ISSN:0003-2654    

DOI:10.1039/AN98005FX041

出版商:RSC    

年代:1980

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98005BX043

出版商:RSC    

年代:1980

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98005FP129

出版商:RSC    

年代:1980

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98005BP135

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

NOVEMBER 1980 The Analyst Vol. 105 No. 1256 Assessing the Analytical Quality of the Clinical Laboratory* T. P. Whitehead Department of Clinical Chemistry, Wolfson Research Laboratories, Queen Elizabeth Medical Centre, Edgbastoia, Birmingham, B15 2TH The clinical laboratory services of the UK Kational Health Service (NHS) is the fastest growing expenditure area of the Service. The clinical chemistry services are probably the largest of the disciplines making up the clinical laboratory service regarding workforce, workload. and expenditure. In the early years of the SHS there was evidence of poor analytical performance by the clinical chemistry laboratories. This was shown by external quality assessment (EQA) techniques. By continuous and frequent use of EQA where portions of the same material are sent to over 400 laboratories and analyses compared it has been shown that the analytical performance has considerably improved. The reasons for this are probably the adoption of methods shown to perform well on EQA and also the widespread use of internal quality control techniques. The difficulties of providing appropriate material for EQA and the method of dealing with poor performance are also discussed.Keywords : Clinical analysis ; analytical quality Approximately 4;4 of the costs of the National Health Service hospital costs are spent on clinical laboratories. Thus, in the current year approximately l250 million will be spent on laboratory services. The labour force is now said to be higher than that of the Scientific Civil Service.I t has for many years in a situation of escalating costs been the forerunner, growing faster than any of the other services. This is a reflection of the increased use of laboratory investigation in the diagnosis and treatment of patients. Within the various disciplines comprising the clinical laboratory services the discipline of clinical chemistry has probably the largest workforce, workload and investment in capital equipment. I t is the youngest of the disciplines only having a history of some four decades. The laboratories comprising the discipline are virtually all situated in hospitals although providing some general practitioner services. The labour force includes graduates in science and medicine and technologists. A basic role of the clinical chemistry laboratory is to provide reliable data on the composition of specimens obtained from patients as an aid to the diagnosis and treatment of disease.During the last three decades there has been abundant evidence both from the UK and from many other parts of the world that where portions of the same material are analysed by a number of laboratories the variance in the results obtained indicates a situation which is unacceptable on the basis of both clinical and analytical criteria. For example, in 1952 Professors Wootton and King from the Royal Postgraduate School of Medicine in London distributed a serum to laboratories in the UK for calcium analysis. Half of the laboratories stated that the level of the calcium was “within the normal range,” one quarter stated it to be above normal and one quarter to be below normal.Similar surveys in the USA resulted in both Federal and State regulations regarding the accreditation of laboratories. In the USA Senate such words as scandalous and criminal were used to describe some of the results produced at high cost to the patient in the USA. The first individuals to practice clinical chemistry were graduates in medicine and bio- chemistry and the technologists were frequently recruited from other disciplines such as * Plenary Lecture presented at the 5th SAC International Conference on Analytical Chemistry, Lancaster, There are approximately 400 such laboratories. J u l y 20-26th, 1980. 10091010 WHITEHEAD : ASSESSING THE ANALYTICAL Analyst, Vol. 105 histopathology. Thus there was not a sound basis of analytical chemktry in the subject, the introduction gradually over the years of more chemists has been a stimulus to better analytical training for all entering the discipline.This paper describes the observation of the analytical quality of clinical chemistry laboratories in the UK and summarises the endeavours made to improve the situation. Quality Assurance All the measures taken to ensure good clinical chemistry practike from preparation of the patient before collection of a specimen to correct interpretation of a result are described as quality assurance techniques. Changes in the composition of biologitxd specimens produced by incorrect collection procedures or inappropriate specimen handling can far outweigh the analytical variance.Assuring quality in patient investigation .is not solely an analytical problem. Internal Quality Control Clinical chemistry laboratories in the UK now normally practise some form of internal quality control, usually the performance of assays on quality control material a t the same time as patients' material. This type of quality control has been introduced as a result of active teaching for all grades of staff, the inclusions of questions regarding the techniques in examinations and active research and development resulting in many publications. External Quality Assessment This is where portions of the same material are sent to several laboratories for particular analyses and the results compared. This paper describes the author's experience with the UK National Quality Assessment Scheme (NQAS) in Clinical Chemistry.The Scheme is sponsored by the Department of Health and Social Security (DHSS), who also sponsor schemes in the other disciplines. The Scheme is operixted from the If'olfson Research Laboratories, Department of Clinical Chemistry, Queen Elizabeth Medical Centre, Birmingham. The United Kingdom National Quality Control Scheme (UKNQCS) An important stimulus to such activity has been external quality assessment. It is financed by the DHSS and employs six people. The main objectives of the Scheme when it started in 1969 wer'e: 1. To send a t 14-day intervals a portion of the bulk human. serum and, on occasions, non-human serum, to all those hospital laboratories in the UK which perform clinical chemistry analysis.A laboratory which did not routinely perform all of the 15 analyses would not necessarily be excluded from participating in the Scheme. To return results from participating laboratories to the organising laboratory quickly; the results from all laboratories to be available to the participants within 10 days of the specimen arriving in the participating laboratories. 2 . To assess results from 15 of the more commonly performed analyses. 3. 4. 5 . 6. 7. To make participation voluntary and preserve anonymity. To present the results in a manner that would enable the participants to make judge- To assess the role of automation, analytical methods, laboratory workload and other To determine whether any improvement in precision and accuracy in the hospital ments of their performance, particularly in relation to the analytical method used.factors possibly affecting the variance of results. laboratories of the UK occurred as a result of such frequent surveys. Organisation of the Scheme The distribution of serum specimens to 200 laboratories in the UK began in July 1969; a t the present time t:he participants number 420. There is reason to think that the vast majority of laboratories within the NHS per- forming clinical chemical analysis have entered the scheme and approximately 90% of The Scheme has been in operation for 11 years.A’otlewabev, 1980 QUALITY OF THE CLIKICAL LABORATORY 1011 participating laboratories return the results for each distribution of serum. At the present time the survey regularly includes 15 different chemical determinations.These are serum sodium, potassium, chloride, urea, glucose, calcium, phosphate, iron, total protein, albumin, bilirubin, alkaline phosphatase, cholesterol, uric acid and creatinine. Surveys of blood lead, thyroid function tests and certain enzymes are also being carried out. The computer in the author’s laboratory has been programmed to perform virtually all the clerical tasks involved in the scheme and the survey involves approximately 4 h of computer time each fortnight. The difference between the NQAS and previous surveys was the frequency of survey, 20 times a year, approximately five times as frequent as many other schemes; also the relatively short delay, 10 days, between receipt of the specimen and the receipt of the results of the survey.T h e material distributed Obtaining sufficient suitable material of acceptable quality for distribution is a major problem. Virtually all the material used a t the present time has been lyophilised, usually by commercial organisations. I t is not always of human origin; animal material is suitable for some tests. Confidence by the participants in the organisation of such schemes is based primarily on the quality of the material distributed; it is common for laboratories with poor performance to blame the material. Proof that the quality of the materials distributed is satisfactory, is based upon the fact that certain participants consistently achieve, over a long period of time, an excellent analytical performance on material randomly distributed for individual surveys.The problem of how well reconstituted lyophilised material simulates patients’ serum will be dealt with later. Timetable of distribution of seva a d results Labelling for distribution is facilitated by the computer line printer, which draws on a disc file of addresses and code numbers of the participating laboratories. The computer also prepares the docunient to be returned by the participants. This document (Fig. 1) indicates ~vhich analyses are to be performed, which units are to be used and also gives any special instructions regarding reconstitution of the material. The documentation and material are packed in a protective polystyrene box. All packs are posted on a Saturday and almost without exception arrive in the participating laboratories early on the following Monday morning (day 1).The laboratories perform the analyses listed and return the results before the following Monday (day 8). On day 8 all results are put into the computer and following verification they are analysed on the Tuesday (day 9). Computer printouts of results are posted to the participating laboratories on the Wednesday (day 10). This timing of distribution enables at least 9076 of all participating laboratories to be included in the printout prepared on day 9. Both these differences were made possible by the use of computer facilities. Lab: Distribution: 3 Last date for receipt of results:?c 4 - E ; U !I t l i i # : / 0 5 3 j f , l ,/I ‘I 1 / 1 J t i (1 i t i ~ J G i ~ u iir 0) I / I [: i 1 i I- u i7 i i-iy31u,mu I / 1 T o L o l I V t - o t c i t < J / l fi 1 k~ u $11 i I., - I 1. i 1 c I 11 ,,J ~ T ~ 8 3 ~ e 1 fil I’ 1 / 1 1 1 - c 8 1 , ~ I’F. lL,,lli>l/l p ! 1 ,, 5 I, I1 <I t # c ~ - & q I r ! - - I,! I? 1 ,’ I L A Please check your laboratory number is correct Fig. 1 . Document to be returned by participants. Format of report to fiavticipaitts It is important that such documentation has information of educational importance comparing performance by Each participating laboratory has its own computer-printed report.1012 WHITEHEAD : ASSESSING THE ANALYTICAL Analyst, Vol. 105 different analytical methods. It is also important that the statistical methods used are clearly understood by all participants. It is frequently those laboratories with least under- standing of the statistics used that need the most help in improving their performance.The following is a description of the information provided in the computer printout. The computer lists the results attributed to each laboratory so that they may be checked for clerical errors by the participating laboratory. The mean, standard deviation (S.D.) and coefficient of variation for each determination are calculated and printed. After removal of all results outside three standard deviations either side of the mean, these statistics are re-calculated and these are termed the re-calculated mean and standard deviation. An example of this portion of the printout is shown in Fig. 2. This technique eliminates those results which are probably due to random mistakes. Following the statistical calculations, there are printouts of histograms of the reported results for each determination.An example of such a histogram is shown in Fig. 3. The range of the histogram corresponds to the re-calculated means, &2 S.D. A result within the limits is shown by a cross and a result outside these limits by a dot. These limits are not "limits of acceptability" but are a con- venient method of presenting the results, which enables each participant's results to be related to all other results. 3 e 7 LABS PARTICIPATED IN THE SCHEME > i n h L L JHLA G L u C C A L C tHLlS EL NO OFRESULTS j L q 3 d q 2 7 3 31.1 307 3 3 5 L b l L L I MEANVALUE 1 3 5 . 0 ~ . 8 4 1 V b . 4 b.U 3 . 7 L . U L 1.55 Z b . 3 COEFF OFVAR 1 . 3 4 . U 2 . 5 7 . 7 11.4 6 .7 b.3 1U.b STD. DEVIATION l . Y d.14 2 . 7 0 . 5 0 . 4 u . 1 4 U . 1 L 3.11 RECALCULATED RESULTS EXCLUDING THOSE OUTSIDE * 3 SD IN THE ABOVE CALCULATIONS:- NO. OF RESULTS j 7 Y 3 7 8 2 7 0 3 6 4 3 b j 3 3 5 2 b 4 2 2 U STD. DEVIATION 1.b U . U 3 2 . 4 U.4 (1.3 U.Ub U . 1 U 2 . r MEANVALUE 1jY.u 2 . b 3 1Ub.4 b . u 3.7 2.bL l . Y > 2 8 . 0 COEFF. OF VAR. 1 . 1 3 . 3 2 . 3 h . 3 Y.3 4.U 5.2 b . 3 Fig. 2. Example of calculation of over-all means and re-calculation. Tlie computer disc file contains information regarding the analytical method in use in the participating laboratories for each determination and the results are classified according to the methods in use. Only results used in the calculation of the re-calculated mean are included. The mean, standard deviation and These are presented as statistical summaries.h l > l U b h h < f'Ud LLUCdhL l*s**I****l**r*~***r~****~****~****l * * /*I 1'1 d L / L * * 4 < 2.80 *. i 4 I.YU * A * 3 3 . l u * A X * 3 a j . j u - - > * A A X X X X X X X * * X X X X A A X A X X X X X x X A A A X h X A X .h X X X X X A A X X * 1 3 3 3.3u 7 9 * 60 3 . 3 u * X A A h A X X X A X X h X A X * 3 . 7 u * X A A h X x X X X X k X A X I; X X k 2.i 4 . 1 0 * x x Y, X A X * U 4 . 3 0 * * 5 4 . 5 0 * A * 1 1 > 4.b0 *.. * * * 4 * * * l****l****I****I****~****I****~****I lvUb',dILK Llt' kLSULTb 2U 4 0 60 b U 1 U O 120 1 4 U NUiluLH Gt RESLiL'I'S 11" MlSTULkAR 3 6 7 Fig. 3. Example of a histogram for glucose. The arrow indicates the actual value.November, 1980 QUALITY OF THE CLINICAL LABORATORY 1013 coefficient of variation of the results of each method are calculated and a summary is typed and included in the report received by each participant.Table I illustrates the format used for the presentation of the results according to analytical method; glucose is used as an example. TABLE I EXAMPLE OF PRINTOUT FOR GLUCOSE SHOWING THE STATISTICS OF INDIVIDUAL METHODS HLSULTS FUk GLUCOSE (MlkOL/L) (CXCLUUING VALUES O U T S l U h + / - 3S.D.) O V t H A L L -->I,, A h u kL - ti L u c us 6 0 x 1 u A s r: b l h E H A IJ 1 O A h A L Y Zt H I H t U U C 1' 10 i i A H 1 1 OH SI4A HEDUCTlUiV A A 1 GLUCD5C O X l D A S E A A 11 O R SEiA GLUCOSC DX. t: k. C K h P h A IY A L Y Z E H ti L IJ C DS E NU. *IEAI'I S.U. C.V. 3 6 3 3.66 0.34 9 . 2 6 6 9 3.70 0.46 12.57 5 1 3.62 U.36 9 .8 1 1 8 3.83 0.47 11.21 1 4 3.91 0.32 8.25 9 9 3 . 6 9 l 1 . 2 5 6.66 6 2 3.68 0 . 2 9 7.8b 4 t 3.52 0.24 6.78 Y U b H H L S U L I : 3.20 The variauce in,dex If EQAS are to play a part in improving the performance of the participating laboratories then it is the author's experience that some type of performance scoring is essential. The performance of individual laboratories should be compared with the performance of all participating laboratories. This means that all laboratories should know what the best laboratories are capable of. Mere declaration of the percentage of laboratories achieving results within a certain acceptable coefficient of variance does not communicate such informa- tion. Performance-in any individual laboratory and in all laboratories taking part in a scheme will alter with time; knowledge of the magnitude of such changes is also important.Laboratories in EQAS should be provided with information indicating quantitative relations of performance of individual laboratories to methods, apparatus, workload, staffing, etc. I t is for these reasons that the variance index was devised. Calcdatioiz of the variame index (VI) The V I is calculated on the results obtained from the participating laboratories for a particular determination. First, the mean value obtained by laboratories using the same method is calculated. Previously, the type of analytical methods used by participants for individual determinations have been classified and those using the same or similar methods are grouped together for calculation of the method mean.The participant is required to agree to such classication. For some determinations participants may use methods which cannot be classified in this way and their results cannot be used in PI calculations. More than 90% of the 420 participating laboratories use methods which can be used for PI calcula- tions. The calculation uses only those values which fall within the mean rf3 S.D. for the results returned by participants for this method. This is to avoid incorporating results which are random mistakes, such as those occurring in clerical transcription into the method mean calculation and thus falsely distorting the value. The method mean ( X m ) is subtracted from the result of an individual laboratory (x) and the percentage variation from the method mean calculated.x m x 100 x - 3 Variation (yo) = V = - .y1014 WHITEHEAD : ASSESSISG THE AXALYTICAL ilnalyst, Vol. 105 (The sign is ignored). The V I is calculated from this figure by dividing it by the chosen coefficient of variation (CCV) given in Table 11. To avoid decimal points this figure is multi- plied by 100. Variance index = V I = c;V ~ x LOO Obviously, the lower tlie V I , the closer the result is to the method mean. The CCV values shown in Table I1 are the lowest CVs obtained for particular determinations during the first 2 years of the scheme. They are kept constant so that improvements in tlie performance of laboratories can be detected. TABLE I1 CHOSEN COEFFICIENT OF VARIATION USED I N V I CALCULATIONS Coefficient of Determination variation, yo Sodium .. . . . . . . 1.6 Potassium . . . . . . 2.9 Chloride . . . . . . . . 2.2 Urea . . . . . . . . 5,7 Glucose . . . . . . . . 7.7 Calcium . . . . . . . . 4.0 Phosphorus . . . . . . 7.8 Iron . . . . . . . . 15.0 Determination Kric acid. . . . . . Creatinine , , . . Bilirubin . . . . Albumin . . . . . . Alkaliiic phosphatase . . Cholesterol . . . . Total protein . . . . Coefficient of variation, yo . . 7.7 . . 8.9 . . 19.2 . . 3.9 . . 7.5 . . 19.6 . . 7.6 Because the coefficient of variation and not the standard deviation is used in the calculation of V I when the mean value falls outside tlie limits listed in Table 111, the VI is not calculated. It is particularly important to avoid V I calculations on low mean values for serum determina- tions with a high variance such a bilirubin, alkaline phosphatase and iron.A formal definition of variance index is “the difference between the result obtained by a participant and the calculated method mean expressed as a percentage of the mean, divided by a chosen coefficient of variation for that determination; the resultant figure is multiplied by 100.” TABLE I11 RAXGE OF VALUES USED I N THE V I CALCULATIOXS Dctermination 1,ow High Units Sodium . . . . . . 110.0 160.0 minol l-5 Potassium.. . . . . 1.5 8.0 mmol 1-1 Chloride . . . . . . 65.0 130.0 niinoll-’ Crea . . . . . . 2.5 66.7 mmoll-’ Glucose . . . . . . 0.8 22.2 minol 1-1 Calcium . . . . . . 1.0 4.0 Innloll-’ Phosphorus . . . . 0.6 3.9 Inmoll-’ Iron . . . . . . 3.6 53.6 pmol 1-1 Urate .. . . . . 179.0 893.0 pmol 1-1 Crcatinine . . . . 62.0 1770.0 p n o l 1-1 Bilirubin . . . . . . 9.0 342.0 pmol1-1 Total protein . . . . 40.0 100.0 g 1-1 Albumin . . . . . . 15.0 60.0 g 1-1 Allialiine phosphatase . . 6.0 100.0 KA units per 100 in1 Cholesterol . . . . 1.3 12.9 Vnriame iiadex stove (\’IS) the same substance may be expressed as the mean variance index. the term variance index score (VIS). score, possibly due to a clerical error, V I values greater than 400 are treated as 400. The performance of an individual laboratory in several analyses of different material for In tlie UKQAS we use To avoid incorporating very high V I values in theNovember, 1980 QUALITY OF THE CLINICAL LABORATORY 1015 The mean VIS may be calculated for different determinations and several distributions ; the resultant calculation is the ovev-all variance index score. In practice it has been found useful to calculate the ‘‘rwnning” over-all variance index score.In this the over-all variance index score for the most recent 40 analyses is calculated. I&‘here the score for more recent results are added, the appropriate number of the earliest results are dropped out of the calculation. Running variance index score plotter. of running V I score. The graphs involve a period of 2 years. that a laboratory did not return a result for that particular determination. and the best 5% lines delineate the area in which 90% of laboratories have VIS. Figs. 4-7 illustrate the running VI score graphs, which are drawn on the computer graph Each square represents a distribution of serum which was used for the calculation A break in the graph indicates The worst 5% 200 200 F 150 150 W G1 ._ .- ; 100 p 100 p 50 m m .- ._ 50 9 0 ‘4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 4 8 12 16 20 24 28 32 3640 44 48 52 56 60 64 Events Events Fig.4. Example of the running variance index Fig. 5 . Example of the running variance index score computer printout. score computer printout. Fig. 4 illustrates the results from a laboratory where performance was poor for the first 6 months of the period, gradually improved in the next 6 months, remained static as regards performance in the next 6 months and then dramatically improved within the last few weeks. Fig. 5 shows the results from a laboratory with a very consistent performance kept up over a period of 2 years.Fig. 6 shows the results from a laboratory which had an average performance which dramatically deteriorated over a period of a few weeks owing to staff difficulties and then improved when these problems had been solved. Fig. 7 is a typical running VIS graph. 200 150 m - 8 100 ? 50 ?? 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 “4 8 12 16 20 24 28 32 36 40 44 4852 56 60 64 Events Events Fig. 6. Example of the running variance index Fig. 7. Example of the running variance index score computer printout. score computer printout. Throughout the scheme it has been shown that the laboratories with the smallest workload The current mean values for various levels of workload are shown There is good evidence that those laboratories making extensive use of auto- Smaller laboratories using more manual have the highest V I S .in Table IV. matic methods of analysis have the lowest V I S . methods are less precise. tories. At intervals, a table of the type illustrated in Table V is distributed to individual labora- It shows the VIS for the individual determinations and the mean VIS for each determination for all laboratories. butions to their over-all V I S . In this way participants can judge which determinations are making significant contri-1016 96 94 92 90 WHITEHEAD : ASSESSING THE ANALYTIC.AL Analyst, Vol. 105 TABLE IV RUNNIXG OVER-ALL MEAN vI.$ FOR LABORATORIES WITH DIFFERENT WORKLOADS The figures in parentheses are the number of tests performed each year. All laboratories .. . . . . 68 Size 3 (101 000-185 000) . . 66 Size 1 (50000) . . . . 83 Size 4 (188000) . . . . . . 61 Size 2 (50000-100000) . . . . 71 - - - TABLE V EXAMPLE O F RESULTS OF A PARTICIPATING LABORATORY SHOWING THE V1.s FOR INDIVIDUAL DETERMINATIONS COMPARED WITH THE, MEAN VIS FOR ALL PARTICIPATING LABORATORIES Detcrmination Sodium . . * . Potassium . . . . Chloride . . . . Urea . . * . . . Glucose . . . . Calcium . . . . Phosphate . . . . Iron . . . . . . Uric acid . . . . Creatinine . . . . Bilirubin . . . . Total protein . . Albumin . . . . Alkaline phosphatase Cholesterol . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . No. of No. of possible results results returned 34 34 29 28 20 11 34 34 33 33 34 34 34 32 20 19 33 33 32 32 31 23 33 31 33 33 28 26 33 32 Mean V I S 46 42 61 21 21 45 33 90 28 51 25 47 17 38 56 Mean V I for all laboratories 86 88 80 79 85 78 75 77 79 98 96 75 79 79 88 Improvement in Performance The mean VIS has consistently fallen over the years since it was introduced in late 1972.Fig. 8 illustrates the fall from November 1972 to June 1975. Over the last 5 years the fall has remained consistent but slower and the mean VIS for all laboratories is approximately 60% of that in 1972. Fig. 8. Variation of mean V I S over the period November 1972 to June 1975. The reasons for the improvement are possibly two-fold. Firstly, there have been con- siderable changes in the methods used by laboratories for nearly all determinations surveyed. Manual reduction methods for glucose, used by a quarter of participants in 1972, are inherently imprecise and had been completely replaced by 1979 with glucose oxidase methods, many automated. Salt fractionation methods of albumin determination, a difficult method usedNovember, 1980 QUALITY OF THE CLINICAL LABORATORY 1017 by one third of the participants in 1972, has been abandoned for the simple and precise bromo- cresol green method.There are several other such changes and it is known that the EQAS has helped to persuade laboratories to change to better methods. Secondly, there has been an improvement in the precision of methods in constant use during the 11 years of the scheme. For example, flame photometry of sodium and potassium has shown consistent improvement and the between-laboratory coefficient of variation is consistently less than 4=2.0y0, Dealing with Poorly Performing Laboratories Three years ago the professional bodies associated with clinical laboratory work set up panels to consider the perforniance of individual laboratories.The results of each laboratory are considered by a panel of four laboratory workers. They decide whether the performance of a laboratory is acceptable; if not, then a letter is sent pointing out the poor performance and offering help. So far this technique has been successful and it is not envisaged that there will need to be compulsory participation such as in the USA and Germany. Choice of the target values critisised. laboratories is more appropriate. means on the Scheme lie very close to the values obtained by definitive methods. Choice of Material for Distribution The Scheme relies on lyophilised material. There is now increasing evidence that some materials lack commutability, that is, the ability to simulate fresh patients’ serum, and this is certainly a problem with some methods. The addition of preservatives and other substances to lyophilised material may also produce problems in certain determinations. Organisers of schemes such as the UKNQXS have to be continually concerned with research into the properties of lyophilised material. An additional problem is that the ability to obtain and lyophilise material in large batches is based solely in commercial organisations. There is no doubt that studies of material properties is required if further development in survey work are to take place. Confidentiality has been preserved throughout the course of the Scheme. The use of the method mean as the target value on which to judge perfrrmance has been For some determinations, such as enzyme, the use of values prov.ded by reference There is evidence that for the ions sodium, potassium, chloride and calcium the method A combination of consensus and reference laboratories is gradually being adopted. This work would not have been possible without the help of many colleagues and the support of the Department of Health and Social Security.

