ISSN:0144-557X    

DOI:10.1039/AP98522FX041

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

Analytical Proceedings Proceedings of the Analytical Division of The Royal Society of Chemistry CONTENTS 31 3 Editorial 31 3 Award of Second Robert Boyle Medal 314 Analytical Division Awards 31 5 New Members of Council 31 7 Summaries of Papers 31 7 321 'The Automated Analytical Laboratory of the Future' 'New Infrared Analysis of Foods' 'Computer Aided Methods in Chromatography' 'Advances in Mass Spectrometry' 'Expert Systems for the Analyst' Health and Safety in the Chemical Laboratory - Where do we go from here? AN PRO1 22( 1 1 ) 31 3-344 (1 985) areas. 323 333 334 336 Equipment News 341 Analytical Chemistry Trust Fund 341 Conferences and Meetings 341 Erratum 342 Course 342 New British Standards 344 Analytical Division Diary This publication provides an overview of health and safety developments in the chemical laboratory and workplace, and will provide essential reading for anyone involved in these Brief Contents: Accident and Dangerous Occurrence Statistics in the United Kingdom; Morbidity and Mortality Studies; Economics of Health and Safety Measures; Procedures and Statistics in France; Professional Negligence, Liability and Indemnity; The System in the United States of America; The System in the United Kingdom; The System in the Federal Republic of Germany; Hazards of Handling Chemicals; Hazards of Apparatus, Equipment and Services; Managing People; What Standards Should We Use? Conflict of Safety Interests with Legislation; The Protection of Workers Exposed to Chemicals: the European Community Approach; Recommendations Arising from the Symposium.Special Publication No. 57 Softcover 206pp 0 85186 945 9 Price f 16.50 ($30.00). RSC Members f 12.00 Ordering: Non-RSC Members should send their orders to: The Royal Society of Chemistry, Distribution Centre, Blackhorse Road, Letchworth, Herts SG6 1 HN, England. RSC Members should send their orders to: The Royal Society of Chemistry, Membership Officer, 30 Russell Square, London WC1 B 5DT. The Royal Society of Chemistry Burlington House, Piccadilly London WlV OBN Electronically typeset and printed by Heffers Printers Ltd, Cambridge, England November 1985

ISSN:0144-557X    

DOI:10.1039/AP98522BX043

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

313 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Award of Second Robert Boyle Medal The Second Robert Boyle Gold Medal of speeches of Professor Burns and Mr. in and the development of analytical the Analytical Division (which is awarded Cobb, made a lengthy and wide-ranging chemistry since the early part of the to distinguished analytical chemists work- speech covering many aspects of his work ing abroad) was presented by the Pre- sident of the Division, Mr P. G. W. Cobb, to Professor I. M. Kolthoff, formerly of the University of Minnesota, on the even- ing of September loth, 1985. The venue for the presentation was Quarry Bank Mill at Styal in Cheshire, which is a working museum of the (princi- pally cotton) weaving industry. The mill was the building chosen for the Analytical Symposium’s reception at the IUPAC Congress, held in Manchester from Sep- tember 8th-13th. The Boyle Medal commemorates the life and work of Robert Boyle, and Professor D. Thorburn Burns mounted a display of Boyle memorabilia in the reception room, including the three parts so far published of his biography of Boyle (Analytical Proceedings, 1982, 19, 222; 1982, 19, 288; 1985; 22, 253). He also spoke briefly about Boyle and the Robert Boyle Gold Medal, drawing attention particularly to Boyle’s American and Dutch connections (Professor Kolthoff having been born in The Netherlands). century. * ProTfessor Kolthoff, in reply to ’the Professor I. M. Kolthoff (L) receives thesecondAD Robert Boyle Medalfrom Mr. P. G. W. Cohb

ISSN:0144-557X    

DOI:10.1039/AP985220313b

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

314 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Analytical Division Awards A number of presentations took place Professor Belcher in the field of chemical Mr. Johnson received a presentation during the Analytical Division’s Research education. The award is made to a gradu- scroll from Mr. Cobb and Mrs. Ruth and Development Topics meeting, which ate student studying in any British Uni- Belcher, the widow of the late Professor was held from June 26th-28th, 1985, versity or Polytechnic. After his lecture Belcher. Dr. M. S. Cresser (L) receives the twelfth SAC Silver Medal from the President of the AD, Mr. P. G. W. Cobb Dr. J. F. Alder receives the eleventh SAC Silver Medal from the President in The Queen’s University of Belfast. The first lecture of the meeting was a Society for Analytical Chemistry Silver Medal Lecture, given by Dr.M. S. Cresser of the Department of Chemistry and Soil Science, University of Aberdeen. At the end of Dr. Cresser’s lecture the President of the Analytical Division, Mr. P. G. W. Cobb, presented him with the twelfth SAC Silver Medal. The afternoon session on June 27th began with another Silver Medal Lecture, this one being given by Dr. J. F. Alder of the Department of Instrumentation and Analytical Science, UMIST. At the conclusion of the lecture Mr. Cobb presented the eleventh SAC Silver Medal to Dr. Alder. The opening event of June 28th was the delivery of the first Ronald Belcher Memorial Lecture by Mr. Simon John- son, an SAC Student at the University of Cambridge. The Ronald Belcher Memorial Lecture is awarded annually, by the Analytical Division, to com- memorate the long career of the late Mr. S. Johnson receives a scroll from Mrs. R. Belcher and Mr. P. G. W. Cobb

ISSN:0144-557X    

DOI:10.1039/AP9852200314

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 315 New Members of Council Dr. Bill Campbell is currently Senior Analyst in the Research and Technology Department of ICI Petrochemicals and Plastics Division, based at Wilton on Teesside. Born at Stevenston, Ayrshire, in 1949, he was educated at the local High School and Ardrossan Academy before embarking on a BSc course in Chemistry at Strathclyde University, Glasgow. Sur- viving this without disillusionment and with a heightened interest in matters analytical he decided to undertake research toward a PhD under the super- vision of (the now) Professor John Otta- way at the same institution: the subject of this research was the better understanding of “Carbon Furnace Atomisation in Atomic Absorption Spectroscopy.” In 1975 he joined the then Petrochemicals Division of ICI at Wilton, where his interests included Energy Dispersive XRF and non-metals analysis.Later, he turned to a more business related analy- tical role which involved him in a diversity of classical, chromatographic and spectro- scopic techniques as they applied to the solutions of problems in the Bulk and Speciality Chemicals Areas. This led eventually to his current interest in the analytical chemistry of surfactants and the analysis of detergent formulations. He is at present Chairman of the international CESIO/AIS working group on the analy- sis of these materials and actively involved with the UK SDIA/GOSIP industry panel and the relevant BSI panel. Dr. Campbell joined the SAC in 1972, and became an FRSC in 1984.He has written for Annual Reports on Analytical Atomic Spectroscopy (ARAAS) since 1977 and has been Topic Editor since 1980. Married, with one son, much of his spare time is occupied with family mat- ters, but when business and family com- mitments permit he takes an interest in winemaking, photography and travel and has been known to resort to indoor footbalVcircuit training in an attempt to hold back the inevitable. A biography of Dr. Christopher Burgess was published in Anal. Proc., 1982, 19, 422, on the commencement of his first period of service on Council. Dr. Arnold Fogg was born in Radcliffe in 1935 and was educated at Chadderton Grammar School. At the age of 16 he joined ICI (Dyestuffs Division) in Black- ley, working in the Department of Phar- macy under the guidance of Dan Corrigan in the pre-Pharmaceuticals Division days.After learning the difference between pipettes and measuring cylinders he likes to think that he made rapid progress. “A” levels and the first year of an external London degree course were taken part- time at the Royal Technical College, Salford, where he became a full-time student after spending three formative years with ICI. He remembers the exhila- ration of cycling round The Crescent in Salford and across an almost deserted Piccadilly late on freezing winter even- ings, and his bicycle was a common sight parked outside the Central Library in Manchester . Return to ICI was impossible as National Service was still in operation. He reckons that he would have enjoyed being a soldier, although others doubted this, but having become a Scotophile, even attending a rather improbable Highland Games event in south Manchester, he went to carry out research with Bob Chalmers and Wolf Moser in Aberdeen.In 3 years he managed to get his golf score down from 136 to 108, to obtain a rather shaky working knowledge of Scottish Country Dancing and to marry Joyce Carson, an English/Biblical Studies grad- uate, in addition to gaining a PhD degree. His research was on the reaction between sulphite and nitroprusside, discovered by Boedeker in 1861; he still remembers the look on Wolfs face in the front row at a conference when he included in his intro- duction (against advice) the comment that 1861 was also noted for the outbreak of the American Civil War. His return to England in 1961 was to Loughborough College of Advanced Technology, initially as an Assistant Lec- turer Grade B in Inorganic Chemistry.He and Mike Ellis, the hammer thrower, became the first sub-wardens in the newly built Faraday Hall. Despite his firmly held view at that time that 3 years in any one place was about right he is still at Lough- borough. He transferred to the analytical section, formed on the arrival of Duncan Burns in 1967 after university status had been gained. In the following years the section flourished; research activity increased and the MSc course in Analy- tical Chemistry, and the analytical tech- niques short courses for industry in their present form, were introduced. He was promoted to Senior Lecturer in 1977 and to Reader in 1981.He has published some 124 research papers, 50 of them in The Analyst, and has supervised 23 successful PhD students. He joined the Chemical Society and the SAC in 1967, and since 1970 has served continuously on the Electroanalytical Group committee, being Honorary Treasurer from 1976-82 and Chairman currently, and on the Midlands Region committee for all but 5 years since 1970, being Honorary Secretary from 1972-5. He has also served on the Special Tech- niques Group committee and on the SAC 1977 committee. Dr. Fogg considers that the main advantage of being a university teacher of some years standing, and of being actively involved in analytical chemistry research and the Analytical Division, is that one gets to know so many pleasant people in Britain and world-wide, as well as having the pleasure of meeting many former students again.He enjoys his work, travelling, history, being a member of the local photographic society, his family and supporting his wife’s career. As well as living in Loughborough he also lives in Hounslow, where his wife is the Head Teacher of a large comprehensive school. A biography of Mr. C. A. Johnson appeared in Anal. Proc., 1983, 20, 196, following the conferring on him by the316 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Analytical Division of the eighth Analy- tical Division Distinguished Service Award. Since that time he has received an Honorary Doctorate of Science from the University of Bradford. A biography of Dr. Diana Simpson appeared in Anal. Proc., 1984, 21, 158, when the ninth Analytical Division Dis- tinguished Service Award was conferred on her.Gyula Svehla was born in 1929 and was educated in Hungary, where he obtained a BSc and later a PhD from the Technical University of Budapest. For a while he taught analytical chemistry there. In 1965 he went to Aberdeen University and since 1966 he has been a member of the teaching staff at the Queen’s University of BeIfast, where he is now Reader in Analytical Chemistry. In 1968 he became a Fellow of the Royal Institute of Che- mistry, and in 1973 he was awarded a DSc by Queen’s. He has spent research periods in the Heyrovsky Polarographic Institute in Prague, at the Max Planck Institute in Schwabisch Gmiind and at the van? Hoff Institute in Amsterdam. He has published more than 100 research papers, mainly in the fields of kinetic and electrochemica1 methods of analysis, pub- lished two monographs and revised Vogel’s Qualitative Inorganic Analysis.Since 1972 he has been the Editor of the series “Wilson and Wilson’s Comprehen- sive Analytical Chemistry,’’ of which 18 volumes have been published so far. For 12 years he served on the Editorial Board of Talantu. At present, he is Chairman of the IUPAC Commission on Analytical Nomenclature (V.3.), Chairman of the Northern Ireland Section of the RSC Analytical Division and is a member of the Analytical Abstracts Editorial Board. Professor Alan Townshend was a Lecturer in the Chemistry Department of the University of Birmingham from 1964 to 1980, in what was Professor R. Belcher’s celebrated Analytical Research School.During this time he supervised 36 success- ful PhD students. The achievements of these years are sumrnarised in an earlier biography (Proc. Anal. Div. Chern. Sac., 1975, 12, 40). In 1980 he moved to the University of Hull as Senior Lecturer in Analytical Chemistry in the Chemistry Department. He was promoted to Reader in 1982 and to a Personal Chair in July, 1984. At Hull he has built up a flourishing centre for analytical chemistry, with the active encouragement of his colleagues, and has been ably supported, since Janu- ary 1984, by Dr. Paul Worsfold. They have introduced the unique BSc course in Chemistry with Analytical Chemistry and Toxicology (which now has an optional year in industry), and MSc and Diploma courses in Analytical Science.A large research team (currently about a dozen PhD students) is involved in flow injec- tion analysis, chemi- and biolurni- nescence, enzymatic methods, piezoelec- tric sensors, cool flame and diode array spectrometry, microemulsions, environ- mental probIems and other subjects. Professor Townshend is an editor of Analytica Chimica Acta. He is also a member of the Editorial Boards of Trends in Analytical Chemistry and the Canadian Journal of Spectroscopy, and was on the Boards of The Analyst and Talantu. He is the author of three books and about 200 papers. He was awarded his DSc in 1972 and became an FRSC in 1978. Professor Townshend has long been involved with the RSC and its Analytical Division, and the preceding organiza- tions. He was an elected member of the Council of the CSlRSC (1974-79) and has served on its External Affairs Board and its Post-Experience and Books and Reviews Committees.He has been an elected member of the SAC and AD Councils on several occasions. He was the Division’s Publicity Secretary (1978-82) and has served on most of Council’s Committees. He has been both Chairman (1975-77) and Treasurer of the Midlands Region Committee, and is currently the Vice-chairman of the North East Region Committee and Chairman of the Atomic Spectroscopy Group Committee. He has also been elected to the Committees of the Special Techniques and Education and Training Groups of the AD. He was awarded the SAC Silver Medal in 1975 and was the AD Schools’ Lecturer in 1984-85. He also participates in scientific activi- ties outside the RSC. He has been Secre- tary of Commission V.2 (Microchemistry and Trace Analysis) of IUPAC since 1977, and a member of the British National Committee on Chemistry, Analytical Chemistry Sub-committee, since 1975. He is a member of the Chemistry Board of the CNAA, and recently served on the Chemistry CASE panel of the SERC. He is Vice-chairman of the Ultraviolet Spectrometry Group, and a Trustee of the Midlands Association for Qualitative Analysis. Despite the above, he still finds the odd moment to devote to his wife, three teenage sons, a large garden, the odd burst of sporting activity and the oc- casional glass of wine. A biography of Mr. Colin Watson appeared in Anal. Proc., 1981, 18, 232, when he served his first term on the Analytical Division Council.

