ISSN:0144-557X    

DOI:10.1039/AP99128FX013

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANPRDI 28(4) 97-136 (1991) Proceedings of the Analytical Division of The Royal Society of Chemistry 97 Annual General Meeting of the Analytical Division 98 Analytical Division Distinguished Service Award 'Chemically and Biologically Tailored Interfaces as Amperometric Sensors' by Joseph Wang 100 European Analytical Column 15 101 Analytical Division Honours 102 SUMMARIES OF PAPERS 102 Sensors and Signals 102 1 04 106 110 115 117 'Conducting Polymer-based Amperometric Chemical Sensors' by Michael E. G. Lyons, Cormac H. Lyons, Declan E. McCormack,Thomas J. McCabe, William Breen and John F. Cassidy 'Ion-selective Electrodes in Clinical Chemistry: State of the Art' by Andrzej Lewenstam 'Some Artificial Intelligence Techniques for the Interpretation of Experimental Data' by Philip G. Barker 'Chemometrics-the Key t o Sensor Array Development' by Stephen J. Haswell and Anthony D. Walmsley 'Multivariate Calibration of Potentiometric Sensor Arrays' by Robert J. Forster and Dermot Diamond 'Recent Developments in Gel Electrophoresis of Proteins' by Michael J. Dunn 'Validity of Analytical Measurements' by Bernard King and Geoffrey Phillips , - - - 123 New Horizons in Electrophoresis 123 125 VAM Programme 125 127 Equipment News 132 BCR Call for Proposals 132 The Heinz-Zumkley Prize 1991 132 Death 133 Conferences and Meetings 135 Courses 136 Analvtical Division Diarv Typeset and printed by Black Bear Press Limited, Cambridge, England April 1991 Analytical Proceedings CONTENTS

ISSN:0144-557X    

DOI:10.1039/AP99128BX015

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 97 Annual General Meeting of the Analytical Division The nineteenth Annual General Meeting of the Analytical Division of The Royal Society of Chemistry was held at 4.45 p.m. on Friday, March lst, 1991, at the University of Wales College of Cardiff. The Chair was occupied by the President, Professor J . D. R. Thomas, DSc, CChem, FRSC. The Report of the Council for the year ending March, 1991, was presented by the President and adopted. The scrutineers, Dr. N. K. Harris and Mr. A. R. Phillips, reported that no ballot had been required for the Officers of the Division and the following would there- fore serve for the coming year- President-J. D. R. Thomas, DSc, CChem, FRSC. President- Elect-E. J. Newman (Chair- man of the Group Liaison and Policy Committee).immediate Past President-D . Thorburn Burns (Chairman of the Northern Ireland Region). Vice- Presidents-H. I . Shalgosky and J . M. Warren. Honorary Treasurer-T. B. Pierce. Honorary Secretary-C. A. Watson. Honorary Assistant Secretary-F. W. Sweeting (Programmes Secretary). Other Members of Council-The Scru- tineers further reported that 600 valid ballot papers had been received, and that votes had been cast in the election of Ordinary Members of Council as follows: M. C. Finniear, 363; D . E. Games, 357; L. A. Gifford, 353; S. J . Hill, 347; J. Marshall, 416; C. W. M. McLeod, 378; R. M. Miller, 404; J . M. Slater, 248; P. J . Stevens, 405; M. Thompson, 356. The President declared the following to have been elected Ordinary Members of Council for the ensuing 2 years: M.C. Finniear, D. E. Games, L. A. Gifford, J. Marshall, C. W. M. McLeod, R. M. Miller, P. J . Stevens and M. Thompson. Council for 1991/2. E. R. Adlard (Chairman o f t h e North West Region), P. G. W. Cobb (Chairman of the SAC 92 Executive Committee), Professor A. Townshend (R) receives the nineteenth SAC Gold Medal from the President of the Analytical Division, Professor J . D. R . Thomas D. Betteridge, A. G. Fogg (Chairman G. C. S. Collins (Chairman of the Western of the Analytical Editorial Board), J. G. Region), C. S. Creaser (Chairman of the Jones, G. F. Phillips, D. C . M. Squirrell, East Anglia Region), N. T. Crosby P. G. Takla, T. S. West and B. W. (Chairman of the Analytical Methods Woodget, having been elected Members Committee), C. L.Graham (Chairman of of the Council in 1990, will, by the Rules the Midlands Region), J. D. Green of the Division, remain Members of the (Honorary Publicity Secretary and Chair- Mr. R. Suwyer receives the twenty-first Analytical Division Mr. F. W. Sweeting receives the twenty-second A D Distinguished Distinguished Service A ward Service A ward98 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 man of the North East Region), D. W. Houghton (Chairman of the South East Region), A. D. Ruthven (Chairman of the Scottish Region), G. Svehla (Chairman of the Analytical Abstracts Editorial Com- mittee), A. Townshend (Chairman of Programmes Committee), and to be appointed (Chairman of the Education and Training Committee) will be ex officio Members of the Council for 1991/2. The ted to Professor A. Townshend, and the Honorary Officers and new Members of twenty-first and twenty-second RSC the Council will assume their duties on Analytical Division Distinguished Service July 18th. Awards were presented to Mr. R. Sawyer The Annual General Meeting was and Mr. F. W. Sweeting. Short biogra- preceded by a meeting entitled ‘Gas and phies of the recipients are published Headspace Vapour Analysers’. In the below. evening, at the Informal Dinner, the nineteenth SAC Gold Medal was presen-

ISSN:0144-557X    

DOI:10.1039/AP9912800097

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

98 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Analytical Division Distinguished Service Award As reported in the July 1990, issue (p. 165), the twenty-first and twenty-second Analytical Division Distinguished Service Awards have been conferred on Mr. R. Sawyer and Mr. F. W. Sweeting. Ron Sawyer is a native of Sheffield, although he has spent most of his adult life in the South of England. His formal education started at New Mills County Secondary School in north Derbyshire and continued through the second world war at the City Grammar School in Sheffield. He continued in the Honours Chemistry School at Sheffield University under the late Professor R. D. Haworth. As a secondary subject he studied Fuel Technology at St. Georges Square, where the name of Ronald Belcher was some- thing of a legend even in those early days.The late Drs. T. B. Smith and A. E. Beet were the respective analytical mentors in the Chemistry and Fuel Technology Departments. From Sheffield he joined the Depart- ment of the Government Chemist in London as a Scientific Officer and spent a two year sojurn in the RASC on National Service, most of which was spent in Ismalia in the Suez Canal Zone as assis- tant to the War Department Analyst (Foods), at that time J . H. A. (Joe) Ruzicka who was also a member of the staff of the Laboratory. Thus the seeds of his later career with LGC were sown. He served in various Divisions of the Labora- tory and concluded his full-time career as Superintendent of the Food and Nutrition Division with responsibility as Super- intendent Chemist Armed Services Food Supplies and Technical Advisor to the Food Procurement Division of the RNSTS.In the latter capacity he also served on a number of Official Advisory Committees, including the Food Addi- tives and Contaminants Committee and later the Food Advisory Committee. He retired from the LGC in 1985. During the course of his career he published over 60 scientific analytical papers and presented an uncounted num- ber of papers at scientific conferences at home and abroad, covering a variety of interests particularly in automated analy- sis and the analysis of foods and beverages. With Harold Egan and Ron Kirk he undertook a major revision of ‘The Analysis of Foods’ by D. Pearson (formerly by H. E. Cox); this appeared in 1981 as the Eighth Edition of ‘Pearson’s Analysis of Foods’.The Ninth Edition is due for publication in 1991. His interests in official life were also represented in activities with the SAC, which he joined in 1956, and later the Analytical Division of the RSC. He was a founder member of the Automatic Methods Group and served that group as Treasurer and Secretary; he was variously a member, Assistant Secretary, Vice- Chairman and Chairman of the Micro- chemical Methods Group, a founder member of the South East Region Com- mittee and later Vice-chairman and Chairman. In 1973 he served his first term on Council and as a member of the Centenary Executive Committee he was blooded in the wider work of the Divi- sion. He served on Council almost con- tinuously from 1977 until 1989 in various capacities, including the offices of Honor- ary Secretary and Vice-president.During these years he served on all the commit- tees of Council and on the Analyst Publi- cations Committee and Editorial Advi- sory Board. His last appointment as Honorary Assistant Secretary was brought to an early conclusion by a serious illness in 1989, followed by an operation in 1990. Since retirement, and until curtailed by illness, he has acted as a part-time tutor in business computing and a freelance soft- ware programmer. Currently, he has revived this interest and, on a part-time basis, is lecturing and tutoring staff from Local Authorities and local businesses at Northbrook College of Technology, Worthing. On the domestic front he spends a deal of time with his ever-expanding brood of grandchildren. As part of his recuperation from the operation early in 1990, he has recently, together with his wife Barbara, visited his son, daughter-in-law and two grandchildren in North America and attended his elder daughter’s wedding in Brunei.This latter event provided a reason for an extended holiday in the Far East, covering Thailand, Mainland Malaysia, East Malaysia, Brunei and Hong Kong. Frank Sweeting was born and brought up in Buckhurst Hill on the Central Line and he can still remember the use of steam engines when he went to watch Isthmian League football at Leytonstone FC. He went to school at Buckhurst Hill County Primary and High Schools and continued his education at the then South East Essex Technical College in Long- bridge Road, Dagenham, where he enrolled on a four year sandwich course leading to Graduate Membership of the Royal Institute of Chemistry (RIC) by external examination.The concentrationANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 99 of varied industries in the area which included such diverse organizations as May and Baker, Howards of Ilford, and Hopkin and Williams, which were recog- nizable as companies based on chemistry, manufacturing metallurgists such as Murex of Rainham, which required many analysts to support the manufacturing process, and large manufacturers such as Ford (UK), which needed relatively small numbers of chemists in support functions such as QA paint shops and steam genera- tion, confirmed to him early on in life that chemistry really was a passport to any industry.His first full-time employment was in the Chemical Research Department at Allen and Hanbury’s at Ware in Hertford- shire, where he was busy synthesizing compounds to be screened for anti-infec- tive properties. He made use of a number of techniques such as IR, UV and NMR spectroscopy, thin-layer, column and gas chromatography and elemental analysis to separate, purify and establish o r con- firm the structure of the predicted (and sometimes unpredicted) reaction prod- ucts that he made. Most winter weekends found him play- ing soccer for the works team on Satur- days, initially at centre half and later as centre forward. On Sundays he changed games and turned out for the works hockey team. Deciding that the synthetic organic chemists who would progress had all done postdoctoral stints in the United States, he left ‘Allenbury’s’ after two years, and with his membership application to the RIC in the pipeline he moved to Trawsfyndd (nuclear) Power Station as a General Assistant Chemist for just 12 months, where he became interested in automatic methods (because chloride in boiler and boiler feed water is very important-and there are many boilers in a large generating station) and the accur- acy and precision of these analyses.Next he moved into the (at the time) recently reorganized Water Industry as Ben Deeble’s assistant at West Pennine Water Board in Oldham, and joined the Society for Water Treatment and Exami- nation before taking up the post of Chemist and Bacteriologist to the North West Sussex Water Board at Horsham, where he stayed for four years. It was here that he became very closely involved in water production from all sorts of sources including direct river abstraction, a period that led to a lasting interest in process chemistry, and he learned of ‘de- bottlenecking’ and other jargon.He was also elected to membership, as a Professional Associate, of the Institu- tion of Water Engineers while working in Horsham. In the laboratory he moved away from colorimetric methods for metal determi- nations and introduced AAS instead, including non-flame techniques such as hydride generation for species of interest. He also introduced QC charts so that accuracy and precision of analyses could be monitored and controlled, and ac- knowledges the work carried out by the late Tony Wilson to introduce and popu- larize statistical methods to the Water Industry.On the works he is particularly proud of his successful introduction of automatic de-sludging cones into the accelator set- tling tanks at the Hardham works in association with Bill Brignal of PCI. He also realized the limitations of the early on-line water quality monitoring stations in live pollution incidents. With the next reorganization of the Water Industry in 1974 he moved again, this time to Wessex Water in Bath, where he took on responsibilities for running Laboratory services for one third of the Region. This was a time of great excite- ment, helping to set up a Divisional Laboratory with much new equipment to extend the breadth of analytical tasks that could be undertaken: also to increase the throughput of the Laboratory very signifi- cantly without increasing staff numbers by introducing second and third generation automated analysers, and automatic sample introduction to analysers-all things that are now the rule rather than the exception.One of the things that most pleased him during this period was the general acceptance that interlaboratory comparability could only be achieved by properly designed control schemes such as those described by the Analytical Quality Control (Harmonized Monitor- ing) Committee. Much work was carried out at this time to allow the management of Water Qual- ity in the Severn Estuary. This required the Chemists Sub-committee to ensure comparability of results between the many laboratories of four Water Authori- ties for 17 determinands within three years.A project of this complexity had not previously been attempted in the UK water industry. The targets set for accu- racy and precision and the results achieved in this work were published in The Analyst. It was at this time also that his involve- ment with the Western Region of the Analytical Division began, initially sup- porting Gareth Jones at one o r two meetings and then being talent-spotted by Geoff Dickes to become Secretary. Joint roles of Secretary and Treasurer followed and currently he is Vice-chairman and Treasurer. Along the way he has been privileged to have served on the organizing committee of SAC 86, and to have been an ordinary Member of Council for a three year term from 1987.In 1980 he left his analytical responsi- bilities behind and started to become involved in pollution prevention and advice around the Severnside area with its concentration of Industry. This involved him in day to day contact with the polluting potential of large plants such as two artificial fertilizer plants, a major zinc smelter, a carbon black producer, many smaller companies and several large oil storage facilities. This experience only serves to confirm that an overview of ‘pollution’ is necessary. Separate regula- tory authorities for discharges to different environments is a nonsense! One also becomes aware of the interplay between company profit and their investment in production facilities and also the environ- ment. Latterly, in Wessex Water he has been the link-man between the Company and the University of Hull in a research project with Paul Worsfold aimed at developing chemistries for, and proving the viability of, quasi-continuous process analysers using FIA techniques in the water industry.This work was prompted by the poor performance of many of the instruments marketed at the time and the expansion of on-line analysis that legis- lation will bring and has brought. During the first hockey match of the 1987/88 season he noticed how cold his fingers had become during a very wet and windy match, and as the season went on he found knocks and bruises seemed to be taking longer to heal and he had chronic tendonitis in his right forearm. He visited his G P but no useful advice was forthcom- ing apart from ‘rest it’. He arranged for regular physiotherapy at the Bath Sports Injuries Clinic and as it had no effect he decided it must be a disease.His wife convinced their G P that he was ill and blood tests immediately showed ab- normal responses. Within 6 weeks he was diagnosed as suffering from schlero- derma, a little known and rarely diag- nosed chronic auto-immune system disease. Fortunately, he suffers relatively mildly but it has led to his retirement from100 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Wessex Water in April 1990 and a re- appraisal of his priorities. With more time, he is able to put more effort into the Division and currently Council has appointed him Acting Assis- tant Secretary (since confirmed by the AGM). In all his work for the Division he acknowledges the support of his wife Phyllis, who these days is ‘just’ a house- wife as their two sons Alan and Clive have taken themselves off to UEA to study History, and Robinson College, Cam- bridge to study Biochemistry, respec- tively, from autumn 1990. They look forward to their graduation in 1993. Away from the Division he played hockey winter and summer until the onset of his schleroderma and is still on his Club Committee. He regrets not having been able to play in the same side as both of his boys. He had played on the same side as each individually but the four years Alan was unable to play because of orthopaedic surgery spanned Clive’s debut in Club hockey and his own retirement from the game. His lifelong interest in motor sport, particularly on two wheels, sees him as newsletter writer and Competitions Com- mittee Chairman of the West Wilts Motor Club, one of the few clubs in the country with the ability to run Grand Prix Mot0 Cross. He has watched Grand Prix in at least half a dozen countries and has been ‘proud to be British’ when watching Britons clinch their World Champion- ships abroad.

ISSN:0144-557X    

DOI:10.1039/AP9912800098

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

100 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 European Analytical Column 15 Working Party on Analytical Chemistry (WPAWFECS) In addition to the publication of the ‘WPAC Recommendations for the Future in Education in Analytical Chemistry’ (see European Analytical Column 14) we are happy to inform you about very positive comments about this major WPAC activity from outside Europe (see editorial by G . Morrison, Analytical Chemistry, November 1, 1990) and about further actions. 1. The complete Special Session on Education will be documented in a con- vener-edited part of the proceedings of Euroanalysis VII (to appear in Mikro- chimicu Actu, 1991). 2. The WPAC Secretariat will stimulate activities in the field of new text-books of analytical chemistry in the ongoing period. 3.A high level postgraduate training system in Europe was proposed under the name ‘WPAC-Euro-Courses’. Final stat- utes are under discussion. 4. Further aspects of education in analytical chemistry at university level will be discussed in depth at the ‘Second FECHEM Conference on Education in Analytical Chemistry’ to be held in Prague, at the famous Charles University, in the beginning of September, 1992. Professor Stulik has already obtained support from the Czechoslovak Chemical Society, which will organize the confer- ence in connection with the 16th Inter- national Competition in Analytical Chemistry (ICAC 16). 5. We would like to thank Professor Krofta and his colleagues from the ICAC for their personal efforts to make the 15th ICAC 1990 a success and finish the column with their report.L. NIINIST~ and R. KELLNER Report of the Fifteenth International Competition in Analytical Chemistry From September 3rd to 7th, 1990, the Department of Analytical Chemistry of the Prague Institute of Chemical Technol- ogy again organized, after a two-year interval, the 15th ICAC (International Competition in Analytical Chemistry). The Competition was held under the sponsorship of the Rector of the Prague Institute of Chemical Technology, the Ministry of Education, Youth and Sports of the Czech Republic, and the Czecho- slovak Chemical Society. The Competi- tion is one of the events supported by the Working Party on Analytical Chemistry (WPAC) of the Federation of European Chemical Societies (FECS). The official opening of the Competition which took place in the Schooling Club was attended by the Chairman of the Working Party on Analytical Chemistry of FECS, Professor L.Niinisto, and its Secretary, Professor R. Kellner, the Rec- tor of the Prague Institute of Chemical Technology, Professor Cerny, and a rep- resentative of the Czechoslovak Chemical Society, Professor Horak. The Competition was attended by students from 24 universities and tech- nical universities, together with their teachers, from nine European countries: Poland ( l ) , USSR (4), GDR ( 5 ) , Hungary (9, Rumania ( l ) , Yugoslavia (l), Italy ( l ) , Holland (1) and Czechoslovakia ( 5 ) . The Competition involved a theoretical part and a practical one. In the theoretical part of the Competition the students were solving 25 tasks and problems covering the most widely used analytical methods.In the practical part they were determin- ing the contents of copper and nickel in a mixture by titration with chelaton 111. The time limit for solving the set task in each part of the Competition was 150 min. The official language of the Competition was English. The results of the respective parts of the Competition were evaluated by an Inter- national Jury formed by the university teachers who guided the individual teams. The Chairman of the Jury was Associate- Professor Pop1 from the Prague Institute of Chemical Technology, the Vice-Chair- man was Dr. F. Gerhartl (Arnhem, The Netherlands). Intermediate results, after being evaluated by the International Jury, were always made public on notice boards.In the individual competition Mr. Kotschy Andras won first place, Mr. Hantosi Zolt second place (both from Eotvos Lorand University in Budapest), and Mr. Granovski Alexander from the Moscow State University was in third place. In the team competition the placement was as follows: 1, Eotvos Lorand Uni- versity, Budapest; 2, Lajos Kossuth University, Debrecen; 3, Moscow State University, Moscow. Apart from the Competition, the organizers also provided for the partici- pants a social programme including a sightseeing tour of Prague, attendance of a performance of Laterna Magica, a social evening and a closing festive dinner fol- lowed by music and dancing. The participants highly appreciated the organization and the professional pro- gramme as well as the actual course of the Competition. ORGANIZATION COMMITTEE OF THE 1 5 ~ ~ ICACANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 101 Analytical Division Honours At its meeting held on December 14th, 1990, the Council of the Analytical Division approved the following recommendations from its Honours Committee. Nineteenth Society for Analytical Chemistry Gold Medal: Professor A. Townshend (University of Hull). Theophilus Redwood Lecturer for 1992: Professor A. Hulanicki (University of Warsaw). Schweizerische Chemische k Gesellschaft HELVETICA CHIMICA ACTA Subscription Still available Vol.74, 1991 sFr. 51 5.- + postage 32.- Europe sFr. 51 5.- + postage 50.- Oversea Please request our price list Reprinted edit ions Original editions VOlS 1-27 (1 91 8-1 944) VOIS 29-73 (1 946-1 990) Vol.28 Out of print Verlag Helvetica Chimica Acta Postfach CH-4002 Basel

