ISSN:0003-2654    

DOI:10.1039/AN99116FX037

出版商:RSC    

年代:1991

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN99116BX039

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 975 Analytical Quality Assurance A Review Robert J. Mesley, W. Dennis Pocklington and Ronald F. Walker" Laboratory of the Government Chemist, Queen's Road, Teddington, Middlesex NVI I OLY, UK Summary of Contents I nt rod uct ion The Need for Improved Quality Principles of Analytical Quality Assurance Introduction Harmonization of lnterlaboratory Study Protocols Models for Collaborative Studies Assessment of the Accuracy of a Method Statistical Analysis of lnterlaboratory Study Results Presentation of the Performance Characteristics of Analytical Methods Acceptance Criteria for the Adoption of Validated Methods Drafting of the Texts of Standardized Methods Introduction Traceability Criteria for Developing Reference Materials Methods of Certification Contents of Certificates Use of Reference Materials Quality Control of Sampling Internal Quality Control Quality Control of Analysis External Quality Assessment Assessment by Inspection lnterlaboratory Comparisons Analytical Methodology Reference Materials Conclusions References Keywords: Review; quality assurance; statistical analysis; validation Introduction The Need for Improved Quality The literature of analytical chemistry consists largely of papers describing the development of analytical methodology and others presenting the results of applying such methods to particular problems.Those of the first type are frequently confined to the analysis of standard solutions or mixtures, and recovery and precision data are often based on simulated, rather than real, samples.Those of the second type may present data on numerous samples but in many instances give little or no independent evidence of their accuracy. Few published papers have quality assurance of analytical data as their main topic. It is no doubt common knowledge amongst those who practise analysis in support of trade and commerce that analysts can obtain different results on the same material, but it may not be in their interests to publicize that fact. It is in the fields of public health and environmental monitoring, where determinand concentrations are often low and where small differences may be significant, that interlaboratory variation has received most attention. During the past decade numerous organizers of inter- laboratory studies have commented that the data obtained were inadequate for their intended purpose, but they offer a * To whom correspondence should be addressed.variety of explanations. Hendzel et al.,1 in a study of the determination of heavy metals in clam tissue by atomic absorption spectrometry (AAS), obtained acceptable results for copper and zinc but found wide interlaboratory variation in the results for cadmium and lead; they concluded that the method was inadequate for the latter metals at the levels at which they were present in the samples. Lack of validated standard methods was blamed for variation in results for heavy metals in river water2 and for organic matter in rocks.3 In a study of polychlorinated biphenyls in contaminated sedi- ments4 standardized procedures were prescribed, as were calibration standards, but the workers concluded that there was still scope for variation.Other studies5-6 allowed partici- pants a free choice of method, and their organizers attributed the poor reproducibility to lack of suitable reference materials, a plea echoed by others.3-7.8 In assessing the acceptability of the results of collaborative studies, the empirical relationship between relative standard deviation and concentration demonstrated by Horwitz and co- workersg-11 is increasingly being used as a yardstick. Few workers are prepared to state that the cause of inadequate data was poor performance by individual labora- tories, perhaps because they rely on the goodwill of partici- pants to carry out the studies. Knochel and Petersen,12 studying variation in results for heavy metals in Elbe river water, commented that the availability of a reproducible method does not yet imply accurate results.Sherlock et al. ,l3 having established acceptable performance levels for the976 ANALYST, OCTOBER 1991, VOL. 116 determination of lead and cadmium in foods from results obtained by official laboratories, found that only 3 out of 27 other laboratories achieved these performance levels, a statement which provoked spirited correspondence in Chem- istry in Britain.“+-22 Mason23 was even more forthright in condemning laboratories engaged in toxicological analysis for failing to meet acceptable standards. Where such unreliable results form the basis of government or other official action the risk of incorrect decisions being taken can be high.There are also economic consequences. It was reported in 197524 that in the United States 10% of all clinical analyses had to be repeated, at an annual cost of about $900 million. In Germany losses due to poor performance by analytical laboratories were estimated in 1984 to be in the region of 1 billion DM.25 Poor performance in interlaboratory studies poses problems for those seeking to obtain consensus values for the certifi- cation of reference materials (RMs). Initial intercompari- sons conducted by the Community Bureau of Reference (BCR)26-29 sometimes exhibit ranges of results extending over several orders of magnitude; refinement of analytical tech- niques, coupled with the exclusion of outliers (and, if necessary, exclusion of aberrant laboratories), usually leads to acceptable reproducibility.Pszonicki,30 reviewing procedures used by the International Atomic Energy Agency (IAEA), concluded that the exclusion of poorly performing labora- tories could result in insufficient results for statistical con- fidence; on the other hand, inclusion of the results from these laboratories, after removing gross outliers, often yields skewed distributions, with the median providing better consensus values than the mean. Despite the horror stories, several workers have reported improvements over earlier studies. Subramanian and Stoeppler7 considered results for lead in blood by electrother- mal AAS to be improved. Byrne et aZ.31 were able to improve on previous certification figures for ‘difficult’ trace elements in two IAEA reference materials, the new figures being in some instances an order of magnitude lower than those previously published.Similarly, Topping32 found that published values for trace metals in sea-water had diminished by an order of magnitude between 1965 and 1983. Such changes in accepted baseline values make it exceedingly difficult to confirm the effects of pollution on concentrations of environmentally sensitive constituents. To combat such problems in the field of water analysis, analytical quality control procedures for harmonized monitoring of river water2.33-41 and for saline waters42 have been introduced in the UK. Several workers43-45 conclude that taking part in interlaboratory comparisons has the effect of improving the proficiency of participating laboratories.However, Horwitz et aZ.9 showed that participation in check sample schemes improved per- formance for the first few years, but thereafter it remained static or even tended to deteriorate.10 Attention has been focused on the need to ensure the quality of analytical data by the increasing adoption of the principles of laboratory accreditation, for example the National Measurement Accreditation Service (NAMAS) in conformance with International Organization for Standard- ization (ISO) Guide 25.46 In the European Community progress towards the completion of the single market will require adherence to harmonized standards and mutual acceptance of data, if barriers to free movement of goods between member states are to be removed.In this context analytical quality assurance will assume increasing import- ance. Schemes for external assessment and accreditation of the quality of laboratory work47-51 generally attach great import- ance to the management of the laboratory, strict control and documentation of operating procedures, instrument calibra- tion, record keeping, staff training, etc. These features are common to all kinds of measurement and testing, but there are other factors which are peculiar to the chemical area. In particular the difficulty in tracing the calibration of most analytical procedures to recognized standards has stimulated the development of certified reference materials and of hierarchies of methods that can be traced to a definitive method. Interlaboratory studies have proved valuable, not only for monitoring public health and environmental prob- lems, but also for validating methods, certifying reference materials and monitoring the performance of individual laboratories.It is these aspects which provide most of the published literature on analytical quality assurance. Principles of Analytical Quality Assurance The principles of quality assurance as applied to analytical laboratories have been outlined in a number of published texts, most of which are based either on training courses or on systems developed in particular organizations.52-55 The topics covered by such texts encompass virtually every aspect of analytical chemistry, but there is a measure of consensus that the more important features of an analytical quality assurance programme include: (i) the use of validated methods; (ii) properly maintained and calibrated equipment; (iii) the use of reference materials to calibrate methods; (iv) effective inter- nal quality control (control charts, etc.); (v) participation in interlaboratory check sample schemes; (vi) independent audits of quality control procedures; (vii) external assessment by accreditation or other compliance schemes; and (viii) properly trained staff. Most of these general principles have also been summarized in a number of other paper~.9J8.22~5-* Published literature on individual aspects is reviewed in the following sections.Analytical Methodology Introduction The quality performance of analytical laboratories is directly related to the use of validated or standardized methods.Here ‘validated’ and ‘standardized’ should be understood to indi- cate that the performance characteristics of the method, for example in terms of precision, have been demonstrated to meet certain specified requirements. It would appear that the key role the analytical method plays in the quality perform- ance of analytical laboratories has not been fully appreciated. Whilst the expertise of the analyst and the use of high-purity reagents and equipment all contribute to the quality of results produced by a laboratory, the value of these results can readily be undermined by the employment of unsatisfactory methods of analysis. Satisfactory laboratory in-house quality control can only be achieved by ensuring that validated methods are used and the defined method procedure is meticulously followed.Such methods may have been developed by the laboratory itself or have been published by standards organizations. The non- availability of satisfactory methods for the determination of certain analytes or parameters has often resulted in the use of methods having dubious performance characteristics. Even published standard methods cannot always be relied upon, as many such methods have not been subjected to rigorous validation. Where an interlaboratory study has been carried out it has often been shown that the study was not properly organized with the result that the performance characteristics of the method were not correctly assessed. The Association of Official Analytical Chemists (AOAC) has a well-deserved excellent reputation for publishing in its journal methods which have been validated by collaborative study, and simultaneously, a full report on the study.The final texts of the methods are subsequently reproduced in the AOAC’s book of Official Methods of Analysis.62 The statistical principles used in evaluating results of AOACANALYST, OCTOBER 1991, VOL. 116 977 collaborative studies are well established and fully docu- mented.63.M One might assume that all methods published by the ISO, and national standardizing bodies such as the British Stan- dards Institution (BSI), have been validated by collaborative study. However, to take IS0 Technical Committee (TC) 34 (Agricultural Food Products) as an example, it was not until 1981 that the attention of its sub-committees was drawn to the necessity of ensuring that all methods being considered for adoption as I S 0 standards were first thoroughly validated, by properly conducted collaborative studies, for their perform- ance in terms of precision before being published as definitive standards.An I S 0 TC 34 committee document circulated in 198665 indicated that the number of I S 0 methods (for the analysis of food and agricultural materials) supplied with precision data is relatively low for general methods, and that very few of the reference methods have such data. This is surprising as I S 0 first convened a working group of experts in 1971 to develop a guide which eventually appeared in 1974 as ISODIS 5725 under the title ‘Precision of Test Methods. Part 1.Guide for the Determination of Repeatability and Reproducibility for a Standard Test Method’.% The I S 0 TC 69 (Applications of Statistical Methods) is currently involved in a substantial revision of I S 0 5725-1986 and the revised standard is to cover much more than the determination of the precision charac- teristics of a method (Table 1). As an example of British Standards (BS), examination of BS 684 (Methods of Analysis of Fats and Fatty Oils)67 shows that the situation is only marginally better than that with I S 0 standards. Only 14 of the 50 or so methods in the Standard contain any information regarding precision, and it would appear that in many instances arbitrary repeatability values have been cited, rather than values derived from the statistical evaluation of collaborative study results. It is a cause for considerable concern that many published standard methods are not entitled to their assumed status as definitive validated methods. Furthermore, even where some reference is made, in the text of a standard method, to its precision, such information is often intelligible only to a statistician. Greater uniformity in the presentation of statis- tical data is urgently needed.The following sections consider recent progress in the harmonization of protocols for: (1) the organization of interlaboratory studies designed for the validation and stan- Table 1 Accuracy (trueness and precision) of measurement methods and results (IS0 5725-long-term revision in the course of prepara- tion) Part 1: Part 2: Part 3: Part 4: Part 5: Part 6: General principles and definitions A basic method* for the determination of the repeatability and reproducibility of a standard measurement method Measures of precision intermediate between repeatability and reproducibility Section 1: Calibration and recalibration Section 2: Changes within a laboratory method? trueness and precision Section 1: Introduction Section 2: Calculating repeatability and reproducibility limits and other limits Section 3: Acceptability of results Section 4: Stability of results within laboratories Section 5 : Assessment of laboratories Section 6: Comparison of test methods A basic method for determination of the trueness of a test Test methods alternative to the basic methods for determining Practical applications of trueness and precision measures * The basic method described is that applicable to results obtained t Includes the use of CRMs. from a uniform-level collaborative study.dardization of methods of analysis, and (2) the presentation of information on analytical quality control (QC) (including statistical parameters related to performance characteristics) in the texts of standardized methods. Harmonization of Interlaboratory Study Protocols A standards organization considering for adoption a method that has already been published by another organization, should satisfy itself that the method has been validated by a properly organized collaborative study. It would also need to know whether the precision of the method has been evaluated by an accepted method of statistical analysis of the results obtained during the collaborative study.Lack of uniformity in the standards organizations in their approach to method validation has impeded the adoption process and in certain instances has resulted in the duplication of analytical work. The need for harmonization in the approach to collabora- tive studies of methods was highlighted in 1980 at a meeting of the major standards organizations involved in the conducting of interlaboratory studies. This meeting, held in London under the auspices of the International Union for Pure and Applied Chemistry (IUPAC), led to a symposium on the Harmonisation of Collaborative Analytical Studies at Helsinki in 1981,68 followed by another symposium at Washington in 1984. The final outcome of these international meetings was the publication in 1988 of the IUPAC Protocol for the Design, Conduct and Interpretation of Collaborative Studies.@ This protocol (the principles of which have been incorpor- ated in the AOAC Guidelines for Collaborative Study Procedure to Validate Characteristics of a Method of Analy- sis)70 provides standards organizations with specific recom- mendations for meeting certain minimum requirements for collaborative studies. A summary of the main recommenda- tions of the protocol is given in Table 2, but three of the most important recommendations can be mentioned here, viz. : (a) not less than five materials should normally be provided for analysis by participants in the collaborative study; (b) a required minimum of eight laboratories reporting valid data for each material analysed; and (c) two analyses to be carried out on each material, this replication being procured by blind duplicates or split-levels.However, a recent paper by Karpin- ski71 makes the claim that ‘the assessment of repeatability may be adequately carried out with a collaborative study involving as few as three laboratories’. Other protocols outlining procedures for conducting collab- orative studies have been published during the last five years by the International Dairy Federation (IDF),72 the Nether- lands Normalisation Institute (NNI)73 and the Codex Alimen- tarius Commission.74 The NNI protocol has been recom- mended for use by the analytical methods working groups of the International Office for Cocoa, Chocolate and Confectionery Products (IOCCC) .Models for Collaborative Studies Studies organized over the last five years or so have generally been based on one of five designs, viz.: (1) a uniform-level design in which the analyst is required to obtain two results for each test material provided; (2) a uniform-level design in which independent duplicate results for each test material are obtained by providing identical (but unidentified) samples in duplicate (‘blind duplicates’); (3) a ‘split-level’ design with its ‘Youden-pairs’a designed to overcome possible analyst bias arising from known or identification of blind duplicates; (4) a ‘double split-level’ design72 introduced by the NNI, this design was intended to overcome what has been claimed by Zaal- berg75 to be serious shortcomings of the I S 0 5725 uniform- level and normal split-level designs, in the NNI design two identical samples (blind coded) are provided for each of the sub-levels; and (5) a replicate analysis design advocated by the978 ANALYST, OCTOBER 1991, VOL.116 Table 2 IUPAC-1987 harmonized protocol for the design, conduct and interpretation of collaborative studies (only the main requirements outlined in the document are given below) The results of collaborative studies should be analysed by 1-way analysis of variance (‘material by material’), but more complex analyses are not precluded. The absolute minimum number of materials to be used in a collaborative study is five. However, when a single level specifica- tion is involved this may be reduced to an absolute minimum of three.The most important objective of a collaborative study should be attaining a reliable estimate of reproducibility* parameters. To the extent that the performance of known (parallel) replicate analyses detracts from this objective, a requirement for the use of this type of replicate analysis should be discouraged. The best estimate of repeatability* parameters is obtained by the following procedures (listed approximately in order of desirability). (a) The use of a split-level design (single values from the analysis (b) The use of both split-levels and blind duplicate analyses in the (c) The use of blind duplicate analyses. (d) The use of known duplicate analyses (two repeat analyses from the same test sample), but only when it is not practical to use one of the preceding designs.The minimum number of participating laboratories in a collabora- tive study is eight. Only when it is impossible to obtain this number may the study be conducted with less, but with an absolute minimum of five laboratories. (Although it is desirable to have more than eight laboratories participating, studies containing more than about fifteen laboratories become unwieldy.) The precision estimates are to be calculated both with no outliers removed and with outliers removed, using the Cochran and Grubbs outlier tests. The Grubbs tests should be applied only to laboratory means, not to individual values