ISSN:0144-557X    

DOI:10.1039/AP9852200315

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 The Automated Analytical Laboratory of the Future 317 The following are summaries of two of the papers presented at a Joint Meeting of the Automatic Methods Group and the Industrial Analytical instrumentation Group of the Institute of Measurement and Control held on November 28th, 1984, in the BP Research Centre, Sunbury-on-Thames, Middlesex. An Integrated Chemical Monitoring and Data Acquisition System at Hunterston B Power Station D. E. Collis South of Scotland Electricity Board, Generation Design and Construction Division, C & I Section, Spean Street, Glasgow G44 4BE The original on-line chemical instrumentation on the feed- water and steam at Hunterston B Power Station required extensive refurbishment. It was decided to install a centralised chemical monitoring room and to use a computerised data acquisition system.This paper describes the scheme and experience gained during its installation, commissioning and subsequent use. Introduction By 1980, the installed, plant mounted feed and steam chemical analysis equipment was, with the exception of feedwater conductivity, non-functional. It was therefore decided to initiate a refurbishment programme to bring the installed equipment up to CEGB GOM 72 standard. The Chemical Measuring Instruments Advisory Group (CMIAG) Guidance Document on Chemical Monitoring Rooms was used as the basis for the new installation. An investigation into the costs of updating the chart recorders and alarm presentation systems led to the conclusion that a computerised data acquisition, display and alarm system would prove cost effective, especially as it could include the reactor gas instrumentation (being refurbished at the same time) and hence provide the chemists and plant operators with a coherent chemical analysis data facility.Chemical Monitoring Scheme As stated, very little of the on-line chemical instrumentation was functional and the operational equipment was nearing the end of its useful life. This provided the opportunity to re-equip the Station with new equipment approved to the CMIAG specifications and to take a fresh approach to its location and mounting. Space was at a premium and hence the centralised chemical monitoring rooms were built on a new platform over the water treatment plant control room (Fig. 1).The instru- ments were selected to be suitable for operation in an air-conditioned environment, with front access only being available. The instruments were mounted on modular panels, or bench mounted in front of them, to facilitate complete measurement system testing at works, fast installation on to pre-fitted racks and easy modification in the future (Figs. 2 and 3). Similarly, new secondary sample conditioning racks were located immediately outside the chemical monitoring rooms, again on single-sided modular racks. Primary coolers were located close to the sampling points on free-standing frames. To ensure the longevity of the installation, the steel sections of the building were coated with a protective compound where splashing could be expected, the floors were sloped and coated with a non-slip chemical resistant finish, all drains were enclosed and the rooms were air conditioned to 20 k 2 "C, 4040% relative humidity. The steel panels and framework were all galvanized before painting and plastics were used Unit 8 (R4) Turbine hall Unit 7 (R3) Fig.1. Chemical monitoring room location and layout. 1, CEP disch.; 2, CPPO; 3, BFW; 4, DHL; 5, doors, 6, samples; 7, bench; 8, computer system; 9, H and V; 10, coolers318 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 where possible for junction boxes, drain and sample pipework; all valves and fittings were in- stainless steel. f- Roof/wall Sample isolation & Modular panel on horizontal unistrut Clean & dirty drains Sealed floor with covered drain Fig.2. Chemical monitoring room construction diagram Data Acquisition System The most important criteria for the selection of the computer system were the facility to scan (slowly) analogue inputs, to scan (quickly) digital inputs, to handle local and remote inputs, to compute engineering values, to store data and display the data locally and remotely, with hard copy facilities. In addition, the software had to be proven, capable of easy assimilation by engineers and chemists (with no computer experience) and allow on-line development and the use of complex calculations using CORAL and FORTRAN. The system selected, based on a Ferranti Argus 700GL 1000 930 p q 7 L - 3 5 Earth stud Flexible Signals Power junction junctior box box Flow con- trol Hydrazine monitor trunking Steel panel with strengthening channels on rear.O t Finish-white gloss Fig. 3. Hydrazine monitor panel processor, is shown in outline in Fig. 4. A part standard, part special software system was developed by Ferranti and SSEB programmers , the displays being largely implemented by engineering staff using the PMS software. One system covers the two units. The applications software provides a com- prehensive analogue data acquisition and storage system with recall to VDU of any data obtained in the last 24 h. All data is validated before display. The alarm system allows the separa- tion of alarms into groups specific to each unit and to plant or equipment areas within the units. Regular printouts of chemical analysis data have removed the need for manual logs.In use, whereas the chemist can obtain information on both units at his VDUs, the CCR operator is unaware that the system is handling both units, the data and alarms displayed being specific to his unit or plant area. Semi-graphic printer VDUi keyboard 1 computer \ - I \ \ Alarm CCR-R4 I I \ \ GAS-DAS _ - 1 Gas analysis laboratory I Fig. 4. Chemical data acquisition system Experience The use of modular instrument mounting panels and standard framework greatly reduced the installation and commissioning time scales, as did the close proximity of the sample condition- ing racks and computer system. The improved environment for the instrumentation has reduced maintenance and calibration manpower requirements and the computerised data acquistion system has greatly improved the data flow to the chemists and operators.The scheme has been extended to cover additional parameters and the modular instrument mounting and com- puter software have lived up to their expectations in this respect. The versatility of the installation has been shown by its easy modification, at short notice, to monitor samples not originally connected, but required to assist the operation of the plant during a partial malfunction.ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Power Fluidics for Ventilation Control E. Rimmer and J. McGuigan UKAEA, Risle y, Warrington, Cheshire The situation often occurs, particularly in the nuclear industry but increasingly elsewhere, where provision must be made to contain process or experimental material in an enclosure to prevent contamination of personnel.The separation of the material from the operator can be effected either by a complete physical barrier, as in the instance of a glove-box, or by a current of air moving at a sufficient velocity to prevent back diffusion of contamination, as in a fume-cupboard. As an additional precaution against adventitious leakage the interiors of glove-boxes are normally maintained at a lower pressure than the surroundings. The exhaust air from the glove-box will often have to be treated by some sort of clean-up system (e.g., filters, scrubbers, etc.) and in order to overcome the resistance of the clean-up system the extractor fans must normally operate at a greater depression than that required in the glovebox. This necessity results in the requirement for some kind of pressure control device on the glovebox.Power Fluidics Power fluidics is a technology in which the dynamic properties of the process fluid flow are used to carry out control functions by adjusting flow-rates, pressure differentials and flow direc- tions in devices that do not themselves have any mechanical moving parts. This offers potentially very high reliability, which is important when the presence of contamination makes maintenance difficult. A particularly useful device in the ventilation field is the vortex amplifier. This consists of a very short cylindrical chamber with three connections in axial, tangential and radial configuration (Fig. 1). The main direction of flow is from the radial inlet to the axial outlet.If some constant pressure differential is applied between the inlet and the outlet, then the amount of flow through the device can be varied by injecting varying amounts of control flow into the tangential connection, with an increase in control flow causing a decrease in the main flow. The effect occurs because the tangential input causes the flow in the chamber to swirl, creating a vortex flow pattern with a relatively large pressure drop. The device is described as vortex amplifier because a small change in control flow causes a large change in the main flow. I Main Outlet c- Fig. 1. Configuration of basic vortex amplifier Use of Vortex Amplifier for Glove-box Depression Control The characteristics of the vortex amplifier are such that the resistance of the device depends strongly on the pressure differential between the main inlet and the control inlet.If a 319 test is carried out by maintaining a constant differential pressure between the control inlet and the outlet, and varying the resistance on the inlet of the device, then a pressure-flow characteristic of the device is obtained, as is shown in Fig. 2. 1 Extract Fig. 2. Vortex amplifier inlet characteristic The requirements of the glove-box application relate to two possible operating conditions. In the normal operating condi- tion the glove-box has only a small purge air flow through it, and in this state the box depression must be held at the required value (usually about -3.75 cm water). An alternative operat- ing requirement may occur if a breach appears in the containment (e.g., a lost or torn glove).In this instance it will be necessary to provide some velocity 1 m s-1 in the nuclear industry) through the breach to prevent back diffusion of contamination to the surroundings. In order to meet these criteria a high resistance is required between the glove-box and the extract fan under normal operating conditions, while in the emergency breach condition this resistance must be low to allow the required extract flow. These requirements can be met by installing a vortex amplifier in the extraction line between the glove-box and the fan. The necessary conditions are then produced at points N (normal) and E (emergency) on the amplifier inlet characteristics in Fig.2. The use of a vortex amplifier in this depression control application results in a system with a fast response time and high reliability. Fume Cupboards A vortex amplifier system for controlling fume cupboard exhaust is currently under development.The objective is to maintain the face velocity over the sash opening constant irrespective of sash position in the interests of prevention of back diffusion and reduction of heat loss. Linkage to sash mechanism p - - - - - - - - - - - - - - - - I,-, Extract -c flow cupboard Fume I Vortex amplifier Y air fan Main extract fan Fig. 3. Schematic diagram of fume cupboard extract system320 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 A vortex amplifier is installed (Fig. 3) between the fume cupboard and the extractor fan, with the admission of control air regulated by a mechanical valve linked to the sash mechanism.Because of the extremely small depression in the fume cupboard interior there is insufficient pressure differen- tial to allow control air to be drawn into the vortex amplifier from the surroundings, so a small fan is used to pressurise the control air upstream of the mechanical valve. In this system the main advantage of using the fluidic device is that there are no moving part dampers in the main extract duct, which could possibly be subjected to very corrosive exhaust gases from the fume cupboard. A prototype system serving a single fume cupboard has already been tested, and the extension of the principle to suites of cupboards is now under consideration. Conclusions Fluidic devices, in particular vortex amplifiers, have now been used for active ventilation control in parts of the nuclear industry for several years.Whenever a ventilation system requires very high reliability and fast response they often offer considerable advantages over more conventional equipment. RSC ANALYTICAL DIVISION CHROMATOGRAPHY AND ELECTROPHORESIS GROUP AND MIDLANDS REGION It is hoped to arrange, at Loughborough on June 4,1986, a meeting under the title ”Research Topics in Chromatography,” the purpose of which is to provide an occasion for junior researchers (such as postgraduate students or research assistants) to present accounts of their current projects in academic or industrial laboratories. The submission of papers is invited in any appropriate area of chromatography or electrophoresis.In the first instance a proposed title and short abstract (100 to 200 words) should be supplied to arrive as soon as possible, and in any case not later than December 10,1985. The intention is that as many papers as possible will be presented orally in periods of about 20-min each, thus giving the authors experience in formal presentation. However, no doubt some papers will be more suited to poster presentation and a poster group is also envisaged. Authors submitting abstracts should indicate whether verbal presentation or the poster session would be preferred. It is the intention that the final programme for the meeting will be settled by mid-January, 1986. Correspondence about the meeting should be addressed to Dr. D. Simpson, Analysis for Industry, Factories 2/3, Bosworth House, High Street, Thorpe-le-Soken, Essex C016 OEA.Environmental Chemistry Vol. 3 Senior Reporter H. J. M. Bowen A review Of the literature published UP to the end of 1982. Disposal and Utilization of Sewage Sludge Possible Consequences of Sewage Sludge Disposal and Utilization and the Need for Monitoring Brief Contents: Tropospheric Ozone Ozone Sources in the Unpolluted Troposphere Photochemistry of the Clean Troposphere Ozone Distribution in the Troposphere Sinks of Ozone in the Unpolluted Troposphere Tropospheric Ozone Budget Ozone Formation and Destruction in Polluted Air Elevated Ozone Levels Biological Effects of Ozone Analytical Techniques The Environmental Chemistry of Organotin Compounds Toxicological Patterns of Organotins Analysis of Organotins at Environmental Levels Modes of Entry into the Environment Aqueous Chemistry Transformations in the Environment Degradation of Organotin Compounds Determination of Heavy Metals in Sewage Sludge Analysis of Sewage Sludge Selected Procedures for Sludge Analysis Inorganic Deposits in Invertebrate Tissues Metal Deposits Ligand Binding Silica Deposition Urates Specialist Periodical Report (1984) Hardcover 153pp 0 851 86 775 8 Price f41.00 ($74.00) RSC Members f27.00 RSC Members should send their orders to: The Royal Society of Chemistry, Membership Officer, 30 Russell Square, London WClB 5DT. Non-RSC Members should send their orders to: The Royal Society of Chemistry, Distribution Centre, Blackhorse Road, 1 etchwot-th, Het-ts SG6 1 HN, England. The Royal Society of Chemistry Burling ton House Piccadilly London Wl V OBN

