised. However, with amounts from 10 to 100 ng the calibration graph can be used on a semi-quantitative basis. EVERETT: THE DETERMINATION OF THE PRECIOUS METALS Analyst, VoZ. 101 Results TABLE I1 CALI B RATI o N DATA Detection Repro- Relative Upper working limit ng ng % deviation, yo ng pg ml-l limit/ ducibility/ standard -- 1% Metal sensitivity*/ Palladium 0.01 0.002 5 0.5 2.1 1.5 0.3 Platinum 0.18 0.05 2.5 2.7 25.0 6.0 Rhodium 0.01 0.007 5 1 .o 1.6 1.25 0.25 Ruthenium 0.04 0.2 2.5 1.7 6.0 1 .o Iridium 0.11 0.5 10.0 3.5 25.0 6.0 Osmium 3.4 10.0 - - 100.0 20.0 * The mass corresponding to 1% absorption.An attempt to obtain better detection limits for ruthenium and iridium was made by investigating alternative spectral lines (Table 111), but in each instance a considerable scatter signal was observed, especially at the intense iridium 264.0-nm line. Of these alternative lines those for ruthenium 343.67 nm and iridium 209.26 nm are in fact non-resonance lines but their lower energy levels, 0.15 eV and 0.35 eV, respectively, are sufficiently close to the ground state for atomic absorption to occur. The lower level of the non-resonance iridium 208.57-nm line (0.88 eV) is considerably higher than the ground state and therefore no atomic absorption is observed.This line is useful in correcting for non-atomic absorption at 208.9 nm. 1 .o 0.8 0.6 (1J e 8 a P 0.4 0.2 0 0.1 0.2 0.3 0.4 0.5 Concentration of Pd or Rh/pg ml-’ I I J I I I 0 1.0 2.0 3.0 4.0 5.0 Concentration of Pt, Ru or Ir/pgml-‘ I t I I I 1 0 20 40 60 80 100 Concentration of Odpg mI-1 Fig. 1. Calibration graphs for various elements (5-pl aliquots).May, 1976 BY FLAMELESS ATOMIC-ABSORPTION SPECTROPHOTOMETRY 351 The sensitivity of the method is critically affected by the condition of the graphite tubes. These tubes are pyrolytically coated in order to reduce surface porosity and to enable high atomisation temperatures to be used without the tube becoming pitted and porous. However, the quality of the tube declines with use and this effect can be monitored by the change in sensitivity with successive atomisations. There is an initial sharp fall in sensitivity before a plateau region is reached where the sensitivity and reproducibility remain fairly constant.Thereafter, the sensitivity and reproducibility fall sharply owing to rapid deterioration of the tube. For the more volatile metals (palladium, platinum and rhodium) this plateau persists for about 120 atomisations but, as the volatility of the analyte decreases, this number falls and is about 30 for iridium. This decrease in sensitivity can be partially rectified by introducing into the sheathing gas flow argon containing 10% of methane at the rate of 0.9 1 min-l.ls When the tube is heated to a very high temperature the methane is cracked and a deposit of carbon is left on the tube surface, which has the effect of regenerating the tube and, with iridium, considerably lengthen- ing the plateau region to about 75 atomisations.The origin of the signals was investigated for palladium, platinum and osmium by use of a hydrogen continuum lamp operated a t 20 mA; for rhodium the 349.9-nm ruthenium line was used and for ruthenium, the rhodium 343.5-nm line. Non-atomic absorption at the iridium 208.9-nm line was measured by using either the hydrogen lamp or the iridium 208.57-nm line. In no instance was any absorption observed and it was therefore concluded that all analytical signals were those of atomic absorption. TABLE I11 1 yo SENSITIVITY FOR ALTERNATIVE LINES Ruthenium Iridium Wavelengthlnm 1 yo sensitivitylng Wavelength/nm 1 % sensitivitylng I A \ ----- - 287.6 343.67 372.8 0.9 0.72 0.43 208.67 209.26 264.0 0.0 0.6 0.36 Reproducibilities and Detection Limits The values for reproducibility obtained were very good but the precision was critically dependent on the condition of the graphite tube.Time must be allowed for the tube to cool after each sample cycle otherwise low precision results, because the sample soaks to a small but significant extent into the graphite surface and, as the porosity of the surface affects the analytical signal, this degree of absorption must be constant. Consequently, a steady supply of cooling water is required and a lapse of at least 30 s must occur between atomisation and the injection of the following sample.With no ashing step a sample cycle time of 1 min was found to be adequate, but if various ashing stages were included the sample cycle time had to be appropriately increased. The detection limit obtained was defined as that amount of analyte which gives a mean signal equal to twice the standard deviation of a set of at least ten sample - blank pairs close to the detection limit. Those obtained for palladium, platinum and rhodium were very good but those for ruthenium, iridium and osmium were equivalent to about 5% absorption owing to a scatter signal of about 0.01 absorbance unit, which could not be reduced as it was caused by the high atomisation temperatures that were necessary for complete atomisation of the analytes. With the use of simultaneous background correction it should be possible to improve on the figures for ruthenium, iridium and osmium.Interference Studies The interference of the precious metals on each other was determined at ratios from 1- to 100-fold mass excess of interferent over analyte. Results for the 100-fold excess level at two levels of each analyte are given in Table IV. In general, as the concentration of interferent increases for a given concentration of analyte, the tendency is towards a decrease in analytical signal, the exception being for 0.2 ng of palladium as analyte, with which the signal is constant from 5- to 100-fold excess. This352 EVERETT: THE DETERMINATION OF THE PRECIOUS METALS Analyst, VoZ. 101 TABLE IV PERCENTAGE CHANGE IN ABSORBANCE WITH ~ ~ O - F O L D MASS EXCESS OF INTERFERENT In terferent Amount/ A \ Analyte ng Pd Au Rh Pt Ru I r Palladium 1 .O 0.2 Rhodium 1.0 - 0.2 - 5 0 0 + 20 0 0 +6 + 35 + 40 + 25 + 10 + 30 Platinum 25.0 - 5.0 20 - 0 10 0 35 - 65 0 0 - 35 - 30 - 40 + 10 +5 0 - - 70 - 30 80 0 Ruthenium 5.0 - 30 - 60 - 20 - 30 - 55 1.0 0 - 15 +5 + 20 0 Iridium 25.0 - 50 - 45 - 20 - 25 - 75 5.0 - 10 - 30 0 - 10 - 10 effect is illustrated in Fig.2, where data are given for interference at the lo-, 50- and 100-fold excess levels. It seems that at higher levels ruthenium and iridium give greater signal depressions than the more volatile precious metals, this effect being due mainly to accumulation of interferent in the tube. This accumulation is so intense that when osmium is involved it is extremely difficult to obtain quantitative results and therefore none are included here. The accumulation of the less volatile elements and the resulting interferences can be reduced by using a higher atomisation temperature at the expense of precision and tube lifetime.In the determination of platinum, by increasing the atomisation temperature, the interference due to rhodium and iridium can be reduced by between 25 and 50% (Table V). A mechanism for the cause of interference has been postulated12 to be the formation of alloys between the metals involved, which thus changes the atomisation characteristics of the analyte. I t is therefore to be expected that as the interferent concentration increases or the t I AU Pd Pt Rh Ru Ir I nterferent Analyte Pd,l ng Pt, 25 ng Ru, 5 ng Ir, 25 ng Rh,l ng Fig.2. Percentage change in absorbance with different interferents. 0, 10-fold; x , 50-fold; and 0, 100-fold mass excess of interferent.May, 1976 BY FLAMELESS ATOMIC-ABSORPTION SPECTROPHOTOMETRY 353 TABLE V EFFECT OF ATOMISATION TEMPERATURE ON THE INTERFERENCE OF IRIDIUM AND RHODIUM ON 25 ng OF PLATINUM Results are expressed as a percentage change in absorbance. Excess of interferent Atomisation I A > Interferent temperature/"C 5 O- f old 7O-fold 100-fold Iridium 2 200 2 400 2 700 Rhodium 2 200 2 400 - 55 - 45 - 30 - 50 - 35 - 55 - 50 - 30 - 60 - 45 - 80 - 56 - 40 - 60 - 45 tube ages, becoming more porous, the analytical peaks will broaden owing to slowing of the rate of atomisation; results that support this conclusion are given in Table VI. By visual comparison of peak shapes for other analyte - interferent pairs it can be seen that a similar broadening effect occurs.TABLE VI EFFECT OF INTERFERENT CONCENTRATION ON SIGNAL DURATION Peak width a t half-height measured in seconds. Excess of interferent over 25 ng of platinum A f \ I n terf eren t 0 x l x 5 x 10 x 20 x 50 x 70 x 100 Rhodium 1.06 1.14 1.17 1.35 1.57 2.14 2.23 2.60 Iridium 1.06 1.24 1.26 1.33 1.68 1.74 2.23 3.08 Various methods for the reduction and removal of these interferences were investigated. Simple dry ashing was found to be of no real use, the interferences being reduced in many instances but not removed. Pearton and Mallett14 investigated the effect of releasing agents on the determination of palladium and found that whereas uranium, vanadium and lanthanum depress the palladium signal, copper and cadmium do show some releasing action.Therefore, the use of these two reagents was investigated for the other precious metals. The two solutions used were a 1000 pg ml-l solution of cadmium and a solution containing 0.5% each of copper sulphate and cadmium sulphate, the ashing setting being 1290 "C for 30 s. By using the analyte and interferent levels described in Table IV it was found that both reagents were successful in removing interferences but the results obtained were not acceptable owing to the high irreproducibility entailed, relative standard deviations being about 20% for five replicates. The copper - cadmium mixture also caused a decrease in sensitivity of 3040% for all elements. The use of oxidising agents to remove ruthenium and osmium as their tetroxides was investigated. Perchloric acid (3-20y0) had no effect but acidified sodium bromate solution (0.5% m/V) was found to be successful in removing these two elements, the only exception being the removal of ruthenium in the presence of iridium.However, again there was a pro- blem of a 50% reduction in sensitivity and a low reproducibility (relative standard deviation of about 10%) from five replicates. Ebdonl' proposed that with the carbon filament technique the degree of interference was governed, not by the ratio of analyte to interferent, but by the absolute amount of interferent present. Preliminary experiments seemed to support this view and therefore simple dilution of the sample was investigated as a means of reducing interferences.By examining the results given in Table IV it can be seen that this step was most successful and, with the exception of palladium as analyte, the remaining small interferences could be removed by a further 5-10-fold dilution. It can therefore be concluded that dilution of the sample is the best method of removing precious metal interferences, although there will be a limit to this practice as regards precision and the concentration of the analyte. When high levels of ruthenium and osmium are present354 EVERETT: THE DETERMINATION OF THE PRECIOUS METALS Analyst, Vol. 101 it would be practicable to use acidic bromate solution to remove them, although some degree of sensitivity and precision would need to be sacrificed in order to achieve an interference-free determination.Applications The carbon rod technique has been used in this laboratory to determine precious metals in a number of samples, some of the results obtained being presented in Tables VII, VIII and IX. TABLE VII DETERMINATION OF PRECIOUS METALS I N COPPER AND NICKEL STANDARDS Results are expressed in p.p.m. Sample Pd Pt Rh Euratom copper standards : - 1.19 p.p.m. of palladium and platinum 1.6 f 0.3 1.2 3 0.4 Ir - 11.9 p.p.m. of palladium and platinum 11.9 f 2.1 10.3 1.0 - - Copper sample A . . .. .. . . 3.2, 3.2 15.5, 16.2 0.3, 0.3 - .. - 12.3 f 2.0 - I B .. .. .. c .. .. .. . . 0.21, 0.25 1.2, 1.3 - <1.3, <1.3 Nickel sample A . . .. .. . . 2.9, 2.3 2.3,2.0 - ~ 1 . 3 , <1.3 TABLE VIII COMPARISON OF RESULTS FOR SOLUBLE PLATINUM IN AIR FILTERS Results are expressed in micrograms.Sample r > Method A B C D E Precision A This work .. 3 4 3 4 6 k0.l X-ray fluorescence 10 2 4 <2 <2 A2.0 TABLE IX DETERMINATION OF PLATINUM IN WHOLE BLOOD Sample Platinumlpg per 100 ml 1 A .. 10 12 11 B .. 8 8 10 c .. 21 14 14 D .. 6 7 <6 E .. 21 36 21 Platinum in air to determine soluble or total platinum. when compared with that obtained by the equivalent flame method. solution used was dependent on the size of the air filter. Platinum was leached from air filters with 1 M hydrochloric acid or 50% aqua regia in order The detection limit was improved by a factor of ten The volume of leaching Platinum in human blood rod. per 100 ml for a flame method in which no pre-concentration was used.Blood was diluted 1 +1 with de-ionised water and 1-pl aliquots were added to the carbon The detection limit achieved was 4 pg per 100 ml of whole blood as compared with 100 pg Palladium and filatinum in organic substances The improvement in sensitivity and detection limits by the non-flame method when com- pared with those obtained with a flame was 10-fold for platinum and 40-fold for palladium.May, 1976 BY FLAMELESS ATOMIC-ABSORPTION SPECTROPHOTOMETRY 355 In the methods used the sample was simply dissolved in an appropriate solvent and applied directly to the carbon rod. Precious metals in copper and nickel Following dissolution in aqua regia the precious metals were pre-concentrated by ion exchange and the resulting solution was applied to the carbon rod.The detection limits compared with those by the equivalent flame procedure were improved by a factor of four for iridium and rhodium, of ten for platinum and of 40 for palladium. Ruthenium was not determined. For most of these methods non-flame atomic absorption was used as the instrumental finish and therefore enhancement factors were due only to the inherent sensitivity of the technique resulting from higher atomisation efficiencies. However, for samples such as human blood, a flame can accept only dilute samples and therefore the enhancement afforded by non- flame atomisation is further increased. With air filters, blood and organic substances the determinations were free of interference from the matrix, but for the copper and nickel samples, although there was no interference from these metals in the final solution, it was necessary to run standards in order to eliminate errors caused by incomplete ion exchange.Conclusion The carbon rod method provides a sensitive and precise method of determining the precious metals down to below parts per million levels. The sensitivity is such that the upper working limit of the method is approximately the same as the lower limit of the corresponding flame procedure. It thus provides a most useful extension of the atomic-absorption technique. Osmium, however, owing to its very low volatility, is the exception and in all but a few isolated instances the flame procedure will be preferred. Although mutual interference between the precious metals is, in some instances, high, they can mostly be removed by simple dilution of the sample and if this device is not possible the standard additions procedure can be used without too great a loss in sensitivity.Also, integration of the analytical signal may prove a better measurement technique as it should compensate for the variation in the rate of evaporation of the analyte. The author thanks the Board of Directors of Johnson Matthey Chemicals Limited for permission to publish this paper, and his colleagues at Royston for their help in analysing some of the copper, nickel and air samples. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 16. 16. 17. Strasheim, A,, and Wessels, G. J., Appl. Spectrosc., 1963, 17, 65. Erinc, G., and Magee, R. J., Analytica Chim. Acta, 1964, 31, 197. Schnepfe, M. M., and Grimaldi, F. S . , Talanta, 1969, 16, 591. Pitts, A. E., van Loon, J. C., and Beamish, F. E., Analytica Chim. Acta, 1970, 50, 181. Mallett, R. C., Pearton, D. C. G., Ring, E. J., and Steele, T. W., Talanta, 1972, 19, 181. Sen Gupta, J. C., Mineral Sci. Engng, 1973, 5, 207. Scarborough, J. M., Analyt. Chem., 1969, 41, 250. van Loon, J. C., 2. Analyt. Chem., 1969, 246, 122. Atwell, M. G., and Hebert, J. Y . , Appl. Spectrosc., 1969, 23, 480. Pitts, A. E., van Loon, J. C., and Beamish, F. E., Analytica Chim. Acta, 1970, 50, 195. L’vov, B. V., Spectrochim. Acta, 1969, 24%, 53. Guerin, B. D., J l S . Afr. Chem. Inst., 1972, 25, 230. Adriaenssens, E., and Knoop, P., Analytica Chim. Acta, 1974, 68, 37. Pearton, D. C. G., and Mallett, R. C., National Institute of Metallurgy, Johannesburg, 1974, Report Donega, H. E., and Burgess, T. E., Analyt. Chem., 1970, 42, 1521. Morrow, R. W., and McElhaney, R. J., Atom. Absorption Newsl., 1974, 13, 45. Ebdon, L. C., Ph.D. Thesis, University of London, 1971. No. 1598. Received September 19th. 1975 Accepted November 28th, 1975