of replicated designs. Outlier removal should be stopped when more than 22% (ie., more than two out of nine laboratories) would be removed as a result of the sequential application of the outlier tests.When standard deviations are expressed in relation to the absolute value of the arithmetic mean the term recommended is relative standard deviation (RSD). The equivalent IS0 term is coefficient of variation (CV). * As defined in IS0 5725-1986. of each of two closely related materials). same study. German Federal Office of Health in which five or more analyses are required for each sample provided. The replicate analysis design has been criticized on the grounds that many participants in the study would work towards submitting results which are as close together as possible, thereby introducing a replication bias which would inevitably lead to an incorrect estimate being made of the precision of the method.However, the protocols adopted for the development (by collaborative study) of certified refer- ence materials (CRMs) by the BCR require the multiple analysis of samples (generally not less than five replicate analyses are required to be carried out by each laboratory involved in the certification exercise). However, in the BCR studies the objective is not to assess the precision of analytical methods but to determine the concentration level of an analyte to the greatest degree of accuracy possible. During the certification exercise an endeavour is made to obtain, whenever possible, results using more than one method in order to prevent the certified values being ‘method depen- dent’. The IUPAC 1987 Protocol lists in ‘approximate order of desirability’ the split-level model, a model based on a combination of blind replicates and split-level, and a blind replicates model; a model requiring known replicates is listed last with the caution that such is not to be used unless it is impractical to use one of the foregoing designs.In view of the fact that the split-level model is recommended as the model of choice, it is perhaps somewhat incongruous that the future revision of IS0 5725 is to designate the split-level model as merely an ‘alternative’ to that of the uniform-level model. Assessment of the Accuracy of a Method International collaborative studies of methods have generally been confined to the determination of precision only. The reason for this would appear to be that it is not always feasible to assess the accuracy (trueness) of a method. This is a problem especially with empirical methods, where the end result is method dependent and is not an entity which can be defined in absolute terms. It is essential that the text of the method gives not only an indication as to the precision of the procedure, but also information regarding the accuracy (truenesshias), the limit of detection and the sensitivity in respect of the analyte to be determined.Furthermore, it is essential to state whether the procedure is subject to interference from materials likely to be present in the samples to which the method of analysis may be applied. These requirements could only be met if they were taken into account when the protocol of the collaborative study designed to investigate the method was drawn up.It would be necessary to provide ‘spiked’ samples (i.e., samples with known amounts of added analytes) and samples contain- ing substances which could be a possible source of interfer- ence, in order to establish the accuracy of the method in the presence of interfering materials. In some samples the levels of analyte for determination would need to be close to the expected limit of detection. In methods published in the UK by the Standing Committee of Analysts of the Department of the Environment (DOE) these important factors are taken into account, for example, see the DOE method for Arsenic in Potable Waters by Atomic Absorption Spectrophotometry.76 The section on the Per- formance Characteristics of the Method for this procedure cites the limit of detection for arsenic.As to the sensitivity of the method, the user of the method is informed that under the conditions of the procedure, arsenic, at a level of 2.0 pg 1-1, gives an absorbance of approximately 0.10. Interferences, such as copper, silver and selenium are listed in another section of the method, together with an indication of the levels above which these elements, when present, may interfere with the determination of arsenic. Volumes of collated analytical methods for pesticides and other hazardous substances at pg kg-1 concentrations have recently been published (collated in book form) in Germany;77978 the book contains information on the limits of determination, sensitivity, limits of detection, along with other performance data, of the methods.Unfortu- nately the differences in viewpoint as to how the detection limit should be defined has contributed to neglect in its use by analysts.79 The use of CRMs for establishing the truenesshias of an analytical method has been restricted by the limited availabil- ity of such materials. Certified reference materials have an important role to play in this context and in view of this the revision of IS0 5725 is to include a section outlining procedures for determining the trueness of a method by the use of reference materials. One of the major difficulties frequently encountered when attempting t o establish the accuracy of a method by collabora- tive study is the provision of samples in which the concentra- tion of the analytes, required to be determined, can be guaranteed to be stable during the course of the collaborative study; this is particularly difficult with labile compounds.The organization of collaborative studies on methods for micro- biological analysis presents serious problems of a kind not normally encountered with chemical methods and for this reason the drafting of guidelines for such studies is now being undertaken by the AOAC. The inherent difficulties in providing samples which will be microbiologically stable will be immediately obvious, and this means in effect that it may beANALYST, OCTOBER 1991, VOL. 116 979 virtually impossible for all participants in the study to be provided with samples of identical composition. Table 3 Example of a statistical report on an interlaboratory study.(The table reproduced below is an amended version of that published in Pure AppE. Chern., 1988, 60,890.) Statistical analysis according to I S 0 5725-1986. Values for reproducibility limit (R) are those applicable to results obtained from a single analysis of an identical laboratory sample by different laboratories Statistical Analysis of Interlaboratory Study Results The 1989-IUPAC Protocols recommend that the principles outlined in IS0 5725 and the 1987-IUPAC protocol should be followed for the statistical evaluation of results obtained from collaborative studies of candidate standard methods. However, many co-ordinators of working groups developing standard methods have found the statistical evaluation of collaborative study results difficult (despite the availability of IS0 5725) and for this reason a simplified approach to the statistical analysis of such results based on the I S 0 5725 procedure was specially prepared for inclusion in the Guide- lines for the Development of Standard Methods by Collabora- tive Study.80 This publication also provides a comprehensive guide to the organization of method performance studies and draws attention to the increasing number of computer programs now available which have been designed for the statistical analysis of collaborative results.Statistical parameters which are of the greatest value in assessing the precision of a method are the estimates of the repeatability/reproducibility standard deviations (s,, sR) . These give an indication of the spread of results experienced when the method was collaboratively studied.In recent years the number of validated standard methods which cite repeatability/reproducibility values/limits ( r , R) in the pre- cision clauses of the methods has steadily increased. These values (which are related to a 95% probability criterion) are derived by multiplying the values for s,, sR by the factor 2 i.e., 2.83 (now generally rounded to 2.8). This criterion means that the values for the differences between results obtained from the analysis of a sample would not be expected, in more than 5% (1 in 20) of instances, to be greater than the values for r (for within-laboratory determinations carried out under repeatability conditions), or R (for between-laboratory deter- minations carried out under reproducibility conditions). Unfortunately the best use of these values in the precision clauses of many published standard methods has not always been made.The Analytical Methods Committee of the Royal Society of Chemistry has questioned the concept that repeatability and reproducibility are properties of the method.61 They point out that the within-laboratory variance of results is often markedly different from one laboratory to another, while the concept of reproducibility as random error ignores the fact that between- laboratory variance represents bias and depends on the selection of laboratories to take part in the trial. These comments might be countered by the view that between- laboratory variation is in fact a reflection on the ruggedness of the method.Statistical analysis of results for tocopherols by high-performance liquid chromatography Sampleflevel A B c D Number of laboratories Number of results Number of laboratories retained after elimination of outliers results Number of accepted Mean value/pg g-I Repeatability standard deviation, s,* Repeatability coefficient of variationt Repeatability limit (r)* (2.8 x s,) Reproducibility standard deviation (sR)* Reproducibility coefficient of variationt Reproducibility limit (R)* (2.8 X SR) 16 32 17 34 16 32 17 34 14 15 13 16 26 263 32 508 28 17 30 69 0.8 3.5 5.3 12.6 2.5% 2% 15 2 5.1 10 17.5 36 36.0 7% 44.5 31% 17% 15 49 126 102 * Values expressed as micrograms of tocopherol per gram of t- Relative standard deviation. sample. future will have associated with them a table of statistical parameters similar to the example given in Table 3.However, reproducing this table in the text of a validated standard method will only be of value to the analyst if the clauses dealing with the precision of the method clearly indicate how the statistical parameters are to be interpreted. The role of precision clauses in standardized methods and the most practical way of drafting them was therefore considered in depth by the IUPAC Workshop. It concluded that informa- tion pertaining to the performance characteristics of a standardized method should not form part of the main text of the method but be reproduced in a separate section covering the principles of analytical QC and presented as an appendix to the method. Acceptance Criteria for the Adoption of Validated Methods The adoption of methods by standards organizations would appear to be, in many instances, very much on an ad hoc basis; before recommending adoption some organizations have not always first established whether the method has been ade- quately studied and its precision shown to meet a required standard.In the absence of guidelines it has often been difficult €or standards organizations to decide whether a method warrants adoption or not. Honvitz'o reported on the average relative standard deviation (RSD) (coefficient of variation) values calculated for a considerable number of AOAC methods that had been collaboratively studied during the past decade or so, and established a mathematical relationship between these values and the concentration of the analyte.It is considered that this criterion could usefully be taken into account when deciding whether a particular method has the degree of ruggedness expected of a standard method and the IUPAC-1989 Protocols indicate how this criterion can be applied. A full consideration of the reproducibility standard deviation (RSDR) criterion in the context of collab- orative studies for milk and milk products methods of analysis has recently been published.82 Presentation of the Performance Characteristics of Analytical Methods Recently there have been discussions, on an international level, regarding the extent to which statistical information can be usefully included in the texts of standard methods. The discussions took the form of the circulation of a questionnaire to organizations involved in the publication of standardized methods and the subsequent consideration of responses to the questionnaire at a Workshop convened by IUPAC in Wash- ington (1989). The conclusions reached at the workshop resulted in the publication of the 1989-IUPAC Protocols for the adoption of standardized analytical methods and for the presentation of their performance characteristics.81 The need for harmonization in this area has been recognized for many years and if the IUPAC-1989 Protocols are adopted by standards publishing organizations it can be expected that the texts of most standard methods to be published in the980 ANALYST, OCTOBER 1991, VOL.116 Start with two test results (xl, x2) Table 4 Format of a standard method.The format is based, in part, on IS0 7812, Layouts for Standards Part 2: Standard for Chemical Analysis. (Note: This IS0 standard is in the course of revision) 1. Scope- This states briefly what (i.e., analyte/parameter) the method determines 2 . Definition- Here a precise definition of what is to be understood by the analyte or parameter referred to in the Scope as being determined by the method 3. Field of application- This section indicates the type of materiavmatrix to which the method is applicable The principle outlines the basic steps involved in the procedure Specific apparatus required for the determination is listed but not that which can be expected to be available in a reasonably well-equipped analytical laboratory Analytical-reagent grade reagents are specified where this is considered desirable; in certain circumstances it may also be necessary to specify the level of purity of the water, e.g., glass-distilledlde-ionized, that should be used A reference to the importance of the laboratory sample being taken according to a standard procedure should be included here together with recommendations as to the storage of the laboratory sample This section is divided into numbered paragraphs or sub-clauses for the sake of clarity and to allow reference to be made to certain steps in the method at earlier or later stages of the procedure.It should include a ‘preparation of test sample’ clause and a reference to ‘quality assurance’ procedures This section indicates how the final results are calculated and the units in which the results are to be expressed In certain instances it may be considered advisable to provide additional information as to the procedure. Such information can take the form of notes which may be placed here or incorporated in the main body of the text (as is generally the case for the texts of IS0 methods) This should include information on analytical quality control, e.g., precision clauses citing repeatability and reproducibility values (limits), a table of statistical data outlining the accuracy (trueness and precision) of the method as established by collaborative study, and also any required illustrations of apparatus, tables required for calculation of results, etc. A reference to the published report on the interlaboratory study which was carried out prior to the standardization of the method must be included 4.Principle- 5. Apparatus- 6. Reagents- 7. Sampling- 8. Procedure- 9. Calculation and expression of results- 10. Notes- Annexe- References- Following the adoption of the validated method by the standardizing body, the final task (and probably one of the most important), is to ensure that the text of the method is drafted in a format which the user of the method will find the most practical. It should be added that all of the above discussion relates to methods of quantitative analysis. There remains a need to establish criteria for acceptance of qualitative methods, i.e., those used primarily to detect the presence of particular analytes in test samples. Workers in the Netherlands83 have proposed criteria for the identification of analytes by various spectroscopic and chromatographic techniques, some of which have received recognition in official European Community methods.Obtain two more test results L I p = median of (xl, x2, x3, x4) [i.e., (x2 + x&2] Fig. 1 Flow diagram illustrating the derivation of the final quoted result when test results do not meet the conditions of repeatability. r = Repeatability limit for the method; and p = final result quoted. x2, x3, are, respectively, the second and third smallest test results. [The flow diagram has been reproduced (with some modification) from IS0 5725: Accuracy (Trueness and Precision) of Measurement Methods and Results Part 6. Practical Applications: Section Three-Methods for Checking the Acceptability of Results and Determining the Final Quoted Result] Drafting of the Texts of Standardized Methods An outline of the recommended I S 0 format for the text of a standard method, amended in line with the IUPAC-1989 Protocols, is reproduced in Table 4.This format is based on that outlined in the IS0 Standard 78/2 (currently undergoing revision).W It is inappropriate to consider here in any detail the drafting of standard methods but three aspects merit particular attention: the inclusion in the text of the method of clauses pertaining to the number of determinations, the expression of results and the precision of repeatability and reproducibility. There is at present little harmonization in the way that these clauses are drafted but the need for harmoniza- tion in their presentation was considered by the IUPAC Workshop and the IUPAC-1989 Protocols give firm recom- mendations as to the most appropriate format for precision clauses. Some standard methods contain a clause instructing the analyst to perform two analyses on the same test sample.In view of the difficulty of catering for all possible situations the IUPAC-1989 Protocols recommend that the text of a standard method should not include a clause pertaining to the number of determinations. However, quality assurance procedures followed in the analytical laboratory should give clear guidance on the approach to checking quality performance of the method by the use of duplicate or replicate analyses of the test sample. In clauses relating to the expression of results, some standard methods instruct the analyst to report as the final result the arithmetic mean of two single test results (obtained under the conditions of repeatability) ‘provided the requirements of repeatability are met’, and if the requirements of repeatability are not met ‘to discard the results and obtain two more test results from a further duplicate analysis of the same test sample’.This is inadequate as it does not cover the situation where the results from two further determinations do not meet the requirements of repeatability. A satisfactory way of dealing with poor dupli- cate results is outlined in Section 3 of Part 6 of the long-term revision of IS0 5725. This section is concerned with theANALYST, OCTOBER 1991, VOL. 116 981 determination of the acceptability of results.The 1989- IUPAC Protocol reproduces a flow diagram (adapted from the I S 0 standard) illustrating an approach to dealing with poor replicates (Fig. 1). Reference Materials Introduction Many detailed reviews have already appeared in the litera- ture85-93 to describe specific reference material categories. It is not the intention of this review to repeat such exercises, but to confine attention to the fundamental aspects and characteris- tics which are common to all reference materials. Such materials were first produced by the National Bureau of Standards (NBS) [now the National Institute of Standards and Technology (NIST)] in the USA in 1906, and were followed by Germany in 1912 and Britain in 1916. The United States reference materials programme arose from a need for better quality control of the cast iron used to make rail wheels following a series of derailments due to wheels breaking and both the German and British programmes were similarly associated with the steel industry.Since these beginnings there has been a significant increase in the importance and usefulness of reference materials, particularly since the 1950s and the accompanying rapid evolution of analytical instrumentation, and there are now about 120 producers (commercial, private and public) in the western world.85 The large increase in the number of reference materials being produced led in 1975 to an I S 0 Council Committee on Reference Materials (ISO-REMCO) being set up in order to establish international guidelines94-96 on principles of certifi- cation, methods of use, needs, availability and nomenclature, which include the following definitions?4 reference material (RM), a material or substance one or more properties of which are sufficiently well established to be used for the calibration of an apparatus, the assessment of a measurement method, or for assigning values to materials; and certified reference material (CRM), a reference material one or more of whose property values are certified by a technically valid procedure, accompanied by or traceable to a certificate or other documentation which is issued by a certifying body.Usually, the property values of RMs are established on the