ISSN:0144-557X    

DOI:10.1039/AP9852200317

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 321 Near Infrared Analysis of Foods The following is a summary of one of the papers presented at a meeting of the South East Region held on February 14th, 1985, at the Leatherhead Food Research Association. Near Infrared Analysis in the Fourier Domain with Special Reference to Process Control A. M. C. Davies AFRC, Food Research Institute, Colney Lane, Norwich NR4 7UA W. F. McClure North Carolina State University, Raleigh, North Carolina 27607, USA Near infrared (NIR) spectra are normally recorded at 2-nm intervals over the range 1100-2500 nm and it has long been recognised that one of the problems of applying NIR analysis is the abundance of highly intercorrelated data, McClure and co-workersl.2 demonstrated that transformation to the Fourier Table 1.Proved and potential applications of NIR in the Fourier domain Data reduction 1 Spectral smoothing 1 Predicting composition 2 Correction for particle size effects Reducing calibration time 2 Reducing magnetic storage requirements 2 Simple computation of derivative spectra 2 Providing rapid calibration updating 3 Providing a check of instrument anomalies 3 Reducing multicollinearity 3 Providing a rapid check of instrument noise 3 Calibration transfer by instrument alignment 3 Composition prediction without regression based calibration 3 Discriminant analysis 3 Process control 3 Application Reference 2 domain, in the form of the trigonometric coefficients, concen- trated the useful information into a small number of variables, which could be applied more readily to analytical problems.The advantages and a possible application of FTNIR, which we have brought together in work that will be the subject of a forthcoming paper,3 are summarised in Table 1. The present paper will concentrate on the last two applications and demonstrate that they could form the basis for general methods of process control in the food manufacturing industry. The application of NIR to process control is attractive because information can be obtained very rapidly, without elaborate sample preparation or contact with the sample. However, in practice it has been found very difficult, mainly as a 'result of the problem of calibration. Conventional NIR analysis is based on the use of regression analysis to compute a calibration and this demands a wide range of analysed samples.In manufactured products this wide range is not normally available and manufacturers are unwilling or unable to produce out of specification material. Attempts are made to overcome these defects by preparing samples in the laboratory or by transferring laboratory calibrations. By using Fourier trans- formed NIR data is it possible to utilise methods which are not based on regression analysis and only require samples that are normally produced by the process. The two methods we have been investigating can both be demonstrated by using the same set of instant coffee samples, which were obtained as retail purchases and were either normal or decaffeinated instant coffee. NIR spectra of these samples were measured on a Pacific Scientific Mark I 6350 Research Composition Analyzer without any sample preparation and then transferred to a DEC VAX computer for transformation to the Fourier domain and subsequent processing of the transformed data.It has been shown previouslyl.2 that while almost all of the useful information is contained in the first 50 pairs of Fourier coefficients, six pairs of coefficients may contain sufficient 15 v) C Decaffeinated - 60 - 30 0 30 60 -60 - 30 0 30 60 0 Discriminant score $ 10 .- b n $ 5 Fig. 1. Discriminant analysis of coffee samples in the wavelength (a) and Fourier (b) domains322 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 a coefficients b coefficients 0 4 8 12 16 20 24 0 4 8 12 16 20 21 Coefficient number Fig. 2. Fourier coefficients for normal instant coffee and decaffeinated instant coffee.Successive sets of coefficients have been slightly offset to the right. Each coefficient number shows a range of values for normal coffee followed by a range of values for decaffeinated coffee. Some clear differences can be observed between the groups for some coefficient numbers information for composition predictions. In this work, 25 pairs have been utilised because of statistical limits arising from the number of samples and computational requirements in the GENSTAT programs that were used for data analysis. The results of discriminant analysis are shown in Fig. 1, which compares data from both Fourier and wavelength domains. The wavelength data were reduced to the same number of variables by averaging every 14 data points.In order to demonstrate that this was not a chance result the samples were shuffled to produce two random groups, which could not be separated even in the Fourier domain. While separation in the wavelength domain was adequate, the improved separation in the Fourier domain suggests that it would be successful with more demanding identification problems. The process control method that is under development is based on using the process to define its normal variability in a multi-dimensioned Fourier space and then testing sample spectra to determine whether they are inside or outside this envelope. In Fig. 2, the FTNIR spectra have been normalised to the instant coffee and large variations are apparent, particularly in some of the A coefficients, for the decaffeinated samples. In order to investigate which Fourier coefficients were most important and also to produce a more readily comprehensible system, principal component analysis was applied to these results.Fig. 3 is a plot of the first and second principal components and demonstrates the rejection of all but two of the decaffeinated samples. These were rejected by the third principal component. This result demonstrates that Fourier data could be used to detect departures from the normal variability of a production line. In Fig. 3 it can be seen that all of the decaffeinated samples appear in the same quarter of the principle component plot. This suggests that there is a “decaffeination” vector and that other differences would give rise to different vectors.Hence, having detected an error condition it would then be possible to carry out further data analysis to identify the nature of the error. Conclusion There is still much work to be carried out on the development of these ideas but we are encouraged by the present results. In order to be able to apply this work to process control it will be necessary to develop on-line instruments which can produce FTNIR data. This could be done either by using an inter- ferometer or by application of fast Fourier transformation of wavelength data. 0 - 15 - - 30 I 1 I 1 I I - 30 -15 0 15 30 1st component Fig. 3. Plot of first and second principal components for process control applied to normal and decaffeinated coffee. A, Control samples; a, de-caffeinated samples; broken line, rejected samples The authors are grateful to Susan Ring for writing the Fourier program, Jeremy Franklin for statistical assistance and advice, Alex Grant and Simon Gonzales for NIR analysis and Annette Baker (Campden FPRA) for sample collection. References 1. 2. 3. Giesbrecht, F. G., McClure, W. F., and Hamid, A., Appl. Spectrosc., 1981, 35,210. McClure, W. F., Hamin A., Giesbrecht, F. G., and Weeks, W. W., Appl. Spectrosc., 1984,38, 322. McClure, W. F., Davies, A. M. C., and Geisbrecht, F. G., in preparation.

