ISSN:0144-557X    

DOI:10.1039/AP98017FP001

出版商:RSC    

年代:1980

数据来源: RSC

ISSN:0144-557X    

DOI:10.1039/AP98017FX016

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

ANALYTICAL DIVISION DIARY 159 April, 1980 Annual Reports on Analytical Atomic Spectroscopy VOLUME 8, 1978 Edited by Dr J. B. Dawson and Dr B. L. Sharp This series provides the practising analytical chemist and spectroscopist with a handbook of current practice and recent advances in instruments and methods for the determin- ation of elements in the form of comprehensive, critical annual reports. As atomic spectrometric analysis has matured, progress in realizing the goal of precise and accurate analysis has led to the appearance of many papers in which information concerning fundamental principles, instrumentation and methodology is found within a single publication. For this reason, the structure of Volume 8 differs from its predecessors in that the two parts, 'Fundamentals and Instrumentation' and 'Methodology,' have been expanded into four chapters, 'Atomization and Excitation,' 'Instrumentation,"Methodology,' and 'Applications,' in order to give a clearer distinction beween work that is of general interest and that dedicated to a specific analysis. Hardcover 285pp 82" x 6" f 17.50 (CS Members f 13.00) (Stil/ available: Vols 3-7 covering 7973 to 7977) Obtainable from : The Chemical Society, Distribution Centre, Blackhorse Road, Letchworth, Herts., SG6 1 HN

ISSN:0144-557X    

DOI:10.1039/AP98017BX018

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

ANPRDI 17(4) 103-160 (1980) ISSN 0306-1 396 April 1980 Analytical Proceedings Proceedings of the Analytical Division of the Chemical Society AD President R. Belcher Hon. Secretary P. G. W. Cobb Hon. Treasurer J. K. Foreman Hon. Assistant Secretaries D. 1. Coomber, O.B.E.; D. C. M. Squirrel1 Hon. Publicity and Public Relations Officer Dr. A. Townshend, Department of Chemistry, University of Birmingham, Birmingham, 61 5 2TT Secretary Miss P. E. Hutchinson Editor, Analyst and Analytical Proceedings P. C. Weston ' Assistant Editors Mrs. J. Brew, R. W. Hazell, R. A. Young Publication of Analytical Proceedings is the responsi- bility of the Analyst Publications Committee: J. M. Ottaway (Chairman) W. H. C. Shaw H. J. Cluley D. Simpson 'P. Gray A. Townshend J. N. Miller G. E. Penketh T.B. Pierce "Ex officio members 'P. C. Weston J . White head All editorial matter should be addressed to: The Editor, Analytical Proceedings, The Chemical Society, Burlington House, Piccadilly, London, W1 V OBN. Telephone 01 -734 9864. Telex 268001. Advertisements: Advertising Department, The Chemical Society, Burlington House, Piccadilly, London, W1 V OBN. Telephone 01 -734 9864. @ The Chemical Society 1980 Editorial With this edition of Analytical Proceedings, the changes initiated in the January 1980 issue are taken a stage further and the style and format of the journal are now in the final form adopted by the Analyst Publications Committee. The advance sale of advertising space allowed the introduction of the red cover from the January issue and we naturally hope that readers will find this and all the other changes attractive and stimulating and that Analytical Proceedings will be of increasing value to their work.The Publications Received items have been expanded to include more detailed information on the contents of the books listed. Another innovation is the introduction of a Reader Enquiry Service, using a reply-paid form as one page of each issue. This will cover advertisements as well as items from Equipment News and, when appropriate, courses and conferences. We trust that readers will make effective use of this extra facility. In future there will be a regular Editorial feature on this page. These articles will be prepared by individual members of the Com- mittee, will represent their views alone, and are intended to be thought-provoking and perhaps controversial. Readers are invited to comment or reply via the Correspondence column! Analytical chemistry is a significant branch of chemistry with a distinct philosophy of its own. In my view the role and the value of the analytical chemist are not always understood or appreciated by chemists in other fields and scientists generally. The Editorials and Correspondence features could provide a means for the expression of views of members of the Analytical Division and others on this subject, and on the current and developing role of the analytical chemist in society as a whole. J. M. Ottaway Chairman, Analyst Publications Committee 107

ISSN:0144-557X    

DOI:10.1039/AP9801700107

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

108 ANALYTICAL DIVISION ANNUAL GENERAL MEETING Anal. PYOC. Reports of Meetings Microchemical Methods Group The thirty-sixth Annual General Meeting of the Group was held at 0.15 p.m. on Tuesday, January 15th, 1980, in the Meeting Room of the British Academy, Burlington House, London, W. 1. The Chair was taken by the Chairman of the Group, Mr. A. C. Thomas. The following office bearers were elected for the forthcoming BR3 3BS. Honorary Treasziver-Ah. M. R. Cottrell. HofLovary Assistant Secreta#*y-Mr. B. T. Saunderson. Members of Committee- Mrs. D. Butterworth (ex oflcio), Dr. G. Ingram, Dr. D. A. Pantony, Mr. R. C. Rooney, Mr. R. Sawyer and M i . W. J. Swindall. Mr. S. Bance and Mr. H. I. Shalgosky were re-appointed as Honorary Auditors. year: Chaivmnn-Mr. A. C. Thomas. Vice- Chairman--Mr.G. J. Dickes. Honovary Automatic Methods Group Secretary-Mr. P. R. W. Baker, Department of The fourteenth Annual General Meeting of the Physical Chemistry, Wellcome Research Labora- Group was held at 1.40 p.m. on Thursday, tories, Langley Court, Beckenham, Kent, December 13th, 1979, at the Scientific SocietiesApril, 1980 REPORTS OF MEETINGS 109 Lccture Theatre, 23 Savile Row, London, W.l. The Chair was taken by the Chairman of the Group, Dr. P. B. Stockwell. The following office bearers were elected for the forthcoming year: Chairman-Mr. S. R. Hill. Vice- CJznirman-Mr. D. G. Porter. Honoravy Secvefnvy-Dr. C. J. Jackson, Health and Safety Executive, Occupational Medicine and Hygiene Laboratory, 403 Edgware Road, London, NW2 6LN. Honorary Treasurer-Mr. J.L. Martin . Sx r e tla ry---ILI r . A. G. Hudson. Membevs of Coinnzitfee-Dr. D. Betteridge, Mr. A. L. Dennis, Mr. F. R. B. Fearn, Mr. D. Rowe, Mr. D. C. 1%. Squirrel1 and Mr. I<. H. Wall. Dr. J . E. Page and Mr. R. Sawyer were re-appointed a.5 Honorxy Auditors. H o no rary A ss i s tant Electroanalytical Group The tenth Annual General Meeting of the Group was held a t 2 p.m. on Friday, December 7th, 1979, at the Institute of Cancer Research, Royal Marsden Hospital, Clifton Avenue, Sutton . The Chair was taken by the Chairman of the Group, Dr. B. J. Birch. The following office bearers were elected for the forthcoming year : Chaivnzan-Dr. B. J . Birch. Vice-CJaairman- Dr. J . D. R. Thomas. Honorary Secretary- Mr. A. E. Bottom, Kent Industrial Measure- ments Ltd., EIL Analytical Instruments, Hanworth Lane, Chertsey, Surrey.Honorary Treasurer-Dr. A. G. Fogg. Honorary Assistant Secretary-Dr. T. H. Ryan. Members of Committee-Dr. J. Burmicz, Mr. J. Comer, Mr. I. Davidson, Dr. J . P. Hart, Dr. D. Inman (representative from Electrochemistry Group), Dr. P. 0. Kane, Dr. H. Thompson and Dr. A. Watson (co-opted). Mr. J. G. Tillman and Dr. R. M. Smith were re-appointed as Honorary Auditors. Education and Training Group The ninth Annual General Meeting of the Group was held a t 6 p.m. on Wednesday, December 19th, 1979, a t Chelsea College, Manresa Road, London, S.W.3. The Chair was taken by the Chairman of the Group, Dr. J. G. Pritchard. The following office bearers were elected for the forthcoming year: Chairman-Dr.J. G. Pritchard. Vice-Chairman-Dr. E. J . Green- how. Honorary Secretary-Dr. J. F. Tyson, Department of Chemistry, University of Tech- nology, Loughborough, Leicestershire, LE 11 3TU. Honorary Treasurer-Mrs. M. I. Arnold. Members of Committee-Dr. D. M. W. Anderson, Professor D. T. Burns (ex oficio), Mr. H. A. Glastonbury, Mr. B. Mills, Dr. J. M. Skinner, Mr. A. F. Smith and Dr. J. D. R. Thomas. Dr. N. T. Crosby and Dr. W. J . Williams were re-appointed as Honorary Auditors. The primary aim of the Group is to improve education and training in analytical chemistry. To this end it takes an interest in all establish- ments wherein analytical chemistry is taught or practised. The Group organises meetings on and makes scrveys of new and changing trends in the practice of analytical chemistry. It considers the impact of these on the teaching of analytical chnmistry and provides a forum for debate on educational policy and training requirements.The group aims to publish on these matters frorn time to time in Analytical Proceedings and elsewhere and also aims to encourage young ptxople to take up careers as analytical chemists. Joint Pharmaceutical Analysis Group The tenth Annual General Meeting of the Group was held a t noon on Thursday, January 17th, 1980, a t the Pharmaceutical Society of Great Britain, 1 Lanibeth High Street, London, SE1 7JN. The Chair was taken by Mr S. C. Jolly, who apologised for the absence of the Chairman, Mr. E. Addison, due to illness, Mr. Jolly thanked the sponsoring bodies for their continuing support.The Annual Report for 197 9 was presented by the Honorary Secretary, Miss I. Ladden. Mr. Jolly reported that the three ordinary members retiring from the Committee of Management this year were Miss &I. C. Cone, Mr. G. I?. Phillips and Dr. L. R. Rowc. On be- half of the Group he thanked all the retiring members for their valuable help in organising discussion meetings ; particular thanks were due to Mr. Phillips for his long service on the Com- mittee, including the Chairmanship in 1974 and 1975. The results of the election to fill the three vacancies on the Committee of Management were announced. Dr. D. H. Calam, Mr. J. C. Deavin and Mr. A. J. Middleton had been successful in the ballot. Mr. Jolly introduced the new Chairman of the Group for 1980/1981, Mr.H. E. Brookes. Mr. Brookes proposed a vote of thanks to the retiring Chairman and the Secretary undertook to send Mr. Addison the best wishes of the Group. Mr. Jolly congratulated Dr. D. C. Garratt on his appointment as an Officer of the Order of the British Empire in the New Year Honours List. Dr. Garratt had been instrumental in forming the Joint Pharmaceutical Analysis Group and110 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY Anal. Proc. had for several years been the Pharmaceutical Society’s nominated member on the Committee of Management; lie was retiring from the Pharmaceutical Society but had accepted an invitation to remain on the Committee as a co-opted member. Group members joined in congratulating Dr. Garratt. Mr. Jolly reported that thz Committce of Management for 1980 would be : Chaivnzaiz- hlr. H. E. Brookes. Nonovav-y Secyetuv-v- Miss I. Ladden. Immediate Past Chaivman- hfr. E. Addison. Ovdinnvy Members of thc Committee-Dr. A. H. Andrews, Dr. D. H. Calam, Dr. L. E. Colcs, Mr. J. C. Deavin, I l r . D. C. Garratt, Mr. A. J. Middleton and Mr. I?. \V. Webb. Meiizbevs Nominated by Spoizsori.ng Bodies-&lr. S. C. Jolly (Pharmaceutical Society of Great Britain), Mr. W. H. C. Shaw (Chcmicnl Society, Analytical Division) and Professor W. B. Whalley (Royal Institute of Chemistry). In the afternoon a Discussion hlccting was held on “Some Aspects of Microbial Control in Pharmaceuticals,” a t which the speakers were Mr. J. Emerson (Fisons Ltd.) on Recent Develop- ments Associated with the Limulus Test, Dr. A. L. Davison (Glaxo Operations) on Bacterial Limit Testing of Non-sterile Products and Mr. R. Feakes (Travenol Laboratories) on the F,, Concept-Theory and Practical Applica- tion.