basis of accuracy (i. e . , demonstrated relationship or traceabil- ity to the ‘true’ value of the property being measured).Often, however, such measurement is not possible. This is always the situation where the relationship between some property and the parameter being measured cannot be established from a sound, theoretical basis that can be expressed in mathematical terms. For example, there is no complete theory that relates the composition of a synthetic rubber to its elastic properties. For systems involving measurements of this type, measure- ment compatibility ( i . e. , measurements agreeing within the limits of uncertainty which reflect the end-use require- ments97-99) is best achieved through the use of arbitrarily selected RMs from a single source based on highly standard- ized methods of analysis and/or tests. In fact, insistence upon following the measurement protocol exactly is usually a prerequisite for successful measurement compatible results.Many ISO, American Society for Testing and Materials, BSI or Deutsche Industrie Norm procedures are in this category. Traceability In an absolute sense the ‘true’ value can be defined only as being that value directly traceable to the base system of measurement units [Systeme International d’Unites (SI)] or their derivatives27.100.101 although, more generally, IS0102 defines traceability as the property or result of a measurement whereby it can be related to appropriate standards, generally international or national standards, through an unbroken chain of comparisons. In the area of chemical measurements, the most often encountered SI base units are the kilogram, the mole, the ampere and the second.In actual practice (with the exception of the kilogram) it is somewhat difficult to relate accurately these base units to actual measurements of chemical compo- sition in the field because many analytical measurements are influenced by a variety of matrix effects and the precision of most analytical measurements is a far larger source of uncertainty than the uncertainties associated with the measurement of the SI base units.103 Exceptions to the latter include measurements involving ‘primary’ chemical standards where uncertainties of one or two parts in 10000 are often encountered and reference to an absolute scale is required to ensure compatibility of these measurements. This compatibil- ity is illustrated in the work of Sappenfield et aZ.104 where several ‘primary’ redox CRMs were inter-compared using the coulomb (ampere second) as a basis of measurement.Another difficulty for chemical measurement traceability is that the SI unit by which chemical composition is ultimately measured, viz., the mole, is only realizable as a pure chemical species. However, there are no means of determining purity directly and unambiguously.105 It is only possible to determine impurities by as many methods as possible. Using the SI base units as absolute or ‘true’ values, several workers103J06-109 have described a measurement tree or pyramid for transferring accuracy throughout a measurement system via methods, transfer standards and CRMs. The function of each component in the hierarchy is to transfer accuracy to the level immediately below it and to provide traceability to the level immediately above it, thus assuring measurement compatibility in the over-all system.Proceeding from top to bottom, accuracy requirements diminish at the expense of increased measurement efficiency. In such a system, a typical accuracy requirement for a field method of 5-10% would require the accuracy of the reference method to be in the range of 1-3%. The definitive method and primary RM should then have accuracies in the 0.1-1.0% range. Depending on the measurement being made, not all of the steps in the measurement hierarchy are necessary. If a ‘primary’ standard is readily available and inexpensive, it may be used and related directly to laboratory samples.103 In other instances, three or four steps in the measurement hierarchy might be required to relate the ‘primary’ standards to laboratory samples.Criteria for Developing Reference Materials Before developing an RM one of the first considerations should be to assess the technical requirements which dictate its end use.99 This will largely govern the certification require- ments such as the accuracy, stability and physical form of the material. It is also necessary to develop an adequate means of measuring the property and preferably one which is traceable to the appropriate SI units. Ideally, the RM should have the same or nearly the same matrix as the samples to be analysed, which it should also match with respect to the levels of trace elements present.110 For example, for natural matrix samples these should be the natural levels not elevated levels due to contamination from handling and preparation.If possible the speciation of the CRM, that is its valency and binding, should also be the same as in the real matrix. Clearly, this requirement implies the need for a large variety of CRMs in the environmental, biological and biomedical fields. The material must be homogeneous, i.e., the difference between representative sample measurements must be smaller than the over-all uncertainty limits of the measurements.111 Before a certification process or even an interlaboratory comparison is started the homogeneity should be studied at982 ANALYST, OCTOBER 1991, VOL. 116 various levels of sample intake.86 In a homogeneity study it is important that the analytical techniques to be used are capable of achieving high precision (reproducibility).The accuracy of the techniques is less important; therefore, it is not necessary to be concerned with an evaluation of systematic errors at this stage. Suitable techniques to study matrix homogeneity for trace elements include X-ray fluorescence, neutron-activation analysis or inductively coupled plasma atomic emission spectrometry. It is often not necessary to investigate all the trace components to be analysed, but trace elements preferen- tially associated with different phases of the matrix should be selected for this study, e.g., the inorganic and organic fractions of coal, soil and plant material and the large and small particle fractions in fly ash and in urban particulates.111 When several elements are envisaged for certification in the same candidate material, homogeneity for one element does not imply that the material is also satisfactory for other elements.When the inhomogeneity is not negligible in comparison with the precision of the measurement methods it may not be possible to certify the material in the usual manner. However, in these instances, the degree of in- homogeneity may still be low enough to make the material of practical interest. It should be borne in mind that segregation is always possible if the material is not a pure compound. Matrix materials and biological materials may contain phases of different particle size and density, and segregation may occur during transit from the producer, general handling of the bottle and even on standing on a laboratory bench for a prolonged period of time.Re-homogenization may therefore be required if only a sub-sample of the material is to be analysed. Certain powdered materials supplied by the National Research Council of Canada, for example, are supplied with a PTFE ball which is used to agitate and mix the material before use. Frauerwieser”2 has discussed the possible errors due to inhomogeneity when an environmental material such as urban particulate matter is used as a CRM, while Guzzi et al. 113 have proposed a method to calculate the degree of inhomogeneity from practical measurements on solid samples. The ideal RM should be physically and chemically stable for an indefinite period of time.However, most materials change in time as a result of evaporation or chemical reactions under the influence of temperature, light, air or humidity, due to precipitation, bacteriological activity or to interaction with the container in which they are stored.86 The values to be certified may therefore change, or at least the speciation or part of the matrix may be altered. This is particularly a matter of concern for organic or volatile (e.g., mercury and organometallics) components and solutions. Therefore, the stability should be tested under conditions which accelerate the normal condi- tions which might occur in the laboratory. The CRMs whose stability is not infinite under normal laboratory conditions should be provided with an expiration date.During storage certain materials may absorb water preferentially in the upper layers of the bottle. The certificate should therefore give instructions for handling and storage. Griepink25 has described the rationale for preparing CRMs in the form of a flow chart while Dragoo1’4 lists the criteria necessary for developing CRMs for X-ray diffraction analysis. Several workers86-115,116 have described the mistakes that have been made by certain producers of CRMs during the preparation of such materials. Methods of Certification Once the representative nature of the sample is established, the stability and homogeneity controlled, the most difficult task is still to come, namely the certification itself, i.e., to obtain property value data which approach as closely as possible the ‘true’ value, together with uncertainty limits.Essentially there are three analytical procedures used to do this.111 Namely: (a) certification by a single laboratory; (6) certification based on a statistical consensus of several laboratories; and (c) certification based on several methods and several laboratories. Procedure (a) is generally only carried out when there is no alternative, for example when one laboratory has developed an expertise far superior to what can be found elsewhere, using definitive methods where all major significant parameters can be related directly or by a solid chain of evidence to the base or derived SI units.108 Examples of definitive methods include isotope dilution mass spec- trometry, coulometry and gravimetry. It is possible that certification by one laboratory may include the risk that some bias or error may pass undetected.Even the so-called definitive methods do not give an absolute guarantee against possible errors. An alternative approach under procedure (a) is for one laboratory to use a range (typically three) of reliable independent (usually reference) methods of analysis each of which has been shown in many experimental situations to yield good results for which both the accuracy and precision are clearly known. Here, independent method means that the basic principles used for the analysis are entirely different, so that possible interferences or systematic errors can reasonably be expected to be different. Such methods should be applied independently by different-operators. In its simplest form, procedure (6) 66kists in obtaining results from several (but possibly many) laboratories, submit- ting the results to a statistical analysis and rejecting the statistical outliers.The mean value is then certified as the estimate of the ‘true’ value; Marchandise and Colinet1*1 note that this approach should be treated with caution because it is often assumed that if the number of participants and data is large enough the mean value will not differ much from the ‘true’ value. This may not be correct. When a large pool of results is classified according to the methods used, the means for each method may differ sometimes by up to 10% or more. If one of the methods is more accurate, mixing with less accurate results can be misleading.Marchandise and Col- inetlll also consider that the statistical treatment usually used in procedure (b) in order to establish the mean value and the uncertainties may be unreliable. For example, some standards describe the statistical procedure to be applied to the population of all single values N of r laboratories having supplied them. This is clearly not correct because the values supplied by one laboratory are not independent estimates of the ‘true’ value. They are bound by the fact that they belong to the same parent laboratory. The calculations to obtain the mean value and the standard deviation should therefore be carried out on the mean values of each laboratory. The third approach to certification, procedure (c), is based on the use of different independent methods carried out by a number of laboratories.As noted earlier, individual labora- tories may well be responsible for producing bias in a method and it is important that at least two or three well trained and experienced laboratories participate for each method. They should work independently and should not use a single detailed protocol. The examination of the results may lead to the elimination of one or several methods if their inaccuracy can be demonstrated. At the limit, only one method applied by only a few laboratories may be valid for certification. This is in contrast to the simple consensus approach [procedure (b)] where an accurate, but minority method could be eliminated because its results appear as outliers. Where possible, definitive methods should be included in procedure (c) type certification campaigns.Marchandise26-27 has summarized the methods of certifica- tion and concludes that a property can only be reliably certified when its value is confirmed by several laboratories working independently and using more than one method of equivalent accuracy, that is by procedure (c). Marchandise points out that this principle is equivalent to that which isANALYST, OCTOBER 1991, VOL. 116 normally carried out in the field of metrology where even the best national physical laboratories obtain confirmation of their results by collaboration with other similar establish- ments. However, Michaelis117 notes that for complex CRMs with many properties to be certified, a combination of procedures (a), (b) and (c) should be used.Rasberryllg comments that at NIST it is not sufficient to rely on assessments of measurement precision to set uncertainty limits. Instead, an individually tailored programme is set up for the project design, measure- ment and certification of each CRM.119 Marscha112° considers that although most CRMs are cur- rently certified by large interlaboratory campaigns, it is possible that this approach will decline due to the highly specialized nature of CRMs to be produced, the number to be produced, a decline in the number of laboratories able to participate for technical or economic reasons and the difficulty in perfectly demonstrating the traceability of such analyses. Marschal suggests that the best approach in the future would be collaborations involving between two and five specialist laboratories.Before selection of the final results which will be used in the calculations leading to the certified value(s) some generally accepted statistical treatments must be carried out. The Kolmogorov-Smirnov test as reviewed by Lillieforsl21 (good- ness of fit of a distribution), the Dixon test and Chauvenet’s criterion122 (outliers), the Cochran and the Bartlett tests123 (homogeneity of the variances of the laboratory means) are the most popular. Several workers26,124-126 give good accounts of the certification of RMs including the statistical assessment of the data from the different approaches to certification. International Organization for Standardization Guide 3596 assumes, in the ideal situation, that a statistically meaningful evaluation of the data should be possible, leading to an expression of the uncertainty as either (a) a statistical tolerance interval (usually constructed so that it will cover 95% of the population with a probability of 95%) or.(b) a confidence interval for the mean (usually expressed at the 95% confidence level).However, data sets from many interlaboratory studies do not always meet the criteria for the valid application of a purely statistical approach (i.e., normal distribution with a minority of outliers, random rather than systematic error as the major source of scatter).llo,127 While statistical tests can be applied in a fully objective manner, the criteria required (such as maximum percentage of outliers, the minimum number of acceptable laboratory averages, the degree of concurrence of the various analytical methods, the relative uncertainty of the over-all mean, etc.) for deciding the status of the recommen- ded value as being of a satisfactory or acceptable degree of confidence or merely an information value, are subjective.Thus, while statistical methods have a most important role to play, they need to be supported by analytical considerations such as the chemical nature of the sample, likely interferences, selectivity or applicability of methods and the degree of risk of contamination. Indeed, several workers12*,129 are critical of the quality of some interlaboratory studies and the fact that the results of many studies are either never published or appear in reports which are not readily available.Equally, it has also been observed130 that few workers publish data that include the testing of their analytical procedures with CRMs. Christie and Alfsen131 and Christie132 noted that skewed frequency distributions are often presented with collaborative study data, and for such distributions the arithmetic mean is often a poor estimate of the most frequently determined value. Christie and Alfsen solved the problem using poly- nomial fitting to a histogram and also using a gamma transformation procedure based on the data transformation philosophy of Box and Cox,*33 whereby the data are trans- formed to make up a symmetric distribution. The latter method allows the construction of a confidence interval which is non-symmetrical with respect to the central value of skewed distributions.Pszonicki30 showed the simple median and its confidence interval, calculated on the basis of a non-para- metric distribution after the rejection of outliers, to be the optimum indicator of the central values. De Goeij et al. 127 and Abbey et al. 134 considered several methods for deriving ‘best values’ from disparate data reported in collaborative studies of rock samples, including the dominant cluster mode scheme135 and the gamma transformation mode.131 Contents of Certificates Rasberry119 notes that at the end of a certification exercise there is far too much data available to put onto a certificate. It is necessary to distil the data into one meaningful uncertainty statement for each value certified. An account must be taken of the precision of all the measurements and the homogeneity of the material but, more importantly, it it necessary to zero in on the ‘true’ value. In instances where certification is for an element near the detection limit of state-of-the-art methods, it is usual for relative uncertainties to be large. For example, an element that can be certified at 1 f 0.8 ng g-1 may seem to have an unacceptably high relative uncertainty of &go%.However, the absolute error is fairly small and defines the presence of the element in the CRM to within kO.8 ng g-1. International Organization for Standardization Guide 3195 describes clearly what information should be given on a certificate. It notes that essentially a certificate should communicate information about a CRM including a statement of the certified property values, their meaning and their uncertainty (confidence limits).The remainder of the infor- mation on the certificate is peripheral to this central statement and has two purposes: to describe the general nature and use of the material and to assure the user of its integrity. Essentially the certificate should contain the following infor- mation: the general particulars of the certifying organization and the reference material (name, sample, number, date of certificate, etc.), a description of the material and its intended use, the certified values, their confidence limits and an explanation of these values and the techniques used for their measurement, the references (e.g. accompanying report giving full details of the procedures used for the preparation and certification of the material, analytical data, etc.) and the names of the participating analysts and certifying officers. In addition, the homogeneity limit (minimum size of the rep- resentative sub-sample) and expiration date are considered important .86 Parr et al.136 report on the large differences noticed between RM producers in how the concentration values and their confidence intervals are defined. They note that the inter- national recommendations on terminology generally do not appear to be followed. For example, words such as ‘certified’, ‘recommended’, ‘indicated’, ‘non-certified’ and ‘information’ values are used to describe the type of concentration value reported for each element. However, according to IS094 only the terms ‘certified value’, ‘uncertified value’, ‘consensus value’ and ‘best estimate’ of a given quantity are officially recognized.Equally, there is much confusion as to the meaning of the uncertainties that are attached to the concen- tration values of trace elements in biological materials.136 It is common to find statements that cannot be interpreted in a meaningful quantitative or statistical way. Such statements reflect the practical problem that analytical data on which the certification procedure is based are frequently made up of several different sub-sets of data, each with its own, imper- fectly known, statistical distribution. Such sets of data may defy conventional statistical evaluation. Use of Reference Materials The only rational basis for using an RM is as a monitor of a measurement system that is in a state of statistical control.56984 ANALYST, OCTOBER 1991, VOL.116 That is, a valid measurement procedure has been implemen- ted using QC procedures that assure a requisite degree of reproducibility. Usually, the RM will only be available in limited amounts so that the statistics of the measurement process should be demonstrated by measurements on other materials. Only in this way can the results of an RM measurement be considered as representative of the measure- ment system. To the extent that its compositional properties simulate those of the sample ordinarily measured, its ‘correct’ measurement can imply ‘correct’ measurement of the routine samples.56 Such a conclusion requires that the protocol of measurement is the same in each instance.Hence, it is necessary that no special care be exercised in measuring the CRM, other than that ordinarily used. Reference materials are best used on a regular basis. The sporadic use of RMs when trouble is suspected is a legitimate use, but systematic measurement will generally be more informative. Certified Reference Materials may be used as the sole RM or they may be used with internal RMs in a systematic manner, thus conserving the former and adding credence to the latter. The CRMs are valuable commodities requiring much time and effort to produce and it is therefore recommen- ded that they are only used when necessary, for example in the validation of a method. Internal RMs, produced in-house by the application of CRMs, should be employed on a regular basis.Reference materials find a wide variety of uses in analytical chemistry, but one of the major uses, and indeed the original driving force behind CRM development, is for the quality assurance (QA) of measurement processes. 