ISSN:0144-557X    

DOI:10.1039/AP9852200321

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 323 Computer Aided Methods in Chromatography The following are summaries of six of the papers presented at a Joint Meeting of the Chromatography and Electrophoresis Group and the Scottish Region held on March 19th-20th, 1985, in the University of Edinburgh. Automated Method Development, J. C. Berridge Pfizer Central Research, Sandwich, Kent CT13 9NJ Chromatography is literally defined as “colour writing.” Could Michael Tswett have foreseen the advances that have developed the principle he first demonstrated at the turn of the century into the high technology technique that so many of us take for granted today? I suggest that, while he may have‘speculated upon the possibilities, he could not have even dreamt of the gleaming, microprocessor controlled chromatographs that are now the standard armoury of practically every analytical laboratory.Whilst we have seen tremendous strides made in the technologies of both gas and liquid chromatography, it must surely be to high-performance liquid chromatography (HPLC) that we should turn to see the embodiment of an analytical technique par excellence. In just two decades HPLC has developed into a highly sophisticated instrumental tech- nique offering speed, sensitivity and selectivity. Arguably representing a pinnacle of separation science, HPLC is now firmly established as a reliable and versatile analytical tool for the separation and quantitation of the mixtures soluble in liquid phases. The scientific literature abounds with examples from all avenues of analysis.There is an ever increasing stream of publications describing new applications and new separa- tions but how were these separations developed? How many of the published separations represent optimum separation condi- tions? Indeed, were the separations really developed at all, or did serendipity play a major role in solving the problem? Separation development in HPLC remains a most difficult and frustrating task. It is first necessary to decide upon the chromatographic mode that will be used, for example reversed- phase, normal phase or perhaps size exclusion, etc. This fundamental decision then dictates the variables that can be considered in the second step, the optimisation of the separation. With the advent of low cost computing, the automated optimisation of HPLC separations is a practical reality through a number of methods, which will be discussed in outline.Automation of the first step, that fundamental decision, is rapidly approaching through the development of “Expert Systems. ’’ Optimisation, the Basic Problems Almost any aspect of separation is amenable to optimisation. For many years attention was focused upon the design of the column, the support it contained and the mobile phase flow dynamics.’-6 This attention mainly resulted from the fact that large particles (i.e., larger than 10 pm) were in use and much still had to be learned about the mechanics of chromatography under high pressure. Today, most of the basic mechanical problems have been overcome, although there is naturally much interest in improving the performance of the chromato- graphic h a r d ~ a r e .~ - l ~ More is known about the influence of solvent composition on retention, selectivity and peak shape and the emphasis has moved towards the optimisation of the mobile phase composition to achieve the desired separation. Is Fact or Fantasy? it possible to define what that separation should be? Many definitions of the goal of separation optimisation have been offered, and include13J4: to achieve the separation goal with the minimum of time and effort; to produce the “best” separation possible of a given sample; to achieve the optimum combination of speed of elution, sample size and resolution of solutes; and to select the mobile phase and column combina- tion that gives base-line separation in a given time.None of these definitions represents the total answer. What is certain is that the ultimate achievement will reflect both the stationary phase and the mobile phase selection. There is still the problem, however, of recognising that the separation goal has been reached. The magnitude of this problem is much diminished if the identities of all components and interferents are known. In fact, there can be no ultimate solution of the optimisation problem if this not the case. For automated separation optimisation it is convenient in many situations to use a mathematical function to describe the quality of a separation in some semi-quantitative manner. There are a large number of such functions, called Chromato- grahy Response Functions or Chromatography Optimisation functions. A selection of these, and the applications to which they have been put, is presented as Table 1.An essential prerequisite of automated separation optimisa- tion is that suitable chromatographic hardware is available. This is rapidly ceasing to be the major stumbling block it was as more and more manufacturers introduce microcomputer con- trolled chromatographs, some with optimisation software provided. Be sure that, if you intend to purchase a system with the idea of using it for automated optimisation, the computer controller has full communication with the autosampler, pumping system and any data processing unit (e.g., integrator). It is much easier to ignore a capability provided by the manufacturer than to try to create a vital missing link.Finally, the optimisation scheme to be used must be chosen. There is no shortage of methods from which to choose. There are simple, almost ignorant, approaches, statistical experimen- tal designs, chemometric methods and a variety of schemes which combine elements of all three. All have their uses in particular situations. None emerges as the clear leader as each has its own advantages and disadvantages and some are easier to implement for automated applications than others. It is interesting to compare the various approaches in turn. Statistical Experimental Designs The first steps towards the goal of automated separation optimisation were taken with the introduction, in the late 1970s, of commercial instruments which could undertake a series of analyses using pre-programmed mobile phase compo- ~itions.23~24 Early users probably did not realise the full capabilities of these chromatographs but a more formal approach to the exploration of mobile phases, given the name324 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Table 1.Chromatographic quality evaluation functions in HPLC Quality function CRF = 7 ln(PjIP,) + a(t,-ti) I 1 Experimental variables Gradient parameters and flow-rate Concentration of organic modifier, pH and buffer CRF = Ri+ La- bl TA- Ti) -c( To- TI) Composition of ternary temperature I mobile phase, pH, flow, r = ll R,l[F R,(i-Z)]t-‘ I 1 CRF = lnCfi/gi)-lOO(M-N) COF = ?ai ln(RiIRd) + b(t,-tj) I I s = ( f b - t a ) / ( f a + f b ) a < b Composition of ternary and quaternary mobile phase Concentration of organic modifier and pH Composition of quaternary mobile phase PH Optimisation Simplex search 15 method Reference Simplex search 16 Simplex search 17,18 Mixture design 19 Simplex search 20 Simplex design, 21 empirical model Window diagram 22 Sequential Isocratic Step chromatography, was described by Berry.25 Particularly if the sample consists of basic drugs, it is possible to use a standardised separation strategy with just one column and two mobile phases to achieve a wide variety of selectivities for both normal and reversed-phase chromato- graphy.2630 More recently, this approach has been shown to be useful for acidic and neutral compounds ,31but now six mobile phase constituents are required.Even so, a very simple experimental design can be used to explore mobile phase compositions. It is likely, however, that a more detailed exploration will be required.All analysts surely appreciate the rational classification of solvent properties developed by Snyder .323 This classification has been translated into two “solvent selectivity triangles,” one for normal phase and one for reversed-phase chromatography. By using a simplex statistical design,21,34,35 seven to ten36 experiments conducted over a triangular region provide sufficient data to be able to describe the chromatographic behaviour of all the solutes over this area. By plotting the resolutions between each set of peak pairs it is possible to build up an “overlapping resolution map” (ORM), which will define the mobile phase composition(s) that can provide an optimum separation.Many literature examples attest to the value of this method (e.g., references 36-39) and it has appeared as the method used in a commercial instrument.35 One of the disavantages of the ORM approach is that individual solutes need to be identified in successive chromatograms. Of course, peak identification is not ncessary if a chromatographic response function is used. By combining sound statistical design, a chromatographic response function and an iterative component in an optimisa- tion scheme a relatively simple way of optimising reversed- phase separations is produced. 19340 Two experiments only are conducted, with different modifiers in proportions expected to give approximately equal analysis times. It is then assumed that the retention times of solutes will be a linear function of the composition of blends of the binary solvents, an assumption which permits an estimate of an optimum compostion to be made.By conducting a separation at the predicted composi- tion, either the optimum will be revealed or the data generated can be added to the knowledge gained so far to revise the prediction of solute behaviour. By repeating this process the approximate model is gradually made more accurate and the optimum conditions located. It has recently been shown that this approach can be extended to both pH optimisation and the optimisation of ion-chromatographic separations.41.42 Chemometric Methods Chemometrics43 is the application of mathematical, statistical and computational techniques to the design and analysis of experiments and the data they produce. General optimisation techniques fall into this classification and one method which has found particular use in chromatography is entitled “win- dow diagrams.”a The first applications of window diagrams to chromatography were for the selection of stationary phases in gas chromatography.45 Being a general method, window diagrams can also be applied to HPLC.Early examples concentrated on a single factor, such as the mobile phase pH,46747 but the ideas have now been successfully extended to embrace multi-factor optimisations.4*.49 The selection of mobile phase pH was used recently as a model for the use of window diagrams in the automated optimisation of HPLC ~eparations.5~ Window diagrams have the advantage that they reveal the conditions that should give the separation with the best possible resolution between the worst separated peak pair, the “global” optimum.Other optimum regions, “local” optima, are also indicated and the chromatographer is free to choose one of these should a secondary criterion (such as cost, analysis time, etc.) need to be considered. The price of achieving this optimum is that all peaks must be identified in each chromatogram and some mathematical expression should be found to describe the retention behaviour of the peaks throughout the experimental space being investigated. There are alternative chemometric methods for searching for an optimum separation, which do not require peaks to be identified in each chromatogram.One way of investigating experimental variables in HPLC is systematically to explore the whole of the available factor space. If this is done without any feedback to the system controller51 the search may be very thorough but will probably be very time consuming. By introducing a degree of feedback into a controlling algorithm it is possible that the number of experiments can be considerably reduced, albeit at the risk of locating a local optimum rather than the global optimum. Such an approach is commercially available52 for both isocratic53 and gradient54 optimisation. One disadvantage of the approach used in this system is that it considers essentially one variable at a time. One of the fundamental things that chemometrics is doing to analytical chemistry is banishing the myth that variables should be varied only one at a time.26 The Sequential Simplex Procedure55.56 is a general multi- factor optimisation method in which all factors are varied as the experimental design seeks to progress sequentially towards the optimum region.The advantages of using the simplex proce- dure are that the calculations are relatively simple, the rules of the procedure are precisely defined and the procedure is ideally suited to automated optimisation.57 For such an attractive method there inevitably must be some drawbacks and these stem mainly from the need to use a chromatographic response function to evaluate the qualities of the separations achievedANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 325 and to guide the procedure towards the optimum separation. From Table 1 it is evident that there is no one function that is the best and the choice of function still rests with the preferences of the chromatographer.58 Early applications of the simplex procedure to chromatographic optimisation were restricted, by the hardware available, to manual applica- tions.15~6,59760 Now that microcomputer controlled chromato- grahs are more widely available it has been shown that the simplex procedures can be used to guide the automated optimisation of normal-phase separations61 and a wide variety of parameters in reversed-phase chromatography.17J8,*0@ Towards Automated Methods Development Much has been achieved in the field of separation optimisation, both for the chromatographic hardware and for the selection of the mobile phase once the choice of chromatographic mode has been made and the column selected.For total automation of method development, the initial decisions will have to made and explored under computer control. Already the first steps in this direction are being made with the combination of the ORM approach and an analogous investigation of columns which encompass most of the selectivity possibilities in reversed-phase chroma tography.63 Perhaps the most exciting developments we can expect to see will be in the field of “Expert Systems” where encouraging progress is already being reported. 