ISSN:0003-2654    

DOI:10.1039/AN9760100348

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

356 Analyst, May, 1976, Vol. 101, pp. 356-366 The Development of Remote Spectrographic Heads for Metallurgical Analysis and their Application to Product Inspection Analysis A. D. Ambrose and J. D. Hobson Research Centre, British Steel Corporation, Tubes Division, Corby, Northamptonshire, "17 1 UA Dunford Hadjields Ltd, East Hecla Works, Shefield, S9 1TZ Direct-reading optical-emission spectroscopy has been widely adopted in the iron and steel industry as a rapid analytical technique for process control and for other laboratory analyses. The conventional spark stand, however, severely restricts the size and shape of the sample that can be accommodated on the instrument and thus limits the more general use of the technique for on-site analysis outside the laboratory. This disadvantage has now been overcome by using a fibre-optic light-guide to link a mobile excitation head to a conventional spectrometer.Parallel but independent work has been carried out in the authors' labora- tories, first on photographic and later on direct-reading spectrometers. At the Corby laboratories, further work has resulted in the manufacture and application of an inspection analyser that incorporates a 5-m guide. Typical calibration graphs, reproducibilities and examples of material identification are given, and possible applications are discussed. Steel is produced to many different compositional specifications according to its intended usage. During recent years there have been outstanding advances in the analytical procedures used for process control, thus making possible the more economic production of ever higher tonnages of steel of assured composition.In the interval between primary production and the finished product and again before the latter reaches the customer, however, there may be many intermediate stages of manipulation, e.g., forming, sub-division and transportation, at any of which normal identification marks may be obliterated. Although great precautions are taken, loss of identity, with the attendant risk of mix-ups, is a very serious matter and has been of concern to quality control engineers for many years. This problem has led to a continuing quest for rapid and reliable methods for the identification of material on-site.' Generally, purely chemical methods of identification are not well suited to large-scale production, although local spot-testing, for example with dimethylglyoxime for nickel, is sometimes useful.Physical methods, based, for example, on the magnetic or electrical properties of the material, have been more widely applied but, unfortunately, they indicate chemical composition only indirectly and are too easily affected by other metallurgical factors. Other miscellaneous methods include spark testing, and the use of portable isotope analysers and direct-vision spectroscopes. Although each may be helpful in specific circumstances, none meets entirely the requirements of quality control inspection. Equipment that does have the potential for this kind of work is that based on measurement of spectral radiation, either optical or X-ray.Traditionally, however, such equipment, while of excellent use in the laboratory and highly developed in this context, has been unsuitable for on-site quality control inspection, as it is too bulky, fragile and easily upset by vibration or temperature variations. Recent develop- ments by the manufacturers have now enabled single-console, robust optical spectrometers to be provided, which are suitable for use in the plant with the minimum of protection. The need to cut and prepare a suitable size of sample limits the scope of the technique. On the other hand, the handling difficulties in locating massive components quickly and accurately on the spectro- meter are obviously formidable. The novel solution to this problem that was conceived independently at the Research Centre of the British Steel Corporation, Tubes Division, Corby, and in the laboratories of Dunford Hadfields Ltd, Sheffield, embodies separation of the electrical discharge into a portable test-probe, which is independent of the main spectrometer but coupled to it by a light-guide several metres long if necessary.Clearly, the outstanding problem is one of sample presentation.AMBROSE AND HOBSON 357 Another spectroscopic approach to inspection analysis involves atomisation of the sample in an arc and transportation of the vapour in a stream of argon through a flexible pipe to a plasma arc coupled to a spectrometer.2 Conveyance of light, however, appeared to have certain attractions and governed the lines of the present investigation. The present paper describes the laboratory work carried out on this concept and then proceeds to describe a production version of the apparatus in operation at Corby since the beginning of 1974.Possibilities for further development are also indicated. Experimental Work at Corby Research Centre Preliminary Trials Using a Photographic Spectrograph In order to determine the feasibility of the remote sampling concept and generally to gain experience in the use of fibre optics, preliminary experiments were conducted by using a Hilger large-dispersion Littrow spectrograph and a glass light-guide that was approximately 2 m long and 3mm in diameter. The only optical modification was the removal of the condenser lens that normally stands before the spectrograph slit, and the insertion, im- mediately before the slit, of a short-focus lens (microscope eye-piece), which was adjusted until the spectra were in focus.In the first experiments the original 15 000-V uncontrolled ax. condensed-spark source supplied with the instrument was employed and many spectra were successfully recorded. In the expectation of possible plant usage, however, it was envisaged that, initially, the remote sampling head would be hand-held, and in view of the elaborate safety precautions that would be required for high-voltage operation, it was decided to replace this source by the power unit from a Metascop direct-vision portable spectroscope. This is a compact unit that is well suited to being built into a portable test head and, when operating at a relatively low potential, presents only a slight safety problem.In operation, a locating pin is in contact with the sample and the body of the unit is tilted until the discharge electrode just makes contact. The current flow energises a coil, which lifts the electrode from the sample against the action of a spring, thus simultaneously drawing the arc and breaking the circuit. With the coil now de-energised, the electrode returns to the sample and the cycle is repeated. Typical calibration graphs for manganese and chromium are shown in Fig. 1, in which arbitrarily selected lines in the 400.0-nm region were used. At this early stage in the feasibi- lity trial, time was not spent in determining exact wavelengths. Alternatively, if these graphs are assumed to represent correctly located calibrations and the points are read from them as for analytical samples, results typified by those shown in Table I are obtained. 7.31 I 1 1 1 1 I I I 0.1 0.2 0.3 0.5 1.0 2.0 3, Concentration of element, % Fig. 1.Determination of chromium and manganese. Preliminary experi- ments with photographic spectro- graph.358 AMBROSE AND HOBSON: THE DEVELOPMENT OF REMOTE Analyst, VoZ. 101 TABLE I COMPARISON OF ANALYSES AND CHEMICAL RESULTS Photographic spectrograph ; Metascop source ; 60-s exposure. Chromium, % Manganese, % -------7 r- Sample Analyser method Chemical method Analyser method Chemical method S.S. 401 0.18 402 0.51 403 0.42 404 0.68 405 0.21 406 2.21 0.18 0.55 0.42 0.68 0.2 1 2.12 0.97 0.16 1.70 0.46 1.31 0.53 1.0 0.19 1.69 0.52 1.28 0.53 These results were regarded as being highly encouraging, although it must be emphasised that they were obtained under controlled laboratory conditions with clean prepared samples and by using a relatively long exposure time of 60 s.Other information to emerge from these photographic experiments was as follows : (a) The useful wavelength range transmitted by the glass light-guide does not extend below about 380.0 nm, thus precluding the determination of carbon, sulphur and phosphorus. Above this wavelength range, however, spectral lines are listed for most of the alloying elements. (b) From the manufacturer’s data, a loss of light intensity to approximately 30% of the original value (70% loss) might be expected along the 2-m glass guide. In practice, inserting the guide between the source and the slit gave a slight gain in intensity, presumably because of the greater solid angle subtended.(c) Comparisons were made with the guide straight, looped and coiled on itself several times. It was found that provided the coil diameter was not less than about 25 cm, no deleterious effect on the spectral lines could be detected. Trials with a Direct-reading Spectrometer The next important stage in the laboratory experiments was the transference of the system to a direct-reading spectrometer, for which purpose Applied Research Laboratories, Luton, kindly lent us a sequential single-channel instrument (the Quantoscan) so as to enable trials to be continued with electronic recording and in regions of the spectrum that were beyond the range of the photographic spectrograph.As before, a short-focus lens was fitted between the end of the light-guide and the spectrometer slit, and adjusted for maximum intensity. The first trials with the Metascop source, which had given good results on the photographic spectrograph, however, proved to be unsatisfactory. With a smooth-running discharge, the light intensity was too small to record in a reasonable length of time. Only by increasing the gap almost to cut-off point was there sufficient sensitivity and then the arc was so erratic that multiple readings were necessary. Accordingly, attention had to be given to an alternative form of discharge. In emission spectroscopy, two extreme forms of discharge are recognised. One is a high- voltage, low-current spark that has a relatively low light output but gives good quantitative results.At the other end of the scale is the d.c. arc (commonly 25-50 V on closed circuit and a current up to 10 A), which gives a high light output but is less accurate and subject to wander. Between these extremes, various interrupted arc sources have been designed so as to give optimum performance in any given situation. With suitable engineering, no doubt any of these sources could be used in a remote sampling head but, bearing in mind the safety aspects mentioned in the previous section and in order to avoid the need for lengthy development, attention was concentrated on the possibilities of the d.c. arc, and it was found that with a simple source operating at 35 V and 2-2.6 A closed circuit (1 10 V open circuit), there was sufficient light intensity for a guide several metres long to be used if necessary.A circuit diagram is shown in Fig. 2. Spectroscopically speaking, the d.c. arc is a crude source and it was realised that the accuracy achieved might not be high but it was considered that provided results were reasonable, refinements could be introduced. For the portable test head, a copper electrodeMay, 19 76 SPECTROGRAPHIC HEADS FOR METALLURGICAL ANALYSIS 359 Fig. 2. The d.c. arc source. was fastened into a simple press-down holder as illustrated in Fig. 3. In practice, the test head is applied to the sample. The electrode, which is gripped in an inner sliding tube, is then pressed down against the action of the spring so as to touch the sample and strike the electric arc.When the pressure is released, the electrode springs back a controlled distance and the arc continues to burn. This device is the subject of a patent application. Insulating cap Ink Sample (-ve) Fig. 3. First experimental test head in laboratory trials. By using this relatively crude source, spectral