ISSN:0003-2654    

DOI:10.1039/AN9800501009

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

1018 Analyst, November, 1980, Vo2. 105, #. 1018-1031 Analytical Instrumentation for the 198Os* Howard V. Malmstadtt School of Chemical Sciences, 41 Roger Adams Laboratory, University of Illinois, Urbana, Illinois 61801, USA Several microcomputer-controlled analytical instruments are described that illustrate certain automation concepts, focus attention on soine recent developments and help us to envisage what may be introduced during the 1980s. Yew types of stopped-flow analysers are shown to have co’nsiderable promise for routine and investigative kinetic and equilibrium anal,ytical/ clinical methods. Throughputs of up to 5.00 samples per hour with O . l - l % precision are practical. Various types of rotating-disc analysers are also described. These include a multi-sample colorimeter, a unique .multiple absorption curve spectrophotometer, and a multi-test or multi-sample analyser for clinical/analytical laboratories. Complete systems are discussed, including preparation and transport of samples and reagents, the: encoding of the desired chemical information as digital electronic signals, optimisation and control of measurement conditions, data processing and final display of the chemical information. Keywords ; Microcomputer-controlled analytical instruments ; stoppe!d-flow analysers; rotating-disc analysers ; automated sample preparation; analytical instrumentation Impressive developments during the 1970s have paved the way foi- an exciting new genera- tion of automated analytical instruments for the 1980s. Microelectronics and micro- computers and other evolving technologies are having a major impact on instrument design.This, indeed, should be the decade of elegant microconiputer-controlled analysers. Many of the recent instrument designs now in production by several companies incorporate microprocessors to provide improved sensitivity, selectivity, precision and accuracy for classical analytical techniques. Spectrophotometers, pH meters, titrators, gas - liquid chromatographs and many other commonly used instruments are all benefiting from the effective control and “intelligence” introduced by inexpensive mass-produced micro- processors. The interlinking of several “intelligent” instruments by. automated communica- tion networks will lead to more and more automated laboratories and even to automated research and development.A complete automated analytical system should, of course, start with reliable sampling and ensure positive sample identification throughout all operations. After automatically preparing the sample for measurement, all of the desired chemical information should be encoded as digital electronic signals so that the data can be readily processed and the results displayed in the best format for aiding the personnel who require the data. Every step in the automated analytical process should involve closed loop feedback, including calibration and checking of all devices and parameters (wavelength, photometric readings, temperature, etc.) against certified standards each day, hour, or even each series of determinations to ensure accurate results. It would also be desirable to provide automatic “trouble shooting” and maintenance of the instrumentation so as to eliminate signrdicant “down time” and loss of critical information.Considering the present status of automation, it is reasonable to expect significant developments along these lines during the 1980s. Perhaps this will be the decade of super analytical robots. If so, we need to consider their impact on an already revolutionary era of analytical chemistry. How intelligent and versatile will they be? Will they provide an economical work force th,at can efficiently handle the complex environmental and clinical bioanalytical problems ? Will they work intelligently on both research and routine laboratory measurements? How will they affect laboratory manpower requirements ? What type of training should student:; and present analytical chemists seek and receive so as to work effectively with them? To gain some insight into * Plenary Lecture presented at the 5th SAC International Conference on Analytical Chemistry, Lancaster, f Present address: Box YWAM, Kailua Kona, Hawaii 96740, USA.July 20-26th, 1980.RIALMSTADT 1019 these questions, I will first describe some of the microcomputer-controlled analytical systems that we have developed in our laboratories. These systems include the automated prepara- tion and transport of samples and reagents, encoding of the desired chemical information as digital electronic signals, optimisation and control of measurement conditions, data pro- cessing and the final display of chemical information, as illustrated in Fig.1. With these examples, it will be possible to illustrate several important automation concepts, to focus attention on present developments and speculate on what to expect in analytical instru- mentation during this decade of the 1980s. - - b Data domain conversion - - b Measurement Presentation of results Fig. 1. General block diagram for a chemical analyser Microcomputer-controlled Stopped-flow Analyser Sample and reagent preparation, transport and mixing operations are often more time consuming and costly than the encoding of the chemical information as electrical signals with subsequent data processing and display. Also, they are subject to frequent human error and bias. Therefore, a successful automated analyser must accomplish these operations simply, economically and reliably.An investigation of automated analysers indicates that the greatest differences are in the methods of sample and reagent preparation, transport, mixing and clean-up. One of the first approaches taken to automated sample and reagent handling and high throughput of measured samples was the air segmented continuous flow analyser (CFX) developed by Skeggs and the Technicon Corporation.l In the past few years, the unseg- mented flow injection analyser (FIA) that was developed independently by the R5iiEka2 and Stewart3 groups has become increasingly popular, and its characteristics are often compared with the CFA. More than a decade ago4?j we developed an analytical stopped- flow analyser (SFA) that has evolved into an increasingly useful instrument for routine or investigative high-speed analytical/clinical chemical determinations.The SFX has the high sample throughput of the CI:A and FIA and has some significant advantages. The reagent and sample are precisely measured and quantitatively mixed so that precisions of 0.1% are possible with the SF-4. Also, quantitative reaction rate methods that are often more specific than the equilibrium methods can be used readily, even for very fast reactions in the 1-s range. Our most recent microprocessor-based SFA6 is much simpler, inexpensive and more com- pact than its hardwired or minicomputer-based relative^.^-^^ The entire SFA system, shown in Fig. 2, is automated using an inexpensive Rockwell AlR'I65 microcomputer with self-contained keyboard, display, printer and ROnlS for control of all operations, communica- tion between modules, data acquisition and reduction, display and print-out of results. The compact sample handling and photometric modules and the typewriter-size micro- computer occupy only about 0.5 m2 of bench space.The programs are provided for specific applications using kinetic or equilibrium quantitative analytical procedures. This system requires no operator attention during normal use. For rapid reaction rate methods or with reactions that come to equilibrium rapidly, the SFA has a sample throughput of 400-500 samples per hour. This assumes pick-up of sample1020 MALMSTADT : ANALYTICAL Analyst, Vol. 105 Filter photometric system W Syringe drive system recorder Fig.2. Stopped-flow analyser (SFA) from an automated turntable. However, for reactions that are slow, requiring 30 s or more to react, the sample throughput of the regular SFA drops to 100 or less samples per hour. To improve the throughput for the SFA with slow reactions, we have introduced a storage coil between the mixer and observation cell, as illustrated in Fig. 2. We have shown11 that an unsegmented solution-storage technique is successful. The throughput of this new stopped-flow - unsegmented storage analyser (SF - USA) is excellent while retaining the previous advantages of the regular SFA. Present studies indicate that the SFA with a segmented solution-storage technique (SF - SSA) has certain advantages over the SF - USA.Quantitative results with the SF - SSA compare favourably with the regular SEA. The sampling - mixing device for the stopped-flow analysers is illustrated in Fig. 3. I t is based on an automatic dual syringe pipette - mixer that was developed in our laboratories. The aliquoting, transporting and mixing of solutions with a precision of 0.1% is readily accomplished. The high precision and sensitivity, excellent sample throughput and negligible start-up time should make the SFA, SF - USA and SF - SSA very popular during this decade. Improvements in design and even more compact instruments than the one illustrated in Addition of aliquots of Transfer to sample and reagent Mixing of solutions observation cell Observation cell --+ From C- reagent 14 t t Fig. 3.Sampling - mixing system for the SFA. ll From sample-November, 1980 IKSTRUMENTATION FOR THE 1980s 1021 Fig. 2 can be expected in the next few years. Plug-in programs for hundreds of specific procedures will make it possible to put these instruments into immediate use in most labora- tories. Pre-packaged reagents will also simplify the start-up procedure for specific methods. Rapid Sequential Analyser The schematic diagram of an SFA that can operate in a rapid sequential model2 is shown in Fig. 4. The chemical methodology can be changed automatically by a simple and rapid change in the position of a multi-port reagent valve and selection of an interference filter to isolate the desired wavelength for absorbance measurements. Analytical techniques based on kinetic, equilibrium or enzyme activity procedures can be specified by the operator through a keyboard.A unique reagent preparation system based on bulk solution reagents was developed13 and used with the rapid sequential analyser, and it is presented in the next section. Reference Light detector; Stopped-flow source beam splitter module Sample detector '_I Fig. 4. Rapid sequential analyser (S is the stopping mechanism; W is waste line; M is the mixer). Automated Solution-handling System Using the Mass of the Solutions An alternative to the automatic pipetting stations which accompany most automated discrete analysers is a computer-controlled solution-handling system based on mass, as illustrated in Fig. 5 . An electronic sensor is used to weigh accurately nominal aliquots of sample and reagent solutions that are added in small increments to a disposable beaker.Each reagent or sample is accurately weighed after addition and the beaker is then auto- matically moved to a stirring station while another beaker is moved into position for mass measurements. The amounts of reagents added to the beaker can be incrementally adjusted as desired. The continuous electronic feedback of mass information allows for automation as opposed to mechanisation. At present, samples equivalent to about 200 pl must be weighed for better than 0.5% accuracy, and about 1 ml for 0.1% accuracy. For clinical samples, it would be feasible to start with one larger sample which can be measured accurately (e.g., 200 11.1 of blood serum) and then dilute 10-100 times to provide a single solution from which nominal aliquots could be taken and weighed for all desired tests.The net amount of sample used for each test might be only 2-20~1, but each mass measurement involves larger nominal aliquots (such as 200 p1 or more) which can be determined more reliably. This procedure is also applicable for pipetting stations as well as electronic weighing stations and should always be considered for more accurate results when only small amounts of sample are available. The rapid feedback of information between the mass sensor and the microcomputer allows the controller to monitor as well as control the operations.13 It is not necessary to rely on the expected delivery of a pre-set amount from the pumping devices. The general utility of the mass sensor system results from the ability of the microcomputer controller to change1022 MALMSTADT : ANALYTICAL Awalyst, Vol.105 quickly, and without operator intervention, the solution delivery parameters, and then moni- tor those changes through mass sensor information. This system has considerable promise for SUPPlY Teletypewriter Fig. 5. Sample - reagent preparation using electronic inass sensor with microcomputer feedback. automated mechanistic studies of chemical reactions, and for preparing samples on a routine basis for standard-addition techniques. Relatively inexpensive commercial sample - reagent preparation instruments based on automatic mass, nominal digital reagent addition and microcomputer feedback could become available in the next few years.Automated System The combination of the automated mass sensor with the stopped-flow unit shown in Fig. 4 provides a completely automated system.9 Its step-wise operations are illustrated in Fig. 6. The operations are tlie same as for a classical spectrophotometric procedure, except each of the operations with this system is automated and rapid. On request by the operator at the keyboard (step 1) the monochromator and electronic balance are calibrated.13s14 The operating parameters for the quantitative determination of a specific chemical constituent are automatically set (steps 2 and 3). This includes the setting of the absorption wavelength, selection of the desired composite reagent and other parameters for optimisation of each procedure. One operation provides the spectropliotometric dark and 100% transmittance information for tlie blank solution.A second set of operations involves the standardisation of the analytical niethod against a set of standards. A third operation is the acquisition of absorbance data for a series of uiiknowii samples. The spectrophotometer is initialised (step 4A), by delivering a measured amount of diluent into an empty cup on the turntable. This cup is now stepped to the next position (step 5 ) , the solution is stirred (step 6), wliicli makes the tfiluent(s) available to the sample channel of the SFA. Aliquots of the blank and reagent are now drawn into the syringes (step 7) and then delivered through the mixer (step 8) into the observation cell (step 9). Steps 7 to 9 are repeated the desired number of times to flush the SFA, and then the dark and 100% transmittance data are acquired and stored (step 10).An empty cup on the turntable is weighed and a nominal amount of desired stock standard solution is delivered into the cup and weighed. A nominal amount of diluent is then added and weighed to Next, it is necessary to sequence through three procedures. The automated system now prepares a working curve (step 4B).November, 1980 IKSTRUhlENTATION FOR THE 1980s \ 250-ml Reagent resewoirs \, \ Aliquot sample and reagent \\ 8 ' 1-b -- t Mixer Observation cell / / / / \, hi, / Mix sample and reagent i 1023 Fig. 6. Automatic system combining mass sensor, nominal incremental pipettes and stopped-flow unit.1024 MALMSTADT : ANALYTICAL Analyst, Vol.105 provide an accurate standard based on the masses of standard and diluent added, The turntable is then stepped to the next position where the solution is stirred and made available to be aliquoted by the sample channel of the SFA. The SFA is flushed and steps 7-10 are repeated for the desired number of standards. Thus, a working curve is automatically prepared and stored in memory. Samples can now be analysed by proceeding to step 4C. A nominal aliquot of sample(s) is placed in a cup on the turntable, weighed, diluent added and again weighed to give a known dilution. The turntable is then indexed and steps 6 to 10 are repeated as for the standards. The unknowns are automatically measured, and the results that are based on the stored working curve are printed out.The selection of reagent requires less than 1 s, setting of parameters only about 5 s, the preparation of four working-curve standards about 2 min and a total time of about 3 min to obtain and store the working curve if the measurement time for the specific analytical procedure is only a few seconds. The throughput of samples will depend on the reaction time after mixing the reactants, and will be typically 100-500 samples per hour for both kinetic and equilibrium procedures. Rotating Disc Analysers The novel concept of using a spinning disc for simultaneously mixing multiple samples with a reagent and determining the concentration of each by nieasuring the absorbance or fluorescent radiation with a single beam photometric system was originated and developed a t Oak Ridge National Lab0rat0ries.l~ The instruments introduced a t Oak Ridge were known as centrifugal fast analysers, and nearly all of the early chemical methodologies were aimed at providing clinical tests for constituents in serum.Samples and reagents are pipetted into individual holding compartments in a disc. The discs typically have about 15-30 sample compartments and one reagent compartment associated with each sample compart- ment. \\'hen the disc is rotated at high speed, the sample and reagent from each pair of compartments are dumped through a mixing orifice into a photometric observation cell. Each of the sample cells passes successively past a photometric measurement system as the disc spins, and usually the photometric data from several rotations of the disc are averaged for each sample.The parallel mixing of many discrete samples and standards with rapid successive measurements provides for unique procedures and high precision. The small discs and compartments on miniature centrifugal analysers enable very small samples to be used with a significant reduction of reagent costs. At present the discrete loading of discs with sample and reagent are time consuming when utilising sequential loaders, but some work has been done on parallel loaders. Also, disposable discs pre-loaded with reagents are possible for some applications. N'e have modified the centrifugal analysers for improved precision16 and versatility1' for clinical analyses, but we have also utilised the rotating-disc concept for applications different from the original purpose.One of these applications is as a versatile automated multi- sample spectrophotometer in which the desired precision and resolution can be pre-set a t the keyboard. Another application is as a versatile multi-sample micro-colorimeter that has several advantages compared to ordinary absorption photometers that are used for quanti- tative chemical determinations of specific analytes. A multi-channel and multi-wavelength rotating-disc analyser (RDA) has also been developed that allows multi-tests as well as multi-samples on a single disc. Clinical profiles can be run on a single disc, and as many as 250-500 tests per hour could be performed with this system. Each of these three systems will be described because they illustrate certain concepts that should be generally applicable.Microcomputer-controlled Multi-sample Rotating-disc Colorimeter Advantages of this RDA instrument1* compared with ordinary colorimeters include the high precision that can be obtained, use of micro-samples without the problem of micro-scale air bubbles causing serious errors, measurement of many samples (easily 1040) within a few seconds, optimum use of available light source energy for shorter measurement times and automated movement of the disc into loading and measurement positions. The design features of the RDA are shown in Fig. 7. All of the steps described in this procedure can be performed rapidly.November, 1980 INSTRUMENTATION FOR THE 1980s Photomultiplier tube 1025 Photomultiplier housing Fig. 7 .Multi-sample rotating-disc colorimeter. The rotating disc unit moves into a measurement compartment as shown in Fig. 7 . The light source is focused so that a sharp, well defined beam passes through the solutions in the sample cuvettes. The filter wheel has been installed on the upper side of the measurement housing in front of the photomultiplier tube. A small motor drives the filter wheel under microcomputer control, but it does have an alternative manual control. Precisely fitted linear motion bearings (bushings) ensure a friction-free movement. A threaded drive shaft connected to a reversible motor is used to drive the centrifugal head. Microswitches operate to stop the movement when the head has reached the proper position inside the colorimeter housing for absorption measurements and outside for loading the discs.The disc drive motor M, the mirror MR and the double convex lens L of focal length 35 mm have been installed on an aluminium base B. A metal adaptor AD is permanently fastened on the motor shaft that holds the sample disc SD and the encoding ring R. The ring R is made from Plexiglas and has 12 encoding marks of black paint. The encoder and the photo- interruptor module P1 serve as the cuvette indexing pulse generator to indicate when each cuvette on the disc is in position for measurement. Another black mark on the same ring is similarly coupled with another photo-interruptor (not shown in Fig. 7) and generates the pulses for rotation indexing, The relative position of the filter FL on the filter wheel FW assembly when the centrifugal head is in the measurement position is also shown in Fig.7 A photomultiplier tube is used as the detector. The centrifugal head slides smoothly on two steel guide rods.1026 MALMSTADT : ANALYTICAL Analyst, Vol. 105 The rotating disc for this colorimeter has 12 sector-shaped observation cells with a path length of 1.00 cm as shown in Fig. 8. The sector shape is best for maximum collection of photon information for a given measurement period.le The body of the disc is made from a machinable glass - ceramic material, MACOR (Corning Co.) , that combines good machining properties with a good resistivity toward most chemical reagents. The windows for the multi-sample disc are made from two silica circular plates.Each plate has the necessary central hole, and the plates are cemented to the disc body with a silicone rubber adhesive. Details of the assembled disc, one of the sample fill ports, the relative position of the solution and light beam and the corresponding encoder are all illustrated in Fig. 8. \ Glass ceramic disc body V Fill port Silica plates Photo-interrupter Fig. 8. Disc design for colorimeter. Measurement Conceptsls To obtain accurate photometric measurements over a wide absorption range two sets of correction factors should be used: the transmittance correction factors, 71, and path length correction factors, $1. The first set is used to correct for small permanent differences in the transmittance properties between the reference cuvette and each sample cuvette.These differences are mainly due to small scratches, glass defects and internal reflections, affecting the measured light intensity for each cuvette loaded even with a non-absorbing solution. The transmittance correction factor T I for the sample cuvette i is calculated by equation (1). where T r and Ti are measured apparent transmittance values for the reference cuvette and sample cuvette i, both of them filled with a non-absorbing solution (e.g., water) in order to avoid the effect of small differences of the path lengths. As seen from equation (1) the transmittance of the reference cuvette is always equal to 1. Therefore, the corrected transmittance (Ti, ,,,.) of the sample cuvette is given by equation (2) :November, 1980 INSTRUMENTATION FOR THE 1980s 1027 Whereas equation (2) is valid for non-absorbing solutions, when the measured absorbance becomes higher another correction should be made to compensate the small differences of the path lengths. To compensate for the path length differences and the match of the sample cuvettes, one of them should be designated as the standard cuvette.The first cuvette (i = 1) is considered as the standard cuvette. Therefore, the path length correction factor is calcu- lated by equation (3) : where A, and Ai are the apparent absorbances of the cuvettes 1 and i, both of them being filled with the same absorbing solution. Therefore, the absorbance when normalised for path length is given by .. * * (3) fJi = AJAi - * . .. .. .. * (4) Ai, cor. = PiAi . . .. Combining equation (2) and equation (4), and using the known relation between transmittance and absorbance, the corrected absorbance is ..’ . (5) Ai, cor. = - pi log(7i Ti) . . . . It is noteworthy that if the centrifugal analyser is t o be used for kinetic methods of analysis, and the measured quantity is the first derivative of absorbance with respect to time, dAi/dt, the set of transmittance correction factors is not necessary. This can be shown by differentiating equation (5) and observing that p i and 7 1 are time independent. A charge to count converter, that gives a count proportional to the number of photons observed by the photocathode over a measured period of time, has been used in this work. The apparent transmittance for the sample cuvette i can be calculated with equation (6): where t i are the total converter counts (a number proportional to the number of photons observed by the photocathode) and the total observation time, for j successive rotations.The totals tr, as well as 7 n b and tb are the analogous quantities measured, respectively, during the time the beam passes through the reference cuvette and the “dark” cuvette. The use of a dark cuvette (a cuvette covered with black tape) allows a direct correction for any background output of the charge-counting system attributable to dark current of the photomultiplier tube and to offset any output of the charge to count ni and 3 and 3 3 3 3 converter. CombininE equations (5) and (6), we obtain the final equation used for the calculation of . . the correctevd a6sorbance values Automated Multi-channel Spectrophotometer with Rotating Sample Disc By using the rotating sample disc concept, a new type of multi-channel ultraviolet - visible spectrophotometer was developed.20 Simultaneous recording of ultraviolet - visible absorp- tion spectra for several samples (ten or more with special discs) can be readily obtained with the spectrophotometer.In addition to the multi-channel capability, the required sample volumes with the disc that we used are only 1OOp1, and the system also provides for pre- setting the desired photometric precision during the spectral scans. A precision of 0.000 1 to 0.1 absorbance unit can be requested from the keyboard. A microcomputer can control the parameters necessary to attain the pre-set precision, trading longer measurement time for higher precision as required by photon statistical considerations.The block diagram of the system is shown in Fig. 9.1028 MALMSTADT : ANALYTICAL Analyst, VoL 105 Fig. 9. Block diagram of automated multi-channel spectrophotometer. By mechanically designing the rotating sample disc system to fit on the optical axis of a modular commercial spectrophotometer, it has been possible to develop a computer-controlled ratio recording ultraviolet - visible spectrophotometer that can simultaneously produce spectral absorption curves for many micro-samples. The automated spectrophotometer allows the operator to select one of two modes of wavelength sequencing. In the scanning mode, the monochromator starts a t one wavelength and obtains and stores all of the reference and sample information needed.Similar data are then acquired at pre-set wavelength increments (e.g., 0.01-10 nm steps) throughout the specified spectral range. In the discrete mode, individual wavelengths of analytical interest are selected by the operator, and the monochromator slews to each of these wavelengths sequentially. In both modes of opera- tion, the grating is stationary during the spectrophotometric measurement so that true absorbance values can be measured when scanning across sharp peaks. The optical encoder for the wavelength drive on the monochromator used in this instrument makes it readily feasible to place the wavelength selection under computer control, Another important feature of the spectrophotometer is the automated, real-time adjust- ment of instrumental conditions to obtain the level of precision desired by the operator.This type of control was originally demonstrated for photon counting systems.21 An analogue output of the photomultiplier tube detector is used in this instrument instead of counting photons, but it is found that with careful design, photon statistical errors are still the only major source of imprecision. The operator may specify the desired precision as either a coefficient of variation in absorbance, or a standard deviation in absorbance, depending upon the application; the computer varies the measurement time a t each wavelength in order to achieve the desired precision. It is believed that this feature will be included on future commercial spectrophotometers. To implement the computer-control functions requires interfaces to the monochromator and to the photomultiplier power supply, an analogue to digital converter system, and the software routines for photometric precision control, wavelength selection and digital data manipulation and storage.The output information is recorded by either a chart recorder or a line printer. Graphs of precision (expressed as percentage relative standard deviations of the measured absorbance values) aeysus the absorbance19 a t various numbers of averaged rotations from R = 1 to R = 1000 are shown in Fig. 10. The theoretical curves that are based on photon statistics are shown as the solid lines. They illustrate that for every 10-fold increase in the number of rotations a t constant speed, there is a dm increase in precision.It can be seen in Fig. 10 that the experimental points are quite close to the theoretical solid lines. ThisNovember, 1980 INSTRUMENTATION FOR THE 1980s 1029 0.01 I , I 0 1 2 Absorbance Fig. 10. Per cent. relative standard devia- tion as function of disc revolutions (at constant speed). indicates that the rotating-disc system follows photon statistical considerations, a t least down to relative standard deviations of about 0.02% for 1000 revolutions (about 30 s total elapsed time) but only 1-2 s measurement time for each cuvette. When the number of revolutions is increased to 10000, the precision improves to about 0.01% relative standard deviation, which indicates that other factors are causing variation in the absorbance measure- ments.I t can be observed in Fig. 10 that the percentage relative standard deviation for any given number of revolutions does not change much from about 0.3 to 2 absorbance units, which is a point previously emphasised,16t22 but often neglected. Spectra for 5 of the 15 solutions of phenol red indicator in buffers of varying pH that were run simultaneously on a single disc are shown in Fig. 11. Other applications include comparison of filter spectra by mounting the filters directly on the disc holder and multi- component spectra. Multi-channel and Multi-wavelength Relative Standard Deviations for Multi-tests and Multi-samples A rotating disc instrument was designedz3 so that several different wavelengths of radiation could be observed on each revolution of the disc.The unit has six different wavelength channels, five for absorbance measurements and one fluorescence. This provides the necessary wavelengths so that 20 of the common clinical chemistry tests can be run on each sample. Thus, stat measurements or routine profiles can be obtained on the analyser within 15 min of receipt of the sample. Also, single chemistry tests can be run on multiple samples in a batch mode of operation. An 8080 microcomputer systemz4 was used for all functions of the instrument. Conclusions Many of the classical analytical measurement techniques and chemical methodologies can be significantly improved by modern microelectronic technology. We have shown how spectrophotometers, colorimeters, stopped-flow analysers for kinetic and equilibrium chemical methods, sample and reagent preparation units and other systems can provide better precision, accuracy, sensitivity and selectivity by re-designing with low-cost microcomputers.Although there have been dramatic improvements in the past few years, there should be1030 MALMSTADT : ANALYTICAL Analyst, VoZ. 105 2 I/ al C m e l SI n a 0 300 400 500 600 Wavelength/nm Fig. 11. Absorption spectra obtained simultaneously at 5 pH values: A , 11 ; B, 8.1 ; C, 7.7 ; D, 7 . 0 ; and E, 4.0. even more significant developments during the 1980s, especially as more instrument designers and analytical chemists gain experience and a thorough understanding of microcomputers. The introduction of low-cost flow injection and stopped-flow analysers will improve the output and relability of analytical results from many laboratories.We can expect that major improvements and applications of these analysers will appear regularly on the market. Also, the development of hybrid systems will probably provide unique and useful analytical features. Automated sampling and the preparation of samples and reagents for quantitative deter- minations should be on the threshold of major advances. Perhaps, the automated mass sensor with incremental pipette and microcomputer feedback that was described in this paper will be refined so that it will become commonplace for routine standard-addition and other analytical techniques. In all instruments, there is a need for automated calibration and self-diagnosis of problems.The addition of these features in the new designs would improve the quality of results from many laboratories. There will be many surprises as breakthroughs in several areas of technology during the 1980s combine to provide powerful new automated analytical tools. I t seems inevitable that nearly all analytical laboratories will become highly automated during this decade. It is certainly important that courses and conferences be provided to train and re-train chemists as they are required to work in an increasingly automated environment. The challenge to educators is very clear. References 1. 2. 3. 4 . 5. 6 . 7. Skeggs, L. T., Am. J . CZin. Patlzol., 1957, 13, 451. Rfitibka, J., and Hansen, E. H., Anal. Chim. Ada, 1975, 78, 145. Stewart, K. K., Beecher, G. R., and Hare, P. E , Fed. Proc. Fed. Am. SOC. Exp. Biol.. 1974, 33, Javier, A . C., Crouch, S. R., and Malmstadt, H. V., Anal. Chem., l!)69, 41, 239. Malmstadt, H. V., Cordos, E. A., and Delaney, C. J.. Anal. Chem., 1972, 44, 26A. Koupparis, M. A . , Walczak, K. M., and Malmstadt, H. V., J . Autom. Cizem., 1980, 2, 66. O’Keefe, K. R., and Malmstadt, H. V., Anal. Chem., 1975, 74, 707. 1439.November, 1980 INSTRUMENTATION FOR THE 1980s 1031 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Krottinger, D. L., McCracken, M. S., and Malmstadt, H. V., Talanta, 1979, 26, 549. Krottinger, D. L., McCracken, M. S., and Malmstadt, H. V., Am. Lab.. 1977, 9(3), 51. Krottinger, D. I,., McCracken, M. S., and Malmstadt, H. V., J . Autom. Chem., 1978, 1, 15. Malmstadt, H . V., Walczak, K. M., and Koupparis, M. A., Am. Lab., 1980, 12, 27. Malmstadt, H. V., “Topics in Automated Chemical Analysis,” Ellis Horwood, London, 1979, p. 95, Renoe, B. W., O’Keefe, K . R., and Malmstadt, H . V., Anal. Chem., 1976, 48, 661. Spillman, R . W., and Malmstadt, H. V., Anal. Chem., 1976, 48, 303. Anderson, N. G., Am. J . Clan. Pathol., 1970, 53, 778. Avery, J. P., Gregory, R. P., IV, Renoe, B. W., Woodruff, T., and Malmstadt, H. V., Clin. Chem., Renoe, B. W., Gregory, R. P., IV, Avery, J . P., and Malmstadt, H. V., Clin. Chem., 1974, 20, 955. Efstathiou, C. E . , and Malmstadt, H. V., Am. Lab., 1979, 11, 19. Efstathiou, C. E., Cordos, E., and Malmstadt, H . V., Anal. Chem., 1979, 51, 58. Avery, J . P., and Malmstadt, H. V., Anal. Chem., 1976, 48, 1308. Timmer, R . B., and Malmstadt, H. V., Am. Lab., 1972, 4(9), 43. Malmstadt, H. V., Franklin, M. L., and Horlick, G., Anal. Chem., 1978, 44, 63A. Wengert, G. R., Jr., PhD Thesis, University of Illinois, 1979. Avery, J . P., PhD Thesis, University of Illinois, 1978. 1974, 20, 942.