64 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. References Snyder, L. R., J. Chromatogr. Sci., 1972, 10,200. Snyder, L. R., J. Chromatogr.Sci., 1972, 10,369. Knox, J. H., J. Chromatogr. Sci., 1977, 15, 352. Martin, M. , Eon, C., and Guiochon, G . , J. Chromatogr. , 1974, 99 , 357. Snyder, L. R., J. Chromatogr. Sci., 1977, 15,441. Guiochon, G., “Optimisation in Liquid Chromatography,” in Horvath, C., Editor, “High Performance Liquid Chromato- graphy, Advances and Perspectives,” Volume 2, Academic Press, New York, 1980. Parris, N. A., “Instrumental Liquid Chromatography,” Second Edition, (Journal of Chromatography Library, Vol. 27), Elsevier, Amsterdam, 1984. Snyder, L. R., and Kirkland, J. J., “Introduction to Modern Liquid Chromatography,” Second Edition, Wiley, New York, 1979. Kaiser, R. E., and Oelrich, E., “Optimisation in HPLC,” Verlag, Heidelberg, 1981. Kucera, P. , Editor, “Microcolumn High Performance Liquid Chromatography,” Elsevier, Amsterdam, 1983.Rosset, R., Caude, M., Desbarres, J., and Schmidt, E., Analusis, 1980, 8, 213. Halasz, I., and Gorlitz, G . , Angew. Chem. Int. Ed. Engl., 1982, 21, 50. Forum on Optimisation Methods, “8th International Sympo- sium on Column Liquid Chromatography, New York, 1984. Hearn, M. T. W., “HPLC of Peptides,” in Horvath, C., Editor, “High Performance Liquid Chromatography, Advances and Perspectives,” Volume 3, Academic Press, New York, 1983. Watson, M. W., and Carr, P. W. Anal. Chem., 1979,51, 1835. Wegscheider, W., Lankmayr, E. P., and Budna, K. W., Chromatographia, 1982, 15, 498. Berridge, J. C., J. Chromatogr., 1982, 244, 1. Berridge, J. C., Analyst, 1984, 109, 326. Drouen, A. C. J. H., Billiet, H.A. H., Schoenmakers, P. J., and De Galan, L., Chromatographia, 1982, 16, 48. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. Nickel, J. H., and Deming, S. N., Liq. Chromatogr. Mag., 1983, 1, 414. Glajch, J. L., Kirkland, J. J. , Squire, K. M. , and Minor, J. M. , J. Chromatogr., 1980, 199, 57. Jones, P., and Wellington, C. A., J. Chromatogr., 1981, 213, 357. Sybrandt, L., and Montoya, E. , Am. Lab., 1977,9(8), 79. Karasek, F. W., Res. Dev., 1977, 28 (6), 38. Berry, V. V., J. Chromatogr., 1980, 199,219. Massart, D. L., and Hoogewijs, G., Pure Appl. Chem. , 1983, 55, 1861. Detaevernier, M. R., Hoogewijs, G., and Massart, D. L., J. Pharm.Biomed. Anal., 1983, 1, 331. Hoogewijs, G., and Massart, D. L., J. Pharm. Biomed. Anal., 1983, 1,321. Hoogewijs, G., and Massart, D. L., J. Liq. Chrom., 1983,6, 2521. Hoogewijs, G., and Massart, D. L., J. Pharm. Belg., 1983,38, 76. de Smet, M., Hoogewijs, G., Puttemans, M., and Massart, D. L., Anal. Chem., 1984, 56,2662. Snyder, L. R., J. Chrom. Sci., 1978, 16,223. Snyder, L. R., J. Chromatogr. , 1974,92,223. Snee, R. D., Chemtech., 1979, 9,702. Leher, R., Ind. Res. Dev., 1983, 25, 116. Issaq, H. J., and McNitt, K. L., J. Liq. Chromatogr., 1982,5, 1771. D’Agostino, G., Mitchell, F., Castagnetta, L., and O’Hare, M. J., J. Chromatogr., 1984, 305, 13. Haky, J. E., Young, A. M., Domonkos, E. A., and Leeds, R. L., J. Liq. Chrom., 1984,7,2127. Antle, P. E., Chromatographia, 1982, 15, 277.Schoenmakers, P. J., Drouen, A. C. J. H., Billiet, H. A. H . , and De Galan, L., Chromatographia, 1982, 15, 688. Haddad, P. R., Drouen, A. C. J. H. , Billiet, H. A. H., and De Galan, L., J. Chrornatogr., 1983, 282,71. Haddad, P. R., and Cowie, C. E., J. Chromatogr., 1984,303, 321. Betteridge, D. , Lab. Pract. , October 1983, 13. Laub, R. J., Int. Lab., May 1981, 16. Laub, R. J., and Purnell, J. H., J. Chromatogr., 1975,112,71. Deming, S. N., and Turoff, M. L. H., Anal. Chem., 1978,50, 546. Price, W. P., Edens, R., Hendrix, D. L., and Deming, S. N., Anal. Biochem., 1979, 93, 223. Sachok, B., Kong, R. C., Deming, S . N., J. Chromatogr., 1979, 199, 317. Weyland, J. W., Bruins, C. H. P., and Doornbos, D. A., J. Chromatogr. Sci., 1984, 22, 31. Nickel, J.H., and Deming, S. N., Am. Lab., 1984, 16(4), 69. PESOS: Perkin Elmer Solvent Optimisation System. OPTIM 1 system from Spectra-Physics. Bradley, M. P. T., and Gillen, D., Spectra-Physics Chromat- ogr. Rev., 1983, 10,2. Bradley, M. P. T., and Gillen, D., Spectra-Physics Chromat- ogr. Rev., 1984, 11, 10. Deming, S. N., and Morgan, S. L., Anal. Chem., 1973, 45, 27814. Deming, S. N., and Parker, L. R., Crit. Rev. Anal. Chem., 1978, 8, 187. Spendley, W., Hext, G. R., and Himsworth, F. R., Techno- metrics, 1962, 4, 441. Debets, H. J. G., Bajema, B. L., and Doornbos, D. A., Anal. Chim. Acta, 1983, 151, 131. Svoboda, V., J. Chromatogr., 1980, 201, 241. Fast, D. M., Culbreth, P. H., and Sarnpson, E. J., Clin. Chem., 1982,28,444. Berridge, J. C. , Chromatographia, 1982, 16, 172.Berridge, J. C., J. Chromatogr., 1984, 316,69. Glajch, J. L., and Kirkland, J. J., Anal. Chern., 1983, 55, 319A. Karnicky, J., Anal. Chem., 1984, 56, 1312A.326 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Signal Acquisition and Processing in the Analytical Laboratory C. J. Adie Edinburgh Regional Computing Centre, University of Edinburgh, The King‘s Buildings, Mayfield Road, Edinburgh EH9 3JZ Data Acquisition Systems There are many commercial offerings in the field of com- puterised signal acquisition and processing in the laboratory environment. For many purposes (particularly in routine “production” situations) such systems may be perfectly satis- factory. However, where the required data analysis cannot be carried out by using the supplied software, or where funds are limited, there is often a strong case for implementing a “home brew” data acquisition and analysis system using a microcom- puter.This article describes some aspects of designing such a system. The signal generated by a detector (such as a photodiode) must pass through several transformations before it can appear as a series of numbers in a computer memory. The analog signal from the detector must first be “conditioned,” which typically involves filtering, impedance changing and amplifica- tion. These operations are usually carried out within the analytical apparatus itself, and the conditioned signal is often available externally to drive a chart recorder or other device. This output can be used by the data acquisition system, but if it is a 0-10 mV chart recorder output, further amplification (up to say 0-1 V) will be necessary. The analog to digital converter (ADC) is the next element in the system.The conditioned analog output from the analytical apparatus is converted to (typically) 8, 12, 14 or 16 digital signals (bits). The number of bits that the ADC generates determines the resolution of the conversion process. An 8-bit ADC has a resolution of one part in 256, a 12-bit ADC one part in 4096; 12 bits are almost always more than sufficient for applications in the analytical laboratory. The ADC can be interfaced to the computer through a “versatile interface adaptor” (VIA). This is a chip which converts the digital signals from the ADC to a form compatible with the computer’s internal digital structure.It also contains the timers which regulate the process of sampling the analog signal at regular intervals, and it is the VIA with which the computer software communicates directly. Sample Rate The computer causes the ADC to sample the input signal at discrete time intervals. Choosing the appropriate sample rate for a particular type of input signal, having regard to the memory limitations of the microcomputer being used, is a complex problem. Looking at the frequency spectrum of the input signal, the rule is that there should be no frequency present which is greater than half the sample rate. The sample rate, the ADC resolution and the available computer memory limit the time for which continuous acquisi- tion can be sustained through the equation Available memory (bytes) N x sample rate (samples s-1) Max.acquisition time (s) = where it is assumed that every sample occupies N bytes. For an 8-bit ADC, N = 1, while for a 12-bit ADC, N = 1.5 (packing 2 samples into 3 bytes) or N=2 (using one two-byte integer for every sample). On-line versus Off-line Analysis In many commercial systems (such as chromatography integra- tors), the data is analysed as it is being acquired. The alternative is to save the data in a floppy disk file after acquisition for later analysis. There are arguments for and against both techniques. On-line real time analysis requires complex programming, and the software overhead involved usually means that the maximum sample rate that can be achieved is lower. Moreover, no matter how good the feature recognition software, auto- matic analysis is always limited in flexibility, and as the data is not stored, later manual analysis of aberrant features is impossible.On the other hand, because only a few samples are stored in the memory at one time, the total length of a run is not limited by memory size. Analysis and reporting are carried out together, so there is relatively little opportunity for human error, making the technique suitable for routine “production” situations. Off-line analysis is much simpler to implement, and because all of the data from a run is available, better algorithms for feature recognition and analysis can be used. There is little software overhead during acquisition, and faster sample rates can therefore be achieved.Manual interaction with the data where automatic analysis fails or is unsuitable is always possible. The main limitations of this method are that the separate analysis phase may become time consuming, and that the number of samples is usually limited by available memory. Digital Filtering Once we have the data in the computer, there are many techniques that can be applied to improve or reduce the data. One such technique is digital filtering, which, although a substantial field of research in its own right, is relatively unknown in some areas of applied science. Commonly, the digitised signal will contain noise as well as useful information, and by designing an appropriate digital filter we can smooth out the noise substantially. If we are interested in the rate of change of the input signal, a digital filter to differentiate it may be designed.A (linear) digital filter acts on a set of input sample points x(n) to produce an output series y ( n ) where n = 0, 1, 2, . . . The filter equation is N M k = - N k = l y ( n ) = c c(k) x(n + k ) + 2 d(k)y(n - k) . . (1) where the c(k) and d(k) are constant coefficients. If at least one d(k) is non-zero, the filter is said to be recursive. Such filters have their advantages, but they are difficult to design and may introduce instabilities. Henceforth we will consider only non-recursive filters where all the d(k) are zero. The filter coefficients c(k) are chosen to give the filter a suitable “transfer function” H ( n , which is the amplitude of the output signal when the input is a sine wave of frequencyf.For instance, a low-pass filter has H(n 1 at low frequencies and H(f) 0 at higher frequencies. The author has developed programs on a BBC microcomputer which calculate and display c(k) and H(n for both smoothing filters [c(k)=c(-k)] and differentiating filters [c(k)= -c( - k ) ] . Note that because a non-recursive filter uses “future’ points x(n + k) with k>O to calculate y ( n ) , it will introduce a delay of N + 1 sample points between the input and output when used in an on-line analysis situation. Reference 1 is a useful introductory text-book on the subject. Peak Detection Automatic peak detection algorithms are less easy to write than might at first be supposed. Problems of noise immunity andANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 327 base-line aberrations are important, and the associated requirements of area integration and base-line correction complicate the situation further.A particularly useful technique for understanding and developing peak detection algorithms is that of the “state diagram.”2 The algorithm has a finite number of “states,” represented as nodes in a graph (see Fig. 1). Permitted transitions between states are represented as arcs, and each arc is labelled by the condition which, if satisified, causes the program to make the corresponding transition between states. The program maintains a state variable, the value of which denotes the current state. A simple three-state algorithm could, for instance, be implemented in PASCAL as shown in the listing. The procedures “start ,” “upslope” and “downslope” handle the details of examining the samples, checking for the conditions for a state transition, and re-assigning the state variable to the appropriate values before returning.As an example, the algorithm3 used for peak detection in the Hewlett-Packard 3390 integrator can be expressed in the form of a state diagram, as in Fig 1. There is no space here to list all the conditions A-H which cause transitions, but as an example, condition A is that the first and second derivatives at three successive sample points are all positive, indicating the start of a peak. Note that the algorithm recognises solvent peaks, and correctly handles peaks on solvent tails. n Solvent / W l / Normal \ Normal downslope F Fig. 1. State diagrams of peak detection algorithm in the HP 3390 reporting integrator A Spectrophotometer System In order to illustrate some of the considerations described above, a system developed by the author to acquire and analyse data from a spectrophotometer on an IBM Personal Computer will be described.The signal from the spectrophotometer’s chart recorder output (0-1.6 V) is digitised by a 12-bit ADC connected to a U-Microcomputers Science Card which is plugged into an expansion bus socket in the IBM PC. Software written in compiled BASIC acquires data from the ADC, and timing is controlled by two cascaded timers in a VIA on the Science Card. The data is acquired into an integer array in memory and written to disk for later analysis, but a chart recorder like display is maintained on the screen during acquisition.The maximum run duration and number of sample points (up to a maximum limit) are determined by the operator at the start of the run, and he is informed of the sample rate which will be used. Typical figures are 20 min run time, 2000 sample points, and one sample every 0.6 s. Analysis involves calculating the derivative of the signal, which is a decay curve obeying the equation This is carried out with a non-recursive differentiating filter containing 31 terms [N = 15 in equation (l)], which also removes high frequency noise. Having obtained dddt, two alternative methods of determining V and K are possible: the first simply linearises (2) by plotting the inverse of the derivative against llx and performing a linear regression analysis. The second method avoids the inaccuracies intro- duced by linearisation by using the simplex method of least-square analysis.4 It is considerably more accurate, but takes longer.Conclusions The design and implementation of a data acquisition and analysis system is a task which requires techniques drawn from a range of fields such as electronics, electrical engineering, physics, mathematics and computer science. Some of these techniques are mentioned above; none of them is new, but hopefully at least some of them are new to readers of this journal. References duldt = V/(x + K ) . . . . ‘ - (2) 1. 