lines were located in the visible region for the following elements of interest: manganese, 476.24; chromium, 520.61 (425.44) ; molyb- denum, 386.41 ; nickel, 471.44; niobium, 405.89; and vanadium, 437.92 nm. Aluminium (396.15nm) also falls within the transmission range of glass and could be added to the programme if required. The nickel line given above is rather insensitive and is not recom- mended for alloying concentrations below about 1%.The chromium line at 620.61 nm is less strongly self-reversing than that at 425.44 nm and is preferred, but it may be outside the range of some spectrometers. As expected, analytical results using this relatively crude source were not as good as had previously been obtained on the photographic instrument, partly because of the inherent variability of the arc and partly because of the much shorter integration times that are now being used. Nevertheless, they were suitable for the identification of a number of important steel specifications, without risk of confusion between them. It should perhaps be emphasised that analytical accuracy, although always desirable, is not necessarily the major consideration in the identification of material on site. Speed of testing assumes a high priority, even at the expense of accuracy, provided, of course, that there is no risk of confusion between the individual specifications.Process control of the primary steelmaking ensures that the concentrations within a given specification are correct. With this aspect in mind, and in order to relate the work still more closely to plant practice, a number of tube samples were then analysed directly on the surface without surface prepara- tion of any kind. With chromium as a test case, it proved to be possible to separate the360 AMBROSE AND HOBSON: THE DEVELOPMENT OF REMOTE ATZ~I?J&~, VOl. 101 principal steel grades of interest (residual, 0.5, 1, 2, 4%, etc. of chromium) unambiguously in testing times of under 10 s, of which 5 s were required for pre-burn.More details of the analytical results are given under the inspection instrument at the Corby works, where the production version of the instrument is described. Experimental Work at Dunford Hadfields Preliminary Investigations In early 1969, the analytical laboratories were receiving fairly frequent requests for analyses to be carried out on rather large pieces of steel in the course of works investigations, and were using a number of the techniques discussed earlier, but at considerable expense and inconvenience. The purchase of a demonstration fibre-optic light-guide kit for general evaluation in the physics department led almost at once to the idea of bringing the light from a discharge, rather than an unwieldy specimen itself, to the spectrometer.Preliminary tests with a 1-m glass light-guide of only 1.5 mm diameter coupled to a 3-m quartz photographic spectrograph at once showed that typical emission spectra of steel could be recorded on Kodak IIL panchromatic plates. It was clear that the idea was sound and patent applications3s4 were filed. The patents envisaged that a widened scope could be achieved for spectrometry outside the laboratory by using an excitation head capable of being used under workshop conditions, for example, as part of a routine inspection on a production line, with the head coupled via a flexible light-guide to an automatic spectrometer housed nearby in a small enclosed space capable of being maintained under controlled condi- tions akin to those of a normal spectrographic laboratory.In subsequent experiments carried out in order to evaluate the idea, the gradual introduction of more sophisticated equipment has shown the validity of the original idea, and exploitation of the patent is now the subject of a licence agreement with Applied Research Laboratories. The first step was to produce some actual analyses by use of the Hilger and Watts 3-m Quartz Littrow spectrograph that had previously been calibrated for steel analysis. However, all of the spectrographic lines used in normal photographic or direct-reading procedures for steel were in the ultraviolet region. Examination of the spectrum of light from a mercury lamp transmitted through the fibre optic gave its transmission range and showed that the use of the conventional Kodak B10 plate was satisfactory.However, constant exposure times gave spectra of decreasing density, although the intensity recovered after a period of inactivity. The fading effect was found to be due to the concentration of heat on the end of the fibre optic by the lens used to gather light from the source discharge, and it was eliminated by introducing a pair of heat filters (Evans Electroselenium, Type HA1). These also removed longer wavelengths and the transmission band of the optical path was reduced to approxi- mately 380.0 to 440.0 nm. Provided that allowance was made for variations in the gamma characteristics of the plate with wavelength, as determined by step exposures on the spectrum of a mercury-discharge lamp, satisfactory calibration graphs could be produced for the line pairs manganese 403.1 - iron 400.7 nm, chromium 425.4 - iron 426.0 nm and molybdenum 386.4 nm with an unidentified iron line of suitable sensitivity.However, the vanadium line at 437.9 nm was unsatisfactory because of an inter-element interference from nickel, and it was shown that 1% of nickel was recorded as being equivalent to 0.1% of vanadium. Experimental Work with Silica Optics Use of a silica optic guide was desirable in order that transmission in the ultraviolet region should match that of the lenses and prisms of the spectrograph, allowing the use of normal working techniques. In the early stages, silica light-guides were not commercially available, so experimental light guides were made, first from home-made hand-drawn fibres, and later from Thermal Syndicate Spectrosil fibres of 40 pm diameter.The home-made optic guides showed satisfactory transmissions in the ultraviolet region but they were internally fragile and gradually lost their power of transmission. Eventually, a commercially manufactured guide that was 1 m long and 3 mm in diameter became available (from Jena Glaswerk Schott, Mainz) and this was used in conjunction with a water-filled fused silica cell of 10-mm path length, substituted for the glass filters because the latter did not transmit in the ultraviolet region, for heat absorption.May, 1976 SPECTROGRAPHIC HEADS FOR METALLURGICAL ANALYSIS 361 Adequate light for photographic recording was available; the silica guide had a quoted transmission of 50% at wavelengths above 250.0 nm, but falling to about 20% at 200.0 nm.Fortunately, this loss was more than compensated for by the improved aperture of the guide and the quartz lenses used for light collection. Satisfactory calibrations for steel analysis were obtained for silicon, manganese, chromium, nickel and molybdenum. As an example, the calibration for chromium using the two-line pair method is shown in Fig. 4. Chromium, % Fig. 4. Calibration of 3-m quartz photographic spectrograph using 1-m silica fibre-optic light-guide. Chromium 282.2371-nm line. Iron lines: A, 282.8634 nm; B, 282.7434 nm; C, 281.9294 nm. An early success for the technique was gained with the large forging shown in Fig. 5 (see Plate). Analysis showed that steel of incorrect composition had been used, thus explain- ing metallurgical problems; a parallel analysis on a piece of steel of correct type gave results that were in close agreement with its composition determined conventionally (Table 11) , which confirmed the reliability of the results for the unidentified material.TABLE I1 ANALYSIS OF A LARGE FORGING USING THE FIBRE-OPTIC LIGHT-GUIDE Standard test piece N.5659 Si Mn Cr Ni MO Known composition, % . . 0.20 0.29 1.94 2.02 0.61 Result by fibre optic, yo . . 0.18 0.27 1.99 1.98 0.64 Result by fibre optic, % . . 0.25 0.47 1.33 4.70 0.23 Stem forging Experiments with a Direct-reading Instrument The next development was to couple the silica guide to an air-path Applied Research Laboratories direct-reading spectrometer, which was built in 1954 but was still operational.At first, in order to avoid major modifications, a pair of glass prisms was also interposed into the optical path. This arrangement limited the choice to lines in the visible region, but a good calibration graph was produced for the vanadium line 437.923 8 nm in low-alloy steels (Fig. 6). The standard deviation for ten consecutive results at 0.50% of vanadium was 0.011%. This finding justified further work and the spectrometer was moved to another laboratory where it could be linked with a Hilger and Watts FS 139 source unit. A manually operated digital measurement unit replaced the original chart-recorder system, The external optics were rearranged so as to eliminate the use of glass, the silica fibre and quartz lenses being used to bring the light to the entrance slit. Experiments were conducted in the use of reflecting362 AMBROSE AND HOBSON : THE DEVELOPMENT OF REMOTE Andyst, VoZ.101 I 0.1 0.2 0.3 0.4 0.5 0.6 Vanadium, % 7 Fig. 6. Calibration graph for vanadium with air-path direct-reading ultraviolet spectrometer incorporating silica fibre- optic and glass components. Vanadium 437.9238-nm line. optics,6 which have advantages over lenses in providing wide aperture and improved coupling to fibre-optic guides, and in eliminating problems of absorption. Fig. 7 shows typical ways in which reflecting systems can replace lenses. Excitation in air gave unsatisfactory reproducibility, but the use of a conventional Petrey table and a triggered discharge in argon (for which the source unit had been designed) gave results that were much more reproducible.By using this hybrid instrument, calibrations were obtained for many of the elements for which lines had been selected when the spectro- meter was built as an air-path instrument, including silicon 251.612 3, manganese 293.306 3, chromium 267.715 9 (Fig. 8), nickel 231.603 7, molybdenum 281.615 4, copper 327.296 2 and the previously mentioned vanadium 437.923 8 nm, with the iron line at 271.441 2 nm as internal standard. Aluminium at 396.15 nm gave results that were too scattered to be of use. Subsequent work has been directed at developing a mobile head; experiments with a rough prototype have been used to aid in the design of a correctly engineered version. The prototype showed the necessity for a good seal in order to exclude air from the argon atmosphere. It also showed the importance of keeping the end of the optic guide free from sputtered steel, which caused gradual loss of light transmission.It is hoped that these problems will have been overcome in the unit now being built. Applications of the Hybrid Instrument Considerable use is made of spark-testing at Dunford Hadfields as a routine test, but this method of identification of a steel type requires very considerable skill and practice. More- Excitation Spectrometer Fig. 7. Use of mirror optical systems with light guide in ultraviolet region. 1 = Radiation source ; 2 = metal; 3 = primary mirror ; 4 = secondary mirror ; 5 = transparent screen ; 6 = fibre-optic light-guide ; 7 = primary mirror; 8 = secondary mirror; 9 = beam; and 10 = spectrometer slit.May, 1976 SPECTROGRAPHIC HEADS FOR METALLURGICAL ANALYSIS 363 Fig.8. Calibration graph for air-path direct-read- ing ultraviolet spectrometer with excitation in argon and l-m silica fibre-optic light-guide. Chromium 267.7169-nm line. over, when two similar steels or two heats of similar nominal composition are involved it cannot be applied. When such an instance arose, involving 101 bars, it was decided to test the capabilities of the fibre-optic experimental apparatus at its present stage of develop- ment. The steels had the nominal compositions shown in Table 111, and vanadium was the obvious choice of element for rapid sorting. COMPOSITION Steel A .. B .. TABLE I11 OF STEELS (yo) I N MIXED BATCH OF BARS, INSEPARABLE BY CONVENTIONAL SPAR K-TEST1 NG C Si S P Mn Cr Ni Mo -4 1 v 0.39 0.28 0.042 0.017 0.65 1.00 0.26 0.11 0.005 0.02 0.45 0.28 0.022 0.022 0.60 1.02 0.20 0.11 0.006 0.15 Linishing the specimens took 16.5 min, the 101 unknown bars were analysed for vanadium in only 59 min, and a further 14 analyses on standard steels and checking of results took 16 min.Thus 115 analyses required 91.5 min. The samples were given a 7.5-s pre-spark and a 7.55 exposure in accordance with the calibration graphs mentioned previously, and the separation into two groups was so obvious as to render calculation unnecessary. Subse- quent examination of the integrated counts from vanadium and iron, using the usual ratio method, produced the statistics given in Table IV. TABLE IV REPRODUCIBILITY OF VANADIUM ANALYSES OBTAINED DURING SORTING OF A BATCH OF MIXED BARS, USING FIBRE-OPTIC SPECTROMETRY Vanadium, % -----7 Steel A, 36 bars .. . . .. .. 