ISSN:0144-557X    

DOI:10.1039/AP980170108b

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

110 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY Anal. Proc. Mass Spectroscopy Coupled with Chromatography The following are summaries of the two papers presented at a Meeting of the Scottish Region held on October 16th, 1979, at Heriot-Watt University, Edinburgh. Combined High-performance Liquid Chromatography - Mass Spectrometry D. E. Games Deeartment of Chemistry, University College, P.O. Box 78, Cardiff, CF1 1XL In recent years high-performance liquid chromatography (HPLC) has become an increasingly important technique for the qualitative and quantitative analysis of compounds which are thermally labile and/or of low volatility, and hence which cannot be directly handled by gas chromatography (GC). The desirability of direct combination of HPLC with a mass spectro- meter (LC - MS) has provoked considerable discussi0n.l One school of thought advocates the collection of HPLC fractions off-line, followed by mass spectral measurements in aconventional or semi-automatic way.The alternative entails the direct coupling of the HPLC to the mass spectrometer (LC - MS), which would enable measurements similar to those currently obtain- able by GC - MS to be produced. Arguments in favour of the first approach are primarily based on the fact that it is much easier to collect HPLC fractions than GC fractions. By using this semi-automatic method there would be no restriction on HPLC solvent systems, and it has been suggested that most of the compounds studied by HPLC are not amenable to study by the two most commonly utilised mass spectral ionisation methods, electron impact (EI) and chemical ionisation (CI).In the author’s opinion these arguments are ill-conceived; while it is true that at the present time LC - MS is not capable of handling all types of solvent system and all organic compounds, as will become apparent in this paper, current LC - MS systems are capable of handling a very large proportion of the studies currently carried out by HPLC. The main reasons for develop- ing LC - MS are as follows. Firstly, on-line LC - MS would result in considerable time saving in obtaining mass spectral data. Quantitative measurements and identification using selected ion monitoring (SIM) techniques can be performed. Also, if a data system is available un- resolved peaks in the HPLC chromatogram can be partially or fully resolved by using al- gorithms of the Biemann - Biller type.Secondly, LC - MS would provide a universal detection system for the HPLC. Thirdly, complex mixtures, e.g., biological fluids, petroleum fractions and crude natural product extracts, where collection is clearly impracticable, could be studied. Lastly, fast eluting and sharp narrow peaks, which present major problems in collection,2 could be studied.April, 1980 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY 111 Off -line and Discontinuous Sampling Off-line LC - MS methods have been reviewed3 and a semi-automated method of collection of samples has been rep~rted.~ This approach is currently the only feasible method for hand- ling compounds which only yield useful mass spectral data with field desorption (FD),5 laser FD,6 californium-252 de~orption,~ desorption chemical ioni~ation,*~~ laser desorption mass spectrometry10 or other technique+ for handling involatile and thermally unstable molecules.It should be pointed out that considerable confusion currently exists as to what are difficult compounds, i.e., many of the compounds studied to date by off-line LC - FDMS12-16 are amenable to study by current LC - MS systems. An alternative discontinuous sampling approach making use of an automated probe has been deve1oped.l' The system consists of a hollowed out direct-insertion probe, at the tip of which is a gold gauze or absorbent material. The probe is directly connected to the HPLC via a capillaryftube and a semi-automatic system is used for introduction into the ion source.A major problem with the system is the 3-5 min cycle time between samples, which necessitates interruption of the chromatography if fairly closely eluting peaks are being studied, resulting in poor chromatography. On-line LC - MS Two fundamental characteristics of HPLC make this technique basically incompatible with the mass spectrometer. Firstly, the volume of mobile phase is much greater than can be handled by conventional mass spectral pumping systems. Typical HPLC separations with flow-rates of 1 ml min-1 result in gas volumes in the range 150-1 200atmmlmin-1, depending on the solvents used. However, a mass spectrometer, if configured for CI, can only handle up to 20 atm ml min-1. Hence, any LC - MS interface will have to provide a large solute - solvent enrichment if all of the HPLC eluate is to be analysed, or alternative types of HPLC with low flow-rates, e.g., micro-columns, will have to be utilised.Secondly, many of the types of sample analysed by use of HPLC present problems in obtaining good mass spectral data. Therefore, in designing an interface particular attention will have to be paid to the form in which the sample is introduced into the source and the ionisation methods used. The various approaches to on-line LC - MS have been the subject of a number of review articles.2s18-21 At the present time the main criteria that would be acceptable for an LC - MS system are : maintenance of chromatographic performance ; no restriction on HPLC systems; capability of giving at least EI and CI spectra without sample decomposition; reliability; sensitivity equivalent to that obtainable by GC - MS; and reasonable cost.At the present time no one system meets all of these criteria but a number closely approach these requirements. In the longer term other criteria can be added, particularly the ability to provide mass spectral data from compounds which do not provide useful EI or CI spectra. However, here other factors become important, e.g., the mass range of the mass spectrometer and the use of other methods of mass spectral ionisation with on-line LC - MS. Current approaches to LC - MS can be divided into four basic types: direct introduction of all the eluate from the HPLC into the mass spectrometer; splitting of the eluate from the HPLC and direct introduction of a portion of it into the mass spectrometer ; enrichment of the solute relative to the solvent of the HPLC eluate, in an interface prior to mass spectral ionisa- tion; and the removal of solvent from the HPLC eluate by using a mechanical transport system. The relative merits and problems encountered in these various approaches will now be discussed.Direct Introduction of HPLC Eluate With this technique the entire HPLC eluate is fed into an atmospheric pressure ion source (API) where it is vaporised. Ion production is effected by nickel-63 or a corona discharge and the ions resulting from solvent-mediated ion - molecule reactions are sampled through a 25-pm pinhole into a quadrupole mass analyser.2s26 Alternatively, a plasma chromatograph can be used.27 The former approach was shown to be effective for the identification and detection of a number of relatively volatile compounds, including drugs, steroids and poly- nuclear hydrocarbons.However, it has not been further developed, probably because of the following inherent problems involved in this approach. Firstly, samples have to be112 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY Anal. PYOC. vaporised at atmospheric pressure. Thus, the sample vapour pressure requirements must be substantially higher than those of normal direct probe introduction into the mass spectro- meter so that there are severe limitations on handling compounds with low volatility. Secondly, usually only quasi-molecular ion information is provided, severely limiting the amount of structural information available.Thirdly, whilst the technique exhibits high sensitivities for compounds with high electron affinities, it is insensitive for other types of compound. Fourthly, because of the high pressure in the API source polar solvents tend to cluster extensively yielding (solvent), H+ ions where n = 1-5; therefore, low relative mole- cular mass solutes are difficult to characterise. Direct Introduction of a Portion of the HPLC Eluate In this approach a fraction of the HPLC eluate is fed directly into the mass spectrometer. The basis for this approach was laid by Tal'roze and co-w~rkers,~*-~~ who conducted a series of experiments and calculations to devise the most effective method of introducing solvents into a mass spectrometer via a capillary tube.McLafferty recognised that if the mass spectrometer were to be operated under CI conditions then much higher source pressures could be tolerated and hence eluate could be introduced at a higher rate. In this approach up to 1 yo of the eluate from the HPLC was introduced directly into the ion source of a magnetic mass spectrometer fitted with a high-capacity pumping system. The solvent is used as a CI reagent gas to yield CI spectra from the sol~te.3~-36 This type of approach should be able to utilise the effect obtained when a solute - solvent mixture is sprayed into a mass spectrometer through an orifice of approximately 5 pm ; CI ionisation of molecules with insufficient vapour pressure for direct probe study can be effected.3' Unfortunately, to the author's knowledge, the effect has not yet been demonstrated with on-line LC - MS.These initial studies have recently been extended to the use of a quadrupole mass spectrometer, where the possibility of high-voltage breakdown in the source, if too high a flow-rate of solvent is fed into the ion source, is reduced. The approaches have varied con- siderably in complexity. Henion, in the simplest approach, successfully interfaced an HPLC with an unmodified commercial instr~ment3~7~~ and has recently demonstrated impressive sensitivities when micro-HPLC columns are u t i l i ~ e d , ~ ~ obviating the necessity for splitting the HPLC eluate. A more complex approach, again utilising a quadrupole instrument with enhanced cryogenic pumping, has also been reported,2 and this system is currently being marketed by the French company Ribermag. A major problem in these approaches has been an inability to handle low-volatility solutes.This problem has recently been overcome by using a cooling system in the capillary interfaceg1 and this approach is also commercially available, being marketed by Hewlett-Packard. Data published to date show, by the wide variety of compounds that are amenable to study with these systems, that this is a suitable approach to LC - MS.2731-41 Its major advantage appears to be an ability to handle eluates containing high proportions (>70%) of water. The disadvantages over other systems are : an inability to handle involatile buffers ; the restriction of mass spectral data to CI, the CI being obtained with the eluting solvent (this can cause a restriction in the choice of eluting solvent as not all are equally efficient CI reagent gases) ; only compounds with relative molecular masses in excess of 150 can be studied; the sensitivity is reduced unless micro-columns are used for HPLC, because only a fraction of the eluate is passed into the mass spectrometer; blocking of the capillary, or the small hole that is used to spray the eluate into the mass spectrometer ion source, occurs ; and if cryogenic pumping is used, de-icing of the pumps must place limitations upon the length of time for which the system can be used. In spite of these limitations this approach presents a working LC - PI'IS system that is capable of assisting in the solution of many currently time-consuming and intractable problems.These studies were confined to EI. Direct Introduction with Enrichment Approaches of this type make use of a variety of techniques to effect enrichment of the solute relative to the solvent prior to ionisation. Jones and Yangg2 used a semi-permeable, silicone membrane LC - MS interface to effect the selective diffusion of a non-polar solute from a polar solvent. Unfortunately, the bulk of com- pounds studied by means of HPLC are polar; a second limitation is that the system only works for compounds that are volatile up to 250 "C.April, 1980 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY 113 A second approach based on GC - MS interfaces utilised a one-stage glass-jet interface. The interface was used in combination with micro-column HPLC ; however, enrichment was not good enough to permit the recording of EI spectra and only solvent-induced CI spectra were obtained.43 A major problem was that compounds of low volatility could not be handled.These problems have recently been overcome by use of a vacuum nebulising interface, which can handle more difficult compounds and higher solvent fl~w-rates.~~ Again, solvent-induced CI spectra are obtained. At present it is difficult to make a reasoned assessment of this system as little on-line LC - MS work has been reported. Two more complex approaches have utilised fast pumping systems in order to generate a free expanding jet from solvent flow-rates of the order of 1 ml min-l. The eluate from theHPLCis vaporised by laser45 or sonic radiations46 or, more recently, by a form of direct heating.Sol- vent is discarded in a nozzle - skimmer system, leaving a well collimated beam of solute molecules. Satisfactory enrichment is achieved and both EI and CI spectra have been reported. However, more recent work has been carried out only with CI. Non-volatile buffers can be tolerated for short periods of time and the system is capable of handling rela- tively involatile molecules. At present the design is very expensive, but it appears that con- siderable simplification is possible and that a commercially viable system may be possible based on this approach. By using the same principle and similar equipment a dense gas chromato- graph has recently been successfully coupled to a mass ~pectrometer.~' Mechanical Transfer Based on his design for a moving wire HPLC detector Scottg8 designed an LC - MS moving wire interface in which solvent was removed in vacuum-lock interface chambers; solute was then flash vaporised into the ion source of the mass spectrometer, where EI or CI spectra could be obtained.A major limitation was that less than 1% of sample eluted from the HPLC was transferred into the mass spectrometer. McFadden et aLP9 provided increased sample utilisation by transporting the eluate on a belt. Initially metal belts were utilised, but more recently polyimide belts of Kapton have been used and transfer yields in the 25-40% range have been obtained.50 The amount of HPLC eluate allowed on to the belt is controlled by a low dead volume splitter. Some solvent is removed by an infrared heater, the remainder being removed in two vacuum locks.Flash vaporisation of the sample occurs when the belt reaches a point behind the ion source and the vapour passes into the ionisation chamber of the mass spectrometer, where EI or CI takes place. From res~lts~l-~6 obtained in the author's laboratory with this interface, it appears that the processes involved in the flash vaporisation are similar to those occurring in the technique of rapid heating from an inert PTFE surface, which has been used to minimise thermal decomposition in studies of underiva- tised peptides.5' In our experience the interface out-performs most conventional direct- insertion probes. The system exhibits good retention of chromatographic integrity, can be used in both EI and CI modes with no restriction on CI reagent gases, is reliable and can handle volatile and involatile buffers and gradient elution systems.It has been stated that EI data cannot be obtained with reversed-phase systems2; in the author's experience this statement is completely unfounded. Studies of a wide variety of compounds have been performed in Cardiff with reversed-phase systems and excellent EI and CI mass spectral data obtained.51-66 At the present time limitations of the system are as follows. Firstly, only compounds with relative molecular masses in excess of 120 can be studied on-line. There is no such limitation when solutions of samples are spotted on to the belt and the system is used as an automated probe. Secondly, solvent systems with a high percentage of water (in excess of 600/,,) give poor results.Small droplets of liquid form on the belt, instead of the normal continuous film of solvent, leading to severe fluctuations in the vacuum in the ion source. Thirdly, sensitivity is not as good as with GC - MS; and lastly, there is a sample volatility limitation with the system. Its ability to handle compounds appears to fall between direct probe and desorption chemical ionisation. However, the system would appear to lend itself well to modification for other methods of ionisation, e.g., laser, SIMS or sputtering techniques. A number of workers have purchased such systems from Finnigan Instruments and most systems appear to be working well. Recently VG Organic have introduced a similar system, but as yet it is not in customers' laboratories so cannot be fully assessed.A clean-up heater removes any residual sample.114 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY Anal. Proc. A Finnigan system has been in the author’s laboratory for the past two and a half years and a wide range of studies have been carried out with it. Recently mono- and disaccharides, glycosides and underivatised dipeptides, glucuronides and bile acid conjugates have been studied by on-line LC - MS. Many other classes of compound have also been studied in the author’s50--56 and other laboratories. 21 ,48 ,49 ,58-62 An alternative approach of this type has been reported, which overcomes some of the surface problems.W,6* The system consists of a continuous stainless-steel belt, perforated with small holes that enable solvent to be entrained. Solvent is removed in an evaporator and the solute that remains is transported to a reactor, where it is converted into hydrocarbons by catalytic reaction with hydrogen.The hydrocarbons that are formed are characteristic of the compounds being eluted and are swept into the ion source of the mass spectrometer operating in the CI mode. Impressive analyses of complex lipid mixtures have been reported with the system. Its major disadvantage is that conventional EI and/or CI spectra are not obtained, and although the technique is extremely useful as a liquid-chromatographic detector, and has some uses in the lipid area, it is unlikely to have widespread applicability. A further interesting approach, which may be capable of handling more difficult compounds, has recently been reported.65 The equipment consists of a conveyor belt that has been pre- treated by oxidation or with alkali metal atoms.The belt is constructed of a high work function metal and ionisation is effected by use of a lo5 V cm-1 field. Unfortunately, no examples of compounds run or LC - MS studies are given, so that one is unable to assess the usefulness of this approach. Conclusions While a perfect system has not yet been developed the current systems enable a wide range of HPLC problems to be solved. Future developments are likely to lead to improved sensi- tivity, and the introduction of alternative methods of ionisation. However, an LC - MS system that is capable of handling the full range of compounds studied by HPLC, and all of the types of solvent systems utilised, is not likely to be with us for some years.It is important that workers intending to enter this area should make a reasoned assessment of their current usage of HPLC and the types of compound involved. In the author’s opinion, in a large number of instances they will find current systems capable of providing considerable assistance in the solution of their problems, at a modest cost, if they already have a mass spectrometer or are considering purchasing one. I am indebted to Miss C. Eckers, Dr. J. L. Gower, Dr. P. Hirter, Mrs. M. E. Knight, Mr. E. Lewis, Dr. I. A. S. Lewis, Dr. M. G. Lee, Dr. K. R. N. Rao, Mr. M. Rossiter, Miss U. Trub, Mr. N. C. A. Weerasinghe and my wife, Dr.M. L. Games, without whom my studies in this area over the past two and a half years would not have been possible. We are all most grateful to our many friends in industry who have helped to make our studies possible. At present, with four commercial systems available, LC - MS is a viable technique. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References West, A. R., Editor, “Advances in Mass Spectrometry,” Volume 7, Heyden, London, 1978, p. 949. Arpino, P. J., and Guiochon, G., A n d . Chem., 1979, 51, 682A. Huber. J . F. K., Van Urk-Schoen, A. M., and Sieswerda, G. B., 2. Anal. Chem., 1973, 264, 257. Elbert, S., Gruhn, B., Wipfelder, E., and Heusinger, H., Anal. Chem., 1976, 48, 1270. Beckey, H. D., “Principles of Field Ionization and Field Desorption Mass Spectrometry,” Pergamon Schulten, H.R., Lehmann, W. D., and Haaks, D., Org. Mass Spectrom., 1978, 13, 361. McFarlane, R. D., and Torgerson, D. F., Science, 1976, 191, 920. Hunt, D. F., Shabanowitz, J., Botz, F. K., and Brent, D. A., Anal. Chem., 1977, 49, 1160. Rapp, U., Dielmann, G., Games, D. E., Gower, J. L., and Lewis, E., in Quayle, A., Editor, “Advances Posthumus, M. A., Kristemaker, P. G., Meuzelaar, H. L. C., and de Brauw, M. C. T. N., Anal. Chem., Daves, G. D., Jr., Acc. Clzem. Res., 1979, in the press. Schulten, H. R., and Beckey, H. D., J. Chromatogr., 1973, 83, 315. Evans, N., Games, D. E., Jackson, A. H., and Math, S. A., J. Chromatogr., 1975, 115, 325. Majors, R. E., Wilson, B., Greenwood, H., and Snedden, W., Biochem. SOC. Trans., 1975, 3, 867.Cole, D. L., Leonard, N. J., and Cook, J. C., Jr., in “Recent Developments in Oligonucleotide Syn- thesis and Chemistry of Minor Bases of t-RNA,” Universytet Im. Adama Mickiewicza Press, Poland, 1975, p. 153. Press, Oxford, 1977. in Mass Spectrometry,” Volume 8, Heyden, London, 1980, in the press. 1978, 50, 985.April, 1980 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY 115 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. DeFilippi, L. J., and Hultquist, D. E., Biochim. Biophys. Acta, 1977, 498, 395. Lovins, R. E., Ellis, S. R., Tolbert, G. D., and McKinney, C. R., Anal. Chem., 1973, 45, 1553. Scott, R. P. W., in Weissberger, A., Editor, “Contemporary Liquid Chromatography,” Volume XI of Dawkins, B.G., and McLafferty, F. W., J. Chromatogr. Sci., 1978, 9, 259. Kendler, E., and Schmid, E. R., in Huber, J. F. K., Editor, “Instrumentation for High Performance McFadden, W. H., J. Chromatogr. Sci., 1979, 17, 2. Carroll, D. I., Dzidic, I., Stillwell, R. N., Haegele, K. D., and Horning, E. C., Anal. Chem., 1975, 47, Horning, E. C., Carroll, D. I., Dzidic, I., Haegele, K. D., Horning, M. G., and Stillwell, R. N., J . Horning, E. C., Carroll, D. I., Dzidic, I., Haegele, K. D., Horning, M. G., and Stillwell, R. N., J . Horning, E. C., Carroll, D. I., Dzidic, I., Haegele, K. D., Lin, Shen-Nau, Oertli, C. U., and Stillwell, Horning, E. C., Carroll, D. I., Dzidic, I., and Stillwell, R. N., Pure Appl.Chem., 1978, 50, 113. Karasek, F. W., and Denney, D. W., Anal. Lett., 1973, 6, 993. Tal’roze, V. L., Karpov, G. V., Gorodetskii, I. G., and Skurat, V. E., J . Phys. Chem. (U.S.S.R.), 1968, 42, 1658, 1664. Tal’roze, V. L., Karpov, G. V., Gorodetskii, I. G., and Skurat, V. E., J . Phys. Chem. (U.S.S.R.), 1969, 43, 198. Tal’roze, V. L., Grishin, V. D., Skurat, V. E., and Tantsyrev, G. D., in Ogata, K., and Hayakawa, T., Editors, “Recent Developments in Mass Spectrometry,” University Park Press, Baltimore, 1970, p. 1218. “Techniques of Chemistry,” Wiley-Interscience, New York, 1976, p. 273. Liquid Chromatography,” Elsevier, Amsterdam, 1978, p. 163. 2369. Chromatogr., 1974, 99, 13. Chromatogr. Sci., 1974, 12, 725. R. N., Clin. Chem., 1977, 23, 13. Baldwin, M. A., and McLafferty, F.W., Org. Mass Spectrom., 1973, 7, 1111. Arpino, P., Baldwin, M. A., and McLafferty, F. W., Biomed. Mass Spectrom., 1974, 1, 80. Arpino, P. J., Dawkins, B. G., and McLafferty, F. W., J. Chromatogr. Sci., 1974, 12, 574. McLafferty, F. W., Knutti, R., Venkataraghavan, R., Arpino, P. J., and Dawkins, B. G., Anal. Chem., McLafferty, F. W., and Dawkins, B. G., Trans. Biochem. Soc., 1975, 3, 856. Dawkins, B. G., Arpino, P. J., and McLafferty, F. W., Biomed. Mass Spectrom., 1978, 5, 1. Baldwin, M. A., and McLafferty, F. W., Org. Mass Spectrom., 1973, 7, 1353. Henion, J. D., in Daly, N., Editor, “Advances in Mass Spectrometry,” Volume 7, Heyden, London, Henion, J. D., Anal. Chem., 1978, 50, 1687. Henion, J. D., “26th Annual Conference on Mass Spectrometry and Allied Topics,” St.Louis, Mo., Serum, J. W., and Melera, A., “26th Annual Conference on Mass Spectrometry and Allied Topics,” St. Jones, P. R., and Yang, S. K., Anal. Chem., 1975, 47, 1000. Takeuchi, T., Hirata, Y., and Okumura, Y., Anal. Chem., 1978, 50, 659. Tsuge, S., Hirata, Y., and Takeuchi, T., Anal. Chem., 1979, 51, 166. Blakely, C. R., McAdams, M. J., and Vestal, M. L., J. Chromatogr., 1978, 158, 261. Udseth, H. R., Orth, R. G., and Futrell, J. H., “26th Annual Conference on Mass Spectrometry and Randall, L. G., and Wahrhaftig, A. L., Anal. Chem., 1978, 50, 1705. Scott, R. P. W., Scott, C. G., Munroe, M., and Hess, J., J. Chromatogr., 1974, 99, 395. McFadden, W. H., Schwartz, H. L., and Evans, S., J. Chromatogr., 1976, 122, 389. McFadden, W. H., Bradford, D.C., Games, D. E., and Gower, J. L., Am. Lab., 1977, 55. Games, D. E., Gower, J. L., Lee, M. G., Lewis, I. A. S., Pugh, M. E., and Rossiter, M., in Reid, E., Editor, “Blood Drugs and Other Analytical Challenges,” Ellis Horwood, Chichester, 1978, p. 185. Games, D. E., Gower, J. L., Lee, M. G., Lewis, I. A. S., Pugh, M. E., and Rossiter, M., Proc. Anal. Div. Chem. Soc., 1978, 15, 101. Games, D. E., Eckers, C., Gower, J. L., Hirter, P., Knight, M. E., Lewis, E., Rao, K. R. N., and Weerasinghe, N. C. A., in Lawson, A., Editor, “Current Developments in the Clinical Applications of HPLC, GC and MS,” Academic Press, 1980, in the press. Games, D. E., Eckers, C., Lewis, E., Rao, K. R. N., Rossiter, M., and Weerasinghe, N. C. A., in Quayle, A., Editor, “Advances in Mass Spectrometry,” Volume 8, Heyden, London, 1980, in the press. Games, D.E., Lewis, E., Haskins, N. J., and Waddell, K. A., in Quayle, A., Editor, “Advances in Mass Spectrometry,” Volume 8, Heyden, London, 1980, in the press. Rapp, U., Dielmann, G., Games, D. E., Gower, J. L., and Lewis, E., in Quayle, A., Editor, “Advances in Mass Spectrometry,” Volume 8, Heyden, London, 1980, in the press. Beuhler, R. J., Flanigan, E., Green, L. J., and Friedman, L., J . Am. Chem. Soc., 1974, 96, 3990. Dymerski, P., Banard, D., and Kaminsky, L., “26th Annual Conference on Mass Spectrometry and Allied Topics,” St. Louis, Mo., U.S.A., 1978, American Society of Mass Spectrometry, p. 429. McFadden, W. H., Bradford, D. C., Eglinton, G., Haj Ibrehim, S., Dark, W. A., and Nicolaides, N., “26th Annual Conference on Mass Spectrometry and Allied Topics,” St.Louis, Mo., U.S.A., 1978, American Society of Mass Spectrometry, p. 422. Wright, L. H., and Oswald, E. O., “26th Annual Conference on Mass Spectrometry and Allied Topics,” St. Louis, Mo., U.S.A., 1978, American Society of Mass Spectrometry, p. 47. Dark, W. A., McFadden, W. H., and Bradford, D. L., J. Chromatogr. Sci., 1977, 15, 454. 1975, 47, 1503. 1978, p. 865. U.S.A., 1978, American Society of Mass Spectrometry, p. 420. Louis, Mo., U.S.A., 1978, American Society of Mass Spectrometry, p. 655. Allied Topics,” St. Louis, Mo., U.S.A., 1978, American Society of Mass Spectrometry, p. 659.116 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY Anal. PYOC. 62. 63. 64. 65. Dark, W. A., and McFadden, W.H., J . Chromatogr. Sci., 1978, 16, 289. Erdahl, W. L., and Privett, 0. S., Lipids, 1977, 12, 797. Privett, 0. S., and Erdahl, W. L., Chem. Phys. Lipids, 1978, 21, 361. Rrunnee, C., Franzen, F., and Meyer, S., Ger. Pat., 1978, 2654057 (Cl. GO1 N 31/08). Micropollutant Analysis by Gas Chromatography - Mass Spectrometry D. E. Wells Freshwater Fisheries Laboratory, Faskally, Pitlochry, Pevthshire The volume and diversity of information generated by the various types of GC - MS systems available have made it extremely difficult to obtain standard mass spectra for detailed com- parison from different types of instrument. Consequently, it has become almost mandatory to analyse a standard of the suspected pollutant alongside the sample, as opposed to making a comparison with reported information, and this demands reproducible instrument calibration. This calibration of the GC - MS instrumentation for quantitative analysis is often conducted on a rather ad hoc basis.Also, with the variations in instrumental conditions, such as the gas- chromatographic programme, column performance, type of interface and ion source and mass filter cleanliness, it is essential to calibrate the system by using a standard, reproducible procedure. This is particularly necessary with quadrupole instruments (e.g., the Finnigan 3200F Quadrupole GC - MS used in the work reported in this summary), where contaminated rods can cause precursor peaks and high mass discrimination, and excessive ion energy can cause “front-end” lift off of the mass peaks, particularly at the lower mass range (< 100 amu).Echelberger et al.1 have proposed the use of decafluorotriphenylphosphine (UM443) as a GC - MS calibrant for this purpose. However, for the initial mass spectrometer setting-up procedure it is much more convenient to have the calibrant bleeding into the instrument via a leak valve rather than via the gas chromatograph where repeated injections are necessary. Perfluorotributylamine (PFTBA) fills this role for the low-resolution instrument, although it suffers from low ion abundance above 220 amu and the usual intense peak of CF3+ at m/e 69. The resolution (R) is set across the mass range to within &5% for Am, at m/e 69-70, m/e 219- 220 and m/e 502-503, and the ion energy is limited by keeping m/e 70 to the same amplitude as m/e 502 using the programmable multiple ion monitor.This setting is then verified by obtain- ing a full spectrum. This is a suitable calibration for obtaining mass spectra, but for most analyses for organic pollutants the single (SIM) or multiple (MIM) ion monitor is used to achieve the desired sensi- tivity and selectivity. At these low concentrations the signal to noise ratio is very much en- hanced, with the electron energy just below the ionisation potential of the helium (24.6 eV),2 so that calibration, again with PFTBA, is best achieved by selecting the required electron energy (24 eV) and adjusting the ratios of m/e 70 to m/e 502 (0.1 : 1) using the same procedure as before. The ratio selected at the chosen electron energy then remains constant for each subsequent calibration.Once the mass spectrometer has been tuned the whole GC - MS system is checked for sensitivity and reproducibility by using a suitable calibration compound. The value of this calibration can be enhanced if the materials selected can also be used as internal standards (Table I). This is achieved on packed columns by injecting the calibrant into the gas chroma- tograph and adjusting the conditions for a constant retention time (e.g., R = 6 min for the Ultramark 443). A check is made on the reproducibility of the spectrum by comparison with duplicate and previous calibrations (Table 11) and the secsitivity and linearity of response are measured by using the programmable MIM on at least two ions (Fig. 1). After first checking for correct installation with no dead volume, by monitoring a methane inje~tion,~ a test procedure devel- oped by Grob et aL4 is used to give information on the flow-rate, temperature programme rate and column efficiency in terms of the TZ number and film thickness (WCOT), as well as on the adsorption of polar material.For qualitative analysis on capillary columns these tests, Similar tests can be made when a capillary column is used.April, 1980 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY TABLE I 117 INTERNAL STANDARDS FOR MICROPOLLUTANT ANALYSIS High-performance liquid Gas chromatography chromatography -7 - System detector MS ECD AFID UV Fluorescence - UM443 (decafluorotriphenylphosphine) + + + + 9-Bromoanthracene + + - + + + 9,lO-Dibromoanthracene + + Hexabromobenzene + + Decachlorobiphen yl 3- + - + + + - - - - TABLE I1 RELATIVE INTENSITIES FOR SELECTED IONS FROM THE SPECTRUM OF THE CALIBRANT ULTRAMARK 443 Relative intensities A f 7 mle 51 mle127 mle198 mle275 mle365 wale442 Proposed rangel 30-60 40-60 100 10-30 (1 >40 +I1 f 7 - +12 +20 zt5 Coefficient of variation (n = 5 ) 4.4 29 100 7.8 0.5 20 Mean value (70 eV) Coefficient of variation (n = 6) 56 50 100 12 1.2 41 Mean value (25 eV) -+60 f 1 4 - f38 f 2 0 4115 coupledlwith a spectrum of a calibrant, are sufficient.However, if quantitative work is con- ducted an internal standard is essential and this should elute as close as possible to the sample peak. A reproducibility test was made during a series of analyses for fenitrothion, monitored at m/e 127 with UM443 as the internal standard (m/e 125).The coefficient of variation for fenitrothion alone 'was &20.0%, n = 7, which was reduced to 47.6% when ratioed to the internal standard. -; = 208 5 8.7% 0 100 1000 10000 Mass weight on Ultramark 443 on column/pg Fig. 1. Calibration graph for Ultramark 443. Malachite green is a triphenylmethane dye, which can be used as a prophylactic or as a thera- A number of eggs taken from the hatchery In this peutic for fungal infection in fish and fish eggs. were believed to have been contaminated with the dye after flushing with clean water.118 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY Anal. Proc. instance the analytical problem was to separate the Malachite green from other interfering co- extracted fluids, but without the need to develop an extensive clean-up and separation pro- cedure.Malachite green as a salt [Fig. 2(a)] does not chromatograph on most columns. However, it is readily converted into its colourless leuco base [Fig. 2(b)] in alkaline solution, which can be chromatographed a t 300 "C on Dexsil300. 100 50 r C C Q) + + v) .- + .- .- - a a 100 50 300 200 m/e Fig. 2. EI spectrum of (a) Malachite green and (b) its leuco base. Electron energy, 70 eV. A sample of the eggs was macerated with 0.1 M potassium hydroxide in methanol, centrifuged and the centrifugate injected into the GC - MS. Under the severe chromatographic conditions much of the co-extracted material was pyrolysed at the injection port or on the early part of the column, and the leuco base, monitored at m/e 253 (M-CC,H,) and m/e 330 (M+) gave a chroma- togram free from interference (Fig.3). The similarity of the retention times and the ratio of the two ions monitored in the standard and sample indicated the presence of the leuco base of Fig. 3. Multiple ion monitoring of the leuco base of Malachite green: (a) extract from salmon eggs suspected of contamination with Malachite green; and (b) 50 ng of Malachite green as the leuco base. Column, 3% Dexsil 300 Chrom G, 100-120 mesh; temperature 300 "C (isothermal).April, 1980 MASS SPECTROSCOPY COUPLED WITH CHROMATOGRAPHY 119 Malachite green. With this gas-chromatographic treatment the performance of the column rapidly deteriorated, but replacement was much less costly than the development of another separation technique for a small number of samples.Eulan WA New, active ingredient polychloro[2-(chloromethylsulphonamido)] diphenyl ether (PCSD) has found increasing use in the woollen industry as an alternative mothproofing agent to dieldrin. The formulation is also known to contain the primary amines of the sul- phonamides as impurities (Fig. 4). The analysis of the PCSDs by GC - MS5 has been depend- ent upon the pyrolysis of the sulphonamide in the injection port. With a clean injector at 230 "C and a good column the chromatography was acceptable (coefficient of variation = &7%). However, with prolonged use of the column for the analysis of environmental samples the reproducibility deteriorated (coefficient of variation = &-20y0), reflecting the changing conditions of the injector and the column head.CI CI CI CI SO2 CH2 CI ? CI Fig. 4. Structure of the PCSD and PAD constituents of Eulan WA New. Derivatisation of the sulphonamide by replacing the active sulphonamino hydrogen was limited owing to steric hindrance, but by modifying a method6 developed for the extractive alkylation of sulphonamide drugs it was possible to methylate the PCSDs. This method uses a two-phase alkylation with aqueous tetrabutylammonium hydroxide as the counter ion and methyl iodide in hexane as the alkylating reagent. The reaction and back-extraction goes to 95% completion after shaking (for 30 min) at room temperature. Analysis of the reaction mixture indicated that the methyl derivative had been formed and chromatographed without decomposition. The mass spectrum (Fig. 5) gave a molecular cluster at m/e 481487, having an isotopic ratio indicative of six chlorine atoms. The cluster at m/e 358-374 (M-113) comes from the loss of the side chain, -SO,CHCI,, followed by the elimination of a chlorine atom to yield the base cluster at m/e 333-339. This stable, even 200 300 400 500 600 m/e Fig. 5. EI spectrum of the methyl derivative of the main active ingredient of Eulan WA New eluting from an OV-101 17m x 0.3mm i.d. capillary column. Electron energy, 70 eV.120 NEW NUMERICAL METHODS Anal. PYOC. electron fragment is probably the tricyclic ion, formed by the nitrogen bridging following the loss of the chlorine, in similar fashion to the N-chloromethyl pyrolysis product of the un- derivatised PCSD.5 Decachlorobiphenyl was used as the internal standard and was added prior to the derivatisa- tion (Fig. 6). The quantitative analysis of the derivative with the internal standard was much improved, with a within-batch coefficient of variation of &2.5%, n = 10, and a between- batch coefficient of variation of &-7.6%, n = 6, at a concentration of 0.5 mg 1-1. Fig. 6. Chromatogram of Eulan WA New and the methyl derivative obtained on an OV-101 17 m x 0.3 mm i d . capillary column. (1) and (3), Polychloro(chloromethy1- sulphonamide) diphenyl ether (PCSD) ; (2) and (4), polychloro 2-aminodiphenyl ether (PAD); (5) and (6), methyl derivative of (1) and (2); and IS, internal standard (de- cachlorobiphen yl) . References 1. 2. 3. 4. 5. 6. Eichelberger, J. N., Harris, L. E., and Budde, W. L., Anal. Chem., 1975, 47, 995. Wells, D. E., in Price, D. P., and Todd, J. F., Editors. “Dynamic Mass Spectrometry,” Volume 5, Jennings, W., “Gas Chromatography with Glass Capillary Columns,” Academic Press, London, 1978. Grob, K., Jr., Grob, G., and Grob, K., J . Chrornatogr., 1978, 156, 1. Wells, D. E., Anal. Chim. Acta, 1979, 104, 253. Gryllenhaal, O., Tjarnlund, U., Ehrisson, H.. and Hartvig, P., J . Ckromatogr., 1978, 156, 275. Heyden, London, 1978, pp. 85-105.