124 By including an RM in each batch of samples and plotting the results on a control chart, one has a graphical means of interpreting the output of a method over time; the repeatability of the results and the precision of the method can be seen at a glance. Further, when various analysts use different methodologies (and even the same method), unacceptable discrepancies can arise, usually attributable to calibration or procedural differ- ences. The analysis of commonly available RMs can identify such problems and lead to their solution.While replicate analyses of an RM indicate that the precision obtained by the system is good enough, the deviation of the mean analytical value from the certified value is a measure of the systematic bias of the system. This strategy can improve the quality of analytical measurements. The availab- ility of an RM together with the certification report enables the analyst to match directly the performance of his laboratory with the certifying laboratories unlike when he compares his performance with other laboratories in a collaborative study when such information is only known retrospectively. By this strategy the analyst takes an important step towards improv- ing the QA of his laboratory. He can verify the performance of his instruments and analytical methods each time there is a risk of bias, for example due to a possible interference. As such, CRMs have been described86 as the keystones of good QA practice.Certified Reference Materials are often used in the develop- ment and calibration of reference methods and assuring their accuracy. Such methods may be suitable for direct use. Alternatively, they may serve as a basis for developing or evaluating other methods. Reference methods are also commonly used for producing secondary RMs which, in turn, are directly used in routine field measurement applications. For some types of instrumental methods the calibration and traceability depends on pure chemical standards certified for specific physical properties, e.g., benzoic acid certified for its melting-point.However, the only means of calibrating a total analysis in most instances is to use an RM of similar matrix in order to assess the reliability of the sample preparation stages such as acid digestion, ashing and separations. Other uses of RMs include establishment of measurement traceability124 ( e - g . , development of secondary CRMs), evaluation and comparison of field methods, and to serve as method evaluators. 119 However, it should be noted that the measurement of a single CRM may not be fully informative. The analytical errors involved may be constant, measurement level related, or a combination of these, and a single right or wrong result will not indicate on which of several possible curves it might lie.56 An intimate understanding of the operation of a particular measurement system may make it possible to eliminate some of the possible sources of error and to interpret better the data from the measurement of CRMs, but ideally the analysis of several CRMs, spanning the concentration range of interest, is the most useful way to investigate measurement bias.137 The three sample approach, analysis of a low, middle and upper range sample, is practical in most instances, provided that the reference samples are sufficiently homogeneous and that the range of analytical interest is covered. Bias is even identifiable using relatively inho- mogeneous samples, provided that a sufficient number (n) are analysed, as it is highly unlikely that a randomly selected sample from a lot would deviate in a systematic manner from a population mean value, provided n is 5 or more.One difficulty to the three sample approach (which is more apparent than real) is that it may be difficult to identify suitable RMs containing the analyte at the desired concentra- tion level and with the desired matrix. However, many analysts attach too much importance to the concept of selecting a reference material with exactly the same matrix as the routine samples to be analysed. Parr et al. 137 consider it is much more convincing if the analyst can validate the analytical method for use with a variety of different matrices rather than with just one. It should be noted that any storage conditions which would alter the certified value of a material must be avoided.106 With inorganic materials, e.g., steel chips, little warning is necess- ary.Furthermore, the usual requirements for a standard are that it be stable, non-hygroscopic and non-efflorescent in most atmospheres. However, there are CRMs which pose storage problems. For example, coal contains volatiles, cement tends to hydrate and biological materials are particularly prone to degradation even in their lyophilized forms. Careful labora- tory storage in desiccators, sealed vials, amber-coloured bottles and similar containers is usually called for at the appropriate temperature. Once the seal is broken by the user and the container opened, the original contents may easily be contaminated. Thus, in a laboratory situation, some policy concerning the use of standards must exist to prevent contamination.The integrity of the measurement process of a laboratory rests on this consideration as much as with the analyst’s technique and the reliability of the method. Quality Control of Sampling It is axiomatic that meaningful analytical results can only be achieved if the analysed sample adequately represents the population or material from which it was drawn. All too frequently effort is concentrated on improving the precision of analytical procedures, when any uncertainty in the analysis may be far outweighed by the uncertainty attached to the taking of the sample. Many analysts have little or no control over the samples submitted to them but they should always satisfy themselves that the samples are worth analysing. Effective sampling must be based on a statistically valid sampling plan.For many purposes (e.g., QC of manufactured products, commercial valuation or acceptance, customs or other regulatory control) the population to be sampled may consist of a large number of discrete items or packages. Depending on whether the analysis or inspection is by attributes, where each item either passes or fails a prescribed limit or test, or by variables, where some property is measured on a continuous scale, there are established standardsANALYST, OCTOBER 1991, VOL. 116 985 prescribing the number of items to be taken to form a representative sample.1387139 The statistical factors affecting the variance of the analytical result have been outlined by Taylor53 and Kratochvil and Taylor. 140 The size and number of samples needed to be representative of a bulk material will depend on the extent to which both the sample and its parent population are homogeneous, and this may need to be established by repetitive analyses.Recom- mended methods for sampling gases, liquids and solids are given in BS 5309.141 Additional problems arise if the bulk material is segregated or stratified. 140 In other circumstances, for example where contamination is suspected, representative sampling of the bulk material would not be appropriate. Non-representative sampling may also be necessary where parts of a bulk or consignment are inaccessible. In these circumstances there is usually no way in which the sample result can be statistically related to the parent population. The relationship between environmental samples and the decisions which may be taken on the basis of their analytical results is particularly critical .59,*42 Where small differences may be interpreted as trends it is vital that the natural variability, both of the sample material and of the analytical process, should be clearly understood.The dangers of sample contamination owing to the use of unsuitable containers or inadequate storage or transport conditions must also be stressed. 143 The literature on sampling for chemical analysis was surveyed in 1981 by Kratochvil and Taylor144 and more recent references are quoted by Garfield.143 Internal Quality Control The terms ‘quality control’ and ‘quality assurance’ are sometimes used interchangeably, and where they are separ- ately defined the definitions often differ in detail.In some instances51952 the distinction may not be easily comprehen- sible. If the term ‘quality assessment’ is also introduced, then quality assurance may be taken59 to embrace both quality control and quality assessment. These are defined concisely by Taylor:56 ‘Quality control is the mechanism established to control errors, while quality assessment is the mechanism to verify that the system is operating within acceptable limits’. The Analytical Methods Committee61 qualifies the latter by defining quality assessment as ‘the objective testing of the performance of a laboratory by an external agency’, while limiting quality control to ‘practical activities’ undertaken to control errors. For the purpose of this review quality control is taken to include also the wider internal aspects of quality assurance which are essentially laboratory management prac- tices.The external accreditation and compliance schemes48-51 place great emphasis on the role of laboratory management in ensuring the quality of analytical results. Consistent quality does not happen by accident, it is the result of detailed planning and positive commitment of resources. The first stage in implementing a quality system is to designate a person to be responsible for QA, whose first task will be the production of a quality plan. At all stages thorough documen- tation is necessary, particularly for the Standard Operating procedures required by Good Laboratory Practice (GLP), so an essential feature is a laboratory quality manual; examples of the contents of such a manual are given by Dux54 and Taylor.53 Records will be required of staff training and competence and of the maintenance and calibration of all equipment used. Garfield52 provides a schedule of regular instrument per- formance checks, while a major concern in demonstrating comparability of data is the need to establish traceability of calibration of measuring equipment to national or inter- national standards. This is relatively easy for instruments measuring basic physical properties such as mass and temper- ature, but may be more difficult for chemical measurements. Certified reference materials or pure chemical standards may be used to calibrate the response of an instrument, such as a chromatograph or spectrophotometer, but the total analytical method may also require calibration with a reference material of comparable matrix. In the absence of traceability to a standard material or property, traceability to a definitive analytical method may be possible.Documentation of sample handling, from receipt in the laboratory to issue of the analytical report and disposal of the sample remnant, is also vital and may be facilitated by the introduction of laboratory information management systems (LIMS). Where samples are required for forensic analysis or other legal purposes demanding continuity of evidence the chain of custody must be clearly documented, together with records of security of custody. In Britain, section 69 of the Police and Criminal Evidence Act 1984 states that evidence generated from computer records will not be admissible unless it can be shown ‘that there are no reasonable grounds for believing the statement is inaccurate because of improper use of the computer’, and that at all material times the computer was operating properly.To meet these requirements most LIMS software now includes audit trail facilities. A system of independent quality audits, by staff not concerned with day to day management of the work of the laboratory, is in any event an essential part of any quality system. Some organizations maintain separate lines of responsibility for QC and QA to ensure the independence of the latter. The audit procedures of the US Food and Drug Administration have been outlined by Garfield.52 Guidelines have also been produced by IS0.46 Quality Control of Analysis The first step in any analysis is the selection of the analytical method.The suitability of a given method clearly depends on the acceptable limits of error in the results. It is therefore necessary to specify these accuracy requirements61 and match them against the performance characteristics of candidate methods. This process is illustrated graphically by Taylor.56 In the area of trace analysis a critical characteristic may be the detection limit for the analyte concerned. A definition of detection limit was proposed by IUPAC’45 in terms of k times the standard deviation of the blank measurements, with a strong recommendation that .the value of k should be 3. This value was adopted by the Analytical Methods Committee79 for the detection limit of a total analytical system, the blank in this instance being a ‘field blank’, i.e., the sample matrix contain- ing zero analyte.It was recognized, however, that analysts evaluating instrumental techniques frequently defined the detection limit as twice the standard deviation of the blank, the blank in this instance being pure solvent rather than a field blank. The Committee on Environmental Improvement of the American Chemical Society59 also recommended the use of three standard deviations for the limit of detection, together with a limit of quantitation of ten standard deviations, the region between these being a ‘region of less certain quantita- tion’. Curriel46 had also recommended ten times the standard deviation as the determination limit.Whatever analytical method is selected, its performance characteristics should be known in order that the degree of uncertainty of bias attached to the results can be quoted. The precision of the method should have been established by collaborative trials, but the Analytical Methods Committee has recently warned that the repeatability (within-laboratory variance) achieved in an individual laboratory may not be the same as that obtained in the interlaboratory trial.61 They also suggest that reproducibility (between-laboratory variance) is not really a property of the method, as it represents bias between laboratories. Day to day repeatability in a laboratory can be monitored by the use of control charts. These require the availability of a986 ANALYST, OCTOBER 1991, VOL.116 significant amount of a stable and homogeneous QC material, a portion of which is analysed with each batch of samples. The result is plotted on a Shewhart chart147 for comparison with the reference value. The latter is usually the mean of a series of determinations, with control limits set at three standard deviations above and below the reference value and warning limits at two standard deviations above and below the reference value. A result outside the control limits, or two successive results outside the warning limit in the same direction, should be taken as an indication that the analysis is out of statistical control, requiring corrective action to be taken.33.61 Where replicate measurements are regularly made the mean is plotted on this chart, while the range of results can be plotted on a second chart.61Ja An alternative type of chart, the cumulative sum or Cusum chart,'@ is more effective for detecting small but consistent errors (for an example see reference 61).Control charts provide an effective measure of in-house repeatability and an assurance that the analysis is under control, and they can also be used in appropriate circum- stances as a check on inter-operator bias. They do not, however, provide any evidence of the over-all trueness, or freedom from bias, of the laboratory results unless the QC material is a CRM. The CRMs are generally too valuable to use for this, purpose, but one can be used occasionally (if available) to supplement the regular QC material.An approach used by groups of laboratories undertaking blood or serum analyses is to distribute common internal QC samples to all laboratories in the gr0up.150~151 More commonly, the internal QC materials of the laboratories may be supplemen- ted by external QC samples of unknown composition dis- tributed by a central laboratory.33,42,55,152 For some types of analysis the materials examined may be insufficiently stable for long-term use of a single QC material. An alternative approach, allowing batch to batch variation to be monitored, is to save one or more samples from each batch -and analyse them again in a subsequent batch.148 A QC chart could also be plotted of instrument response to blanks and standard solutions. In the authors' own laboratory a bulk flour is used as a QC material for the microbiological assay of B-group vitamins; the vitamin content has decreased gradu- ally but steadily over a period of years, but at any given time the expected value is predictable.For a large general purpose laboratory performing a wide variety of analyses the amount of QC data collected may be enormous, and this will serve no purpose unless it is regularly assessed. Dux58 described a minicomputer-based system which provided regular printouts of cumulative quality control data for each type of analysis, highlighting any results exceeding the 95% confidence limit. This facility is now well within the capability of personal computers and should be available as part of a LIMS package. External Quality Assessment The increasing requirement to demonstrate comparability of analytical data, both for environmental monitoring and to ensure mutual acceptance within the single European market, demands some form of external assessment of the quality of results produced by individual laboratories.There are two main ways of doing this, which are essentially complementary. The first is by physical inspection of the laboratory to ensure that it complies with externally imposed standards; the second is by assessment of performance in interlaboratory compari- sons using centrally distributed samples. Assessment by Inspection In the UK there are currently three schemes which undertake inspection of chemical laboratories, though none of them was specifically designed for this purpose. All three have counter- parts in other countries and are subject to international standards of guidelines.For UK testing laboratories offering a service the appro- priate form of external assessment is usually accreditation by NAMAS. This service was formed in 1985 by the amalgama- tion of the British Calibration Service (BCS) and the National Testing Laboratory Accreditation Scheme (NATLAS). Prior to the introduction of NATLAS in 1981, national schemes for accreditation of testing laboratories had been introduced in Australia, National Association of Testing Authorities (NATA)153 and New Zealand Testing Laboratory Registra- tion Council (TELARC).154 A similar scheme was also in operation in the US National Voluntary Laboratory Accredi- tation Program (NVLAP). 