ISSN:0003-2654    

DOI:10.1039/AN9800501018

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

1032 Analyst November 1980 Vol. 105 $9. 1032-1044 Plasma Spectroscopy Comes of Age* S. Greenfield 37 Fiery Hill Road Barnt Green Birmingham R45 8LE A “state of the art” review of ICP - OES is given in which the author attempts to give an impartial account of the technique with respect to detection limits, precision interferences and the low power - high power controversy. Keywords Inductively coupled p l a s m a ; detection l i m i t s ; precision; inter-ferences; high- and lowpowered plasmas The age of majority in Great Britain is now 18 years. At that age a person is considered to have reached a level of responsibility sufficient to fit him or her for a place in society as a respectable adult. Eighteen years ago experiments were started which were to lead to the production of an exciting new source an inductively coupled annular plasma which in turn led to the renaissance of emission spectroscopy; plasma spectroscopy as it became known.Today there are some 600-700 of these systems in use throughout the world and their number is growing fast Truly the technique of plasma spectroscopy can be said to have achieved respectability. I t is interesting to look a t the circles to which this once precocious infant is now admitted. In the UK plasma systems are in daily use in such prestigious establishments as the Labora-tory of the Government Chemist the Institute of Geological Sciences Metropolitan Police Laboratories Water Research Centre Macaulay Institute and the UKAEA. Users in the USA include the National Bureau of Standards the Aerospace Corporation and the Montana Bureau of Mines; in Sweden the Geological Survey and the Swedish Institute for Metal Research; in Israel the Nuclear Research Centre; and in Germany the Spectrochemical Institute a t Dortmund; South Africa makes extensive use of ICI’s in The National Institute for Metallurgy and at CSRI.In the UK there are the British Steel Corporation British Rail at Derby British Drug Houses a t Poole and of course, Albright & Wilson at Oldbury. Many more examples could be cited. ICP - OES has been used to analyse geological samples ferrous and non-ferrous alloys precious metals and rare earths marine sediments, organisms and coastal sea water oils for fuels foodstuffs urine beer and spirits animal feeds animal wastes clinical samples forensic samples such as tissue glass and steel-again there are many more examples; any material which can be taken into solution can be analysed for a large number of elements by ICP - OES.In fact detection limits have been determined by ICP for nearly 80 elements and whilst on the subject of detection limits it might be interesting to compare those obtained by ICP - OES with those obtained by atomic absorption (AA). In the early days of atomic absorption it was pointed out that there are always more atoms in the ground state ready to absorb than there are in an excited energy level ready to emit. Absorption techniques therefore might be expected to be more sensitive and a t ordinary temperatures so much so that regardless of the noise characteristics the detection limits in absorption would always be much better than in emission.However a large absorption signal depends not only on the number of atoms ready to absorb but also on the intensity of the incident beam which is to be absorbed. You cannot absorb more than is there. The dependence on the incident beam destroys the simplicity and generality of the argument on sensitivity and hence on detection limit and indeed it was found in practice It has come of age. An equally prestigious list can be cited for industry. In Germany we have the Hoesch Huttenwerke AG. Equally long is the list of applications. * Plenary Lecture presented at the 5th SAC International Conference on -4nalytical Chemistry Lancaster, July 20-26th 1980 GREENFIELD 1033 that for many elements the detection limits found in flame atomic absorption were very similar to those found in flame emission.The detection limits published by Christian and Feldmanl in 1971 for the two techniques are plotted on log - log paper. The principal diagonal shows where the two detection limits are equal. Those elements above this line have a greater that is worse detection limit in flame emission than in flame absorption while those below the line have a smaller that is better detection limit in emission than in absorption. For the 66 elements compared 29 have lower limits in AA 34 have lower limits in emission and 3 are identical. The faint lines are contours representing factors of ten so that zinc for instance near the top of the figure and lying on the fourth contour above the locus of equality has therefore a detection limit in emission lo4 times that in absorption.Similarly for Se again a t the top of the figure and lying on the third contour the ratio is 103. We can sum up by saying that the detection limits are better for more elements in emission but of the elements which are better in absorption some are very much better. Many AA practitioners seem to think that those obtained in plasma emission are also similar. This is not so. Recent promotional literature from a prominent instrument manu-facturer includes a convenient table of AA detection limits which are similar with perhaps a modest over-all improvement to those published by Christian and Feldmanl in 1971 and by Berman2 in 1975. These are compared with ICP detection limits taken from Boumans and Barnes' 1975 collcction3 in Fig.2. Of the 59 elements compared potassium is notably better in absorption with a factor of 30 ruthenium and bismuth just better with factors of 1.5 and 1.25 and indium lead and silver are equal. The other 53 elements are all better in emission with factors going up to 2000 for calcium and 5000 for boron. One might have thought that in Fig. 1 the two This is illustrated in Fig. 1. 100 000 10 000 1000 100 4 10 2 $ 1 - k r' ._ w .-al c > 0.1 2 0.01 n * -2 w 0.001 0.0001 0.00001 0.0001 0,001 0.01 0.1 1 10 100 1000 10000 100000 Result by flame atomic absorption p,p.b. Fig 1. emission. Comparison of detection limits in flame atomjc absorption and flam 1034 GREENFIELD PLASMA Alzalyst Vol.105 techniques of flame emission and flame AA were more or less equally balanced. I t is now apparent in Fig. 2 that the use of a plasma source in emission has swung the balance greatly in favour of emission. At one time when flame emission was being compared with flame absorption it was thought that lines with short wavelengths might give better results in absorption and those with long wavelengths in emission. This was because for lines whose lower energy level is the ground state the energy of the upper level is proportional to l / h . Thus short-wavelength lines are more difficult to excite and their sensitivities in emission might be low. This does not seem to be a valid argument in plasma emission since the worst line for the ICP is potassium at 766.5 nm.The reason is almost certainly that for very easily excited lines the plasma is too hot. There can be no doubt that for the great mijority of cases plasma spectrometry gives lower detection limits than flame AA and by factors that are often large. However when AA is used with non-flame atomisation exceedingly low detection limits can be obtained. In methods of this kind discrete samples are u!jed and the signal measured is a transient as distinct from the more common type of method we have been considering where a solution is continuously fed to the source and a steady-state signal is measured, For this the appropriate unit for detection limit is concentration while for discrete sampling the mass of analyte is more commonly used. In comparisons between the two types some assumption must be made about the amount of sample used.Frequently a sample volume of 1 ml is postulated for solution analysis so that an absolute detection limit of 10-9 g would correspond to a detection limit of Considerations of this kind make exact comparisons difficult; however data due to L’VOV~ indicate that detection limits for AA with graphite tube atomisation are much superior to those obtained with conventional flame atomisation. The factors by which they are g ml-l that is a concentration of 1 ng ml-l. 0.000 01 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000 Result by flame atomic absorption p.p.b. Fig. 2. Comparison of detection limits in flame atomic a’bsorption and ICP emission with conventional nebulisation November 1980 SPECTROSCOPY COMES OF AGE 1035 improved range from 80 (for potassium) to 2 x lo6 (for silicon) with most in the range 10s-105.A comparison with ICP spectrometry with conventional nebulisation is shown in Fig. 3. However it is likely that L’vov’s very low limits would be difficult to obtain in a routine analytical laboratory and the careful work of Chakrabarti and co-workers5 confirms this. Using a commercially available carbon rod atomiser they found detection limits typically an order of magnitude higher than L’vov’s. Electrothermal atomisation can also be applied to plasma spectrometry and was first so applied by Kleinmann and Svoboda,6 who used a graphite disc. Nixon et ale7 used a tantalum filament to vaporise their samples and for 100-pl samples reported detection limits which are smaller than the present-day values for continuous pneumatic nebulisation ; when their values are transformed to correspond to 1-ml samples the factors by which they are superior range from 2 to 2000 most being of the order of tens.In Fig. 4 these values are compared with L’vov’s for graphite tube atomisation. They are mostly rather worse although if we apply Chakrabarti’s factor of 10 to L’vov’s values (which would effectively make the contour marked x 10 the locus of equality) there is little to choose. Kirkbright’s schools used a graphite rod for vaporisation and Fig. 5 compares their results with L’vov’s values. Again if Chakrabarti’s factor of 10 is applied there is not a great deal to choose. Actually this comparison is rather unfair to Kirkbright’s results as mentioned earlier if one is com-paring results from transient signals with those from steady-state signals one has to make an assumption about the volume of sample used when a solution is fed continuously to the excitation source.This disguises the fact that Kirkbright was using lo-$ samples and Fassel 100-pl samples so that in absolute terms that is with the detection limit in grams Kirkbright’s results are in most instances superior to Fassel’s. The swing to AA is impressive. Commonly the value taken is 1 ml and this has been done here. 100 000 10 000 1 000 100 4 2 10 c‘ .- In ._ E l n-r u I 0.1 a R 3 a -0.01 0.001 0.000 1 0,00001 0,000 1 0.001 0.01 0.1 1 10 100 1000 10000 100000 Result by graphite tube atomic absorption p.p,b.Fig. 3. Comparison of detection limits in graphite tube atomic absorption (L‘vov) and ICP emission with conventional nebulisation 1036 GREENFIELD PLASMA AIzalyst Vol. 105 I I V J J . 0s 0.000 01 0.000 01 0.0001 0,001 0.01 0.1 1 10 100 1000 10000 100000 Result by graphite tube atomic absorption p.p.b, Fig. 4. Comparison of detection limits in graphite tube atomic absorption (L’vov) and ICP emission with tantalum filament atomisation. Kirkbright also reportedg that for elements which form refractory carbides or oxides and which are therefore difficult to remove completely from the atomiser the detection limit is much improved typically by a factor of one or two orders of magnitude if a small amount of a halocarbon is mixed with the argon which carries the sample from the atomiser.The mechanism postulated for this is the preferential formation of halides which precludes the possibility of the formation of carbides at high temperatures. However since similar tech-niques are applicable to AA as well this should not be thought. of as necessarily favouring the ICP. Recently there have been two reports by Salin and HorlicklO and by Ohls and Sommer,ll of a technique where the sample is placed in a graphite cup which is then inserted into the plasma. An improvement in the detection limit of two orders of magnitude is reported by Ohls. This with Fassel’s and Kirkbright’s results gives us reason for expecting the improve-ments in detection limits when electrothermal vaporisation is used with an ICP to result in values which may well be comparable to those obtained in practical applications of graphite furnace AA.Another technique used in AA to improve the detection limits of the set of difficult elements arsenic antimony bismuth selenium tellurium germanium and tin is to convert the analyte element into its gaseous hydride. When the sample is introduced as a gas rather than as a solution requiring to be nebulised the large losses associated with aerosol formation are eliminated and so a much larger fraction of the analyte is useful. This technique has been used in plasma emission spectrometry also and detection limits of 1 p.p.b. (parts per l o p ) or better have been reported.12 In Fig. 6 it can be seen that the values for the two techniques are similar the score being 3 to 2 in favour of ICP with two draws.Hydrogen water vapour and carbon dioxide are produced as by-products of the hydride formation and some plasmas have inadequate power to cope with these. Systems have been designed to eliminate these by-products and at the same time to collect the hydride and feed it to the plasma as a concentrated plug. With such a system an improvement in the detection limit for arsenic by two orders of magnitude has been reported.13 Without playing the numbers game too hard it is probably fair to say that detection limits for ICP and AA are similar when hydride generation is used. When one reads of detection limits of fractions of a part per billion one sometimes wonders how meaningful these small numbers are in practice.There are two ways of looking at this November 1980 SPECTROSCOPY COMES OF AGE 1037 100 000 10 000 1000 n 100 2 d g 10 P L a 1 n U .-A= g 0.1 > n - - 0.01 a k? = 0,001 0.000 1 0.00001 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000 Result by graphite tube atomic absorption p.p.b. Fig. 5. Comparison of detection limits in graphite tube atomic absorption and ICP emission with graphite rod atomisation. The estimated population of the world in 1975 was 4 x 109 so that each of us represents 114 p.p.b. of the world population in terms of heads. (No doubt if we were interested in concentrations in terms of weight my contribution would be a little higher!). In any case you might think that a visitor from outer space searching for traces of Greenfield might think himself very lucky if he found him at all.The other view depends on the very large numbers of atoms involved. In a gas a t room temperature there are about lOI9 particles per millilitre so that 1 p.p.b. will correspond to 1010 particles in a 1-ml sample. Thus if the world’s population is atomised (as it may well be) then given time for the system to come to equilibrium by mixing it is exceedingly likely that a few atoms of Greenfield may be found after all. Although considerations of this kind support the respectability of very low detection limits sometimes one may enquire how generally useful these are in practical analysis. In April 1978 there was an ICP conference a t Noordvijk in The Netherlands and I was asked to introduce a session on “Analytical Performance-Limits of Detection Precision and Stability.” I opened my comments by saying perhaps rather abrasively that “the present league table of detection limits must be extremely valuable to those people who spend their lives analysing distilled water.” I then went on to suggest a different approach, namely to carry out detection limit determinations in a variety of matrices which were to be chosen so as to be likely to introduce difficulties of different kinds.I suggested a 10% solution of sodium chloride to start with. With such a solution many nebulisers run under conventional conditions will block up. If so the solution must be diluted by a factor x , say and when the detection limit of a trace element is determined in the diluted solution it must be multiplied by this factor x to give a true value of the detection limit in the 10% solution.Thus if a solution has to be diluted for any reason the detection limit will suffer compared with the idealised case of a dilute aqueous solution requiring no dilution. I also suggested using a 10% solution of calcium carbonate as a matrix the idea being to test how detection limits depend on a high level of stray light. I also suggested solutions of iron or chromium or nickel so that problems of spectral interference could be looked at. Finally I suggested a matrix of olive oil knowing that organic compounds sometimes cause undesirable effects. These suggestions were made because it seemed to me that the marvellous values of detection limit which were appearing in the literature were becoming ends in themselves an 1038 GREENFIELD PLASMA Analyst Vol.105 ~ 000001 0,0001 0.001 0.01 0.1 1 10 100 1000 10000 100000 Result by atomic absorption p p.b. Fig. 6. Comparison of detection limits in atomic absorption and ICP emission when the sample is introduced as hydride. that the true objective of spectrochemical analysis was becoming almost forgotten. This objective is of course the analysis of samples of interest to the client. With this in mind let us turn to a topic that undoubtedly is of major importance in any kind of chemical analysis namely precision. While there are laboratories in which the determination of ultratrace amounts with moderate or poor precision is of importance it seems to me that the determination with high precision of high concentrations is a problem of milch wider application.Think for instance of assay determinations of costly metals in ores. Many workers have found by experiment on ICP systems that the uncertainty in a measurement expressed as its standard deviation is proportional to the magnitude of the signal so that a constant relative standard deviation (or RSD) is obtained over a wide range. This is found not only if the signal is varied by varying the concentration of analyte but also if the integration time is varied with the surprising conclusion tha,t provided the signal is not ridiculously small a short integration gives as precise a value as a long one. This point will not be elaborated here except to say that it is probably due to the signal drifting rather than being scattered randomly about a mean.However the result that the RSD of the gross signal is constant can be used quite simply to relate the RSD of the net signal to the signal to background ratio. The RSD of the net signal which is often also the RSD of the analysis is shown in Fig. 7 as a function of the signal to background ratio xlb with the assumption that the gross signal has a constant RSD of 2%. When the signal to background ratio is much greater than 1 the RSD of the analysis approaches 2%; as the ratio approaches 1 the RSD increases rapidly until when the ratio is 1 the RSD goes to infinity as here the net signal is zero the gross signal consisting entirely of background. It is not difficult to show that if the detection limit is defined as the con-centration corresponding to a net signal which is 2 f i times the standard deviation of the background it corresponds to an RSD of 50% as shown in the figure.Now for some purposes an RSD of 50% is fine although some people would need con-vincing of this. For instance if a specification demands less than 10 p.p.b. of an element and it is determined as 5 p.p.b. the chances of it genuinely passing are 19 out of 20 which may be adequate. In other circumstances such poor precision might be intolerable November 1980 SPECTROSCOPY COMES OF AGE 1039 u-D v) a 1 “ 10 100 “DL Log 7 , Fig. 7 . Relative standard deviation of net signal (x-b) It is assumed that the relative Precision calculated for as a function of x / b . standard deviation of x is 2%.u(x) = 0.02.7. DL = detection limit. The value of 20/0 RSD is used in the calculations here because for some years this was a typical value obtainable at concentrations well above the detection limit. For the majority of purposes this precision is adequate but not for assay work. As already mentioned, increased precision is not obtainable by increasing the integration time and unless the time scale of the drift is taken into account replicate measurements do not help either. Thus, the way to achieve higher precision is to take steps to reduce the 2% RSD of the gross signal. They were interested in assays of feed materials for nickel and cobalt refineries with an annual value of one billion dollars. Wet chemical analysis can give results with an RSD of about 0.1% when conducted with the utmost care while more common routine results have an RSD of about 1%.They therefore tried to match these figures for wet analysis with an ICP spectrometer. Using the commercial system as delivered they found it impossible to control drifts in the calibration of perhaps 10% in an hour or to maintain precision under 1% RSD. They investigated the problem and the first part of the apparatus to come under scrutiny was the nebuliser. The nebuliser supplied required frequent adjustment. They therefore tried a non-adjustable type which gave improved stability against drift but poorer short-term precision. After observing that the thin glass walls vibrated they designed and patented a non-adjustable model with thick walls and incorporated a baffle in the expansion chamber.With this they were able to reduce the RSD to 0.52y0 or if an internal standard were used to 0.33%. In addition the calibration drift over several hours was reduced to about 1% under normal circumstances. During further investigation it was noticed that the signal was strongly temperature dependent. This was traced to the gas flows so mass flow-controllers were installed. Finally the effect of fluctuations in the power was investigated. The effect was different for different elements in some instances the signal rose with power in some it fell and for some the effect was minimal. For zinc the worst example an apparent change in con-centration of 0.01 p.p.m. was caused by a change of only 0.29 W which is about 0.03% of their applied power.Thus if a precision of 0.1% at concentration levels