ISSN:0144-557X    

DOI:10.1039/AP9912800100

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

102 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Sensors and Signals The following are summaries of six of the papers presented at a Joint Meeting of the Institute of Chemistry of Ireland, the Royal Society of Chemistry and the British Council, held on September 21st, 1990, in Dublin City University. Chemically and Biologically Tailored Interfaces as Amperometric Sensors* Joseph Wang Department of Chemistry, New Mexico State University, Las Cruces, NM 88003, USA The interest in developing small sensing devices for use in environmental monitoring, clinical assays or process control is growing rapidly. Amperometric sensors satisfy many of the requirements for such tasks, particularly owing to their inherent sensitivity, selectivity toward electroactive species, speed of response, size and simplicity.However, there are still problems of selectivity and stability associated with the utility of amperometric probes. In particular, co-existing surface- active materials or electroactive species may result in a degraded response or overlapping signals. Surface modifica- tion represents one promising avenue by which to address these stability and selectivity problems. A judicious design of new interfaces and surface microstructures can thus meet many sensing needs. Various approaches for using modified elec- trodes for amperometric sensing, based on the surface im- mobilization of various (active and passive) moieties, are currently being developed and assessed. Coverage of the sensing surface with an appropriate perm- selective film can greatly enhance the sensor selectivity and stability.This is accomplished by rejection from the surface of undesired species, while allowing transport of the analyte. An additional separation step is thus performed in situ on the electrode surface. Different avenues to control the access to the surface, based on various discriminative films, have been explored (Table 1). Improved selectivity is being accomplished Table 1 Discriminative coatings for amperometric sensors Transport mechanism Coating Size exclusion Cellulose acetate Base-hydrolysed cellulose acetate Polyaniline Poly(viny1pyridine) Poly(ester-sulphonic acid) Charge exclusion Nafion Polarity Phospholipid Mixed control Cellulose acetate-Nafion Cellulose acetate-poly(viny1pyridine) Reference 1 2 3 4 5 6 7 8 9 by taking advantage of analyte properties such as size, charge, shape or polarity.Of particular interest to us is the ability to manipulate the permeability of these coatings to meet specific sensing needs. For example, highly controllable size-exclusion films can be obtained by controlling the amount of charge consumed during an electropolymerization process.3 Such films also form an effective barrier to the transport of large surface-active materials, thus minimizing electrode poisoning problems. Additional sensing advantages can be achieved * Plenary Lecture. using multi-functional coatings, consisting of different overlaid permselective films8 or mixed layers.9 Also promising are sensor arrays, based on electrodes coated with different permselective films ( i .e . , tuned toward different analytes), and operated in connection with statistical pattern recognition methods. 1 0 Electrocatalysis represents another promising route to improve the response of amperometric sensors. Catalytic modified electrodes are used to lower substantially the overvoltage of numerous analytes (with otherwise slow elec- tron-transfer kinetics). Such electrodes rely on the surface immobilization of an appropriate redox mediator, which mediates the electron transfer between the analyte and the electrode. Recent work in this laboratory has illustrated the unusual catalytic reactivity and stability of ruthenium dioxide-modified electrodes. The resulting interfaces have been shown to be effective for facilitating the oxidation of hydroxylated compounds, such as carbohydrates, 11 alcohols12 or saccharide-related antibiotics.13 Sensitive measurements of such compounds have thus been accomplished with low and fixed (+0.4 V vs. Ag-AgCI) potential. Other catalytic surfaces have been successfully employed, including Meldola blue, cobalt phthalocyanine or mixed-valent ruthenium deposits, which greatly facilitate the sensing of NADH, thiols or hydrazines, respectively. 14-16 The incorporation of biocatalytic moieties, particularly enzymes, on amperometric electrodes can greatly enhance the specificity of electrochemical measurements. We have developed various novel approaches for fabricating fast-responding and sensitive enzyme electrodes. One scheme involves the incorporation of the enzyme within a rigid graphite epoxy matrix.17 This results in polishable ( i . e . , re-usable) enzyme probes, where the bulk of the electrode serves as a source (or reservoir) for the biocatalytic activity. Another promising avenue, particularly for fabricating miniaturized sensor surfaces, is to deposit the enzyme electrolytically (in the presence of platinum ions) on carbon fibre microelectrodes.18 We have recently incorporated such a glucose oxidase fibre electrode as a rapid on-valve flow-injection detector, which allows the detection of glucose while it is still in the injection valve with exceptionally high injection rates of a few thousand samples per hour.19 Such a detector is based on placing the enzyme-modified microfibre within the centre of the valve outlet (with the reference electrode located outside). Poly- (ester-sulphonic acid) films also permit convenient fabrication of enzyme electrodes.For example, the rapid entrapment of glucose oxidase within these negatively charged coatings offers a fast and sensitive response to glucose, while simultaneously rejecting anionic interferences (e.g. , ascorbic acid) and protecting the surface against large proteins (Fig. l ) . z O We are also exploring the incorporation of biocatalysts on electrodeANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 103 -7 I t I '1 "L I t I I Time - Fig. 1 Current-time recording on increasing ( a ) the glucose and ( b ) the ascorbic acid concentrations (in 1 and 0.1 mmol dm-3 steps, respectively), at (A) bare and (B) poly(ester-sulphonic acid)-glucose oxide-modified electrodes surfaces for in sctu elimination of potential interferences.21-23 Enzymes such as papain, uricase, ascorbic acid oxidase or tyrosinase have thus been immobilized and employed to address (via conversion to non-interfering species) the effects of albumin, uric and ascorbic acids or acetaminophen, respective 1 y .Surface characterization plays an important role in under- standing the behaviour and performance of tailored sensor surfaces. We have developed a high-resolution scanning electrochemical microscopic technique for characterizing biologically modified electrodes, based on a nearby microelectrode tip which probes the biological generation or consumption of electroactive species (Fig. 2). Two-dimensional X-Y images of the biological activity can be obtained by moving the tip over the surface.24 The technique can be employed also for exploring the kinetics of biological surface reactions.This is accomplished by repetitive and rapid measurements of the biologically generated (or consumed) marker, performed at a fixed position above the surface. The dynamics of metal uptake by alga-containing surfaces, of the hydrophobic partitioning of drugs into lipid layers or the biocatalytic activity of tissue surfaces, can thus be explored.25 Scanning tunnelling microscopy (STM), using a similar arrangement, represents another powerful tool for characteriz- ing sensor surfaces. In particular, recent advances permit STM imaging of electrode surfaces in solution under potentiostatic control. For example, high-resolution topographic images, provided by STM, offer a better understanding of sensor activation processes (e.g., electrochemical pre-treatments leading to fine changes in the surface roughness).26 Similarly, STM can be used to probe sensor passivation reactions, such as the formation of polymeric phenoxy films during anodic measurements of phenolic compounds.27 High-resolution map- pings of conductive sites of composite surfaces can also be obtained through tunnelling/out-of-tunnelling changes.28 Work is in progress employing STM and scanning electron microscopy for the microfabrication of biosensors via the high-resolution deposition of enzymes. This includes the formation of extremely narrow and pre-defined enzyme 'lines' during the controlled movement of the tip.A similar approach for depositing pre-defined membrane barriers is also being explored. Overall, electrochemical sensors based on modified elec- trodes are very attractive, as they couple the inherent sensitivity of amperometry with new dimensions of selectivity provided by the analyte-modifier interaction. Many exciting developments are expected based on this unique coupling, and the diversity of potential (biological and chemical) surface modifiers. Financial support from the American Chemical Society (Petroleum Research Fund) is acknowledged. References 1 Sittampalam, G.. and Wilson, G . S.. Anal. Chern.. 1983, 55. 1608. 2 Wang, J., and Hutchins. L. D., Anal. Chem., 1985, 57, 1536. 3 Wang, J . , Chen. S . P., and Lin, M. S . , J. Electroanal. Chem., 1989, 273, 231.4 Wang, J.. Tuzhi, P . . and Golden, T., Anal. Chirn. Acta, 1984, 194, 129. 5 Wang, J . , Golden, T., and Tuzhi, P.. Anal. Chern., 1987, 59, 740. Positioner P' P R Fig. 2 Scanning electrochemical microscopy of biological surfaces104 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 6 Wang, J., and Golden, T., Anal. Chem., 1989, 61, 1597. 7 Wang, J., and Lu, Z., Anal. Chem., 1990,62, 826. 8 Wang, J., and Tuzhi, P., Anal. Chem., 1986, 58, 3257. 9 Wang, J., andTuzhi, P., J. Electrochem. SOC., 1987, 134, 586. 10 Wang, J., Rayson, G., Lu, Z., and Wu, H., Anal. Chem., 1990, 62, 1924. 11 Wang, J., and Taha, Z., Anal. Chem., 1990,62, 1413. 12 Leech, D., Wang, J., and Smyth, M. R.. Electroanalysis. 1991, 3,37. 13 Leech, D., Wang, J., and Smyth, M. R., Analyst, 1990, 115, 1447.14 Gorton, L., Torstensson, A., Jalgfledt, J., and Johansson, G., J. Electroanal. Chem., 1984, 161, 103. 15 Halbert, M. K., and Baldwin, R. P., J. Chromatogr., 1986,345, 43. 16 Wang, J., and Lu, Z., Electroanalysis, 1989, 1, 517. 17 Wang, J., and Varughese, K., Anal. Chem., 1990, 62, 318. 18 Wang, J., Li, R., and Lin, M. S., Electroanalysis, 1989, 1, 151. 19 20 21 22 23 24 25 26 27 28 Wang, J., and Li, R., Anal. Chem., 1990, 62, 2414. Wang, J., Leech. D . , Ozsoz, M., Martinez, S., and Smyth, M. R., Anal. Chim. Acta. in the press. Wang, J., Wu, L., Martinez, S., and Sanchez, J., Anal. Chem., in the press. Wang, J., Naser, N . , and Ozsoz, M., Anal. Chim. Acta, 1990, 234, 315. Wang, J.. and Naser, N., in preparation. Wang, J., Wu, L., and Li, R., J.Electroanal. Chem.. 1989,272, 285. Wang, J., and Wu. L.. in preparation. Wang, J.. Martinez. T.. Yaniv, D. R.. and McCormick. L. D., J. Electroanal. Chem., 1990, 278, 379. Wang, J., Martinez, T., Yaniv, D. R., and McCormick. L. D., in preparation. Wang, J., Martinez, T., Yaniv, D. R., and McCormick, L. D., J. Electroanal. Chem., 1990, 286, 265. Conducting Polymer-based Amperometric Chemical Sensors Michael E. G. Lyons, Cormac H. Lyons, Declan E. McCormack," Thomas J. McCabe and William Breen* Physical Chemistry Laboratory, University of Dublin, Trinity College, Dublin 2, Ireland John F. Cassidy Chemistry Department, Dublin Institute of Technology, Kevin Street, Dublin 8, Ireland The design, fabrication and application of novel amperometric chemical and biological sensors has been a topic of consider- able interest in recent years.l.2 This research activity has occurred in tandem with rapid developments in the area of polymer-modified electrodes.3 In this paper we present aspects of our current work on the development of polymer-modified electrodes for use as amperometric chemical sensors.We describe two types of sensor electrode material: conducting polypyrrole and Nafion-metal oxide composites. In both instances the sensing material is cast on an inert support surface as a thin film. We illustrate the utilization of the sensors for the amperometric detection of ascorbate and catechol. A quantita- tive description of the operational mechanism of each of these sensor probes is also briefly presented. Polypyrrole- based Ascorbate Sensor Electronically conducting polymers such as polypyrrole (PP) offer much potential as materials for amperometric detection.They are readily synthesized via in situ electropolymerization, are highly conducting over a relatively wide potential window when in the oxidized state and can exhibit a very low background current if a suitable choice of dopant counter ion is made. Our studies have indicated that polypyrrole doped with the dodecylbenzenesulphonate ion (DBS-) exhibits excellent activity for the electro-oxidation of ascorbate4 when the reaction is carried out in 0.1 mol dm-3 NaCl. A rotating disc configuration is used in these experiments to ensure well defined hydrodynamics at the sensor surface. The PP-DBS- layer is readily formed on a glassy carbon support by potentiostatic electrodeposition [0-SO0 mV (versus Ag-AgCl)] from a solution containing pyrrole (50 mmol dm-3) and DBS- (0.1 mol dm-3) for a given period of time. The oxidation of ascorbate occurs via an electron transfer- chemical mechanism and is chemically irreversible.At an unmodified glassy carbon surface the redox behaviour of ascorbate is also electrochemically irreversible owing to sluggish electron transfer kinetics, and the quality of the electrochemical signal deteriorates rapidly owing to electrode fouling. A more satisfactory situation is observed at the * Also at the same address as J. F. Cassidy. PP-DBS- electrode, however, which is illustrated in the cyclic voltammogram outlined in Fig. l(a). The polymer is in an electronically conducting state for potentials more positive than -100mV.The current response for a 0.3 mmol dm-3 ascorbate sample sharply rises prior to the oxidation peak, indicating facile electrode kinetics at the oxidized polymer surface. Rotating disc and theoretical studies have indicated that the ascorbate does not partition into the polymer film but I 20 s w 0 0.5 EN 1 Fig. 1 ( a ) Cyclic voltammogram of a dilute ascorbate solution (0.3 mmol dm-3,O. 1 mol dm-3 NaCI) recorded at a sweep rate of l0mV s-1 at a PP-DBS- electrode. (b) Evaluation of the modified electrode response to a sequence of ascorbate injections in the amperometric mode. The stock ascorbate solution was 8 mmol dm-3 in 0.1 mol dm-3 NaC1, and the volumes added in sequence to the cell were 5 x 50 pI followed by 2 x 75 p1 simply reacts at the polymer-solution interface.4-5 Thus the sensor operates in a particularly simple manner.It is also of interest to note in Fig. l ( a ) the very low background current in the 0-100 mV region, which suggests that low concentrations of ascorbate may be readily determined amperometrically. The voltammogram also indicates that the reaction is under mass -ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 105 transfer control (current response proportional to bulk concen- tration of substrate) when E = 500 mV. The amperometric response [at E = 500mV (versus Ag-AgCl)] of the PP-DBS- sensor to various ascorbate injections is illustrated in Fig. l ( 6 ) . The response time is very rapid (typically 5 s) and a good steady-state current response without any evidence of electrode fouling is evident.Linear calibration graphs extending from 2 mmol dm-3 to about 5 pmol dm-3 were obtained. Further experiments have indicated that the polymer sensor may be used to determine readily and accurately the ascorbate content of commercial vitamin C tablets. The sensor configuration must be modified if it is used to monitor the vitamin C content in fruit juices, however. For this application a permselective membrane such as cellulose acetate or polyphenol must be attached to the polymer-coated probe. In this latter configuration the response time is lengthened but the probe functions well in practice, exhibiting little fouling and excellent sensitivity. We finally present some data of interest in bioelectro- chemistry and neurochemistry .Differential-pulse voltammo- grams of an ascorbate-dopamine mixture in phosphate buffer (pH 7.4) at an unmodified platinum and at a PP-DBS-- modified platinum electrode are illustrated in Fig. 2(a) and ( 6 ) . We note that only a single composite peak corresponding to the oxidation of the two analytes is observed using the unmodified electrode, whereas two distinct oxidation peaks are obtained for the polymer-modified electrode. The peak identification was readily accomplished by initially examining the pulse voltammetry of each component separately. We conclude, therefore, that the polymer layer exhibits some discrimination for the oxidation of dopamine and ascorbic acid. ~ ~~~~~ ~~ -0.15 0.4 -0.15 0.4 E N E N Fig. 2 Differential-pulse voltammogram of a mixture containing 0.4 mmol dm-3 dopamine and 1.2 mmol dm-3 ascorbate in phosphate buffer (pH 7.4) at ( a ) bare platinum and (6) PP-DBS- coated platinum electrodes.Sweep rate, 10 mV s-1; pulse amplitude, 50 mV. In (b) the first wave corresponds to ascorbate and the second corresponds to dopamine Metal Oxide-Nafion Composites for Catechol Detection Conductive metallic oxide electrodes such as Ru02-coated titanium have found wide application in the area of electro- catalysis,h because the outermost region of the oxide surface is hydrated; the catalytically active centres are surface bound oxymetal groups which act as binding centres for solution- phase substrates.’ Nafion, an ionicaliy conductive perfluori- nated polymer, has been used as a matrix for the incorporation of various electroactive substances to form novel classes of chemically modified electrodes.8 We now present the results of preliminary work carried out on composites of these two interesting materials, and specifically describe Ru02-Nafion composite-modified electrodes for catechol detection .g The procedure for preparing the composite sensor has been reported elsewhere.1O The sensor consisted of an oxide-loaded (3% RuOz) thin Nafion film (the latter formed from a 1% Nafion solution in ethanol, layer thickness 1.5 pm) deposited on a glassy carbon support electrode. The voltammetric response of a 10 mmol dm-3 catechol solution at the Ru02-Nafion composite in 0.2 mol dm-3 H2S04 is shown in Fig. 3(a). The redox chemistry is quasi- reversible, as implied by the large peak separation between the anodic and cathodic peaks (LIE, = 325 mV at 20 mV s-l).The heterogeneous electrochemical rate constant for catechol electro-oxidation has been determined from rotating disc voltammetry to be in the region of 5.3 X 10-3 cm s-1, which confirms the assumption of quasi-reversibility. The most important point to note from Fig. 3(a) is that catechol oxidizes a) I I I 0 500 1000 NmV vs. SCE 6 12 (SJmrnol dm-3)-1 Fig. 3 (a) Cyclic voltammetric response of a dilute catechol solution (0.8 mmol dm-3,0.2 mol dm-3 H,SO,) at an Ru02-Nafion composite modified electrode. Sweep rate, 20 mV SKI. (b) Plot of Koutecky- Levich intercept versus inverse substrate concentration. Data obtained from rotating disc experiments conducted on a series of dilute catechol solutions in the potential region where the RuIV-RuVI redox chemistry predominates.6 Catechol oxidation is a 2e-, 2H+ process.One therefore has a match between the oxyruthenium surface electrochemistry and the catechol oxidation process, and one may obtain mediation of the substrate oxidation via the surface-bound oxyruthenium groups. The oxyruthenium redox chemistry may be described in terms of the following expression : (-0-)2R~(OH)2(OH2)2 = (-O-)2Ru(OH)4 + 2H+ + 2e- RuIV Ru v1 The mediation occurs via the RuV’ group as follows: RuV’ + S = RuIV + P where S and P denote substrate and product, respectively. This mediation process may be described in terms of Michaelis- Menten kinetics,’ 1 where a precursor and a successor complex are formed as part of the reaction sequence.We propose the following reaction mechanism: kD K kC S(m) --+ S(a) S(ads) + P(ads) bulk interface adsorbed solution region state where kD represents a rate constant defining the rate of material transport to the electrode surface, K is the adsorption equilibrium constant and k, is a first-order rate constant for the substrate-product transformation in the adsorbed state. The latter reaction scheme may be confirmed’ using rotating disc voltammetry where a plot of the Koutecky-Levich intercept I K L versus inverse substrate concentration should be linear, with a positive slope and a positive intercept. The latter analysis is the electrochemical equivalent of the well estab- lished Lineweaver-Burk plot in enzyme kinetics.Such a plot is shown in Fig. 3(b). The relevant theoretical equation corre- sponding to the plot in Fig. 3(b) is ZKL = (llnFKkcT) S,-l + l/nFk,T106 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Given that the surface coverage of oxyruthenium sites, = 6.5 X 10-9 mol cm-2 (obtained by integration of the cyclic voltammetric response at 20 mV s-1 in 0.5 mol dm-3 H2S04), we obtain from the intercept in Fig. 3(b) a value of k, = 2.85 s-1. This corresponds to a relatively facile transformation of substrate to product. The adsorption equilibrium constant K is obtained from the slope of the plot in Fig. 3(b) as K = 0.35 dm3 mmol-1. For low values of substrate concentration we may show that9 the heterogeneous electrochemical rate constant is given by kME = KkJ (2) Inserting the numerical values obtained for r, K and k , yields k, = 6.5 x 10-3 cm s-1.Again, this value is characteristic of quasi-reversible electron-transfer kinetics. Hence we con- clude that the kinetics of catechol oxidation on the composite electrode are sufficiently rapid for the electrode to be a good detector in an amperometric sensor device. Further work is in progress to examine the full potential of these oxide-ionomer composite materials as amperometric sensors for substrates of biological interest. 1 2 3 4 5 6 7 8 9 10 11 References Borman, S . , Anal. Clzem.. 1987, 59, 1091A. J . Chem. Soc., Furaduy Truns. I , 1986, 82, 1033-1270; Special Issue, New Electrochemical Senyors. Hillman, A. R., in Electrochemical Science and Technology of Polymers, ed.Linford, R. G., Elsevier Applied Science, Barking, 1987, pp. 103-291. Lyons, M. E. G., Breen, W., and Cassidy, J . F., J . Chern. Soc., Faraduy Truns., 1991. 87, 115. Lyons, M. E. G., Bartlett, P. N.. Lyons, C. H.. Breen. W., and Cassidy, J . F., J . Electrounul. Chem., in the press. Lyons. M. E. G.. and Burke, L. D., J . C‘hem. Soc., Faruduy Tram. I . 1987, 83. 299. Burke, L. D., and Lyons, M. E. G., Mod. A.\pect.\ Electro- chern.. 1986. 18, 169. Whiteley. L. D., and Martin, C. R . , J . Phys. Chern., 1989. 93. 4650. Lyons, M. E. G., Lyons, C. H . , and Michas. A . , J . Efectroanal. Chem., submitted for publication. McCormack, D. E., PhD Thesis, University of Dublin, 1990. Albery, W. J., and Bartlett. P. N., J . Chem. Soc., Chem. Cornmun.. 1984, 234.lon-selective Electrodes in Clinical Chemistry: State of the Art Andrzej Lewenstam Laboratory of Analytical Chemistry, Abo Akademi University, SF-20500 Turku-Abo, Finland Electrochemical methods have attracted the interest of work- ers in clinical chemistry together with the growing demand from medical and hospital laboratories for fast, reliable, inexpensive and fully automated analyses. A breakthrough in the clinical application of electrochemical sensors came in 1966 with the pioneering invention of a valinomycin-based potass- ium electrode by Stefanac and Simon. The subsequent development of new generations of ion-selective (ISE) and gas-sensitive electrodes supplemented recently with a rapidly growing family of electrochemical biosensors provided ana- lytical tools that permit the specific and precise determination of a number of species of medical importance.As a conse- quence of these trends, a number of commercial analysers based on electrochemical methods became available and for this reason the determination of pH, blood gases (PO?, pCO?), electrolytes (Na+, K+, C1-, Ca2+, Li+, Mg”) and some metabolites (glucose, urea, uric acid, lactate) is now predomi- nantly performed with the use of electrochemical sensors. Extensive reviews of existing and potential applications of electrochemical techniques in clinical chemistry have recently appeared.1-3 Different aspects connected with the use of ion-selective electrodes for clinical purposes have also been discussed.- The status of electrochemical methods with emphasis on the challenges and problems which they are imposing on both the analytical chemist and the clinician has recently been discussed.7 This paper provides a general overview of the present status of ion-selective electrodes employed in commercially available electrolyte analysers.Special attention is paid to recent developments and some analytical and clinical difficulties originating from the use of ion-selective electrodes. Direct and Indirect Measurement-Xhemical and Medical Terminology Prior to any further description and to allow for a logical coherence, it is necessary at this point to stress a very basic terminological difference in the language of analytical and clinical chemistry, concerning the meaning of direct and indirect measurement. From the medical point of view, direct measurement is defined as a measurement carried out directly in an undiluted sample (whole blood, plasma, serum or urine), whereas indirect measurement is undertaken with sample dilution.In the chemical sense, direct measurement takes place when the sensor (electrode) is directly sensing the substance of interest with resulting Nernstian behaviour, i.e., with a linear dependence of electrode potential on the logarithm of the activity of the analyte. Hence direct measurement is possible only with use of a sufficiently specific or selective electrode. An indirect measurement takes place when Nernstian behaviour of the sensor (electrode) is obtained only if a special procedure (mathematical, physical and/or chemical) of signal trans- formation is introduced.This type of measurement has to be performed if the electrode is not selective enough or is not directly sensing the analyte. From these simple definitions, it is obvious that what is considered a direct measurement in the medical sense is not necessarily direct in the chemical sense and vice versa. It is beyond doubt that from both the chemical and clinical points of view a direct determination is the most beneficial as one leaving both the sample and over-all signal ‘untouched’. Practical Problems Arising from the Application of Electrochemical Methods in Clinical Analysers Apart from the various classifications, it is important at this point to reiterate the basic goals of chemical and clinical analysis. There is no doubt that an analytical chemist should provide a working method that satisfies the demands of a clinician and conforms with the requirements of analytical chemistry.On the other hand, the clinical chemist has to satisfy the expectations of a physician. The numerical result of an assay has to be presented in such a way that supports the final diagnosis. This means that the clinical chemist is not only constrained by the chemical measurement and its interpreta- tion but also by the particular-in the hospital or generally in medicine-me thods of sample collection, preservation, stor- age and handling and also the traditions in data interpretation specific to his or her field. The areas of action of these two chemists, including professional standards (and unfortunately sometimes ‘common sense’), do not always overlap.However, if the analyticalANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 107 chemist is not able to give an explicit resolution to the clinical problem in chemical terms, there is little alternative but to recognize the fact that ultimately clinical chemistry is serving the patient. Consequently, one has to compromise on trans- forming the analytical method to conform with the knowledge and traditions built up in clinical chemistry. This problem is well illustrated by the chemical and clinical aspects of the determination of sodium and potassium in blood. Ion-selective Electrodes-Direct Determinations in a Chemical Sense Measurement of Sodium and Potassium The most routine method for the determination of these two ions in clinical samples is flame emission spectrometry (FES).In the medical sense it is an indirect method, although direct in the chemical sense. As a consequence of the concept of the method, the measured concentrations of sodium and potass- ium are related to the total volume of the blood sample, including insoluble proteins, lipids and crystalloids (soluble salts, sugars, etc.). The development of sodium and potassium ion-selective electrodes challenged FES methodology. With these electro- chemical tools it is possible to measure directly (in both chemical and medical senses) the activity of sodium and potassium in water fractions of plasma instead of a total volume o f plasma. From the chemical point of view it would be rational to say that because activity is a driving force in physiological processes, one should interpret the results from now on by use of the activity scale rather than in terms of the concentration of diluted sample.However, to make such a