155 Harmonization of the accreditation standards of these bodies has been pursued through the International Labora- tory Accreditation Conference (ILAC) and through ISO.46 In Europe the accreditation criteria have been formalized in European Standard EN 45OO1.47 The NAMAS Accreditation Standard M1048 is compatible with both IS0 Guide 25 and EN 45001, but specifies its requirements in more detail.Conform- ity with the international standards has enabled NAMAS to negotiate mutual recognition agreements with schemes in France, Australia, New Zealand and Hong Kong. The NAMAS has also concluded agreements with bodies conduct- ing assessments to BS 5750 and GLP, so that joint assessments may be carried out by laboratories wishing to be covered by more than one scheme. Accreditation of testing laboratories had its origins in areas of physical and mechanical testing, where the majority of tests are of an objective nature and usually intended to demonstrate conformity with specifications.Assessment procedures there- fore attach great importance to traceability of the calibration of measuring equipment to physical standards. Analytical chemistry is more subjective and may depend on chemical standards of more doubtful traceability. Mesleyl56 has de- scribed the NAMAS accreditation of an analytical laboratory and commented on the shortage of objective assessment criteria. Guidance on the interpretation of the accreditation standard for chemical laboratories has recently been produced by NAMAS.157 The NAMAS Accreditation Standard includes sections on organization and management, quality manuals, quality aud- its, staff, equipment, calibration, methods, records, sample handling and test reports.It says virtually nothing about QA of results (though this is covered by NIS 45157), and assessors must rely largely on documentary evidence of quality. This is perhaps the main shortcoming of accreditation schemes, and to overcome this they may ask candidate laboratories to analyse spot check samples provided by assessors or to provide evidence of satisfactory performance through participation in recognized proficiency testing schemes. For UK laboratories other than those covered by NAMAS, accreditation to IS0 9000 or GLP should be considered. The IS0 9000 series of International Standards relates to quality systems in the context of suppliers of goods and services, and these correspond to the EN 29000 series of European Standards and the various parts of British Standard BS 5750.51 Of particular relevance to laboratories is BS 5750 Part 3 (IS0 9003) which lays down requirements for QA in final inspection and test. British Standard 5750 can be used as a basis for assessment, by purchasers or by independent third parties, of a supplier's quality management system and capability to assure goods and services of the required quality.Some major purchasers, such as British Rail and the Ministry of Defence, carry out their own assessments of BS 5750, but in order to minimize the burden of multiple assessments several indepen- dent approval bodies, including BSI Quality Assurance Services and Lloyds Register Quality Assurance, will carry out assessments and grant BS 5750 approval.To ensure uniform-ANALYST, OCTOBER 1991, VOL. 116 987 ity of assessment standards the UK Government has set up the National Accreditation Council for Certification Bodies to accredit bodies offering BS 5750 approval. Companies seeking BS 5750 approval may have their own testing laboratories included, but approval is restricted to tests carried out on goods manufactured or supplied by that company. The relevance of BS 5750 to the generality of analytical labora- tories is therefore limited. Good Laboratory Practice is concerned with the organiza- tional processes and conditions under which laboratory studies are planned, performed, recorded and reported for the non-clinical testing of chemicals for the protection of man, animals and the environment.The principles of GLP were introduced in the US to assure the quality and integrity of the data generated and to allow their use by regulatory authorities in hazard and risk assessments of chemicals. They were subsequently recommended for international adoption by the Organization for Economic Co-operation and Development (OECD)49 and have received legal authority in the European Community. 158.159 Within the UK compliance is monitored by inspectors of the Department of Health.5" Good Laboratory Practice is aimed mainly at studies, often using laboratory animals, intended to establish the toxicity of chemicals, and as such is of concern particularly to the producers of pure chemicals and of pharmaceuticals, cosmet- ics, pesticides and other chemical products offering potential risks.It specifically includes facilities designed to ensure that the identity, amount and composition of test and reference substances are in accordance with the study plan, so analytical laboratories serving these industries will normally be required to comply. Inspections involve regular monitoring of labora- tories, including a critical review of any studies being carried out, and may also involve in-depth study audits carried out at the request of regulatory authorities in this country or abroad. Study audits are likely to concentrate mainly on animal studies, but regular monitoring involves checks on manage- ment systems, including detailed records of staff, and, in particular, documented standard operating procedures for every aspect of the work.Interlaboratory Comparisons Interlaboratory comparisons are an essential feature of method development and validation (when they are usually referred to as collaborative studies) and also play a major part in those RM programmes which depend upon collaborative certification. In the form of proficiency testing, or external quality assessment schemes, they provide independent evidence that analytical quality assurance is effective. Many surveys aimed at environmental monitoring or epidemioiog- ical studies also make use of data provided by a number of different laboratories. In practice, whatever the prime objec- tive of the exercise, many such comparisons fulfil more than one purpose.In certification exercises carried out by BCR the first intercomparison frequently shows a lack of agreement between the results from the collaborating Iaboratories.2"29 This often necessitates a study of the available methodology in order to identify sources of error, in the course of which the performance of individual laboratories is necessarily scruti- nized, and those that show poor repeatability or bias with respect to other laboratories may be eliminated from the study. Only when acceptable agreement is obtained can the results be used to certify the RM. The statistical basis for such certification was reviewed by Pszonicki,30 who concluded that the distribution of data may differ significantly from the normal distribution, and that the simple median was therefore a better indicator of the central value than the arithmetic mean.Parshin et al.43 found that regular participation in certification studies of RMs contributes to the maintenance of higher quality of analytical work, an observation which provides support for the BCR practice of relying upon a relatively small number of expert laboratories to carry out collaborative certification. The monitoring of river water quality requires the compari- son of data from numerous laboratories, and in Britain a co-ordinated system of analytical QC was set up in which all the regional water authorities took part. The principles of this system having been agreed,33 a series of studies were then made of the accuracy of determination of chloride,34 nitrogen in various forms ,353 suspended solids,37 electrical conductiv- ity and pH,38 biochemical oxygen demand (BOD)39 and various heavy metals.2.40.41 Accuracy requirements had been set in advance and in some instances these were not achieved; however, within-laboratory precision was generally adequate.The apparent bias between laboratories was attributed to lack of validated analytical methods for the levels concerned and to possibly biased calibration. Similar conclusions were reached by Smith,'" following a series of interlaboratory comparison studies by water analysis laboratories in southern Africa. The UK study on pH identified problems resulting from the instability of the samples, and a further study was undertaken in which the participants met in one place and performed their analyses simultaneously.161 Good agreement was obtained in the laboratory using a variety of pH meters, but corresponding field tests showed significant variation, apparently attribu- table to the commercial equipment used.A co-ordinated analytical QC system, comparable to that for river waters, was subsequently set up for filterable cadmium and mercury in saline water.42 Whereas collaborative studies of meth0ds56~162-164 demand strict adherence to prescribed methods of analysis, studies carried out for other purposes may allow participants a free choice of method. Indeed, in the certification of RMs use of a variety of independent methods is positively encouraged in order to ensure that results are not subject to instrumental or other methodological bias.In environmental monitoring opinions differ on the advisability of method standardization. Dembicki3 found wide variations in results from 19 labora- tories for kerogen in rocks, attributed primarily to differences in procedures and recommended development and adoption of standard methods. Eastwood and Jackson165 found signifi- cant interlaboratory differences between soil lead results obtained by acid digestion and AAS and those determined by X-ray fluorescence spectrometry; they too commented on the need for standardization of methods in addition to regular interlaboratory comparisons. Studies on the determination of polychlorinated biphenyls in herring5 and in sediments4 both concluded that variations in procedures were responsible for a wide spread of results.Subramanian and Stoeppler7 found that experienced laboratories using electrothermal AAS to determine 55-200 ng ml-1 of lead in blood showed a 10% bias relative to results obtained by isotope dilution mass spec- trometry, but considered that this degree of accuracy was acceptable. Delves151 had previously observed that distribu- tion of a common internal QC sample to all laboratories taking part in surveys of lead concentrations in blood resulted in improved interlaboratory agreement. He concluded that the use of the laboratories' own methods, coupled with the distribution of reference samples, was a practical alternative to the use of the same reference method by all laboratories. On the other hand, Wagemann and Armstrong,'@ studying trace metal determination in animal tissue, recorded good agreement between a wide range of different instrumental techniques.Gardner and Dobbs,167 while deploring poor comparability of analytical results, asserted that uniformity of methodology is only a minor factor in achieving comparability between laboratories and that it was a misconception that analytical accuracy was governed by methodology. This view was challenged by Mesley and Sargent,168 who agreed that988 ANALYST, OCTOBER 1991, VOL. 116 valid analytical measurement depends upon many factors, but felt that it was sensible to eliminate any disagreement which might result from the choice of method. The use of interlaboratory comparisons as a means of improving the quality of analytical requirements is exempli- fied by proficiency testing or external quality assessment schemes, in which individual portions of a uniform bulk material are distributed for analysis by all the participating laboratories.Results are returned to the organizers, where they are subjected to some form of statistical analysis, resulting in the issue of a report to participants. Depending upon the purpose of the scheme, this report may include an assessment of the performance of individual laboratories, either in terms of the achievement of some pre-determined level of performance or by calculation of an index which allows ranking of laboratories. Some exercises, such as that which revealed difficulties in determining lead and cadmium in foodstuffs,13 are carried out on an ad hoc basis, primarily in order to establish whether there is a significant problem.Others are organized with regular distributions of material at intervals ranging between 2 weeks and 3 months, sometimes with very large numbers of participants. Some of the latter type may be used to assess professional competence and failure to reach a specified level of performance may result in loss of a licence to carry out analyses for official or regulatory purposes. Within the UK there are established schemes covering water analysis (Aquacheck'@); occupational hygiene, includ- ing asbestos fibre counting [Regular Interlaboratory Counting Exchange (RICE)]170 and toxic substances in workplace atmospheres (WASP14*), replacing the earlier AQUA scheme;171 the United Kingdom External Quality Assessment Scheme for Clinical Laboratories, 172 which includes schemes for general clinical chemistry, trace elements in biological fluids173 and drug assays, in addition to microbiology and several biochemical areas; and food analysis [Food Analysis Performance and Assessment Scheme (FAPAS)].174 There are also several privately organized schemes, covering partic- ular types of commercial products, and restricted proficiency schemes within individual companies and government depart- ments. Some of the schemes mentioned above are not limited to UK participants, and several schemes are organized on an international basis by United Nations agencies such as The Food and Agricultural Organization (FAO), The World Health Organization (WHO) and The International Associa- tion for Research into Cancer (IARC).The situation in the UK is no doubt mirrored in many other countries. The proliferation of proficiency testing schemes could act as a barrier to mutual recognition of analytical data unless there is some degree of uniformity in their procedures. Some general guidelines are provided in the ISO/IEC (International Electrochemical Commission) guide on development and operation of laboratory proficiency testing,175 and a standard practice has been developed in the US.176 The essential feature is the establishment of a soundly based statistical protocol, both for the design of the exercise and the analysis of the results. When a performance index or other measure of performance is to be calculated it is necessary for the organizer to establish a target value for each sample, which may be based on a value determined by a definitive method or a certified figure where a reference material is used.More commonly the consensus mean is used, but this can be affected by outliers or a skewed distribution of results. This has been demonstrated in a study reported by the Analytical Methods Committee of the Royal Society of Chemistry ,177 which has suggested the use of other criteria, such as the median or a trimmed mean.178 Performance assessments also require an estimate of variance, usually expressed as the standard deviation, which again is dependent on the policy adopted for the treatment of outliers. Here again, the Analytical Methods Committee has highlighted the effects of including or exclud- ing results from aberrant laboratories.179 There is clearly still scope for rationalization.Conclusions Despite the numerous reports in the literature which have highlighted the inadequacy of analytical measurement data, it is clear that considerable progress has been made to improve matters during the past decade. For example, there is significant evidence for the increasing validation of standard methods, together with much improved formats adopted for the presentation of their texts, the provision of information regarding the precision of methods and the increasing availability of RMs which allows the truenesshias of methods to be monitored. Nevertheless, there is no room for com- placency at this point in time. Much analytical data is still of poor quality as many analysts and laboratories who have started to implement QA procedures have only done so in a less than rigorous and comprehensive manner.For example, it is not sufficient to participate in proficiency testing schemes if the methods employed have not been thoroughly validated and the laboratory has not been properly assessed (accredited) by a third party. Thus, although there are now grounds for optimism that analytical QC in many laboratories can meet the high standards demanded by accreditation bodies such as NAMAS, many more still have much to do. With the increasing need for measurement compatibility between countries which will be required following completion of the European Community's single market in 1992 it is essential that those in the latter category should address the problem now.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 References Hendzel, M. R., Fallis, B. W., and deMarch, B. G. E.,J. Assoc. Off. Anal. Chem., 1986, 69, 863. Analytical Quality Control (Harmonised Monitoring) Cornmit- tee. Analyst, 1985, 110, 1. Dembicki, H., Geochim. Cosmochim. Acta, 1984,48. 2641. Alford-Stevens, A. L., Budde, W. L.. and Bellar, T. A., Anal. Chem., 1985,57, 2452. Musial, C. J., and Uthe, J. F.,J. Assoc. Off. Anal. Chem., 1983, 66,22. Stack, M. V., and Delves. H. T., in Collaborative Interlabora- tory Studies in Chemical Analysis, eds. Egan, H., and West, T. S., Pergamon. Oxford, 1982. p. 115. Subramanian, K. S., and Stoeppler, M., Fresenius 2. Anal. Chem., 1986,323, 875. Taylor, A., Starkey, B.J., and Walker, A. W., Ann. Clin. Biochem., 1985, 22, 351. Horwitz. W., Kamps, L. R., and Boyer. K. W., J. Assoc. Uff. Anal. Chem.. 1980,63, 1344. Horwitz, W., Anal. Chem., 1982, 54, 67A. Horwitz, W., and Albert, R., Anal. Proc.. 1987, 24, 49. Knochel, A., and Petersen. W., Fresenius Z. Anal. Chem.. 1983,314, 105. Sherlock, J. C., Evans, W. H., Hislop, J., Kay, J. J.. Law, R., McWeeny, D. J., Smart, G. A., Topping, G., and Wood, R., Chem. Br., 1985, 21, 1019. Woodford. R. C., Chem. Br., 1986, 22, 111. Bee. H. M., Chem. Br., 1986.22, 111. Westway, H., Chem. Br., 1986, 22, 111. Buekett, J.. Chem. Br., 1986. 22. 112. Burns, D. T., Chem. Br.. 1986, 22, 306. Rickwood, P. C., Chem. Br.. 1986, 22, 412. Anderson, R. K., Chem. Br., 1986.22.523. Gardner. M. J., Chem.Br., 1986,22, 523. Sherlock, J. C., Evans, W. H., Hislop. J., Kay, J. J.. Law, R., McWeeny, D. J., Smart, G. A., Topping, G., and Wood, R., Chem. Br., 1986,22, 112, 627. Mason, M. F., J . Anal. Toxicol., 1981, 5 , 201. Cali, J. P., Fed. Proc. Fed. Am. SOC. Exp. Biol., 1975.34.2123. Griepink, B., Fresenius Z. Anal. Chem.. 1984. 317, 210.ANALYST, OCTOBER 1991, VOL. 116 989 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 New Reference Materials. Improvement of Methods of Measure- ment, ed. Marchandise. H., EUR 9921 EN, Commission of the European Communities, Brussels, 1985. Marchandise. H., Fresenius Z. Anal. Chem., 1987, 326, 613. Van Egmond, H. P., and Wagstaffe. P. J., Food Addit. Contam., 1989,6, 307.Griepink, B., Quim. Anal., 1989, 8, 1. Pszonicki, L., Anal. Chim. Acta, 1985, 176, 213. Byrne, A. R., Camara-Rica, C., Cornelis, R., de Goeij, J. J. M., Iyengar, G. V., Kirkbright, G., Knapp, G., Parr, R. M., and Stoeppler, M., Fresenius Z . Anal. Chem., 1987,326, 723. Topping, G.. Sci. Total Environ.. 1986, 49, 9. Wilson, A. L., Analyst, 1979, 104, 273. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1979, 104, 290. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1982, 107, 680. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1982, 107, 1407. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1983, 108, 1365. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1984, 109,431.Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1985, 110, 97. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst, 1985, 110, 103. Analytical Quality Control (Harmonised Monitoring) Commit- tee, Analyst. 1985, 110, 247. Gardner. M. J., Hunt, D. T. E., and Topping. G., Water Sci. Technol.. 1986, 18, 35. Parshin, A. K., Obol'yaninova. V. G., and Sul'dina, N. P., Zh. Anal. Khim., 1986, 41, 453; Chem. Abstr., 1986, 105, 17343~. Smart, N. A., Analyst. 1984, 109,781. Weber, J. P., Sci. Total Environ., 1988, 71, 111. Guidelines for Assessing the Technical Competence of Testing Laboratories. I S 0 Guide 25, International Organization for Standardization, Geneva, 1978. General Criteria for the Operation of Testing Laboratories.European Standard EN 45001, 1989; British Standard BS 7501, 1989. NAMAS Accreditation Standard M10, NAMAS Executive, National Physical Laboratory, Teddington, 1989. Good Laboratory Practice in the Testing of Chemicals, Final Report of the OECD Expert Group on Good Laboratory Practice. OECD, Paris, 1982. Good Laboratory Practice. The United Kingdom Compliance Programme, Department of Health, London, 1989. Quality Systems, British Standard BS 5750; International Standards I S 0 9000-9004. Garfield, F. M., Quality Assurance Principles for Analytical Laboratories. Association of Official Analytical Chemists, Arlington VA, 1984. Taylor, J. K., Quality Assurance of Chemical Measurements, Lewis, Chelsea, MI, 1987. Dux, J.P., Handbook of Quality Assurance for the Analytical Chemistry Laboratory, Van Nostrand Reinhold, New York, 1986. Cheeseman, R. V., and Wilson. A. L.. Technical Report TR 66, Water Research Centre, Medmenham, 1978. Taylor, J . K., Anal. Chem.. 1981.53, 1588A. Taylor, I. S., Chem. Aust.. 1983, 50, 82. Dux, J. P., Am. Lab., 1983, 15(7), 54. ACS Committee on Environmental Improvement, Anal. Chem., 1983, 55, 2210. Obrusnik, I.. Chem. Listy, 1987, 81, 1256; Chem. Abstr.. 1988, 108, 123410q. Analytical Methods Committee, Analyst, 1989, 114, 1497. Official Methods of Analysis, Association of Official Analytical Chemists. Arlington, VA, 15th edn., 1989. Youden. W. J . . and Steiner, E. H.. Statistical Manual of the AOAC, 3rd printing, Association of Official Analytical Chemists, Arlington. VA, 1982.Wernimont, G. T., in Use of Statistics to Develop and Evaluate Analytical Methods. ed. Spendley, H., Association of Official Analytical Chemists, Arlington. VA. 1985. ISO/TC 34/SC 11, Committee Document N 836, International Organization for Standardization. Geneva, 1986. 