0.01, 0.00, Steel B, 65 bars . . .. .. .. 0.15, 0.01, Identification of bars from mixed batch Mean value Standard deviation Other tests have established that a wide variety of low-alloy steels can be successfully analysed. As an example, Table V gives the results obtained from 15 successive analyses of one piece of a bar of EN 353 quality, and for one analysis from each of ten bars from a single cast of EN 8 steel.364 AMBROSE AND HOBSON: THE DEVELOPMENT OF REMOTE Analyst, VoZ. 101 TABLE V REPRODUCIBILITY OF SPECTROMETRIC ANALYSES FROM EXPERIMENTAL EXCITATION HEAD LINKED BY SILICA FIBRE OPTIC Si Mn Cr Ni Mo c u EN 353 steel Mean (1 bar, 15 exposures), yo .. 0.23 0.95 1.02 1.34 0.11 0.22 Standard deviation, % . . . . 0.01, 0.01, 0.01, 0.00, 0.00, 0.OO6 EN 8 steel Mean (10 bars from one cast), % . . 0.23 0.85 0.15 0.16 0.03 0.16 Standard deviation, yo . . . . 0.04, 0.02, 0.00, 0.00, 0.00, 0.01, The Inspection Instrument at the Corby Works The success of the experiments described in the Corby trials with a direct-reading instrument gave convincing proof of the feasibility of inspection analysis within the plant, using optical spectroscopy and a remote sampling head, and led the Corby Research Centre to have dis- cussions with Applied Research Laboratories about the construction of a production version of the instrument for tube inspection at Corby. This was based on a modified version of the Quantometer 28, a compact, single-console instrument with internal temperature control that is sufficiently robust to operate under plant conditions with the minimum of protection.The conventional spark stand is replaced by a flexible metal sheath about 5 m long, carrying the light guide, the electrical cables and water-cooling pipes and terminating at the portable test head. This head, a robust version developed from the experimental model, is illustrated in Fig. 9 (see Plate). Below the main block, two adjustable rails serve to accommodate tubes of different diameters. The vertical cylinder, which is of insulating material, contains the electrode-locking mechanism. A quarter turn of this cylinder enables the electrode to be gripped or released at will.After setting the initial gap, it is simply necessary, on transferring the head from one tube to another, to depress the cylinder until the electrode makes contact with the sample and then to release it again so as to draw the arc. This test head is also subject to a patent application. The apparatus can be operated in two modes: manual and semi-automatic. Manual Operation : Graphical Evaluation In the manual mode, a meter reading is obtained for each element in turn and plotted against concentrations in the usual way. Typical calibration graphs are shown in Fig. 10 (a, b and c). Niobium suffers interference from manganese but this effect is automatically allowed for by signal subtraction within the instrument. Rather surprisingly, it was found that manganese formed two graphs, the low-alloy steels forming a separate line from the plain carbon steels.This is probably a function of the spectral line used and the excitation, and is obviously undesirable, Nevertheless, it does not invalidate the use of the instrument in practice, given familiarity with the specifications being sought. Typical results for the instrument used in the manual mode are shown in Table VI, and results of reproducibility tests on tube samples with unprepared surfaces in Table VII. The coefficients of variation can be seen to be of the order of 10%. TABLE VI ANALYSIS OF TUBE SAMPLES (UNPREPARED SURFACES) Manganese, yo +- A L 0.73 0.72 0.52 0.53 1.6 1.50 0.62 0.64 0.57 0.58 0.44 0.43 Chromium, yo m 0.35 0.33 2.5 2.33 < 0.1 0.04 0.43 0.40 2.4 2.23 1.0 0.96 Molybdenum, % Vanadium, % +- * A L A L 0.55 0.60 0.29 0.25 0.88 0.93 - - 0.26 0.26 - - 0.52 0.62 0.22 0.22 1 .o 0.94 - - 0.55 0.50 - - A, analyser; d.c.arc, 5-s pre-burn, 8-s integration. L, laboratory analysis of parent metal.Fig. 5. Analysis of a large forging using a silica fibre-optic coupling between the excitation source and a photographic spectrograph. [To face p . 364Fig; 9. Production version of portable test head. Fig. 11. General view of the inspection analyser in operation. [To f a 9 p. 355May, 1976 SPECTROGRAPHIC HEADS FOR METALLURGICAL ANALYSIS TABLE VII REPRODUCIBILITY OF RESULTS (TUBE SAMPLES, UNPREPARED SURFACES) Element Mean, % Standard deviation, yo Coefficient of variation, 365 Manganese . . . . 0.69 0.052 Chromium .. . . 0.46 0.053 Molybdenum . . . . 0.60 0.059 Vanadium . . . . 0.23 0.017 7.5 11.5 9.5 7.4 Automatic Operation: Alloy Selector More usually, it is used in a semi-automatic mode in which the meter readings for each element are automatically compared with pre-set values that correspond to the steel specifications being sought. If thesevaluesagree, agreenlight shows on the console and is repeated in the test head. If the readings do not agree, red lights are illuminated and an audible signal sounds. If desired, the readings for the individual elements can then be examined manually in order to decide on which element the discrepancy has occurred. While it is not proposed to give a detailed description of the instrument design, it may be said briefly that, in principle, the instrument makes provision for the selection of 60 different specifications by means of alloy selector switches on the front of the console.On turning the switch from one position to another, permanently wired circuits come into operation so as to provide an acceptance “gate” for each element corresponding to a segment of the calibration graph, appropriate to the specification being sought. Since its installation at the beginning of 1974, many thousands of tubes have been successfully tested. With added experience, the measurement time normally employed for automatic inspection comprises a pre-burn period of 3 4 s and an integration time of approximately 4 s. Allowing for occasional changing of electrodes, wiping the glass window that protects the end of the light-guide and transference of the test head from one tube to another, testing time averages 12 s per tube over a large batch of determinations.If other ancillary operations, such as untying the roped bundles, laying out the tubes on the inspection table and affixing identifica- tion tapes on the samples after test, are also taken into account, throughputs of over 200 tubes per hour can be achieved. Fig. 11 (see Plate) shows a general view of the equipment in a tube-making mill at Corby. Such equipment is now being manufactured for retail sale by Applied Research Laboratories Ltd. under agreement with the British Steel Corporation. In practical inspection analysis, the instrument is seldom used in the manual mode. 0 1 2 3 Chromium, % 1 1 1 1 1 1 1 1 Molybdenum, % 0 0.2 0.4 0.6 0.8 1.0.1.2 1.4 Niobium, % Fig. 10. Typical calibration graphs for (a) chromium (425.44 nm), (b) molybdenum (386.41 nm) and D.c. arc. (c) niobium (405.89 nm) (corrected for manganese). Discussion and Further Developments The success of these various experiments offers considerable promise for applications in This optimism is borne out by the successful application of the metallurgy and engineering.3 66 AMBROSE AND HOBSON production analyser at Corby. This particular apparatus was designed to meet a specific set of circumstances, which it appears to be doing very well. Taking the work of the two labora- tories together, it is obvious that variations and further developments are possible. Alterna- tive source units, for example, can be used to improve analytical accuracy, especially in a fully automated system in which the probe is not handled manually.When necessary, a form of specimen preparation could be introduced in order to deal with more heavily scaled materials, and instruments can be designed with a conventional spark stand and a remote sampling head leading to the same spectrometer. Light-guides made of silica can be used to extend the wavelength range into the ultraviolet region, thus allowing the determination of most of the elements commonly required in steel analysis. The conventional manganese line at 293.3 nm used in the Dunford Hadfields instrument does not suffer from the need for separate calibrations for plain carbon and low- alloy steels mentioned earlier.So far the determination of carbon has not proved possible with the guide materials available, but is not necessarily permanently ruled out. The use of a mini-computer for processing the data into individual specifications should provide more flexibility than the gating system described. Clearly there are two distinct roles for very rapid analysis obtained with this type of equip- ment. One application is for routine inspection for quality assurance, with the instrument as part of a production line, for example in billet or bar-finishing departments. With dedica- ted instrumentation there may well be a need for several such units in various departments of the works. The other role is in the operation of a mobile metallurgical analytical service made available to several works within a group. In this instance the instrument would be housed in a suitably sprung and insulated vehicle, which could be taken to the site of any metallurgical problem and plugged into a local power supply, or supplied from its own gener- ator. This arrangement would enable large masses of steel to be analysed non-destructively and on site, for example when required by a customer’s quality check. It could be used in the stockyards on ingots or blooms of uncertain identity, for example during stocktaking, or for the analysis of scrap, or of materials returned for re-processing, for example, large rolls returned for re-hardening after use. There is little doubt that this development in optical spectroscopy, embodying remote sampling and flexible coupling between source and spectrometer, frees the technique from traditional restrictions in sample presentation, should open up many new analytical possibili- ties and seems certain to prove an invaluable weapon in the annoury of makers and users of many kinds of metals and alloys. The authors are grateful to Mr. J. W. Shaw, Technical Director, British Steel Corporation, Tubes Division, and Mr. G. F. Smith, Technical Director, Dunford Hadfields Ltd, for permis- sion to publish this paper. They also thank their colleagues who have collaboratedin this work, in particular acknowledging the contributions of Mr. D. W. Swingler (BSC Corby) and Mr. T. W. Lomas (DH). References 1. Ambrose, A. D., Croall, G., Henrys, F., and McGavin, R. F., in “Proceedings of the Conference 011 the Determination of Chemical Composition,” Brighton, September, 1970, Iron and Steel Institute, London, pp. 153-163. Hoyt, R. E., Dryer, H. T., and Runkler, L. D., Pittsb. Conf. Analyt. Chem. AppZ. Spectrosc., Cleveland, Ohio, 1975. British Patent No. 1 335 657, 1973. US. Patent No. 3 909 133, 1975. Res. Disclosure, August, 1974, No. 124, p. 12, 12419. 2. 3. 4. 5 . Received October 15th, 1976 Accepted November 1 lth, 1975