ISSN:0144-557X    

DOI:10.1039/AP9801700110

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

120 NEW NUMERICAL METHODS Anal. Proc. New Numerical Methods, Optimisation and Pattern Recog nit ion The following are summaries of six of the papers presented at the Analytical Division Sym- posium held at the Chemical Society Autumn Meeting on September 19th and ZOth, 1979, at the University of Lancaster. This Symposium was organised in conjunction with the Education and Training Group. Review of Statistical Methodology Applied to the Results of Chemical Analysis Irene M. Calus Department of Engineering Mathematics, University of Technology, Loughborough, Leicestershire, LE11 3T U Many problems have solutions that involve dealing with numbers, e.g., finding the values of x for which x2 + 3% + 2 = 0 or the dissociation pressure, T, of a chemical compound when 'E = 0.8 log, T + 3.In the first instance, simple algebra leads to exact answers, x = -1 T or -2. In the second, there is no exact answer but a successive approximation technique will give a value of T correct to any required number of decimal places. In either instance, an answer can be given of which we are 100% confident.A$ril, 1980 NEW NUMERICAL METHODS 121 Now suppose the problem is to find the iron content in a sample of material and five deter- minations made on it give: Iron, yo 13.99 14.15 14.28 13.93 14.30 The variation from one reading to another introduces an element of uncertainty; an answer cannot be stated with 100~o confidence. Where such variation is present, statistical methods are applicable. Reliability of an Estimate mean, 14.13. by another method give : Iron, Yo 14.12 14.10 14.15 14.11 14.17 These values also give a mean of 14.13, but show less scatter, suggesting greater precision in the second method and hence a more reliable estimate.Thus, instead of just stating an estimate, it would be more informative to indicate how reliable it is likely to be. In order to do this a mathematical model is needed, and some assumptions must be made, one of which is that the variation from one result to another is due to random error. Then, if there is no bias, each observation can be expressed as : Only an estimate of the percentage iron content can be given. This would usually be the Now, for the sake of argument, suppose that analyses of this sample of material 1 e Observation True value + Random error c - X The e-component arises from errors of judgement, slight fluctuations in experimental condi- tions, etc.For example, in the precipitation step of a gravimetric analysis, slight differences in the rate of addition of the reagent or in the speed of agitation are unavoidable. The average value of e in the long run (i.e., in an infinite number of measurements) will be zero. Hence, if an infinite number of measurements (in statistical terms, the population) were possible, their mean p would be the true value, i.e., p = c. If bias exists, as a result of faulty calibration, personal prejudices, etc., c b e -/- Random error - X Observation True value + Systematic error If b is constant, p = c + b. Conclusions made about p on the basis of, say, five measurements (a statistical sample, as distinct from a chemical sample) must take account of the likely effect of the e-components. This necessitates the assumption of a model that describes the behaviour pattern of e over-all, i.e., in the population.Many statistical procedures in common use depend on the assumption of the normal (Gaussian) model, although text-books with a practical emphasis (e.g., analytical chemistry texts) often neglect to mention it. This paper concentrates on variation caused by experimental error because of its importance to analytical chemists; the normal model may well then be suitable. However, where random variation has other causes, as in radioactive counts, the normal model may be inappropriate. In assessing the reliability of a sample mean as an estimate of the population mean, we must bear in mind all the samples that might have arisen from that population and consider the distribution of sample means.For random samples of size n : Population A r 3 Mean Standard deviation Individual observation x P cr Sample mean 3 P a/ 6 (standard error of the mean) The 2’s can often be taken to be normally distributed even when the x’s are not, as illustrated by a computer simu1ation.l It can then be said that there is a 95% chance that a value of 5 will not differ from p by more than 1.960/<n, i.e., that the interval 3 - 1.96 o/+i to Z + 1.96a/2/I;Z will contain p. This is the 95% conjidence interval for p. Intervals can beAnal. Proc. associated with any degree of confidence, e.g., 5 & 1.640/2/i-i for 90% confidence.Their calculation requires the value of 0. It may be known from past experience; some manufac- turers express their instruments’ error as a standard deviation. If not, 02 is estimated from the sample using s2 = - C ( X - ~ ) ~ and values from the appropriate Student’s t-model must then be substituted for 1.64, 1.96, etc. Note that the use of t here still assumes a normal model for the Z’s. For the (iron, yo) deter- minations this gives : 122 NEW NUMERICAL METHODS 1 n-1 When n = 5, the 95% confidence interval for p is I 2.78~145. 1st set of data: 14.13 & 0.207 i.e., 13.923 to 14.337 2nd set of data: 14.13 -+ 0.036 i.e., 14.094 to 14.166 A Test of Significance Sometimes the question may not concern the value of p, but whether it differs from a particular value.An example is the question of bias in a method of measurement, illustrated here by data reported by Florence and Farrar.2 When the amount of beryllium present was known to be 3.179 mg, ten repeat titrations using their new photometric method gave (in mg) : 3.167, 3.177, 3.177, 3.169, 3.173, 3.177, 3.177, 3.177, 3.171, 3.169 The mean of these values is 3.173. This question can be answered by using a signijcance test. A Student’s t-distribution shows a less than 1% chance of a sample mean being as far away from 3.179 as 3.173, i.e., the difference is signijcant at the 1% level, representing strong evidence of bias. The 95% confi- dence interval for the bias, b, of the beryllium method is -0.009 to -0.003, possibly negligible for practical purposes. In this respect a confidence interval is more informative than a significance test.Is there evidence of bias? We test the hypothesis that no bias exists, i.e., p = 3.179. A statistically significant difference may not be of practical significance. Control Charts Regarding the need for an ongoing check on the results being obtained from a chemical measurement process routinely used in a laboratory, Mande13 recommends running reference materials periodically and using Shewhart control charts in order to interpret the results. These charts are used to check that p and (T remain unchanged, i.e., that a state of statistical control exists. Only control charts for means will be considered here, but in practice an accompanying chart for standard deviation or range would be necessary.Fig. 1 shows the essential features. A point falling outside the control lines suggests a shift in population mean, an indication of bias if reference material is being analysed and the central line corresponds to the known concentration (e.g., nitrogen content of standard urea sample3). In another example of the use of control charts in the laborat~ry,~ the central line represented the value adopted for the water equivalent of a calorimeter. A monthly check was made by plotting a mean value obtained from four calibration runs. Sample means are routinely plotted. I Upper control line I @. Centre line lz 0 I @ Lower control line Fig. 1. A control chart.Afwil, 1980 NEW NUMERICAL METHODS 123 Shewhart placed each control line at a distance of 3a/@ from the central line.If sample means are normally distributed, that gives a 0.3% chance of a “false alarm,” i.e., of a non- existent change being signalled, but the risk of an actual change going undetected may be unacceptably high. This has led to some modifications, e.g., a pair of inner control lines that act as “warning limits.” Another development has been the cumdative sum (CUSUM) chart, found particularly suitable for use in the chemical process industry. Outliers The occurrence of a gross error (mistake or blunder) may result in an observation that is manifestly inconsistent with the other data, i.e., an outlier. Here attention will be confined to a single outlier in a univariate sample, illustrated by data from an IAEA seawater c~mparison.~ Levels of g5Zr-95Nb initially reported were 2, 10, 9, 24, 0.4, 84, 14 and 0.1 pCi kg-l. A quick glance suggests that one result, 84, is unusually far away from the rest.Whether this impression is justified can be decided by testing the hypothesis that the suspect observa- tion is from the same population as the others, i.e., by a test ~fdiscordancy.~ One such test uses the statistic T = -, x being the suspected outlier. A value of x unusually far away from 3 will give an unusually large value (positive or negative) for T. Critical values of T corresponding to various probability levels and sample size are available,’ based on the assumption of a normal population (not always mentioned). Ordering the observations so that x, < x2 < x3 < . . . < xn-, <x,, the suspected outlier must be either xl or x,.So far we have only considered the effect of random and systematic errors. X-Z S Alternatively, a Dixon ratio, denoted by Q or rlo, can be used as test statistic. For a lower outlier x,, Q(or rlo) For an upper outlier x,, Q(or rlo) = xn - xn-l xn - x, Reference to a table of critical values8 of Q will show whether the value of Q obtained is unusually high. A test of discordancy represents a detection, not necessarily a rejection, procedure. When it has confirmed the labelling of an observation as an outlier, the next stage is to seek an explanation. Firstly, the outlier arose from a gross error and should be rejected. Currie5 warns that it might not be the outlier that is at fault. The mercury content of fish muscle was given by six laboratories as 0.38, 0.37, 0.28, 0.66, 0.36 and 0.86 p.p.m.Although 0.86, and perhaps 0.66, appear discordant, chemical evidence suggested that the other results were low because of mercury loss. Secondly, an unusual combination of circumstances has arisen and further investigation may lead to new knowledge (as in the discovery of the element argon). Thirdly, the population model assumed in the test was wrong. Decisions about the rejection of outliers become less important if the statistical procedures used are resistant to their influence. For example, the effect that an outlier can have on a sample median is limited and there is a case for using this instead of the mean. Such a table assumes some model (usually the normal) for the population. Some possible responses then are as follows.References 1. 2. 3. 4. 5. 6. 7. 8. Bajpai, A. C., Calus, I. M., and Fairley, J. A., “Statistical Methods for Engineers and Scientists,” John Wiley, Chichester, 1978, p. 170. Florence, T. M., and Farrar, Y. J., Anal. Chem., 1963, 37, 712. Mandel, J., in Kolthoff, I. M., and Elving. P. J., Editors, “Treatise on Analytical Chemistry,” Part I, Volume 1, Second Edition, John Wiley, New York, 1978. Wernimont, G., in de Voe, J. R., Editor, “Validation of the Measurement Process,” ACS Symposium Series 63, 1977. Currie, L. A., in Kolthoff, I. M., and Elving, P. J., Editors, “Treatise on Analytical Chemistry,” Part I, Volume 1, Second Edition, John Wiley, New York, 1978. Barnett, V.. and Lewis, T., “Outliers in Statistical Data,” John Wiley, Chichester, 1978. Grubbs, F.E., and Beck, G., Technometrics, 1972, 14, 847. Dixon, W. J., Ann1 Math. Stat., 1951, 22, 73.124 NEW NUMERICAL METHODS Anal. Proc. Correlation and Regression Analysis : Applications to the Analysis of Chemical Data B. A. Moore Department of Mathematics, Loughborough University of Technology, Loughborough, Leicestershire, LE 11 3TU The problem of statistical relationship arises from an interest in the joint distribution of random variables; here we could be interested either in the interdependence between some of the variables or in the dependence of one or more variables upon others. In general, we can say that the study of interdependence leads to the investigation of correlations, while the study of dependence leads to the theory of regression.An interesting example of the former was provided by an entry in the 1975 finals of the BBC TV “Young Scientists of the Year’’ contest, in which the team from Aberdeen submitted a project exploring possible links between the contents of drinking water and health. Samples of tap water from major towns in the UK were obtained and analysed, and data relating to deaths from various diseases were examined. (Previous findings had indicated that coronary mortality rates were higher in soft water areas than in hard water areas.) A correlation matrix was formed using data obtained from 32 towns, the first six variables being concerned with the chemical anlayses carried out, and the remaining four representing male and female coronary and leukemia deaths.We note, however, that a measure of statistical relationship, such as a correlation coefficient, should never be used to deduce a causal connection; our ideas on causation must come from outside statistics. Evidence from an exercise such as that described above of a tendency for coronary deaths to increase as pH value decreases should not be claimed as establishing that low pH itself causes coronary disease, but could support any medical evidence that had been found for this. It is probably due to dissatisfaction with interpretational aspects that “correlation analysis” as such has lost its earlier popularity, but there are still areas where patterns of causation are insufficiently understood for a more structured approach to be used. This was used as the basis of a correlation analysis.Correlation and Regression variables is the correlation coefficient. are made on two random variables X and Y. defined by The most important measure of the strength of an association between two random Let us suppose that n pairs of measurements (xi,yi) Then, the sample correlation coefficient is where S,,, S,, are sums of squares about the mean for the two variables, and S,, is the corresponding sum of cross-products. Because an uncorrelated population may give rise to a sample showing correlation there is, for each sample size, a level that Y must reach in order to indicate a correlated population. This critical value of Y also depends on the risk that we are prepared to take of making an incorrect decision. The coefficient can usefully be used when the pattern of the plotted points (xi, yi) suggests an approximately linear relationship.Estimates of the model parameters are usually found by using the principle of least squares. The results of this are well known, and are not given here, but it is interesting to note that the subsequent analysis of variance breakdown can be expressed as The first term on the right-hand side gives the portion of the total variability in the response, Sww, which is explained by the fitted regression; the remaining term represents variation “about” the fitted line. Substituting the expression for r given above shows that these components can also be represented as Sz,/S,. and S,, - (S~y/Szz), which are convenient computational forms.April, 1980 NEW NUMERICAL METHODS 125 Application to Calibration Graphs Often, in analytical chemistry, quantitative analysis to determine an unknown concentra- tion is accomplished by means of a calibration graph that has previously been determined by using known samples.Responses can be optical, electrical or mechanical measurements, for example, and are usually related to concentration in an approximately linear manner. (Alternatively, a transformation can be used to convert the relationship into a linear form.) With the usual form of model where the residuals, E , are independently and normally distributed, let xo represent the concentration for an unknown sample, which is analysed to provide a value of the response, yo. Then, the required estimate of concentration for the sample is given by (yo - a)/b where a and b are the usual least-squares estimates of a and 18.The fact that the measurement yo has some implied uncertainty means that the estimate of concentration cannot bz precise, and this often leads to the calculation of a confidence interval for xo. If n points were used to fit the calibration graph, it is not difficult to show that for a 95% confidence interval we can solve the quadratic equation 1 (x, - ;F)2 X ( x - x)2 (yo - a - b ~ , ) ~ = t:-,(0.05)s2 for the only unknown, x,). [The value of .fz-2(o.05) is obtained from a table of Student's t-distribution and s2 = [S,, - S2,,/Szz]/(n - 2) is the usual estimate of residual variance.] The two solutions represent the two ends of the required interval. In chemical analysis it is sometimes necessary to compare calibration graphs in order to see whether any differences can be explained purely in terms of experimental error. To illustrate this, we discuss a published paper by Z&tka,l in which three methods of preparation of standard titanium(1V) solutions are compared in terms of their calibration graphs, each relating absorbance measurements ( y ) to the amount of titanium added (x).One set of published results can be summarised as follows: Standard solution 1 n, = 9 a, = 0.000 7 b, = 0.014487 S: = 0.71 x 10-o Standard solution 2 Standard solution 3 n2 = 9 n3 = 9 a, = 0.001 3 a3 = 0.001 2 b, = 0.014468 b, = 0.014 494 S: = 0.57 x lo-' S; = 0.29 x The problem is to decide whether the differences that exist between these three fitted lines can be explained as being solely due to sampling error.The author seeks to answer this point by comparing each pair of regression coefficients b, using the usual t-criterion. As none of these pairs yields a t-statistic exceeding the critical value obtained from tables, he concludes that all the graphs can be considered to be identical. (i) Do the a values differ, so that the three lines are not identical, but parallel? (ii) Can the variability about each line really be assumed to be the same, as is implicity assumed in the above t-tests? (iii) Is a comparison in pairs really valid, when we have more than two lines? In fact, a re-analysis of the data gave This, however, leaves unanswered a number of questions : S: = 0.419 x lo-', ~ 2 , = 0.413 x sf = 0.417 x so that the variability about each line can certainly be regarded as being the same.Com- paring intercepts and slopes in pairs again shows no significant differences so that, despite computational differences, we can agree with the author's findings that the three lines can be regarded as identical. Hence, it is indeed reasonable to choose one of the standards (ammo- nium titanyl oxalate) in preference to the others on such grounds as being simpler to prepare.126 NEW NUMERICAL METHODS Anal. Proc. The approach used so far, however, is appropriate to comparing a pair of lines. For comparing more than two lines, a different approach should be used, as otherwise we have difficulty in interpreting probabilities when apparently significant differences are found.We use the Z&tka data again to illustrate an alternative method, which avoids these logical difficulties. We seek to distinguish between three possible models, represented diagramatically