proposal one should have an unambiguous activity scale for these two ions in plasma and also the power to convince physicians that the traditional way of reporting on sodium and potassium in blood is outdated. Despite advanced work on the thermodynamic interpreta- tion of electrolytes in blood, there is insufficient reason to propose the use of the activity scale in clinical chemistry and in this situation a compromise had to be proposed. Two possible resolutions were outlined: one was an introduction of the indirect method in a medical sense, i.e., sample dilution with direct measurement of sodium and potassium by use of ion-selective electrodes, and the other the direct method in both senses amended with the imposition of a range of overlap of this and the FES method.The first approach involved the traditional way of data interpretation. In the second, the adoption of calibrators to match the concentration of sodium and potassium in blood plasma with a normal content of plasma water, 0.93 kg I - l , was necessary.9 As a consequence of this strategy, naturally a discrepancy appeared between FES and direct measurements for samples diverging in water content, i.e., the samples with abnormal concentrations of proteins and/or lipids. This last outcome revealed the possibility of clinical misinter- pretation in abnormal patient samples and was in fact an argument f o r the use of direct measurements with ion-selective electrodest" (see Table 1).However, any further step to broaden the use of ion-selective electrodes demands more careful inspection of calibration procedures and the develop- ment of reference materials and methods. The independent measurement of water content may still be necessary, as advocated recently.11 The problem of calibration, reference materials and reference methods In the mid-1970s it became evident that the direct, in both chemical and medical senses, measurement of sodium and potassium with ion-selective electrodes was feasible. A few companies constructed ISE-based clinical analysers and started sales. However, it soon became apparent that there was no reference material or recommendation on which the manufac- turers could rely in quality control.It was not surprising that the comparison of performance undertaken with different commercial analysers using the same serum-based material showed a scatter among reported values of more than 6 mmol I- 1 for sodium coneentration.7 This situation resulted in a clear need for rectification and redefinition of the procedures for the clinical assessment of sodium and potass- ium. The reconciliation of chemical, analytical and clinical points of view was found to be no easy task, but the resulting documents are now almost ready for acceptance by the International Federation of Clinical Chemistry (IFCC) .I2 In these documents a convention for reporting the results of the direct determination of sodium and potassium is proposed. The main ideas of the proposal can be summarized as follows: the results are reported in terms of concentration (mmoll-1); the results of measurements on standard normal specimens should conform exactly with those obtained by FES on the same specimens; standard plasma specimens are defined as having a mass concentration of plasma water of 0.93 kg 1-1, a plasma hydrogen carbonate concentration of 24 mmol I - I , a plasma pH of 7.40 and concentrations of albumin, total protein, cholesterol and triglycerides within the reference ranges of healthy subjects.Consequently, it is proposed that the amount reported should be termed as 'the adjusted active substance concentration'. Finally, it is stated that although the results given by ion-selective electrodes for plasma of normal healthy subjects are equivalent to the substance concentration of the ion, in samples with abnormal concentrations of complexes of the ions the results are numerically different from true substance concentration.Table 1 Dedicated ISE-based electrolyte analysers Class of Producer analyser Parameters * Amdev (USA) Lytening Na, K, Li. (Ca?, Cl?) AVL (Austria) AVL Na, K, Ca, pH, C1. C02 Corning (UK) Corning Na. K, Ca, pH. Li. C1 Fresenius (Germany) Kodak (USA) Kone (Finland) Nova (USA) Nova Na.K.Ca,TCa,tpH,C1.Li.CO2 Radiometer (Denmark) ICA, KNa Na, K, Ca, pH Ionometer Na, K, Ca Ektachem Na, K. C1, CO2 Microlyte Na, K, Ca, pH, Li, CI, CO?, Mg * Charges of ions arc omitted. t TCa = total calcium. The above recommendation would virtually remove the discrepancy between direct and indirect (in the medical sense) determinations in normal patient sera. Extensive studies of sodium and potassium binding to inorganic ligands and proteins, water binding to proteins, liquid-junction effects and the influence of ionic strength have shown that the bias between sodium and potassium reports obtained from an average ISE-based commercial analyser and FES on patient pooled sera, although statistically significant, is below 1% .s Important work on prototype reference material, coordi- nated by the National Institute of Standards and Technology (USA), has also been accomplished.The Standard Reference Material selected for production consists of three levels of ultrafiltered human serum from a single pool, certified free of hepatitis B and HIV-11.It will be stored in 3 ml portions in 6 ml glass ampoules at -80 "C. l3 It can be concluded from what was said above that a compromise between chemical and clinical points of view is attainable but it is not an easy goal and the price to be paid also is not low. Measurement of pH and Chlorides-Two Extremes Similar arguments to those for sodium and potassium can also108 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 be applied to the ISE-based direct determination of pH and chlorides. If this is done it could be seen that the status of these two measurements reflects the basic dilemma discussed, i.e., the need to use a well defined activity scale versus an adjusted substance concentration concept. For hydrogen ion determina- tion, the routine method involving well defined activity standards, measurement recommendations14 and reference material13 was developed.The status of the determination of chloride, although performed by a number of commercially available analysers (see Table l), can at best be compared with that of sodium and potassium discussed above. Although the chemical status of this ion resembles that of sodium and potassium, no accurate recommendation or reference material has yet been proposed. The chloride electrode itself accounts for increased difficulty in this respect, namely a relatively poor selectivity for hydrogen carbonates and especially for some more lipophilic anions of clinical importance, such as bromides, iodides and salicy- late,l5>’6 results in a need for reconsideration of the definition of a normal patient sample.The situation discussed so far shows that apart from the difficulties with the most advantageous type of measurement, i.e., direct in the chemical and medical sense, one can expect a cumulation of problems and consequent complications with the application of less specific sensors. Measurement of Calcium Ionized calcium In principle, the direct measurement of ionized calcium with a calcium ion-selective electrode encounters similar difficulties to the measurement of ions discussed above. However, the construction of an appropriate convention in this instance is less straightforward because of two complications: the appar- ent lack of a reference method (which can also be seen as advantageous as imposing no constraints) and the appreciable complexation of calcium ions in blood, depending on the total calcium concentration, proteins, calcium-binding ligands and pH.Except for pH there is no simple way to assess these influences, especially in instruments employing only electro- chemical methods. However, even in general this task is undoubtedly difficult, even if possible. This means that the way to find a satisfactory recommendation for expressing the results for ionized calcium is not easy. However, the importance of this parameter in clinical chemistry is now well recognized.17.18 Therefore, a rational resolution must be found and in fact work in this direction is well advanced.19JO Already an important step has been made by Siggaard-Andersen et al.,Zl who proposed normalizing the readouts of ionized calcium versus pH by the introduction of a special correction function.The idea of compensating for the influence of the changes in sample pH due to hydrogen carbonate status and reporting the value for ionized calcium ‘corrected’ to pH 7.40 seems to be sound, especially for separate plasma or serum and aerobic sampling. On the other hand, the obvious influence of various calcium- binding ligands and a narrow applicable pH range of the correction function severely limit the uncritical use of this type of report and at the moment the decision regarding its use is still open. Total calcium Two basic approaches for the potentiometric determination of total calcium have been proposed.22-23 Both of them are based on the hydrolysis of calcium complexes (due to sample dilution, typically more than 1 : 10) with concurrent decomplexation of calcium by hydrogen ion22 or complexation by a fixed amount of a calcium-binding ligand.23 So far only a few manufacturers have implemented a potentiometric way of assessing total calcium (see Table l), although it may soon offer a good alternative to the still dominating spectrophotometric method.Measurement of Lithium The introduction of a lithium ion-selective electrode added a new analyte accessible by direct measurement. This assessment cumulates all the difficulties discussed above, still being complicated by the insufficient selectivity of a lithium sensor for sodium ions. As a specific lithium electrode does not exist, the procedure for the determination of lithium has to include a means of compensating for the influence of sodium.This means that the sodium concentration has to be determined (or pre-set arbitrarily) and accordingly used to obtain a final report on the concentration of lithium ion? For this reason, one can expect that the report on lithium will be influenced not only by the error of this sensor itself but also by the errors coming from sodium readouts and the over-all status of the procedure, including calibration, data acquisition and transformation.25 Still the complex problem of clinical relevance of reported data remains. The fact that few companies offer ISE-based lithium analysers (Table 1) shows that even more complicated work on the recommendations than that discussed above must be completed.Measurement of Ionized Magnesium The importance of ionized magnesium in physiological and metabolic processes creates a natural interest in measuring this parameter. Moreover, the possibility of monitoring ionized calcium increased the demand for reports on ionized mag- nesium because only a joint report on these two parameters (supplemented with pH measurement) can provide a clinically relevant picture of divalent metal ion status. Despite prolonged efforts and obviously due to an accumulation of the various difficulties discussed above, this analytically challenging prob- lem remained unsolved.26 Only very recently was the first successful method for the fully automated determination of ionized magnesium announced27 and, correspondingly, the first commercial analyser able to report ionized magnesium was introduced (see Table 1).As already mentioned, the development of a reliable automated method for ionized magnesium encounters several problems that have to be handled. Thus a relatively selective and stable sensor must be employed, signal formation properly interpreted, the influence of over-all ionic strength and diffusion potential taken into account, activity and substance concentration being bound to appropriate standard composi- tion and all technical problems such as carryover, washing, data acquisition and over-all automation solved. The method for ionized magnesium is reported to work in the ion concentration range 0.2-3 mmol l-1 with a precision of better than 3%.No ion seems to interfere; the apparent influence of calcium is compensated for numerically. An influence of magnesium-binding ligands, including proteins and some anticoagulants, is indicated. A need for codification of sampling, preservation and storage has been stressed .28 With ionized magnesium a new chapter in electrolyte analysis has been opened. Many aspects resemble those known already from the history of ionized calcium, hence the same or perhaps even more difficult work on a reference method and recommendations must be carried out. Determination of Hydrogen Carbonates and Total Carbonates The studies with lithium and magnesium electrodes clearly show that by losing the specificity of the electrode the most desirable direct measurement in both medical and chemical senses starts to be unattainable.This difficulty increases with the determination of any form of carbonates, as no sensor is available that would directly sense these species. Indirect measurement is then the only alternative. As one can judge from existing instruments, several ideas have been tried to overcome this obstacle. In blood-gas analysers, chemically indirect measurement is performed with use of a C02 gas electrode (Severinghaus type). The system is calibrated with appropriate mixtures of gases and the partial pressure of C02 is then measured in the samples withoutANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 109 dilution. Consequently, any other forms of carbonate are then calculated from the Henderson-Hasselbalch equation with the use of C02 and pH values measured for a particular sample.There is a long and continuing discussion of the medical relevance of this procedure, which basically refers to the question of the reliability of C02 measurement and/or the ‘stability’ of the dissociation constant of carbonic acid (for the latest data see reference 29 and references cited therein). Undoubtedly, the indirect (in the chemical sense) measure- ment of C02 is the main problem in this instance. The other electrochemical methods for the determination of carbonates also suffer from this problem. As there is no direct electrode for hydrogen carbonate ions, three ways of measurement are utilized. The first, direct in the chemical sense and indirect in the medical sense, employs a carbonate ion-selective electrode to measure the sample prediluted with base to convert the more acidic form of carbonates.30 The second, indirect in both chemical and medical senses, uses the C02 gas electrode for the measurement of total C02 evolved after addition of acid.31 The third, also indirect in both senses, is based on distortion of the pH of the carbonate buffer by the selective binding of hydrogen carbonates in the sample.32 The different chemistries used in these three methods unavoidably affect the final results and therefore a certain number of discrepancies must be expected when these methods are compared.The almost total indirectness of the electro- chemical methods in the assessment of carbonates complicates a thorough interpretation of the medical relevance. However, as nobody is questioning the clinical importance of such a report the work on appropriate conventions in this respect would be-sooner or later-initiated. Prospects for Use of New Ion-selective Electrodes By taking into account the general properties of ion-selective electrodes (concentration ranges, selectivity, expected life- time, etc.) and by listing the ions of physiological importance, it is possible to deduce that some further improvements or application of new ISEs are still feasible.One can expect further advances with lithium and magnesium electrodes as the existing types suffer from relatively poor selectivity. The introduction of an electrode for the direct determination of hydrogen carbonates would be welcome. Inorganic phosphate and sulphate ions are other candidates for direct determination with ISEs.Undoubtedly a growing application of electrochem- ical biosensors related to ISEs is also to be expected. Conclusions The use of ion-selective electrodes in clinical analysers has provided an effective answer to growing clinical demands for specific, fast, inexpensive and fully automated measurements. A variety of ionophores, analytical methods, clinical concepts and technical inventions account for the expansion of this technique. Although the successful use of ion-selective elec- trodes, as with any other analytical method, has inherently been accompanied by various chemical and medical constraints they are of great value in clinical analysis. References Wang, J . . Electroanalytical Techniques in Clinical Chemistry and Laboratory Medicine, VCH, New York, 1988.Biosensors, eds. Turner, A. P. F., Karube, I.. and Wilson. G. S . , Oxford University Press, Oxford. 1987. Czaban, J . D.. Anal. Chern., 1985. 57, 345A. Landenson, J. H.. Anal. Proc.. 1983, 20, 554. Oesch. U., Ammann. D., and Simon, W.. Clin. Chern.. 1986, 32, 1448. Byrne. T. P., Ion-Sel. Electrode Rev.. 1988, 10. 107. Lewenstam, A., Fresenius J . Anal. Chern., 1990, 337, 518. Covington, A. K., in Methodology and Clinical Applications of ton-selective Electrodes, eds. Maas, A. H. J., Boink, F. B. T. J . , Saris, N. E., Sprokholt, R., and Wimberley, P. D., IFCC Workshop, Helsinki, 1985, published privately, Copenhagen, 1986. 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Boink. A.B. T. J . , Buckley. B. M., Christiansen, T. F., Covington, A. K., Maas, A. H. J.. Mueller-Plathe, O., Sachs, Ch., and Siggaard-Andersen, O., in Methodology and Clinical Applications of Ion-selective Electrodes, eds. Maas, A. H. J . , Buckley, B., Marsoner, H., Saris, N. E . , and Sprokholt, R., IFCC Workshop, Graz, 1986, published privately, Utrecht, 1987. Maas, A. H. J . , Siggaard-Andersen, O., Weisberg, H. F., and Zijlstra, W. G., Clin. Chern., 1985, 31, 482. Flear, C., in Methodology and Clinical Applications of Ion- selective Electrodes, eds. Mass, A. H. J., Buckley, B., Manzoni, A., Moran, R. F., Siggaard-Andersen, O., and Sprokholt, R., IFCC Workshop, Stresa, 1988, published privately, Utrecht , 1989. Maas, A. H. J., in Methodology and Clinical Applications of ton-selective Electrodes, eds.Burritt, M. F., Cormier, A. D., Maas, A. H. J., Moran, R. F., and O’Connell, K. M., IFCC Workshop, Danvers, 1987, published privately, Rochester, MN, USA, 1988. Koeh, W. F., and Paule, R. C., in Methodology and Clinical Applications of ton-selective Electrodes, eds. Maas, A. H. J., Buckley, B., Manzoni, A., Moran, R. F., Siggaard-Andersen, O., and Sprokholt, R., IFCC Workshop, Stresa, 1988, pub- lished privately, Utrecht, 1989. Mass, A. H. J . , Weisberg, H. F., Burnett, R. W., Mueller- Plathe, O., Wimberley, P. D., Zijlstra, W. G., Durst, R . A., and Siggaard-Andersen, O., J. Clin. Chern. Clin. Biochem., 1987. 25, 281. Lewenstam. A., Saarinen, K.-S., and Hulanicki, A., in Methodology and Clinical Applications of ton-selective Elec- trodes.eds. Maas, A. H. J . , Buckley, B., Marsoner, H . , Saris, N. E., and Sprokholt, R., IFCC Workshop, Graz, 1986, published privately, Utrecht, 1987. Lewandowski, R., Sokalski, T., and Hulanicki, A., Clin. Chern., 1989. 35, 2146. Bowers, G. N.. Jr., Brassard, C., and Sena, S. F., Clin. Chern., 1986,32, 1437. Fogh-Andersen, N., Dan. Med. Bull.. 1988,35, 575. Boink, A. B. T. J., Buckley, B. M., Christiansen, T. F., Covington, A. K., Maas, A. H. J., Mueller-Plathe, O., Sachs, Ch., and Siggaard-Andersen, O., in Methodology and Clinical Applications of Ion-selective Electrodes, eds. Maas, A. H. J., Buckley, B., Marsoner, H., Saris, N. E., and Sprokholt, R., IFCC Workshop, Graz, 1986, published privately, Utrecht, 1987. Covington, A. K . , Kelly, P. M., and Maas, A.H. J . , in Methodology and Clinical Applications of Ion-selective Elec- trodes, eds. Maas, A. H. J . , Buckley, B., Manzoni, A., Moran, R. F., Siggaard-Andersen, O., and Sprokholt, R., IFCC Workshop, Stresa, 1988, published privately, Utrecht, 1989. Siggaard-Andersen, O., Thode, J., and Wandrup, J., in Proceedings of the 5th Meeting of the tFCC Expert Panel on pH and Blood Gases, IFCC Workshop, Copenhagen, 1980, pub- lished privately, Copenhagen, 1980. Anker. P., Wieland, D.. Simon, D., Dohner, R. E . . Asper, R., and Simon, W., Anal. Chern., 1981, 53, 1970. Hulanicki, A.. Lewandowski, R., Michalska. A., and Lewen- stam, A., Anal. Chirn. Acta, 1990, 233, 269. Metzger, E., Dohner, R.. Simon. W., Vonderschmitt, D. J., and Gautschi, K., Anal. Chern.. 1987, 59, 1600.Dolowy, K., and Lewenstam, A., unpublished results. Muller. M., Rouilly, M., Rusterhotz. B., Maj-Zurawska, M., Hu, Z . , and Simon, W., Mikrochirn. Acta, Part ItI, 1988, 283. Maj-Zurawska. M., and Lewenstam, A., Anal. Chirn. Acta, 1990, 235. in the press. Lewenstam, A., Maj-Zurawska, M., Blomqvist, N., and Saarinen, K.-S., in Methodology and Clinical Applications of Ion-selective Electrodes. ed. Moran, R . F., IFCC Workshop, Monterey. 1990, in the press. O’Leary. T. D., and Langton, S. R., Clin. Chern., 1989, 35, 1697. Scott, W. J.. Chapoteau, E . , and Kumar, A., Clin. Chern., 1986. 32, 137. Buzza, E. E . , in Methodology and Clinical Applications of Ion-selective Electrodes, eds. Maas, A. H. J., Boink, F. B. T. J., Saris, N. E., Sprokholt, R., and Wimberley, P.D., IFCC Workshop, Helsinki, 1985. Lewenstam, A., Ivaska, A.. and Wanninen, E., Talanta, 1986, 33, 739.110 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Some Artificial Intelligence Techniques for the Interpretation of Experimental Data Philip G. Barker Interactive Systems Research Group, School of Computing and Mathematics, Teesside Polytechnic, Middlesbrough TS1 3BA In a recent book on information physics and the internal structure of the universe, Stonier1 claims that information (like matter and energy) is a fundamental and intrinsic component of the physical universe in which we exist. That is, all objects and processes (both animate and inanimate) possess informa- tion. Much of scientific research is therefore concerned with ‘discovering’ the intrinsic information associated with the entities and processes that surround us.Many research methods have been devised in order to facilitate the collection of data and information relating to those scientific processes which are of interest to us. However, the monitoring and collection of data from many processes can be difficult, costly and time consuming when undertaken manually. Therefore, the fully automatic and intelligent collection and interpretation of data in a reliable way is, for many scientists, one of the ultimate goals of research into artificial intelligence (AI). Of all living species, humans are unique in their ability to design and construct sophisticated tools and machines to facilitate problem-solving activity. Undoubtedly, electronic computers are ranked amongst the most important of these tools.Like the electric motor, the computer is often described as a meta-machine, that is, it can be used as a building block in order to fabricate a wide range of other more sophisticated machines. The computer is extremely important in the physical sciences both for processing raw data and as a basic control element within experimental equipment .z This is particularly so within the area of ‘chemical instrumentation’. Indeed, the range of analytical instruments that is now, in some way or another, dependent on computer technology is growing almost exponentially. As computers have become more sophisticated, so too has the equipment in which they are embedded. Over the last few years a number of important hardware and software develop- ments have taken place.These developments mean that new types of analytical problem (hitherto impossible to solve) can now be tackled with relative ease. The obvious hardware developments that have taken place lie in the areas of processing speed and the allowable memory address space. Both have increased dramatically. Develop- ments in interface standards now make it possible to add a wide range of peripheral equipment to a basic low-cost computing facility (a micro- or minicomputer system) in order to construct an experimental workstation or a sophisticated laboratory information management system. Of course the development of, new types of sensor, sensing mechanisms and signal processing hardware also considerably extends the range of processes that it is possible to monitor and control.Each of the above developments is important in its own right. However, the two most fundamental hardware developments that have taken place lie in the areas of parallel processing and neural network architectures (NN computers). The ability to link large numbers of basic processing elements together into a powerful concerted whole (parallel processing) is fundamental to the achievement of ultra-high- speed computing. Very fast computation is important in complex real-time control situations, pattern-matching algorithms and in the processing of image data. The import- ance of NN computers lies in their ability to enable computer systems to ‘break away’ from the use of classical von Neumann architectures and the possibilities that they offer for imple- menting automatic machine learning.Despite their underlying importance, this paper is not primarily concerned with hardware developments. Instead, its essential objective is to review some of the important progress that has taken place in the area of software, particularly that which is available for the automatic and intelligent interpreta- tion of experimental data. The remainder of this paper therefore addresses this issue. Following a short historical prelude, three important areas of A1 research are described: virtual instruments, expert systems and neural networks. The role of each of these areas of research in the context of data interpretation is discussed. Some possible future directions of development are then outlined. Historical Perspective Within virtually all computer systems, some form of software is a fundamental requirement.The importance of computer programs lies in their ability to tailor a given hardware configuration into an almost unlimited number of different kinds of ‘application engine’. Typical engines are those which are used to support applications such as word processing, desk-top publishing, spreadsheet analysis, data acquisition, process control and expert systems. Because most of these applications involve end-user participation, they are often referred to as ‘interactive computer systems’.3 Today, a large proportion of interactive computer applications involve the use of a low-cost personal computer (PC). The various tasks involved in producing computer programs are collectively referred to by the term ‘software engineering’ (SE).When interactive computer systems are involved, software engineers must concern themselves with three im- portant issues: understanding the semantics of the application domain, the design of suitable observation, management and control mechanisms and the design, creation and maintenance of appropriate end-user interfaces. It is primarily through this last aspect of SE that computer systems can be made to simulate (and substitute for) real-world objects (such as instruments and controllers) within a given target application domain. Of course, the effectiveness with which this goal can be achieved invariably depends on the quality of the ‘meta- phors’ and ‘myths’ that an interface embeds.”.“ This topic is discussed further under Virtual Instruments.Obviously, A1 techniques are important in each of the three areas of software engineering mentioned above. Of course, A1 methods are also extremely relevant in the context of automat- ing software engineering processes themselves. Indeed, one of the major goals of A1 research in this area is to eliminate the need for programming activity, through the use of develop- ment shells, software generators or automatic machine learn- ing. A classic example of the shell approach is illustrated in the many different types of expert system shell that are now available commercially; some of these are discussed in more detail under Expert Systems. The use of automatic machine learning using neural network computers has already been briefly mentioned in the previous section; this topic will be further discussed under Neural Networks.Undoubtedly, one of the major attractions of currently available personal computer systems is their ‘user-friendliness’ and ease-of-use compared with conventional mainframe and minicomputer systems. Ease-of-use of PCs has been achieved primarily as a consequence of the availability of technology to support the transition away from text-based interfaces towardsANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 111 the extensive use of reactive graphical user interfaces (GUIs). A typical GUI employs pictorial communication methods (involving pictures, icons, windows and pull-down menus) and pointing operations made with a mouse. Often such interfaces are referred to in the literature as WIMP interfaces.s Graphical user interfaces are made possible because of the availability of low-cost, high-resolution, bit-mapped, colour display monitors.In addition to allowing the display of high-quality static pictures and images, such monitors also permit the display of animation; they also facilitate the viewing of TV-quality video pictures (either full-screen or part-screen). Through the appropriate use of pictorial forms it is possible to achieve highly realistic simulations, emulations and surroga- tions. As we shall discuss in the following section, pictorial interfaces are extremely important in the context of virtual instruments and consultation dialogues with expert systems. Some AI Techniques Amongst other things, the study of artificial intelligence concerns itself with the design and fabrication of ‘intelligent systems’. Stonie+ suggests that intelligence should be con- sidered as ‘a property of advanced information systems which allows such systems to analyse their environment then engage in processes which enhance the survivability or reproducibility of the system’.On this basis, all living organisms possess some measure of intelligence because they are able to survive and reproduce. But what about machines such as computers, process controllers, automatic analysers and autonomous robots? Can they show intelligence too? Stonier cogently argues that they can.’ Intelligent behaviour is multi-faceted. If machines can emulate some subset of intelligent behaviour then they must be deemed to show some level of intelligence.The intelligence of a machine is often gauged by its ability, or otherwise, to pass the Turing test.s Five important facets of human intelligence that can be fairly easily reproduced within suitably designed machines are: (1) the ability to mimic or copy the behaviour of other entities; (2) goal-seeking behaviour; (3) participation in problem-solving activity based on inference and decision making; (4) the potential to learn from experience; and ( 5 ) the ability to adapt internal and external behaviour in order to accommodate the requirements of changing conditions. Examples of systems that exhibit one or more of these facets are a robot system (facet I ) , a game-playing program (facet 2), an expert system (facet 3), a neural network (facet 4) and a senstor7 (facet 5 ) .Of the five facets of artificial intelligence listed above, as far as this paper is concerned the three most important are mimicry, problem solving through inference and decision making and the ability to learn from experience. These form the basis for the material that is discussed in the remainder of this paper. Virtual Instruments Reference was made earlier to the fact that the computer was a meta-machine. This means that through suitable design pro- cedures it can be incorporated into other systems in order to enhance their functionality, intelligence and/or performance. Alternatively, through the use of appropriate ‘myths’, comput- ers can also be made to look like any other kind of machine.Virtual instruments9 provide one example of the sophisticated use of myths to give the user the impression that the computer is, in fact, acting like some other piece of instrumentation. One of the most impressive implementations of the virtual instru- ment paradigm is the LabVIEW system developed by National Instruments for use on the Apple Macintosh PC.10 LabVIEW is an acronym for Laboratory Virtual Instrument Engineering Workbench. It is a sophisticated software system that enables engineers and scientists to use an interactive computer system to design complex programs that emulate the behaviour of real instruments. Essentially, a virtual instrument (V1) i s an instrument whose general function and capabilities are determined in software. The success of LabVIEW’s approach to virtual instrumentation is based primarily upon its graphical programming language (called G).G is an object- oriented language in which programs are created through the use of icons. These icons are used to represent either complete instruments or instrument components; if necessary, these can be held in various software libraries. When designing a new instrument, the user simply organizes a particular set of icons into an appropriate spatial arrangement on the screen of the host PC’s CRT and links the icons together in various ways to indicate the data flow pathways that are needed. & File Edit Go Tools Obiects 1 URUEFORH Frequency SawTooth Triangle Square Sine 5 6 AMD I i tud2 H e r e i s a cornDutatiorml V I w h i c h s i m u l a j i s p l a y t h e waveform.Note f r o m t h e disl LabVlEW a l l o w s t h e u s e r to a d i u s t t h e fr Fig. 1 Simulated function generator panel Three basic steps are involved in creating a VI: first, the design of a front-panel; second, the design of a block diagram that describes how the instrument should perform; and third, extensive testing to prove that the instrument actually performs in the correct fashion. The instrument front-panel provides the means by which the user of the VI actually controls it. This control is effected by means of ‘point and click’ operations made with the host PC’s mouse. Examples of instrument front-panels are illustrated in Fig. 1 (a function generator) and Fig. 2 (a two-channel spectrum analyser). Fig. 3 shows the underlying instrument block diagram for the function genera- tor depicted in Fig.l . & File Edit Go Tools Objects 1 ,005 .010 ,015 .020 ,025 .030 Time (msec) 0 300 600 ‘300 1200 1500 1800 2100 2400 2700 3000 Freouencu CHz) l H e r e i s t h e f r o n t oanel o f t h e s o e c t r u m a n s l u r e r imolemented w i t h t h e MIO- 16 p l u g - i n board. F a k i n g f u l l sdvaniage of t h e 68’881 iil t h e Mac /I, a i024 point complex FFT e x e c u t e i i n 0.5 s e c 0 n d s . a Fig. 2 Two-channel spectrum analyser panel LabVIEW can interface to the outside world of data acquisition by means of an IEEE 488 interface, CAMAC, VXI, RS232/422 interfaces and various types of A/D (and D/A) converter and digital I/O board. These can be plugged into the back of the host PC in which the system is running.Through these boards it is possible to acquire data from an experiment (and/or a process control rig) and then use a VI within the PC to process and display it. Alternatively, a VI that is resident within the PC can be used to control a real instrument that is112 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 situated remotely within another part of a laboratory or which is located in some inaccessible position within process plant equipment. An example of the successful way in which the LabVIEW system can be used to acquire and process experimental data is illustrated in the work of Kamal et al. 11 They used LabVIEW in order to monitor and control rheological tests relevant to the design of optimum blends of motor oil.The LabVIEW software sends test data to a design guidance system through a ‘DSPT Workbook’ that is written in HyperCard.3 DSPT is an acronym for Decision Support Problem Technique; the work- book is essentially an end-user interface myth that makes the computer system appear to act like an interactive, intelligent electronic book.12 An important component of the over-all system is its inductive learning package. This takes the test data from the LabVIEW system and uses them to generate classification rules. These rules can subsequently be incorpor- ated into an inference engine that can be used as an expert system to formulate oil blends. The use of expert systems and inductive learning are each discussed in subsequent sections of this paper. & File Edit Go Tools Obiects 1 Fig.3 Block diagram for a function generator Expert Systems An expert system is essentially a computer program that embeds human problem-solving expertise relating to some specialized problem domain. The program uses its knowledge in order to solve particular types of problem using decision- making and inferencing techniques. The basic software archi- tecture of expert systems and the way in which they can be developed and used are described in considerable detail elsewhere.13-19 Therefore, in this paper we concentrate pri- marily on the use of development shells, rule induction and the use of blackboard systems. Development shells Until the advent of suitable development shells, the production of expert systems was always fraught with technical difficulties and productivity issues. For some types of application this is still true.However, for many others the availability of simple, easy-to-use expert system shells has enabled many of the development problems to be overcome. Using such shells, users are able to generate their own expert systems for the particular applications in which they have an interest. Natu- rally, the power and utility of an expert system shell depend on the facilities it provides and the methods of inferencing that it makes available. Undoubtedly, the simplest possible type of expert system shell is that based on the use of a simple decision tree processor. Such systems enable the author of an expert system to create a decision tree and the user of that expert system to prune the tree in order to reach a successful goal state (problem solved) or an unsuccessful goal state (problem not solved).The ‘Generic Expert’20 is an example of a low-cost, easy-to-use system of this sort. Although useful for certain types of problem (simple trouble-shooting and classification tasks), systems of this type have their limitations. Some of their major limitations stem from their inability to generate new know- ledge (through inferencing techniques), their lack of facilities for handling uncertainty and the difficulties of implementing inexact reasoning. If these facilities are required, then a more sophisticated type of expert system shell must be employed. In our work, we have made extensive use of a knowledge processing system called KnowledgePro.21 We have found this system attractive because of the rich repertoire of knowledge processing primitives that it makes available. Facilities exist to permit inferencing (both forward and backward chaining), the creation of frames, list processing, the implementation of sophisticated input-output strategies, picture manipulation and the ability to link to software modules that are written in any of the standard data processing languages such as FORTRAN, BASIC, PASCAL or C.The system also makes available a range of ‘add-on’ interfaces to enable data to be imported from (or exported to) other packages such as spreadsheets and database systems. Using the Knowledgepro development shell we have found the creation of sophisticated expert systems to be relatively straightforward.A variety of different tools and techniques can be employed in order to ease the tasks involved in developing expert systems. One extremely useful technique is that known as rule induction. This is a powerful technique that can be used to analyse tabular collections of data in order to discover the nature of the rules that can be used to represent the various relationships embedded within the data. Rule induction is briefly described in the following section. Rule induction Rule induction is a mathematical procedure that can be used to deduce the rules involved in classifying data. The successful use of this approach depends on the availability of a set of example data (the training set) to be classified (D), a set of attributes or characteristics which describe the data (A), a collection of classes into which the data will fall (C), an induction algorithm (I) and the sought-after rule set that classifies the data (R).The functional relationship between these five items is as follows: I(D,A,C) + R The mechanism used in most induction systems is based on Quinlan’s ID3 algorithm,lI.** which uses Shannon’s theory of information. It selects, as a basis for classification, the attribute that results in the largest decrease in entropy of classification in the training set. Once the rule set has been deduced it can be used to classify data (E) not contained within the original training data set, that is, R(E) + C In much of our work we have used a rule induction system called KnowledgeMaker22 in order to classify different types of experimental data.KnowledgeMaker is capable of producing decision trees or sets of IF.. .THEN.. .ELSE rules that can be incorporated directly into an expert system (in this instance Knowledgepro). A good example of the way in which rule induction can be used in an analytical chemistry context can be found in the work of Kamal et al. ,I1 who used rule induction in order to deduce rules that describe the relationship between oil viscosity measurements at different temperatures and the grades of motor oil to which they belong. The utility of rule induction is also reflected in the work of Harrington and co-~orkers,*3,24 who used a ‘rule-building’ expert system for classifying mass spectra. Their work demonstrates that the rule induction approach can perform better than linear discriminant analysis for classification of pyrolysis mass spectra and laser ionization mass spectra.ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 113 Blackboard systems Many complex problems (such as speech understanding, image analysis and molecular structure elucidation) often require the use of several different expert systems simultaneously in order to achieve a satisfactory solution within a reasonable time span.This requirement can often be achieved through the use of a ‘blackboard architecture’. 18-25 Within such systems the term ‘blackboard’ is used to describe the central database facility that coordinates and controls the independent groups of rules (known as knowledge sources) that are together responsible for solving a particular problem.The knowledge sources communicate with each other by writing messages to the blackboard and reading messages from it. An important component of a blackboard system is its ‘scheduler’; this is responsible for managing the various knowledge sources which are each competing with each other in order to handle the incoming pieces of data. Various types of scheduling strategy can be employed.25 An example of the way in which complex chemical problems can be solved using this approach is illustrated by the CRYSALIS project at Stanford University. 18.25 The task domain of CRYSALIS is X-ray protein crystallography; the system was designed to produce the best possible three- dimensional structural description which could be derived from a given electron density map (EDM).In order to infer this structure, CRYSALIS must interpret X-ray diffraction data consisting of position and intensity values for diffracted waves. This is achieved through the use of multiple independent knowledge sources that generate and test multi-level hypotheses about plausible protein structures using knowledge about protein composition and X-ray crystallography and heuristics for analysing EDMs. Fundamental to the operation of CRYSALIS is its blackboard structure, which consists of two ‘panels’, one for hypotheses and one for representing EDMs. One of the limitations of CRYSALIS is the substantial amount of computing that needs to be performed in order to find solutions. However, this limitation can be overcome if the system can find a ‘nearly optimum’ sequence of rules to apply in response to a specific problem.Large computer systems are often needed in order to solve complex problems using the blackboard approach. However, as the computational power of microcomputers (and their ease of interconnection) increases, so it becomes possible to implement blackboard problems using multiple inter- connected microcomputer systems. Neural Networks Conventional computers are extremely good at executing algorithms that apply procedural logic to well-defined prob- lems. Unfortunately, they are not very good at emulating human cognitive skills such as those involved in image recognition, speech understanding, general pattern recogni- tion, learning, inference, induction, association and dealing with ill-defined or poorly structured situations.However, computers that are based on the use of neural networks are seen as being a possible solution to these kinds of artificial intelligence problems, since they attempt to emulate the way in which the human brain works. An artificial neural network is composed of a large number of parallel processing elements called neurons. Each neural element has a set of inputs and a set of outputs. The outputs from one neuron can be connected to the inputs of other neurons, thereby creating the network structure. Functionally, each neuron performs two basic tasks: first, a summation of its inputs (each of which is weighted), and second, a transfer of ‘firing’ of its output lines when and if the input threshold is met. The output of a neuron is usually computed using a simple rule such as If Input > Threshold then Output = 1 else Output = 0 Neural network elements are organized into layers.The most primitive are single-layer nets that operate like percep- trons.7 Although they are very useful, single-layer nets have many limitations. However, the limitations of single-layer nets are overcome in multi-layer networks. These consist of three parts: the input layer, the hidden layer and the output layer. The operation of a neural network involves two basic phases, learning and recall. In the learning phase, training stimuli are presented to the input layer and the connection weights between elements within the hidden intermediate layers are then adjusted in order to produce the desired response at the output layer.During the recall phase the network is ‘put to work’. It is shown input patterns and expected to produce correct responses to those patterns on the output units. Typical tasks involve associative recall, pattern recognition and classifi- cation. An essential feature of any neural network is its learning rule, i . e . , how it adjusts its weights or modifies its responses in various input-output patterns. A learning rule will usually seek to bring actual output closer to the desired output. Neural network learning takes place in one of three basic ways: supervised, unsupervised or self-supervised. Supervised learn- ing occurs when the user presents the network with trial-and- error inputs and then teaches it correct and incorrect responses.In unsupervised learning, data are simply entered without human intervention; this process leads to internal clustering-the desired result. Self-supervised learning occurs when the network monitors itself and corrects errors in the interpretation of data by feedback through the network. In self-supervised learning the network weights are adjusted according to pre-defined algorithms. One of the commonest learning strategies is ‘back-propagation’.26 Back-propagation is a statistical method in which network weights are learned from experience in a trial-and-error fashion. Other examples of learning strategies include drive-reinforcement theory and the outstar method. Artificial neural networks are most commonly implemented as software simulations, integrated circuit chips and optical systems.Currently, many simulators are available that run on standard computer architectures. Neural networks are good at finding rules and discovering patterns in unorganized data. They are widely used in sonar, radar and image processing. One interesting chemical application has recently been ini- tiated by the Federal Aviation Administration in the USA, who have installed a neural network-based thermal neutron analysis (TNA) bomb detector. The system is called SNOOPE (System for Nuclear Online Observation of Potential Explo- sives) and is based on a back-propagation supervised learning algorithm. The TNA identifies explosive materials by bom- barding them with low-energy neutrons and then identifying the resulting gamma-ray emissions.New materials can easily be learned by inserting them into the TNA and commanding the system to learn the materials’ characteristics. A neural network is also being used at the Los Alamos National Laboratory in the USA in order to analyse DNA structure. The neural network examines the nucleotide-base sequence of DNA, determining which regions are coded for proteins. The benefit of using the neural network (actually a software simulation) is speed, trading off some loss of accuracy. Low-cost neural network simulation software is starting to become available for PC environments. Systems such as NeuroShell (Ward Systems Group, Frederick, MD, USA), for example, can learn by example, handle fuzzy logic and can be used in many situations as an alternative to expert systems, the ID3 algorithm and regression analysis.NeuroShell runs under MS-DOS and creates neural network applications using a modified back-propagation algorithm. Systems that run on PC-based equipment are useful for teaching and student projects. However, for more sophisticated applications such as those described above the use of a more powerful workstation would be a necessary requirement.114 Future Direction Towards Intelligent Machines A very large number of interesting chemical processes take place in locations other than conventional chemistry labora- tories, e.g., beneath the oceans, on remote planets and in outer space. Many of these processes can have a significant impact on the Earth and the way in which we live. There are therefore many compelling reasons why we should wish to take analytical science to many remote and hazardous places.However, in order to achieve this we must effect ways of analysing materials and processes that do not require human intervention but instead depend on automation techniques, tele-robotics and the use of intelligent machines. Automation techniques have been widely used by chemists for many decades in order to gain freedom from dull, routine and repetitive tasks. Descriptions of various types of auto- mated laboratory equipment and automation systems can be found in the literature.2-27 One example of these, PASS (Perdue Automatic Synthesis System”), provides an auto- mated environment for the development and optimization of both the synthesis and the analysis of organic materials. The operational system consists of two work areas (a synthesis table and an analysis table), each of which is serviced by its own robotic arm; essentially, these function as static pick-and-place robot systems.Material flow between the two work areas is achieved using a vial transportation system that is based on a model train engine running on a computer-controlled track. Other examples of automation systems include APOCALYPSE (for growing suitable crystals for use in X-ray crystallography) and GASP (an acronym for Generic Analy- tical Sample Preparation). As its name suggests, GASP is used to automate the processes involved in preparing samples for analysis; it is designed to handle everything from solids through ointments to liquids. Tele-robotics is another important aspect of the develop- ment of automatic analysis systems that are able to function in remote and/or isolated locations without direct human inter- vention.28 In terms of autonomy, a tele-robotic system offers more independence than tele-operation or tele-sensing, but considerably less than would be available from an intelligent autonomous robot.” The essential components of a tele- robotic system suitable for chemical analysis are either a static or mobile robot facility, appropriate sampling and analysis modules, a bi-directional communication channel to facilitate tele-operation and an operator interface that facilitates remote control of the system.The operator interface is usually implemented through virtual machine and virtual instrument systems similar to those described under Virtual Instruments.Indeed, LabVIEW is being extensively used by a number of organizations for the remote control of robot systems. Of course, one of the ultimate aims of A1 research is to produce intelligent machines that can function completely autonomously once they have been given a problem to solve. A possible high-level functional architecture for such machines is illustrated schematically in Fig. 4. Unfortunately, unlike humans, the intelligent machines that have been built to date are not good at general problem solving. Most of them operate in a tight, fairly restricted domain, such as chess playing, mathematical theorem proving, maze solving and spectra interpretation. The basic principle underlying the way in which such machines work is the automatic identification of the goals and sub-goals that need to be achieved in order to attain a satisfactory problem solution.This activity involves substantial planning, that is, finding multi-step solutions to complicated problems, and requires the coordination of tasks with available resources. In many simple robot systems a repertoire of domain specific plans, agendas and scripts are programmed into the control unit in order to facilitate problem solving. In more intelligent and autonomous systems the creation of plans, agendas and scripts is undertaken dynamically by the robot itself as it encounters problems to be solved; candidate solutions can be analysed and assessed and then used to Sensing - ANALYTICAL PROCEEDINGS. APRIL 1991, VOL 28 INTELLIGENT MACHINE - , optimize the performance of the system.Although substantial progress has been made with respect to the development of ‘truly’ intelligent machines, few (if any) are capable of meeting Stonier’s definition of an intelligent system which was introduced earlier in this paper. For many types of problem this does not really matter since the combined collaborative intelligence of human(s) and machine(s) together represents an optimum approach to intelligent behaviour. However, for some problems (where human collaboration is not feasible), highly intelligent machines will need to have the ability of learning to cope for themselves. For this reason A1 researchers will continue to look for ways in which machines can be made ‘to think like people’.INPUTS OUTPUTS Pe rfor ma nce Diagnosis optimizer - Reports Commands - 1-1 Repair Processing -Messages -Alerts Problem solving Problems --+ Fig. 4 Top-level functional architecture for an intelligent machine Conclusion Many of the tasks associated with collecting and analysing experimental data can be automated through the use of appropriately designed computerized instrumentation. For many applications in analytical chemistry this instrumentation needs to be mobile, robust and capable of autonomous operation. Chemical robotics and laboratory automation equipment can be used very effectively to meet many of the requirements of automated data collection. However, to facilitate the automatic interpretation of experimental data many novel types of computer program are often needed. This paper has discussed some potentially useful A1 techniques (surrogation, expert systems, rule induction and neuraI networks) which, when applied to instrumentation systems, could make them more intelligent and autonomous. 1 2 3 4 5 6 7 8 9 10 11 12 References Stonier. T., Information und the Internal Structure of the Universe-An Exploration into Information Physics. Springer, London, 1990. Barker, P. G., Computers in Analytical Chemistry, Pergarnon Press, Oxford, 1983. Barker. P. G.. Basic Principles of Human-Computer Interface Design. Century-hutch inson, London. 1989. Rubenstein. R., and Hersh, H., The Human Factor-Designing Computer Systems f o r People. Digital Press, Burlington. MA, USA. 1984. Perry, T. S.. and Voelcker. J . , IEEE Spectrum. 1989,26. No. 9, 46. Stonier, T.. A l S o c . , 1988, 2, 133. Stonier, T.. in preparation. Aleksander, 1.. Designing Intelligent Systems-An Introduction, Kogan Page. London. 1984. Santori. M.. IEEE Spectrum, 1990, 27. No. 8 , 36. LubVIE W User’s Guide, National Instruments, Austin. TX, 1987. Kamal, S. Z., Mistree, F., Sorab, J., andVanArsdale, W. E., in Artificial Intelligence in Design, Proceedings of the Fourth International Conference on the Applications of Artificial Intelligence in Engineering, Cambridge, UK, July 1989, ed. Gero, J. S., Springer, London, 1989, pp. 521-538. Barker, P. G.. Interactive Electronic Books, Interactive Multi- media, in the press.ANALYTICAL PROCEEDINGS, APRIL 1991. VOL 28 13 14 15 16 17 18 19 Barker, P. G.. Expert Systems in Analytical Chemistry, Irish Barker, P. G., in Artificial Intelligence in Industry and Gowrn- ment, Proceedings of the International Conference on Artificial 20 21 22 Intelligence in Industry and Government, November 23-25, 1989, Hyderabad, India, ed. Balagurusamy. E., Macmillan 23 India, New Delhi, 1989, pp. 413423. Dessy. R. E., Anal. Chern.. 1984, 56, 1200A. 24 Dessy, R. E.. Anal. Chem., 1984. 56. 1312A. 25 Chemistry. ACS Symposium Series. No. 408. American Chem- ical Society. Washington. DC, USA, 1989. 26 Waterman. D. A., A Guide to Expert Systems, Addison- Wcslcy, Reading, MA. USA. 1986. 27 Giarratano, J . C., and Riley, G. D., Expert Systems-Principles 28 and Programming, PWS-Kent Publishing. Boston. MA. USA. 29 1989. Chem. News, 1989, 28. Hohnc. B. A., and Pierce, T. H., Expert System Applications in 115 The Generic Expert-User Guide. PI Technologies, Lafayette, CA. USA, 1986. Barker, P. G., Eng. Appl. Artif. Intell.. 1989. 2, 325. Thompson. B.. and Thompson, B . , KnowledgeMaker User's Guide. Knowledge Garden, Nassau. NY, USA, 1987. Harrington, P. de B., Street, T. E.. Voorhees, K. J., di Brozolo, F. R., and Odom, R. W., Anal. Chern.. 1989.61.715. Harrington, P. de B . . and Voorhees, K. J.. Anal. Chem., 1990, 62.729. Craig, I . D., The Cassandra Architecture: Distributed Control in a Blackboard System, Ellis Horwood, Chichester, 1989. Wasserman. P. D.. Neural Cornpuiing: Theory and Practice, Van Nostrand Reinhold. New York, USA, 1989. Newman, A. R., Anal. Chem.. 1990, 62, 29A. Rokey, M., and Grenander, S . , IEEE Expert. 1990,5, No. 3,s. Iyengar, S . S., and Kashyap. R. L., IEEE Comput., 1989, 22 ( 6 ) . 14.