of 10 p.p.m. is required a very stable generator is necessary. When these precautions were taken precisions of 0.1-0.4~0 RSD were obtainable with ideal aqueous solutions with concentration 10 p.p.m. When real samples were used the values for nickel were 0.4% as compared with 0.25% for pure solutions. Thus provided sufficient care is taken with the nebuliser design and with effective controls for the gas flows and the power supply it appears that RSD values of less than 0.5% even at levels of 10 p.p.m. can be achieved with a “reasonable prospect,” as the authors say of reducing this to 0.1%. Up to this point I have been describing the attributes of plasma emission spectroscopy from the evidence of factual data and well documented experiments; I am now going on to discuss the interferences to which the technique is subject.Whilst a number of these inter-ferences are also well documented others are not and for these the evidence is less conclusive, often partially contradictory and sometimes open to a contrary interpretation. A recent report by Meddings et al.14 is concerned with this problem. These values are means for 25 elements 1040 GREENFIELD PLASMA Afialyst Vol. 105 There is no doubt that the technique like any other form of emission spectroscopy is subject to spectral interferences. These are of the usual kind such as instrument broadening of the spectral lines leading to spectral overlap of close packed spectral lines and also of molecular band spectra.In addition to these the high temperature of the plasma causes the introduction of a few extra spectral interferences. Thus the broadening of the lines as a consequence of the high temperature causes strong lines to have “wings” that may overlap lines of interest even when these are 0.1 nm or so away. Some of these strong lines result in a magnification of the problem of stray light and necessitate great care to be taken in the designing and making of tLe apectrometer. Radiative recombination which occurs when a free electron combines with an ion to yield a neutral atom causes emission of a continuum, which can also cause spectral interference. Although it is important that the spectroscopist is aware of these effects they in no way invalidate the technique and proper selection of lines and operating conditions and the use, where necessary of background corrections will nullify these interferences.Another interference effect for which there is substantial evidence is that due to the effect of viscosity and other solution properties of the analyte on the uptake rate of the nebuliser. The effect of this interference can be and often is mistakenly attributed to other causes. It can be minimised in a number of ways by feeding the nebuliser with a pump by matching the solution properties of the standards to those of the sample by mathematical correction for the solution properties and by the use of internal standards. There is little if any evidence for another type of interference chemical interference associated with the formation of stable refractory compounds.Compound formation is precluded by the high kinetic temperature of the plasma. Such evidence as there is for compound formation arises in the main from experiments conducted with torches and injectors producing solid or non-annular plasmas or from torches operating under eccentric conditions. It must be presumed that under these circumstances the analyte does not reach a temperature high enough to dissociate refractory compounds. Enhancements of the signal which have been reported are of course due to other causes and are not due to compound formation which is always a depressive effect. In passing it is worth mentioning that there is little if any evidence for anion interference. It might be expected that the influence of the addition of an easily ionised substance such as sodium to the analytical matrix would be to produce ionisation interference.In this instance one might expect the emission from atomic lines to be enhanced and that from ionic lines to be depressed as the increase in electron density perturbs the equilibrium. In fact although some of the evidence would seem to support this expectation some does not. For instance Abdallah et al.15 traced the enhancement of the 253.5-nm atomic line of phosphorus to a desolvation effect. The effect varied with the molecular state of the phosphorus and with the temperature of the desolvation oven and it was absent when the desolvator was omitted. These same authors and also Fassel and his colleagues found an enhancement of the atom line of calcium and a reduction of the ion line with increasing amounts of sodium.It is interesting to note that they also found that the effect disappeared at high powers that is we presume at higher temperatures. As the two sets of experi-menters were using two very different types of nebuliser the indication was that the inter-ference lay within the plasma and could be consistent with a variation in electron density. However as Mermet and Robinls pointed out in an earlier paper experimentally sodium did not appear to have an appreciable effect on the electron density or the excitation temperature. Abdallah et aL15 confirmed these findings by measuring the ratio of an atom and an ion line of magnesium and the ratio of two ion lines of titanium with increasing sodium concentration.The first ratio gives a measure of the electron density and the second the excitation tempera-ture. They found that the electron density and the excitation temperature remained practically constant. From these results they concluded that Saha’s equation could not be applied and suggested that the interference was an atomisation interference. The authors conclude their paper with the comment “what these interferences stem from and the means of freeing oneself from them remains to be deeply studied.” We at Oldbury have calculated the electron density in an argon plasma as a function of sodium concentration at various temperatures and these are shown in Fig. 8. This shows that when a sufficiently low concentration of sodium is introduced even if it is highly ionised its contribution to the electron density is negligible compared with tha November 1980 SPECTROSCOPY CONES OF AGE 7 I 10'3 1 ~ 1 r 7 5 0 0 0 c T I , x-x-x-x-x-x-x 1041 10-13 10 11 10-9 10-7 10-5 10-3 Partial pressure of Na in Ar Electron density in Na -Ar plasma as a function of sodium concentration.Fig. 8. due to argon. The figure also shows that the higher the temperature the greater is the con-centration of sodium before the electron density shows a detectable increase. From this it might be concluded that sodium should not affect the electron density unless considerable amounts are present or the temperature of the system is low. Several workers have suggested that a matrix such as sodium can coat the analyte particles causing a change in the rate of volatilisation and thus accounting for the different optimum observation zones which are found in the presence of sodium.From this it follows that if the system has been optimised for shall we say detection limits in the absence of a matrix, then it will not be optimised for detection limits in the presence of a matrix. Put another way the system will not be optimised for minimum interference. Brockaert et ~ 1 . ~ 7 following an examination of the radial distribution of excitation tempera-tures found that sodium metaborate reduced the excitation temperature of the central zone of the plasma. Kornblum and de Galan18 found that the uptake rate of the nebuliser also altered the temperature of the central zone and hence the emission.Brockaert et a1.17 found that the metaborate reduced the intensity of yttrium lines whereas sodium chloride did not. As the sum of the ionisation energy of the yttrium and the excitation energy of the lines is close to the energies of the argon metastables they concluded that the decrease in excitation of the yttrium was due to the quenching of the metastables by the borate. This of course, assumes that excitation is via argon metastables. Perhaps for clarification purposes I can summarise what I see as the present position with regard to interferences in ICP - OES : (i) There are spectral interferences. (ii) Annular plasmas do not exhibit chemical matrix effects. (iii) Solution effects can alter the uptake rate of nebulisers. ( i v ) There is evidence that desolvation effects do occur.( v ) Some effects which occur in the presence of sodium may be ionisation suppression, others may be caused by changes in excitation temperature others may be volatili-sation effects, ( v i ) If a system has been optimised for detection limits it will not necessarily be optimised for minimum interference 1042 GREENFIELD PLASMA Analyst voz. 105 Thus it can be seen that our knowledge of the interferences that occur in plasma spectro-scopy is not complete and there remain anomalies that are not easily explained. It is indeed fortunate that the effect of these interferences is small. For instance Larson et a1.19 com-mented that an easily excited element sodium in concentrations of up to 6900 p.p.m., exerted an unusually low influence on the observed emission intensities of three selected elements calcium chromium and cadmium.Boumans and de Boer2(' observed that con-comitants a t concentrations of 1000 p.p.m. had small effects. Kornblum and de Galan16 found that interferences under normal conditions were so small that they were lost in the imprecision encountered in Abel inversions and so they had to iidopt unusual conditions to make them bad enough to be measured reliably. A similar controversial area is that concerning the power rating of the high-frequency generator used to produce the plasma. The number of watts put into the plasma will affect not only its temperature but also its size particularly its length and both these properties will affect the temperature of the analyte stream flowing through it.From this it is easy to see that power supplied to the plasma will affect the detection limit. This is so because any variation in temperature will affect the spectral radiance of the analyte signal and also that of the background or continuum and a t different rates. Because detection limit is related to both signal and to background it too will be affected. The temperature will also affect the electron density of the plasma and of the analyte stream. It can be imagined that if the electron density is very lugh to start with as it will be if the temperature is high then the addition of an easily ionised material may be expected to have less effect than it would in a low-powered shall we say < 1.5-kW plasmasystem, which will have a lower temperature.As is seen in Fig. 8 calculations show that the higher the temperature the greater is the concentration of sodium which can be tolerated before the electron density is significantly increased. Again it can be demonstrated that higher power in the plasma will dissociate molecules which are not dissociated in lower power plasmas. A classic example is that of the so-called Swan bands or carbon - carbon banding which undoubtedly are present when organic solvents are introduced into low-powered plasmas but are not present in high-powered plasmas as Figs. 9 and 10 show. There is reason to believe that the plasma resembles a tube furnace and normally it can be expected that the analyte will pass through the central tunnel without having much effect on the plasma surrounding it.However if as can happen analyte material does enter the plasma it can well be imagined that the extra demand on power that this makes may be an appreciable fraction of the total power supplied to a low-power plasma but is likely to be negligible for a high-power plasma. \Yhy then in view of these apparent advantages of high-powered systems are most systems offered for sale low power? I believe the reason has less to do with science than it has to do with economics and attractive packaging that is smallness and other salesworthy points. It is said by a t least one manufacturer that everything which can be done on a high-power system can be done equally well on a low-power system. [ do not know whether this is true or not and I doubt whether the manufacturer does.With two exceptions I know of no systematic work with optimised systems which has been carried out to compare high-and low-powered plasmas for any of the properties which are likely to be affected by power such as those described above. The exceptions were studies of Greenfield and Thorburn Burnsz1 and some by Ebdon et a1.22 The main part of the former work lay in comparing the net signal to background ratios of three optimised plasma systems on one spectrometer. The three systems were the Scott torch with argon coolant and the Greenfield torch with argon and nitrogen coolants. High- and low-power generators were used as necessary in the hope that they would give the power range required. I n Ebdon's work a comparison was made between a Greenfield torch and a demountable torch of their own construction operating with argon and with nitrogen coolants.The basis of the comparison was the signal to background ratio obtained when these torches were run under optimised conditions with the same generator and spectrometer. The Scott torch was found to require powers ranging from 0.53 to 1.21 kW in the plasma, when a series of lines varying in difficulty of excitation were optimised for signal to back-ground ratio. The Greenfield torch with an argon coolant could not be optimised in mos Fig. 9. Effect of power in the plasma on Swan bands. Fig. 10. Effect of power in the plasma on Swan bands. [to face page 104 November 1980 SPECTROSCOPY COMES OF AGE 1043 instances as the minimum power obtainable from the generator was greater than the indicated optimum value for the gas flows required.In two instances however true optimisation was obtained. In one a power of 3.92 kW was required for the other 1.86 kW and it may be said that the bottom end was < 1.39 kW. At the two powers where optimisation was achieved the net ratios were similar to those obtained with the Scott torch. When the Greenfield torch was run with a nitrogen coolant the same set of lines gave in five instances out of eight and with powers varying from less than 0.79 up to 2.6 kW better signal to background ratios than the Scott torch. Of the other three instances two of the lines were subject to molecular interference and the other was not truly optimised as the optimisation required a lower power than could be obtained from either generator.Ebdon and his colleagues also found that the nitrogen-cooled plasma required more power to reach an optimum for the manganese 257.6-nm line than did the argon-cooled plasma and indeed his generator was unable to supply enough power to give a true optimisation a t the gas flows indicated. The point about these two series of experiments is that in both instances the workers had no doubt what the basis of their comparison was and took steps to ensure that all their systems were operating under optimum conditions before any comparisons were made. In both series of experiments each of the two groups optimised the observation height the power in the plasma and the injector plasma and coolant gas flows in the one instance using an alternating variable search method of optimisation in the other by the use of a variable step size simplex procedure.To emphasise this lack of reliable information and to highlight the way in which confusion is engendered in the minds of the scientific layman let us consider the following example. If we wish to compare say a Radyne generator with a Greenfield torch against a Plasma-therm generator with a Scott torch it is perfectly valid to do so providing it is understood that it is the total systems which are being compared and that the operating conditions are optimised for the same figure of merit whatever that may be. If the instrument on which the comparison is being made is a simultaneous instrument then compromise conditions can be used. Properly a set of compromise conditions should be derived from a knowledge of the individual optimised conditions.The adjective compromise should not be used as a euphemism for arbitrary. The comparison of one generator and one torch with a different generator and different torch is not a comparison of individual generators or torches and if such a comparison is required then each generator should be run with the same torch and each torch mounted on the same generator. Above all unless conditions are optimised for each combination any comparison will be invalid. These remarks may appear obvious to some but the literature would suggest that they may not be quite so obvious to others. Let me return to the claim that anything that can be done on a high-powered plasma can be done on a low-powered plasma.I have said there is a lack of systematic evidence but I do know of examples from personal experience and from reports in the literature of low-powered plasmas which extinguish if the sample cup is removed from the nebuliser inlet and, as a consequence air is drawn through the plasma. As I have demonstrated low-powered plasmas through which organic solvents are nebulised exhibit Swan bands and the charac-teristic green hue is readily seen. It is also said that carbon formation occurs on the injector tip under these circumstances. When hydride-forming elements are introduced into the plasma as hydrides following a borohydride reduction the hydrogen formed is not readily tolerated by low-powered systems or so it has been reported. Similarly it has been reported that low-powered systems tend to extinguish when solutions of high acid concentration are nebulised and the resulting aerosol injected through them and it is noticeable that many workers with low-powered plasmas do seem to restrict the salt content of the solutions they nebulise to around 1%.I t is my experience when experimenting with unusual operating conditions which produce instability in the plasma that stability can often be restored by increasing the power. It is also my experience that high-powered plasmas are not subject to the limitations outlined above. Now all this adds up to an over-all impression that low-powered systems are not as versatile and adaptable as the high-powered ones. However I would again emphasise it is only an impression and I do not know of any work that I would regard as reliable which substantiates or denies this impression other than that which I have reported.Until our knowledge o 1044 GREENFIELD plasma spectroscopy has reached the point where we know exactly what power we require for all circumstances it seems to me preferable to have a reserve of power and to work with a generator which will produce about 5 kW of power in the plasma whilst still being capable of maintaining a plasma of less 0.5 kW. A high-powered generator can always be turned down but there are limits to which a low-powered generator can be turned up. I have tried to give an honest report on topics in plasma spectrometry which are sometimes contentious sometimes not and which I hope will be fully resolved by future research, Perhaps I may conclude by asking the question what else does the future hold? A field where I think substantial advances might be made is that of spectrometers as distinct from sources.The recent developments in gratings are important and the speeds, both optical and mechanical which scanning monochromators are reported to possess make the use of these very attractive especially if more than one is used to view a single source under computer control. The developments with echelle gratings have led to the development of two new types of spectrometer. The first is the analogue of a scanning monochromator, but with no moving parts the scanning being done electronically. The second is the corresponding analogue of the conventional direct reader where the whole spectrum is stored on a vidicon screen and as many wavelengths as are wanted can be chosen to have their intensities measured sequentially.Both these ideas seem to me to be exciting and excellent in principle and it is to be hoped that the practical difficulties will be overcome soon. Instead of dispersing the radiation emitted by the source and identifying wavelengths associated with an element we ignore the radiation entirely and use the plasma as a source for a mass spectrometer identifying the elements by mass. I am convinced that now having come of age it will survive to a ripe old age. Another interesting idea was reported by Gray some years Papers were given on this topic a t this conference. Finally I believe the future holds a lot for the ICP.1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. References Christian G. D. and Feldman F. J. A p p l . Spectrosc. 1971 25 660. Berman E. Appl. Spectrosc. 1975 29 1. Boumans P. W. J . M. and Barnes R . M. I C P I n f . Newsl. 1978 3 445. L’vov B. V. “Atomic Absorption Spectrochemical Analysis,” Adam Hilger London 1970 p. 228. Sturgeon R. E . Chakrabarti C. L. Maines I. S. and Bertels P. C. Arzal. Chem. 1975 47 1240. Kleinmann I. and Svoboda V. Anal. Chem. 1969 41 1029. Nixon D. E. Fassel V. A. and Kniseley R. N. Anal. Chem. 1974 46 210. Gunn A. M. Millard D. L. and Kirkbright G. F. Analyst 1978 103 1066. Kirkbright G. F. and Snook R. D. Anal. Chem. 1979 51 1938. Salin E. D. and Horlick G. Anal. Chem. 1979 51 2284. Ohls I<. and Sommer D. I C P I n f . Newsl. January 1980 5 (Special Edition) Paper 78. Thompson M. Pahlavanpoor B. Walton S. J . and Kirkbright G. F. Analyst 1978 103 568. Fry R. C. Denton M. B. Windsor D. L. and Northway S. J . Appl. Spectrosc. 1979 33 399. Meddings B. Kaiser H. and Anderson H. ICP In$ Newsl. January 1980 5 (Special Edition), Abdallah M. H. Mermet J . M. and Trassy C. Anal. Chim. Acta 1976 87 329. JIermet J . BZ. and Robin J. Anal. Chivn. Acta 1975 70 271. Brockaert J . A. C. Leis F. and Laqua K. Spectrochim. Acta 1979 34B 167. Kornblum G. R. and de Galan L. Spectrochim. Acta 1977 32B 455. Larson G. F. Fassel V. A . Scott R. H. and Kniselcy R. N. Anal. Chem. 1975 47 238. Boumans P. W. J . M. and de Boer F. J Spectvochim. Acta 1976 31B 355. Greenfield S. and Thorburn Burns D. Anal. Chim. Acta 1980 113 205. Ebdon L. Cave M. R. and Mowthorpe D. J. Anal. Chim. Acta 1980 115 179. Gray A . L. Analyst 1975 100 289. Paper 65