as : X 4 Model I (includes interaction) x - Model II (additive effects) x + Model Ill (common line) It is, of course, the last of these models that corresponds to the case of identical calibration graphs. If we assume a model with three general lines, which can differ in both slope and intercept (or, in statistical terms, we allow for the presence of a group-concentration interaction com- ponent in the model), then the residual sum of squares will be given by 5 {Sg - [S2]2 = 8.741 9 x i - 1 and is on 21 degrees of freedom (as we have 27 data points, and 3 slopes and 3 intercepts have been estimated from the data).If we assume a model consisting of three lines with a common gradient, but possibly having different intercepts (again, in statistical terms this corresponds to an additive model, with no interaction term) then, noting that the best estimate of the common gradient is Z S ~ ~ / X S ~ , the residual sum of squares now becomes with 23 degrees of freedom. Finally, if we postulate that the 27 data points are scattered around a common line (i.e., the three lines have identical slopes and intercepts), then the residual sum of squares becomes s,, - - ”’ - - 10.591 4 x 10-6 s,, and carries 25 degrees of freedom.April, 1980 NEW NUMERICAL METHODS 127 Differences between the respective sums of squares correspond to differences in the model assumptions. The analysis comparing the three models can be set out formally in an analysis of variance table (Table I).TABLE I ANALYSIS OF VARIANCE Source of variation x 106 freedom x 106 Between groups (after 1.782 2 2 0.891 1 Sum of squares Degrees of Mean square adjusting for concentration) Interaction 0.067 3 2 0.033 6 Residual 8.741 9 21 0.416 3 Comparing the interaction mean square with that for the residual, we see that there is certainly no evidence of an interaction term, so we can assume at least that the three lines have equal gradients. Similarly, after comparing Models I1 and 111, we reject Model I1 and conclude that the three calibration graphs can be taken to be identical in all respects. Multiple Regression Models Correlated regressor variables can give rise to interpretational difficulties when we allow for the possibility of more than a single variable being required to explain the response.It is often useful to make a preliminary exploration of the relationships among the regressors, perhaps by determining the eigenvalues of their correlation matrix and then using a factor analysis approach, before attempting to fit the model. Orthogonal factors extracted from the correlation matrix can be used to identify a nearly orthogonal subset of regressors. Regression techniques can usefully be applied to designed experiments when it is required to optimise a process. Designs specifically intended for fitting planes and quadratic response surfaces can be used to determine the most appropriate approximation to the response in the experimental region.This fitted response surface can then be used to locate the level at which each of the process variables should be set in order to maximise the response. References 1 . 2. ZAtka, V., Analyst, 1970, 95, 47. Draper, N. R., and Smith, H., “Applied Regression Analysis,” John Wiley, New York, 1966. Numerical Optimisation of Kinetic Parameters by Unidimensional Search M. J. J. Holt, A. C. Norris,* M. I. Pope and M. Selwood Defiartment of Chemistry, Portsmouth Polytechnic, Portsmouth, PO1 2DT The introducton of laboratory computers has opened the way for more powerful methods of data handling and analysis. As the applications are so varied, however, the development of applications software is likely to remain the responsibility of the user rather than the manu- facturer.There is therefore a need to convey the principles of numerical methods to chemists in a way that avoids unnecessary mathematical rigour and allows them to see the relevance of the techniques to their own chemical problems. In this paper, we demonstrate the application of one such technique, the unidimensional search (UDS), to the optimisation of kinetic para- meters. Numerical Optimisation Many chemical phenomena are described by models that contain parameters or constants. The experimenter wishes to know the values of these constants that predict the results that he either observes or specifies. The optimum values are determined by an optimisation proced- * To whom correspondence should be addressed.128 NEW NUMERICAL METHODS Anal.Proc. urel that has two essential features: a criterion to judge the “best” values and a systematic strategy for finding them. If a measured response y , sub- ject to experimental error, is a function of a precisely-known variable x, then the least-squares principle attempts to find the value of the sum Q that minimises The criterion used most frequently is that of least squares.2 n 2 = 1 Q =.E [yi - f ( ~ i ) ] ~ ~ i . . .. .. * * (1) where there are n discrete data points (xi, yi) and y = f (x) is the functional relationship derived from the model; wi is a weighting factor3 such that wi = p/oii (i = 1, 2, . . ., n) where p is a constant and 4i is the statistical variance in the value of yi. When oii is sniall, wi is large and vice versa, so that the (wi) terms in equation (1) weight the sum in favour of the more precise data points.If the points are equally precise then the (wi) have the same constant value, which does not affect the minimisation of Q. The assumption of constant (wi) may be necessary if the variances are unknown, but the weighting factors should always be included if they are available (see later). The strategy for finding the unknown parameters, $j ( j = 1,2, . . ., m), of the model depends upon how the parameters appear in the function f(x). In the simplest cases, they occur in a linear way and a linear, least-squares (LLS) analysis can be applied directly. In contrast, non-linear optimisation meth0dsl9~ are needed when there are many non-linear parameters. Frequently, however, there are not more than three parameters, at least one of which occurs linearly.In these cases, the function f(x) can often be transformed to give a new response variable Y such that the parameters are obtained from an LLS analysis using the (xi, Yi) data. If the variable Y does not involve an unknown parameter, then the linear analysis can be applied directly, but if it involves a parameter$, then a further stage is necessary. Following transformation, therefore, an initial value $k(0) is chosen and used to find the values of the remaining (LLS) parameters of the transformed function F(x) that minimise n $k is then varied systematically to give values t&(l), $k(2), . . ., etc., and the linear analysis is repeated at each step to find the value $; that yields the minimum sum of squares Q($Z).The parameter values corresponding to this minimum are then the optimum values. As the procedure seeks a minimum as a function of a single parameter $k it is known as a unidimen- sional (UDS) search. The UDS method can be implemented in various ways, but one of the most efficient, due to Davies, Swann and Campey,ly4 chooses a fractional increment (or decrement) 8 of C$,~(O) and calculates Q with the values $p = +p) + (2‘ - 1)s (Y = 0, 1, 2, . . .) until the minimum in Q is bracketed. quadratic in the region of the minimum and the value Q(fk) is found by interpolation. parameters can then be refined by re-starting the search with a smaller value of 8. details of the algorithm are given in refs. 1, 4 and 5. tion of kinetic parameters.The sum of squares function Q(&) is then assumed to be The Full In the following section, we demonstrate the application of the UDS method to the optimisa- UDS Applications in Chemical Kinetics Homogeneous, First-order Reactions property h (=y) rather than concentration such that k In an isothermal study of the elementary reaction A-tproducts, we usually measure some “0 Gc I ho - &!3 I [A], I hi i- IApril, 1980 NEW NUMERICAL METHODS 129 Here the suffixes 0, i and ca refer to the values at the start, time ti and the end of the reaction, respectively. The integrated rate equation in terms of h is hi = Am + (A, - hw)e-kti (i = 1, 2, . . ., n) . . .. .. (2) Provided that A, is known accurately, we can determine the rate constant k and the value h, by rearranging equation (2) to carry out an LLS analysis with the transformed variable Yi given by This analysis cannot be used directly, however, if A, is highly inaccurate or unknown.The data could be subjected to a constant time interval method6 but such techniques place con- straints upon experimental procedure and magnify error due to the way in which they com- bine data.7 A better approach,* therefore, is to use the UDS method to search for the optimum value A,* that minimises Yi (A,) = In ( I &-A, 1) = 1 n (Ih, - LJ I) - kti where k and A, are found by the LLS method. gives the weighting factor In equation (3), the propagation of error in hi wi = p/a,; = p/{[aY,(h,)/ahi]Ohi)2 = p(xi-h,)2/oAT; Notice that we cannot ignore the weighting factors even though the values of identical.observed value of A, is clearly inconsistent with the earlier data. may be The UDS method is particularly valuable when the reaction has a large half-life or the Homogeneous, Second-order Reactions To illustrate the application of the UDS approach to second-order kinetics, we consider the k process A + B-tproducts, which is followed by measuring the property h such that [Bl,/[Al, = y > 1 “ 0 = QO(0 4 0 ) ; [BIO = QY(h0 - A W ) [AIi = q(hi - A,); [BIi = Q [y(AO - - (A, - &)I where q is a calibration constant. The integrated rate equation in terms of h is Xi = A, + (y - 1) (4 - Xoo)/(yek’2t - 1) . . . . - * (4) where the pseudo-first-order rate constant k’ = k(y-1) [A],. Suppose that the value of A, is measured accurately but the initial concentrations are not available and y and A, are unknown.We rearrange equation (4) to give the transformed function Yi(8) = ln[(8-hi)/(A, - hi)] = lny - k’ti where 8 = A, + ?(Am - A,). optimum value 8* and the consistent (LLS) parameters y* and k’* that minimise The UDS method is then used to search over 6 and find the n i = 1 Q(0) = C [Yi(8) - l n y + k’ti12wi where wi = Y [(e - hi) (A, - hi)/(e - A,)I~/$~ As y* is now known, can be found from the definition of 8. Independent measurements with just A, B or product allow q, and hence [A],, and finally k to be found. The same approach can be used to determine any three of the parameters A,, A,, y and k knowing the value of only one of them.130 NEW NUMERICAL METHODS And. PYOC. Isothermal, Solid-state Decomposition Reactions For the decomposition solid,+solid, + gas, the reaction is readily followed by thermogravi- metry, i.e., the property A is the total mass of solid , + solid, remaining at a given time.When the decomposition follows a contracting interface mechanism, the integrated rate equation has the form5 hi = A, + (A, - A,) [l - (1 - n)K’t,] lI(1-n) . , .. * (5) where n is the so-called “order of reaction.” formed variable = (A,, - Rearrangement of equation (5) gives the trans- yi = (Ai - - (A,, - Aa)l-n (1 - n)& and an LLS analysis with the (ti, Yi) data gives k and A,, directly, provided that A, and n are known. For most solid-state reactions, however, the value of n is fractional and cannot be predicted theoretically. When A, is available, therefore, the optimum values of k and A, can be estimated by a UDS search over n, coupled with a linear, least-squares analysis with the [ti, Y , ( n ) ] data to minimise n r = l Q(n) = -Z [Yi(n) - (A, - Am)l-n + (A, - A,)l-n (1 - n)kti]2wi where wi = p(Ai - A,)2n/[(l - n)2~A:] Non-isothermal, Solid-state Decomposition Reactions Kinetic parameters for solid-state decompositions are often obtained from experiments in which the temperature of the sample is varied at a constant rate p.Thennogravimetry is used to determine the total mass A of the remaining solid and the theory leads to the relation- shipQ f(&) = f(h,) + (l//3) Jrk(T)dT . . .. .. (6) where f(h) is a (frequently empirical) function of h determined by the mechanism. This function has the values f(&) at zero temperature (no decomposition) and f(Ai) at temperature Ti.If the rate constant k(T) is assumed to have an Arrhenius-type temperature dependence k = Ae-EIRp, equation (6) can be written as f(hi) = f(A,) + (A//3) Jr e-BIRTdT or The Arrhenius integral bi has to be determined numerically and, to do this, the value of E must be known. As is readily seen, the optimum values of both constants A and E are obtained by a UDS search to find the value of E that minimises yi = f(&) + (A/P)bi Q(E) 2 [ y i ( ~ ) - f ~ c o ) - (~//3)bi12wi i e 1 where wi is determined by the form of f(A). elementary, homogeneous reactions. The same approach can, of course, be applied to Conclusions The above examples demonstrate the versatility of the UDS method in application to kinetic problems, and similar applications are available in other areas.The method is com- putationally efficient and it is particularly advantageous when values such as zero or infinity readings are missing from the data. The method is restricted to the optimisation of not more than three parameters but its range of application is still very large. Perhaps its most serious disadvantage is that the statistical reliability of the optimum parameters depends on the un- known precision of the search parameter.April, 1980 NEW NUMERICAL METHODS 131 References 1. 2. 3. 4. 5. 6. 7. 8. 9. Adby, P. R., and Dempster, M. A. H., “Introduction to Optimization Methods,” Chapman and Hall, Bevington, P. R., “Data Reduction and Error Analysis in the Physical Sciences,” McGraw-Hill, New Sands, D.E., J . Chem. Educ., 1974, 51, 473. Box, M. J., Davies, D., and Swann, W. H., “Non-linear Optimization Techniques,” Oliver and Boyd, Norris, A. C., Pope, M. I., and Selwood, M., J . Thermal Anal., 1976, 9, 425. Swinbourne, E. S., “The Analysis of Kinetic Data,” Nelson, London, 1971, Chapter 4. Holt, M. J. J ., “Improved Numerical Approaches in Chemical and Pharmacokinetic Data Analysis,” Holt, M. J. J., and Norris, A. C . , J . Chem. Educ., 1977, 54, 426. Brown, M. E., and Phillpotts, C. A. R., J . Chem. Edztc., 1978, 55, 556. London, 1974. York, 1969. Edinburgh, 1969, Chapter 2. PhD Thesis, Portsmouth Polytechnic, 1978. N urnerical Methods for N on- I i near Opt i misation W. H. Swann Production Systems Group, Corflorate Laboratory, Imperial Chemical Industries Limited, P.O.Box No. 11, The Heath, Runcorn, Cheshire, WA 7 4QE Introduction The non-linear optimisation problem can be stated as follows: find values for the variables xl, x2, . . . . . . . . . ., xn = which minimise (or maximise) the objective function f(x) subject to the constraints ci(x) > O , z = 1,2, ... , m, where f(x) and ci(x) are general, non-linear functions of the variables x. Such problems occur frequently in analytical chemistry. For example, the optimisation of chromatographic systems involves adjusting such variables as column temperature, carrier gas flow-rate, stationary phase loading and analysis time to optimise an objective function representing some measure of over-all chromatographic separation, such as total overlap, response function or informing power.Alternatively, in reaction kinetics studies, the objec- tive function may be the yield of a desired product, which is to be maximised by a suitable choice of variables such as reagent concentrations, temperature, dielectic constants and catalyst concentration. A typical constrained problem might be to minimise the cost of an analytical method with respect to reagent concentration, solvent composition, acidity and temperature, subject to upper and lower bounds on values that the variables can take, and subject also to the precision of the method not falling below some lowest acceptable limit. Iterative Approach Many numerical methods have been proposed for tackling this problem in an iterative manner.lS2 Starting from an initial approximation xo to the solution they proceed by genera- ting a sequence of improving approximations (xk), k = l, 2, .. . . . , in such a way that f(xk+l) < f(xk) (for minimisation) so that xk -+ x* as k -+ co where x* = min (f(x)}. The sequence of approximations is generated by repetitive use of the iterative equation where dk is the direction to be followed from x k and ak is a positive scalar that determines the size of the step along dk. Some methods require ak to be chosen so that f(xkfl) is the minimum along the direction dk through x k ; this procedure is known as a line search, and methods vary in the accuracy to which they require ak to be found. The vector dk is determined according to changes in f during previous iterations, so a means of obtaining values for f(x) and ci(x) for any specified x must be available, either via a computer sub-routine or directly from measure- ments on an experimental rig in a laboratory.Clearly a basic requirement of a good optimisation method is that it should locate the solution to the problem with as few steps (ie., function evaluations) as possible. Furthermore, it should be reliable and robust in that it should not break down or terminate at a non-optimal point. x k + l = xk + akdk132 NEW NUMERICAL METHODS Anal. Proc. Unconstrained Methods One method that has been used in a number of analytical chemistry applications is the simplex method, which searches the parameter space by means of a sequence of regular simplex designs. Each iteration consists of replacing the worst vertex in the design by its reflection in the centroid of the others.The method is thus very simple in concept and many developments have been suggested, the most effective of which is that due to Nelder and Mead in which the simplex expands and contracts according to the local geometry of the function. A different approach was proposed by Davies, Swann and Campey in their DSC method, in which an iteration consists of a line search along each of a set of orthonormal directions. Between iterations the directions are rotated according to progress made in an attempt to align them with the axes of the function contours. A more theoretical approach is to require that the step from the current point should yield maximum improvement in the function which is achieved by the choice@ = - g k , where g k is the vector of first partial deriva- tives of the function evaluated at xk.This is the well known steepest descent direction, which turns out to be a poor choice because it takes into account function slope but not curvature. Curvature is defined by higher order derivatives, and a more suitable choice of direction would be dk = - G-lkgk, leading to the Newton iteration xk+l = x k - akG-lkgk, where Gk is the Hessian matrix of second derivatives evaluated at x k , and G-lk its inverse. For quadratic functions this method converges in a single step, but for general functions G may be indefinite and/or singular, and various modifications have been proposed offering different ways of overcoming the difficulties which this introduces. In order to avoid the calculation of second derivatives and inversion of the Hessian it is possible to use the iteration xk+l = x k - akWkgk, where Wk is an approximation to G-l initially set equal to the identity matrix and subsequently updated via Hk+l = Hk + A".This defines a family of Quasi-Newton methods, each characterised by its particular choice of the update matrix Ak. If Ak is chosen as a single rank matrix the result is a method which does not require line