2. Hamming, R. W., “Digital Filters,” Prentice Hall, Englewood Cliffs, NJ, USA, 1983. Aho, A. V., Hopcroft, J. E., and Ullman, J. D., “The Design and Analysis of Computer Algorithms,” Addison Wesley, London, 1984.3. Hewlett-Packard 3390 Reporting Integrator Manual, Appendix K. 4. Caceci, M. S., and Cacheris, W., Byte, 1984 (May), 340. APPENDIX A Possible Program PROGRAM peakdetect (input, output); (* This program skeleton illustrates how to implement a simple VAR state: (start, upslope, downslope); PROCEDURE startstate(. . .); three-state peak detection algorithm in Pascal. *) . . . . . . BEGIN (* This procedure might for instance get the next sample, work out the first derivative, and depending on its value either repeat the process, or exit after setting the value of the global variable “state” appropriately. *) END; (* of startstate *) PROCEDURE upslopestate (. . .); . . . BEGIN (* Similarly, this routine might get the next sample, calculate the first derivative, and exit with “state” set to “downslope” only if the derivative becomes sufficiently negative.*) . . . END; (* of upslopestate *) PROCEDURE downslopestate (. . .); . . . BEGIN (* This routine might get the next sample, calculate the first derivative, and exit with “state” set to “upslope” only if the derivative becomes sufficiently positive. *)328 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 END; (* of downslopestate *) BEGIN (* start of main program block *) ' . . state: = start; WHILE true DO BEGIN CASE state OF start upslope downslope : downslopestate (. . .); END; (* of case *) : startstate (. . .); : upslope (. . .); END; (* of while *) END. An Efficient iterative Procedure for N RPLC Involving Photodiode Array De obile Phase Optimisation in :ection L.de Galan Laboratory for Analytical Chemistry, Jaffalaan 9, 2628 BX Delft, The Netherlands As witnessed by the recent appearance of four commercial systems, the computer aided optimisation of mobile-phase conditions in reversed-phase liquid chromatography (RPLC) is fashionable. This is not surprising. First, RPLC is being applied to increasingly complex samples that are difficult to separate. Secondly, small variations in the mobile phase composition can produce significant improvements, so that optimisation is profitable. And finally, RPLC is fairly robust, so that it lends itself to computer control. The first step in qn optimisation strategy is the selection of the number and the range of the mobile phase conditions considered.Because such parameters as temperature and flow-rate are much less effective in producing solute shifts, the attention should be concentrated on the composition of the mobile phase, i.e., the concentration of organic modifiers, ion-pairing reagents, buffer pH, etc. The boundaries are usually chosen so as to keep the total analysis time approxi- mately constant throughout the procedure. The second step is the choice of a criterion that expresses the goal of the optimisation or the quality of a chromatogram in a single, quantitative value. Although many optimisation criteria have been proposed in the literature, they are largely similar in that they all aim at the separation of as many solutes as possible. A typical and useful criterion is the resolution product where N is the constant plate count of the column and q, ti and ki are the peak width, the retention time and the capacity factor of solute i, respectively.It is important to note that the multiplication extends over all solutes, rather than over the discernible peaks in the chromatogram. Indeed, the determi- nation of this number in the initially poorly resolved chromato- gram of an unknown sample is a key issue, which will be addressed below. Now the optimisation procedure proper will vary the mobile phase composition systematically, until the highest value for the criterion has been found. This results in the best possible chromatogram within the boundaries set by the operator. Current commercial procedures derive the criterion directly from the discernible peaks in the chromatogram by measuring or estimating peak retentions and widths.The major advan- tages are that the procedure is insensitive to peak reversals, that the nature of the solutes need not be known and that the procedure is easily programmed to run under computer control. This last aspect is important, because the procedures invariably require many test runs, ranging from 20 for self-searching Simplex routines1.2 to as many as 60 for lattice designs.3 An inherent weakness is that the number of solutes remains unknown, so that the intended separation of all solutes is sometimes not realised, not through chromatographic limitations, but by fault of the procedure. The Iterative Regression Design The major reason for the many runs needed in commercial procedures is that the optimisation criterion usually varies highly erratically with a change of mobile phase composition. By contrast, the retention times of the individual solutes vary much more regularly.In fact, the logarithmic capacity factor (In k j ) frequently varies almost linearly. Consequently, when all solute capacity factors are known at the extreme ends of the parameter range, they are easily calculated over the entire range and, hence, the optimisation criterion can be predicted indirectly through equation (1). This forms the basis of the iterative regression procedure developed in the author's laboratory.4-5 bi 3 T 0 200 400 600 Time/s Fig. 1. Chromatograms of: 1, phenol; 2, benzaldehyde; 3, o-cresol; 4, m-dinitrobenzene; 5 , benzene; and 6, p-iodophenol in (a), aceto- nitrile - water and (b) tetrahydrofuran - water As an example, the separation of six aromatic solutes is shown in Fig.1. When run in either tetrahydrofuran - water or acetonitrile - water the separation is incomplete, although the presence of six solutes can be observed. In the phase selection diagram the logarithmic capacity factors are connected by straight lines so that their values and, hence, the resolution product can be calculated after equation (1) over the full range of ternary mixtures composed of acetonitrile tetrahydrofuran and water. An optimum composition of 29% acetonitrile and 14% tetrahydrofuran is predicted, but when this mixture is tried, the result is disappointing (Fig. 2). However, the newANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 329 data obtained for the capacity factors at this mobile phase composition can be entered into the phase selection diagram and a new optimum composition is predicted, which indeed provides an excellent separation (Fig.3). In three years of practical experience with the procedure the following advantages have emerged. 0 $-TH F -t 0.37 1 29% ACN, 14% THF, 57% H20 2 0 200 400 600 Time/s Solute Recognition by Photodiode Array Detection Obviously, a key issue in the iterative regression procedure is the identification of corresponding solutes in successive chro- matograms. In the example of Fig. 1 this was accomplished by separate injection of the solutes, which is only possible when they are known and available to the operator. Before we discuss an approach for unknown samples, we should recall that the need to identify at least the number of solutes in the sample also emerges in other optimisation procedures.More- 0 @-TH F + 0.37 36% ACN, 8% THF,56%H20 Fig. 2. Phase selection diagram constructed from the retention data in Fig. 1 and the chromatogram run at the predicted optimum composi- tion. Note the coincidence of solutes 5 and 6, contrary to expectation 0 200 400 600 Tirne/s 1. The procedure is very efficient. In the present instance the optimum is reached in four runs. From our experience with many samples we can state that the number is always less than ten. 2. By its iterative nature the procedure is self correcting. An incorrect prediction is restored by entering new data into the procedure.In this way, the true, non-linear retention behaviour of the solutes is approximated with increasing accuracy by a series of linear segments. A complete picture over the parameter range is maintained. 3. The procedure is versatile in that it allows the use of various optimisation criteria. It can also be applied to several mobile-phase parameters, such as organic modifier con- tent ,475 pH6 and ion-pairing reagent concentration.’ 4. Recently, the procedure has been extended from the one-parameter optimisation illustrated in Figs. 1 and 2 to the simultaneous optimisation of two parameters.6 In this instance the straight lines connecting two sets of data points in a one-dimensional space are replaced by triangles determined by three data points in a two-dimensional space.Through iteration the true undulating retention surface of a solute is now approximated by a series of planar triangles. 5. However, the computing time to convert the logarithmic retention behaviour to the optimisation criterion of equa- tion (1) increases from l min in the one-parameter situation to 10 min for the simultaneous optimisation of two parameters. Fig. 3. Corrected phase selection diagram, now based on retention data taken from Figs. 1 and 2; the final chromatogram run at the revised optimum composition shows excellent separation over, once the solutes can be traced through the procedure, it becomes feasible to optimise for the separation of a few key solutes rather than all solutes. We consider such an optimisa- tion to be of great practical significance. Strictly speaking, for a successful optimisation of an unk- nown sample it is not necessary to identify the nature of the solutes.It is sufficient to know their number and to recognise their elution order in successive chromatograms. Given the poor separation in the initial stage of an optimisation proce- dure, peak area measurements are clearly inadequate. We have previously shown that two-wavelength absorbance ratios have a limited potential.8 Currently, however, we have obtained promising results with full absorbance spectra collec- ted during the chromatogram with a photodiode array detec- tor. The four-step procedure is illustrated in Fig. 4 for the example of polycyclic hydrocarbons. Throughout the chromat- ographic run in acetonitrile - water full ultraviolet absorption spectra are collected every second and stored in computer memory.Thereafter the data are processed as follows.9 Firstly, a chromatogram is plotted based on the maximum absorbance measured in each spectrum. This ensures that no solutes are missed, provided they can absorb ultraviolet radiation within 200 and 600 nm. By inspection, the chromatogram is divided in base line separated groups [(I to VI Fig. 4(a)]. Secondly, the330 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 4 7 230 400 4.9 5.9 Ti me/m in Wavelengthtnm Time/min F'ig. 4. Solute recognition from ultraviolet spectra. (a), Maximum absorbance chromatogram of unknown polyaromatic hydrocarbons in acetonitrile - water; (b), superposition of spectra recorded during elution of peak groups containing a single solute or more than one solute; (c), superposition of elution profiles recorded for different wavelengths to derive the number of solutes in a peak group; (d), representative spectra of solutes hidden underneath a composite profile (im)purity of a group is checked by plotting spectra collected during that group.When all spectra coincide the group contains a single solute. When there are more solutes with a minimal separation (0.1 (J is sufficient), this will show in a confusing picture of varying spectra [Fig. 4(b)). Thirdly, certain wavelengths are now selected to re-plot elution profiles. Again, for a single-solute peak these profiles coincide. The presence of multiple maxima in the elution profiles confirms peak impurity and allows a reliable estimate of the number of solutes present in each group [Fig.4(c)]. Fourthly, representative spectra of the solutes are selected at appropriate elution times. Although in the instance of unresolved peaks these spectra will not be pure, they are sufficiently characteris- tic to allow matching of corresponding solutes in successive chromatograms [Fig. 4(d)]. So far, the solute recognition from ultraviolet absorption spectra has been executed visually by human interpretation. At present, we are developing algorithms and strategies to permit evaIuation by computer. The determination of the number of solutes can be performed by principal component analysis. The matching of corresponding spectra is possible through correla- tion analysis.Conclusion The iterative regression design provides an efficient and versatile procedure for mobile phase optimisation in RPLC. The need to recognise solutes throughout the procedure opens novel optimisation perspectives. Photodiode array detection is a powerful tool for recognising the number and elution order of unknown solutes provided that they absorb ultraviolet radia- tion. Computer handling of a complete procedure appears feasible. 1. 2. 3. 4. 5. 6. 7. 8. 