ISSN:0144-557X    

DOI:10.1039/AP9912800102

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, APRIL 1991. VOL 28 - 115 Computer 1 Chemometrics-the Key to Sensor Array Development Stephen J. Haswell" and Anthony D. Walmsley School of Chemistry, Thames Polytechnic, Wellington Street, Woolwich, London SE 78 6PF In recent years, considerable effort has been invested in establishing reliable transducers based on a wide range of physical or chemical properties for the detection of numerous analytes in the vapour or liquid phase. A recurring problem that severely limits the application of such devices to 'real' situations is thc presence of interferences. One philosophy to adopt in overcoming such problems is that of the 'smart sensor'. The underlying premise of a smart sensor is that Sensor array LJ E I ect ro n ics Y I n t erface '1 acquisition I Signal processing r l Array Control Fig.1 Test protocol of the construction of a smart sensor array * Present address: School of Chemistry, The University. Hull HU6 7RX. so-called interferences do not exist but that the transducer produces complex signals consisting of all the available analyte information. Obtaining and interpreting complex signals is not without problems. However, the degree of information con- tained within a transducer response can be considerable and lead to a more reliable device which offers a higher level of matrix analysis. The aims of the smart sensor research programme, part of which will be described in this papcr, are to produce a device that will be (i) a more reliable intelligent device, (ii) capable of complex data interpretation, (iii) able to function as a stand-alone or networked device, (iv) inexpensive and (v) suitable for a wide range of applications in atmospheric and breath monitoring. Taguchi metal oxide semiconductor gas sensors were used to illustrate a proposed protocol for the selection of sensing components in a gas sensor array.Semiconductor gas sensors have been investigated by many workers*-3 for several years. They typically respond to a whole range of combustible analytes, such as hydrocarbon vapours, but their full potential 10 r 8 $ 6 0 Q. i 4 2 0 Sensor 1 Sensor 2 Sensor 3 Sensor 4 Sensors Ethanol Diethyl ether Hexane Light petroleum fl Chloroform Benzene Fig. 2 Bar chart of mean sensor responses116 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 for identification of analytes has never been exploited owing to the cross-selectivities they show to analytes in a mixture.Much work has been carried out on investigating the use of dopants4 to reduce the cross-selectivities and more recently work has been carried out on arrays.5-7 The underlying principle behind the array approach is that each sensor will respond to all analytes but slightly differently, giving a response pattern that can then be used to identify the analyte. However, in order to do this, each sensor in the array must give a characteristic response. Therefore, the main problem in the development of an array is the choice of which sensors to use. To assist in this choice, chemometric techniques can be used as very valuable tools. Various multivariate data analysis methods can be applied to assist in selecting suitable sensors for an array.8 Cluster analysis has been used to identify which sensors are acting indepen- dently of each other in an array,7 as also has factor analysi~.9.*~) As an empirical method, cluster analysis is very useful as it can be used in two ways-first, one can treat the sensors as variables and investigate the array component, and second, using the analytes as variables one can investigate possible analyte-sensor interactions.C h I o roform Benzene Star plot key Fig. 3 Star symbol plots for array sensors Experimental A schematic diagram of the apparatus used to develop and evaluate the smart sensor philosophy is shown in Fig. 1. The computer was a Research Machine Nimbus AX with all software purpose-written in Turbo Pascal to permit full control of the array via the IEEE interface.Through the software the various parameters of the array could be controlled, such as heating temperatures, sensor voltages, necessary for the Taguchi sensors used, and data acquisition times. Before any results were obtained, the computer self-calibrated the array to a set of pre-defined levels, e . g . , a pre-selected level of stability for each sensor in the array must be met. The Microlink(Bi0- Data) provided the necessary analogue to digital and digital to analogue conversions and the gain-amplifier provided the current for the heating circuit and the gain for the sensor response circuit. Statgraphics (Mercia Software) and Clustan (Wishart) software packages were used for data analysis.The sensor array was mounted in a 10 1 glass vessel, with the sensors being placed near the top, but in no particular order. The array consisted of four Taguchi gas sensors, three different types with one duplicate. Here examples will be given to illustrate the design protocol using organic solvents, although other sample types have been studied with the system. Samples were introduced as liquids through a septum in the top of the vessel and a constant flow of air (0.5 1 min-I) was introduced at the gas inlet at the bottom of the vessel. The furnace was used to provide efficient solvent vaporization. Results and Discussion Characterization of Sensor Array Responses The mean responses for each sensor, for each analyte, over a range of concentrations are shown in Fig.2. One can clearly see that the responses for each sensor are not the same, but the duplicate sensors (sensors 2 and 4) do show some similarities. A more informative graphical method for representing these data is the star symbol plot. In these plots the individual sensors are represented as the centre of each star and each point on the star corresponds to the analyte response. The distance between the centre and the points of the star is proportional to the response. These plots are shown in Fig. 3. The similarity between sensors 2 and 4 from the star symbol plot can now be seen more clearly, and it is also apparent that the three sensor types gives different response patterns. Hierarchical clustering was then used to show the exact clustering of the sensors, in order to quantify the similarity or dissimilarity of the sensors.As is common practice, several methods of clustering were used. Only the clusters that remained the same throughout the various methods were considered robust clusters. Commonly, one would plot the cluster 'tree' or dendrogram with a simple point for each of the variables, but in this instance it was thought to be more useful to plot the star symbol at each position of that variable on the dendrogram (Fig. 4). From this combined dendrogram and star symbol plot it can clearly be seen that there are two distinct clusters. The first cluster contains the duplicate sensors and shows that their similarity is very high. The second, a much less significant cluster, was between sensors 1 and 3 and it is worth noting that there is no significance at the 5% level between the two clusters.Q- - _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ p I I I I I I I I I I I 1 I 3 I 1 ' I I I &) I I I I I I I I I I I I I I I I I I I I I 1 I Fig. 4 Combined dendrogram and star symbol plots to show clustering of the gas sensors Characteristic Responses of the Analytes Fig. 5 shows the individual sensor responses for each analyte, and it can be seen that the individual analytes produce a series of patterns associated with the sensors in the array. These patterns or differential responses form the basis by which identification of the analytes of interest can be made. These 10 I 8 a) 2 6 n $ 4 2 0 Ethanol Diethyl ether Hexane Light Chloroform Benzene Solvent petroleum Sensor 1 Sensor 2 Sensor 3 Sensor 4 Fig.5 Bar chart of array response to each solventANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 117 Sensor Sensor 2 3 @ Sensor 1 Sensor 4 Star plot key Fig. 6 Star symbol plots for analytes data have also been represented as star symbol plots (Fig. 61, with the centre of the star being the analyte and the points the four sensors. The one analyte that stands out as giving a totally unique response is chloroform. It is interesting to note how much easier it is to identify similar star plots than it is to identify similar patterns from the bar chart data. Cluster analysis was also carried out on these data and the resulting clusters were plotted as before. The dendrogram (Fig. 7) indicates that there are three clusters in these data: ethanol and diethyl ether (points 1 and 2), hexane, light petroleum and benzene (points 3, 4 and 6) and chloroform (point 5 ) .As each point o n the dendrogram is a mean response 5 5 I ' I I .- Z I I Fig. 7 Combined dendrogram and star symbol plots to show clustering of analytes for the analyte over a concentration range, i . e . , one point is the mean of several replicate values ( n = 81, it is therefore valid to say that chloroform forms a real cluster and not simply an isolated point. There could also be some uncertainty as to whether hexane (point 3) is a member of the first or second cluster. On closer investigation one can see that if the star plot at point 3 were rotated through 180" it would be possible to superimpose it on point 6, indicating that the response patterns for hexane and benzene are negatively correlated. This further illustrates the value of combining the dendrogram with the star plot as it allows one to understand why the clusters are formed and it assists in distinguishing between clusters.The clusters observed relate closely to the chemical nature of the compounds investigated, indicating a correlation between chemical functional groups and sensor response. Conclusions This paper has attempted to illustrate how chemometric techniques, namely cluster analysis, can assist in the selection of sensors for an array device. By adapting such an approach, the selection and design of a sensor array will ensure that the differential response eventually obtained describes the data or analyte(s) with the maximum information.This is a different approach from post-experimental multi-variable data reduc- tion techniques which attempt to identify trends in a data set. In this work we attempted to design a device using chemo- metric techniques to give us a predicted response. In this way, real-time data analysis can be more reasonably achieved and the use of knowledge-based system post-signal generation can be adapted in a simpler way. As an aid to selecting the components of an array the clustering technique will indicate the level of resolution that can be expected for each array sensor for chemically similar compounds. In the example given in this paper, it can be seen that the Taguchi sensors are limited in the selectivity they will show to different organic solvents; this is partly a function of their selectivity and noise characteris- tics. Clearly, the analyst could use the technique outlined to select suitable sensor components from a range of sensor types to custom build a smart sensor. The figures presented in this paper have been reproduced with permission from Elsevier and full details of this work can be found in reference 7 .1 2 3 4 5 6 7 8 9 10 References Logoethetis, E. M., Park, K., Meitzler, A. H., andLand, K. P., Appl. Phys. Len., 1975. 28. 209. Park, K.. and Logocthetis, E. M., J . Electrochem. Soc., 1977. 124, 1443. Watson, J., Sensors Acruators. 1984, 5 . 29. Nitta, M., and Haradome, M.. IEEE Trans. Elecrron Devices. Hidetsugu, A., Tadayosi. Y., Shigehiko, K.. Yoshimasa, T., Yoshikatsu, M., and Shin-Ichi.S., Anal. Chim. Acra. 1987, 194, I . Muller, R . . and Lange. E., Sensors Actuators, 1986, 9, 39. Walrnsley, A. D., Haswell. S. J., and Metcalfe. E., Anal. Chim. Acra, 1991. 242, 31. Sharat, M., Illman, D., and Kowalski. B., Chemometrics. Wiley. New York. 1986. Carey, W., Beebe. K., Kowalski. B., Illman, D., and Hirsch- field, T.. Anal. Chem.. 1986, 58, 149. Rose-Pehersson. S., Grate, J . , Ballantine, D.. and Jurs, P.. Anal. Chem.. 1988, 60, 2801. 1979, ED-26. 247. Multivariate Calibration of Potentiometric Sensor Arrays Robert J. Forster and Dermot Diamond" School of Chemical Sciences, Dublin City University, Dublin 9, Ireland The application of chemometrics in multi-component analysis is a rapidly expanding area of analytical chemistry. In the past, analysis centred on the use of single probes, which were required to be analyte specific (or as near specific as possible) in order to obtain accurate information.More recently, the emphasis has moved from the search for increasingly more selective sensors to the use of sensor arrays and chemometrics. The use of several sensors gives more information about the sample composition while chemometrics allows those contribu- tions arising from analytes and interferents to be decoupled. Recently we described the application of an array of poten- * To whom corrcspondcncc should be addressed.118 ~~ Make an exploration ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 A - I Find function at base point Set new base point t t Make pattern step I I Make an exploration :rease steD Fig.1 Flow chart for method of Hooke and Jeeves tiometric sensors, whose response was modelled via a Nickol- skii-Eisenmann equation including two selectivity coefficients, to the determination of sodium, potassium and calcium in mixed solutions. 1 Central to this approach is the use of iterative optimization techniques to determine simultaneously the parameters of the Nickolskii-Eisenmann equation:2 where Eij is the potential of the jth electrode, measured in the ith sample, Eio is the standard cell potential, aik and ail are the activities of the primary ( k ) and interfering ions (I) in sample i, respectively, K,kl is the selectivity coefficient of the jth electrode with respect to the Ith interfering ion and Sj is the slope of the electrode with respect to the primary ion in the absence of interferents.In contrast to traditional approaches, where parameters are optimized singly or at best painvise, the optimization methods employed in this stud] simultaneously evaluate the four parameters of the Nickolskii-Eisenmann equation. For ex- ample, in order to characterize an ion-selective electrode (ISE) in terms of the Nickolskii-Eisenmann equation using manual methods, the electrode slope and standard cell potential would be determined from the slope and intercept, respectively, of a plot of Ecell versus log(primary ion concentration), for a series of single ion calibration solutions. The determination of selectivity coefficients would require further samples contain- ing interfering ions. This manual, stepwise optimization procedure is both cumbersome and inefficient.More import- ant, a model derived using such a manual procedure is unlikely to reflect accurately the sensor response in real multi-com- ponent samples, as pure solutions would be used in its construction. These problems can be avoided by using mixed calibration solutions and non-linear optimization techniques such as direct search and gradient methods. to an array of ISEs for the determination of sodium, potassium and calcium. p-t-Calix[4]methyl acetate,3 valinomycin and ETH 129 (Fluka) immobilized within plasticized poly(viny1 chloride) matrices as described previously3,4 formed the basis of the sodium-, potassium- and calcium-selective electrodes, respectively, and are labelled Na, K and Ca. A sparingly Xe ExDansion A Fig.2 simplex method Basic operations. reflection. expansion and contraction of selective sensor based on p-t-calix[4]tetraoximes was similarly constructed and is labelled SSE. The data required for sensor modelling via the Nickolskii-Eisenmann equation were obtained in mixed solutions of sodium, potassium and calcium chloride. A three-factor, two-level, factorial experimental design was chosen, giving 24 mixed solutions for the calibration of the €our-sensor array. In the calibration procedure the potential of each electrode within the array was recorded in each of the 24 calibration solutions. The data were stored in a file in ASCII format, allowing post-run analysis with software applications packages to be performed. These data were then used in conjunction with non-linear optimization techniques to determine simultaneously the cell potential (E!)), the slope (S,) and selectivity coefficients (K,kl).The objective function for these optimization procedures was to minimize the residual sum of squares (RSS) to less than 10-1(), this being defined by Experimental These non-linear optimization procedures have been appliedANALYTICAL PROCEEDINGS, APRIL 1991. VOL 28 119 Table 1 Model parameters (with standard deviations obtained using five separate calibration sets on a single array in parentheses) Ej’lmV Sj/mV per decade Kj,Na+ X lo3 K j . K + X lo3 Kj,Ca7+ X lo3 Sensor TY PC ISE Na. j = Na+ -69.99 (0. IS) 54.9 (0.05) 1 20.4 (0.9) 17.3 (1.54) 4.06 (0.05) 1 5.09 (0.05) K, j = K+ 54.0 (0.21) 53.9 (0.04) Ca.j = Ca2+ 54.0 (0.19) 27.1 (0.08) 0.01 (0.0005) 0.016 (0.0013) 1 ; = Ca2 t 56.61 (0.89) 50.98 (0.11) 7.54 (0.38) 5.39 (0.28) 1 SSE 1 = Na+ 82.0 (0.43) 5 1 .o (0.09) 1 715 (35.4) 364.2 (18.9) 1398 (66.9) 1 185.5 (8.90) j = K+ 74.56 (0.54) 50.98 (0.10) where E,,’ are the estimates of the potential of electrode j in sample i provided by the model and Ell are the recorded experimental potentials. This means that the sum of squared deviation between the predicted potentials and those recorded experimentally for all 24 calibration solutions is required to be less than 10-lOmV. Initial estimates of El() and S, were obtained from the intercept and slope of a plot of E,] versus log(activity) of the primary ion. For each of the interfering ions, initial estimates of 3 x 10-3 and 0.5 were used for the selectivity coefficients of the highly selective and sparingly selective electrodes, respectively.Optimization Procedures The optimization procedures employed were based on direct search and gradient methods. Two direct search techniques were employed, namely the Hooke and Jeeves method6 and modified simplex optimization.7 A gradient search method, that of Davidon-Fletcher-Powell,x was also examined. The two approaches differ in that direct search methods use only functional evaluations for optimization, whereas gradient methods also use information about the rate of function decrease to improve algorithm efficiency. Hooke and Jeeves Method This algorithm initially carries out an exploration of the function behaviour in a region about a base point b l .At this base point the function value is calculated. Each variable is then incremented by a step length h, (which can be variable for each point) and the functional value evaluated at each of these points. If this move reduces the function value then the original base point is replaced with this new value. Otherwise the base point is reduced by the step length h, and, provided that the function is reduced at this point, bl - h, becomes the new base point. If neither case decreases the function then the same base point is retained and the process repeated with a smaller step size. When the function has been successfully reduced, i.e., the base point changed to point b2, a pattern move is initiated. The pattern moves seek to exploit the information gained in the exploration step.The direction given by b2 - bl is most likely to produce a further function decrease, hence the function is evaluated at the next pattern point: and in general Exploratory moves are continued about P l ( P i ) . If this produces a function reduction, then a new base point has been reached and a pattern move is then made. Otherwise the pattern move is abandoned and further exploration made. These processes are continued until the step size has been reduced to some predetermined small value. Fig. 1 gives a simplified flow chart of the procedure. P , = bl + 2(b3 - bl) P, = b/ + w/ + 1 - 6,) Simplex Optimization The method described here is based on Nelder and Mead’s method and utilizes the notion of a simplex, which in n-dimensional space is simply a set of n + 1 mutually equidistant points.Thus, for example, in two dimensions the simplex is a triangle whereas in three dimensions it is a regular tetrahedron. The optimization is achieved by comparing the function value at the n + 1 vertices and moving this simplex over the response surface in a structured manner so as to minimize the objective function. The simplex is allowed to distort and become non-regular to increase the algorithm efficiency. In this contribution the performance of the method is further improved by constraining the variables to regions where realistic solutions will be obtained. In the present instance this involves restricting the electrode slope to between 45 and 60 mV per decade and maintaining positive selectivity coeffi- cients. If any of these conditions is violated then the operator is informed.I . From the vertices of the initial simplex xl, x2, ..., x,, the function values f ( x I ) , f ( x 2 ) , .. ., f(xII) are evaluated. 