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 Precision of Test Methods-Determination of Repeatability and Reproducibility for a Standard Test Method by Interlaboratory Tests, International Standard IS0 5725-1986. Methods of Analysis of Fats and Fatty Oils, British Standard BS 684. Collaborative Interlaboratory Studies in Chemical Analysis, eds. Egan, H., and West, T. S., Pergamon, Oxford, 1982.Honvitz, W., Pure Appl. Chem., 1988, 60, 855. J. Assoc. Off. Anal. Chem., 1988,71, 161. Karpinski, K. F., J. Assoc. Off. Anal. Chem., 1989, 72, 931. Precision Characteristics of Analytical Methods-Outline of Collaborative Study Procedure, International Dairy Federation Standard 135A: 1988. Vegetable and Animal Oils and Fats-Determination of Repeatability and Reproducibility of Methods of Analysis by Interlaboratory Tests, Netherlands Normalisation Institute Stan- dard NEN 6303, 1988. Recommendations for a Check List of Information Required to Evaluate Methods of Analysis, Codex Alimentarius Commission CWMAS 81/70, Food and Agriculture Organization, Rome, 1981. Zaalberg. J., J. Assoc. Off. Anal. Chem.. 1989, 72, 34. Arsenic in Potable Waters by Atomic Absorption Spectropho- tometry, Methods for the Examination of Waters and Asso- ciated Materials, Department of the EnvironmentWater Research Council, HM Stationery Office, London, 1982.Analyses of Hazardous Substances in Biological Materials. Verlagsgesellschaft, Weinheim, 1988, vol. 2. Manual of Pesticide Residue Analysis, Deutsche Forschungs- gemeinschaft, Bonn, 1987. vol. 1. Analytical Methods Committee, Analyst, 1987, 112, 199. Pocklington, W. D., Guidelines for the Development of Stan- dard Methods by Collaborative Study, Laboratory of the Government Chemist, Teddington, 5th edn., 1990. Pocklington, W. D., Pure Appl. Chem., 1990, 62, 149. Peeler, J . T., Honvitz, W., and Albert, R., J. Assoc. Off. Anal. Chem., 1989, 72, 784. deRuig, W. G., Stephany, R. W.. and Dijkstra, G., J.Assoc. Off Anal. Chem., 1989, 72, 487. Layouts for Standards-Part 2: Standard for Chemical Analysis, International Standard I S 0 78/2, 1st edn., 1982-04-01. Directory of Certified Reference Material (CRM) Sources of Supply and Suggested Uses, International Organization for Standardization, Geneva, Switzerland, 1982. Dams, R., Pure Appl. Chem., 1983, 55. 1957. Boumans, P. W. J. M., and Roelandts, I., Spectrochim. Acta, Part B , 1989, 44, 1. Alvarez, R., Fresenius Z . Anal. Chem., 1986, 324, 376. Landsberger, S., Chem. Ecol., 1984, 2, 3. Wells, D. E., Fresenius 2. Anal. Chem., 1980, 332, 583. Roelandts. I . , Spectrochim. Acta, Part B , 1989, 44. 5. Roelandts, I., Spectrochim. Acta, Part B , 1989, 44, 281. Gladney, E. S., O'Malley, B. T., Roelandts, I., and Gills, T.E., Standard Reference Materials: Compilation of Elemental Con- centration Data for NBS Clinical, Biological, Geological and Environmental Standard Reference Materials, NBS Special Publication 620-111, US Government Printing Office, Washing- ton, DC, 1987. I S 0 Guide 30, Terms and Definitions Used in Connection With Reference Materials, International Organization for Standardi- zation, Geneva, Switzerland, 1981. I S 0 Guide 31. Contents of Certificates of Reference Materials, International Organization for Standardization, Geneva, Swit- zerland, 1981. IS0 Guide 35, Certification of Reference Materials-General and Statistical Principles, International Organization for Stan- dardization, Geneva, Switzerland, 1985. Cali, J. P., and Stanley, C. L., Ann.Rev. Mat. Sci., 1975,s. 329. Cali. J. P., Fresenius Z. Anal. Chem., 1979, 297, 1. Coleman, R. F., J. Autom. Chem., 1980, 2, 183. Cali. J . P., Med. Instrum., 1974, 8, 17. I S 0 Recommendation R-1000, International Organization for Standardization. Geneva, Switzerland, 1969. International Vocabulary of Basic and General Terms in Metrology, International Organization for Standardization, Geneva, Switzerland, 1984. Reed, W. P., The Role of Standard Reference Materials in Achieving Measurement Traceability. NBS Special Publication 582, US Dept. of Commerce, 1980.990 ANALYST, OCTOBER 1991, VOL. 116 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 Sappenfield, K.H., Marinenko, G., and Hague, J. L., Comparison of Redox Standards, NBS Special Publication 260-24, US Dept. of Commerce, Washington, DC, 1972. Head, A. J., Hierarchy and Traceability of Certified Reference Materials, ISO-REMCO paper 181, International Organization for Standardization, Geneva, Switzerland, 1989. Cali, J. P., Mears, T. W., Michaelis, R. E., Reed, W. P., Seward, S. W., Stanley, C. L., and Yolken, H. T., The Role of Standard Reference Materials in Measurement Systems, NBS Monograph 148, US Dept. of Commerce, Washington, DC, 1975. Uriano, G. A., and Cali, J. P., ACS Symp. Ser., 1977,63, 140. Cali, J. P., and Reed, W. P., The Role of National Bureau of Standards SRMs in Accurate Trace Analysis, Proceedings of a Symposium on Accuracy in Trace Analysis, NBS Special Publication No.422, Government Printing Office, Washington, DC, 1976. Uriano, G. A., and Cravatt, C. C., CRC Crit. Rev. Anal. Chem., 1977,361. Taylor, J. K., J. Test. Eval., 1983, 11, 385. Marchandise, H., and Colinet, E., Fresenius Z . Anal. Chem., 1983,316,669. Frauerwieser, G., Mikrochim. Acta, Part I I , 1982, 55. Guzzi, G., Colombo, A., Girardi, F., Pietra, R., Rossi, G., and Toussaint, N., J. Radioanal. Chem., 1977, 39, 263. Dragoo, A. L. Powder Diffr., 1986, 1, 294. De Goeij, J. J. M., Volkers, K. L., Tjioe, P. S., and Kroon, J. J., Radiochem. Radioanal. Lett., 1978, 35, 139. Pszonicki, L., Veglia, A., and Suschny, O., Report on Intercom- parison Air-311 on the Determination of Trace Elements in Simulated Air Filters, Report IAEA/RL/95, IAEA, Vienna, 1982.Michaelis, R. E., Pure Appl. Chem., 1977,49, 1483. Rasberry, S. D., Am. Lab., 1987, 19, 130. Rasberry, S. D., J. Res. Natl. Bur. Stand. (US), 1988, 93, 213. Marschal, A,, Future Trends for CRM Development, Certifica- tion and Use, Proceedings of ISCRM 89, Beijing, China, International Academic Publishers, Oxford, 1989. Lilliefors, H. W., Am. Stat. Assoc. J., 1967, 62, 309. Methodes Sratistiques, Commissariat a I’Energie Atomique, Dunod, Paris, 1969. Snedecor, G. W., and Cochran, W. G., Statistical Methods, Iowa State University Press, Ames, IA, 6th edn., 7th print, 1974. Taylor, J. K., Handbook for SRM Users, NBS Special Publica- tion 260-100, US Dept. of Commerce, Washington, DC, 1985. Sutarno, R., and Bowman, W. S., in Workshop on Reference Materials, CANMET, Energy, Mines and Resources, Ottawa, Canada, 1982, p. 13. Sutarno, R., and Steger, H. F., Talanta, 1985,32, 1088. De Goeij, J. J. M., Kosta. L., Byrne, A. R., and Kucera, J., Anal. Chim. Acta, 1983, 146, 161. Marlow, W. F., and Lafleur, P. D., Proc. Symp. Nucl. Tech. Environ. Pollut., 1971, 91. Brzezinska, A., Eur. Spectrosc. News, 1982,40, 19. Landsberger, S . , Chem. Ecol., 1984, 2, 3. Christie, 0. H. J., and Alfsen, K. H., Geostand. Newsl., 1975, 1, 47. Christie, 0. H. J., Fresenius Z. Anal. Chem., 1979, 297, 20. Box, G. E. P., and Cox, D. R., J. R. Stat. Soc., 1964, B26,211. Abbey, S . , Meeds, R. A., and Belanger, P. G., Geostand. Newsl., 1979, 3, 121. Ellis, P. J., Copelowitz, I., and Steele, T. W., Geostand. Newsl., 1977.3, 123. Parr, R. M., Muramatsu, Y., andclements, S. A., FreseniusZ. Anal. Chem., 1987,326, 601. Parr, R. M., Schelewz, R., and Ballestra, S., Fresenius 2. Anal. Chem., 1988,332, 518. Sampling Procedures for Inspection by Attributes, British Standard BS 6001. Specification for Sampling Procedures and Charts for Inspection by Variables for Percent Defective, British Standard BS 6002, 1979. Kratochvil, B., and Taylor, J. K., Anal. Chem., l981,53,924A. Methods for Sampling Chemical Products, British Standard BS 5309, 1976. 142 Taylor, J. K., TrAC, Trends Anal. Chem. (Pers. Ed.), 1986,5, 121. 143 Garfield, F. M., J. Assoc. Off. Anal. Chem., 1989, 72, 405. 144 Kratochvil, B. G., and Taylor, J. K., NBS Tech. Note (US), 1153, National Institute of Standards and Technology, Gaithers- burg, MD, 1982. 145 Nomenclature, Symbols, Units and Their Usage in Spectrochem- ical Analysis-II, Spectrochim. Acta, Part B , 1978, 33, 242. 146 Currie, L. A., Anal. Chem., 1968, 40, 586. 147 Shewhart, W. A., Economic Control of the Quality of Manufac- tured Products, Macmillan, London, 1931. 148 Analytical Monitoring in Workplace Air Monitoring, Health and Safety Executive, Methods for the Determination of Hazardous Substances MDHS 71, London, 1991. 149 Data Analysis and Quality Control Using Cusum Techniques, British Standard BS 5703, 1982. 150 Needham, L. L., Burse, V. W., Korver, M. P., Lapeza, C. R., Liddle, J. A., Bayse, D. D., and Price, H. A., J. Anal. Toxicol., 1983, 7,279. 151 Delves, H. T., Anal. Proc., 1984, 21, 391. 152 Jorhem, L., and Slorach, S., Fresenius Z. Anal. Chem., 1988, 332, 738. 153 National Association of Testing Authorities, Chatswood, NSW, Australia. 154 Testing Laboratory Registration Council of New Zealand, Auckland, New Zealand. 155 National Voluntary Laboratory Accreditation Program, Gaithersburg, MD. 156 Mesley, R. J., Int. Analyst, 1987, 1(6), 15. 157 NAMAS Information Sheet NIS xx, Accreditation for Analytical Laboratories, NAMAS Executive, Teddington, 1990. 158 Council Directive 87/18/EEC, Off. J. Eur. Comm., 17.1.87, L15, 29. 159 Council Directive 88/320/EEC, Off. J. Eur. Comm., 11.6.88, L145, 35. 160 Smith, R., Talanta, 1984, 31, 537. 161 Davison, W., and Gardner, M. J., Anal. Chim. Acta, 1986,182, 17. 162 AOAC Committee on Collaborative Studies, Anal. Chem., 1978,SO. 337A. 163 Horwitz, W., in Collaborative Interlaboratory Studies in Chem- ical Analysis, eds. Egan, H., and West, T. S., Pergamon Press, Oxford, 1982, p. 45. 164 Horwitz, W., J. Assoc. Off. Anal. Chem., 1983, 66, 455. 165 Eastwood, I. W., and Jackson, K. W., Environ. Pollut., Ser. B , 1984.8, 231. 166 Wagemann, R., and Armstrong, F. A. J., Talanta, 1988, 35, 545. 167 Gardner, M. J . , and Dobbs, A. J., Chem. Br., 1988,24, 875. 168 Mesley. R. J . , and Sargent, M., Chem. Br., 1989, 25, 30. 169 Aquacheck. Water Research Centre, Medmenham, Bucking- hamshire. 170 Regular Interlaboratory Counting Exchange (RICE), Institute of Occupational Medicine, Edinburgh. 171 West, N. G., Anal. Proc., 1986, 23, 330. 172 Whitehead, T. P., and Woodford, F. P., J. Clin. Pathof.. 1981. 34,947. 173 Taylor, A., and Briggs, R. J., J. Anal. At. Spectrom., 1986, 1, 391. 174 Wood, R., and Patey, A., Food Analysis Performance Assess- ment Scheme, Ministry of Agriculture. Fisheries and Food, Norwich. Development and Operation of Laboratory Proficiency Testing, ISO/IEC Guide 43. International Organization for Standardiza- tion, Geneva, 1984. Standard Practice for Conducting an Interlaboratory Test Pro- gram to Determine the Precision of Test Methods, ASTM E 691-79, 1980 Annual Book of ASTM Standards, Part 41, American Society for Testing and Materials, Philadelphia. Analytical Methods Committee, Analyst, 1989, 114, 1489. Analytical Methods Committee, Analyst, 1989. 114, 1693. Analytical Methods Committee, Analyst, 1989, 114, 1699. 175 176 177 178 179 Paper 1 /00077B Received January 7th, 1991 Accepted March 28th, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911600975

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 991 Voltammetric Behaviour of Morphine at a Glassy Carbon Electrode and Its Determination in Human Serum by Liquid Chromatography With Electrochemical Detection Under Basic Conditions Phillip H. Jordan Department of Chemical Patholog y, Royal Devon and Exeter Hospital, Church Lane, Heavitree, Exeter EX2 5AD, UK John P. Hart* Department of Science, Bristol Polytechnic, Coldharbour Lane, Frenchay, Bristol BS16 IQY, UK The electrochemical oxidation of morphine was studied at pH values of between 7.00 and 12.00 by cyclic voltammetry and chronoamperometry at a planar glassy carbon electrode. The peak potential was dependent on pH over the range 7.00-9.75; it was independent of pH above the latter value, indicating a pK, value of 9.75. The peak current was found to be independent of pH, ionic strength of phosphate buffer (0.02-0.1 rnol dm-3) and percentage of acetonitrile (040% v/v).The oxidation was found to occur in three steps; these are considered to result from a one-electron oxidation of the phenoxide group, followed by a one-electron loss from the oxidation product, pseudomorphine, and finally a two-electron loss from a tertiary amine group. A simple method of analysis by high-performance liquid chromatography was developed which employed a column packed with a reversed-phase, pH-sta ble, octadecylsilane-modified silica. Separation was achieved with a mobile phase containing 20% v/v acetonitrile in 0.05 rnol dm-3 phosphate buffer, pH 11.0. Amperometric detection was carried out with an applied potential of +0.45 V versus Ag-AgCI.The detection limit was 1.24 x 10-13 rnol of morphine injected. The detector gave a linear response from 1.2 x 10-12 to 4.0 x 10-10 rnol of morphine injected. The extraction method required 0.5 ml of serum, and no solvent evaporation was needed. The recovery of morphine was 80.9%. The method gave a linear response to at least 15.0 x rnol dm-3. The relative standard deviation was 4.95% at 0.75 x 10-7 rnol dm-3 and 3.73% at 3.0 x 10-7 rnol dm-3. Keywords: Morphine; cyclic voltammetry; liquid chromatography; electrochemical detection; basic conditions When morphine (I) is given for the treatment of severe pain in cancer patients by continous subcutaneous administration,’ or for the relief of pain following surgical procedures,2 excessive doses must be avoided as they cause symptoms which are similar to shock and which can be fatal, whereas an insufficient dose will fail to reduce the pain.Hence, monitoring of blood plasma levels is often performed.3~~ Therapeutic concentra- tions of morphine in plasma are at the nanomolar level.5 Any method for clinical application must therefore be sufficiently sensitive. The presence of metabolites which are very similar in structure demands selectivity. Radioimmunoassays have been developed that are very sensitive, and are therefore suitable for the levels of morphine found in plasma; however, antibody cross-reactivity with glucuronide metabolites has been reported.6 Gas chromato- graphic methods have been available for many years,’ and are still being reported, particularly with the use of mass spectrometry,*-lO but almost all require prior derivatization of morphine.Gas chromatography-mass spectrometry is sensi- tive and selective , but requires expensive equipment and skilled operators. A method for the determination of morphine by ‘high- * To whom correspondence should be addressed. speed’ liquid chromatography was reported as early as 1972.” A true ‘high-performance’ method was reported in 1975.12 This procedure made use of a fluorescence detection tech- nique by reacting extracts of urine samples with potassium hexacyanoferrate( HI) to form pseudomorphine. The reaction could proceed on-column as an alkaline mobile phase was used. Post-column derivatization with the same reagent has also been achieved.13 Both techniques suffered from a lack of sensitivity; the detection limit for the latter technique was 3.4 x 10-11 mol injected.Several reports have been published which discuss the use of liquid chromatography with electrochemical detection (LCEC) for the determination of morphine. These methods involve the use of acidic mobile phases, and therefore detector potentials in the range from k0.614 to +1.0 V15716 are necessary. It would be desirable to employ basic mobile phases for the detection of species undergoing oxidation processes because lower applied potentials could be used; this might be expected to enhance both sensitivity and selectivity for morphine. Unfortunately, most reversed-phase columns have to be used within the pH range 3-7 in order to avoid dissolution of the stationary phase. Modified silica-based packing material which is pH-stable has recently become available however, and this allows for the possibility of using mobile phases at high pH values.This paper describes the use of a modified silica material which can be used from pH 2 to 13 for the separation of morphine at a basic pH. It was thought that the use of a high pH would facilitate the electrochemical oxidation of mor- phine, requiring the use of a low applied potential, which would improve selectivity, and hence enable a simple extrac- tion procedure to be employed. It was also assumed that the operation of the column at a pH greater than the pK, of morphine would avoid the silanol interactions which cause severe peak asymmetry when silica columns are used to separate basic solutes.992 ANALYST, OCTOBER 1991, VOL.116 As no systematic detailed studies of the oxidation of morphine at a glassy carbon electrode over the pH range of interest have previously been reported, these formed the first part of the present investigation. Experimental Reagents and Standards Acetonitrile, methanol and dipotassium hydrogen phosphate were of HiPerSolv quality and were supplied by Merck (Poole, Dorset, UK). All of the other solvents and chemicals were of AnalaR grade, unless stated otherwise, and were also obtained from Merck. Morphine sulphate pentahydrate, hydromorphone hydrochloride and codeine were all supplied by Sigma (Poole, Dorset, UK). In order to avoid any confusion which might result from the use of gravimetric units, as morphine can be obtained in the form of various salts and hydrated salts, stock solutions that contained 0.03 and 6 X 10-5 rnol dm-3 morphine in methanol were prepared.Molar units are used in the text wherever possible. Apparatus Cyclic voltammetry A triangular wave generator (Model PP2, Oxford Electrodes, Oxford, UK) was connected to a three-electrode cell consist- ing of a glassy carbon working electrode, a platinum wire counter electrode and a saturated calomel reference electrode (SCE). The signals were registered on an x-y recorder (Model PN4, J.J. Instruments, Southampton, UK). Chronoamperometry A three-electrode cell, having the same construction as that used for cyclic voltammetry, was connected to a laboratory- constructed potentiostat containing an LF351 operational amplifier as the voltage follower, and an AD549JH opera- tional amplifier for the current to voltage conversion.The potentiostat was controlled via a Flight Electronics 12 bit analogue to digitavdigital to analogue converter by an Opus PC-XT computer, which was programmed in Turbo Pascal. l6 The signals were registered on an x-y recorder and simul- taneously captured by the computer via the analogue to digital converter; the computer then performed all of the calcula- tions. High-performance liquid chromatography The pH-stable modified silica packing material (2.5 g) (Type S5W-C18, Phase Separations, Queensferry, Clwyd, UK) was suspended in 40 ml of acetone by placing in an ultrasonic bath (Decon, Hove, Sussex, UK) for 10 min, and the slurry thus prepared was then pumped into the column at 44.8 MPa, using acetone, by means of an air-driven pump (Phase Separations). The packed column was then washed by pumping with methanol at 1 ml min-1 for 1 h before use.The high-performance liquid chromatography (HPLC) system consisted of a high-pressure pump (Model 510, Millipore, Waters Chromatography Division, Harrow, Mid- dlesex, UK) and a valve injector (Model 7125, Rheodyne, Cotati, CA, USA) which contained a 50 p.1 loop, and a Tefzel I pH-resistant rotor seal. The amperometric detector was equipped with a thin-layer cell having a glassy carbon electrode (Type TL5, Bioanalytical Systems, West Lafayette, IN, USA) and an Ag-AgC1 reference electrode, connected to a potentiostat/amplifier (Model LC4B, Bioanalytical Systems).The chromatograms were recorded on a chart recorder (Model LS14, Linseis, Selb, Germany) or an integrator (Model 3380A, Hewlett-Packard, Winnersh, Wokingham, Berkshire, UK). Extraction Disposable 15 ml polypropylene centrifuge tubes and 0.5 ml conical tubes (Cat. Nos. 57.513 and 72.669, Sarstedt, Beau- mont Leys, Leicestershire, UK) were used as supplied. Procedures Cyclic voltammetry Cyclic voltammograms were recorded from -0.5 to +1.0 V (versus SCE) at pH values between 7.0 and 12.0, and the effect of pH on the peak current ( i p ) and peak potential ( E p ) values was determined. The effect of buffer ionic strength (0.02-0.1 rnol dm-3), acetonitrile concentration (040% v/v) and scan rate on the i, values was then investigated. The glassy carbon working electrode was cleaned between each scan by rinsing with distilled water, polishing the surface by rubbing with aluminium oxide, rinsing again with distilled water and finally drying with soft, absorbent tissues. Chronoamperometry An initial potential ( E l ) of +0.03 V versus SCE was applied to the cell for 15 s, followed by a final potential (E2) of +0.55 V. These potentials correspond to the foot of the first oxidation peak of morphine determined by cyclic voltammetry, and the trough after the oxidation peak.The current-time response signal was then recorded for approximately 50 s using a current sampling rate of 9.5 samples s-1. Three tests were performed at each pH value. The area of the glassy carbon working electrode had previously been calculated from the chronoamperometric data of hexacyanoferrate(u), the oxidation number of which was known.As the diffusion coefficient of morphine was not known, the diffusion coefficient for heroin (diaceytl mor- phine), which is a very similar molecule, was taken from published data17 and was used in the calculations. Hydrodynamic voltammetry The hydrodynamic voltammogram for morphine was recorded by the injection of 20 p1 of a 6 x 10-7 rnol dm-3 solution of morphine in the phosphate buffer mobile phase. Determination of morphine in human serum by LCEC Without internal standard. ( a ) Place 0.5 ml of serum in a disposable 15 ml poly(propy1ene) centrifuge tube and add 0.5 ml of 0.2 rnol dm-3 borate buffer at pH 9.0. ( b ) Add 2 ml of chloroform-propan-2-01 (9 + l ) , stopper, then vortex mix for 2 min and centrifuge for 5 min at 2000g.