ISSN:0003-2654    

DOI:10.1039/AN9800501032

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

Agzalyst, November, 1980, Vol. 105, pp. 1045-1059 1045 Moral Ageing of Analytical Methods* George E. Baiulescu Department of Aizalytical Chemistry, National Institute of Chemistry, Splaiul Independenlei 202, 77208 Bucharest, Romania The paper is based on progress, development and optiinisation of analytical methods. On the basis of the examination of three kinds of correlations, the method - instrument, the man - instrument and the man - method - instrument correlations, the best conditions to diminish the moral ageing of analytical methods are proposed. Keywords : Moral ageing; obsolescence ; optilnisation One of the features of modern analytical chemistry as a science is its continuing improvement, both as such and in its close co-operation with other disciplines such as physics, automatics and data processing.Recently, the rank-and-file analytical chemist has found that it becomes increasingly difficult to characterise a sample objectively, particularly when the sample belongs to a material with a very complex structure and composition or when one’s aim is to determine traces of a particular component in a matrix. The introduction, for objective reasons, of novel physical methods of analysis, which need not be detailed here but are responsible for the explosive advance in the methods of analysing surfaces, and which are ultra-rapid because of their non-destructive and easily automated character, has brought to the analyst a wide spectrum of analytical methods, some of which are highly sophisticated. Competition among instrument manufacturers is, of course, responsible for the fact that an up-to-date instrument may become outmoded overnight as far as its operational para- meters are concerned, or else the method of analysis itself may become impracticable in some cases.This way, it becomes harder and harder to make an educated choice of the most adequate method for analysing a certain material and of the most suitable instrument available on the market. When considering the evolution of analytical methods and of analytical instrumentation, we may see that from time to time, as required by the chemical - analytical control of industrial output or by the necessities of new fields in scientific research, new techniques of analysis occur which complement the older ones. Under such conditions, the older analytical methods, which are no longer appropriate with respect to sensitivity, selectivity, accuracy and rapidity for the considered scope of the particular problem, may be regarded as being morally aged.The degree of moral ageing depends on the very nature of the method, but the analytical methods nonetheless preserve, within certain limits, their viability and validity. These limits are set by the nature and composition of the sample. No matter how much the analytical methods may be improved, a degree of the operational parameters with which we could know the exact composition of the sample would never be reached. We might say that “no analysis is better than the sample itself.” During the latter decades of this century, in order to obtain satisfactory results, an analyst who investigates various materials from a compositional or structural point of view must be an excellent connoisseur of analytical chemistry as a science and of chemical analysis as a technique.To obtain a high degree of confidence, an analytical chemistry specialist must be aware, in our opinion, of three groups of correlations: (a) the method - instrument correlation; ( b ) the man - instrument correlation; and (c) the method - man - instrument correlation. These correlations are simple repercussions of the modern development of science and tech- nology. The analytical chemist is aware that by his activity he contributes both to the advancement of science and to the optimisation of the industrial processes by the chemical - analytical control of its output. Lancaster, July 20-26th, 1980.* Plenary Lecture presented at the 5th SAC International Conference on Analytical Chemistry,1046 BAIULESCU : MORAL AGEING Analyst, Vol. 105 As early as during his undergraduate studies, the future analyst should be made aware of the necessity for a great correlation between teaching, research and industrial production, which could be schematised by a triangle (Fig. l), which may become equilateral under proper conditions. This may only happen if perfect optimisation of the whole information system exists, the information moving in the direction indicated by the arrows. Only the man who can be integrated into this cycle can eradicate gradually, step by step, his own degree of moral ageing. Teaching Research production Fig.1 . The great correlation. As Professor Laitinen appropriately showed in a recent editorial in Analytical Chemistry,l “the essence of the modern approach is a quest for the fundamental understanding of a problem rather than an empirical determination of composition.” Before proceeding to scrutinise the three types of correlation, let us consider for a moment the notion of moral ageing insofar as the operator is concerned. The continuing increase in the amount of information, in analytical chemistry as a whole, and in its various branches may cause man, its beneficiary, to be overwhelmed; he must therefore re-adapt himself continuously to this informational universe. To avoid his own moral ageing, the analyst has to embrace the most efficient information system, a system which he would use in accordance with the complexity of the analytical system.The most difficult problem apparent today is that of the mentioned correlations, particu- larly the method - man - instrument correlation. The appearance of new synthetic materials, the demands associated with substances of higher and higher purity and the needs of bio- chemical, biomedical and clinical research have caused this triple correlation to become the most difficult to preserve within normal parameters. Horwitzz appropriately remarked that “as the level of measurement goes down, analytical methodology and instrumentation become increasingly complex and sophisticated, operations may become too expensive and results too unreliable for practicability,” and, what is still more important, “a practical method will always be practicable, but a practicable method is not necessarily practical.” These facts, which may seem very clear to many people, are nevertheless by their nature very complex and have implications upon the general analytical process itself.The progress, development and optimisation of analytical methods would have to take account of these correlations and to investigate them within their interdependence. This is the only way to ensure that the notion of moral ageing would be understood: that means would eventually be found to harmonise older analytical methods with new requirements and that entirely novel methods would be continually devised. In what follows we shall discuss this group of three correlations, based on the daily practice of the analyst and on pure analytical research.The Method - Instrument Correlation It is the personal opinion of the author, always conveyed to his students, that any great scientific discovery in chemistry in general, and in analytical chemistry in particular, is based on facts which, once shared by the general public, may appear to be extremely simple. A certain phenomenon is noticed by most of the specialists in a certain field, but it takes a certain type of person or research team to think somewhat differently as to how this phenomenon could be applied. The history of analytical chemistry provides us with a large number of such instances; one need only remember polarography and atomic absorption. Perhaps the discovery of these techniques is the surest proof of this method - instrument correlation.The polaro-November, 1980 OF ANALYTICAL METHODS 1047 graphic technique was introduced by Heyrovsky in the early 1920s and it developed explosively mainly because of improvements in instrumentation. A method of analysis is really good when it may develop horizontally, i.e., when it is applicable to a wide class of compounds. The application of polarography to systems of organic compounds was made possible by a knowledge of the mechanisms of the organic reactions in general and of the electrode reactions in particular. However, the polarographic technique began to show signs of ageing after only a few decades; it could no longer face the sensitivity requirements of modern science and practice.Polarography had t o be revived- but how? It revived by itself, not through such palliatives as oscillopolarography (a tech- nique which may well be more sensitive but which surely is much more demanding), but by renewing the principle itself. By analogy with the concentration by extraction or by chromatographic techniques, an increase in sensitivity was sought by electrochemical pre- concentration. “This is the principle of electrochemical stripping methods, in which the substance to be determined is concentrated electrolytically on the measuring electrode (in the form of an amalgam or of a film on the surface of a mercury or solid electrode) and is then transferred back into the solution by the reverse electrolytic process.”3 Anodic-stripping voltammetry has now become a technique of prime analytical importance, which equals in its performance, especially with respect to its sensitivity, electrothermal atomic-absorption spectroscopy, and is comparable to many radiochemical and radiometric methods.Owing to instrumental improvements, it has become a current working method for trace analysis. It has the advantage over atomic absorption of allowing multi-element determinations in a single run, and in addition it is possible t o determine elements in several of their different valence states. Atomic-absorption spectroscopy emerged as a necessity at a precise time, when any developments in flame photometry could no longer maintain it in a better shape than a t a level of moral ageing. It is to the merit of all those who had studied the flame-Kirchoff, Bunsen, Teclu as predecessors, Mavrodineanu and others-that they opened the way for Walsh to initiate that analytical infovmafion explosion which is atomic absorption : a very simple idea indeed, but highly efficient, that of using the atoms in their ground state and not in their excited states.A short time thereafter, L’vov, West and Massmann improved it, bringing it to spectacular performances with respect to sensitivity (10-14 g for some elements, e.g., Cd and Zn) and a high degree of automation. Modern instruments, provided with microprocessors, can carry out hundreds of analyses per hour in a totally objective mannev. We cannot overlook, however, not even in the framework of this general presentation, a technique which has not become morally aged, at least formally, although it is as old as this century.This is mass spectrometry, which has developed so far in three stages: isotope separation, organic structural analysis and, through adaptation of spark sources, an invaluable integral, high-sensitivity technique for analysing traces a t the parts per billion levels. Every aspect of this technique still preserves its viability and validity. By modernising and computerising the instrumentation, mass spectrometry has become a highly efficient detector for GC and HPLC. Spark-source mass spectrometry is today an unsurpassed weapon in analytical cosmochemistry and in the investigation of high-purity materials. We cannot, however, discuss the method - instrument correlation without mentioning the moral support which we, the analysts, have-namely the reaction itself. The chemical reaction is our stronghold which, no matter how highly automated or ultrasophisticated is the instrument (and we mean instruments of various generations, even those provided with self-servicing), will always remain viable. The most perfect instrument, even when totally automated, will never be able to do what man can do, what he thinks and imagines.The researcher in the field of analytical chemistry will never be replaced by an automaton-and this is what gives us some hope. Knowledge of the reaction mechanisms in solution, in flames, etc., is a major problem, which can ultimately affect the advance of analytical chemistry as a science and of chemical analysis as a technique.We shall illustrate this with a few significant examples taken from the domain of chemical reactions in solution, We shall refer to some studies carried out on organic reagents. If you wish, this is a problem of atomic statistical population. Introduced in late 1954, this technique revived by itself.1048 BAIULESCG : MORAL AGEIIGG Analyst, Vol. 105 Perhaps the most significant example on these lines is the introduction by Belcher et al." of the alizarin complexan 3-di(carboxyniethyl)aminoetliyl-1,2-dihydroxyanthraquinone or 1,2- dihydroxyanthraquinon-3-ylmethylamine-~~~-diacetic acid. The cerium complex of this reagent (ALC-Ce) has the very interesting feature of giving a direct colour reaction for the fluoride ion. Leonard and \Vest5 believe that the blue complex formed by ALC-Ce with the fluoride iun is a ternary complex.the formula 2: The ALC-Ce complex 6as the formula 1, and the ternary complex Can anyone say why must we determine the fluoride ion in this way, when we have today on hand an electrode which is particularly selective for this ion, the LaF, electrode, as a single crystal if you wish? Let us not forget, however, that these solid-membrane electrodes themselves owe their existence to a study on the mechanism of the precipitation process. We can see that, a t certain moments, analytical research grows by itself and that new fields start to develop from some field which had beforehand appeared outmoded. We may draw from this the very interesting conclusion that the d2gree of moral ageing of a particular method depends on a particular period of time, that any analytical method remains potentially valid provided that it adapts itself as well as possible to the increasing demands of the analytical process.Let us argue the importance of the study of the reactions by two examples picked up from our own practice. We shall refer first to an element whose ions have an extremely pro- nounced tendency to hydrolyse and to form inorganic polymers. This is zirconium, which forms very strong bonds with oxygen, thereby accounting for its hydrolysing and poly- merising capacities. For these reasons, a very limited number of organic reagents are known which form unitary compounds with zirconium, detectable as such by a physical method. \Ye had begun this study of the organic reagents for zirconium more than 20 years ago, based on the examination, statistical if you wish, of a wide group of reagents; we concluded that a reagent containing a pyrazolone nucleus, namely tartrazine (3), used as a dye and indicator in titrimetry, forms a unitary compound with zirconium,B whose formula, tested by EM, IR, NhlR and MS studies, is (4): FOONaNovem,ber, 1980 OF ANALYTICAL METHODS 1049 The reagent allowed the zirconium compound to be directly weighed and the method turned out to be highly selective.Owing to this selectivity, it was used for determining zirconium in Romania and in the USA, Poland and the USSR. IVhereas zirconium is a difficult element to study analytically, palladium is, on the contrary, too submissive- to such an extent that its analytical chemistry is very rich.We ourselves proposed, also many years ago, systematising the reactions of this element with organic reagents used to detect and determine palladium. We concentrated on the study of the analytical functional groups and of the analytically active groups (a terminology we believe is still valid today). As a consequence of our study, it became clear that palladium(I1) reacts with organic reagents containing two general types of analytical functional groups (A and B) shown below: The second example belongs to the analytical chemistry of palladium. By noticing' that reagents which contain both A and B groups in their molecule, e.g., Methyl Red, give a very sensitive reaction with palladium(II), we synthesised a new reagent, a derivative of chromotropic acid which contains symmetrically both types of groups.8 In this way, we obtained reagent 6, which is very sensitive and selective for palladium.Its performance is comparable to that of the palladiazo reagent (5).9 In fact, by examining the formula of palladiazo we may see that this reagent belongs to the class of reagents containing analytical functional groups for palladium : OH OH OH OH We believe that these examples show to a sufficient extent the importance of studying reactions in solution and the contemporary relevance of such studies. It is appropriate to repeat that our main role as analysts lies in doing basic research in the framework of analytical chemistry as a science ; the most sophisticated inst,rumentation merely serves the analytical aims of this science, i.e., the chemical analysis and control of industrial production.We do not deny that for a series of analytical researches such as the study of some kinetic reactions or of some enzymatic reactions, we need sophisticated instruments. Let us not forget, however, that all major discoveries in analytical chemistry were made based on a knowledge of the essence of the phenomena and, moreover, using fairly primitive instru- mentation. A very fashionable field of rnodern analytical chemistry, that of chemi- and bioluminescence (CL and BL), may illustrate this point; it is based to a great extent on the study of the nature and mechanisms of these reactions and makes use of less sophisticated instruments.1050 BAIULESCU : MORAL AGEING Analyst, Vol.105 At the 11th Annual Symposium on Advanced Analytical Concepts for the Clinical Labora- tory, held in Oak Ridge, Tenn., April 26-27th, 1979, Seitz, from the University of Durham in New Hampshire, suggested, in connection with the instrumentation available today for CL analysis, that “the instrumentation for the most part is not fully automated; I’m not convinced that if you operate manually you’re going to get the kind of operator-to-operator precision you’d like for a clinical instrument. Also, if you do the firefly or bacterial re- actions, you have to have some provision for not using up too much of your reagents.” Seitz continues : “My prediction is-and this is my personal belief-that you won’t find the present instruments used on a wide scale in the clinical laboratory.My feeling is that someone will either adapt some of the present automated equipment so it can measure chemiluminescence, or come out with some special instruments dedicaied to specific analyses. The whole instrument will be much more automated-you’ll put a sample in a tube and the instrument will do everying else.”1° These are opinions only. What is positive is that the complexity of this kind of reaction (the CL and BL reactions), and their restricted number, ought to guide forcibly research in this field in order to discover new reactions of this type and, what is still more important, in replying to Seitz-to find the optimum method- instrument correlation. After that, the problem would be left in the hands of the electronic people and of the automation specialists to optimise the system. It is certain that, during some period of time, ex$losive processes occur, which finally lead to the very rapid advancement of some laboratory techniques and in which new methods of analysis appear.Do these appear randomly? Of course not. Let us think of the ultimate beneficiary, this is solely industrial production or is, in other words, the control of industrial output. That great teaching - research - industrial production correlation closes in upon itself for a while, having to be developed thereafter on another, higher level, once new necessities of production demand it. The general analytical process is therefore dependent on production and the information explosion in analytical chemistry has, as its eventual purpose, the very progress, development and optimisation of the analytical methods in order that they be taken over by the beneficiary, which is chemico-analytical control, and finally the optimisation of the industrial process by and with analytical chemistry.Several fields, every one implying a large number of analyses, have contributed to a great extent to the improvement of the method - instrument correlation; in what follows we shall refer to only three of these fields-environmental pollution, clinical analysis and forensic science. As each of these fields has particularities of its own, the analytical systems had to be adapted to the specific purpose of each field. The necessity for continuous determinations, for automatically recording the analytical information given by the study of the air and water pollution, imposed a selection of the analytical methods and their harmonisation with the monitors.Not every method of analysis has withstood this shock. The methods used today for selectively determining air pollutants, for example, have to depend on high sensitivity, selectivity, accuracy and rapidity (the rapidity being mainly determined by the rate of the reaction involved). Only a limited number of analytical methods passed such a test. Although manufacturers have produced various portable instruments for controlling the atmospheric pollutants (03, SO,, NO,), most of these instruments, while being sufficiently sensitive, do not have the desired selectivity (we have especially in mind electrometric analysers).The only selective and accurate method for determining SO, in the atmosphere is, for example, the method proposed by West and Gaeke,ll although it has the drawback of being slow in transmitting the signal. The circle closes again, after all, and we return to the necessity for a method based on a principle which is adaptable to some empirical requirements, the analysis of pollutants in this instance. Only those methods based on chemical reactions or the entirely physical methods of analysis which give due credit to the operational parameters of the analytical process will be adaptable on the new, automated and computerised instrumentation. Horwitzz is right again: “A practical method will always be practicable, but a practicable method is not necessarily practical.” The problem becomes markedly difficult when we have to arialyse a complex mixture.A typical example is the analysis of organic pollutants in the environment. Their large number and great diversity require a previous separation and a suitable detector. It is in this field that the efficiency of the separation techniques, such as GC:, and the necessity for exigent and expensive detectors, such as the mass spectrometer, can be verified. TheNovember, 1980 OF ANALYTICAL METHODS 1051 modern GC - MS - computer systems turn out to be very suitable for resolving complex mixtures of air pollutants. Their acceptance avoids the use of a large number of individual instruments; owing to their lack of selectivity, these would be morally aged starting from the very moment of their manufacture (see the unsuccessful attempts at producing separate analysers for various groups of hydrocarbons).Clinical analysis has also imposed a selection of those methods of analysis which are sufficiently sensitive, selective and fast, and have an advanced degree of automation. A substantial contribution to the understanding of the method - instrument correlation in this very important field has been made by the study of enzymatic reactions. Broadly, the classical analysts were afraid of using enzymes, maybe due to the uncommon working conditions required by their manipulation. The instability of the enzymes caused their application in analytical practice to be difficult. A technique was then developed for immobilising the enzymes by bonding them to an inert support ; their stability thereby increased, while their catalytic activity was preserved.Based on the study of enzymatic reactions on the one hand, and on the progress made in the enzyme immobilisation techniques on the other, a series of instruments were soon developed for determining glucose, lactose, sucrose, galactose, etc. Immobilised enzymes have evolved from simple chemical deter- mination to more complex fields, where they do remarkable work. For instance, the immobilised cholinesterase is used successfully in the analysis of pesticides. We can see from all these examples that the fundamental basis is again the study of the reactions; the instrumentation is merely an annex6 which is necessary and sufficient to the general analytical process insofar as it is adequately selected.In contrast with biology, this selection is not natural, but is directed by laws suitably formulated (by man) to reach the optimisation parameters of the method - instrument correlation. Use of chemical analysis in criminology and forensic science is an interface connecting the broad group of methods specific to clinical analysis with proper forensic procedures; this is valid in the large number of cases in which components of an organic nature are involved. Forensic science has got rid, to a great extent, of the older research techniques. The electron microprobe replaced microscopy, and atomic-absorption spectroscopy, particularly electro- thermal atomisation, replaced almost completely the classical emission and absorption spectroscopic techniques. Radioimmunoassay (RIA), a technique developed by Yalow and Berson in the 1950s, has become a current technique in modern forensic laboratories.We mention here only two examples: the detection of curare in the Jascalevich muder t r i a P and the detection of drugs of abuse in hair by radioimmunoassay.13 The Jascalevich case could only be solved by close co-operation between several high-performance analytical techniques such as RIA, HPLC and MS. You can see how many things analytical chemistry can do, and all this merely to identify a substance such as d-tubocurarine(7) : The achievements of analytical chemistry in forensic science are spectacular indeed. 