11. These function values are then sorted in descending order fil, fg, ..., fl with corresponding vertices xh, xg, ..., x I . 111. The centroid of all points except xI, is then found and labelled xo: xo = l / n C x, i # h f(xo), i.e., ,fo, is then evaluated. IV. An attempt to reduce the objective function is then made by moving away from x/, by reflection. xII is reflected in xo to give x, andf(x,.) = f r found. This process is illustrated in Fig. 2. If a is the reflection factor then x, - xo = a(x0 - x h ) and x , = [(l + a)xo] - axl,. It is evident that where (Y = 1 that a mirror image is obtained.V. f, is now compared with fi; if f r < fi then the lowest function value has been obtained and so the direction from xo to x, appears to have been successful in reducing the function. An expansion along this direction is carried out to find x, and f(x,) = f, evaluated. Fig. 2 illustrates this process. Iff, <fi then xh is replaced with x, and the n + 1 points of the simplex are tested for convergence to the minimum. If this has been achieved then the procedure is terminated, otherwise step I1 is repeated. If fe is not less than x , the simplex has extended too far in the direction of xo - x, and hence x, is abandoned. Instead xh is now replaced with x,, which yields an improve- ment. If the convergence criterion has been satisfied the optimization is complete; if not, step I1 is repeated.If f, > fi but f r is not greater than fg (where fg is the second largest functional value), x, is an improvement on the two worst points of the simplex and x , replaces xh. Satisfaction of the convergence criterion results in termination, otherwise step I1 is repeated. If f, > fi and fr > fg, then step VI is performed. Table 2 Electrodc characteristics as determined using manual methods (with standard deviations for three different sensors in parentheses) Sj/mV per decade Kj.Na+ X lo3 Kj.K+ x 10’ K ~ , , - ~ Z + x 103 Sensor Type ISE Na, j = Na+ 54.7 (0.87) 1 12.5 (1.10) 3.16 (0.31) K , j = K+ 53.8 (0.91) 0.41 (0.03) 1 0.16 (0.02) Ca,j= CaZ+ 27.2(1.13) 0.020 (0.002) 0.028 (0.002) 1 SSE j = N a + 53.8 (1 .05) 1 158 (16.1) 75 (8.1)120 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 VI.f r andfh are now compared; iffr >fh a contraction step is performed, otherwise xh is replaced with x, before performing the contraction. Iff, > fh the simplex has moved too far in the direction xh to xo. For this reason a contraction is performed. If f r > f h , x, is found from x, - xo = P(Xh - xo) where 0 < P < 1. Thus x, = fixh + (1 - P)xo (Fig. 2 ) . Iff, <fi,, xh is replaced with x, and x, is given by x, = Pxr + (1 - P)xo VII. If at this point f, < f h , then x h is replaced with x, and convergence checked. Failure to satisfy convergence requires a return to step 11. If > f h it appears that the search for a value less than fh has failed and so the simplex size must be reduced.VIII. The simplex is now reduced by halving the distance of each point of the simplex from XI, the point generating the lowest functional value. Therefore, xi is replaced with xI + 1/2(xi - XI), i.e., xi is replaced with 1/2(xj + x,). The functional values are again calculated at each of the n + 1 nodes and the convergence tested; if convergence is not achieved then step I1 is repeated. IX. The test of convergence is based on the standard deviation of the n + 1 function values being less than a predetermined value. Davidon-Fletcher-Powell Method Gradient methods use descent properties of the function to be minimized in order to improve the efficiency of the computa- tional algorithm. Several gradient methods employ quadratic functions as a Taylor expansion can approximate any function in the vicinity of its minimum, unless all of its second derivatives are zero.The quadratic function F(x) = a + x% + 1/2xTGx where a is a constant, b a constant vector and G is a positive definite symmetric matrix which has a minimum at the point x*, where x* is given by x* = -G-lb Subject to certain continuity conditions, any function can be approximated in the region of a point xo by +(x) = f ( X 0 ) + (X - xo)TVf(xo) + 1/2(x - xo)TG(xo>(x - xo) where G(xo) is the Hessian matrix at xO. A reasonable approxi- mation forf(x) might be the minimum for +(x). If the latter is at x, then Wxo) + G(xo)(xm - xo) = 0 Therefore, X, = - G-'(xO)V~(XO) xrn = xo - G-l(xo)g(xo) Thus from the point xi the next approximation for the minimum should be xi + 1 = xi - G- '(xJg(xJ or x;+ 1 = X; - h;G-l(xi)g(xi) where hi is determined by a search in the direction G- l(xi)g(xi).The Davidon-Fletcher-Powell method avoids calculating the inverse Hessian matrix @'(xi) at each step by setting the search direction at stage i as - H i g ( x j ) , where Hi is a positive definite symmetric matrix which is updated at each stage. An initial point xo and a symmetric positive definite matrix HO, usually the unit matrix, are selected. I. At stage i there is a point xi and a positive symmetric definite matrix Hi. 11. The search direction is set to d. = - H . . lgl 111. A linear search is performed along the line xi + Adi to IV. Set find the value hi which minimizesf(xj + hidi).v . = h.d. I 1 1 V. Set xi+ 1 = xj + vi VI. f ( x i + and gi+ I are calculated. The procedure is terminated if 1 gi+ 1 or 1 v; I is sufficiently small, otherwise the procedure is continued. VII. Set ui = gi + 1 - gi VIII. The matrix H is updated by setting H;+ 1 = Hi + Ai + Bi where A j 1 v~v;'/( ~i'ui) Bi = - H;uiu;'Hi/( u;'Hiui) IX. i is increased to i + 1 and step I1 repeated. Results and Discussion The algorithms described use different approaches to achieve a common objective, i.e., minimization of the cost function [equation ( 2 ) ] and the accurate determination of the paramet- ers of the Nickolskii-Eisenmann equation. The convergence of the experimental potentials and those predicted by the model, as indicated by the dependence of the residual sum of squares on the iteration number, is illustrated in Fig.3 for the three algorithms. Fig. 3 shows that the three algorithms are capable of satisfying the convergence criterion of reducing the residual sum of squares to less than 10-10. However, the rate of convergence is dependent on the method employed, with the gradient method being superior. The computational capacity of second- and third-generation personal computers is such that the increased complexity of the simplex or gradient methods does not significantly increase the time taken per iteration. This means that the more rapid convergence rate of the simplex and gradient methods results in a direct decrease in the time required for model building. This reduction in the model building time can be significant. For instance, the application of simplex optimization to the modelling of the sodium sensor required only 100 s compared with 340 s for the method of Hooke and Jeeves.Gradient methods frequently provide more rapid convergence, most noticeably where there are a large number of variables involved. This gives enhanced performance particularly where the function is quadratic or near quadratic, with such functions converging in n iterations where n is the number of variables. The disadvantage of gradient methods, however, is that as they require the calculation of a differential, noise may represent a greater problem. As an illustrative example of the effect of noise on the convergence behaviour of the Davidon- Fletcher-Powell method, the dependence of the residual sum of squares on the iteration number is given in Fig.4 for unfiltered data. Fig. 4 shows that the optimization procedure is robust in that the convergence criterion is ultimately satisfied. However, noise does significantly increase the number of iterations required. Where the signal to noise ratio is low, a direct search method such as simplex can converge more rapidly than the gradient method and produce a minimum time for model building. However, for computer-controlled poten- tiometry such as that described here, a large number of data points can be rapidly acquired, allowing digital signal process- ing and filtering techniques to be applied. The use of a software digital pre-filter can avoid the problems associated with the calculation of a differential in gradient methods of optimiza-ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 121 tion, allowing the rapid convergence of the gradient method to be exploited. Further, such signal processing can greatly improve the accuracy and precision of the electrode charac- teristics calculated using non-linear optimization procedures compared with traditional single-electrode measurements.I-------------I 90 s 200 400 600 - 45 s I I - I I 40 80 120 160 - 20 s 20 40 60 80 - 20 s I I 0 35 70 105 140 Iteration No. Fig. 4 Dependence of residual sum of squares (RSS) on the iteration number for the Davidon-Fletcher-Powell method applied to un- filtered data mixed solutions.9 The non-linear optimization procedure described here circumvents these problems by using mixed calibration solutions and by modelling the sensor response over an extended activity range.The accuracy of the modelling procedure is ultimately reflected in the predictive ability of a modelled array. Table 3 gives the errors obtained for a series of independent samples within (samples 1-S), above (samples 9-11) and below (samples 12-14) the original concentration range used for modelling. The errors observed are typically no larger than 3%, which is superior to traditional single-electrode tech- niques.") Significantly, the errors are independent of the sample composition, suggesting that the modelled sensor array can decouple signals arising from primary and interfering ions. This means that the modelled array is capable of determining low levels of a single cation in the presence of a large and widely varying excess of the other two with significantly improved accuracy and precision compared with traditional methods.Iteration No. Fig. 3 Convergence of experimental potentials with those predicted by the model as indicated by the residual sum of squares. (a) Method of Hooke and Jeeves; ( h ) constrained simplex optimization; ( c ) Davidon-Fletcher-Powell method Table 3 Prediction of sodium, potassium and calcium concentrations using modelled four-sensor array (with % error in parentheses) Sample No. [Na+]/mmol dm-3 [K+]/mmol dm-3 [Ca*+]/mmol dm-3 Model Parameters Table 1 gives the model parameters for the array determined using the non-linear optimization methods described, and shows that the model parameters are all well defined and have good precision. The precision of these parameters is a consequence not only of the modelling procedure, but also of the precision of the individual potentials measured, which results from the capture and processing of a large number of data points. Table 2 gives the electrode slope and selectivity coefficients as determined by traditional methods.The slope observed is the same for both methods whereas significant differences exist between the measured selectivity coefficients. Selectivity coefficients have traditionally been determined using a single concentration of separate ion solutions for ions of the same charge and the fixed interference (mixed solution) method for ions of differing charge. These methods are unsatisfactory for the characterization of sensors to be used in real samples as they do not take account of the known concentration dependence of the selectivity coefficients and the possibility of interactive effects occurring between ions in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 126.7 (2.89) 120.9 (2.11) 125.7 (-0.91) 119.9(-2.85) 121.5 (3.00) 126.8 (2.82) 124.2 (-0.99) 122.4 (-0.38) 187.6 (-2.56) 207.4 (-0.67) 50.0 (-2.85) 210.3 ( I .22) 65.4 (1.77) 60.2 (-0.33) 3.11 (-1.31) 3.01 (2.11) 5.22 (2.33) 4.87 ( - 1.48) 3.12 (2.44) 2.89 (-2.31) 5.09 (2.27) 4.94 (-0.44) 20.88 (3.08) 14.56 (1.09) 0.51 (3.01) 0.31 (1 .00) 9.22 (-2.33) 0.11 (-2.01) 0.88 (2.97) 0.77 (1 .OO) 1.22 (2.67) 1.22 (0.87) 0.78 (2.11) 1.22 (-2.09) 0.75 (- 1 .O) 1.27(-0.18) 9.67 (-0.23) 5.23 (-3.09) 5.09(1.19) 0.21 (0.38) 0.08 (3.10) 0.098 (-2.86) Conclusions The application of non-linear optimization techniques allows the potentiometric sensor response to be characterized in terms of the Nickolskii-Eisenmann equation.The equation param- eters can be simultaneously determined, unlike traditional sequential evaluation. The approach described also has the considerable advantage that the sensor response is determined in mixed solution, a procedure which cannot be easily accomplished using more traditional parameter determination methods. When applied to sensor arrays the non-linear model characterizing the sensor response gives superior analytical performance compared with traditional single-electrode122 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 methods using calibration graphs.Several different optimization algorithms have been des- cribed and their respective merits considered. In general, the more sophisticated approaches exhibit more rapid con- vergence, which typically results in a direct reduction in the time required for model building. Although gradient tech- niques are more prone to noise, the application of a digital pre-filter will in most circumstances allow the increased efficiency of gradient methods to be exploited. The calibration and modelling procedure described avoids the problems associated with traditional methods of defining selectivity coefficients and produces a model that can ac- curately predict unknown concentrations. Given the known inaccuracy and imprecision of the separate solution method and the availability of inexpensive, powerful computer data acquisition systems, perhaps it is now time to re-assess the methods of selectivity coefficient determination and to re-define standard techniques.1 2 3 4 5 6 7 8 9 10 References Forster. R. J., Regan, F., and Diamond, D., Anal. Ciiem.. in the press. Morf, M., Tulanta, 1979, 26, 719. Cadogan, A. M., Diamond, D.. Smyth. M. R.. Deasy. M., McKervey. A. M.. and Harris. S. J . , Analyst. 1989, 114. 1551. Amman, D.. Morf. W. E . , Anker, P.. Meier, P. C., Pretsch. E., and Simon, W., ion-Sel. Electrode Rev.. 1983. 5. 3. Forster, R. J . , Cadogan, A., Telting-Diaz, M., Diamond, D., Harris, S . J . , and McKervey, M. A.. Sensors Actuators. in the press. Hooke. R., and Jeeves, T. A . , J . Assoc. Cotnpr. Much., 1961. 8. 212. Nelder, J . A., and Mead, R., Cornput. J . , 1965, 7, 308. Fletcher, R.. and Powell, M. J . D., Compur. J . , 1963. 6, 163. Ion-selective Electrode Methodology, ed. Covington. A. K., CRC Press, Boca Raton. FL, 1979, ch. 1. p. 18. Tclting-Diaz, M.. Regan. F.. Diamond, D.. and Smyth. M. R . , J . Phurm. Biomed. Anal.. in thc press. RSC ANALYTICAL AND FARADAY DIVISIONS ELECTROANALYTICAL AND ELECTROCHEM ISTRY GROUPS A Meeting on MICROELECTRODES FOR ANALYSIS will be held in Oxford, May 22nd, 1991 The meeting will take place in St. Catherine's College, Manor Road, Oxford, commencing at 10.30. The papers will be: 'Microelectrodes for Analysis' by A. M. Bond (La Trobe University, Bundoora, Australia); 'Micro Liquid-Liquid Interfaces' by H. H. Girault (Edinburgh); 'Microelectrodes and Enzymes' by P. N. Bartlett (Warwick); 'Microelectrode Sensors for Gas Analysis' by D. Pletcher (Southampton); 'Microband Electrodes' by D. E. Williams (University College, London); and 'New Instrumentation for In-vivo Monitoring of Catecholamines' by D. Crespi (Nottingham). Registration, including lunch and refreshments: Members of RSC, f25; Non-mem bers, f40; Students, retired and non-employed members, f 15. Cheques payable to 'Electroanalytical Group RSC' should be sent t o A. E. Bottom, ABB Kent-Taylor Ltd., Oldends Lane, Stonehouse, Gloucestershire GLIO 3TA. Information on the meeting is available from the same address.