ISSN:0003-2654    

DOI:10.1039/AN9760100356

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

Analyst, May, 1976, Vol. 101, @. 367-378 367 Detection and Determination of Polynuclear Aromatic Hydrocarbons by Luminescence Spectro- metry Utilising the Shpol'skii Effect at 77 K Part 11." Detection Systems An Evaluation of Excitation Sources, Sample Cells and 6. S. Causey, G. F. Kirkbright and C. G. de Lima? Department of Chemistry, Imperial College of Science and Technology, London, S W7 2A Y h comparison has been made of the use of a xenon arc continuum source, a mercury-vapour lamp and a fixed-wavelength helium - cadmium laser source for excitation in the production of quasi-linear luminescence emission spectra of some polynuclear aromatic hydrocarbons in n-alkane solvents a t 77 K. The performance of a new design of cryostat sample cell for work a t 77 K has been evaluated and compared with that of a commercially available Dewar- flask sample cell.The use of d.c. integration, photon counting and of repetitive optical scanning in conjunction with a signal averager has been investigated for registration of the luminescence obtained from PAH compounds by use of the Shpol'skii effect. Although numerous papers have described the application of quasi-linear luminescence spectro- metry at 77 K to the determination of polynuclear aromatic hydrocarbons (PAH), in most instances the apparatus employed has incorporated a mercury-vapour discharge lamp1-* as source and a glass or quartz D e ~ a r ~ - ~ tube as the sample cell. In continuation of our previous work,1° which was concerned with the application of low-temperature luminescence spectro- metry utilising the Shpol'skii effect at 77 K, we have undertaken an evaluation of the use of different sources of excitation, two different types of sample cell and a number of techniques of signal registration in order to develop the optimum instrument assembly and to exploit the inherent high sensitivity and selectivity of which the technique is capable.This paper therefore describes a comparison of the use of a xenon arc, a mercury-vapour discharge lamp and a helium - cadmium laser as excitation sources. The use of a commercially available Dewar-flask sample cell and the performance of a new design of cryostat sample cell for work at 77 K is described. The application of d.c. integration techniques, photon counting and repetitive optical scanning with signal averaging using a multi-channel time averager for registration of low-temperature luminescence signals is reported.Experimental The basic assembly of the spectrometer was similar to that described previouslylo; a 1-m grating monochromator (Rank Hilger Ltd., Monospek 1000) with an aperture of f / S and a reciprocal linear dispersion of 0.8 nm mm-1 at the exit slit was employed in conjunction with an EM1 62565 or 9601B photomultiplier for most of the work. In the study of d.c. integration for registration of luminescence, a modified Quantoscan monochromator (Applied Research Laboratories Ltd., Luton, Bedfordshire) was used. The electronics associated with this monochromator made it possible to use both constant-time and constant-charge modes. For the constant-charge mode a second (reference) photomultiplier was used; the radiation originating from the sample was partially reflected in a quartz fibre and the reflected beam was used as a reference beam.The reference and measuring photomultipliers were side-window Hamamatsu R-106 (response type S-19) instruments. The reciprocal linear dispersion of this monochromator was also 0.8 nm mm-1. * For details of Part I of this series, see reference list, p. 378. t Present address : Departamento de Quimica, Universidade de Brasilia, Brasilia, Brasil.368 CAUSEY et a,!. : DETECTION AND DETERMINATION OF POLYNUCLEAR Analyst, VoZ. 101 Sources A 150-W xenon arc lamp (Osram XBO W/l) continuum source powered by a suitable supply unit (Perkin-Elmer, Model 150), which provided the start voltage and an operating voltage of 20 V d.c., was employed.The medium-pressure mercury-vapour discharge lamp (Philips, Type MBW/U) was operated at 125 W after removal of the outer envelope of Wood’s glass. A 3-mW helium - cadmium laser (Electro-Photonics/Liconix, Model 401/301) was used in the laser experiments. This laser has its principal emission at a wavelength of either 325 or 441 nm, which can be selected by changing the appropriate mirrors. Sample- handling Systems A commercially available Dewar-flask sampling system (American Instrument Co., Mary- land) similar to that described previously was employed. This system included a sample cell (length 200 mm, i.d. 3 mm and wall thickness 0.6-1 mm) constructed from silica tubing and which required sample volumes of between 0.3 and 0.5 ml.A new design of liquid-nitrogen cryostat sample cell for use with small liquid samples was assembled. The construction of this sample-handling system is shown in Fig. 1. The system is similar to those proposed by Parkerll and Svishchyov,12 without the inconvenience of the latter, with which it is necessary to warm the whole system in order to change the sample. Also, for compounds with biological activity, for example, some of the carcinogenic polycyclic aromatic hydrocarbons and aflatoxins, the system must be of practical use without the possibility of spillage or contamination C 7- E I E Ln Fig. 1. Copper cryostat sample cell assembly (for key, see text). The copper reservoir, A, into which the liquid nitrogen is introduced was surrounded with expanded polystyrene, H, as insulator.In the upper end of the reservoir a vacuum space, B, was provided in order to decrease heat transmission. At the lower end of the reservoir is a plane, vertical surface, C, to which the sample cell (D, E, F and G, when assembled) is attached by means of four copper screws, g. The copper cell, D, forms a sample compartment with a capacity of about 0.5 ml. Two small-diameter steel tubes (i.d. 1.58 mm), which end in syringe-type connections, a , provide for filling and flushing the cell. The sample compartment extends slightly outside the excitation area in order to allow for solvent contraction on cooling. The sample compartment was isolated from the ambient temperature by a quartz microcell, E, which was evacuated several times, filled to a low pressure (10 torr) with dry nitrogen and sealed in order to create a vacuum chamber.This chamber was used as a window and a thin rubber washer, c, was used to seal it against the wall of the copper cell. The chamber was tightened in position by using an aluminium front plate, F, and nylon screws, d. TheseMay, 1976 AROMATIC HYDROCARBONS BY LUMINESCENCE SPECTROMETRY. PART II 369 screws are slightly flexible at room temperature and have the advantage of being poor conduc- tors. Care must be exercised not to over-tighten the nylon screws as this would cause damage to the vacuum chamber. The front window of the chamber was defrosted by using a single turn of Nichrome wire (2.18 SZ m-l), G, which was heated by using the mains supply applied at 2-3 V a.c.across a rheostat. Temperature measurements In order to compare the cooling rate of the conventional Dewar-flask cell and the new cryostat assembly two copper - constantan thermocouples were used; the e.m.f. developed a t these was measured by using a Servoscribe recorder (Model Re 511.20). The thermocouples were tested and the measurements were calibrated by using liquid nitrogen, ice and solid carbon dioxide; the e.m.f. was related to temperature by reference to the appropriate published calibration ~alues.1~ The measuring thermocouple was introduced into the solvent either inside the silica tube in the Dewar flask or in the sample compartment of the cryostat cell. The reference thermocouple was introduced into a Dewar flask containing melting ice.Detection Systems D.c. integration For the experiments with constant-time and constant-charge integration the Quantoscan spectrometer was used with a sample cell and a mercury-vapour lamp excitation source, as described above. The measurement electronics employed consisted of simple integrator circuitry, which permitted display of the luminescence signal integration to constant charge or for a constant time. Photon counting A simple single-channel system (Model 300, EDT Research, London) having a capacity of lo7 counts was used for the photon-counting experiments and the output from the photo- multiplier was developed across a 50-SZ load resistor. A pulse-pair resolution of 10 ns was obtained with this system. The EM1 62565 photomultiplier tube used for the photon- counting experiments had low dark current (at ambient temperature about 40 counts s-l and at dry-ice temperature about 4 counts s-l).A specially built cooling chamber based on the design used by Sharp14 was used in order to cool the photomultiplier during the experi- ments. A constant flow of dry nitrogen was circulated around the photomultiplier so as to avoid condensation either during the cooling or the warming cycle. In most of the photon-counting experiments the copper cryostat cell was used. Signal averaging Experiments on the use of signal averaging were made with a 1 000-channel signal-processing system (Unimac 4 000, Data Laboratories Ltd., Mitcham, Surrey). This apparatus has an on-line digital processing system for the storage and analysis of waveforms in both the time and frequency domains. The instrument is modular and has four basic units.The first unit is a memory with 1 024 words and the second is a sweep timer, which generates timing pulses so as to control the rate at which the data is sampled during analysis. Delay times after each sweep (from 1 ps to 999 ms) and before each sweep (from 1 ms to 999 ps) can be selected. A programme unit is provided in which not only single-channel averaging is possible but also multi-channel averaging, as four single-channel inputs, which can be averaged simultaneously, are available. A display control module provides for the selection of the number of sweeps in intervals of 2” from 1 (2O) to 16 384 (214). In order to produce rapid repetitive scanning of small wavelength ranges in the luminescence emission spectra an oscillating, transparent refractor-plate mechanism, the operation of which is similar, in principle, to that of those proposed by McWilliam,ls Roldan16 and Snelleman et aZ.,17 was constructed.This assembly was located within the monochromator, behind the entrance slit. As the refractor plate, a poly(methy1 methacrylate) (Perspex) block of path length 25 mm, which was transparent from 345 nm throughout the visible region, was employed. The plate was driven by an oscillator circuit, similar to that described by Snelleman et aZ.,17 by which the A frequency of oscillation of about 5 Hz was used in all the experiments.370 CAUSEY et al. : DETECTION AND DETERMINATION OF POLYNUCLEAR Analyst, VOl.101 amplitude of oscillation could be controlled. The wavelength range scanned with this system was typically between 2 and 3 nm. A simple pre-amplifier (of gain 50) was constructed and used between the photomultiplier and the signal averager. The signal being averaged was continuously observed at a display oscilloscope (Hewlett-Packard, Model 175A) and after averaging was either photographed or point plotted by using a potentiometric chart recorder. Reagents The n-alkanes used in this work were of laboratory reagent grade and were purified by percolation through silica gel (60-120 mesh), which had been activated overnight at 110- 130 "C. Materials Samples of pure polynuclear aromatic hydrocarbons were kindly donated by Tobacco Research Council Laboratories, Harrogate, British American Tobacco Co., Southampton, and Shell Research Ltd., Chester. Results and Discussion Application of Different Excitation Sources A comparison was made of the use of the 150-W xenon arc source, the medium-pressure mercury-vapour discharge lamp and the helium - cadmium laser for excitation in the produc- tion of the quasi-linear luminescence spectra, at 77 K, of selected PAH compounds.Radiation from the xenon and mercury sources was selected by using interference filters of peak-trans- mission wavelength either 250 or 300 nm (with corresponding spectral half-band widths of 14 and 30 nm and about 30% transmission at the central wavelength). The laser excitation wavelength was either 325 or 441 nm without the use of an interference filter in the irradiation optics.Table I shows the detection limits (expressed as that concentration of the compound studied which produced a luminescence signal to noise ratio of 2 at the detector electronics) for the compounds selected for study. As shown for coronene and 3,4-benzopyrene, the detection limits obtained were typically lower by a factor of about two when the mercury- vapour lamp rather than the xenon arc source was used. The use of the mercury-vapour lamp rather than the xenon arc lamp also gives rise to less scattered radiation from the sample at wavelengths near to the wavelength of measurement of the luminescence emission. Fig. 2 shows the luminescence spectra obtained with the xenon and mercury sources in the region of 400 nm for a solution of 3,4-benzopyrene in octane at 77 K; the effect of the scattered back- ground radiation and noise from the continuum source is clearly visible.Although the mercury 404.7-nm line also contributes scattered radiation to the 3,4-benzopyrene spectrum its presence is readily identified and does not distort the shape of the observed emission TABLE I DETECTION LIMITS, p.p.m., OBTAINED AT 77 K WITH DIFFERENT EXCITATION SOURCES FOR PAH COMPOUNDS Excitation source Compound Curonem* 3,4-Benzopyrene 3 1,2-Benzanthracene: 1,2,5,6-Dibenzanthraccne: Perylene* 3,4.8,9-Dibenzopyrene Luminescence emission wavelength/nm 445.05 403.00 383.75 394.25 443.95 449.15 7 Mercury-vapour He - Cd laser 150-W discharge (-A---, xenon arc lamp 325 nm 441 nm 2 x 10-31. 