2c1- Analytical chemistry is practically a field without frontiers, the analytical techniques revive by themselves, they are apparently inoculated against moral ageing, they complement and supplement each other.Let us consider now the second of our correlations.Analytical chemistry aspects Any analytical determination is based on a very simple relationship: * . * . ’ * (1) P = f(C) * . . . that is, on setting up a dependence between a physical property and the concentration of the substance to be determined. Hence the most appropriate name for the methods used in - QuantityNovember, 1980 OF ANALYTICAL METHODS 1053 analytical chemistry: Physical Methods of Chemical Analysis. Any other names, such as physico-chemical methods or instrumental methods, are either incomplete or biased, or sometimes even confusing.It might be recommended that we define our own speciality in very precise terms, a t the very least in order to be in a position to promote it. The name of physico-chemical analytical methods was much appreciated by some physical chemists who, wishing to appropriate these methods and being somewhat ignorant of the matter, believed them to belong to physical chemistry. They were nevertheless willing to concede that gravimetry and titrimetry, regarded as classical methods of analysis, rightly belong to analytical chemistry. They overlooked the obvious fact that these methods also make use of a physical property in their measurements, hence the validity of the above-mentioned name. By its very nature, the sample to be analysed is the culprit in the division on the analytical methods into non-destructive and destructive : The man - instrument correlation is carried out by the intermediary of the sample.,-Non-destructive methods ‘-Destructive methods Man - Sample - + Instrument- Let us go back to our octahedron; in fact, the sample itself leads us to it. On this occasion we can solve a much discussed, generally valid problem, the competition among specialists in acquiring increasingly more sophisticated laboratory instruments. When (inorganic and organic) samples with comparatively simple composition, especially in solution, are analysed, emphasis is given to the analytical chemist as a performer of the chemical analysis. An altogether different situation is met when the sample has a more complex composition (biological samples or synthetic materials), or when trace analysis is necessary (ultrapure and nuclear materials, or semiconductors) ; such cases require more and more advanced non-destructive methods (let us only mention surface analysis techniques) and the analytical chemist is in many cases excluded or replaced by a physicist, For this reason, in order to make efficient use of the complete arsenal of physical methods, non-destructive or destructive, the analytical chemist must know the nature of the sample, to get the utmost information about it. This knowledge of the nature of the sample implies, in turn, being well informed horizontally in three corners of the parallelogram : inorganic, organic and physical chemistry.This is the reason why we used this particular geometric representation; in our opinion, these branches play equal parts in the formation of a good analyst.The fourth corner of the parallelogram, representing the horizontal development of the analytical chemist, is automatics. The analytical chemist of the 1980s and the future analytical chemist must take as an indisputable fact that a compulsory prerequisite for using the new microprocessor- and microcomputer-equipped instruments is to know, a t least a t an elementary level, the basic problems of automatics. This parallelogram, which connects the sample to automatics by means of the three funda- mental chemical disciplines, accounts for the fact that today the analytical chemist is no longer a simple performer, but also a thirtker, a man who solves problems-by using an automated instrument if you so wish.As Grasselli aptly remarked in a recent paper,16 “the modern analytical chemist no longer just answers questions, but rather solves problems or suggests solutions.” Though equivalent, each of the four disciplines represented on the parallelogram has a decisive role in the formation of the analytical chemist. He will have to acquire, in the first place, a sound basic knowledge of inorganic and organic chemistry; this is the way for him to understand and approach any kind of sample; he will understand many mechanisms of reaction in solution or the complexation (masking and demasking) processes. As shown above, knowing the nature of the sample is almost an unmistakable sign that the analysis would be successful.Understanding solution chemistry, equilibrium reactions and reactions in partially aqueous or in non-aqueous media requires the analytical chemist to possess a deep knowledge of physical chemistry. Several high-performance analytical methods, such as kinetic and enzymatic methods, in order to be understood and thereby used to their best effects, also require a thorough study1054 BAIULESCU : MORAL AGEING Analyst, Vol. 105 of physical chemistry. Similarly, the analytical chemist cannot carry out thermal analyses, even the simplest thermogravimetric one, if he does not have a sound knowledge of chemical thermodynamics. If the parallelogram has equivalent coyneys, this means that the analytical chemist has got a sound instruction in the horizontal direction, that he has got a scientific foundation. This scientific base is furnished by the school and by the teacher; if good himself, the teacher injects him with the creative spirit needed by research. It is commonly known that a good science teacher can only be a person who has worked in the scientific field and who continues to do so; to put it into our terms, one who can understand the man - instrument correlation.From their very beginning all chemical sciences, and especially analytical sciences, have progressed as a result of some instrumentation. The man - instrument correlation is, however, much more complex than it appears to be at first sight, To illustrate this we shall replace the plane parallelogram with a solid body and consider the two apices of the ocatahedron, that is, the development of the analytical chemist in the vertical direction.This development involves physics and mathematics. The very name of 9hysical methods of chemical analysis is an argument for the necessity for having a sound knowledge of physics in order to be able to use laboratory instrumentation and to understand the physical significance of the various phenomena which occur in a chemical interaction or reaction. On the other hand, cleaning of the analytical signal, its conversion and transmission in a suitable form, requires a deep knowledge of mathematics, and not only of mathematical statistics for data processing as some analytical chemists still believe. The efficient use of computer terminals or of microcomputers needs a basic know- ledge of programming methods and of data processing, particularly in cases where a krge amount of analytical information is involved; this is the case with taridem techniques such as GC - MS - computer or of transform techniques such as FT-IR or FT-NMR spectroscopy.Owing to the complexity of the data to be interpreted, the structural analysis branch is the user of an ample mathematical apparatus. Many undergraduate students who just begin to acquire the basis of qualitative analysis in their first year of study, regard them- selves as detectives who try to identify a number of ions in a solution. The true detective is, however, that analytical chemist who deals with structural analysis. Let us take the example of a complex mixture resolved by the GC - MS - computer tandem system or, still more convenient, merely the MS analysis of some organic components.The analytical chemist has to reconstitute the original molecules knowing only the mass fragments. You might call it a new chemistry. In most instances he cannot solve this problem by his own means and he calls for the computer. This might be very helpful, but only if the analyst knows enough chemistry to be able to interpret the results. As Kowalski, a specialist in a new field, chemo- metrics (which includes the application of mathematical and statistical methods to the analysis of chemical measurements), rightly remarks,17 “computer pattern recognition methods extend the ability of human pattern recognition but, in the end, it is the chemist who must do the chemistry.” Looking again at the “octahedron of analytical chemistry,” we can be in some measure satisfied, having specified every coYneY of the octahedron. Some people will ask, nonetheless, where is electronics? In my opinion, electronics, in spite of its impressive recent development and of its being a rather separate discipline, is too close to physics to elude being included in it.An analytical chemist who has a sound knowledge of physics can also understand electronics and can be the beneficiary of the information given by the most up-to-date physical methods of analysis. We shall refer in what follows to one of the most recent physical methods in surface analysis, solid-state photoelectron spectroscopy with synchrotron radiation. This is the only one of the many techniques of modern science benefitting from synchrotron radiation.As Weaver and Margaritondo show,18 “Synchroton radiation sources, providing intense, tunable, polarised and stable beams of ultraviolet and X-ray photons, are having a great impact on biology, physics, chemistry, materials science and other areas of research. Synchrotron radiation has revolutionised solid-state photoelectron spectroscopy by enhancing its capabilities for investigating the electronic behaviour of solids and solid surfaces.” The new techniques of analysing surfaces can only be used through a healthy knowledge of physics. As Hercules so appropriately remarks,lg “one of the new frontiers in analytical chemistry is the chemical analysis of surfaces.” We benefit today from a wide spectrum of techniques for analysing surfaces, techniques which often allow trace analysis on surfaces.Is it necessary for chemists?November, 1980 OF ANALYTICAL METHODS 1055 We cite here ESCA (X-ray photoelectron spectroscopy), AES (Auger electron spectroscopy), SIMS (secondary ion mass spectroscopy) and ISS (low-energy ion scattering spectroscopy), These techniques have emerged to supplement X-ray fluorescence spectroscopy, an old surface analysis method which, though morally aged at a certain moment, has revived by automation and computerisation. XRF and its grandsons are successfully used today for studying a large variety of inorganic, organic and biological materials. Perhaps the most spectacular field which proves the involvement of physics in analytical chemistry is, however, the development of ICP - AES (inductively coupled plasma - atomic- emission spectroscopy).Research in this domain has led to the emergence of a powerful technique for chemical analysis. To give history its due, we shall quote Professor Fassel,20 a specialist not in physics, but in physical chemistry: “For the past 17 years my associates and I have devoted a fraction of our efforts to the development of the basic science, the investigative methods and the hardware for an analytical approach that would eventually provide the capability of determining the chemical elements selectively, at all concentration levels, i.e., major, minor and trace constituents, simultaneously if so desired, or in a rapid sequential manner, with a single analytical technique, and with accuracy and precision.A new analytical approach, usually identified as inductively coupled plasma - atomic emission spectroscopy (ICP - AES), has emerged from these studies.” This technique has necessarily appeared as an addition to a revival of emission spectroscopy, a technique which had become morally aged from several points of view. The highest ageing coefficient of classical emission spectroscopy concerned its sensitivity. Together with flame and electrothermal AA, ICP - AES furnishes today the best high-performance instru- ments in the hands of analytical chemists. The discovery of ICP - AES and its use in modern analytical chemistry are one of the best proofs of the existence of a man - instrument correla- tion. In order that this technique could emerge, the octahedron had to work perfectly on both directions in space.The development of such a technique needs a sound knowledge of chemistry (organic and inorganic), physical chemistry, physics, mathematics and auto- matics. Let us now take up the last, and perhaps most important correlation. The Method - Man - Instrument Correlation This correlation is closely related to the method - instrument and the man - instrument correlations. Whereas in discussing the first correlation we emphasised the role of the reaction and of the reaction mechanisms and, in the framework of the second, the role of the teaching process in analytical chemistry, this third correlation will have as a pivotal term the chemico- analytical control of industrial production.In most instances this control is carried out today with automatic analysers and has as its final purpose the optimisation of industrial Does this mean that, within this last correlation, man is included in an inflexible tandem, that he becomes a one-man instrument, a man ultraspecialised in a certain field? Perhaps many of those who are not familiar with education problems in analytical chemistry would tend to believe that this is so. The problem must, however, be stated in totally different. terms. To adapt himself to this correlation, to interpose himself between method and instrument for optimising the analytical process, the analytical chemist of the future must be a cultured man in his own speciality, and indeed be generally cultured. This will only be ensured by the continual improvement of the teaching process in chemistry in general and in analytical chemistry in particular.This is one of our common problems, but it is solved differently from case to case and from country to country. In an editorial entitled “Education for the 21st Century,” AbelsonZ1 states that, “Whatever the changing shape of society, scientists and engineers will have essential roles. The un- certainties, though, make it advisable to caution against excessive specialisation. In contrast, it seems desirable to adopt policies of maximum flexibility, of preservation of options, of being prepared to pursue lifelong learning.” I t is certain that in order to preserve his form, a good specialist must continuously read, must have a good development on the horizontal-but of course not only on the surface.output.1056 BAIULESCU : MORAL AGEING Analyst, Vol. 105 The chemico-analytical control of industrial production could have brought both the analyst and analytical chemistry itself to a virtual standstill. Fortunately, basic analytical research has solved and still solves the method - man - instrument correlation. Man has to adjust himself to the necessities of the continuous control required by the production process, To the analytical chemist, the necessity of knowing the problems of automation and of analytical sciences has proved to be a necessary and assertive forward step. In this way the analyst has learned to operate automatic analysers, but not to become part of them. We must realise once and for all that analytical chemistry as a science used automation for doing chemical analyses and does not automate itself. As Whalley correctly remarks,22 “now in industry, and in the chemical industry in particular, we are swinging over more to automation of analytical control, and by this I do not mean doing analytical chemistry automatically.” Analytical chemistry is subtantially involved in the control of the product’s quality and this by the very observance of the method - man - instrument correlation. Taking account of the intended purpose and of its implications, this correlation must be considered very seriously.This action of the correlation upon the control process, and hence ultimately upon the optimisation of quality control, may be pictured by a triangular pyramid (Fig.3). Quality control Method Fig. 3. Optimisation of It would be interesting to consider quality control from a historical point of view and from which the analytical chemist sees it. In a first stage, the instrumentation used by the analyst was rather primitive and many of us can recall the time when the quality control in steel mills was still carried out by rapid titrimetric methods. Today, titrimetry has become morally aged in this branch of industry, where it was replaced by automatic emission spectro- scopy or X-ray fluorescence spectroscopy. The current requirements of quality control in industry have led to extremely fast progress in automatic control instrumentation, The presence of a process chromatograph in an industrial installation is now regarded as common- place.Two domains of prime consequence, air and water pollution, are perhaps among those for which the necessity for automation and automatic quality control appears most obvious. A formidable competition has existed and still exists among instrument manufacturers to come out with more and more new devices, new analytical automata for continuous measure- ments. The control of the quality of air and of water is, however, by the very nature of the determinations involved, very pressing. Analytical chemistry had to co-operate with a series of related disciplines in order to work out novel principles, new analysis methods. These had to be provided with optimised operational parameters which could make them adaptable to automatic systems.In this sense, a s@ectacular example of method - man - instrument correlation is perhaps the use of piezoelectric sensors as continuous monitors of atmospheric pollutants. The piezoelectric crystals, coated with various adsorbent materials, have been used as sensors for a number of pollutants, including sulphur dioxide, nitrogen dioxide and ammonia. Cooke et al.23 have recently proposed such a piezoelectric sensor as a continuous monitor of atmospheric pollutants. Even if, as in any new field, these sensors still have drawbacks, we are certain that in the not too distant future analytical chemistry will prove the possibilities of such devices being incorporated as parts of an automatic system for controlling the quality of air. This point of view is based on the theoretical feasibility of such a system and on the relatively simple way of building up the necessary instrumentation.quality control.November, 1980 OF ANALYTICAL METHODS 1057 Investigation of water pollutants, or the analysis of waste waters, has been the main beneficiary of the research carried out in connection with electrochemical sensors. The ion-selective electrodes originally proposed for analysing some anions (a branch of analytical chemistry which had previously been ignored) have found today practically unlimited applications in the analysis of organic and biological compounds. They are now a highly efficient means in continuous analysis in many industrial branches, especially in the control of the quality of water. As early as the 1960s, research in the field of ion-selective electrodes carried out worldwide by various analytical chemistry groups led to the advancement of the theory and practice of this prominent technique.Research progressed at an unexpectedly quick pace, if we take into account that, in the analysis of waste waters, most of the pollutants which could potentially be determined are organic. The references found in the literature are so rich that this led us in 1977 to publish a monograph24 on the application of ion-selective electrodes in organic analysis. There are today many electrochemical sensors for on-line process monitoring, using a variety of techniques such as potentiometry, conductivity measurements and v ~ l t a m m e t r y . ~ ~ In order that the method - man - instrument correlation is able to lead us to automatic quality control, the man factor has to ensure a perfect matching of the operational para- meters of the method with those of the instrument, that is the adaptation of the most suitable method t o the most suitable instrument.Let us picture this fact in the form of the most perfect spatial body, a sphere (Fig. 4). Following the choice of the most suitable method for analysing a certain material, simple or complex, in order to get the most out of it the analyst will have to choose from the great variety of instruments available, the most appropriate to his actual analytical problem. In the case of on-line analyses, he would have to get and process the analytical information continuously. m Instrument Method Fig. 4. Adaptation of the A very significant problem which arises with automatic on-line analysers is that of the reproducibility of the analytical signal.Instruments which are able to self-calibrate periodically have to be designed. An auto- matic device is up to the expected task insofar as the angle between the two great circles (one standing for the method and the other one for the instrument) approaches zero, that is, when one great circle superimposes the other. A method - instrument symbiosis is then ensured, which remains valid for a practically unlimited time with no need for man to interfere. The analytical chemist is the only one in a position to judge, or assess, when this “couple” is no longer necessary and to decide which other “couple,” operationally more favourable, should replace it.Besides providing the automatic control systems used directly in the industrial process, chemical analysis becomes an indispensable weapon in a string of public services, among which is the quality control of food. In this field of direct concern to us, as it is related to public health, the contribution of analytical chemistry is tremendous. All the examples already given, as well as the many others which could not be included, prove that analytical chemistry, by the intermediate of chemical analysis, is of great consequence to our daily life. This is mainly because we seek products of better and better quality; quality means quality control, which, in turn, implies chemical analysis, which has to rely on analytical chemistry in order to get from it the best quality control system.This is perhaps why Markland26 believes that “analysis includes any process for determining any fact as to the nature, substance or quality of any material.” Let us look now, in the framework of this third correlation, at the general analytical process. To do this we should take into account, as shown above, the operational para- method to the instrument. These devices should be extremely reliable.1058 BAIULESCU : MORAL AGEING Analyst, Vol. 105 meters of the method and the functional parameters of the instrument. To become competi- tive, any analytical method must meet the following requirements. ( a ) The highest possible sensitivity (also dependent on the concentration of the analysed component). In other words, a method which is suitable for trace analysis should be used in this field only, and not for major components; to do otherwise would mean impairment of the method - instru- ment correlation. We should be able to determine a chemical species, irrespective of its concentration, in the presence of a complex matrix, (c) Good accuracy, a parameter which depends on both method and instrument and which varies as a function of the concentration of the analysed component.However advanced the laboratory instrumentation, in going down to lower and lower concentrations, parts per million, billion, trillion or even lower levels, the accuracy of the determination decreases, ( d ) The rapidity of obtaining the analytical information which, in the case of the destructive analysis methods, depends on the reaction mechanisms in solution and, for non-destructive methods, is to a great extent affected by the functional parameters of the instrumentation.In order to be correlated with the most adequate method, the analytical instruments should also perform in the optimum range of their functional parameters. In today’s so- called third-gemration instruments, the high degree of automation is provided by a micro- processing coye. We might say that our discussions today about miniaturised and automated analytical instruments would not have been possible without the invention of microprocessors. In most analytical fields, the modern instrumentation is almost completely automated, starting with sample positioning (automatic sample switching) and ending with remote information transmission if you wish.The most typical example is that of the totally automated instru- ments for analysing the Moon’s soil, placed on the lunar module and directly sending the analytical information to Earth; but let us not forget that an automaton does not think, that it is an artefact of the human mind, whom it may in some measure mimic, but whom it could never substitute. ( b ) The best possible selectivity. Conclusions Following this general inquiry into the correlations involved in the analytical process, we have gained a clear understanding of the role of analytical chemistry in education, in research and in the quality control of industrial products (by means of chemical analysis). The analytical process can thus be schematised as in Fig.5. \ / chemistry 1 control Tea chin 9 Fig. 5. The analytical system. notion, including All stages have We are usine the term analvtical svstem because it is a comDrehensive all the stages Gf the analyticd proceis (teaching - research - apilications). to be maintained with optimum parameters in order that the system be viable. If not, moral ageing of the whole system sets in. The system is cyclical, but the circle never closes upon itself. Improvements in teaching lead to an amplification of research which, in turn, benefits industrial production, and the present and future production needs to use men and instruments which should be adaptable to its optimisation, therefore having a reduced degree of moral ageing. We saw above how we can partly reduce the degree of moral ageing of laboratory instrumentation (for research and control) and of that used in automatic on-line 0November, 1980 OF ANALYTICAL METHODS 1059 control, but the task is much more difficult with regard to the operator, man.Man cannot be bettered but within a continuous educational process. In chemistry, in particular in analytical chemistry, education can be carried out as a continuation of the teaching process. The possible educational forms are very diverse, but a principle which always holds is that the best form of education is the steady team work in the laboratory. No modern large-scale analytical investigation can be carried out except by a well blended team, consisting of persons equal as far as their scientific background is concerned or, as often happens, by a mixed, interdisciplinary team.The most significant example is that of the new physical methods of chemical analysis, which owe their emergence to a continuous, often long-term research effort of a mixed team. This is the only way in which automatic analysis and control systems could appear, in which the existing instrumentation can be improved. This is the only way in which we can steer clear of moral ageing. Any analyst who can understand this truth will become persuaded that : 1. Analytical chemistry is a science which will always remain young. 2. Research in analytical chemistry is required not only by chemistry, but also by a wide range of other frontier scientific fields whose names begin with bio: biophysics, bio- chemistry, biology, bioenergetics, biochemical engineering, etc., to mention but a few of the beneficiaries. 3. Chemical analysis-the applicative side of analytical chemistry-has now become a constituent, integral part of the production process ; it contributes to the optimisation of the process of controlling the quality of the industrial products. In order that all this becomes true we shall have to believe in the boundless possibilities of this science, analytical chemistry. References 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Laitinen, H. A., Anal. Chem., 1979, 51, 2065. Horwitz, W., Anal. Chem., 1979, $1, 741A. Vydra, F., Stulik, K., and Julakova, E., “Electrochemical Stripping Analysis,” Ellis Honvood, Belcher. R.. Leonard. M. A., and West. T. S., T . Chem. Soc., 1958, 2390. Chichester, 1976, p. 18. Leonard, M’. A,, and West, T . S., J. Chem. Soc:, 1960, 4477. Baiulescu, G., and Turcu, L., Anal. Chim. Acta, 1959, 21, 33. Popa, G., Negoiu, D., and Baiulescu, G., Zh. Anal. Khim., 1959, 14, 322. Baiulescu, G., Greff, C., and Dgnet, F., Analyst, 1969, 94, 354. Perez-Bustamante, J. A,, and Burriel-Marti, F., Anal. Chim. Ada, 1967, 37, 49. Anal. Chem., 1979, 51, 826A. West, P. W., and Gaeke, G. C., Anal. Chem., 1956, 28, 1816. Hall, L. H., and Hirsch, R. F., Anal. Chem., 1979, 51, 812A. Anal. Chem., 1980, 52, 43A. Laitinen, H. A , , Anal. Chem., 1970, 42, 37A. Stock, J . T., Anal. Chem., 1973, 45, 974A. Grasselli, J . G., Anal. Chem., 1980, 52, 30A. Kowalski, B. R., Anal. Chem., 1975, 47, 1152A. Weaver, J. H., and Margaritondo, G., Science, 1979, 206, 151. Hercules, D. M., Anal. Chem., 1978, 50, 734A. Fassel, V. A,, Anal. Chem., 1979, 51, 1290A. Abelson, P. H., Science, 1979, 205, 1087. Whalley, C., Analyst, 1974, 99, 817. Cooke, S., West, T. S., and Watts, P., Anal. Proc., 1980, 17, 2. Baiulescu, G. E., and Cosofret, V. V., “Applications of Ion-Selective Membrane Electrodes in Organic Bailey, P. L., Anal. Chem., 1978, 50, 698A. Markland, J., Analyst, 1974, 99, 810. Analysis,” Ellis Honvood, Chichester, 1977.