ISSN:0144-557X    

DOI:10.1039/AP9912800115

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 123 New Horizons in Electrophoresis ~~ The following is a summary of one of the papers presented a t a Meeting of the Analytical Division held on October 30th, 1990, in the Scientific Societies’ Lecture Theatre, London W1. Recent Developments in Gel Electrophoresis of Proteins Michael J. Dunn DeDartment of Cardiothoracic Surgery, National Heart and Lung Institute, Dovehouse Street, London SW36LY Proteins are charged at a pH other than their isoelectric point (pl) and migrate in an electric field in a manner dependent on their charge density. If the sample is present as a narrow zone, proteins of different mobilities will travel as discrete zones and separate during electrophoresis. Electrophoresis is an ideal technique to resolve the individual components of protein mixtures.Such separations are best carried out in a support medium to counteract the effects of convection and diffusion and to facilitate immobilization of the separated proteins. The high resolution capacity of techniques based on the use of polyacrylamide makes these the methods of choice for most applications. Polyacrylamide has a considerable toxicity hazard, but this problem can now be overcome using pre-mixed reagents available as pre-weighed powders or ready-made stock solutions from a variety of suppliers. A range of electrophoretic techniques are available, but the most popular and important are polyacrylamide gel electrophoresis (PAGE) in the presence of the anionic detergent sodium dodecyl sulphate (SDS), by which proteins can be separated on the basis of their molecular size, and isoelectric focusing (IEF) in polyacrylamide gels, which can be used to characterize proteins in terms of their pls.A third technique, two-dimensional (2-D) PAGE, which combines IEF in the first dimension with SDS-PAGE in the second dimension, is particularly useful for the analysis of complex protein mixtures. In this paper, some of the recent developments in the use of these three procedures are discussed. In addition, some recent advances in the visualization and analysis of electrophoretic patterns and the characterization of proteins separated by gel electrophoresis are described. Readers requiring a more general introduction to the field are referred to some recent books. 1-5 SDS-PAGE Sodium dodecyl sulphate is an effective solubilizing agent for a wide range of proteins.Most proteins bind 1.4 g of SDS per gram of protein, thereby masking the intrinsic charge of the polypeptide chains, so that the net charge per unit mass becomes approximately constant. Electrophoretic separation is then dependent only on the effective molecular radius (rM) of the proteins and occurs as a result of molecular sieving through the gel matrix. The polyacrylamide gel concentration used determines the effective separation range ( e . g . , a 5%T gel will separate proteins in the range 25-200 kD). The separation range can be extended using gradient gel systems, which are particularly useful for samples containing proteins with a wide range of relative molecular masses.Ready-made homogeneous and gradient SDS-PAGE gels are now available for a variety of apparatus and gel formats and their use should improve the reproducibility of protein separations. The speed of separation is often of great importance to the analyst and recent developments i n gel miniaturization have now made it possible to achieve SDS-PAGE separations in less than 1 h. Advances are also being made in the automation of electrophoretic procedures, again providing advantages of speed and con- venience. Isoelectric Focusing (IEF) Isoelectric focusing is a high-resolution method in which proteins are separated in the presence of a continuous pH gradient. Under these conditions the proteins migrate accord- ing to their charge until they reach the pH values at which they have no charge, i.e., their pl.The proteins attain a steady state of zero migration and are concentrated into narrow zones. The most popular method for the generation of pH gradients for IEF is the incorporation of low relative molecular mass carrier ampholytes (CA) into the gel matrix. Using CA it is possible to generate wide (pH 2-10) or narrow (1-2 pH units) pH gradients able to separate proteins whose pls differ by 0.01 pH unit. In theory, IEF is a stable, equilibrium technique, but in practice the electroendosmotic properties of IEF gels result in instability of pH gradients. This phenomenon results in cathodic drift so that with time the pH gradient and the proteins within it migrate towards the cathode, causing decay of the pH gradient and loss of proteins from the gel.This also results in poor reproducibility of separations, which is exacer- bated by batch to batch variability in CA preparations. In addition, certain proteins can interact with CA, resulting in the formation of complexes with altered pl. An important recent innovation has been the development of Immobiline reagents (Pharmacia-LKB) for the preparation of gels containing immobilized pH gradients (IPG). A detailed review of this topic has been published.6 These reagents are a set of acrylamide derivatives with different pK values. The IPG gels are prepared as slabs on plastic supports by generating a gradient using the appropriate Immobiline solutions. In this way the buffering groups forming the pH gradient are covalently attached to the polyacrylamide backbone, resulting in pH gradients that are infinitely stable owing to the elimination of cathodic drift (but not electroendosmosis).Using this technique narrow pH gradients spanning as little as 0.1-1.0pH unit can be generated, capable of separating proteins whose pls differ by as little as 0.001 pH unit. The original Immobiline reagents, particularly the alkaline species, suffered from problems of hydrolysis and auto-polymerization, resulting in poor reproducibility of separations and reduced shelf-life. These problems have been overcome with the Immobiline I1 reagents in which the basic species are dissolved in anhydrous propan-1-01 and the acidic species are stabilized with an inhibitor. Two-dimensional PAGE Two-dimensional PAGE, in which proteins are separated by IEF under denaturing conditions followed by SDS-PAGE, is124 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 generally used for the characterization of complex protein mixtures.In the method most frequently used, based on that developed by O’Farrell,7 the IEF dimension is carried out in CA IEF gels in glass tubes and is able to resolve up to 2000 proteins (16 x 16cm gels). Resolution capacity can be improved using larger gel formats ( e . g . , 32 x 41 cm; 5000 proteins) or faster separations can be achieved using miniature formats but at the cost of resolution (<lo00 spots). A major problem in 2-D PAGE is variability between spot profiles, and this can be significantly improved using large batches of gels in each dimension. Of course, CA IEF for 2-D PAGE suffers from the same problems as one-dimensional (l-D) IEF discussed above.Indeed, electroendosmosis is so severe in glass tube IEF gels that pH gradients rarely extend much above pH 7. The use of IPG gels for the IEF dimension is an attractive alternative but there were initially considerable problems associated with their use. Many of these problems have now been overcome (for a review, see reference B ) , allowing good 2-D separations with improved inter-gel spatial reproducibility to be obtained. Visualization The most popular staining procedure for polyacrylamide gels is still that based on the use of Coomassie Brilliant Blue (CBB) R-250 in methanol-water-acetic acid (45 + 45 + 10). However, this method involves staining and destaining steps and the method is relatively insensitive (about 0.5 pg of protein per band).A 10-fold increase in sensitivity can be achieved using CBB R-250 or G-250 in a colloidal form9 with the additional advantage that a destaining step is not required. Silver staining methods are even more sensitive (about 1 ng of protein per band). A variety of protocols have been described (reviewed in reference 10) and commercial kits are now available which should give improved reproducibility. Radio- actively labelled proteins are usually detected using standard autoradiographic or fluorographic procedures, but alternatives with increased dynamic range and sensitivity are now becoming available, such as the use of storage phosphor imaging technology. 11 Quantitative Analysis Quantitative data can be obtained from l-D gels using the many commercial densitometers and software packages avail- able.Similar data can be retrieved from 2-D gels and it is possible using sophisticated computer packages (reviewed in reference 12) to generate 2-D protein databases (reviewed in reference 13), which can be used to search for changes in patterns of protein expression. Western Blotting Gel electrophoresis provides no direct information on the identity and biological activity or function of the separated proteins. One of the most important approaches to this problem is provided by techniques of Western blotting. In this methodology, proteins are transferred (blotted), usually by the application of an electric field perpendicular to the plane of the gel (electroblotting), onto the surface of a support such as nitrocellulose. Proteins immobilized on the surface of a support are readily available for interaction with a variety of ligands such as antibodies and lectins, which can be used to characterize particular components of the electrophoretic profile.A good review of this topic and its applications is available.14 A major recent advance in this area has been the development of systems based on the use of enhanced chemiluminescence for the detection of proteins on blots. 15 This method is extremely sensitive (<1 pg compared with 10 pg using horseradish peroxidase-diaminobenzidine), allowing expensive and rare antibodies and other ligands to be used at very high dilutions. Chemical Characterization A recent major advance also based on the use of blotting technology has been the development of methods for the direct chemical characterization of proteins from 1-D and 2-D gels.In this approach proteins separated by electrophoresis are elec- troblotted onto special supports such as poly(viny1idene difluoride) (PVDF) , polypropylene or glass-fibre (nitrocellu- lose is not generally used as it is not resistant to many of the reagents used in protein chemistry). The blot is then stained and the bands or spots of interest are cut out and subjected directly to amino acid analysis, peptide mapping and N-termi- nal or internal amino acid sequence analysis (reviewed in reference 16). These techniques have sensitivities in the low picomole range so that it is now a practical possibility to isolate proteins to purity from an initial complex mixture in one step by gel electrophoresis and obtain direct information on their chemical composition This approach will be invaluable for protein identification and provides a route for the cloning of genes for hitherto unidentified proteins through the use of synthetic oligonucleotides and the polymerase chain reaction (PCR).1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Gel Electrophoresis of Proteins: a Practical Approach, eds. Harnes, B. D.. and Rickwood, D., IRL Press, Oxford, 2nd edn., 1990. Righetti, P. G., Isoelectric Focusing: Theory, Methodology and Applications. Elsevier, Amsterdam, 1983. Allen, R. C., Saravis, C. A., and Maurer, H. R.. Gel Electrophoresis and lsoelectric Focusing of Proteins, Walter de Gruyter, Berlin.1984. Gel Electrophoresis of Proteins, ed. Dunn, M. J . , Wright, Bristol, 1986. Andrews, A. T., Electrophoresis: Theory, Techniques, and Biochemical and Clinical Applications, Clarendon Press, Oxford, 2nd edn., 1986. Righetti, P. G., Immobilized p H Gradients: Theory and Methodology. Elsevier, Amsterdam, 1990. O’Farrell, P. H.. J . Biol. Chem., 1975, 250. 4007. Gorg. A.. Postel, W.. and Gunther. S., Electrophoresis. 1988, 9, 531. Neuhoff, V., Arold, N., Taube, D., and Ehrhardt. W., Electrophoresis. 1988, 9, 255. Rabilloud. T.. Electrophoresis, 1990, 11, 785. Johnston, R. F.. Pickett, S. C., and Barker. D. L., Electro- phoresis, 1990. 11, 355. Miller. M. J.. in Advances in Electrophoresis. cds. Chrambach. A., Dunn, M. J.. and Radola, B. J . , VCH, Wcinheim, 1989, vol. 3. p. 181. Celis, J . E., Madsen. P., Gesser. B., Kwee. S . , Nielsen, H. V.. Rasmussen, H. H., Honore, B., Leffers. H., Ratz, G. P., Basse, B., Lauridsen. J . B., and Celis. A., in Advances in Electrophoresis. eds. Chrambach, A.. Dunn, M. J., and Radola. B. J., VCH, Weinheim, 1989, vol. 3, p. 1. Baldo. B. A.. and Tovey. E. R., Protein Blotting: Methodology, Research and Diagnostic Applications. Karger. Bask, 1989. Durrant. I.. Nature (London). 1990, 346, 297. Aebersold. R., in Advances in Electrophoresis. eds. Chram- bach, A.. Dunn, M. J., and Radola, B. J., VCH, Weinheim. 1991. vol. 4. p. 81.

ISSN:0144-557X    

DOI:10.1039/AP9912800123

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYTICAL VAM Va I id i t y PROCEEDINGS, APRIL 1991, VOL 28 Programme of Analytical Measurements* Bernard King and Geoffrey Phillips Laboratory of the Government Chemist, Teddington, Middlesex The Need for Measurement The analysis of a material to determine its composition is one of the earliest branches of chemistry and, together with physical measurement, is basic to modern society. An estimated one billion analytical measurements made annually in the UK all play a vital part in ensuring the quality of commercial products, in addition to, for example, assisting the Government in the development of policy and the proper enforcement of regula- tion. Assessment of environmental issues (such as the depletion of the ozone layer, the quality of waterways and acid rain), food and agriculture enforcement, the quality of diet, the protection of public health, the development of novel agrochemicals, the quality of raw materials and finished products and new product R&D in the chemical and pharmaceutical sectors, all critically depend on the validity of data provided by analytical chemists.Many aspects of Government business need valid results of chemical analysis. National and international trade depend on measurement of chemical composition both for revenue classification and for the removal of technical barriers to trade in the Single European Market and for wider free-trade through the GATT. In order to reduce costly and time-wasting duplication of effort caused by re-testing, measurements undertaken in one country must be recognized and mutually acceptable in others.The Validity of Analytical Measurement If, as has been estimated, 10% of the Gross National Product (GNP) of industrialized countries is associated with the costs of measurement and, as shown in one US study, at least one in ten of tests need to be repeated, the costs associated with obtaining the wrong answers are enormous, apart from unnecessary actions taken as a consequence of inaccurate measurements. Obtaining the necessary accuracy and precision when under- taking analytical measurements is, therefore, not only impor- tant in terms of product quality and community welfare, but it also contributes to the economic well-being of the nation. The Laboratory of the Government Chemist (LGC) has for some time been concerned that the chemical analysis sector has lagged far behind the physical and engineering sectors in its ability to demonstrate the validity of its data and the compatibility of results between different laboratories.Government policy on the National Measurement System is detailed in the White Paper ‘Measuring up to the Competition (Cmd. 728)’. Although the primary responsibility for ensuring the validity of test data rests with the producers and users of the data, chemical analysis was identified as an area of special concern. The LGC was tasked with co-ordinating the develop- ment of a national and international advisory structure and * This article is a summary of a paper expected to appear in the August issue of Chemistry in Britain. It describes the background of the LGC initiative aimed at improving the ‘Validity of Analytical Measurements’ and facilitating the mutual recognition of analytical data between European countries.Bernard King is Deputy Director/ Government Analyst and is VAM Programme Manager; Geoffrey Phillips, who was recently appointed as Visiting Professor in Analytical Chemistry at Glasgow College of Technology, has responsibility for special VAM projects. 125 with DTI financial support to promote the concept and benefits of analytical quality assurance (AQA) and to encourage adoption of systematic quality systems as an integral part of laboratory procedures. The quality of analytical results largely depends on the skill and ability of analytical chemists, continuously attentive to the capabilities and limitations of the techniques they employ.It has been recognized that analytical results obtained in one laboratory might not be comparable to those produced in another, or even when obtained in the same laboratory at a different time by an alternative method or analyst. In such instances, the analyst would be failing to meet a fundamental obligation, to provide the customer with results of demon- strable and requisite accuracy. The VAM Programme The VAM programme is a long-term commitment to develop- ing an analytical dimension to the National Measurement System (which already provides a sound basis for the compari- son of physical measurements). Five identifiable strands in the first phase include: the support for validated methods; increased availability and regular use of certified reference materials; universal adoption of internal AQA procedures; wider provision of proficiency testing; and increased third party audit and assessment of a laboratory’s QA protocols.The VAM programme has variously tackled these com- ponents: by establishing a technical infrastructure wherein analytical confidence and co-operation can develop; promoting the benefits of AQA and of laboratory accreditation; produc- ing and promoting reference materials; developing and vali- dating methods, and advising on their application and perfor- mance; encouraging the use of proficiency testing and inter- comparison studies throughout Europe; and conducting or supporting research into emerging analytical techniques. A video has been made to explain the importance of VAM in the analytical laboratory, the benefits of its adoption and the difficulties caused by disagreement over results.A central aim of VAM is to promote a UK and European infrastructure within which analytical laboratories will be able to demonstrate the validity of their data. This will extend to the mutual recognition of test data and facilitate the free move- ment of trade within Europe by 1992. CHEMAC The Chemical Measurement Advisory Committee (CHEMAC) was set up in order to help define the require- ments of VAM and to provide a focus for issues in the UK. CHEMAC has been formed as a committee of 25 senior analytical chemists representing a wide range of interests and includes the President of the RSC’s Analytical Division. The main aims of CHEMAC are to address problems associated with traceability, variability and uncertainty in analytical measurement. It also seeks to stimulate the development of a UK Chemical Measurement System including laboratory QA, the supply and training of analytical chemists, promotion of co-operation in areas of mutual interest and provision of a co-ordinated input into national and international organiza- tions which affect the practice of chemical analysis.126 The symposium will provide a forum where interesting and useful applications of atomic spectroscopy can be reported and discussed.In addition to plenary, invited and submitted lectures, a particular feature of the meeting will be the presentation of posters. There will also be an exhibition and a social programme for delegates and their guests.This meeting is organized by the Atomic Spectroscopy Group, Analytical Division of The Further information can be obtained from the Chairman of the organizing committee: I ~ 1 ~ Royal Society of Chemistry. ~ ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 EURACHEM This last aim has been addressed mainly through EURACHEM. Launched at a meeting in Frankfurt in November, 1989, by a group of laboratory directors from ten European countries, this now consists of delegates drawn from government, academe and industry in 16 of the 18 EC and E n A nations. In this forum, European analysts can collabor- ate through a network of national laboratories with a common interest in validated measurements. It works closely with other organizations already active in this area variously to promote awareness of quality problems, training of analysts, develop- ment of validated methods, traceability of reference materials, expansion of proficiency testing schemes, and AQA systems based on good laboratory practice and EN 45000.Activities have included support for development of certifi- cation and testing infrastructure required to realize the Single European Market. Promotional material has been prepared to alert technical organizations, educational establishments and senior decision makers to the importance of valid analytical measurements. Workshops have been planned for experts from across Europe to exchange information and develop best practice in areas of proficiency testing and education and training; and working groups will pursue these aspects.National specialist groups have also been formed in Ger- many and The Netherlands; Norway has held an open meeting involving 150 laboratories. In Sweden, consultation is focused through the Swedish Chemical Society and their National Testing Laboratory, while Belgium and France are promoting EURACHEM directly through their QA and accreditation networks. Education, Training and Status The education and training of analysts is seen as a cornerstone of validation of analytical measurement. The LGC runs a regular series of QA training courses for analytical personnel. More generally, a EURACHEM working group chaired by Professor Tolg (Institut fur Spektrochemie, Dortmund) aims: to provide professional and academic bodies with an input on chemical metrology and quality issues; to promote the exchange of analytical chemists between national laboratories; and to facilitate technology transfer through short QA training courses directed to working analysts and managers.The first workshop, in Strasbourg, is to exchange views in QA training and to make recommendations on a core syllabus for short courses. Following a recent review by CHEMAC, a working group has been set up to examine the supply, education and training of analytical chemists in the UK. This group will build on the report, ‘A Strategy for Analytical Chemistry’, prepared for the RSC’s Analytical Division by a working party chaired by its former President, Doug Squirrell. Although the VAM programme is essentially concerned with tactics, many of its general objectives march comfortably within the strategic comments on industrial importance, applied research, integration of analysis in secondary and tertiary education and professional status subsumed in the Squirrell Report. The question of status for analysts is further endorsed by the RSC in the promotion of an Indicative Register for appropriately experienced and qualified analytical chemists. Conclusion The VAM programme has involved considerable resource commitment because adoption of systematic and transparent laboratory QA protocols, bolstered by appropriate training of analysts and managers, is believed to be the most important factor likely to influence the development of chemical analysis into the next decade. Collaboration with the RSC and identification of common interest can be mutually beneficial. The Sixth Biennial National Atomic Spectroscopy Symposium will be held at the Polytechnic South West, Plymouth gfh BNASS c3 6th BNASS 22-24 July 1992 Dr S Hill, Department of Environmental Sciences, Polytechnic South West, Drake Circus, Plymouth, Devon PL4 8AA, UK.