9. References Berridge, J. C., J. Chromatogr., 1982,244, 1. Berridge, J. C., Chromatographia, 1982, 16, 172. For example, the PESOS and OPTIM procedures of Perkin- Elmer Corp. and Spectra-Physics. Schoenmakers, P. J., Drouen, A. C. J. H., Billiet, H. A. H., and de Galan, L., Chromatographia, 1982, 15, 688.Drouen, A. C. J. H., Billiet, H. A. H., Schoenmakers, P. J., and de Galan, L., Chromatographia, 1982, 16, 48. Haddad, P. R., Drouen, A. C. J. H., Billiet, H. A. H., and Billiet, H. A. H., Drouen, A. C. J. H., and de Galan, L., J. Chromatogr., 1984, 316, 237. Drouen, A. C. J. H., Billiet, H. A. H., and de Galan, L., Anal. Chem., 1984, 56,971. Drouen, A. C. J. H., Billiet, H. A. H., and de Galan, L., Anal. Chem., 1985, 57, 962. de Galan, L., J. Chromatogr., 1983, 282, 71. The BBC Micro in Chromatography Alan G. Rowley Department of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 3JJ Of the various low cost microcomputers on the market the BBC Model B is perhaps the most attractive for consideration as the basis of a potential chromatography data system.The BBC computer has a built-in analogue to digital converter, the specifications of which call for 10-bit resolution and a conversion time of 10 ms. In practice, however, the converter falls rather short of the specification, being not perfectly reproducible to 10 bits but the error is within k 1 % at full scale reading on the converter. This error is adequate for most purposes, bearing in mind the other sources of error in typical chromatography systems. A more serious problem is the temperature sensitivity of the reference voltage used by the converter, as the BBC micro heats up considerably in use. This can be overcome by replacing the three diodes, D6,7 and 8, in the BBC with proper reference voltage devices.' Once this modification is carried out the BBC can serve adequately as a chromatography data logger and we were able to collect data at up to 10 samples per second, although in routine use four samples per second are sufficient to reproduce the performance of a typical chart recorder.The memory available in the BBC computer is quite small, and it is not possible to display the chromatogram on the screen in real-time whilst storing any substantial amount of data. This problem could be solved completely by adding a second processor, but this is an expensive option. One must, if the chromatogram is to be displayed continuously, process the data as it is collected and carry out peak detection and integration in real time, or, alternatively, retain the chart recorder for display and log all of the data points from the chromatogram into memory for transfer to disc and later processing and display in high resolution.We opted for the latter method because it lent itself well to our particular objective, which was the provision of an interactive interrogation and integration facility for the chromatogram.ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 33 1 The software that was developed allows the collection of data from around 30-40 min of chromatography, at four samples per second. The chromatogram can then be displayed on the computer screen, reproportioned along the time or mass axis, and interrogated for peak height and retention time by means of a cursor driven by the user from the keyboard. Similarly, integration is carried out interactively by marking the start- and end-points on the displayed chromatogram and inserting a base line, which can be sloping if necessary.Because the complete set of raw data from the chromato- gram is collected and can be stored on disc it is possible to return to it at a later date and to re-examine the information in a different way. Although the approach that we have chosen is not highly automated we feel most content and safe with it because all the difficult decisions, such as where peaks start and end and the base-line position, are made by the chromatographer and not by software, which must always refer back to more or less arbitrary assumptions built in when it is written. Reference 1. Aidie, C. J . , User Note No. 39, Edinburgh Regional Computing Centre, 1983.The Resolution of Overlapping Bands M. F. Fox School of Chemistry, Leicester Polytechnic, Leicester LEI 9BH The best conferences evolve as they proceed. My original objective was to discuss the effects of band overlap and the means of resolving these bands into their initial component bands. That approach assumes that little further can be done about the way in which the bands occurred in such close proximity to each other in elution time records or charts. The developments described earlier in this meeting require re-examination of this initial assumption, for it is now possible to optimise the separation of bands by varying solvent composition. The optimisation method proceeds through an elegant combination of a sophisticated, programmable multi- pump HPLC and an expert programme in a dedicated mini-computer using a simplex analysis.By repetitive analy- sis, the optimum solvent ternary composition can be found which will optimise individual band separation. The need for numerical separation of overlapped bands is reduced but not removed. The use of dual-wavelength monitoring coupled with solvent composition optimisation is a further method of resolving overlapping bands. The other point which bears in on me is the very extensive use of (mainly) HPLC chromato- graphic methods in the quality control of pharmaceuticals. The direction of my paper has therefore been re-orientated towards answering the question: “What is the limit of detection of two bands superimposed and appearing as one band?” My first consideration must be of the shape function for a chromatographic band.Usually this function is a Gaussian distribution, provided that the detector has a linear response. This band shape is not normally recognisedas the bands are presented as very sharp, narrow peaks. The second consideration is what happens when two bands come together, gradually merging, and the effect on the profile band shape. Several limits of resolution will apply. The first limit of resolution occurs when for two bands of equal or dissimilar intensity, but of equal width, merge sufficiently for the minimum between the peaks to disappear. If the width is Y and the separation of the actual peak is A, then the quantity A h is 0.854 for Gaussians. This ratio is dimensionless. Further, it is still possible by various techniques such as the LOGDIFF method to identify further bands visually and separate down to a second limit of resolution, where the point of inflection on the over-all band profile disappears at 0.64 A h .Is it possible to go further and separate two strongly overlapped peaks? We have evolved a method of resolving two bands based upon removal of a first band from a profile and testing the residue for significance as a further band. This method works well but runs into a further and final problem. This final problem is composed of two bands, dissimilar in intensity, equal in width, with a noise component, being identified as such, compared with one asymmetric band with the same noise level. The asymmetric band has the form of a log-normal distribution.We find that for an r.m.s. level of 0.01, the limit of decision between two bands of unequal intensity (but equal width) and one asymmetric band is 0.36. The distinction is not absolute, because it depends upon the r.m.s. level of noise. I hope that the statement of these limits for resolution of overlapping bands will be useful to chromato- graphers . Standardisation in High-performance Liquid Chromatography Roger M. Smith and Tony G. Hurdley Department of Chemistry, Loug h bo roug h University of Tech no logy, L oug h boroug h, L eicestersh ire LE 1 I 3TU Richard Gill and Anthony C. Moffat* Central Research Establishment, Home Office Forensic Science Service, Aldermaston, Reading, Berkshire RG7 4PN One of the attributes of the computer is its ability to search, sort and collate information from large data bases such as Chemical Abstracts, thus saving much laborious and time consuming manual searching.It should also be possible to apply this ability to the identification of analytes ( e . g . , drugs) * Present address: Home Office Forensic Science Laboratories, Hinchingbrooke Park, Huntingdon PE18 8NP. using their chromatographic properties by searching a library of retention values. However, although databases are available for gas - liquid chromatography (GLC) so far little work has appeared on the application of computer searching in high-performance liquid chromatography (HPLC), despite the widespread use of this technique in many areas of chemistry. This, in part, arises because HPLC is relatively new but undoubtedly a major332 ANALYTICAL PROCEEDINGS.NOVEMBER 1985, VOL 22 factor contributing to this lack of development has been that HPLC suffers from poor reproducibility when considered by the normal criteria of analytical chemistry. Although the precision of analyte quantification can be good, the precision of retention measurements used for analyte identification is very poor. The reason is partly that the same flexibility in separation and selectivity that makes HPLC so valuable means that if different laboratories use slightly different experimental condi- tions these can have marked effects on the absolute retention times. In addition, capacity factors (k'), the conventional method for reporting retentions, are also susceptible to errors arising from the measured value of the column void volume (to).This value is dependent on the method of measurement1 and, so far, chromatographers have not agreed on a standard method. Retention Index Scale Similar problems were also encountered in the early stages of development in GLC, but a considerable degree of standardi- sation was achieved by using the Kovats retention index system.2 In 1981 this concept lead to the proposal for a corresponding scale in HPLC based on the homologous alkyl aryl ketones,3 following an earlier report of a scale based on the alkan-2-ones.4 Since then, the alkyl aryl ketone scale has been shown to provide robust and reproducible retention values and, together with a set of column test compounds, can be used to characterise stationary phases in a similar way to the use of McReynolds and Rohrschneider constants in GLC.5-7 Applications to Drug Analysis More recent studies have examined a number of selected drug analyses to determine the suitability of retention indices and alternative methods for recording retention data for the construction of a retention data base.Barbiturate Analysis The first system examined was the separation of barbiturates on ODS - Hypersil using a methanol - buffer (40 + 60 V/V) eluent at pH 8.5.8 Using the established conditions, the intra-laboratory reproducibility of the retentions was good. Each of the experimental parameters was then varied in turn to test the effect that small differences in conditions would have on retention.9 Changing the proportion of methanol or the temperature had a marked effect on the capacity factors but only a small effect on the retention indices (Fig.1). Under the conditions of the analyses the barbiturates were partly ionised and therefore both their capacity factors and retention indices were sensitive to pH shifts. Clearly, this is the factor which must be closely controlled for good results in inter-laboratory comparisons. A major problem in HPLC is the differences between the selectivities and retentions of nominally comparable CIS reversed-phase columns. Consequently, the separation of the barbiturates was repeated on columns from different batches of the same commercial material. Both the retention indices and capacity factors were highly reproducible under these closely controlled conditions. The study was then extended to a comparison of several other packing materials and this time large differences in the capacity factors were found.The retention indices showed some variation, but each value had a much smaller range with less overlap between the different compounds.10 It was also found that, under these experimental conditions, weakly basic aromatic amines (which would be non-ionised) did not show any specific silanophilic interaction with the columns. Further, their retention indices were fairly constant.11 This was in contrast to the large differences in retention and selectivity on different columns reported for methanol - water systems.'* L 40 K~LB 30 t'. \ Capacity factors 1000 Retention indices I'b' QB 600 Y 30 35 40 45 50 Methanol,% Fig. 1. The variation of (a) capacity facotrs and (b) retention indices of barbiturates separated on an ODS - Hypersil column, methanol - buffer eluent, pH 8.5 (40 + 60). A, Barbitone; x, phenobarbitone; 0, talbutal; 0, quinalbarbitone (results taken from reference 9) Local Anaesthetics Further studies examined the more complex eluent system methanol - water - phosphoric acid - 0.1% hexylamine (30 + 70 + 100 + 1.4 V/V) at pH 2.5, which is used for the analysis of local anaesthetics on an ODS - Hypersil col- umn. 13914 Changing the hexylamine concentration caused the greatest variations in capacity factors and retention indices and this is the major factor that must be controlled in order to obtain reproducible inter-laboratory results. Although it was hoped that any differences in silanophilic interactions on different column materials would be masked by the hexyl- amine, there was a greater variation in the retention of N-methylaniline than had been observed under conditions used for the barbiturate study. Conclusion For both of these HPLC systems involving the separation of barbiturates and local anaesthetics on ODS - Hypersil the retention indices provide a more robust method than capacity factors for reporting retentions and are able to compensate for small changes in conditions such as occur between laboratories. To some extent they can also compensate for differences between packing materials. The greatest variations in retention are caused by changes in the degree of sample ionisation and this would need careful control if the results were to be used to create a data base of retention values. The authors thank the Science and Engineering Research Council for a CASE studentship (to T. G. H).ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 References 1. 2. 3. 4. 5. 6. 7. 8. Wells, M. J. M., and Clark, C. R., Anal. Chem., 1981, 53, 1341. Kovats, E., Helv. Chim. Ada, 1958, 41, 1915. Smith, R. M., J. Chromatogr., 1982,236, 313. Baker, J. K., and Ma, C.-Y., J. Chromatogr., 1979, 169, 107. Smith, R. M., J. Chromatogr., 1982,236, 321. Smith, R. M., Anal. Chem., 1984, 56, 256. Smith, R. M., Trends Anal. Chem., 1984, 3 , 186. Gill, R., Lopes, A. A. T., and Moffat, A. C., J. Chromatogr., 1981, 226, 117. 333 9. 10. 11. 12. 13. 14. Smith, R. M., Hurdley, T. G., Gill, R., and Moffat, A. C., Chromatographia, 1984, 19,401. Smith, R. M., Hurdley, T. G., Gill, R., and Moffat, A. C., Chromatographia, 1984, 19, 407. Smith, R. M., Hurdley, T. G., Gill, R., and Moffat, A. C., J . Chromatogr., 1985, submitted for publication. Engelhardt, H., Dreyer, B., and Schmidt, H., Chromato- graphia, 1982, 16, 11. Gill, R., Abbott, R. W., and Moffat, A. C., J. Chromatogr., 1984,301, 155. Smith, R. M., Hurdley, T. G., Gill, R., and Moffat, A. C., J. Chromatogr., submitted for publication.