1 x 10-3t 5 x 10-4 - 2 x 10-4t 1 x 10-4t 1 x 10-4 - - 2 x 10-3t 5 x 10-4 - - 5 x 10-37 3 x 10-3 - - 1 x 10-25 - 2 x 10-3 - 1 x 10-411 - 7 x 10-4 * In hexane.t 300 nm interference filter. $ In octane. $250 nm interference filter. 11 300 or 325 nm interference filter.May, 1976 AROMATIC HYDROCARBONS BY LUMINESCENCE SPECTROMETRY. PART 11 37 1 spectrum. The use of the xenon arc source has an advantage, however, when the compound to be determined exhibits one of its principal, narrow, Shpol’skii luminescence emission bands at a wavelength at which it would be overlaid by scattered radiation from mercury line radia- tion from the mercury-vapour discharge-lamp source. An example of this effect was observed for the weak quasi-linear emission of dibenzofuran in tetrahydrofuran at 77 K, which occurs at 434.10 nm; this emission is obscured by scatter of the strong broad mercury line from the mercury source at 435.8 nm, whereas with xenon arc lamp excitation the dibenzofuran lumines- cence at 434.10 nm is readily observed.For compounds whose luminescence excitation maxima lie at short wavelengths near to the mercury 253.7-nm line, however, the use of the mercury-vapour lamp produces greater sensitivities, the output intensity of the xenon source at these wavelengths being relatively low. For perylene, for example, whose wavelengths of maximum excitation occur at 246 and 266 nm, xenon arc source excitation resulted in a luminescence signal of intensity only one tenth of that obtained with the mercury-vapour lamp source. We are of the opinion that in instrument assemblies for low-temperature luminescence spectrometry it is useful to have both mercury and xenon-discharge lamp excitation sources available as for particular analyses the best performance will1 be attainable with only one of these sources.? (h) I 400 400 Wavelengthhm Fig. 2. Quasi-linear lumines- cence emission a t 403 nm for 3,4- benzopyrene in octane a t 77 K. (a), Mercury-vapour discharge-lamp source; ( b ) , xenon arc lamp source. Investigations in our laboratories, which are to be described in a later paper,l* have demon- strated that in n-alkane matrices at 77 K the luminescence excitation spectra of PAH com- pounds that exhibit the Shpol’skii effect in their emission spectra also exhibit characteristic narrow-band absorption maxima. In order to achieve efficient excitation at these wave- lengths a source which has high energy at the specific excitation wavelengths required for each compound is therefore desirable.Ideally, a tunable dye laser source would be employed. The only laser excitation source available to us, however, was a CW helium - cadmium laser with power output at 325 nm of 3 mW or at 441 nm of ‘7 mW. Obviously the use of this source can only be fully effective if the principal excitation maximum of the luminescence of the com- pound investigated overlaps the narrow-line output from the source at 325 or 441 nm. With excitation at 325 nm no gain in signal intensity was observed, compared with that obtained by use of the mercury-vapour lamp, for most of the compounds studied. As shown in Table I, however, some gain in detection limit for coronene, 1 ,e-benzanthracene and 1,2,5,6-dibenzan- thracene was observed with excitation at 325 nm with the laser source.For excitation at 441 nm an enhancement in both signal intensity and signal to noise ratio (detection limit) was obtained only for perylene. This results from the fact that the 441.6-nm laser line almost372 CAUSEY et al. : DETECTION AND DETERMINATION OF POLYNUCLEAR AnaZyst, VoZ. 101 exactly overlaps the longest wavelength (0,O) band in the absorption spectrum of perylene.10 This finding confirms our opinion that the use of a tunable dye laser source would be beneficial for low-temperature luminescence spectrometry of PAH compounds utilising the Shpol’skii effect. Sample-handling Systems The system consists of a silica sample tube, which is immersed in liquid nitrogen contained in a Dewar flask with a transparent silica bore so that incident radiation can enter in order to excite luminescence in the sample, which can then be viewed by an analysing spectrometer.While this system has a number of advantages, including a fast cooling rate and a stable final temperature of 77 K, the system suffers from a number of disadvantages. Poor reproduci- bility can arise from difficulty experienced in the positioning of the sample tube in the Dewar flask; this disadvantage has already been noted by Hollifield and Winefordner20 and Kirk- bright et aZ.,21 who have recommended alternative arrangements for use with a Dewar-flask or tube assembly. Other problems result from the condensation of atmospheric water vapour on the external surface of the Dewar flask and within the liquid-nitrogen coolant.Signal noise is also increased owing to the constant boiling of the liquid nitrogen in the optical path surrounding the sample cell. In order to eliminate these effects, and to improve the sensitivity and precision of low-temperature luminescence spectrometry, a new sample cell and cryostat system was designed and constructed (Fig. 1). The performance of this cell, with particular reference to sample cooling rate, final temperature achieved and precision and detection limits obtainable for PAH compounds, was compared with that of the conventional Dewar- flask sample cell. The Dewar-flask sampling system is in widespread use and is commercially available.Cooling rate and j n a l temperature The rate of cooling of sample solutions plays an important role in determining the width and intensity of the quasi-linear luminescence emission observed from PAH compounds in n-alkane solvents at 77 K.22 Temperature measurements for both the Dewar-flask and copper cryostat assembly were made by using a copper - constantan thermocouple inserted directly into sample solutions of coronene in hexane in each cell. As shown in Table 11, the mean cooling rate for the copper cryostat assembly was found to be considerably slower than that for the silica tube and Dewar-flask system, although the cooling rate for the latter system is a function of the wall thickness of the sample tubes employed. As expected, the final temperature attained by the sample in the silica-tube sample cell immersed in liquid nitrogen within the Dewar flask was 77 K, while for the copper cryostat assembly the equili- brium temperature attained by the frozen sample solution was higher, Le., 80 K.TABLE I1 COMPARISON OF MEAN COOLING RATES FOR THE TWO SAMPLE CELLS EMPLOYED Sample cell Silica tube: 0.6 mm wall thickness 1.0 mm wall thickness Copper cryostat cell Ratio of intensities of components Mean cooling of coronene doublet rate/K min-l (I445.1, m d 1 4 4 9 . 4 4 nm) 720 540 76 3.3 f 0.3 3.7 & 0.3 4.05 f 0.1 Personov2 has observed an increase in the ratio of the intensity of the coronene 445.15-nm emission to that of its emission at 443.44 nm in hexane at 77 K when slow rather than rapid cooling is employed. As shown in Table I1 we have confirmed this observation, although the change in the intensity ratio observed is relatively small in relation to the large difference in cooling rate that occurs in the two sampling systems.Precision attained with Dewar--ask and copper cryostat sample cells A comparison was made of the reproducibility of quasi-linear luminescence signals obtained at 77 K for coronene and 3,kbenzopyrene, each at two different concentrations in both sampleMay, 1976 AROMATIC HYDROCARBONS BY LUMINESCENCE SPECTROMETRY. PART 11 373 cells. The emission intensity obtained at the principal luminescence wavelengths was measured repetitively for sample solutions introduced into each cell. The relative standard deviations obtained for these sample solutions are shown in Table 111.I t is evident that a substantial improvement in precision is possible with the copper cryostat assembly. This results principally from the reproducible alignment of the sample in the optical path of the spectrometer and the elimination of noise, which is caused in the silica-tube system by the boiling of the coolant liquid nitrogen in the path of the incident and luminescence radiation. TABLE I11 REPRODUCIBILITY OF RESULTS OBTAINED WITH SILICA-TUBE AND COPPER CRYOSTAT CELLS Relative standard deviation, % A Luminescence I 7 Concentration/ emission Silica- tube Copper Compound M Solvent wavelength/nm cell cryostat cell Coronene 2.5 x 10-6 Hexane 445.05 9.7 1.6 Coronene 2.5 x 10-7 Hexane 445.05 12 2.6 3,4-Benzopyrene 1 x Octane 403.00 4 1.3 3,4-Benzopyrene 1 x Octane 403.00 18 11 Techniques Employed for Registration of Luminescence Signals The use of three techniques for recording the luminescence signal received by the photo- multiplier at the exit slit of the monochromator was evaluated.These techniques were d.c. integration, photon counting and repetitive optical scanning used in conjunction with a 1 000- channel signal-averaging system. D.c. integration An investigation was made of the use of simple signal-integration techniques as an alterna- tive to direct presentation of the voltage generated across a load resistor on the output of the photomultiplier. Experiments with integration for a constant time and also for the time required to reach a constant charge were undertaken with sample solutions of 3,kbenzopyrene in octane at 77 K.In the constant-time mode of operation an integration time of 60 s was used with an applied voltage of 700 V at the side-window photomultiplier fitted in the Quanto- scan spectrometer. As the background luminescence intensity was high in these experiments the integrated background signal at 400 nm was measured and backed-off electrically before integration of the luminescence at the 403-nm emission line of 3,4-benzopyrene. Table IV shows that, as expected, a considerable improvement in the relative standard deviation of the luminescence signals is obtained when constant-time integration is employed rather than direct display of the instantaneous luminescence signal recorded from the photomultiplier across a load resistor.The detection limit for 3,4-benzopyrene is improved from about 5 x lo-' M to TABLE IV RELATIVE STANDARD DEVIATIONS OBTAINED IN MEASUREMENT OF LUMINESCENCE OF 3,4-BENZOPYRENE I N OCTANE AT 77 K WITH INSTANTANEOUS AND D.C. INTEGRATION TECHNIQUES OF SIGNAL REGISTRATION ConcentrationlM 1 x 10-4 i x 10-5 5 x 10-7 1 x 10-7 6 x 2.5 x 2.6 x 10-0 5 x 10-6 Relative standard deviation, % Integration for fixed Integration to Direct read-out time period constant charge 2.5 3.0 - 3.0 3.0 - 9.0 4.6 0.9 25.0 3.0 1.4 - 9.4 0.7 - 6.5 1.6 - 6.0 4.4 - 4.4 6.0 A 1374 CAUSEY et al. : DETECTION AND DETERMINATION OF POLYNUCLEAR Analyst, VoZ. 101 about 1 x M. Table IV also shows that further improvement in the relative standard deviation of the luminescence signals was obtained when the constant-charge mode of integra- tion was adopted.This further improvement in precision is explained by the availability of the signal from the reference photomultiplier, which is employed to correct the measured signal integration period for variation in the over-all light input due to fluctuation in the intensity of the excitation source during this time. It should be noted that, while the use of signal integration can result in substantially im- proved detection limits for PAH compounds by measurement at a known wavelength corres- ponding to a peak in the luminescence emission spectrum, this improvement obviously cannot be achieved in qualitative identification studies at low concentrations, for which the lumines- cence emission spectrum must be scanned over a wide wavelength range.Photon counting In the technique of photon counting the current pulses that result at the anode of the photomultiplier due to photon impact at the photocathode are counted directly. When sufficient resolution is available electronically this technique provides a direct digital measure- ment of the intensity of the radiation incident upon the photomultiplier. Advantages established for the technique include the ability to provide detection at low light levels, improvement in signal to noise ratios, the decrease of effective dark current, accurate signal integration, improved precision of analytical results and direct digital output of data.23 A number of workers have described the advantages of the application of the photon-counting technique to spectrofluorimetry and have demonstrated that enhanced detection limits can be obtained for many corn pound^.