ISSN:0144-557X    

DOI:10.1039/AP9912800125

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 127 Equipment News Mass Spectrometer The Optima stable isotope ratio mass spectrometer offers total accessibility: the ion optical components, the dual inlet and the trapping systems are all easily acces- sible above the bench. When the 10 or 20 sample manifold is fitted, it too is totally accessible for cleaning or maintenance. Reliability of small samplc performance of the sample preparation, manifold and inlet systems has been enhanced by the use of a new fabrication system which produces seamless joints in the pipework. The Optima features multi-tasking soft- ware controlled through the mouse and windows system. VG Isotech, Aston Way, Middlcwich, Cheshire CWlO IHT. Data System for Mass Spectrometry The MACH 3 data acquisition system for use with existing mass spectrometers makes use of a SUN workstation for ultra-high-speed 32 bit processing with 4 Mbytes of main memory (expandable to 16 Mbytes).A 16 in FST colour monitor with 1152 x 900 pixel resolution displays the graphics based intuitive user inter- face. Control of the instrument and the acquisition process are handled by the mass spectrometer control processing unit connected to the SUN by an industry standard VME bus, providing close-cou- pled high-speed 32 bit data transmission throughout the system at up to 40 MHz. One unique feature of the system is its ability to display data in real time during acquisition. The MACH 3 system has built-in Ethernet networking capability. Many network protocols are supported, including DECnet, allowing the system to be integrated into a local area network.Kratos Analytical Ltd., Barton Dock Road, Urmston, Manchester M31 2LD. Mass Spectrometer Data System A new data system features the COM- PLEMENT multi-tasking software pack- age. Multi-window displays and mouse functions make MS operation and the reduction of data extremely easy. The system uses the 32 bit HP 9000 series engineering workstation, which supports high capacity hard disks, a 650 Mbyte erasable optical disk drive and many peripherals, including laser printers. With its UNIX-based operating station, the system can be linked to other instruments, such as NMRs, via Ethernet. COMPLE- MENT also contains an embedded library search algorithm using library databases including chemical structures.The new data system is available with all new Jeol mass spectrometers and users can retrofit it to existing SX, DX, AX and HX instruments. Jeol (UK) Ltd., Jeol House, Silver Court, Watchmead, Welwyn Garden City, Hertfordshire AL7 1LT. Mass Spectrometry Software General Multi Dynamic software GMD-2 has been included as a new analysis task in the latest software for the VG Sector 54 mass spectrometer. It enables the user to construct his own multi-dynamic collector analysis routine without the need for formal programming. The new software uses a spreadsheet format in the Dynamic Editor so that masses can be assigned in a Mass Array to an individual collector (up to 9), as in the case of a static multi- collector array. In addition, magnetic field jump sequences can be assigned and integration times set up.Thus, it is possible to measure more isotopes than there are collectors on the instrument. The second half of the Dynamic Editor is a function array, consisting of 20 func- tions, which are template equations that can be applied to the data generated. VG Isotech, Aston Way, Middlewich, Cheshire CWlO 1HT. Microwave Sample Preparation System The MDS-2000 microwave sample prepa- ration system simplifies preparation of samples for atomic absorption and emis- sion spectroscopy and other analytical procedures. The versatile menu driven programming allows users to enter the sample mass, number of vessels, volume per vessel and chemical composition. The program for each sample can consist of up to 5 separate stages including microwave power, run time, pressure control set point, and run time at the control point.Twenty or more multi-step sample prepa- ration programs can be stored and printed in hard copy or transferred to a computer via the system interface ports. CEM Corporation, P.O. 200, Matthews, NC 28106, USA. Atomic Absorption Spectrometer The Model 41OOZL’s unique transverse heated graphite tube design with uniform temperature profile and integrated small mass L’vov platform allows direct analysis of most samples against aqueous refer- ence solutions. It is the first AA spec- trometer to use the longitudinal Zeeman background correction, with magnetic field parallel to the optical light beam. Direct compatibility with the FIAS-200 flow injection system extends system capabilities to include fully automated FIA mercury hydride analysis.Perkin-Elmer Ltd., Maxwell Road, Beaconsfield, Buckinghamshire HP9 1QA. Nebulizer The Sharp Spray Cone Nebulizer was developed for ICP spectrometry by Dr. Barry Sharp of Loughborough University in conjunction with the British Technol- ogy Group. It is designed for analysing samples with high dissolved solids and solutions containing particles and slurries, and it can be used with all the makers’ ICP instruments. Perkin-Elmer Ltd., Maxwell Road, Beaconsfield, Buckinghamshire HP9 1QA. Spray Chamber A new glass spray chamber and torch unit has been introduced for the PU7000 advanced inductively coupled plasma spectrometer. It offers improved wash through times, better precisions and pro- longed aspiration periods for aqueous solutions containing a high level of dis- solved solids.A special conversion kit allows current PU7000 users to benefit from the new sample systems. Philips Scientific, Analytical Division, York Street, Cambridge CB1 2PX. Spectrophotometer An updated version of the Spectronic 3000 diode array spectrophotometer offers even faster scanning and features an improved version of the user-friendly operating software, Rapid*Scan. This latest software enables the exceptional speed of the diode array detection to be used to the full, with scanning rate improving from 6 scans per minute to 15 scans per minute over the entire analytical range. Milton Roy European Laboratory Group, Oaklands Park, Fishponds Road, Wokingham, Berkshire R G l l 2FD.128 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Spectrophotometer Accessory The LISR-3100 integrating sphere attach- ment from Shimadzu can be used for measurement of reflectance spectra of solid samples, such as powders, papers and cloth, into the near infrared part of the spectrum. It is also used for trans- mission measurement of solutions and solid samples, featuring high stability and excluding the influence of the state of the sample surface.It is used in conjunction with the Shimadzu UV-visible-NIR scan- ning spectrophotometer and has a wavelength range of 240-2400 nm. V. A. Howe and Co. Ltd., Beaumont Close, Banbury, Oxfordshire OX16 7RG. Spectrometers Two new spectrometers bring the benefits of fast FTIR at start-up costs typical of dispersive IR systems.The PU9600 Series features a keypad control model for rou- tine FTIR and a PC-driven instrument for more sophisticated analytical applica- tions. Designed and manufactured to BS.5750 and IS09000 standards, the PU9600 series combines advanced elec- tronics, precision optics and powerful software integrated into a high quality optical bench. The instruments operate in point for fundamental parameters deter- mination of element concentrations. PX is supplied in PC and VAX versions and is compatible with either the Philips or Criss XRF 11 fundamental parameters pack- ages. Philips Analytical, P.O. Box 218, 5600 MD Eindhoven, The Netherlands. Transfer Line for Thermal Desorber A fused silica lined, heated transfer line and splitter kit for the TD 4 single shot thermal desorber is now available.De- signed for direct connection to a GC analytical column, the system provides uncompromised high resolution capillary chromatography without the use of a Philips PU9600 FTIR spectrometer the mid-IR range between 7000 and 360 cm-* with a resolution of better than 2 cm-1, and an option is available to extend the range down to 200 cm- 1. Both models benefit from built-in self diagnostics, ultra-fast Fourier transform for rapid sample analysis, and sealed and desic- cated optics for a fast, purge free start-up. High energy throughput is available for difficult, highly absorbing samples. Philips Analytical, York Street, Cam- bridge CBl 2PX. Software for XRF Analysis A choice of two packages for the makers’ sequential X-ray spectrometers enables users to opt for either an interactive or fully automatic approach to semi-quanti- tative analysis of large or small unknown samples.Requiring a minimal number of standards, PSA (Philips Semi-quantita- tive Analysis) uses qualitative scans from the X40 analytical software as the starting capillary injector on the GC. The new line will connect to any GC and is compatible with all types of GC column from ‘/i in packed to narrow bore capillary. It can be retrospectively fitted to any TD 4 unit. Perkin-Elmer Ltd., Maxwell Road, Beaconsfield, Buckinghamshire HP9 1QA. Gas Chromatography Column The GS-Alumina GC column is designed for the analysis of saturated and unsatu- rated hydrocarbons. It can separate closely related compounds in light hydro- carbon blends, and trace impurities can be determined.The GS-Alumina column’s large internal diameter (0.53 mm) makes it easily adapted to packed column GC systems, even those using valves for sam- pling and/or column switching. The col- umn is available in both 30 and 50 m lengths. J and W Scientific, 91 Blue Ravine Road, Folsom, CA ( 3330-4714, USA. Sample Preparation Consumables An extensive range of Hyperclean syringe filters and Hypersep solid-phase extrac- tion columns for rapid sample preparation by selective absorption and elution prior to analysis has been introduced. Hyper- clean syringe filters are available in sterile and non-sterile form in four sizes from 4 to SO mm. Colour coded according to which of six different membranes they contain, they have acrylic or poly- propylene housings.Pore sizes are typic- ally 0.2 and 0.45 pm, but other sizes are available. Hypersep columns can have one of 14 different sorbent bed packings, through which samples can be driven by positive pressure or drawn by vacuum or centrifugation. A 12- or 24-position vac- uum manifold is available for large scale sample preparation. Shandon Scientific Ltd., Chadwick Road, Astmoor, Runcorn, Cheshire WA7 1PR. Syringe Filters A range of syringe filters have a 1 pm borosilicate glass-fibre pre-filter mounted on top of either a nylon or a cellulose acetate membrane and are especially use- ful for laboratories doing tissue culture and cell culture media filtration. They offer 300% more filtrate than plain mem- branes, eliminate pre-filtration require- ments, reduce hold-up volume, minimize sample loss and prolong membrane life.HPLC Technology Ltd., Wellington House, Waterloo Street West, Maccles- field, Cheshire S K l l 6PJ. Microfiltration Capsules Suitable for preparative filtration for HPLC and for sterilizing air or gases, the disposable Polycap TF versatile microfil- tration capsule consists of a compact polypropylene housing with a hydro- phobic PTFE membrane. This combina- tion will effectively filter aqueous aerosols and provide gases that are dry and particle free. Polycap TF is offered in two size configurations and four porosities. The 0.1 and 0.2 pm pore sizes are for filtration sterilization: the 0.45 L L r n one is for large scale HPLC solvent filtration, and the 1.0 pm unit, which incorporates a poly- propylene pre-filter, offers exceptional high loading capability.Whatman Scientific Ltd., Springfield Mill, Maidstone, Kent ME14 2LE. Refractive Index Detector The ERC-7535 refractive index detector is intended for use in semi-preparative HPLC. I t operates at flow rates of up to 150 ml min-1 over a measuring range from 1 to 256 x lo4 refractive index units. It offers excellent linearity and features a sophisticated optical autozero, which ensures high accuracy and repeatability. Applied Chromatography Systems Ltd., The Arsenal, Heapy Street, Macclesfield, Cheshire S K l l 7JB.ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 129 Chiral Workstation The Chiral Workstation is a culmination of the development of the Chiramonitor and the high quality of the ACS pumps and UV detector. As an integrated pack- age it features central control of opera- tions via the PC-based Midas+ data management system.The ACS Chiramo- nitor, in its new Mark I 1 version, is available separately with 10-fold more sensitivity than previously. Applied Chromatography Systems Ltd., The Arsenal, Heapy Street, Macclesfield, Cheshire SKI 1 7JB. Solid-phase Extraction Cartridges A range of chromatographic cartridges for the removal of ionic compounds prior to ion chromatography is available. There are five Hypersep-IC cartridges, covering most ion chromatography applications involving hydrophobic components, anions, cations, halides and sulphate. They can be selected to retain either the component of interest for subsequent analysis or to remove interfering ions.For large scale sample preparation a 12- or 24-position vacuum manifold is avail- able. Shandon Scientific Ltd., Chadwick Road, Astmoor, Runcorn, Cheshire WA7 IPR. Ion Chromatograph The 2.690.0030 ion chromatograph is suitable for the determination of anions and cations simultaneously, in a large number of samples. The system accom- modates two separating columns, i.e., an anion and a cation column. By manual switching one or other eluent stream can be directed through the detector. After switching, samples can be injected imme- diately because the use of two pumps allows both columns to be constantly equilibrated and hence always to be ready for measurement. An integrator suffices for the evaluation and for the automation of an autosampler.V. A. Howe and Co. Ltd., Beaumont Close, Banbury, Oxfordshire OX16 7RG. Gel Permeation Chromatography-Viscometry System A fully automatic GPC-V system com- prises a 160-position autosampler, iso- cratic pump, viscosity detector and dedi- cated GPC-viscometry software. At the heart of the system is the Viscotek vis- com e t e r-re f r a c t o m e t e r detector , w h i c ha makes use of the universal calibration curve for all polymers. The system can therefore give accurate relative molecular mass information of even novel polymers using any convenient standard. Roth Scientific Co. Ltd., Alpha House, Alexandra Road, Farnborough, Hamp- shire GU14 6BU. Trace Oxygen Analyser The Model 3060 ultra-trace parts per billion oxygen analyser, tested and evalu- ated under a joint effort with Union Carbide Industrial Gases Inc., features the following specification: 0-50 ppb of oxygen low range, 1 ppb of oxygen sensitivity and +2% full-scale accuracy.The Model 3060 has four ranges of analysis with excellent linearity over them, so that there is no need to recali- brate when changing ranges. A brochure giving information on the Series 300 analysers is available. Teledyne Analytical Instruments, The Harlequin Centre, Southall Lane, South- all, Middlesex UB2 5NH. Residual Gas Analyser The current model of the Dataquad RGA, first introduced in the early 1980s, has been designed for the more demand- ing needs of the 1990s. It features 3-decade logarithmic scans, split-screen display, background store and subtract, automatic library search, leak detection with built-in audible signal, alarm out- puts, remote video output, printer port and an RS232 interface for data logging and full remote control.Spectramass Ltd., Radnor Park Industrial Estate, Congleton, Cheshire CW12 4XR. Air Pollution Monitors A range of toxic gas air pollution monitors makes possible portable monitoring for ten commonly encountered toxic gases including carbon monoxide, hydrogen sulphide, oxides of nitrogen, sulphur dioxide and phosphine. The Model TX40 instruments are ideal for COSHH surveys and provide hard copy via a recorder- logger output. A full calibration service is offered. Bedfont Technical Instruments Ltd., Bedfont House, Holywell Lane, Upchurch, Sittingbourne, Kent ME9 7HN. Multi-gas Monitor The Bruel and Kjaer Type 1302 is a highly accurate, microprocessor controlled monitor which can selectively measure up to five component gases and water vapour in any air sample.Optical filters installed in the 1302 can be chosen to select those gases requiring most frequent detection in a particular application. The measure- ment is based on the photoacoustic infrared detection met hod. Livingston Hire Ltd., Livingston House, 2 4 Queens Road, Teddington, Middlesex TW11 OLB. pH Electrode The 6.0219.100 combined pH electrode from Metrohm uses a sleeve type dia- phragm instead of a conventional ceramic type; this eliminates the possibility of blocking. It is of double-junction design, so that the outer (bridge) electrolyte can be easily changed to suit the analysis solution. The electrode is ideal for measurement in a range of solutions from low ionic strength water to non-aqueous samples.V. A. Howe and Co. Ltd., Beaumont Close, Banbury, Oxfordshire OX16 7RG. Diluting System Hamilton now has a complete series of OEM-instruments available in the UK through V. A. Howe. Microlab OEM versions are modular built-in systems and are individually customer tailored. V. A. Howe and Co. Ltd., Beaumont Close, Banbury, Oxfordshire OX16 7RG. Viscometer Rheocord90 is a computerized torque rheometer. As an extruder capillary vis- cometer with rod or slit die capillaries it will subject thermoplastic melts (which are non-Newtonian liquids) to variable levels of flow shear rates. The flow rate of the melt passing through the capillary dies is automatically measured by weighing the extrudate during a given time span on a balance with an RS232 computer port.The pressure drop along the capillary is also measured and correlated to the flow rate. The computer transforms these data to both shear stresses and viscosity as a function of shear rate. All of these data are given as hard copies both in table form and/or as a graphical plot. Haake Mess-Technik GmbH u. Co., Dieselstrasse 4, D-7500 Karlsruhe 41, Germany. Thermal Analysis Instrument The DSC12E is a compact, simple to operate thermal analysis instrument with built-in measuring cell. The measured values are presented either on an attached dot matrix printer or via the built-in serial interface to a computer. The easy to use TA89E control and evaluation software runs under the familiar and widespread MS Windows 286.It is used for the comparison and storage of measurement data and for evaluations with selectable programs. Mettler-Toledo AG, CH-8606 Greifensee, Switzerland. Thermo-mechanical Analysis Module The TMA 92 module from Setaram of France is designed to be used alone with its own control system and personal com- puter, or as part of the Setaram System 92 modular package, which also includes thermogravimetry, differential scanning calorimetry and differential thermal analysis modules. The temperature range of the TMA 92 is from - 150 to + 1000 "C and the instrument is capable of measur- ing dimensional changes of just 0.01 pm by applying loads of up to 150 g. Roth Scientific Co. Ltd., Alpha House, Alexandra Road, Farnborough , Hamp- shire GU14 6BU.130 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 Micropore Analysis System The ASAP 2000 micropore system includes four independent data reduction techniques, identified in the literature as the Horvath-Kawazoe method, the MP method, the t-plot method and the Dubinin method (Dubinin-Radushkevich and Dubinin-Astakhov models).Types of analyses include micropore size and pore volume, micropore distribution and micropore surface area. Micromeritics, One Micromeritics Drive, Norcross, GA 30093-1877, USA. Sedigraph Software The new Quickstep Program for the Sedigraph 5100 provides all the functions of the standard operating modes, plus short cuts. It eliminates almost all key- board input by automatically entering the keystrokes necessary to perform analyses, rinse the system and print reports.Micromeritics, One Micromeritics Drive, Norcross, GA 30093-1877, USA. Hydrometers Nalge Company and the Ever Ready Thermometer Company introduce safe, break -resis t a n t E RTCO-N a1 ge ne plain- form hydrometers made from poly- carbonate. Each is individually cali- brated. The hydrometers are available in three specific gravity ranges (1 .000-1.200, 1.200-1.420, 1.400-1.620) for fluids heavier than water, two Sugar Brix ranges (0-35 and 0-50%) and one salt range Nalge Company, 75 Panorama Creek Drive, Box 20365, Rochester, New York (0-100% ) . 14602-0365, USA. Furnaces A new chamber furnace is the first to offer the capability for reliable, fully automatic firing at temperatures up to 2000 "C. The ZCF 2000 has a chamber of 2.7 1 capacity.Also announced is the VTF Series of horizontal tube furnaces, specifically developed for vacuum firing at nominal temperatures of 1200 and 1500°C. The VTF Series is available in two standard inside tube diameters of 50 and 75 mm, offering heated lengths of 600 mm. Lenton Thermal Designs Ltd., Unit C2, Valley Way, Welland Industrial Estate, Market Harborough, Leicestershire LE6 7PS. Microfiltration Devices Puradisc 25 AS microfiltration devices are now supplied in a new handy dispenser pack. Puradisc 25 AS is recommended for optimum sample recovery from aqueous samples in small-scale purifications. They are ultra-clean, chemically inert disc devices and, being primarily intended for use with syringes, have female luer lock inlet and male slip lock connections. Whatman Scientific Ltd., Springfield Mill, Maidstone, Kent ME14 2LE.Microfiltration Capsules Polycap AS is a compact microfiltration capsule housing a pleated microporous nylon filter membrane of 0.2, 0.45 or 1.0 pm porosity. Being naturally hydrophilic, the membrane can be used with most aqueous solutions, for which it provides high flow rates and throughput. All Poly- cap AS devices incorporate a glass-micro- fibre pre-filter, which extends the life of the nylon membrane and substantially accelerates the speed of filtration. Of the two sizes available, Polycap 36 AS offers a filtration capacity of up to 10 1; Polycap 75 AS offers up to 20 1. Whatman Scientific Ltd., Springfield Mill, Maidstone, Kent ME14 2LE. Filter Discs Sartolon 250 polyamide membranes are one of the first of many new filtration products scheduled to be produced at the Company's West German manufacturing facility.Replacing the Sartolon SM 200 series, the new product offers a maximum sterilization temperature of 134 "C and improved flow rate characteristics. Sartorius, Longmead Business Centre, Blenheim Road, Epsom, Surrey KT19 9QN. Water De-gassing System A complete water de-gassing system which will deliver pure water at a rate of up to 1.6 I min- 1 is announced. The unit is safe and economical and provides stable de-gassing without heating or use of pressurized gases. It prevents the forma- tion of air bubbles, removes any dissolved oxygen or carbon monoxide and ensures that bacteria levels are kept to a mini- mum.Applied Chromatography Systems Ltd., The Arsenal, Heapy Street, Macclesfield, Cheshire SK11 7JB. Molecular Modelling Software ChemMod I1 is an interactive 3-D mol- ecular modelling software package, which runs under the Unix operating system and has the ability to handle large molecules. Optimized graphics performance provides an ideal environment for the interactive manipulation of solid structures. Fraser Williams Group, Port of Liver- pool Building, Pier Head, Liverpool L3 1BY. Molecular Modelling Software Oxford Molecular Ltd. has announced its appointment as value added reseller for Apple Computer (UK) Ltd. Oxford Molecular developed the NEMESIS molecular modelling package, which, in conjunction with the new low-priced Apple Macintosh computers, makes high- performance molecular graphics available from under f2.500.Oxford Molecular Ltd., Terrapin House, South Parks Road, Oxford OX1 3UB. Ion Gun The AS10 is a new ion gun producing an intense ion beam at a fixed energy of 500 eV. Removal of surface contamination at this low energy minimizes sample damage and damage to deeper layers. The AS10 fits to a standard 2% in chamber port and has a long working distance to accommo- date most chamber and sample sizes. An optional precision leak valve and gas manifold for the source gas supply is also available. VSW Scientific Instruments Ltd., Warwick Road South, Old Trafford, Manchester M 16 OJT. Strip-chart Recorders Following the recent launch of its PX105 circular chart recorders and MX control- lers, the makers have redesigned their former PlOOM and P600M strip-chart recorders under the new designations: PXl00 for continuous writing recorders and PX600 for 6-point dotted versions.ABB Kent-Taylor Ltd., Howard Road, Eaton Socon, St Neots, Huntingdon, Cambridgeshire PE19 3EU. Database and Analysis System TekBase is a combination of a relational database management system and a rich variety of analytical? statistical and graph- ical presentation facilities. It allows scien- tists to intcrpret and present data in a variety of new ways. The system is partic- ularly suitable in analytical and spectro- scopic applications, as it has been de- signed to capture and analyse the often massive volumes of complex data that are the hallmark of chemical analysis. Tek- Base allows a laboratory to link up equipment from a variety of techniques, for example, mass spectroscopy, gas chro- matography, infrared chromatography, scintillation techniques, etc.Leading Technology Products, 11 Grosvenor Place, London SWlX 7HH. Equilibrated Phenol for Molecular Biology Applications A pre-equilibrated phenol solution elimi- nates all of the preparatory handling of phenol solutions associated with this reagent. The preparation involves the addition of 8-hydroxyquinoline to a satu- rated phenol solution. This is then extrac- ted with a 1.0 mol dm-3 Tris buffer solution initially, followed by a 0.1 mol dm-3 Tris buffer solution until the pH of the aqueous phase is greater than 7.6. Fisons Scientific Equipment? Bishop Meadows Road, Loughborough, Leices- tershire LEI 1 ORG.Bottle and Vial Crimper Crimpmate is a sturdy benchtop crimper that enables the operator to crimp even the largest autosampler load with mini- mum fatigue. The vial or bottle is simply placed in the adjustable jaw and then crimped in a single swift action by pullingANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 down the Crimpmate lever. Summers Row, London N12 OLD. Chromacol Ltd., Glen Ross House, Sharps Container The new Clinipak benchtop sharps con- tainer offers superior safety and hygiene in a compact disposable unit. The lid of the Clinipak incorporates a device that enables the user to remove and discard the needle from the body of a syringe without having to touch the needle. The Clinipak measures 200 mm in diameter and is 170 mm in height; it is packed in cases of 40 units.Radleys, Shire Hill, Saffron Walden, Essex CBll 3AZ. Broken Glass Disposal Box The new Scienceware broken glass dispo- sal box is a disposable, highly visible receptacle for use in laboratories, hospi- tals and schools. Made of sturdy corru- gated cardboard, it is supplied with a heavy duty 0.05 mm polythene bag to contain glass fragments and spilled liquids. I t is available as a floor (12 x 12 X 27 in) or benchtop (8 x 8 x 10 in) model. Radleys, Shire Hill, Saffron Walden, Essex CBll 3AZ. Spill Treatment Kits Each Spill-X kit comprises three specially formulated powders, safety goggles and gloves, pans, scraper, waste bags, instruc- tions and treatment guide. The kits, which are easily replenished, are supplied in durable plastic cabinets, suitable for wall mounting but easily detachable when needed.BDH, Broom Road, Poole, Dorset BH12 4”. Fabric for Protective Workwear A leading UK analytical laboratory in- volved in research related to public health has changed from 100% cotton fabrics to blended cotton-Dacron fabrics for its laboratory coats. The laboratory under- took a programme of high temperature (70°C) wash tests on Scala cotton rich blended fabric. After repeated laundering at these high temperatures Scala showed no signs of shrinkage or degradation. Klopman International, Film House, 142-150 Wardour Street, London W1V 3AU. Polythene Backed Absorbent Sheets Benchkote Plus is intended for the safe capture of drips and accidental spills to prevent major hazards to health in the laboratory.This improved version of the original Benchkote is thicker to cushion falling glassware. It also provides a larger area of protection from stains and corro- sion than other bench coverings. BDH, Broom Road, Poole, Dorset BH12 4”. Literature A new booklet has been produced by the Medical Working Group of the Chemical Industry Safety, Health and Environment Council of the CIA to assist companies to provide safe working conditions. It deals with the protection of the eyes. CIA Publications Department, Kings Buildings, Smith Square, London SWlP 355. A leaflet explains how to avoid health risks from exposure to solvents at work. Published by the Health and Safety Executive, it is called ‘Solvents and You’. HSE, Baynards House, 1 Chepstow Place, Westbourne Grove, London W2 4TF.The CIA has produced a new guidance which deals with transfer connections for the safe handling of anhydrous ammonia. CIA Publications Department, Kings Buildings, Smith Square, London SWlP 355. A leaflet explains the COSHH com- pliance audit designed to identify areas that still need attention and to provide professional advice from top hygienists. The National Occupational Hygiene Service Ltd., Skelton House, Manchester Science Park, Lloyd Street North, Manchester M15 4SH. The CIA has issued two more booklets of guidance on the implementation of spe- cific aspects of legislation for the control of substances hazardous to health. One deals with the problem of how to set in-house occupational exposure limits for substances which have not yet been as- signed official exposure limits.The other provides guidance on the need to provide information, instruction and training as required by Regulation 12. Other book- lets on Assessments (Regulation 6) and 131 Health Surveillance (Regulation 11) have already been produced. CIA Publications Department, Kings Buildings, Smith Square, London SWlP 355. Two brochures are announced: ‘Ion and Atom Guns’ and ‘PRISM-Imaging Time of Flight Mass Spectrometer’. Cambridge Mass Spectrometry Ltd., Saxon Way, Bar Hill, Cambridge CB3 8SH. An application sheet entitled ‘The Opti- misation of UV-VIS Spectrophotometric Samples’ is now available. Beckman, Progress Road, Sands Indus- trial Estate, High Wycombe, Bucking- hamshire. A brochure describes the new PC con- trolled scanning spectrophotometers from Shimadzu. It outlines the features of the UV2101PC UV-visible spectrophoto- meter and the UV310lPC UV-visible- NIR instrument. V. A. Howe and Co. Ltd., Beaumont Close, Banbury, Oxfordshire OX16 7RG. A two-page flyer explains why the new Spectrolab F spectrometer for metal analysis can perform reliable QC tests at a reasonable price. Spectro Analytical Instruments, Bosch- strasse 10, D-4190 Kleve, Germany. A brochure entitled ‘Chromatography- Customised Systems’ gives details of spe- cially engineered chromatography solu- tions to a host of analytical problems. Philips Analytical Chromatography, York Street, Cambridge CB1 2PX. A brochure describes the Metrohm range of equipment for the automation of elec- trochemical techniques. It also includes application articles for many common analyses. V. A. Howe and Co. Ltd., Beaumont Close, Banbury, Oxfordshire OX16 7RG. Literature is available describing the PL- ETA multi-tasking thermal analysis workstation, which features the simul- taneous operation of up to 7 modules, including DMTA, DSC, TGA, TMA and STA. Thermal Sciences, Polymer Labora- tories Ltd., The Technology Centre, Epi- nal Way, Loughborough, Leicestershire LE1 OQE. A leaflet outlines a range of five field emission scanning microscopes. The models featured include the JSM 6400F, the JSM 6600F, the JSM 6300F and the established JSM 840F and JSM 890F. Jeol (UK) Ltd., Jeol House, Silver Court, Watchmead, Welwyn Garden City, Hertfordshire AL7 1LT.132 ANALYTICAL PROCEEDINGS, APRIL 1991, VOL 28 A 28-page brochure details Nikon’s tometry systems, and micro-manipula- Haybrook, Halesfield 9, Telford, Shrop- extensive range of microscopes and acces- tors. shire TF7 4EW. sories, photomicrography and micropho- Nikon UK Ltd., Instruments Division,

ISSN:0144-557X    

DOI:10.1039/AP9912800127

出版商:RSC    

年代:1991

数据来源: RSC