ISSN:0144-557X    

DOI:10.1039/AP9852200323

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 333 Advances in Mass Spectrometry The following is a summary of one of the papers presented at a Joint Meeting of the Special Techniques Group and the North West Region held on March 20th, 1985, at ICI Pharmaceuticals, Alderley Edge, Cheshire. Recent Developments in Mass Spectrometric Instrumentation for Chemical Analysis J. H. Beynon Royal Society Research Unit, University College of Swansea, Swansea, West Glamorgan SA2 8PP The analytical applications of mass spectrometry have spawned the development of new instrument designs, of new ionisation and scanning methods and the coupling of the mass spec- trometer to various devices for the pre-separation of samples. Many of the developments have been made possible by the parallel development of computers capable of handling the vast amount of data generated and aiding in its interpretation.Many have come about as a result of fundamental studies by physical chemists of the properties of gas-phase ions and the fundamental mechanisms of energy transfer leading to reac- tion. Such studies have been seized upon by analytical chemists and the methods adapted and improved by them. Other developments have been the direct result of attempts to solve specific analytical problems. This paper attempts to assess the current situation, particularly with regard to sector mass spectrometers, which still make up the majority of instruments used in analysis. It also suggests some directions in which future developments may be expected. Separation techniques such as gas or liquid chromatography are now routinely coupled to mass spectrometers for the analysis of complex mixtures, the mass spectrometer acting as a rapid, sensitive and versatile device for the characterisation of the individual components of the mixtures.Several new and powerful ionisation methods have been developed which greatly extend the range of physical forms and chemical constitution of samples that can be studied and which have, inter alia, extended the mass range of the ions that can be studied to beyond 10 000 u. Hand in hand with these advances have come the development of very fast scanning of the mass scale (<1 s per decade of mass) and the use of powerful computers that can handle the vast amount of data produced, rapidly and efficiently.All of these developments have been advantageous but, in order to achieve the desired increase in mass range, speed of analysis, or to enable very small samples to be investigated, the specifications of the instruments have sometimes had to be reduced in other respects. Also, the fact that the same basic design of sector instrument has been progressively improved, in a series of small steps, has sometimes diverted attention from alternative, and possibly fundamentally better, ways of achieving the same end. This paper gives a brief survey of the main developments that have taken place in instrument performance over the past several decades and suggests some new directions for research. Recent Developments of Interest in Analysis Developments of several new ionisation methods will not be covered, neither will recent developments in Fourier transform ion cyclotron resonance (FT/ICR) spectrometers.Mass Range Recent commercial instruments can study ions of mass to charge ratio (rnlz) up to 10 000 at full accelerating voltage. This range has been achieved by the use of large, fully laminated, high-field magnets coupled with changes in the ion optics using either non-normal entry to and exit from the magnetic field or a specially contoured magnetic field. Problems still remain with regard to the sensitivity that can be achieved at such high masses at adequate mass resolution, especially when the magnetic field needs to be scanned rapidly. Comparatively little attention has been given to fundamentally different approaches to obtaining an extended mass range, particularly the possible use of time of flight (TOF) instruments, or of overcoming the problems of maintaining adequate resolution and sensitivity by operating at a lower accelerating voltage.Resolution Mass spectrometry became important in analysis largely because accurate mass measurement, which necessitates high mass resolution, enabled the individual components of multi- plets to be separated and the mass of each component measured sufficiently accurately that molecular formulae and the formulae of fragments could be determined. To aid in this determination, isotopic abundance measurements were also used. Such measurements are now used less. This is partly a result of the high speeds of scanning that are commonly used, which reduce resolution and the signal to noise ratio and make both mass and abundance measurements less accurate. It is worth remembering that slower scanning (possible unless the334 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 sample amount is very small) can overcome these difficulties.It is surprising that no attempt has been made in analytical instruments to improve the mass accuracy beyond 1 p.p.m. because this would greatly increase the certainty of formula elucidation. Computers The most interesting developments have come about because of the use of sophisticated computers to collect and process, rapidly and reliably, the information contained in a spectrum. Widespread use is also made of computers to search spectral libraries and match the spectrum of an unknown.There are, however, disadvantages in the methods that are used at present. The usual presentation of spectra in bar-graph form gives no hint of the, sometimes large, uncertainties in abundance of peaks that may contain comparatively few ions; more informative methods of comparing spectra with that of an unknown have received relatively little attention, although they may require a far less sophisticated computer system. Detectors Focal-plane detectors offer advantages in the speed and sensitivity of recording spectra. So far, little attention has been paid to their use in analysis. New Instruments One of the major developments in the recent past has been the availability of “reversed” geometry double focusing mass spectrometers, in which the ion beam traverses the magnetic sector first.These offer several advantages in analysis. The first of these is that often the molecular ion of a minor component in a complex mixture can be isolated by using the magnet, and its characteristic fragmentation pattern determined (using, for example, collisional excitation). This technique is sometimes referred to as mass spectrometry - mass spectrometry (MS - MS) or tandem mass spectrometry. In order to improve the resolution at which the parent ion can be selected and also the resolution at which the fragmentation pattern can be observed, multi-sector instruments are now available. In these the first two sectors, which form a double-focusing mass spectrometer, are used to select the parent ion of interest; a further two sectors will enable the observation of daughter ions, also at high mass resolution, as the double-focusing action overcomes the problem of the spread of ion translational energies brought about during the fragmentation process.For the study of pure compounds the method also has considerable analytical advantages. The fragmentation pat- terns of all of the major ions in the mass spectrum can be investigated in turn and a complete “fragmentation map” built up; the functional groups present in an unknown compound can be identified in this fashion, while large organic molecules can be identified using this method, even when conventional mass and NMR spectrometry have failed to elucidate its structure. Another interesting recent development has been the construction of so-called “hybrid’ instruments, in which one or more quadrupoles are coupled to sector machines. Such coupling necessitates retardation of the ion beam entering the quadrupole, the retardation voltage being linked to the voltage across the electric sector plates. The advantage of such “hybrid” designs is that three different measurements can be made on the ion beam as it traverses the instrument, momentum to charge ratio (using the magnetic sector), energy to charge ratio (using the electric sector) and mass to charge ratio (using the quadrupole) . Future developments are likely to be influenced by the variety and reliability of the electronic circuitry now available. Multi-sector and hybrid machines capable of scanning the fields in various linked fashions will become more widely used and should give increased performance and versatility over quadrupole instruments of comparable size. Rapid develop- ments are also taking place in TOF and ICR instruments that will also be of undoubted benefit to chemists dealing with complex analytical problems.

ISSN:0144-557X    

DOI:10.1039/AP9852200333

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

334 ANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 Expert Systems for the Analyst ~ ~~ ~~ The following is a summary of one of the papers presented at a Joint Meeting of the Special Techniques and Automatic Methods Groups held on June 4th, 1985, at the Scientific Societies Lecture Theatre, 23 Savile Row, London, W.1. Application of Expert Systems to Corrosion Problems D. E. Williams and C. Westcott Hatwell Corrosion Services, Materials Development Division, AERE Harwell, Oxfordshire OX I I ORA 1. F. Croall and S. Patel Computer Sciences and Systems Division, AERE Harwell, Oxfordshire OX 1 1 ORA The prediction of corrosion limited service life, and the evaluation and specification of materials in terms of corrosion resistance, are tasks which require expert knowledge and judgement.However, corrosion is a very wide field, corrosion experts are few in number, and expertise in particular areas is scattered. Corrosion related problems, often very simple in origin, may appear because the necessary expertise might not be readily or immediately available at the design stage, or during maintenance or repair operations, or when modifica- tions to operating procedures are considered. Large financial penalties, and sometimes loss of life incurred as a result of corrosion induced failures, may draw attention to these problems from time to time, but the less obvious costs in terms of avoidable repair and maintenance are an ever-present problem, sometimes revealing a lack of even basic corrosion knowledge on the part of the design engineers and plant operators.At the present time corrosion expertise resides in persons with experience or with access to pertinent informa- tion, but current developments in information technology, data-base management and knowledge engineering offer the possibility of pulling together corrosion expertise and data from many sources into a so-called “Expert System,” that is, a computer system which combines a data base with expert knowledge coded in the form of rules, and which appears to interact with a user in the same way that a consultant does. Some of the questions posed in the selection of materials for process plant are: what should any given component be made from?; how can it best be protected?; how will it affect or beANALYTICAL PROCEEDINGS, NOVEMBER 1985, VOL 22 335 affected by the behaviour of the rest of the structure?; and will it last for the design life? These questions are not simple to answer, because there will be matters other than corrosion to be taken into account.Corrosion may be one of the more significant factors affecting any decision, but the designer may need to give equal or more weight to other factors such as strength, fabricability , cost and availability. The expert system, like the human expert, must therefore interact with the designer, to discover where experience suggests a preferred material or protection method, but should also be able to respond to the designer’s requirements by suggesting alterna- tive approaches and compromise solutions. By setting up a comprehensive system of rules and providing access to an extensive data base the expert system will provide a sound basis for design against corrosion in the majority of instances.This is clearly an area where a collaborative research programme will be of benefit to all, as no one organisation will have available all the necessary expertise and resources to provide a com- prehensive corrosion expert system. Harwell has already developed a considerable degree of refinement in the methodology of constructing and utilising expert systems. In collaboration with the National Corrosion Service, NPL, a simple prototype expert system has been developed in order to formulate the requirements for and to identify the problems involved in building an expert corrosion system as described above. This system has been developed to the stage where it can demonstrate a significant ability to provide interactive advice to the user.It has been demon- strated to a number of potential users, and has met with an enthusiastic response. The development has allowed many of the computer software problems to be identified and solved. In particular, the problem of linking an expert system to an information retrieval system has been tackled so that best use can be made of the most powerful features of both types of software. The place of the information retrieval system in the whole concept can be understood by analogy with a consultant, who refers as necessary to databooks, standards, codes of practice and manufacturers’ data sheets in the course of a consultation. Our objective is now to develop a set of expert systems aimed at materials engineers, plant designers and operators.The systems will incorporate a substantial digested database of corrosion information and will give appropriate recommenda- tions derived from this database. Each system will deal with a specific corrosion theme. They will embody current good practice and will be based both on information freely available in the literature and on expertise and specialist data provided by the sponsors of the project. The construction of a comprehensive suite of expert systems will be a considerable task. Our view is that such a system can be built by first using check-lists to break down a complex field into its smaller components, then by commissioning authorita- tive reviews of specified narrow areas of corrosion problem derived from these check-lists.These reviews will clearly state the requirements of current good practice, and will give critical appraisals of the strengths and weaknesses of particular materials. Key papers in the literature will be identified and appropriate numerical data will be given. No doubt the success of this procedure will rest to some extent on the goodwill of the sponsors in providing such data. These reviews, and the associated literature, will be put into the database. Rules for the expert system will then be derived from them. The expert system will point to the appropriate reviews as the source of the advice given. In this way, we feel that a large and complex system, involving contributions from a large number of people, can be constructed in a simple and logical fashion.The linkage of expert system and information retrieval system is the particular feature of our work that makes such an approach feasible. We believe that it will be possible to build on these foundations to construct a general system of real usefulness. The authors would like to acknowledge the particular contribu- tions by Mr. N. J. M. Wilkins (Materials Development Division) and Dr. J. M. Wanklyn (Consultant) in helping to establish the logical and philosophical basis of the project and by Mr. J. Bernie (National Corrosion Service, NPL) in the support of this work. Reprints of Important Analytical Chemistry Reviews In an attempt to ensure that major developments in chemistry reach as wide an audience as possible RSC has made available reprints of important reviews which were published in RSC Journals Modern Analytical Methods for Environmental Polycyclic Aromatic - - Compounds by K.D. Bartle, M. L. Lee, S. A. Wise Polycyclic aromatic compounds are major pollutants of the environment, originating from many sources. This paper reviews the techniques that are available for identification and analysis and provides the reader with a comprehensive and authoritative source of information on the subject. The paper is divided into the following sections: Introduction; Sample Preparation; Chromatographic Methods; Mass Spectrometry; Spectroscopic Methods. The review, which contains more than 400 references, will b of interest to environmental, petroleum and analytical chemists.Price f2.50 ($5.00) The Royal Society of Chemistry Burlington House Piccad i I I y London W1V OBN Standardised Thin-Layer Chromatographic Systems for the Identification of Drugs and Poisons by A. H. Stead, R. Gill, T. Wright, J. P. Gibbs, A. C. Moffat The October ‘82 issue of The Analyst featured a review of this area entitled Standardised Thin-Layer Chromatographic Systems for the Identification of Drugs and Poisons. The wide use of TLC for the analysis of drugs and poisons in biological fluids and pharmaceutical preparations suggests that this article will be of great interest to many analysts working in the field. Consequently, The Royal Society of Chemistry has decided to make separate reprints available. Coverage This review gives criteria for good systems and applies them to the selection of the eight most effective. The selected systems are standardised by the use of standard running conditions and the use of reference compounds. Rf x 100 values are given for 594 basic, 48 neutral and 152 acidic drugs on the selected systems both in alphabetical order and ascending order of Rf for each system to aid the identification of unknown drugs. Further identification is enhanced by the inclusion of various locating procedures. Price f5.75 ($1 1.50) Prices inclusive of p & p to UK and European destinations and Surface Mail outside Europe. Airmail outside Europe at cost. To order the above reprints please send payment and a self addressed envelope measuring 6 x 9 minimum to: The Royal Society of Chemistry, The University, NOTTINGHAM NG7 2RD, England.

ISSN:0144-557X    

DOI:10.1039/AP9852200334

出版商:RSC    

年代:1985

数据来源: RSC