ISSN:0003-2654    

DOI:10.1039/AN9800501045

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

1060 Analyst, November, 1980, Vol. 105, p p . 1060-1067 Determination of Polybrominated Biphenyl and Related Compounds by Gas - Liquid Chromatography with a Plasma Emission Detector Kevin J. Mulligan and Joseph A. Caruso and Fred L. Fricke Department of Chemistry, University of Cincinnati, Cincinnati, Ohio 45221, USA United States Food and Drug Administration, Elemental Analysis Research Center, 1141 Central Parkw ay, Cincinnati, Ohio 45202, USA Polybrominated biphenyl and related compounds were separated by gas - liquid chromatography and monitored by a microwave-induced plasma emission detector. The helium plasma was sustained a t atmospheric pressure in a Beenakker cavity. By following the bromine emission a t 478.65 nm, linear ranges of at least 103 with detection limits (signal to noise ratio := 3) of the order of 1 ng were achieved.These results are compared with those obtained with an electron-capture detector and the advantages of the element selectivity of the plasma system are illustrated. Keywords: Polybrominated biphenyl determination; gas - liquid chvomato- graphy; plasma emission detector Polybrominated biphenyls are used as pesticides, insecticides, flame retardants, etc. Chemically, these species are soluble in lipids and insoluble in water and, as a result, tend to accumulate in fatty tissue. Moreover, they are resistant to biodegradation and certain members, such as DDT, polychlorinated biphenyl (PCB) and polybrominated biphenyl (PBB), have been shown to have deleterious effects on the health of individuals who have been exposed to relatively low 1evels.l Current methods of analysis2 utilise gas chromatography with electron-capture detection and are very sensitive (typically of the order of 1 ng for half full-scale deflection), but tend to be time consuming because of the late elution of these compounds.There is a requirement for good separation because of the lack of specificity of this detector in distinguishing between the halogens. The application of an element-selective detector should enable one to relax the chromatographic constraints (separation from hydrocarbons will still be necessary) on the analysis and enhance sample throughput. Moreover, such a technique would provide additional qualitative information. In 1965, McCormack et aL3 interfaced a gas chromatograph to a microwave-induced plasma emission source in order to obtain element selectivity by means of atomic-emission spectro- scopy.The applicability of their technique was limited by the necessity that the plasma be sustained in argon. Certain elements, such as chlorine, bromine, fluorine, nitrogen and oxygen, exhibit only diatomic band emission in this plasma, which decreases the sensitivity and selectivity of the method. Also, it was subsequently shown that the intensity of such emission was not necessarily a linear function of concentration.* Bache and Lisk5 sub- stituted a helium plasma maintained at low pressure (5-10 Tom). This more energetic medium enabled them to determine a variety of halogenated pesticides in amounts approach- ing the nanogram level.Recent improvements in cavity design,6-8 which permit a more efficient transfer of micro- wave power from the generator to the plasma, have enabled workers to sustain the helium plasma a t atmospheric pressure. Beenakkers demonstrated that a helium plasma sustained a t atmospheric pressure in a cavity of his design’ was capable of providing detection limits for elements such as bromine, chlorine and nitrogen that were superior to those previously achieved. Quimby and co-workers interfaced this cavity to a gas chromatographlo and used the system to determine a variety of trihalomethanes in drinking waterll and aqueous chlorination products of humic substances.12 This paper describes the development of a system similar to that described by Quimby et u Z . , ~ O contrasts its performance with that of the electron-capture detector for the analysis ofMULLIGAN, CARUSO AND FRICKE 1061 brominated organic compounds of relatively high molecular mass, and illustrates the advan- tages to be gained from the element-selective nature of the device.Experimental The system is illustrated in Fig. 1 and the components are described individually below. 0 2 in purifier He in tuner .. gain I ILL monochromator I I - Fig. 1 . Schematic diagram of the gas chromatography - micro- R, Metering valve; V, 2-way valve; wave induced plasma system. C, coupler: and GS, glass sleeve packed with glass-wool. Gas Chromatograph Several modifications were made to an F & R.I Scientific 5750 research gas chromatograph (Hewlett-Packard). A high-temperature microvolume three-way valve (Carle Instruments, Model 2033) was inserted after the column to provide a means of venting the solvent.In order to maintain the plasma during the vent interval, additional support gas (helium) was introduced a t a flow-rate of 10 ml min-l. A coupler was constructed from a &-in to +in stainless-steel reducing union to join the silica containment tube (4.7 mm 0.d. x 2.9 mm i.d., Heraeus-Amersil Inc.) to the remainder of the system. This tube passed from the oven through a glass conduit (Pyrex) packed with glass-wool and entered the cavity, which was immediately adjacent to the external wall. Gas-flow Network A series of metering valves (R) and two-way valves (V) served to regulate the flow of helium, which was purified by passage over Drierite and molecular sieves. Provision was made for introducing small amounts of bromine continuously into the plasma by passing helium over carbon tetrabromide, a solid with an appreciable vapour pressure.This facilitated the location of the analytical line and optimisation of operating conditions, and enabled spectra to be obtained for purposes of background examination. The same procedure was used with hexachloroethane (C,Cl,) for characterisation of the chlorine response. This solid required slight warming in order to attain the desired sample introduction rate. Finally, the capability of adding oxygen to the support gas offered a means of cleaning carbon deposits from the walls of the silica containment tube. These deposits were observed when carbon tetrabromide was introduced over extended periods of time or some solvent was inadvertently routed through the plasma. Microwave-induced Plasma The cavity is shown in Fig.2. A Beenakker cavity was constructed from copper stock following established guidelines61062 MULLIGAN et al. : POLYBROMINATED BIPHENYL AND RELAT:ED Analyst, Vol. 2 in -4 Water Front view Water in Cross section Back view 105 U Fig. 2. The Beenakker cavity. The drawing is approxima.tely t o scale. The inner diameter of the cavity proper is 91.5 mm, with a depth of 8 mm. A copper coupling loop (2 mm 0.d.) connects the jack of a UG-58U R F connector to the front wall after penetrating a distance of 12 mm into the cavity. Two holes bored through the body of the cavity as shown and capped with Q-in pipe with *-in 0.d.fittings allowed the cavity to be water-cooled. The skin temperature during con- tinuous operation remained in the neighbourhood of 40 "C. Finally, a port (2 in diameter) bored into the front of the cavity provided good optical access to the plasma. A short length (24 in) of RG-9U coaxial cable (Beldon) linked the cavity to an impedance- matching device (three-stub tuner, Maury Microwave Inc., Model 1823), which was attached to the microwave power generator (Kiva Instrument Co.). This generator was capable of providing microwave energy a t 2450 MHz from 0 to 120 W forward power. Although this configuration was adopted largely for convenience, it was found that the tuner remained cooler during operation, particularly at the higher power levels, than when it was affixed to the cavity with the cable interposed between it and the generator.Heating of the tuner and the cable became more pronounced with the passage of days, but occasional cleaning restored satisfactory operating characteristics. The effect seems to be due to surface oxide formation and might be minimised by gold-plating the conductive surfaces. Detection System Radiation from the plasma was focused on the 0.050-mm entrance slit of a 0.50-m Ebert scanning monochromator (Jarrell-Ash, f-number = 8.6) by a fused silica lens (f = 75 mm, 32 mm diameter) placed 96 mm from the source. The monochrornator has a 1180 groove mm-1 grating blazed a t 300.0nm. An RCA 1P28 photomultiplier tube was positioned directly behind the 0.50-mm exit slit.Signals from the photomultiplier tube were conditioned by a current to voltage converter arranged as a variable-gain amplifier and then recorded on a strip-chart recorder (Hewlett- Packard, Model 7101B). Operating Parameters A silanised glass column (1.5 ft x 4 mm i.d.) was packed with 2% OV-101 on Chromosorb W H P (80-100 mesh) and operated a t 238 "C with a helium flow-rate of 60 ml min-l. These conditions are similar to those which have been suggested elsewhere,2 except for the short column length.November, 1980 COMPOUNDS BY GLC WITH A PLASMA EMISSION DETECTOR 1063 , 1.5 nrn, 467.87 nrn 1481 6 7 nrn 518.24 nrn He Fig. 3. Comparative spectra: CBr, + He vevsus He. The analytical line for bromine determination was chosen after examination of the spectral region between 520 and 466 nm while small amounts of carbon tetrabromide were being introduced into the plasma.The monochromator was scanned at a rate of 1 nm min-l to provide the spectra shown in Fig. 3. A comparison of these spectra with those obtained for the pure helium plasma suggested that optimum results would be obtained by using the 478.55-nm line. This is in agreement with the observation of Bache and Lisk5 on the be- haviour of the low-pressure helium plasma. Workers using the atmospheric helium plasma9*l0 have reported that the 470.5-nm line is more sensitive. The discrepancy may be due to differences in the resolving power of the optical systems employed. Further studies based on repeated injections of a standard amount of analyte revealed that the signal to noise ratio increased as a function of forward power and decreased as a function of the helium flow-rate.The power level subsequently employed (85-90 W forward power at 0-1 W reflected power) was the highest that could be used without excessively heating the tuner and cable. The flow-rate behaviour has been noted by Quimby et al.,1° and, although they suggested that a flow-rate of 50 ml min-l is optimum for bromine and chlorine deter- minations, the chromatographic requirement (60 ml min-l) and the amount of helium needed to maintain the plasma during the vent interval (10 ml min-l) dictated an operational flow- rate of 70 ml min-l. Compounds for investigation were obtained from the US Food and Drug Administration and were used without further purification.Their names (with standard numbers in paren- theses) included $$'-DDT (200), methoxychlor (87), FireMaster BP-6 (PBB) (1050), Citex BN-21 (1217), Citex BC-26 (1218), 2,3,4,5,6-pentabromomethylbenzene (1310), 2,S-dibromo- propyl 2,4,6-tribromophenyl ether (1324) and pentabromodiphenyl ether (1397). While Citex BN-21 is moderately soluble in benzene and toluene, the other compounds dissolve readily. The solubility of certain members, particularly Citex BN-21 and 1310, in light hydrocarbon solvents such as light petroleum (boiling range 35-60 "C) is very limited. Benzene was used as the solvent in spite of its toxicity to provide as rapid an elution of the solvent peak as possible. Results and Discussion The structures of some of the compounds selected for study and the chromatograms obtained for each are shown in Fig.4. The sharp upward spikes reflect the point a t which the column flow was directed into the plasma. This produced a disruption of the plasma that persisted for some time, as indicated by the sloping base line. Initially, this appeared to be due to tailing of the solvent (benzene) peak, but the same behaviour was observed with light petroleum (boiling range 35-60 "C). It seems that the sudden increase in flow-rate (from 10 to 70 ml min-l) when the column flow is introduced slightly relocates the plasma within the containment tube. Although the chromatography can be improved, the time scale of the experiment makes these conditions attractive for this particular study. When utilising complex samples, the1064 MULLIGAN et al.: POLYBROMINATED BIPHENYL AND RELATED Analyst, Vol. 105 , 30s, J Br Fig. 4. Structure and chromatographic properties of several analytes. Identification of some peaks is by the FDA standard number (thc numbers in parentheses are the amounts injected in nanograms) : A, PUB (60) : B, 1397 (86) : C, BC-26 (63); D, BN-21 (80): and E, 1310 (33). chromatography will be improved because of probable hydrocarbon interferences. The retention times follow the relative order presented elsewhere2 but are about 40% shorter. Adsorption effects are known to be a source of difficulty with this separation, particularly when packed columns are employed. Markedly improved performance has been observed with capillary columns.2 Data collected on the recorder were analysed by plotting the ratio of the peak height (signal) to three times the standard deviation of the base line (noise) against the amount of analyte injected.The graphs exhibit a linear dynamic range for each compound of a t least three orders of magnitude. Detection limits were defined as the amount of analyte required to produce a signal to noise ratio of 3. These values are contrasted with the performanc,e of the electron-capture detector in Table I. Although it is not possible to compare the two systems exactly, as the chromatography and the definition of detection limit are different, the electron-capture detector appears to be more sensitive. Dividing the detection limit by the width of the analyte peak at the base line provided a measure of the absolute sensitivity of the emission detector. After adjustment of these values to reflect the amount of bromine present in each of three compounds that exhibited only one chromatographic peak, a comparison showed reasonably good agreement, as shown in Table I.This suggests that molecular fragmentation is not a factor at the power level employed. Moreover, these values are similar to the value reported for the low-pressure Typical calibration graphs are represented in Fig. 5.November, 1980 COMPOUNDS BY GLC WITH A PLASMA EMISSION DETECTOR 1065 5 10 50100 1000 Amount injectedhg Fig. 5. Calibration graphs. The graphs are arbi- trarily placed with respect to the ordinate for clarity. Each point represents an average of a t least three deter- minations. 1, BN-21; 2, 1310; 3, BC-26; 4, 1397; and 5, PBB.helium p l a ~ m a . ~ but, as mentioned earlier, they employed a spectrometer with a higher resolving power. amounts of 1310, phenanthrene and pP'-DDT a t 478.55 nm. signals was determined to be 95 : 25 : 1 (carbon : chlorine : bromine). Quimby et al.1° noted a four-fold improvement by using the 470.49-nm line Selectivity was evaluated by comparing the total response (area) obtained for known The molar ratio for equivalent TABLE I COMPARISON OF DETECTION LIMITS WITH PROPOSED METHOD AND ELECTRON-CAPTURE DETECTOR Detection limit by proposed Normalised Compound method/ng (at 3 X S B ~ ~ ~ l i n e ) * bromine/pg s-l 1310 . . . . 1397 . . . . BN-21 . , . . BC-26 . . . . 0.60 2.7 1.7 1.7 53 48 35 t PBB . . . . 2.8 t * s = standard deviation.t These compounds were not treated in the way described. Electron-capture detection limit/ng (for 3 f s d . ) 0.5 1.4 1.3 0.9 1.5 Relative standard deviations (eight determinations) were evaluate1 in the m i J l e of the working range and are shown in Table 11. PBB and 1310 were also determined at approximately ten times the detection limit and showed only a They varied between 1.4 and 3.3%. TABLE I1 RELATIVE STANDARD DEVIATIONS (?Z = 8) Amount injected/ Relative standard deviation, Compound ng % 1310 . . . . . . 35.6 6.2 1397 . . . . . . 224.0 BN-21 . . . . . . 316.0 BC-26 . . . . . . 266.0 PBB . . . . . . 300.0 30.0 1.4 3.3 1.1 3.3 3.0 2.8 4.41066 MULLIGAN et al. : POLYBROMINATED BIPHENYL AND RELATED Analyst, Vol. 105 2-3-fold increase in relative standard deviation.Much of the scatter is probably due to variation in the 4-yl injection volume. The plasma emission detector may approach the sensitivity of the electron-capture detector and, in addition, it offers greater selectivity. A mixture was prepared that contained DDT, methoxychlor, 1310, 1324, Citex BN-21, Citex BC-26 and PBB. Chromatograms obtained by using the bromine line at 478.55 nm and the chlorine line at 481.01 nm are shown in Fig. 6. 1 Bc-26 PBB (50) Chlorine line Methoxychlor (137) C30 s4 I DDT (107) Fig. 6. Chromatographic charac- ter of a mixture when monitored using either the Br emission or the C1 emission. Analytes are identified on the peaks by FDA standard numbers; the numbers in paren- theses represent the amount of each component present in nanograms.Note that the chlorine emission chromatogram has been inverted to facilitate a comparison of the two. If an electron-capture detector had been used, the mixture would have been largely un- resolvable. Of particular interest is the response to Citex BC-26, which contains two bromine and six chlorine atoms. While there is significant overlap of the BC-26 and BN-21 peaks on the bromine channel, BC-26 is cleanly resolved on the chlorine channel. Knowing the response factors on each channel should enable one to quantify BC-26 on the chlorine channel and then to obtain BN-21 on the bromine channel indirectly. This illustrates the potential usefulness of the plasma emission detector as a means of relaxing chromatographic constraints by virtue of its element-selective capability.Conclusion Although the plasma emission detector may not be as sensitive as the electron-capture detector, it offers the advantage of element selectivity, which can be of use in minimising the degree of separation required for analytical purposes. This feature can serve to reduce the analysis time and can also provide qualitative information. We extend our appreciation to Roy C. Kuennen for construction of the Beenakker cavity and to Charles S. Finsterwalter (USFDA) for providing the compounds studied. K. J.M. and J.A.C. gratefully acknowledge the National Institute for Occupational Safety and Health for partial support of this work under the grant NIOSH R01-OH-00739-~01A1.November, 1980 COMPOUNDS BY GLC WITH A PLASMA EMISSION DETECTOR 1067 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. References Carter, L. J., Science, 1976, 192, 240. Miller, L. J., and Puma, B. J., FDA Lab. Ifif. Bull., No. 2270, March 1979. McCormack, A. J., Tong, S. C., and Cooke, W. D., Anal. Chem., 1965, 37, 1470. Dagnall, R. M., West, T. S., and Whitehead, P., Anal. Chem., 1972, 44, 2074. Bache, C. A , , and Lisk, D. J.. Anal. Chem., 1967, 39, 786. Beenakker, C. I. M., Spectrochim. Acta, Part B, 1976, 31, 483. Skogerboe, R. K., and Coleman, G. N., Anal. Chem., 1976, 48, 611A. Beenakker, C. I. M., and Boumans, P. W. J. M., Spectrochim. Acta, Part B, 1978, 33, 53. Beenakker, C. I. M., Spectrochim. Acta, Part B , 1977, 32, 173. Quimby, B. D., Uden, P. C., and Barnes, R. M., Anal. Chem., 1978, 50, 2112. Quimby, B. D., Delaney, M. F., Uden, P. C., and Barnes, R. M., Anal. Chem., 1979, 51, 875. Quimby, B. D., Delaney, M. F., Uden, P. C., and Barnes, R. M., Anal. Chem., 1980, 52, 259. Received May 8th, 1980 Accepted May 22nd. 1980