( c ) Discard the aqueous phase and transfer the organic phase into a clean 15 ml poly(propy1ene) centrifuge tube. ( d ) Add 0.2 ml of 0.05 rnol dm-3 orthophosphoric acid, vortex mix for 2 min, then centrifuge for 5 min at 2000g. ( e ) Place 0.1 ml of the aqueous phase in a disposable 0.5 ml poly(propy1ene) tube, add 0.02 ml of 0.5 rnol dm-3 phosphate buffer at pH 11.0, mix and inject 0.05 ml onto the column. With internal standard. A solution of the internal standard (hydromorphone) was prepared so that the final concentra- tion in distilled water was 6.0 x 10-6 rnol dm-3. A 0.05 ml aliquot of this solution was added to 0.5 ml of serum at stage ( a ) described above. Calibration, Recovery and Precision of LCEC A calibration graph of peak current versus mass of morphine injected was constructed over the range from 1.2 X lo-'* to 1.2 X 10-9 mol.Aliquots of serum (0.5 ml) were spiked with from 75 x 10-12 to 750 x 10-12 rnol of morphine; a further 0.5 ml aliquot was not spiked and all of the samples were extracted and analysed by the method described above, without the addition of an internal standard. A similar study wasANALYST, OCTOBER 1991, VOL. 116 993 performed with the internal standard, hydromorphone, included in the extraction procedure. In order to assess the precision of the method, aliquots of drug-free serum were spiked with morphine to produce two concentrations, 0.75 x 10-7 and 3.0 x 10-7 mol dm-3; ten samples were analysed at each concentration. A I Results and Discussion Cyclic Voltammetry, Chronoamperometry, Hydrodynamic Voltammetry and Optimization of Conditions Figs.1 and 2 show the cyclic voltammograms recorded at pH 10.0 with two different y-axis scale expansions. The scan rate in each instance was 50 mV s- 1 . In Fig. 1 , the first forward scan shows one well-defined anodic peak (l), followed by another less obvious peak (2). The first peak is probably due to oxidation of the phenoxide ion at the 3-position (see under Hydrodynamic voltammetry), and the second could be due to either further oxidation of the pseudomorphine generated (11), or to oxidation of the tertiary amine; this is discussed later, but evidence suggested that peak 2 is almost certainly due to oxidation of pseudomorphine.Examination of the structure of pseudomorphine (11) reveals that it possesses two phenolic groups. These will be in the form of phenoxide ions at basic pH values and will therefore be susceptible to further oxidation. &OH "3c"& HO Two cathodic peaks (3 and 4) can be seen in Fig. 2 on the first and second reverse scans. Atlthough the exact position of peak 3 is difficult to determine accurately, it seems likely that anodic peak 1 and cathodic peak 3 are a reversible couple as the separation between the peak potentials (AE,) was shown to be about 59 mV.18 This assumes that the number of electrons is 1; in fact this was found to be true by using chronoamperometry , and is discussed later. The absence of a well-defined cathodic peak on the reverse scan (Fig.1) is probably due to the formation of pseudomor- phine, which is then oxidized at a more positive potential. Cathodic peak 4 and anodic peak 5 also seem to be a reversible couple. As peak 5 is only observed on the second forward scan, it is clear that a new species is being formed by the reduction process resulting in peak 4. Peak 5, which occurs at approximately -0.1 V uersus SCE, suggests that a highly conjugated species might be formed, as this type of structure readily undergoes oxidationheduction reactions. These phenomena were observed at an of the pH values examined. It was found that the value of Ep for peak 1 shifted to more negative potentials with increasing pH. This variation is shown in Fig. 3. It can be seen, however, that above pH 10.0, the oxidation was independent of pH. - 4 , I" , -0.4 0 +0.4 +0.8 PotentialN versus SCE Fig.1 Cyclic voltammogram of 1 x 10-4 rnol dm-3 morphine in 0.05 mol dm-3 phosphate buffer at pH 10.0 using a glassy carbon electrode. Initial potential, -0.5 V versus SCE; scan rate, 50 mV s-l. 1F and 2F, first and second forward scans; 1R and 2R, first and second reverse scans. For details of peaks see text 1 I I 0 +0.4 PotentialN versus SCE Fig. 2 Cyclic voltammogram of 1 x 10-4 rnol dm-3 morphine in 0.05 rnol dm-3 phosphate buffer at pH 10.0 using a glassy carbon electrode. Other conditions as in Fig. 1 but using a different y-axis scale expansion. 1F and 2F, first and second forward scans; 1R and 2R, first and second reverse scans. For details of peaks see text Extrapolation of the pH-dependent and pH-independent sections of the relationship shows that a break occurs at approximately pH 9.75, which indicates a pK, of approxi- mately 9.75.This is in agreement with published data. Below this break, the E, values obey the relationship: E, = 0.8843 - 0.067pH. The value of i, was calculated for peak 1 at each pH value. The mean i, was 4.8 PA, with a standard deviation of 0.255 PA and a relative standard deviation of 5.3%. Hence, i, is independent of pH over the range examined. Values of the electrochemical transfer coefficient, an, (where a is the transfer coefficient and n, is the number of electrons involved), were calculated for peak 1 at each pH value, according to the following equation: 19 0.048 an, = ~ where EPl2 is the the potential at ip12.The results are shown in Table 1. The values of an, suggest that the same over-all mechanism occurs at each pH value and they are consistent with a one-electron oxidation. E p - Ep12ANALYST, OCTOBER 1991, VOL. 116 994 0.5 $i 0.4 7 2 0.3 2 2 0.2 v) 0.1 0 I I I I I I I I I I 1 I I 7.0 8.0 9.0 10.0 11.0 12.0 PH Fig. 3 Effect of pH on peak potential for peak 1. Morphine (1 x mol dm-3) in 0.05 rnol dm-3 phosphate buffer Table 1 Values of ana for peak 1 (cyclic voltammetry) at each pH value PH aria 7.0 0.87 8.0 0.96 9.0 0.80 10.0 0.96 11 .o 0.80 12.0 0.80 ?Oo0 > ? 600 400 0 0.1 0.2 0.3 0.4 0.5 I &; s-: Fig. 4 Graph of idcv' versus va. Morphine (2.5 X 10-5 rnol dm-3) in 0.05 rnol dm-3 phosphate buffer at pH 11.0 14 12 10 P 2 8 F $ 6 4 2 Well-defined signals were obtained with all of the phos- phate buffers investigated.At higher pH values, however, the Ep became less positive, and at pH 210 was least positive. Electrochemical detection is best performed at lower poten- tials, as the noise levels are lower; the signal-to-noise ratios are therefore better. It was considered that a phosphate buffer of pH 11.0 would be the most appropriate, as buffering capacity is at a maximum at this pH. The effect of buffer molarity on the height of peak 1 was examined at pH 11.0 using phosphate buffers ranging from 0.02 to 0.1 rnol dm-3. The i, was independent of buffer molarity over the range studied. A molarity of 0.05 rnol dm-3 was selected as lower molarities would have less buffering capacity and higher molarities would result in higher back- ground currents.The effect of the percentage volume of acetonitrile added on the i, of peak 1 was examined by diluting a stock solution of 0.5 rnol dm-3 phosphate buffer of pH 11.0 with various amounts of aceonitrile and water, so that the final buffer molarity of the mixtures obtained was 0.05 rnol dm-3. The i, was independent of the acetonitrile concentration over the range examined. Acetonitrile can there therefore be added to the HPLC mobile phase as required for the chromatographic separation, with no effect on the detector sensitivity. Methanol was not investigated as it was found that the chromatographic peak shapes deteriorated if this solvent was used (results not shown). It can be shown from the Randles-Sevcik equation20 that a plot of iplcvd versus vt will yield a positive slope if the species undergoes adsorption at the electrode surface, where c is the concentration of the species and v is the scan rate.Fig. 4 shows the graph of idcvi versus vt, which clearly does not exhibit a positive slope, and the trend is actually negative. It can therefore be concluded that the oxidation process is diffusion controlled with some degree of irreversibility. Chronoamperometry The current i at time t , following a stepped voltage at a planar electrode, is described by the Cottrell equation; this equation can be solved to determine the number of electrons involved in the electron transfer.16 The mean number of electrons transferred per molecule of morphine oxidized was 1.02 at pH 7.0 and 1.00 at pH 11.0.0 0.2 0.4 0.6 0.8 1 .o 1.2 PotentialN versus Ag-AgCI Fig. 5 Hydrodynamic voltammogram for morphine. Amount of morphine injected, 1.2 X lo-" mol. Voltammogram recorded with applied potentials from +0.15 to +1.10 V in increments of 0.05 V This confirms that the first morphine oxidation wave involves the loss of one electron, and that the same oxidation mechanism takes place at each pH value, as suggested by the electrochemical transfer coefficient. It is likely that the initial oxidation reaction produces free radicals which then dimerize to form pseudomorphine; further evidence for this was obtained by hydrodynamic voltammetry (see below). This has also been suggested by other workers using different electrode materials. 21.22 Hydrodynamic voltammetry The results of the cyclic voltammetric study showed that the best stability would occur at pH 11.0.A mobile phase of 0.05 rnol dm-3 phosphate buffer of pH 11.0 containing 20% v/v of acetonitrile was therefore used for the remainder of this investigation. A hydrodynamic voltammogram was recorded by injecting 1.2 x 10-11 rnol of morphine (3.5 x 10-9 g of morphine base) with the column in position. (The injection of larger amounts of morphine overloaded the detector at the higher applied potentials . ) Fig. 5 shows the voltammogram recorded with applied potentials from +0.15 to +1.10 V in increments of 0.05 V, and by means of duplicate injections. Three distinct waves can be seen. The first wave reaches a maximum at +0.45 V, the second at approximately +0.8 V and the third at approxi- mately +1.0 V.It can also be seen by comparing the sizes of the currents for the three oxidation waves that if the number of electrons transferred for the first oxidation is 1, as shown by chronoamperometry, then the number of electrons trans- ferred during the second oxidation is 1, and the number transferred during the third oxidation is 2. The first oxidation occurs at the phenoxide ion at the 3-position, the second is possibly due to further oxidation of the product, pseudomor- phine, and the third is consistent with a two-electron oxidation of the tertiary amine group to form a secondary amine and aANALYST, OCTOBER 1991, VOL. 116 i> 500 5 400 .- s 0 Y- E 300 .- z g200 E 100 995 - - - - - ketone. Oxidation with the loss of two electrons has been observed for diacetylmorphine,17 which is acetylated at the 3- and 6-positions. The mechanism, which was originally pro- posed for aliphatic tertiary amines,23 consists of the formation of an intermediate ammonium radical, which is converted into a quaternary Schiff's base, which is then rapidly hydrolysed to give a secondary amine and a ketone.This proposed mechanism was further investigated by injecting 3.33 x 10-9 rnol of codeine(II1). The only difference between the morphine and codeine molecules is that codeine has a methoxy group at the 3-position instead of a phenolic group. No response was observed at a potential of +0.45 V, and potentials of greater than +0.8 V were required to obtain any significant response. Hence, oxidation of the tertiary amine occurs at applied potentials of greater then +0.8 V, and the responses observed for morphine at potentials of less than +0.8 V must be related to the phenolic group. HO /v In order to determine accurately the potential required for the initial oxidation of the molecule, the voltammogram was repeated from +0.2 to +0.6 V using increments of 0.025 V between 0.35 and 0.55 V.This is shown in Fig. 6. It can be seen from this voltammogram that an applied potential of +0.45 V versus Ag-AgCI would achieve the best stability and selectivity. As the signal at +0.45 V is situated on a plateau of current values, a drift of k0.02 V would not significantly affect the current. Although higher potentials would achieve greater sensitivity, the selectivity for morphine afforded by the use of a lower potential could be lost.An applied potential of +0.45 V was therefore used for the remainder of the studies. Calibration, Recovery and Precision of LCEC A linear response was observed from 1.2 x 10-12 to 4 x 10-10 rnol of morphine injected. At the applied potential of +0.45 V versus Ag-AgC1, and for a signal-to-noise ratio of 3 : 1, the detection limit was found to be 1.24 X 10-13 rnol of morphine injected. 3.5 I 0 0.2 0.3 0.4 0.5 0.6 Potentia IN versus Ag-Ag C I Fig. 6 Hydrodynamic voltammogram for morphine. Amount of mor hine injected, 1.2 x 10-11 mol. Voltammogram recorded from +O.!to +0.6 V using increments of 0.025 V between 0.35 and 0.55 V The absolute recovery of morphine was determined by spiking 0.5 mi fractions of drug-free human serum with from 75 x 10-12 to 750 x 10-12 rnol of morphine (from 1.5 X 10-7 to 1.5 x 10-6 mol dm-3). These were then extracted using the proposed method.The same amounts of morphine were also diluted in mobile phase and injected directly without extrac- tion; these data are shown in Fig. 7. The recovery was 80.9 k 5.1% at the 95% confidence level. In order to compensate for the loss of morphine during the extraction procedure, an internal standard was included in the procedure. It was found that the addition of approximately 3.0 x 10-10 rnol of hydromorphone to 0.5 ml of serum would give a peak of similar magnitude to that of morphine at the upper limit of the therapeutic range (Fig. 8). The recovery of morphine was determined by the addition of an internal standard to serum samples; the recovery of morphine added to serum in the range from 0.75 x 10-7 to 15.0 x 10-7 rnol dm-3 was found to be linear.Under the conditions described, and for a signal-to-noise ratio of 3 : 1, it was found that the limit of quantification was 8.4 x 10-9 rnol dm-3 (2.4 pg 1-1 of morphine base). This limit 700 I 1 - E600 I N r V I I 1 1 I I I I 0 100 200 300 400 500 600 700 800 Morphine added/lO-'* rnol Fig. 7 Recovery of morphine added to serum nA I I I 0 3 6 1 nA rcrr I I I 0 3 6 Ti me/m i n Fig. 8 Chromatograms of ( a ) a blank serum and (b) a spiked serum. Morphine concentration, 2.8 x lo-' rnol dm-3. A, Morphine; and B, internal standard (hydromorphone)996 ANALYST, OCTOBER 1991, VOL. 116 could obviously be reduced, if required, by either extracting more than 0.5 ml of serum, by injection of more than 0.05 ml or by a combination of both.In order to assess the precision of the method, aliquots of drug-free serum were spiked with morphine to produce concentrations of 0.75 x 10-7 and 3.0 x 10-7 mol dm-3. These levels represent the lower and upper limits, respec- tively, of the therapeutic range. Ten 0.5 ml samples of each aliquot were then analysed by the proposed method. The relative standard deviation was 4.95% at 0.75 x 10-7 mol dm-3 and 3.73% at 3.0 x 10-7 mol dm-3. The studies described above clearly indicate that the proposed method is suitable for the determination of mor- phine in human serum. In conclusion, the main advantages of the proposed method of analysis are the simplicity and low cost of the extraction procedure, and the improvement in the chromatographic separation of morphine due to the use of a pH of 11 .O.Conventional modified silica packing materials are often completely unsuitable for the analysis of basic com- pounds. The high pH of the mobile phase also allows the use of low applied potentials for the detection of morphine, which results in an improved signal-to-noise ratio. The approach adopted in this work could be used as a basis for the reassessment of methods for the analysis of basic solutes by HPLC with electrochemical detection. There are many other alkaloids that come into this category, plus phenolic biological molecules such as the catecholamines, which normally require the addition of ion-pairing reagents to acidic mobile phases in order to achieve retention.The addition of such reagents is clearly to be avoided, if possible, when electrochemical detection is employed, and there is no doubt that they cause degradation of pump seals. The authors thank S. Wring and M. Norman of Bristol Polytechnic for technical assistance. References 1 Drexel, H., Dzien, A., Spiegel, R. W., Lang, A. H.. Brier, C.. Abbrederis, K., Patsch, J. R., and Braunsteiner, H., Pain, 1989,36, 169. 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Yamaguchi, H., Watanabe, S., Fukuda, T., Takahashi, H., Motokawa, K., and Ishizawa, Y., Anesth. Analg. (New York), 1989,643, 537. Conterno, F., and Pacifico, P., Pharmacol. Res., 1989,21,107. Wasels, R., Belleville, F., Paysant, P., Nabet, P., and Krakow- ski, I., J. Chromatogr. Biomed. Appl., 1989, 489, 411. Moore, R. A., Baldwin, D., McQuay, H. J., and Bullingham, R. E. S., Ann. Clin. Biochem., 1984. 21, 125. Stanski, D. R., Paalzow, L., and Edlund, P. O., J . Pharm. Sci., 1982,71,314. Cashaw, J., Walsh, M. J., Yamakana, Y., and Davis, V. E., J. Chromatogr. Sci., 1971, 9, 98. Bowie, L. J., and Kirkpatrick, P. P., Clin. Chem., 1989, 35, 1355. Bowie, L. J., and Kirkpatrick, P. P., J. Anal. Toxicol., 1989,13, 326. Chen, B. H., Taylor, E. H., and Pappas, A. A., J . Anal. Toxicol., 1990, 14, 12. Done, J. N., and Knox, J. H., Process Biochem., 1972. 7 , 11. Jane, I., and Taylor, J. F., J . Chromatogr., 1975, 109, 37. Nelson, P. E., Nolan, S. L., and Bedford, K. R., J. Chromat- ogr., 1982, 234,407. White, M. W., J. Chromatogr., 1978, 178,229. Wallace, J. E.. Harris, S. C., and Peek, M. W., Anal. Chem., 1980, 52, 1328. Wring, S. A., Hart, J. P., Thompson, J. F., and Birch, B. J., Anal. Proc., 1990, 27,209. Barreira Rodriguez, J. R., Cabal Diaz, V., Costa Garcia, A., and Tuiidn Blanco, P., Analyst, 1990, 115,209. Kissinger, P. T., and Heineman, W. R., J . Chem. Educ., 1983, 60, 702. Hart, J. P., Smyth, M. R., and Smyth, W. F., Analyst, 1981, 106, 146. Laboratory Techniques in Electroanalytical Chemistry, eds. Kissinger, P. T.. and Heineman, W. R., Marcel Dekker, New York, 1984, p. 82. Proksa, B., and Molnar, L., Anal. Chim. Acta, 1978. 97, 149. McLeod, C. W., and West, T. S., Analyst, 1982, 107, 1. Masui. M., Sayo, H., and Tsuda. Y., J. Chem. SOC. B, 1968, 973. Paper 1/01 9286 Received April 24th, 1991 Accepted May 21st, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911600991