searches, but which is highly unstable unless suitably modified. If Ak is chosen to be of rank 2, then a sub-class of methods results including the Davidon - Fletcher - Powell method, which has been used widely for some time, and the more recent Broyden - Fletcher - Goldfarb - Shanno update, which is emerging as probably the best Quasi-Newton method currently available.Such methods are generally efficient, but require exact line searches. Both Newton and Quasi-Newton methods require values of derivatives of the function. If these are difficult or impossible to obtain analytically, they can be calculated numerically by using finite difference approximations. In this instance considerable care must be taken in selecting a suitable finite-difference step length in order to avoid introducing significant errors, which can erode the efficiency of the search. The power of the Quasi-Newton approach is that for quadratic functions it generates conjugate directions, which possess the property that a single search along each of n such direc- tions will locate the exact minimum of the n-dimensional quadratic.An alternative method of generating conjugate directions is the method of conjugate gradients, in which the initial step is along the steepest descent direction and subsequent directions are chosen as the gradient plus an appropriate linear combination of the previous directions. A different approach avoiding the need for derivatives was suggested by Powell. In this method an iteration consists of a line search along each of n directions, initially chosen as the co-ordinate axes, followed by a search along the vector of total progress in the iteration. For the next iteration this total progress vector is added to the direction set and the first of the directions deleted. The procedure generates conjugate directions, but under certain conditions the directions can become linearly dependent, thereby restricting the search to a sub-space ; Powell and others have suggested suitable modifications to overcome this difficulty.The simplex and DSC methods are heuristic in nature. Constrained Methods The effect of constraints on the problem is to define a restricted feasible region in which Box has suggested a useful way of extending the simplex search for the solution must lie.April, 1980 NEW NUMERICAL METHODS 133 this situation by using a design involving additional vertices and contracting the design near constraint boundaries whilst expanding it in the interior of the feasible region. A more generally applicable technique is to transform the constrained problem into a sequence of unconstrained ones which can be solved using one of the methods described above.This transformation can be achieved by adding to the objective function a penalty term which increases steadily as a constraint is approached or violated and whose magnitude is controlled by an associated parameter. Careful choice of the form of the penalty can result in a trans- formation such that the required constrained solution is the limit of the unconstrained minima corresponding to a sequence of values of the controlling parameter. An alternative approach for linear constraints is to use the gradients of active constraints to form a projection matrix, which if post-multiplied by a search direction, projects that direction into the intersection of the constraint hyperplanes.This strategy effectively reduces the dimensionality of the problem, and it has been used successfully in conjunction with the DSC, Newton and Quasi-Newton methods. It can be extended to cover non-linear constraints, in which event the projected direction will lie in the constraint tangent hyper- planes and may be non-feasible, so that some means of returning to the constraint must be devised. Conclusion Optimisation techniques have found wide use in many branches of science and engineering, often with highly beneficial results. In applying them three main points need to be borne in mind. Firstly, an obvious but often overlooked consideration, that it is vital that careful attention is paid to problem definition. The true objective function must be determined and the appropriate variables (which must be independent) and relevant constraints identified.Insufficient preparation and thought on these issues can lead to the wrong problem being solved and incorrect, perhaps invalid, conclusions being drawn. An inappropriate choice can result in an inefficient search involving many additional function evaluations, which may be expensive in terms of one or all of cost, time and resources. Even worse, the search may break down, thus wasting evaluations, or “converge” to a non-optimal point , leading again to incorrect conclusions. Finally, it is not sufficient just to accept the terminal point of the search as the required solution. It is important that the nature of this point be investigated via such questions as: is the indicated point really a minimum (or maximum) ? ; is the minimum sharply defined or is the objective function relatively flat?; and what would be the benefit of relaxing one or more of the active constraints? No optimisation study is complete without an appropriate sensitivity analysis. The structural information necessary for such an analysis is provided automatically by some of the available procedures, but with others the user must generate the information by some perturbation device.Secondly, care must also be taken in selecting the optimisation method to be used. References 1. 2 . Murray, W., Editor, “Numerical Methods for Unconstrained Optimization,” Academic Press, London Gill, P. E., and Murray, W., Editors, “Numerical Methods for Constrained Optimization,” Academic and New York, 1972.Press, London and New York, 1974. Improved Methods for Numerical Differentiation of Spectroscopic Curves P. Gans Defiartnaent of Inorganic and Structural Chemistry, The University, Leeds, LS2 9 J T Numerical differentiation is a well established procedure for the location of peak maxima underlying complex spectroscopic curves. The technique most commonly used for this pur- pose was presented in detail by Savitzky and Golayl and corrected by Steinier et aZ.,2 but it134 NEW NUMERICAL METHODS Anal. PYOC. suffers, in common with all techniques for numerical differentiation, from sensitivity to spectroscopic noise and usually the differential shows a lower signal to noise ratio than the parent curve. This makes the detection and location of small peaks difficult.We have investigated the use of spline functions for smoothing and differentiation and have found that the deterioration in signal to noise ratio is generally much smaller with this method. A spline function consists of a number of pieces, each of which is a polynomial of degree n. The pieces are joined together end-to-end and the positions of the joins are termed knots. In addition, the function and all of its derivatives up to that of order n - 1 are continuous in a mathematical sense at the knots. Any spline function of degree n can be constructed as a linear combination of a set of Basis, or B-splines of degree n, which are completely defined by the knot sequence. In our method the coefficients of the linear combination are found by a linear least-squares fitting process applied to the spectroscopic curve.The smoothed curve and its differentials are computed from these coefficients. The method has been applied to the analysis of Raman spectra. Both cubic and quintic spline functions were investigated, and various knot placement schemes were examined. The second differentials obtained by this method had much higher signal to noise ratios than those obtained by the Savitzky and Golay method, and therefore gave a more reliable estimate of the derivative of the noise-free spectrum. The derivative in (b) is that which is expected in the absence of noise. The best results were obtained using a scheme in which knots are so placed as to give the smoothest spline function to fit the data, shown in (e) and (g).With this so-called optimum distribution a smaller number of knots were needed than when they are equally spaced, and hence the amount of noise in the deriva- tive, which tend to increase with the number of knots, decreased. A full description of the method and results has been submitted for publication to Analytical Chemistry. Typical results are shown in Fig. 1 for a curve of known composition. Fig. 1. (a), Experimental curve consisting of two Lorenzians + noise. (b), D2y ( c ) , Dzy obtained using the Savitzky and ( d ) , Dzy obtained from a cubic (e), D2y obtained as in (d), but with an optimal ( f ) , Dzy obtained from quintic spline function with equispaced obtained by differentiating the Lorenzians. Golay method with a 39-point quadratic function.spline function with equispaced knots. knot distribution. knots. (g). Dzy obtained as in (f), but with an optimal knot distribution.April, 1980 NEW NUMERICAL METHODS 135 The method can be applied to the smoothing, differentiation and integration of any curve in digital form, regardless of the origin and shape of the underlying functions. It utilises the whole information contained in a spectrum but rejects most of the information content regard- ing derivatives of order n or higher. References 1. 2. Savitzky, A., and Golay, M. J. E., Anal. Chem., 1972, 44, 1906. Steinier, J., Termonia, Y., and Deltour, J., Anal. Chem., 1972, 44, 1906. Computational Practice in Pattern Recognition Lars Kryger Department of Chemistry, Aarhus University, Langelandsgade 140, 8000 A arhus C, Denmark Todays analytical chemists are frequently faced with chemical problems of extreme complexity involving, for example, the characterisation and comparison of multi-component samples.A wealth of chemical information from modern analytical instruments may be available, but the number of parameters necessary for the characterisation of a sample, e.g., an oil spill, can be very large, and a simultaneous comparison of all of the parameters may not be feasible for the human brain. Further, a large proportion of the information available may be irrelevant to the problem considered, Thus, in order to obtain useful knowledge the chemist must recognise and reject such superfluous information, which otherwise might obscure the relationships sought, During the past decade pattern recognition techniques have been used successfully in order to interpret complex chemical data.This paper attempts to give a brief introduction to some of the computational procedures that have proved successful for the solution of selected analy- tical chemical problems. Many analytical chemical problems can be formulated as questions about classification. By a class we shall denote a set of samples that exhibit a certain amount of chemical similarity, e.g., oil spills originating from two distinct geographical locations could be said to belong to two classes. Two typical approaches frequently used in chemical pattern recognition will be considered, namely the supervised learning approach and the unsupervised learning approach.In supervised learning the goal is to classify one or more samples, the “prediction set,” by their similarity to samples of known classification, the “training set.” In unsupervised learning the object is firstly to reveal relationships between groups of samples by searching for subsets of the samples being particularly similar, and secondly, to interpret such relationships in terms of the physical and chemical properties of the samples. Concepts Pattern In chemical analysis a pattern is mostly a collection of data obtained by chemical measure- ment on a sample. A pattern may consist of raw measurements, but often functions derived from the raw data constitute the patterns. If, for example, a chemical sample is characterised by its gas chromatogram, a pattern might be a series of peak heights obtained from the chromatogram.If we initially assume that, say, ten components are essential for the analy- tical problem under consideration, we can represent the information concerning each sample as a point in a ten-dimensional space, in which any axis represents the concentration of a component. Feature In the above example a feature could be the peak height for one of the ten components, but also some more complex function, such as the ratio of two peak heights, could be interesting to include as a feature. The term feature will denote each of the components of a pattern.136 NEW NUMERICAL METHODS Anal. PYOC. Feature Definition The process of feature definition involves a careful consideration of those factors which might be important for the solution of the chemical problem under consideration.This step calls for the ingenuity and knowledge of the chemist practising pattern recognition. Clearly, features defined on the basis of an educated chemical guess stand a better chance of eventually turning out useful for the classification than do features picked at random. In the chromatographic example the width, height or retention time of the solvent peak are examples of features carrying no information about the sample. Such features would be useless for classification and might obscure useful relationships. Distances and Similarity Very often a simple Euclidean measure of distance is used in chemical pattern recognition; thus, the distance between patterns i and j , d f j , can be written as where x, denotes the qth feature of pth pattern and n is the dimension of the pattern space.A measure of similarity can be based on the above definition: if the similarity of identical patterns equals 1 and the similarity of patterns separated by the maximum distance equals 0, Sij = 1 - dij/max. (diJ .. .. .. . . (2) Examples Essentially, pattern recognition techniques allow the chemist to make comparisons of data of high dimensionality (normally larger than 3). However, for illustration purposes we shall confine ourselves to three-dimensional data, which are readily visualised. This is no over-simplification, as distances and similarities can be computed according to (1) and (2) for any dimension of the pattern space. The following two examples both use the same data base, namely, the X-ray ff uorescence spectra of six samples of a silver - copper alloy plated with nickel and chromium (shown in Fig.1). The Supervised Learning Approach Assume that two factories (1 and 2) produce the plated alloy and that we want to determine from which of the two a sample (A) originates. Two samples of alloy (D) and (F) from factory 1 and three (B) (E) and (C) from factory 2 are secured, and the X-ray spectra of the unknown (the prediction set) and the five known samples (the training set) are recorded. I t is reasonable to believe that the composition of a sample, and hence the relative inten- sities of the &-lines of the four elements listed above, in some way reflects the process (and factory) that produced the sample.If we define three-dimensional patterns by using the intensities of the K,-lines for silver, copper and nickel as features, these patterns fall in two distinct clusters (or classes), namely: 1, A, D and F; and 2, B, E and C, as shown in Fig. 2. The unknown, A, clearly belongs to the cluster A, D, F rather than the cluster B, E, C, and it is reasonable to assume that this cluster reflects the origin of the sample (factory 1). If chromium is also considered, and hence a four-dimensional space is searched for clusters, the picture becomes more blurred, ie., the content of chromium is a less useful feature for classification. In order to improve the prob- ability of a correct prediction the training set should be expanded with more samples of known classification.In this context the K nearest neighbour rule may be used for classification. This rule assigns any sample to the class to which the majority of its K nearest neighbours belong. K here is a small odd number. The Unsupervised Learning Approach so as to obtain a uniform product. Let us now assume that a factory producing the plated alloy wants to optimiseits production Samples are taken at regular time intervals from theApril, 1980 NEW NUMERICAL METHODS 137 r - A B C CU-KCX A l K a ,llnNi-KCX Cr-Ka Fig. 2 . Three-dimensional patterns repre- senting the six samples of the alloy. The intensities of the Ka-lines of silver, copper and nickel are used as features. Fig. 1. X-ray fluorescence spectra of six samples of a silver - copper alloy plated with nickel and chromium.production line and subjected to X-ray spectroscopy. For simplicity we shall use the same data as before and assume that all six samples originate from the production at this factory. The problem is now to recognise any non-uniformity, i.e., cluster formation, in the data. If such clusters are found then these should be explained by studying the experimental parameters (pressure, temperature, plating bath, etc.) employed when the samples were produced. If an explanation is found this may suggest which experimental parameters should be better controlled. Cluster formation is easily recognised by visual inspection when n < 3. For higher dimensionality a matrix of similarities based on equation (2) can be computed. This matrix is then searched for its largest element, which identifies the two most similar patterns. These two patterns are then replaced by one pattern a t their centre of gravity, thus forming a cluster. Then a new similarity matrix is computed by using this first cluster and the re- maining patterns. The matrix is again searched for its largest element, and so on. In this manner the patterns are gradually joined to form clusters. In this example, of course, the ideal situation is the formation of one cluster only. Conclusion Chemical pattern recognition is a rapidly expanding field, and the techniques have been successfully applied within several fields of analytical chemistry. The bibliography is by no means exhaustive, but the text-books and papers mentioned provide a good starting point for the interested reader. Bibliography Books Jurs, P., and Isenhour, T., “Chemical Applications of Pattern Recognition,” Wiley, New York, 1975. Fukunaga, K., “Introduction to Statistical Pattern Recognition,” Academic Press, New York, 1972. Wold, S., and Sjostrom, M., “SIMCA,” in Kowalski, B. R., Editor, “Chemometrics,” Symposium Series 52, American Chemical Society, Washington D.C., 1977.EQUIPMENT NEWS Anal. R o c . Papers Topic General papers on chemical applications of pattern recognition. Classification of obsidian samples based on X-ra.y Classification of severely overlapped voltammetric emission from trace elements. data. Classification of organic compounds based on voltammetric behaviour. Screening of prospective anti-cancer drugs, using Prediction of the long-term behaviour of Ni - Cd Analysis of polymers by pyrolysis - GLC. structural information. space craft cells based on short-term measurements. Reference Kowalski, B. R., and Bender, C. F., J . Am. Kowalski, B. R., and Bender, C. F., J . Am. Chem. SOC., 1972, 94, 1536. Chem. SOC., 1973, 95, 686. Kowalski, B. R., Schatzki, T. F., and Stross, F. H., Anal. Chem., 1972, 44, 2176. Thomas, Q. V., de Palma, R. A., and Perone, Burgard, D. R., and Perone, S. P., .4nal. Kowalski, B. R., and Bender, C. F., J. Am. Arthur, B. W., and Perone, S. P., J . Electrochem. S. P., Anal. Chem., 1977, 49, 1376. Chem., 1978, 50, 1366. Chem. SOC., 1974, 96, 916. SOC. 1979, 126, 720. Kullik, E., Kaljurand, M., and Koel, M., J . Chromatog., 1975, 112, 297.