ISSN:0003-2654    

DOI:10.1039/AN9800501060

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

1068 Analyst, November, 1980, Vol. 105, pp. 1068-1075 Microbiological and Chemical Analysis of Polymyxin B and Polymyxin E (Colistin) Sulphates A. H. Thomas, J. M. Thomas and I . Holloway National Institute f o r Biological Standards and Control, Holly Hill, Hanzpstead, London, N W3 6RB Samples of the sulphates of polymyxin B and polymyxin E (colistin) have been assayed microbiologically. The content of phenylalanine was deter- mined in the samples of polymyxin B. The compositions of the samples obtained by using thin-layer chromatography, gas - liquid chromatography and high-performance liquid chromatography were compared. The relative merits of each method are discussed. Keywords 1 Polymyxin B ; polymyxin E ; phenylalanine; chromatography; microbiological assay The polymyxins are a group of closely related decapeptide antibiotics produced by species of The two polymyxins used therapeutically, polymyxin B and polymyxin E (colistin), can each be sub-divided into at least four components, differing only in the fatty acid moiety attached to the cyclic peptide, i.e., 6-methyloctanoic acid in polymyxins B, and E,, 6-methylheptanoic acid in polymyxins B, and E,, octanoic acid in polymyxins B, and E, and heptanoic acid in polymyxins B, and E,.Little is known about the antibacterial activity and toxicity of the single poly- myxin components. Data on the relative activities of polymyxins B and E are conflicting. It is claimed that polymyxins containing the longer 6-methyloctanoic acid are generally more active than those containing 6-methylheptanoic acid.The composition of the fermenta- tion medium can influence the composition of the polymyxin complex; the addition of L-isoleucine, the precursor of 6-methyloctanoic acid, resulted in the biosynthesis of poly- myxin B containing mainly the B, component.2 Laboratories sometimes find it difficult to agree on determinations of potency for the polymyxin antibiotics, as exemplified in the collaborative assay of the Second International Standard for Polymyxin B.3 This difficulty can be attributed, in part, to the heterogeneous composition of these antibiotics, Gas - liquid cnromatography has been used t o measure the fatty acids present in polymyxin B4 in order to discover the relative composition of the material and it is included in the monograph for colistin in the British Pharmac~poeia.~ The complex composition of the polymyxins has recently been demonstrated by thin-layer6 and higli-performance liquid chromat~graphy.~-~ In this study samples of polymyxin B and E from different sources have been examined for potency and composition by use of microbiological and chemical assays.Materials and Methods Polymyxin B and E the British Pharmacopoeia. potency of the samples. Microbiological Assay The samples were assayed by using the diffusion method suggested in the British Pharma- copoeia, except that the assay medium was modified by replacing the agar with agarose (1% m/V) and by the addition of 2.0 x lo-, M disodium EDTA to improve the sensitivity. The automated assay procedure has been describedlO and the following solutions were used : polymyxin B, 160, 80 and 40 IU ml-l; and colistin, 100, 50 and 25 IU ml-l.The potencies of the samples were expressed in terms of the respective International Standards for poly- myxin B or colistin. They differ from each other in amino acid composition. Experimental All of the samples of polymyxin B and E complied with the tests for identity specified in Table I shows the source of manufacture and the declaredTHOMAS, THOMAS AND HOLLOWAY TABLE I POTENCIES OF SAMPLES OF POLYMYXIN B AND POLYMYXIN E OBTAINED BY MICROBIOLOGICAL ASSAY, INDICATING SOURCE OF MANUFACTURE 1069 Sample Source 1 A Polymyxin B- 2 B 3 B 4 B 5 C 6 C 7 A 8 A 9 Not known 10 Not known Polymyxin E (Golistin)- 1 D 2 E 3 E 4 E 5 E 6 D 7 F 8 F Declared Determined Confidence potency/ potency/ limits, & % IU mg-l I U mg-' (P = 0.95) 8 403 8332 8 345 8 302 8 300 8 043 6 970 7 863 8 079 - 20 500 19500 20818 19 930 20 000 20 600 18733 - - 7 882 8209 8995 8479 8 947 7 161 8 005 8 052 6 206 - 18817 20296 19319 19823 20617 20 608 18815 - 3.7 2.8 4.5 3.0 - - 4.8 4.3 5 .4 - - 4.8 4.6 2.5 3.2 1.3 - Departure from parallelism P < 0.05 P < 0.001 P < 0.05 P < 0.01 P > 0.001 P > 0.001 P < 0.05 P < 0.05 P < 0.05 - P > 0.001 P < 0.05 P < 0.05 P < 0.05 P > 0.001 P < 0.05 P < 0.05 Difference between determined and declared potency, % - -5.41 $8.34 +2.15 +11.23 $2.74 + 1.80 - 1.63 -0.34 I - -3.50 -3.07 -2.51 - + 3.08 $0.04 + 0.43 Phenylalanine Content The phenylalanine content of polymyxin B was determined by a method proposed for inclusion in the Second Edition of the European Pharmacopoeia.Polymyxin B sulphate, 0.375 g, accurately weighed ( M ) , was dissolved in 100 ml of 0.1 N hydrochloric acid. The absorbances (A) at the maxima of 264, 258 and 252 nm, as well as the absorbances at 300 and 280 nm, were measured. The phenylalanine content was determined from Thin-layer Chromatography Pre-coated silanised silica gel plates (20 x 20 cm, 0.25 mm thick, E. Merck) were used and the mobile phase consisted of acetone-0.1 N hydrochloric acid (25 + 75) containing 1% m/V sodium chloride. Prior to chromatography the plates were placed in a filter-paper lined chromatographic tank that had been saturated with the mobile phase, developed to their upper edges, removed and dried at 110 "C for 30 min. Aliquots (1 p1) of aqueous solutions (20 mg ml-l) of the samples were applied to the plates and developed over a 15-cm path in the same direction as the previous run.The plates were then removed from the tank and the separated components detected either with ninhydrin reagent or by bioautography as previously described.6 The ninhydrin-treated chromatoplates were scanned at 570 nm, using the thin-layer chromatography accessory of a fluorescence spectrophotometer (MPF-4, Perkin-Elmer) as this was found to be the most sensitive method of comparing the separation profiles. The areas of the ninhydrin zones and the areas of the zones of inhibition of growth were then measured with an image analyser (Optomax, Micromeasurements) .ll Because the purified individual components were unavailable, no calibration graphs could be obtained.Regression coefficients ( b ) were measured for the two major components of polymyxin B and polymyxin E (b = 12853 biological assay, b = 3018 ninhydrin response). The relative concentrations of the two components were calculated from the following: concl areal - area2 log (cone,> = b1070 THOMAS et d. : MICROBIOLOGICAL AND CHEMICAL ANALYSIS Analyst, VOL, 105 Gas - Liquid Chromatography Each sample was examined according to the method of the British Pharmacopoeia.5 High-performance Liquid Chromatography Analysis of the samples by high-performance liquid chromatography was based on a previously described method.* The equipment consisted of a reciprocating pump (Consta- metric IIG, Laboratory Data Control) and a variable-wavelength spectrophotometer fitted with a 75-pl flow-through cell (CE 272, Cecil Instruments); the column (15 x 0.4 cm) was packed with Spherisorb ODS 5 pm diameter particles.The mobile phase was composed of tartrate buffer (0.005 M), pH 3.0, containing sodium butane-1-sulphonate (0.005 M), sodium sulphate (0.2 M) and acetonitrile (18% V / V ) . The mobile phase was filtered through a glass microfibre filter and de-gassed prior to use. The samples were dissolved in the mobile phase a t a concentration of 10 mg ml-l and 7.5 pl of the sample solution were injected on to the column, the flow-rate being 2 ml min-l and the inlet pressure 1800 lb in-,. The detector was operated at 220 nm with a sensitivity of 0.2 a.u.f.s.The peak areas were measured with an electronic integrator and the results are given as the percentage of the total area produced by each component. Results When the microbiological assays were analysed all the assays of the polymyxin B samples and two of the polymyxin E samples were found to be invalid because of a significant departure from parallelism. Inspection of the logarithm dose - response lines showed that many of the assays were non-parallel as a result of a departure from the linear response a t the lowest concentration of the sample (test) solution. Therefore, all of the polymyxin B assays were re-analysed as 2 + 2 blocks. The results of the polymyxin E assays and the re-analysed polymyxin B assays are shown in Table I. All of the samples complied with the minimum potency requirements of the British and European Pharmacopoeias, although not all of the determinations agreed with the manufacturers’ declared values.Polymyxin E was also assayed in terms of polymyxin B by using the standard assay procedure; it was found to be twice as potent on a mass basis. Only two samples of polymyxin B did not meet the proposed European content require- ment of phenylalanine [9.0-12.0~0 m/m (Table 11)] but there was no correlation between the bioassay results and the phenylalanine content of the polymyxin B sample (Y = 0.14, n = 10). TABLE I1 PHENYLALANINE CONTENT IN SAMPLES OF POLYMYXIN B Sample number . . . . 1 2 3 4 5 6 7 8 9 10 Phenylalanine, % m/m . . 11.19 9.29 9.29 9.70 10.05 10.79 9.67 12.97 8.92 9.88 The complex compositions of polymyxin B and E were demonstrated by thin-layer chromatography (Figs.1 and 2). The polymyxins that could be identified were B,, B,, El and E, (RF values 0.17, 0.25, 0.21 and 0.32, respectively). Some of the unidentified minor components, which were just detectable with ninhydrin, did exhibit antibacterial activity. The samples varied in the number and amount of the minor components detected. Two of the polymyxin E samples (7 and 8) contained an additional bioactive component migrating between El and E, (RF 0.28) that was not detected in the other samples. The densitometric scan of a ninhydrin-reacted chromatogram was characteristic and it was possible to group samples from a common source according to their scan profile.The proportions of the two major components in the polymyxins were determined following thin-layer chromatography on the basis of biological activity and ninhydrin reactivity (Table 111). The correlation between the two procedures was good, r = 0.93 for each component (n = 10 for polymyxins B, and B,, n = 8 for polymyxins E, and E,), but the relative proportions of the two major components in a sample differed depending on the assay method used. It was assumed, because the two components differed only by a fatty acid residue, that the ninhydrin response4 5 9 6 1 7 2 5 9 6 1 7 2 4 Fig. 1. Thin-layer chromatogranis of polymyxin B sulphate detected with ninhydrin ( a ) and with Numbers refer to sample number, method as stated in text. Bovdetella bronchiseptzca XCTC 8344 ( b ) .E2 E l E2 E, 5 3 6 2 1 7 8 4 s ~ t i 2 1 7 8 4 Fig. 2. Thin-laycr chromatograins of polymyxin E sulphate detected with ninhydrin ( a ) and with Numbers refer to sample number, method as stated in text. Bordetella hroizchiseptica NCTC 8344 ( b ) . l t o /ace page I070November, 1980 OF POLYMYXIN B AND POLYMYXIN E (COLISTIN) SULPHATES 1071 of each was equal. On this basis the bioautographic results indicated that the polymyxins containing 6-methylheptanoic acid (B2, E,) have about 3.5-5.0 times more specific anti- bacterial activity than the corresponding polymyxins containing 6-methyloctanoic acid The relative proportions of the fatty acids in the hydrolysed polymyxins determined by gas - liquid chromatography are shown in Table IV. The two major fatty acids found in polymyxins B and E, 6-methylheptanoic acid and 6-methyloctanoic acid, accounted for a t least 97% of the total fatty acids present, assuming that each had an identical response.Although the assay procedure was complex, the results (a typical chromatogram is shown in Fig. 3) were reproducible. The coefficients of variation for polymyxin E, sample 7, were 0.44, 3.14 and 6.59 for the determinations of El, E, and E,, respectively (n = 4) : the smaller the response the greater was the percentage variation. A good agreement of results was obtained by workers in two laboratories (Table V). (BIJ 'I)* TABLE I11 COMPARISON OF THE RELATIVE PROPORTIONS OF THE TWO MAIN COMPONENTS IN SAMPLES OF POLYMYXIN B AND POLYMYXIN E AS DETERMINED BY THIN-LAYER CHROMATOGRAPHY AND BIOAUTOGRAPHIC AND SPECTROPHOTOMETRIC ASSAY Sample No.1 2 4 5 6 7 9 Polymyxin B- Polymyxin E- Relative proportions, % ------ - Bl B2 Bl Bz * r -l Bioautographic assay Spectrophotometric assay 53.39 46.59 80.26 19.74 45.78 54.21 63.80 36.20 50.35 49.65 72.05 27.95 55.45 44.55 83.00 17.00 54.76 45.24 84.95 15.05 54.59 45.41 84.24 15.76 51.51 48.49 82.92 17.08 El E2 El Ez 49.29 50.71 75.46 24.54 65.69 34.31 91.76 8.24 51.89 48.11 81.40 18.60 55.66 44.34 75.61 24.39 67.22 32.78 94.80 5.20 70.7 1 29.29 93.44 6.56 The results of high-performance liquid chromatography demonstrated the heterogeneous composition of the polymyxins. The elution profile of each sample was characteristic and sufficiently distinct to identify the samples by manufacturer.Typical chromatograms are shown in Fig. 4. Retention times of the major components were 510 and 1110 s for poly- myxins B, and B, and 375 and 810s for polymyxins E, and El. The detection of the numerous minor components reduced the proportion of the two major components in each polymyxin to between 70 and 90%. Only two minor compounds (retention times 75 and 405 s) were detected in both polymyxin B and E. There was no evidence that the two major polymyxin E components were present in the samples of polymyxin B or vice versa. A quantitative evaluation of the components was made by assuming that each component had the same absorbance. The method was reproducible within the laboratory; coefficients of variation for polymyxin B, sample 5, ranged from 18.26 for the minor components to 1.37 for the major components (n.= 3) (the values being 13.56, -, 17.01, 18.26, 4.04, 7.87, 1.37, 12.47, 3.62, 1.79 and 1.66 for components 1-11, respectively). The relative compositions of the samples of polymyxin B and E are shown in Tables VI and VII. The correlation between gas - liquid and high-performance liquid chromatography for determining the proportion of each of the major components was good ( r = 0.93, 0.93,1072 THOMAS ef al. : MICROBIOLOGICAL AND CHEMICAL ANALYSIS Analyst, Vol. 105 TABLE IV RELATIVE COMPOSITION OF SAMPLES OF POLYMYXIN B AND POLYMYXIN E BASED Oh' THE CONTENT OF THE FATTY ACIDS DETERMINED BY GAS - LIQUID CHROMATOGRAPHY Relative composition, yo Sample No. 1 2 3 4 5 6 7 8 9 10 Polymyxin B- 77.50 21.40 1.10 0.00 68.52 28.28 3.18 0.11 68.65 27.58 3.62 0.14 75.14 21.41 3.41 0.03 84.03 12.58 3.19 0.19 84.64 12.47 2.68 0.20 76.60 22.36 0.96 0.00 76.10 23.30 0.60 0.00 81.39 17.89 0.70 0.08 72.34 25.53 2.12 0.02 Polymyxin E- El 1 68.84 2 84.42 3 74.40 4 74.50 5 75.39 6 64.60 7 90.70 8 91.01 =2 E, 30.14 1.00 13.67 1.75 23.20 1.70 23.50 1.80 22.80 1.78 34.60 0.80 7.10 2.20 6.87 2.10 E, 0.00 0.15 0.40 0.40 0.00 0.00 0.00 0.00 0.80, 0.99 for polymyxins B,, B,, E, and E,, respectively).There was no correlation between the total proportion of the two major components in each polymyxin determined by either gas - liquid or high-performance liquid chromatography and microbiological potency. 0 5 10 15 20 T ime/m in Fig. 3. Gas - liquid chromato- grams of fatty acids from poly- myxin E sul- phate, sample 3, method as in text.a, n-Hept- anoicacid; b, 6- methylheptanoic acid; c, n-oct- anoic acid; and d, 6-methyl- octanoic acid.November, 1980 OF POLYMYXIN B AND POLYMYXIN E (COLISTIN) SULPHATES 1073 Sample 2 u L 0 3 6 9 1 2 0 3 6 9 1 2 0 3 6 9 1 2 0 3 6 9 1 2 0 3 6 9 1 2 0 3 6 9 1 2 0 3 6 9 1 2 Tim& x lo2 conditions as stated in text. polymyxin E sulphate, conditions as stated in text. Tim& x 10' Fig. 4. (a) High-performance liquid chromatograms of samples of polymyxin B sulphate, (b) High-performance liquid chromatograms of samples of Discussion The disadvantage of the microbiological assay of the polymyxins is the lack of reproduci- bility of the potency determinations between laboratories. Where the dose - response lines of the standard and test are non-parallel the potency determinations are invalid and can be expected to vary from laboratory to laboratory.It has been our experience that the test organism used for the assay of the polymyxins, Bordetella bronchiseptica NCTC 8344, is liable to lose its sensitivity unless frequently sub-cultured. It is possible that different laboratories are using very different dose levels as the working ranges quoted in the pharmacopoeial- suggested method are wide (20-200 IU ml-l for polymyxin B and 50-5000 IU ml-' for poly- myxin E). If this is the case the relative potency of the sample in terms of the standard will vary according to the dose level at which the comparison is made. TABLE V RELATIVE COMPOSITION OF SAMPLES OF POLYMYXIN E BASED ON THE CONTENT OF FATTY ACIDS DETERMINED BY GAS - LIQUID CHROMATOGRAPHY RESULTS OBTAINED FROM TWO DIFFERENT LABORATORIES Laboratory 1 Laboratory 2 Sample f A > r h \ 1 84.42 13.67 1.75 0.00 85.54 13.40 0.93 0.00 3 74.40 23.30 1.70 0.40 73.33 24.60 1.49 0.50 7 90.70 7.10 2.30 0.00 90.40 7.49 1.91 0.00 number El E, E3 E, El E* E, E, The phenylalanine content is used as a quantitative assessment of polymyxin B; the theoretical content of phenylalanine in polymyxin B is 11.4%.Although the method allows for light-absorbing impurities it is non-specific because phenylalanine-containing impurities or free phenylalanine would be included in the determination. The absence of a correlation between the potency and the phenylalanine content of the polymyxin B samples demonstrates that these two parameters are not closely related.The greater specific antibacterial activity of polymyxin B, and the presence of other antibacterial components in polymyxin B, indi- cated by the thin-layer chromatographic results, could explain this lack of correlation. Thin-layer chromatography readily demonstrated the complex nature of the polymyxins. It can be used to identify polymyxin B and E and detect differences in the composition of samples of these antibiotics. The semi-quantitative assessment of thin-layer chromatograms by bioautography and densitometry was a useful means of comparing the specific anti- bacterial activity of the major components. Polymyxins containing 6-methyloctanoic acid (Bl, El) have been claimed to be more active than the corresponding polymyxin containing 6-methylheptanoic acid (B2, EJ,1 although when using Escherichia coli no differences in1074 THOMAS et al.: MICROBIOLOGICAL AND CHEMICAL ANALYSIS Analyst, VOl. 105 TABLE VI RELATIVE COMPOSITION OF SAMPLES OF POLYMYXIN B BASED ON PEAK AREAS OF COMPONENTS SEPARATED BY HIGH-PERFORMANCE LIQUID CHROMATOGRAPHY The values represent the percentage relative proportions of the components. Component number (retention time/s) Compound retention timeis 1 2 3 4 5 6 7 8 9 10 Sample number , , .. .. 1 0.30 0.70 0.64 0.52 0.72 0.85 0.41 0.35 0.31 0.42 (75) 2 0.25 0.33 0.34 0.81 0.00 0.00 0.14 0.29 0.26 0.31 (165) 3 (270) 0.38 1.38 1.34 0.73 0.b4 0.37 0.00 1.03 0.40 0.00 4 0.64 2.11 2.01 1.14 1.02 0.00 1.46 1.26 1.14 (300) 2.23 5 (405) 0.62 1.74 1.53 1.94 1.59 1.48 0.56 0.70 1.04 0.85 6 (450) 1.11 2.08 2.02 1.47 1.69 1.64 1.70 1.81 1.30 2.05 7 (510) ., 22.82 0.13 24.86 3.72 26.85 4.31 22.11 4.84 13.89 4.40 13.47 4.28 22.84 2.75 25.20 1.09 21.10 28.68 0.96 2.26 5.36 3.30 3.60 4.38 9.07 8.74 7.13 7.35 4.96 6.32 10 (1 110) 1x311 57.82 55.99 62.22 67.04 67.84 63.99 60.39 66.60 57.45 11’ (1 185) 2.55 1.50 0.80 0.00 0.60 0.58 1.47 1.11 1.79 1.09 antibacterial activity could be detected between the two major constituents of polymyxin B or polymyxin E.12 Our results indicated that the 6-methylheptanoic acid-containing poly- myxins were more potent than those containing 6-methyloctanoic against Bordetella bronchi- septica, but there was no correlation between the potency of the samples and their content of the 6-methylheptanoic acid-containing component.A similar difference in sensitivity between test organisms has been reportedla; the antibiotic activities of polymyxins B, and E, were almost equal against Escherichia coli, but against Brucella bronchiseptica polymyxin El was 1.6 times more active. We found polymyxin E to be twice as potent as polymyxin B. TABLE VII RELATIVE COMPOSITION OF SAMPLES OF POLYMYXIN E BASED ON PEAK AREAS OF COMPONENTS SEPARATED BY HIGH-PERFORMANCE LIQUID CHROMATOGRAPHY The values represent the percentage relative proportions of the components. Component number (retention time/s ) Compound retention time/s 1 2 3 4 5 6 7 8 9 10 1 1 1 2 13 , Sample number , . , . . . (75) (210) (240) (285) (315) (375) (405) (465) (570) (645) (810) (960) (I 140) 0.00 0.00 1.90 0.54 1.39 31.89 0.00 2.08 0.00 4.07 58.30 0.85 0.00 0.00 0.75 1.15 2.16 0.00 11.89 2.22 2.67 2.73 4.28 64.19 2.11 4.59 0.77 1.74 2.00 2.52 2.68 22.96 2.29 5.38 4.34 4.24 49.68 0.00 0.00 0.11 0.90 1.36 1.86 3.12 22.67 2.03 5.37 3.97 5.02 52.43 0.51 0.00 0.19 1.13 1.98 1.64 1.91 22.42 0.00 7.52 3.80 6.01 48.29 1.05 2.03 0.00 0.00 2.41 0.00 2.09 35.61 0.00 4.29 0.00 3.70 50.65 0.50 0.00 0.00 0.21 1.66 0.19 1.28 7.33 1.86 7.60 0.00 13.37 66.43 0.12 0.00 0.00 0.26 0.84 0.76 1.37 7.24 1.53 8.50 0.00 13.25 66.46 0.80 0.80 The assay of the fatty acids by gas - liquid chromatography was a very reproducible and reliable method for determining the relative proportions of the main polymyxin components present in a sample.However, because of the hydrolysis and extraction procedures it would be difficult to adapt this method to the quantitative determination of the polymyxins on a mass basis.The results of the high-performance liquid chromatography were repro- ducible and the method was uncomplicated. The additional unidentified components made interpretation of the results difficult although the correlation of the major components by gas - liquid and high-performance liquid chromatography was good. The atypical compo- sition of samples (7 and 8) of colistin was only revealed by thin-layer chromatography and high-performance liquid chromatography. High-performance liquid chromatography provides the most information on the composition of samples of polymyxin B and E and it would be the method best suited to monitoring the composition of polymyxin samples in order to ensure that material of a known composition is used.However, because of the complex nature of the polymyxins, it is difficult to suggest a suitable chemical assay to replace the microbiological assay. In order to define the desired composition of the polymyxins reliable data would be required on the toxicity, antibacterial activity and sensitivity to the assay procedure of the individual components.November, 1980 OF POLYMYXIN B AND POLYMYXIN E (COLISTIN) SULPHATES 1075 We thank Dr. A. Islam of the British Pharmacopoeia Commission Laboratory for details of the gas - liquid chromatographic method and for allowing analyses to be performed in his laboratory. References 1. 2. 3. 4. 5. 6. 7 . 8. 9. 10. 1 1 . 12. 13. Storm, D. R., Rosenthal, K. S., and Swanson, P. E., Annu. Rev. Biochem., 1977, 46, 723. Withander, L., and Heding, H., J . Antibiot., 1976, 29, 774. Lightbown, J . W., Thomas, A. H., Grab, B., and Outschoorn, A. S., Bull. W.H.O., 1973, 48, 85. Haemers, A., and De Moerloose, P., J . Chromatogr., 1970, 52, 154. “British Pharmacopoeia 1980,” HM Stationery Office, London, 1980, p. 125. Thomas, A. H . , and Holloway, I., J . Chromatogr., 1978, 161, 417. Tsuji, K.. and Robertson, J . H., J . Chromatogr., 1975, 112, 663. Terabe, S., Konaka, R., and Shoji, J., J . Chromatogr., 1979, 173, 313. Fong, G. W., and Kho, B. T., J . Liq. Chromatogr., 1979, 2, 957. Lightbown, J . W., Broadbridge, R. A., Isaacson, P., Sharpe, J . E., and Jones, A , , Analyst, 1979, Thomas, A. H., and Thomas, J . M., J . Chromatogr., 1980, 195, 297. Nakajima, K . , Chem. Pharm. Bull., 1967, 15, 1219. Vogler, K., and Studer, R. O., Experientia, 1966, 22, 345. 104, 201. Received M a y 15th, 1980 Accepted June 9th, 1980

ISSN:0003-2654    

DOI:10.1039/AN9800501068

出版商:RSC    

年代:1980

数据来源: RSC