^^-^^ In our work we have examined the use of a simple single-channel photon-counting detection system for registration of the quasi-linear lumines- cence of 3,4-benzopyrene and 3,4,9,10-dibenzopyrene at 77 K.Fig. 3(a) shows the variation with the voltage applied to the photomultiplier of the quasi- linear emission count and background count (over 5 s) for a 1 x M solution of 3,4-benzo- pyrene in octane. The signal was recorded at the 403.0-nm peak for 3,4-benzopyrene and the background signal was recorded at 400.0 nm. The influence, at this low concentration, of the relatively high level of background radiation due to scattered source radiation with the “frontal” illumination of the polycrystalline octane matrix is clearly visible.Fig. 3(b) shows the variation of the observed signal to noise ratio with the voltage applied t o the photo- multiplier, the signal to noise ratio having been calculated from the expression proposed by Franklin, Horlick and Malm~tadt~~ for situations in which the background count rate is high. Applied photomultiplier voltage X Applied photomultiplier voltage X Fig. 3. (a), Variation of signal count with photomultiplier voltage for 3,4-benzopyrene: A, total signal count a t 403.00 nm in 5 s; B, background count at 400.00 nm in 5 s; and C, dark-current count in 5 s. ( b ) , Variation of signal to noise ratio with applied voltage a t photomultiplier using photon counting for 3,4- benzopyrene luminescence at 403.00 nm.Thus, if the signal to noise ratio is given by 2, thenMay, 1976 AROMATIC HYDROCARBONS BY LUMINESCENCE SPECTROMETRY. PART II 375 where Rs is the signal count rate, R, is the background count rate and T is the total counting time. The anomolous decrease in signal to noise ratio observed at 2 200 V is an anomalous characteristic of our particular experimental assembly and the photomultiplier tube employed; no explanation for this effect can be offered. Further experiments were conducted with an applied voltage to the photomultiplier of 2 000 V. Fig. 4 shows the luminescence growth curves obtained for 3,4-benzopyrene, utilising both photon-counting and analogue signal registration: the counting time was 5 s.Both photon- counting and analogue readings correspond to the net difference between the signals obtained at the luminescence maximum (403.0 nm) and for the background at 400.0 nm. A similar range of linearity is observed with each system, ie., in the concentration range 10-7-10-9 M. The deviation from linearity at concentrations above lo-’ M can be attributed to an “inner- filter” effect caused by self-absorption of luminescence radiation. Similar experiments were carried out with 3,4,9,10-dibenzopyrene and it was again observed that the ranges of linearity in the luminescence growth curves obtained for this compound at 77 K by photon-counting and analogue measurement techniques were similar. 107 1 06 1 o4 1 o3 lo2 10’ I I I I 1 0 - l ~ I O - ~ 10-8 I O - ~ 10-5 Mo larity Fig.4. Comparison of analytical growth curves for 3,4-benzopyrene obtained with photon-counting and analogue signal registration : A, photon count- ing (count obtained for 5 s count time) ; and B, analogue read-out (mv). Although the growth curves of Fig. 4 might suggest that an improvement in detection limit is obtainable by using photon counting, the determination of the relative standard deviation for replicate measurements of low concentrations indicates that the limit of detection for the compounds investigated is similar for photon-counting and analogue techniques. Table V shows the relative standard deviations obtained for solutions of 3,4-benzopyrene and 3,4,9,10-dibenzopyrene at different concentrations. It is evident that, whereas the detection limits for the compounds are similar, at determinable concentrations above the detection limit the over-all precision attainable with the photon-counting technique is superior to that of simple analogue measurements.The effect of variation in the counting time on the signal to noise ratio obtained for 3,4- benzopyrene was investigated for a M solution of the compound in octane (Table VI). As shown in Table V, the signal to noise ratio obtained is approximately proportional to TI, where T is the counting time, as would be expected for situations in which the background count rate is high (equation 1). The counting times required in order to obtain a substantial improvement in detectability, however, are unacceptable for routine practical application.From the results obtained in this study of the use of photon counting for detection of quasi-376 CAUSEY et d. : DETECTION AND DETERMINATION OF POLYNUCLEAR Analyst, vd. 101 TABLE V PRECISION OF PHOTON COUNTING AND ANALOGUE SIGNAL REGISTRATION FOR LUMINESCENCE OF PAH COMPOUNDS AT 77 K Compound 3‘4-Benzop yrene Concentration/nr 5 x 10-10 1 x 10-0 1 x 10-8 1 x 10-6 1 x 10-6 1 x 10-8 1 x 10-8 1 x 10-6 1 x 10-6 1 x lo-' 1 x 10-7 1 x 10-7 Relative standard deviation, yo A r \ Photon counting Analogue 46 32 2.0 26 1.3 6.3 0.5 3.9 0.2 1.6 0.7 1.4 5.0 6.2 0.6 4.4 0.3 4.0 0.6 2.2 0.9 1.1 44 33 linear luminescence from PAH compounds we must conclude that the fundamental advantages are difficult to exploit. This difficulty undoubtedly results from the presence of high back- ground light levels due to scattered source radiation with the copper cryostat sample cell and optical arrangement employed.Signal averaging In order to permit rapid spectral-data acquisition over chosen small wavelength ranges in the region of the principal quasi-linear luminescence emission maxima for PAH compounds, and to obtain enhanced sensitivity of detection, a simple repetitive optical scanning device was constructed for use in conjunction with a multi-channel analyser. The repetitive optical scanning was achieved in a manner similar to that described by Snelleman et aZ.,l7 using an oscillating refractor plate mounted within the monochromator behind the entrance slit. This device then causes repetitive displacement of the monochromator and allows rapid scanning of small wavelength ranges. The lateral displacement of the light beam, which governs the wavelength range that can be scanned, is given, approximately, by the relationship ..(2) A, = t ~ ( f i - l ) / n .. .. where t is the thickness of the plate, cc is the angle of incidence (radians, in air) and n is the refractive index of the refractor-plate material. In addition, the light beam is displaced longitudinally owing to change in the optical path length so that the image is defocused at the exit slit .I7 This displacement approximates to .. The longitudinal displacement can be corrected for either by using a refractor plate whose surface has been ground so as to produce a convex curvature of long focal length or by corres- ponding displacement of the exit slit.In order to scan a relativelylarge wavelength range a Perspex refractor plate (ngo = 1.49) of 25 mm thickness was used. With the oscillator drive circuit used the maximum amplitude of oscillation obtainable for this plate was &13", which produced a limit of 3 nm on the wavelength range scanned. The output from the photomultiplier was led to the pre-amplifier and then to the 1000- TABLE VI VARIATION I N SIGNAL TO NOISE RATIO WITH COUNTING TIME FOR 3,4-BENZOPYRENE Counting time/s Signal to noise ratio 5 10 10 15 15 19 30 26 60 36365.0 365.47 \--2nrn-I Fig. 5. Calibration spectrum from mercury-vapour lamp obtained for repetitive optical scanning with signal averaging. Fig. 6. Quasi-linear luminescence emission a t 403.0 nm of 3,4-benzo- pyrene (5 x lo-0 M) in octane obtained by repetitive optical scanning with time averaging.Fig. 7. Quasi-linear luminescence emission of coronene a t 445.05 nm obtained by repetitive optical scanning with time averaging (forward and return scans shown).May, 1976 AROMATIC HYDROCARBONS BY LUMINESCENCE SPECTROMETRY. PART 11 377 channel signal-averaging system. The output from the signal averager was displayed on an oscilloscope or printed out on to a chart recorder. The wavelength calibration and scan range were determined by using the emission of the mercury triplet from a low-pressure mercury- vapour discharge lamp; as shown in Fig. 5 the attainable scan range is between 2 and 3 nm, depending upon the power applied to the oscillator. With repetitive presentation of spectral data to the signal averager the signal increases linearly in the register, depending on the number of sweeps, N , whereas random fluctuations (noise) will accumulate only as the square root of the number of sweeps, N*.Consequently the signal to noise ratio obtained increases as N*. An evaluation has been made of the application of repetitive scanning with signal accumula- tion by time averaging to the detection of the quasi-linear luminescence of low concentrations of 3,4-benzopyrene in octane and coronene in hexane at 77 K. Typical spectra obtained for their luminescence emission at 403.00 and 445.05 nm are shown in Figs. 6 and 7. Early experimentswith this technique indicate that it is likely to prove most valuablein low-tempera- ture luminescence spectroscopy.The technique allows the high resolution of the spectro- meter to be utilised rapidly to study the effect of such parameters as cooling rate and solvent type on the half-width and intensity of quasi-linear emission bands and also enables study of the effect of solute concentration and the presence of other PAH compounds on these features of the luminescence emission. In addition, and most importantly from the viewpoint of practical application to the determination of PAH compounds, the technique of repetitive optical scanning used in conjunction with signal averaging permits improved analytical sensi- tivity to be obtained. Conclusions The instrument system assembled permits the quasi-linear luminescence of PAH compounds at 77 K to be used for their detection and determination with high sensitivity.For practical analytical work the use of a xenon arc or mercury-vapour lamp source with the copper cryostat sample cell and d.c. integration of the luminescence signals provides a simple and reliable system. The use of the photon-counting technique for registration of quasi-linear luminescence signals with our instrument assembly does not result in a substantial gain in sensitivity, probably because of the high background light levels encountered with front-surface illumina- tion using our sample cell. Repetitive optical scanning shows considerable promise for both qualitative and quantitative analytical work using quasi-linear luminescence spectrometry. The use of this technique with signal averaging permits operation of the spectrometer at the low light levels obtained when the maximum spectral resolution is required at narrow slit-widths for fundamental studies or when only very low concentrations of the analyte are present.Pilot studies with a simple helium - cadmium laser indicate that a significant gain in sensitivity and spectral selectivity would be obtained by the use of a tunable laser source for excitation of quasi-linear luminescence of PAH compounds. We thank the Science Research Council for an equipment grant for construction of the luminescence spectrometer. One of us (C.G. de L.) thanks the University of Brasilia for study leave and UNESCO for the grant of a Fellowship. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. References Dikun, P. P., Vop. Onkol., 1961, 7, 42. Personov, R.I., J. Analyt. Chem. USSR, 1962, 17, 503. Varshavskii, I. L., Shabad, L. M., Khesina, A. Ya., Khitrovo, S. S., Chalabov, V. G., and Pakhol’nik, Eichhoff, H. J., and Kohler, N., 2. Analyt. Chem., 1963, 197, 272. Fedoseeva, G. E., and Khesina, A. Ya., J . Appl. Spectrosc. USSR, 1968, 9, 838. Personov, R. I., and Teplitskaya, T. A., J . Analyt. Chem. USSR, 1965, 20, 1176. Jager, J., and LugrovA, O., ChemickB Zvesti, 1965, 19, 774. Dikun, P. P., Krasnitskaya, N. D., Gorelova, N. D., and Kalinina, I. A., J . Appl. Spectrosc. USSR, Florovskaya, V. N., Teplitskaya, T. A., and Personov, R. I., Geochem. Int., 1966, 3, 419. Kirkbright, G. F., and de Lima, C. G., Analyst, 1974, 99, 338. Parker, C. A., “Photoluminescence of Solutions,” Elsevier, Amsterdam, 1968. Svishchyov, G. V., Optics Spectrosc., N.Y., 1965, 18, 360. A. I., J . Appl. Spectrosc. USSR, 1965, 2, 68. 1968, 8, 254.378 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. CAUSEY, KIRKBRIGHT AND DE LIMA Weast, R. C., Editor, “The Handbook of Chemistry and Physics,” Forty-eighth Edition, The Chemical Rubber Publishing Co., Cleveland, Ohio, 1967-68. Sharp, B. L., PhD Thesis, University of London, 1972. McWilliam, I. G., J . Scient. Instrum., 1959, 36, 51. Roldan, R., Rev. Scient. Instrum., 1969, 40, 1 388. Snelleman, W., Rains, T., Yee, K., Cook, H., andMeins, O., Analyt. Chem., 1970, 42, 304. Farooq, R., and Kirkbright, G. F., to be published. Personov, R. I., Al’shit, E. I., Bikovskaya, L. A., Optics Commum., 1972, 6, 169. Hollifield, H. C., and Winefordner, J. D., Analyt. Chem., 1968, 40, 1759. Kirkbright, G. F., Mayne, P. J., and West, T. S., Analytica Chim. Acta, 1971, 54, 353. Dokunikhin, N. S., Kizel, V. A., Sapozhnikov, M. N., and Solodar, S. L., Optics Spectrosc., N.Y., 19G8, Franklin, M. L., Horlick, E., and Malmstadt, H. V., Anulyt. Chem., 1969, 41, 2. Aoshima, R., Iriyama, K., and Asai, H., Appl. Opt., 1973, 12, 2 748. Curry, R. E., Pardue, H. L., Mieling, G. E., and Santini, R. E., Cliqz. Chem., 1973, 19, 1259. Tuan, V. D., and Wild, U. P., Appl. Opt., 1973, 12, 1286. 25, 42. NOTE-Reference 10 is to Part 1 of this series. Received November 28th, 1975 Accepted January 6th, 1976