ISSN:0144-557X    

DOI:10.1039/AP9801700120

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

138 EQUIPMENT NEWS Anal. R o c . Equipment News Automatic Nitrogen Analyser The Carlo Erba ANA Model 1400 will analyse nitrogen in a wide range of organic and inorganic compounds. The principle of operation consists of a flash combustion in an oxygen-enriched atmosphere of helium, with a katharometer detection system. Fifty analyses providing both a digital display and a printout of the results can be carried out sequentially, with results directly comparable with those obtained by traditional methods of analysis (i.e. Kjeldahl, Dumas, etc.), accuracy better than 0.2% and detection of amounts less than 0.01%. Insert El on the Reader Enquiry Service form for further information. Erba Science (UK) Ltd. Oxygen Annlyser The 2-Ox trace oxygen analyser is used for pollution monitoring, inert gas analysis in lamp manufacture ancl semiconductor work.It is also suitable for monitoring and controlling combustion processes, heat treatment plants and inert gas blankets. Three ranges to cover oxygen determinations from less than 1 v.p.m. up to lOOyo are available. A three-stage antilogger converts results to a linear output for accurate recording. A stabilised zirconium oxide tube is used as an oxygen concentration cell, producing an e.m.f. dependent on the ratio of the oxygen concentra- tion inside the cell to the reference gas (air) out- side the tube. EOC Ltd. Insert E2 on the Reader Enquiry Service form for further information. Gas Sensor The gas sensor, sensitive to propane, butane, methane and hydrogen, is a ceramic semi- conductor type which uses sintered iron com- posite oxide as a sensing element, with a response time of 10 s cr less.The sensor operates without a noble metal catalyst, and is thus sensitive to water vapour and vaporised alcohol. The sintered fine-grain (0.05-0.2 pm diameter) alpha-type iron(II1) oxide ceramic changes its resistivity when it comes in contact with combustible gas, the change being detected by buried platinum wires. In- sert E3 on the Reader Enquiry Service form for further information. Matsushita Electric Industrial Co Ltd. Colour Measurement The Automatic Tintometer measures the optical density of a sample at two points in the spectrum equivalent to the Lovibond values of red and yellow and indicates these values automatically on two meters. A deep red reference filter corrects any erroneous readings caused by the presence of dirt or suspended solids in the sample.Tintometer Ltd. Insert E4 on the Reader Enquiry Service form for further information.138 EQUIPMENT NEWS Anal. R o c . Equipment News Automatic Nitrogen Analyser The Carlo Erba ANA Model 1400 will analyse nitrogen in a wide range of organic and inorganic compounds. The principle of operation consists of a flash combustion in an oxygen-enriched atmosphere of helium, with a katharometer detection system. Fifty analyses providing both a digital display and a printout of the results can be carried out sequentially, with results directly comparable with those obtained by traditional methods of analysis (i.e. Kjeldahl, Dumas, etc.), accuracy better than 0.2% and detection of amounts less than 0.01%.Insert El on the Reader Enquiry Service form for further information. Erba Science (UK) Ltd. Oxygen Annlyser The 2-Ox trace oxygen analyser is used for pollution monitoring, inert gas analysis in lamp manufacture ancl semiconductor work. It is also suitable for monitoring and controlling combustion processes, heat treatment plants and inert gas blankets. Three ranges to cover oxygen determinations from less than 1 v.p.m. up to lOOyo are available. A three-stage antilogger converts results to a linear output for accurate recording. A stabilised zirconium oxide tube is used as an oxygen concentration cell, producing an e.m.f. dependent on the ratio of the oxygen concentra- tion inside the cell to the reference gas (air) out- side the tube.EOC Ltd. Insert E2 on the Reader Enquiry Service form for further information. Gas Sensor The gas sensor, sensitive to propane, butane, methane and hydrogen, is a ceramic semi- conductor type which uses sintered iron com- posite oxide as a sensing element, with a response time of 10 s cr less. The sensor operates without a noble metal catalyst, and is thus sensitive to water vapour and vaporised alcohol. The sintered fine-grain (0.05-0.2 pm diameter) alpha-type iron(II1) oxide ceramic changes its resistivity when it comes in contact with combustible gas, the change being detected by buried platinum wires. In- sert E3 on the Reader Enquiry Service form for further information. Matsushita Electric Industrial Co Ltd. Colour Measurement The Automatic Tintometer measures the optical density of a sample at two points in the spectrum equivalent to the Lovibond values of red and yellow and indicates these values automatically on two meters. A deep red reference filter corrects any erroneous readings caused by the presence of dirt or suspended solids in the sample.Tintometer Ltd. Insert E4 on the Reader Enquiry Service form for further information.138 EQUIPMENT NEWS Anal. R o c . Equipment News Automatic Nitrogen Analyser The Carlo Erba ANA Model 1400 will analyse nitrogen in a wide range of organic and inorganic compounds. The principle of operation consists of a flash combustion in an oxygen-enriched atmosphere of helium, with a katharometer detection system. Fifty analyses providing both a digital display and a printout of the results can be carried out sequentially, with results directly comparable with those obtained by traditional methods of analysis (i.e.Kjeldahl, Dumas, etc.), accuracy better than 0.2% and detection of amounts less than 0.01%. Insert El on the Reader Enquiry Service form for further information. Erba Science (UK) Ltd. Oxygen Annlyser The 2-Ox trace oxygen analyser is used for pollution monitoring, inert gas analysis in lamp manufacture ancl semiconductor work. It is also suitable for monitoring and controlling combustion processes, heat treatment plants and inert gas blankets. Three ranges to cover oxygen determinations from less than 1 v.p.m. up to lOOyo are available. A three-stage antilogger converts results to a linear output for accurate recording.A stabilised zirconium oxide tube is used as an oxygen concentration cell, producing an e.m.f. dependent on the ratio of the oxygen concentra- tion inside the cell to the reference gas (air) out- side the tube. EOC Ltd. Insert E2 on the Reader Enquiry Service form for further information. Gas Sensor The gas sensor, sensitive to propane, butane, methane and hydrogen, is a ceramic semi- conductor type which uses sintered iron com- posite oxide as a sensing element, with a response time of 10 s cr less. The sensor operates without a noble metal catalyst, and is thus sensitive to water vapour and vaporised alcohol. The sintered fine-grain (0.05-0.2 pm diameter) alpha-type iron(II1) oxide ceramic changes its resistivity when it comes in contact with combustible gas, the change being detected by buried platinum wires.In- sert E3 on the Reader Enquiry Service form for further information. Matsushita Electric Industrial Co Ltd. Colour Measurement The Automatic Tintometer measures the optical density of a sample at two points in the spectrum equivalent to the Lovibond values of red and yellow and indicates these values automatically on two meters. A deep red reference filter corrects any erroneous readings caused by the presence of dirt or suspended solids in the sample. Tintometer Ltd. Insert E4 on the Reader Enquiry Service form for further information.April, 1980 EQUIPMENT NEWS 141 Stereo Microscope The A 0 Stereostar Zoom 580” microscope with integral photographic and coaxial illumination facilities, has a basic magnification range of 1 x to 6 x at a constant working distance of 4 in, and can bc extended to 300 i: with avail- able auxiliaries. A manual camera or the Expostar Camera system can be fitted.A high-density quartz halogen 20-W lamp enables surface details and irregularities to be accurately defined. Reichert- Jung UK, Scientific Instrument Division of British American Optical Company Limited. Insert E5 on the Reader Enquiry Service form for further information. Gas Absorption Cells Long path-length gas absorption cells (LPGAC), complete with built-in transfer optics, are avail- able for the analyses and other measurements of gases by spectrophotometric means. They may be operated in the ultraviolet, visible or infrared ranges. The cell walls are Pyrex, enabling photochemical studies to be carried out by irradiation of the cell contents.Four standard sizes are available. The Model GC1, the single path-length cell, is available in 5 and 17 m path-lengths: the Model GC2, the dual path-length cell, is avail- able in 20 and 40 m and in 36 and 72 m path- lengths. Changing of the path-length in a dual path-length cell is done externally by changing the positioning of the incoming beam, by adjust- ment (reversal) of the transfer optics assembly. The multi-passing optics are aluminised, over- coated telescope quality mirrors with permanent alignment. The cells are vacuum tight, with a built-in vacuum gauge. Newman-Howell Associates Ltd. Insert E6 on the Reader Enquiry Service form for further information. Picture Analysis The MOP Videoplan is designed to provide an.economical method of solving difficult measuring problems in the field of image analy- sis. It contains a built-in computer giving a computer-aided manual collection of geometrical structure data in video interactive mode. The complete system consists of a basic computt:r unit, data monitor, graphical measuring tablet and printer - plotter for output of the measured values. Measuring tablet areas are in ten sizes from the standard 28 x 26 cm to 106 x 152 cm and accuracy of f 0.1 mm. Reichert- Jung UK, Scientific Instrument Division of British American Optical Company Limited. Insert E7 on the Reader Enquiry Service form for further information. Integrating Computers for Chromatography Supergrators 1A and 3A supersede the existing Supergrator 1, 2 and 3 models of integrating computers for gas and liquid chromatography.The latest developments in microprocessor technology and solid-state electronics are used to give operational simplicity, modular expandable construction and full computer capability. Both models provide facilities for six measure- ment modes for calculating peak area, height or both, along with area per unit time or average height over a set time period and a digitise mode, which can also be used for checking signal ampli- tude and polarity. The Supergrator 3A can also be programmed to carry out repeat analyses or re-calibration automatically. The Model 1A will accurately measure peak area, height or both, but its automatic calcula- tion capability is limited to area yo.It can be upgraded to the Model 3A specification by fitting available plug-in expansion units. Kemtronix Ltd. Insert E8 on the Reader Enquiry Service form for further information. Portable Calculator The HP-41C, a hand-held programmable calcu- lator, has a keyboard that can be re-defined to suit tlie user’s personal needs. It provides 448 bytes of programme memory or 63 data storage registers. Hewlett-Packard Ltd. Insert E9 on the Reader Enquiry Service form for further information. Strip - chart Recorders Three Euroscribe flat-bed chart recorders have a capacitance transducer in place of the usual slide-wire potentiometer, thus overcoming the use of sliding contacts which are prone to wear142 EQUIPMENT NEWS Anal. Proc. and dirt. The range gives a choice of a single pen, a double pen or an integrating recorder.A. Gallenkamp & Co. Ltd. Insert E l 0 on the Reader Enquiry Service form for further information. Electronic Recorder A compact three-pen strip-chart electronic recorder, the P103L, records three independent traces on a 100-mm chart. It can record different ranges and types of variable on its three recording channels, for instance, record the temperature, pressure and flow of a liquid or monitor the temperature, pressure and humidity within an air duct. Range changing can be affected by means of plug-in cards. The list of ranges includes temperatures up to 1 GOO "C and electrical signals up to 300 V or 1 A d.c. or 600 V or 5 A a.c. Other variables such as pressure, power, power factor and relative humidity can be accommodated by use of suitable transducers, as well as the level of liquids, powders and granular solids.Foster Cambridge Ltd. Insert El 1 on Reader Enquiry Service form for further information. Multi- channel Recorders Three models in the BS300 series accept one or two channels (Model BS312), onc to four chan- nels (Model BS314) and one to six channels (Model BS316). By fitting the 1-12-channel multi-point option, Model BS314 has 13 channels available, 12 multiplexed and 1 line channel, and Model BS316 has 14 channels, 12 multi- plexed and 2 line channels. The multi-point option is not available with Model BS312. Multiplexed time is variable, with a choice of 1,2,4,10,20,50 and 100 s per print. The pen response is 0.3 s full scale, and the frequency response 1.5 Hz (3 dB) .Reproducibility is 0.15% and there are 16 chart speeds. Bryans Southern Instruments Ltd. Insert E l 2 on the Reader Enquiry Service form for furthcx information. Microprocessor -based Recorders The Series BS8000 microprocessor-based data acquisition and manipulation recorders can have up to eight analogue input channels, each with a differential input. Each model provides the following mathematical functions, to 16-bit precision : add, subtract, multiply, divide, average any two curves (500 points each) point by point, differentiate, integrate, logarithm, reciprocal, normalise, and smooth any curve (600 points) point by point. Plotting functions comprise offset, scale, expand, compress and shift. A floppy disk allows up to 300 graphs, each of 500 points, to be stored, and to monitor the output a 6-character numerical display.Bryans Southern Instruments Ltd. Insert E l 3 on the Reader Enquiry Service form for fui-ther information. Abrasive Cut-off Machine The CM20 abrasive cut-off machine is designed to prepare smooth, precise sample surfaces for spectrographic, X-ray or solute analysis. The abrasive wheel (either 10 or 12 in) will section samples up to 63 mm in diameter. A water recirculating system to allow submerged or jet spray cutting is incorporated. Lcco Instruments (UK) Limited. Insert E l 4 011 tlic Reader Enquiry Service form for further information. Portable Photometer A compact aerosol photometer, the Phoenix Precision Instruments JM-8000, is designed for particulate monitoring indoors or out.With a near forward scattering optical system, the JM-SO00 responds to particles down to 0.1 pm, and possesses a wide range, from 100 to 0.0001 pg 1-1 of 0.3-pm dioctyl phthalate aerosol. This upper limit is higher than the turbidity of any atmospheric aerosol except dense water fogs,April, 1980 EQUIPMENT NEWS 143 while lower limits correspond to an atmospheric visibility of 930 miles. Techmation Ltd. Insert E l 5 on the Reader Enquiry Service form for further informat-ion. Filter Sets Fifteen different filter sets, e.g., neutral density, coloured glass, interference variable band pass, ultraviolet and visible step, infrared and colour additive/subtractive filter sets, have each an individual, original spectrophotometric curve measured on a Cary 14 or equivalent.Insert E l 6 on the Reader Enquiry Service form for further information. Melles Griot B.V. Sub -micron Filter Elements Sinterecl 316 stainless-steel filter elements, designed by Nupro, are suitable for high purity systems, gas supply systems and analytical instrumentation. The sub-micron elements are a nominal 0.5 pm arid can handle differential pressures up to 1000 lb in-2 (6 890 Pa) and temperatnres up to 482 “C. The elements can be installed in Nupro in-line (“F” series) and tee-type (“T” series) filter housings. End-connections include Swagelok-type fittings, male and female NPT, socket and butt weld, and integral tube stubs. Techmation Ltd. Insert El7 on the Reader Enquiry Scrvice form for further information. Pipette Filler A pipette filler has three sensitive and accurate valves operated through control pads within easy reach of the fingers.The tapered neck design will fit pipette sizes from 4.5 to 10.5 mm diameter. The pipette filler is produced from rotationally moulded PVC, which resists attack from most laboratory chemicals. Insert E l 8 on the Reader Enquiry Service form for further information. Saffron. BUZZ - Plug Four new power failure alarms to supplement the Buzz-Plug range have temperature sensors attached. The alarms thus warn of undesirable temperature changes within the equipment, as well as total power failure. Four mqdels are available: the 37 “C model wains when the temperature within the appara- tus rises oi- falls by more than 2 “C; the 4 “C model warns when the temperature riscs above 6 “C; and the -20 OC model warns whcn the temperature inside, for example, a freezer, rises to an unacceptable level. The variable- temperature model is variable from -75 to 200 OC, with any temperature accurately set to within f0.25 “C of the desired point.Insert E l 9 on the Reader Enquiry Service form for further iniornia tion. Laboratory Impex Limited. Gas Chromatographs SIGMA lB, one of the “B” versions of the Sigma gas chromatographs, includes the BASIC programniability option as standard and the BASIC I1 facility available as an option, enabling the user to read BCD code and control an autosampler. SIGMA 2B offers all the analytical functions of SIGMA lF3, and can now be equipped with an auxiliary heat zone for the gas sampling valve.SIGMA 3R contains in its memory a service diagnostic chip to enable checks to be made on the various operations. Independent control of injector and detector temperatures areavailable on SIGMA 3B andalso on SIGMA 4B. An addition to the range of accessories includes the AS-100 automatic liquid sampler. Perkin-Elmer Ltd. Insert E20 on the Reader Enquiry Service form for further information. Chromatograph Back-flush and Cut Acces- sory A back-flush and cut accessory, compatible with the complete range of Sigma instruments, consists of an automatic system for back- flushing and “heart cutting” using a pressure balance system with two columns. Insert E2 1 on the Reader Enquiry Service form for further information. Perkin-Elmer Ltd. Catalytic Reactor Accessory The Catalytic Reactor makes possible the analysis of carbon monoxide and carbon dioxide at low levels.The reactor converts carbon monoxide and carbon dioxide in the carrier144 EQUIPMENT NEWS Anal. PYOC. gas by catalytic hydrogenation into methane, which can then be detected at low levels using an FID. Insert E22 on the Reader Enquiry Service form for further information. Perkin-Elmer Ltd. Atomic-absorption Systems The SP9 system of flame or flameless spectro- meters incorporate computer data handling, a video furnace programmer and flame and flameless autosamplers. Pye Unicam Ltd. Insert E23 on the Reader Enquiry Service form for further information. Ultraviolet - Visible Spectrophotometers The SP6 and SP8 series’ performance has been improved by fitting master, blazed holographic gratings and silica-coated optics.In addition the SP8-150 extends the wavelength range of the SP8-100, and the SP8-250 incorporates a double, master holographic monochromator. Insert E24 on the Reader Enquiry Service form for further information. Pye Unicam Ltd. Infrared Spectrophotometers The SP3 series of ratio-recording infrared spectrophotometers is announced. SP3-100 or SP3-200 both cover the frequency range 4 000 to 600 cm-l, with the latter providing addi- tionally calibrated transmittance expansion and back-off and five scan times. The Spectraset control system will select the optimum operating conditions. The SP3-300 covers the range 4000 to 200 cm-l by incorpora- tion of a fast-response pyroelectric detector and a ceramic source.Pye Unicam Ltd. Insert E25 on the Reader Enquiry Service form for further information. Electrophoresis Equipment The gel electrophoresis apparatus GE 2/4 is suitable for both polyacrylamide gel rods up to a maximum of 20 per run, and gel slabs. Two gel slabs sized 180 x 140 x 0.7mni (or 2.7 mm) thick can be run simultaneously, incorporating dissociating media if desired. It can also accommodate up to four 140 x 80 x 2.7 mm or 80 x 80 x 2.7 mm gels. Poly- acrylamide gradient gels PAA 4/30 and PAA 2/16 can similarly be used. The apparatus is powered by the electrophoresis power supply EPS 500/400. Pharmacia (Great Britain) Ltd. Insert E26 on the Reader Enquiry Service form for further information. Molecule and Crystal Structure Models A new List of models produced by Beevers Miniature Models will shortly be available.These models are compact, of particularly attractive appearance, are accurately made and are relatively cheap. All hydrogen atoms are inserted. Models can be readily made of new structures. Insert E27 on the Reader Enquiry Service form for further information. Beevers Minature Models. New Materials Affi-Gel 731, a solid-phase support for mem- brane isolation, consists of polyethyienimine covalently bonded to the surface of poly- acrylamide beads. The polycationjc f unc- tionality of Affi-Gel 731 binds cells to the bead surface by ionic attraction. Subsequent lysis of the cells lea\Tes a patchwork of membranes covering the bead surface. The cell and organelle derived membranes can be eluted from the support without causing damage.Also available is a high-strength matrix for anchorage-dependent cell culture, BIO-Carriers, a high-strength polyacrylamide microcarrier suitable for culture of primary and diploid cells. The beads have a diameter range of 120-180 mm with a specific gravity of 1.04, which promotes homogeneous suspension. The polyacrylamide matrix withstands stirring and cell growth without breaking up, and maintains their transparency for microscopic viewing. The beads provide a very large surface area; thus 76ml of swollen beads, equivalent to 4 g dry mass, incorporated in a 1-1 flask, are equivalent to 150 Petri dishes or 160 75-ml flasks, yielding 2 x l o 9 cells. Bio-Rad Laboratories Ltd. Insert E28 on the Reader Enquiry Service form for further information.The Mercury Indicator (Product 1-4509) is a powder formulation which is applied directly to a surface, or as a paste on vertical surfaces. Within several hours, pink or black spots will begin to develop wherever mercury is present. No additional analytical work is necessary. J. T. Baker Chemicals B.V. Insert E29 on the Reader Enquiry Service form for further information.April, 1980 CHEMICAL SOCIETY : ANALYTICAL DIVISION 145 ION CHROMATOGRAPHY The Fastest Growing Major Analytical Technique for Ions in Solution Ion Chromatography analyzes ions in solution. Its unique characteristics include: specificity, and rapid, sequential analysis. Although based on ion-exchange chromatography, IC uses conductimetric detection to achieve unprecedented sensitivity 1evek.e.g.less than 10 ppb. IC is especially suited for the analysis of - complicated matrices - several ions in a single sample - ions in low concentration in the presence of a large concentration of other ions 'trace (ppb) levels of ions. Dionex offers this powerful analytical technique in four practical instruments that perform Ion Chromatography rapidly and routinely either manually or automatically. I F- 0 5 lo 15 20 I 1 I I 1 MINUTES Application areas Air Pollution Analysis of ambient aerosols for nitrate and sulfate; trace ions in rain water; SO, in the atmosphere; anions in auto/ diesel exhaust; sulfuric acid (SO,) in stack gas; sulfite, sul- fate in scrubber liquors. Water Pollution Ion characterization of waste effluents; routine ion analysis of ground waters, nitrate-N and phosphate-P in hatchery and bio-pond water; chloride, sulfate and oxalate in paper mill effluent and Kraft black liquors.Elemental Analysis Trace level and interference free analysis of organic fluorine, chlorine, bromine, iodine, sulfur and phosphorus after com- bustion of organic compounds, polymers, and coal. Soil Analysis Direct anion analysis in KCI, LiCI, ammonium acetate, ammonium fluoride,sulfuric acid or bicarbonate soil extracts. Brine Analysis Direct analysis of chlorate, sulfate, calcium and magnesium in 25% brine and 50% caustic solutions; phosphate, bromide, nitrai:e, sulfate, calcium and magnesium in 2% brine. Power Production Trace chloride, phosphate, sulfate, sodium, potassium, mag- nesium and calcium analysis in boiler, boiler feed, steam generator, turbine condensate; fuel cell effluents.Anions in geothermal waters, coal liquification and gasification. Quality Control/Process Oxalate in Bayer liquors, phos- phate, ammcnia in fertilizers; fluoride, sulfate, chromate in plating baths; trace anions, ca- tions, amines in electronic device process water, glycolate in sur- factan ts; monobu ty I, di bu ty I phosphates in uranium refining; halides and sulfates in foods and food additives, primary, secon- dary, tertiary, quaternary amines in monomers; nitrite, nitrate in spent sulfuric acids, cutting fluids and engine coolants; amines in DMF. I Conccnlration (ppb) ct- 5 PO4 3 20 NOj 20 so4 2 20 m i i i r b h i 4 MINUTES Trace Analysis (ppb) of'Anions with a Concentrator Column One of the most significant additions to the Dionex product line is the Concentrator Column. I t has improved Minimum Detection Limits (MDL) significantly. The Concentrator retains ions as sample solution is injected through it; thus, ionspresent in traceamounts (often too small to be detected by routine analysis) accumulate to measurable concentrations in the resin bed. MDL for many common ions are now less than 1 ppb with a 10 ml injection (e.g. SO,-2 may now be easily measured at a concentration of 1 ppb). c/o Roger George, 4 The Buchan, Camberley, Sur'rey GU 15 3XB. Tel: 0276 26373 Telex: 896691 A7 for further information. See page 150146 CHEMICAL SOCIETY : ANALYTICAL DIVISION Anal. Proc. Reference Standards for NMR including TMS, HMDS etc. Full range of Shift Reagents A Complete I NMR Accessories ' m n * I r I including Coaxial I W Cells and Microcell Wilmad Glass 3ervlce lor I Assemblies from 1 Spectmxopists Deuterated Solvents including '100%' purity for NMR Quartz tubes ESR Spectroscopy Precision bore glass and quartz tubing from Wilmad Glass FOR FULL SETOF CATALOGUES: APPLY TO: FLUOROCHEM LIMITED DINTING VALE TRADING ESTATE, DINTING LANE, GLOSSOP, DERBYSHIRE, SKI3 9NU. Tel: Glossop 4917 or 2855 (STD code 04574-1 Telex: 669960 A8 for further information. See page 150April, 1980 CHAIRS FOR ANALYTICAL CHEMISTS Chairs for Analytical Chemists Dr. D. Betteridge has been awarded a Personal Chair in Chemistry a t the University College of Swansea, and Dr. G. F. Kirkbright has been Dr. D. Betteridge Dr. G . F . Kirkbright appointed to the Chair in Analytical Sciences a t the University of Manchester Institute of Science and Technology. 147