ISSN:0003-2654    

DOI:10.1039/AN9760100367

出版商:RSC    

年代:1976

数据来源: RSC

摘要:

Anmlyst, May, 1976, Vol. 101, pp. 379-380 379 Determination of Dicumenyl Peroxide by Gas Chromatography P. Hudec, B. Novotnh and J. Petrbj Reseavch Institute of Macromolecular Chemistry, Tkalcovskd 2, Bmo, Czechoslovakia A method is described for the determination of small amounts of dicumenyl peroxide in liquid samples or powdered materials. Dicumenyl peroxide is isolated from the latter by extraction. Dicumenyl peroxide is one of the additives that are widely used in the industrial processing of polymers. Continuous checking of its concentration during processing is desirable because its residues could considerably affect the product. For the determination of dicumenyl peroxide the method of reduction in a mixture of acetic acid and alkali-metal iodide, followed by quantitative determination of the liberated iodine,l is frequently used.This method, combined with thin-layer chromatography, has also been successfully applied to the determination of dicumenyl peroxide in mixtures with other organic peroxides.2-4 The determination of low relative molecular mass peroxides and hydroperoxides by use of gas chromatography has been described in the l i t e r a t ~ r e . ~ , ~ The usefulness of this method is restricted by the volatility, as well as the stability, of the substance under analysis, 2,5- dimethyl-2,5-bis(tert-butylperoxy) hexane being the limiting case reported. The thermal stability of dicumenyl peroxide is relatively high,7 and we therefore attempted to find suitable conditions for its gas-chromatographic determination.8 Experimental Reagents and Materials Dicumenyl peroxide, more than 98% pure.Octadecane, more than 99% pure. Cumen-7-01, boiling point 102 "C at 2.6 kPa. Phenol, acetone, toluene, acetophenone and methanol. Chromosorb W DMCS (60-80 mesh). Fluorosilicone oil QF 1. The material to be analysed for determination of the dicumenyl peroxide content was in the form of clays, modified by the reaction with butyl l13-diacrylate in the presence of dicumenyl peroxide as initiator. The determination of unreacted dicumenyl peroxide served as the test for the efficiency of modification. Obtained from Noury van der Lande. Prepared in this Institute. Analytical-reagent grade. Apparatus and Conditions A Perkin-Elmer F11 gas chromatograph, fitted with a flame-ionisation detector, was used.A 3 ft x $ in 0.d. glass column was packed with 3% of QF 1 on 60-80-mesh Chromosorb W DMCS and argon was used as the carrier gas at a flow-rate of 40 ml min-l. The optimum column temperature was found to be 110 "C, whereas the injection port was not heated. The sample (2 p1) was introduced by use of a 10-pl syringe that was fitted with a 4-in needle. A Hitachi Perkin-Elmer 2.5-mV recorder, Model 159, was attached to the instrument. Preparation of Samples A sample of modified clay (1 g) was weighed into a 25-ml beaker and mixed with toluene (10 ml). After stirring it for 5 min the slurry was filtered through a sintered-glass filter stick and the filtrate collected; another portion of toluene was then added to the clay and the procedure was repeated. The extracts were combined, 5 ml of octadecane solution in toluene (internal standard) were added and, by making the total volume up to 50 ml, the solution was prepared for gas-chromatographic analysis.380 HUDEC, NOVOTNA AND P E T R ~ J Optimum Gas-chromatographic Conditions Initially, when the analysis was carried out at column temperatures above 140 “C, the decomposition of peroxide was observed even when the injection port was not heated.In view of the fact that the kinetics and mechanism of dicumenyl peroxide decomposition have been described elsewhere,g we employed the published rate constants in order to find the optimum column conditions for the gas-chromatographic determination. The chosen column temperature of 110 “C was a compromise value at which the decomposition could still be neglected (it was less than 1%) although the retention time was acceptable. Under the column conditions described above the peaks due to the solvent, octadecane and dicumenyl peroxide are well resolved.Results and Discussion Fluorosilicone oil QF 1 was selected for its low polarity and short retention time. The retention times are given in Table I. TABLE I RELATIVE RETENTION TIMES OF CHROMATOGRAPHED MATERIALS Compound Relative retention time Toluene . . . . . . 0.04 Octadecane . . .. . . 1.00 (7.2 min) Dicumenyl peroxide . . . . 2.13 Calibration On plotting a graph of the mass ratio of the dicumenyl peroxide and octadecane peaks versus the ratio of their concentrations, a straight line passing through the origin was obtained. For equal concentrations of dicumenyl peroxide and octadecane the mass ratio of the peaks was found to be 0.788.The mean deviation of experimental points from the calibrated plot was &3.2y0 over the entire concentration range. The main products of the decomposition of dicurnenyl peroxide are methane, acetone, acetophenone, phenol and cumen-7-01; their relative concentrations are governed by the reaction mechani~m.~ With the column used in this work all of these compounds are eluted before octadecane so that the determination is not affected. Quantitative Study of the Extraction of Dicumenyl Peroxide from Modified Clays A sample (1 g) of surface-modified clay was repeatedly extracted with toluene as described above. The toluene extracts were analysed in order to determine the dicumenyl peroxide content. Three extractions were found to recover 99% of the peroxide from the clay.Example Modified clay with an initial dicumenyl peroxide content of O . l O ~ o was used for the experi- ments. Following the modification process the samples were analysed to determine the residual dicumenyl peroxide concentration by using the method described. The mean value of the results of four analyses was 0.089 & 0.006% m/m. Thus, in the course of modification, only about 10% of the dicumenyl peroxide that was present initially participated in the reaction. The authors are grateful to Ing. K. Otto of this Institute for fruitful discussions, 1. 2. 3. 4. 5. 6. 7. 8. 9. References Mair, R. D., and Graupner, A. J., Analyt. Chem., 1964, 36, 194. Brammer, J. A., Frost, S., and Reid, V. W., Analyst, 1967, 92, 91. Buzlanova, M. M., Stepanovskaja, V. P., and Antonovskij, V. L., J . Analyt. Chem. USSR, 1966, Hayano, S., Ota, T., and Fukushima. Y.,BunsekiKagaku, 1966,15,365; Chem. Abstr., l967,67,39941r. Bukata, S. W., Zabrocki, L. L., and McLaughlin, M. F., Analyt. Chem., 1963, 35, 885. Abraham, M. H., Davies, A. G., Llewellyn, D. R., and Thain, E. M., Analytica Chim. Acta, 1957, Swern, D., “Organic Peroxides,” Wiley-Interscience, New York, 1971. Hudec, P., and Petrfij, J., Chemikj Pram., 1975, 25/50, 95. Markert, H., and Wiedermann, R., Siemens Forsch. -u. Entwicklungs-ber., 1973, 2, 85. 21, 1324. 17, 499. Received October 21st, 1974 Amended October 31st, 1975 Accepted November 24th, 1976

ISSN:0003-2654    

DOI:10.1039/AN9760100379

出版商:RSC    

年代:1976

数据来源: RSC