ISSN:0144-557X    

DOI:10.1039/AP9801700138

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

April, 1980 CHAIRS FOR ANALYTICAL CHEMISTS 147 Publications Received Electroanalysis in Hygiene, Environmental, Clinical and Pharmaceutical Chemistry. Proceedings of a Conference, organised by the Electroanalytical Group of the Chemi- cal Society, London, held at Chelsea College, University of London, April 17th- 2Oth, 1979. Edited by W. Franklin Smyth. Analytical Chemistry Symposia Series, Volume 2. Pp. xii + 473. Elsevier. 1980. Price $70.75; Dfl145. ISBN 0 444 41850 4 (volume 2); ISBN 0 444 41786 9 (series). These Proceedings of the International Sym- posium are principally concerned with the way in which potentiometric and voltammetric methods of electroanalysis are used to solve a wide variety of analytical problems in the fields of clinical chemistry, hygiene, pharmacy, pharmacology and environmental chemistry.Although papers here reflect the many topics under discussion, certain themes predominate : investigation of the electrochemical techniques and instrumentation ; importance of the mechanism of the electrochemical reaction at the chosen indicator electrode in optimising the electroanalytical “finish” ; direct rapid and sensitive measurements in complex biological matrices ; results on the determination of inorganic, organic and organometallic sub- stances in complex matrices such as body fluids, factory air and the aqueous environment; on- line analysis with and without a prior separa- tion process; speciation studies by the applica- tion of “cold” electroanalytical methods ; and finally the details of novel electrode construc- tion, instrumental design and working analytical methods.Vogel’s Elementary Practical Organic Chemistry 1. Preparations. Third Edition. Revised by B. V. Smith and N. M. Waldron. Pp. xviii + 407. Longman. 1980. Price L9.95. In this Third Edition the range of preparations has been extended and all apparatus and descriptions brought up to date. The book begins with a completely re-written section on modern experimental techniques of organic chemistry. A new chapter on essential laboratory operations follows, which details widely used purification and separation methods including chromatography. All of these methods are developed through the introduction of new experiments. The majority of the preparations of aliphatic and aromatic compounds are retained and examples of more recent reactions are given.A new chapter on the preparation of selected heterocyclic compounds is added. Stress is laid on the applications of infrared, ultraviolet and nuclear magnetic resonance spectroscopy and spectroscopic data from some 300 com- pounds are included in a form which is also suitable for tutorial work. ISBN 0 582 47009 9. Electrophoresis. A Survey of Techniques and Applications. Edited by 2. Deyl, co-edited by F. M. Everaerts, 2. Prusik and P. J. Svendsen. Journal of Chromatography Library, Volume 18. Elsevier. 1979. Price $83; Dfl170. ISBN 0 444 41721 4 (volume 18) ; ISBN 0 444 41616 1 (series). This, the first volume in a two-part set, deals with the principles, theory and instrumentation of modern electromigration techniques.The second part will be concerned with the detailed applications of electromigration methods to all diverse categories of compounds, although to a limited extent some applications are discussed in Part A. Electromigration methods have been so extensively used in the past for both analytical and preparative separations that several of these methods have become standard pro- cedures. These axe discussed In the book, together with newer developments in the field. Part A : Techniques.148 PUBLICATIONS RECEIVED Anal. Proc. Hints are included to help the reader to over- come the difficulties that often arise from the lack of available equipment. Adequate theo- retical background of the individual techniques is also included and a theoretical approach to the deteriorative processes is presented in order to facilitate further development of a particular technique and its application to a special problem.In each chapter practical realisations of dj ff erent techniques are discussed and examples are presented to demonstrate the limits of each method. The mathematical and physico- chemical background is arranged so as to make it as coherent as possible for both non-pro- fessionals, such as postgraduate students, and experts using electromigration techniques. Topics in Current Chemistry, Volume 85. Instrumental Inorganic Chemistry. G. K. Wolf, R. W. Kiser and G. Schwedt. Pp. iv + 220. Springer-Verlag. 1979. Price DM98; $53.90. ISBN 3 540 09338 9; 0 387 09338 9. This book consists of three sections: Chemical Effects on Ion Bombardment , Doubly-charged Negative Ions in the Gas Phase and Chromato- graphy in Inorganic Trace Analysis, all with comprehensive reference lists.The Inorganic Trace Analysis section includes chapters on Gas Chromatography, Liquid Column Chromatography, Paper and Thin-layer Chromatography and Chromatographic Methods for Enrichment. Isoenzyme Analysis. D. W. Moss. Analytical Sciences Monogmafilzs No. 6. Pp. viii + 163. The Chemical Society, 1979. Price L9; $22 (CS Members fl6.75). ISBK 0 85186 800 2. The present recognition that enzymes frequently exist in multiple molecular forms has resulted from the application of a wide variety of experimental techniques. This monograph attempts to draw together the most important of these techniques and to indicate their advantages and limitations in isoenzyme studies.Those which have been developed primarily for, and find their main applications in, other or wider fields of research have not been described in detail ; more detailed attention has been given to principles and methods of analysis, such as separation procedures, which have played a particularly important part in isoenzyme investigations. Examples have been drawn most frequently from isoenzyme systems that occur in human tissues, as the clinical and genetic implications of the existence of iso- enzymes have provided much of the impetus for the development of appropriate analytical techniques, although the techniques themselves are of general applicability. This approach should provide those entering the field of iso- enzyme research with guidance in their selection of methods, together with access to sources of experimental detail.The term “isoenzynies” has been used throughout as a general description of multiple enzymes forms, except when current knowledge compels the exclusion of particular forms from the more restricted definition of this term. Similarly, enzymes are designated by their trivial names, although the identifying numbers allotted by the Enzyme Commission of the International Union of Biochemistry have also been given for enzymes that are referred to more extensively, usually when first mentioned. Biogeochemical Cycling of Mineral-form- ing Elements. Edited by P. A. Trudinger and D. J. Swaine. Studies an Environmental Science 3. Pp. viii + 612. Elsevier.1979. Price $97.50; Dfl200. ISBN 0 444 41745 1 (volume 3); ISBN 0 444 41696 X (series). The elements from which minerals are formed undergo continual cycling within the environ- ment, the cycles being influenced by a variety of factors. This book reviews current know- ledge of the major biological processes which are involved in these geochemical cycles and which influence, directly or indirectly, the formation, dissolution and transformation of minerals. Chapter 1 outlines some general aspects of the biogeochemical cycling of elements , while subsequent chapters deal specifically with the cycles of carbon, phosphorus, iron, manganese, sulphur, silicon and uranium. Two final chapters deal with agricultural aspects of element cycling and the future consequences of mineral exploitation.The book should be of value to geoscientists concerned with such diverse fields as ore genesis, environmental protection, mineral exploitation and extractive metallurgy, as well as to biologists interested in the metabolism of inorganic substrates and the influence of organisms on the geochemistry of the earth’s surface.148 PUBLICATIONS RECEIVED Anal. Proc. Hints are included to help the reader to over- come the difficulties that often arise from the lack of available equipment. Adequate theo- retical background of the individual techniques is also included and a theoretical approach to the deteriorative processes is presented in order to facilitate further development of a particular technique and its application to a special problem.In each chapter practical realisations of dj ff erent techniques are discussed and examples are presented to demonstrate the limits of each method. The mathematical and physico- chemical background is arranged so as to make it as coherent as possible for both non-pro- fessionals, such as postgraduate students, and experts using electromigration techniques. Topics in Current Chemistry, Volume 85. Instrumental Inorganic Chemistry. G. K. Wolf, R. W. Kiser and G. Schwedt. Pp. iv + 220. Springer-Verlag. 1979. Price DM98; $53.90. ISBN 3 540 09338 9; 0 387 09338 9. This book consists of three sections: Chemical Effects on Ion Bombardment , Doubly-charged Negative Ions in the Gas Phase and Chromato- graphy in Inorganic Trace Analysis, all with comprehensive reference lists.The Inorganic Trace Analysis section includes chapters on Gas Chromatography, Liquid Column Chromatography, Paper and Thin-layer Chromatography and Chromatographic Methods for Enrichment. Isoenzyme Analysis. D. W. Moss. Analytical Sciences Monogmafilzs No. 6. Pp. viii + 163. The Chemical Society, 1979. Price L9; $22 (CS Members fl6.75). ISBK 0 85186 800 2. The present recognition that enzymes frequently exist in multiple molecular forms has resulted from the application of a wide variety of experimental techniques. This monograph attempts to draw together the most important of these techniques and to indicate their advantages and limitations in isoenzyme studies. Those which have been developed primarily for, and find their main applications in, other or wider fields of research have not been described in detail ; more detailed attention has been given to principles and methods of analysis, such as separation procedures, which have played a particularly important part in isoenzyme investigations. Examples have been drawn most frequently from isoenzyme systems that occur in human tissues, as the clinical and genetic implications of the existence of iso- enzymes have provided much of the impetus for the development of appropriate analytical techniques, although the techniques themselves are of general applicability.This approach should provide those entering the field of iso- enzyme research with guidance in their selection of methods, together with access to sources of experimental detail. The term “isoenzynies” has been used throughout as a general description of multiple enzymes forms, except when current knowledge compels the exclusion of particular forms from the more restricted definition of this term.Similarly, enzymes are designated by their trivial names, although the identifying numbers allotted by the Enzyme Commission of the International Union of Biochemistry have also been given for enzymes that are referred to more extensively, usually when first mentioned. Biogeochemical Cycling of Mineral-form- ing Elements. Edited by P. A. Trudinger and D. J. Swaine. Studies an Environmental Science 3. Pp. viii + 612. Elsevier. 1979. Price $97.50; Dfl200. ISBN 0 444 41745 1 (volume 3); ISBN 0 444 41696 X (series). The elements from which minerals are formed undergo continual cycling within the environ- ment, the cycles being influenced by a variety of factors.This book reviews current know- ledge of the major biological processes which are involved in these geochemical cycles and which influence, directly or indirectly, the formation, dissolution and transformation of minerals. Chapter 1 outlines some general aspects of the biogeochemical cycling of elements , while subsequent chapters deal specifically with the cycles of carbon, phosphorus, iron, manganese, sulphur, silicon and uranium. Two final chapters deal with agricultural aspects of element cycling and the future consequences of mineral exploitation. The book should be of value to geoscientists concerned with such diverse fields as ore genesis, environmental protection, mineral exploitation and extractive metallurgy, as well as to biologists interested in the metabolism of inorganic substrates and the influence of organisms on the geochemistry of the earth’s surface.148 PUBLICATIONS RECEIVED Anal.Proc. Hints are included to help the reader to over- come the difficulties that often arise from the lack of available equipment. Adequate theo- retical background of the individual techniques is also included and a theoretical approach to the deteriorative processes is presented in order to facilitate further development of a particular technique and its application to a special problem. In each chapter practical realisations of dj ff erent techniques are discussed and examples are presented to demonstrate the limits of each method.The mathematical and physico- chemical background is arranged so as to make it as coherent as possible for both non-pro- fessionals, such as postgraduate students, and experts using electromigration techniques. Topics in Current Chemistry, Volume 85. Instrumental Inorganic Chemistry. G. K. Wolf, R. W. Kiser and G. Schwedt. Pp. iv + 220. Springer-Verlag. 1979. Price DM98; $53.90. ISBN 3 540 09338 9; 0 387 09338 9. This book consists of three sections: Chemical Effects on Ion Bombardment , Doubly-charged Negative Ions in the Gas Phase and Chromato- graphy in Inorganic Trace Analysis, all with comprehensive reference lists. The Inorganic Trace Analysis section includes chapters on Gas Chromatography, Liquid Column Chromatography, Paper and Thin-layer Chromatography and Chromatographic Methods for Enrichment. Isoenzyme Analysis.D. W. Moss. Analytical Sciences Monogmafilzs No. 6. Pp. viii + 163. The Chemical Society, 1979. Price L9; $22 (CS Members fl6.75). ISBK 0 85186 800 2. The present recognition that enzymes frequently exist in multiple molecular forms has resulted from the application of a wide variety of experimental techniques. This monograph attempts to draw together the most important of these techniques and to indicate their advantages and limitations in isoenzyme studies. Those which have been developed primarily for, and find their main applications in, other or wider fields of research have not been described in detail ; more detailed attention has been given to principles and methods of analysis, such as separation procedures, which have played a particularly important part in isoenzyme investigations.Examples have been drawn most frequently from isoenzyme systems that occur in human tissues, as the clinical and genetic implications of the existence of iso- enzymes have provided much of the impetus for the development of appropriate analytical techniques, although the techniques themselves are of general applicability. This approach should provide those entering the field of iso- enzyme research with guidance in their selection of methods, together with access to sources of experimental detail. The term “isoenzynies” has been used throughout as a general description of multiple enzymes forms, except when current knowledge compels the exclusion of particular forms from the more restricted definition of this term. Similarly, enzymes are designated by their trivial names, although the identifying numbers allotted by the Enzyme Commission of the International Union of Biochemistry have also been given for enzymes that are referred to more extensively, usually when first mentioned.Biogeochemical Cycling of Mineral-form- ing Elements. Edited by P. A. Trudinger and D. J. Swaine. Studies an Environmental Science 3. Pp. viii + 612. Elsevier. 1979. Price $97.50; Dfl200. ISBN 0 444 41745 1 (volume 3); ISBN 0 444 41696 X (series). The elements from which minerals are formed undergo continual cycling within the environ- ment, the cycles being influenced by a variety of factors. This book reviews current know- ledge of the major biological processes which are involved in these geochemical cycles and which influence, directly or indirectly, the formation, dissolution and transformation of minerals. Chapter 1 outlines some general aspects of the biogeochemical cycling of elements , while subsequent chapters deal specifically with the cycles of carbon, phosphorus, iron, manganese, sulphur, silicon and uranium. Two final chapters deal with agricultural aspects of element cycling and the future consequences of mineral exploitation. The book should be of value to geoscientists concerned with such diverse fields as ore genesis, environmental protection, mineral exploitation and extractive metallurgy, as well as to biologists interested in the metabolism of inorganic substrates and the influence of organisms on the geochemistry of the earth’s surface.

ISSN:0144-557X    

DOI:10.1039/AP9801700147

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

NEW BRITISH STANDARDS 151 Conferences and Meetings Modern Analytical Instrumentation for Quality Control in Foundries M a y 8-9, 1980, Coventry Programmes and booking forms are now avail- able for BCIRA’s forthcoming Seminar on Modern Analytical Instrumentation for Quality Control in Foundries. BCIRAs Analytical Seminar will present the latest developments in traditional techniques such as optical emission, X-ray fluorescence and atomic-absorption spectroscopy, and thermal analysis. Trace analyses for elements such as lead are of increasing importance so that the new techniques of inductively coupled plasma torch and glow discharge will be discussed. These latter techniques, which will be available during the next 5 years, offer simultaneous multi-element analysis for both trace (sub-parts per million) and macro (per cent .) concentrations in such diverse materials as iron, ferro-alloys, slag and environmental samples.An important part of the Seminar will be the equipment display by leading manufacturers of analytical instruments. Copies of the programme and booking form are available from the Head of Information Services, BCIRA, Alvechurch, Birmingham, B48 7QB. Tel.: 0527-66414. Safe Food for All June 29-July 3, 1980, West Berlin This meeting represents the first World Congress on Foodborne Infections and Intoxications and is being co-sponsored by the International Association of Microbiological Societies, the International Epidemiological Association and the World Association of Veterinary Food Hygienists. The working groups will consider the following topics: Public Health Aspects of Food Hygiene ; Etiological Agents and Toxic Substances ; Epidemiology and Surveillance ; Prevention and Control ; and Special Regional Problems.The Etiological Agents and Toxic Substances section will include consideration of Diagnostic and Analytical Methods. In addition to the scientific programme there will be visits to research institutes, mass catering establishments, etc., a social pro- gramme and an exhibition. The venue of the Congress will be the recently constructed International Congress Center, Berlin.152 CONFERENCES AND MEETINGS Anal. Proc. For information on the scientific programme The address for correspondence is Euro- write to Generalsekretariat Weltcongress, c/o analysis IV, Association of Finnish Chemical Institut fur Veterinarmedizin, Thielallee 88-92, Societies, Mr.Veikko Velamo, Executive Postfach 33 00 13, D-1000 Berlin 33, and for Director, Pohj. Hesperiankatu 3 B 10, SF-00260 general information contact German Con- Helsinki 26, Finland. vention Service, Joachimstaler Strasse 19, D-1000 Berlin 15. Senilab ’81 June 2-5, 1981, London Semlab is an international scientific, educa- tional, medical and industrial laboratory equip- ment exhibition which, in 1981, will be held in the Grand Hall at Olympia. Among the products represented at the exhibition will be analytical instrumentation, biomedical equip - ment, chemicals and reagents, computers and data processing, diagnostic equipment, electronic test and measurement equipment, electrical equipment, environmental and safety equipment, furnishings and fittings, general laboratory apparatus, glassware, laboratory services and optical equipment, including microscopes. There will also be a full schedule of seminars run during the exhibition. For further details contact Industrial and Trade Fairs Ltd., Radcliffe House, Blenheim Court, Solihull, West Midlands, B91 2BG.Euroanalysis IV August 23-28, 1981, Helsinki, Finland This conference is to be held on the Otaniemi campus of Helsinki University of Technology, which is situated in the town of Espoo. All those intending to participate in the conference are invited to submit papers to the Scientific Committees. The following is a list of possible topics, although papers on other areas of analytical chemistry will be considered : atomic spectrometry ; automated techniques ; chroma- tography; clinical analysis; computers in analytical chemistry ; electroanalytical chemis- try ; emission spectroscopy ; environmental analysis ; ESCA and related techniques ; geo- chemical analysis ; food analysis ; forensic science ; kinetic methods of analysis ; materials science ; mass spectrometry ; microchemical techniques ; molecular spectrometry ; nuclear techniques ; pharmaceutical analysis ; photo- metric analysis ; polymer analysis ; reference materials ; spectroscopic techniques ; thermal analysis ; trace analysis ; and X-ray techniques. There will also be two special sessions, “Analyti- cal Chemistry, the Analyst and Society” and “Mass Spectrometry in Inorganic Analysis,” and a special poster session on “Symbolism in Analytical Chemistry.”

ISSN:0144-557X    

DOI:10.1039/AP980170151c

出版商:RSC    

年代:1980

数据来源: RSC