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 997 Amperometric Enzyme Electrode for Theophylline Joseph Wang, Eithne Dempsey and Mehmet Ozsoz Department of Chemistry, New Mexico State University, Las Cruces, NM 88003, USA Malcolm R. Smyth School of Chemical Sciences, Dublin City University, Dublin 9, Ireland An amperometric biosensor for theophylline, based on the recently isolated enzyme theophylline oxidase, is described. The enzyme is entrapped, together with a ferricytochrome C cofactor, within a polymeric (Nafion) coating. The anodic detection (at +0.4 V versus Ag-AgCI) is facilitated by the addition of a redox-mediating hexacyanoferrate(iii) ion. The influence of various experimental variables is described. The limit of detection is 2 x 10-6 mol dm-3 theophylline, with linearity prevailing up to 3 x 10-4 mol dm-3.The fast response and wash times permit rapid flow-injection measurements, with a frequency of 180 samples h-1 and a relative standard deviation of 3.040%. Prospects of using this electrode for clinical diagnostics are discussed. Keywords: Theoph ylline; biosensor; theoph ylline oxidase enzyme electrode; amperometry Theophylline is a widely used bronchodilator drug employed in the management of various asthmatic conditions. Because of its therapeutic significance, theophylline has been the focus of intense bioanalytical efforts. Many methods have therefore been developed for measuring theophylline concentrations, including gas and liquid chromatography, ultraviolet spec- trometry, immunoassays and pulse voltammetry.1-5 An enzyme electrode offers many prospects for clinical diagnosis, particularly where self-testing or in vivo monitoring is concerned, and also for high-speed automated analysis.An indirect electrochemical assay for theophylline , based on its inhibitory action on alkaline phosphatase, has been reported recently .6 This paper describes the use of theophylline oxidase for direct amperome tric biosensing of theophylline. The recent isolation of theophylline oxidase (ThO)7 has opened the way to enzymic methods for theophylline. This enzyme oxidizes theophylline in the presence of ferricytochrome C as the electron acceptor: Theophylline oxidase Theophylline + ferricytochrome C - 173-dimethyluric acid + ferrocytochrome C (1) Gupta et al.8 have described an optical enzymic approach for theophylline based on monitoring the ferrocytochrome C produced at 550 nm.The amperometric assay, described in the present paper, couples the enzymic reaction [of eqn. (l)] with the use of a soluble hexacyanoferrate(Ii1) mediator that reacts with the ferrocytochrome C generated: Fe(CN)b3- + ferrocytochrome C --., Fe(CN)& + ferricytochrome C (2) The hexacyanoferrate(i1) ion thus produced is then oxidized at +0.4 V (versus Ag-AgC1) to yield a sensitive and selective anodic detection of theophylline. Simultaneous regeneration of the cofactor is also obtained. The results of the optimization and characterization of a theophylline electrode are reported here. Experimental Apwratus A 10 ml electrochemical cell [Model VC-2, Bioanalytical Systems (BAS)] was used in the batch experiments.The cell was joined to the working electrode, reference electrode (Ag-AgC1, Model RE-1, BAS) and platinum wire auxiliary electrode through holes in its poly(tetrafluoroethy1ene) (PTFE) cover. The three electrodes were connected to an EG & G Princeton Applied Research Model 264 A polarographic analyser, the output of which was displayed on a strip-chart recorder (Model 4500 Microscribe, The Recorder Co.). The flow-injection system consisted of a carrier reservoir, a Rainin Model 5041 sample injection valve (20 p1 loop), interconnect- ing tubing and a platinum thin-layer detector (Model TL-lOA, BAS). Electrode Preparation Prior to coating, the platinum disc (Model 2013, BAS, 1 mm diameter) was polished with a 0.05 pm alumina slurry and then sonicated in water for 2 min.Modification of the platinum electrode was achieved by covering the surface with two 5 pl aliquots of the mixed polymer-enzyme-cofactor solution (1% Nafion-8.3 U ml-1 of Tho-30 pmol dm-3 femcytochrome C). The first layer was dried at 35 "C (using a heat gun) before the second aliquot was applied. The resulting film was subsequently covered with a thin layer of a Nafion solution [5 pl of 1% Nafion (diluted with ethanol)]. Reagents All solutions were prepared with distilled, de-ionized water. Experiments were conducted in 0.05 rnol dm-3 phosphate buffer (pH 6.9). Theophylline oxidase [20 U ml-1 (1 U = 16.67 nkat)] and cytochrome C (70 pmol dm-3) were obtained from DGS Diagnostics. Stock solutions of theophylline (Sigma) were dissolved in 0.4 mol dm-3 HCl and appropriate dilutions made in phosphate buffer.Potassium hexacyanofer- rate(rrr) (Fisher), ascorbic acid, uric acid, acetaminophen, glucose, caffeine, L-a-phosphatidylcholine (Type XI-E from egg yolk) and cholesterol were obtained from Sigma. The 5% Nafion solution was obtained from DuPont. Procedure Batch and flow experiments were performed at room tempera- ture by applying a potential of +0.4 V and allowing transient currents to decay. For batch measurements, the electrolyte solution [containing 1 mmol dm-3 potassium hexacyanofer- rate(~ii)] was stirred at 400 rev min-1. Theophylline additions were made and the current-time data recorded. In flow injection, the sample and carrier phosphate buffer solutions contained 1 mmol dm-3 hexacyanoferrate(n1); a flow rate of 2.0 ml min-1 was employed.Results and Discussion The amperometric and chronoamperometric responses of the unmodified and the Tho-modified electrodes to successive998 ANALYST, OCTOBER 1991, VOL. 116 0.4 $0.3 0, u 5 0.2 0.1 0 0.2 0.4 0.6 Concentration/mmol dm ~- 3 A - Time Fig. 1 Typical current-time recording for successive 5 x 10-5 rnol dm-3 increments of theophylline concentration obtained at A, the unmodified and B , the Tho-containing electrodes. Applied potential, +0.4 V; solution stirring rate, 400 rev min-I; electrolyte, 0.05 mol dm-3 phosphate buffer (pH 6.9), containing 1 mmol dm-3 K3Fe(CN),. Also shown (inset) is the resulting calibration plot for the modified electrode 0.1 yA I E D C B A 0.1 yA I B-E .-d A - I I - Time Fig.2 Chronoamperometric response to 5 x rnol dm-3 increments in theophylline concentration (B-E), together with the blank response (A), as obtained at (a) Tho-containing and (b) unmodified electrodes. The potential was stepped to +0.5 V standard additions of theophylline, each addition effecting a 5 x 10 rnol dm-3 increase in concentration, are compared in Figs. 1 and 2. In the absence of biocatalytic activity, the unmodified electrode is not responsive to the addition of theophylline. In contrast, the enzyme electrode responds rapidly to micromolar changes in the substrate concentration. A steady-state amperometric response is produced within 3 min. The low noise level allows convenient quantification of low theophylline concentrations. Both techniques yield cur- rent signals proportional to the substrate concentration.With d.c. amperometry, linearity prevails up to 3.5 x 10-4 rnol dm-3 (see inset, Fig. l), with a sensitivity (slope of the linear portion) of 0.807 pA dm3 mmol-1. The maximum therapeutic level of theophylline, encountered in cases of toxicity, 3.3 x 10-4 rnol dm-3 (reference 6), therefore falls within the linear portion. Coated electrodes, containing no enzyme (only the cofactor within the Nafion film), yielded no response to theophylline (not shown). The senshhity 07 the theophyhe electrode is affected by various preparation and operational conditions. As expected, the response increases rapidly with an increase in the enzyme surface loading [between 25 and 60 mU (Fig. 3, curve A)]; a levelling-off is observed for higher loadings.The solution pH also has a profound effect on the sensitivity (Fig. 3, curve B). The current increases with the pH over the range 5.4-6.9, with a gradual decrease at higher values. Fig. 4, curve A, shows the 0.2 5- 2 g 0.1 3 0 0 PH 5 6 7 8 9 I I 1 I I I 20 40 60 80 100 Enzyme activity/mU Fig. 3 Dependence of the current on A, the enzyme loading and B, the solution pH. Response to 5 x rnol dm-3 theophylline. Other conditions as in Fig. 1 PotentialN 0.2 0.4 0.6 0.8 -3. . CI 2 5 0.1 1 4 I I I I ( 0 0 1 2 3 4 5 Concentration/mmol dm-3 Fig. 4 Dependence of the current on A, the K3Fe(CN), concentra- tion and B, the applied potential. Response to A, 5 x and B, 5 x 10-5 mol dm-3 theophylline. Other conditions as in Fig. 1 dependence of the theophylline response on the concentration of the redox-mediating hexacyanoferrate(Ir1) ion.The response increases rapidly with the hexacyanoferrate(iI1) concentration up to 1 X 10-3 rnol dm-3, after which it starts to level off. The small response observed in the absence of hexacyanoferrate(ir1) is attributed to the direct oxidation of the 1,3-dimethyluric acid product [see eqn. (l)]. The signifi- cantly higher sensitivity observed in the presence of hexacyan- oferrate(n1) is therefore attributed to the regeneration of the cofactor. As expected when monitoring the oxidation of the hexacyanoferrate( 11) product, the response increases rapidly with the operating potential between 0.2 and 0.4 V (Fig. 4, curve B). A gradual decrease in the response, observed at higher potential values, is attributed in part to electrochemical ‘consumption’ of the substrate.Although at the optimum potential for biosensing operation (+0.4 V) no direct oxida- tion of theophylline is observed ( e . g . , Fig. 1, curve A); such a reaction can occur at higher potential values. The T h o electrode, operated under the optimum con- ditions described above, permits quantification of low concen- trations, as is desired for clinical theophylline monitoring. Successive standard additions of 5 x 10-6 rnol dm-3 theophyl- line were used to calculate the detection limit (Fig. 5 ) . The signal-to-noise characteristics (S/N ratio = 3) indicated a detection limit of 2 x 10-6 mol dm-3. Although this value is for a synthetic (and not a biological) solution, it is well below the normal therapeutic range for theophylline [5.5 x 10-5-1.1 x 10-4 rnol dm-3 (reference 6)].The fast response of the Tho-containing electrode makes it suitable for use in dynamic flow systems. Flow-injection response peaks for a series of 26 repetitive injections of aANALYST, OCTOBER 1991, VOL. 116 999 40 t $ 1 2 J If - Time Fig. 5 Current-time recording for successive 5 x mol dm-3 increments of theophylline concentration. Other conditions as in Fig. 1 t 4- - Time Fig. 6 rnol dm-3 theophylline solution at (a) the bare and (b) the Tho-containing electrodes. Flow rate, 2.0 ml min-1. Applied potential, +0.4 V. Electrolyte and carrier, 0.05 rnol dm-3 phosphate buffer (pH 6.9), containing 1 mmol dm-3 hexacyanoferrate(iI1) Detection peaks for repetitive injections of a 5 X 5 x 10-4 rnol dm-3 theophylline solution at the bare and T h o detectors are compared in Fig.6 . As expected, no response is observed with the naked surface. In contrast, the coated electrode exhibits well-defined and sharp peaks. A repro- ducible response is observed over this prolonged series [relative standard deviation (RSD) = 3%, mean = 19 nA, range = 18-20 nA], indicating the stability of the enzyme layer under vigorous hydrodynamic conditions. The fast response and wash times result in peak widths of 20 s and an injection rate of 180 samples h-1. The peaks shown in Fig. 6(6) were part of a long run of 90 successive injections for which an RSD of 4% was obtained. The flow-injection response is linear for samples containing up to 1 X 10-3 rnol dm-3 theophylline (Fig.7). A deviation from linearity is observed at higher concentrations. The detection limit (of the flow-injection operation) is 1 X 10-5 rnol dm-3 theophylline; the sensi- tivity (slope of the linear portion) corresponds to 32 nA dm3 mmol-1. Interference studies were conducted in the presence of several relevant compounds. Whereas 1 x 10-4 rnol dm-3 glucose, acetaminophen and caffeine exhibited a negligible effect on the response for 2.5 x 10-4 rnol dm-3 theophylline, a large interference was observed in the presence of 1 x 10-4 rnol dm-3 uric and ascorbic acids (not shown; batch operation as in Fig. 1). Apparently, the Nafion film loses its permselec- tive (charge exclusion) properties in the presence of the immobilized T h o and ferricytochrome C.However, it was possible to address the interference of uric and ascorbic acids 0 1 2 3 4 Concent rat ion/m mol dm - 3 Fig. 7 Calibration plot for theophylline obtained in flow injection. Conditions as in Fig. 6 by placing a lipid layer (20 p1 of 20 mg of phosphatidylcholine- 14 mg of cholesterol per millilitre of chloroform) on top of the Nafion-Tho film. Similar selectivity improvements have been reported for a lipid modified glucose sensor.9 The T h o electrode had an analytically useful response for up to 35 d (with intermittent usage and storage at 4 "C in phosphate buffer). The sensitivity was observed to decrease slowly during this period (about 15 and 30% within 10 and 20 d, respectively). Because of the rapid fabrication and low cost of the T h o electrode, the reasons for its slow deactivation were not elucidated.In conclusion, the feasibility of using T h o for amperometric sensing of theophylline has been demonstrated. Owing to the significance of theophylline in the clinical arena, additional work is required to improve the performance (particularly the stability and selectivity) of the sensor and to assess thoroughly its suitability for analyses of relevant biological fluids. In particular, detailed studies with serum, plasma and blood samples are essential before the full diagnostic value of this sensor can be established. Further improvements could be achieved by co-immobilization of the hexacyanoferrate(Ir1) mediator to obtain a reagentless device. When coupled with the fast and sensitive response, these and other improvements and studies should lead to reliable and rapid testing for theophylline based on disposable strips or automated flow analysis. In vivo monitoring of the drug could also be envisaged (following additional improvements and miniatur- ization). E. D. gratefully acknowledges a scholarship from the Irish Department of Science and Technology. Helpful discussions with Ulla Wollenberger are also gratefully acknowledged. References Least, C. J., Johnson, G. F., and Soloman, H. M., Clin. Chem., 1976, 22, 765. Adams, R. F., Vandemark, F. L., and Schmidt, G. J., Clin. Chem., 1976, 22, 1903. Jatlow, P., Clin. Chem., 1975, 21, 1518. Rubenstein, K. E., Schneider, R. S., and Ullman, E. F., .Biochem. Biophys. Res. Commun., 1972,47,846. Munson, J. W., and Adbine, H., Talanta, 1978,254 221. Foulds, N. C., Wilshere, J. M., and Green, M. J., Anal. Chim. Acta, 1990, 229, 57. GDS Enzymatic Theophylline Reagents, GDS Diagnostics, Elkhart, IN, 1988. Gupta, S. K., Agarwal, A. K., and de Castro, A. F., Clin. Chem., 1988,34, 1267. Amine, A., Kauffmann, J.-M., Patriarche, G. J., and Guil- bault. G. G., Anal. Lett., 1989, 22, 2403. Paper 1 I01 327K Received March 19th, 1991 Accepted April 30th, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911600997

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 1001 Determination of Ethaverine and Papaverine Using Ion-selective Electrodes Christian Eppelsheim, Ralph Aubeck, Norbert Hampp* and Christoph Brauchle lnstitut fur Ph ysikalische Chemie der Universitat Miinchen, Sophienstrage I I , 8000 Munchen 2, Germany Ion-selective poly(viny1 chloride) membrane electrodes for the opium alkaloids papaverine and ethaverine are presented. The electrode membranes contain ion pairs of the alkaloids with the anionic counter ion tetraphenylborate. The detection limits for all electrodes were approximately 2 x 10-6 mol dm-3 at pH 5.0 in 100 mmol dm-3 buffered solutions and the measured slopes were close to the values theoretically expected. The selectivity coefficient observed for the ethaverine-tetraphenyl borate electrode is 10-1.1 with respect to papaverine.The suitability of the membranes for single-use electrodes is discussed. Keywords: /on-selective electrode; ethaverine; papaverine Ion-selective electrodes (ISEs) have become important and reliable devices for chemical and medical analysis, which are inexpensive, easy to use and have a wide field of application. In general ISEs can be used for the quantitative determination of inorganic1 and organic2 ions. One of the many existing principles for the construction of ion-selective membranes is the addition of a lipophilic ion-pair complex into a highly plasticized polymer membrane. Most of the ISEs, sensitive to medically important ionic compounds such as psychopharma- cological drugs,3 alkaloids4 and muscle relaxants ,5 belong to the class of ion-pair based liquid-membrane electrodes.An ISE based on an ion-pair composed of papaverine (PAPA) and aurin tricarboxylic acid for the determination of the opium alkaloid PAPA (Fig. 1) has already been described.6 This drug is used as a spasmolyticum and promoter of blood circulation. The chemically very similar alkaloid ethaverine (ETHA) (Fig. 1) is used medically as an antiarrhythmicum. Several methods for the determination of this drug in vitro have been reported, e.g., fluorimetry,7 thin-layer chromatography,* high-performance liquid chromatography (HPLC)9 and spec- trophotometry. 10 Most of these methods are either time- consuming or require expensive instrumentation, however, some of them have excellent detection limits.The present paper reports on the use of ISEs with poly(viny1 chloride) (PVC) matrices for the determination of the opium alkaloids PAPA and ETHA. Ion pairs of the alkaloids with the anionic counter ion tetraphenylborate (TPB) were used as electroac- tive membrane compounds. The electrodes presented are flexible and inexpensive analytical devices for the determina- tion of the quality and uniformity of pharmaceutical prepara- tions, e . g . , in routine analysis and on-line control, where the observed detection limits at the 2 x 10-6 mol dm-3 level are acceptable and are suitable for the determination of the physiological activity of ETHA and PAPA. Experimental Reagents and Materials The precipitating reagent sodium TPB (Fluka) was of analy- tical-reagent grade.Papaverine hydrochloride (Roth), ETHA hydrochloride (Sigma), noscapine hydrochloride (NOSC, Sigma), laudanosoline hydrobromide trihydrate (LAUD, Aldrich) and atropine sulphate (ATRO, Roth) were used. The membranes were prepared from tetrahydrofuran (THF, Fluka) solutions of PVC (Fluka) of high relative molecular mass and dibutyl phthalate (DBP, Merck). Buffer salts (sodium phosphates) were of at least analytical-reagent grade. All solutions were prepared with doubly distilled water. * To whom correspondence should be addressed. Preparation of the Electroactive Compounds The ion-pair complexes were prepared by mixing stoichio- metric amounts of a 0.01 mol dm-3 aqueous solution of sodium TPB with an equimolar aqueous solution of PAPA hydrochloride and ETHA hydrochloride, respectively.The white precipitates obtained were sedimented by centrifugation at 7000g, extensively washed with doubly distilled water and dried over phosphorus pentoxide for 3 d in an evacuated desiccator. Characterization of the Ion-pair Complexes The predicted compositions of the complexes having molar ratios of 1 : 1 for PAPA-TPB and ETHA-TPB were ascer- tained by elemental analysis with deviations from the theoret- ically expected values of less than 0.4%. The melting-points observed are given in Table 1. Preparation of the Ion-selective Membranes The THF solutions (1 .0 ml) containing various amounts of the ion-pair complexes (see Table 1) together with 62 mg of PVC and 140 pl of DBP were poured into small polypropylene rings (diameter 23 mm) lying on glass cover slips.The rings were covered by a sheet of filter-paper and a second cover slip to obtain slow evaporation of the solvent over a period of 12 h at room temperature. Preparation of the Electrodes The electrodes were prepared by gluing the cut out PVC membrane discs onto the cleaned ends of PVC modules by means of a PVC-THF solution (62 mg of PVC in 1.0 ml of THF). The modules are made from PVC tubes and have a length of 2 cm and a thread fitting to the electrode bodies. The complete electrodes were assembled by screwing the mem- brane modules onto the electrode bodies, which contained an PAPA ETHA Structures of the alkaloids PAPA and ETHA Fig. 11002 ANALYST, OCTOBER 1991, VOL. 116 Table 1 Characterization of the ion pairs Relative Molar molecular Melting- Content Ion pair ratio mass pointJ"C (YO m/m) PAPA-TPB 1:l 659.6 168-170 1.0,2.5,5.0 ETHA-TPB 1 : l 715.8 164-166 1.0,2.5* * Maximum content that could be achieved was 2.5% owing to the limited solubility of ETHA-TPB in the PVC-DBP membrane.Table 2 Characterization of the PAPA and ETHA membranes. Measurements made in 0.1 mol dm-3 sodium phosphate buffer at pH 5 .O Detection limit Ion pair Slope/mV Linear range/ Electrode (YO m/m) decade-' mol dm-3 pg ml-l mol dm-3 PAPA-TPB 1.0 52+3* 2 X 10-6 0.8 10-5-10-3 2.5 57*3* 5 x 10-6 1.9 10-5-10-3 5.0 60*3* 1 x 10-5 4.9 10-4-10-2 2.5 65*2* 2~ 10-6 0.9 10-5-10-3 ETHA-TPB 1.0 61?2* 2X 10-6 0.9 10-5-10-3 * Statistical deviation for ten electrodes. Ag-AgCI internal leadout (Ingold). As an internal filling 3.0 rnol dm-3 sodium chloride (saturated with AgCI) was used.The electrodes were stored in glass tubes in 0.1 mol dm-3 sodium phosphate buffer (pH 5.0) at room temperature. The assembled membrane modules were stored dry at room temperature before use. E. m. f. Measurements Potentiometric measurements were made versus a double junction Ag-AgC1 reference electrode (Type K 801, Radio- meter) with saturated KCI solution in the outer compartment. The electrode potentials were recorded with a digital micro- processor pH/ion-meter (pMX 2000, WTW Weilheim) and plotted on a strip-chart recorder. All measurements were made at room temperature with continuous stirring of the solutions (25 ml) at a constant rate by using a magnetic stirrer. Small volumes of concentrated solutions of the different compounds were added in order to obtain the desired final concentrations.The potentiometric selectivity coefficients kQ' were determined by the separate solution technique" and calculated from the equation log kflp' = (Ej - EJ/S where E represents the e.m.f. readings measured for the primary ion (i) and the interfering ion (j), respectively, and S is the slope observed for the primary ion. The values for kf'?' were calculated from the e.m.f. values measured for 1 x 10-3 rnol dm-3 solutions of the primary and the interfering ions in 0.1 mol dm-3 sodium phosphate buffer at pH 5.0. This pH value was chosen in order to prevent flocculation of the free bases of PAPA and ETHA. Results and Discussion Calibration Data The response characteristics of the investigated PAPA and ETHA electrodes are summarized in Table 2.The slopes exhibited close to Nernstian behaviour. A slight dependence on the ion-pair content was observed for both types of electrodes. Increasing concentrations of the ion pair were correlated to an increase in the slopes. In general the slopes of the ETHA-TPB membranes were somewhat higher com- pared with the PAPA-TPB electrodes. The reason for the increasing slopes is possibly the influence of the ion-pair concentration on the internal diffusion potential of the membrane.12 The detection limits observed for the PAPA- TPB and ETHA-TPB electrodes, which were determined Table 3 Logarithmic selectivity coefficients of the PAPA-TPB and ETHA-TPB electrodes. The ion-pair content of the electrodes is 2.5% m/m Interfering ion Log kFP' ETHA PAPA NOSC ATRO LAUD PAPA-TPB + 1 .O 0 -0.9 < - 2 .0 -3.0 ETHA-TPB 0 -1.1 -1.9 -2.5 <-3.0 * For monovalent (K+, Na+) and divalent (Ca2+, Mg2+) cations values of log kfq' below -3.0 were found. according to the IUPAC recommendations,13 were found to be in the range 2 x 10-6 rnol dm-3 (0.8 pg ml-l)-l x 10-5 rnol dm-3 (4.9 pg ml-1) (Table 2). The detection limits of the ETHA-TPB electrodes were independent of the content of the ion pair. Selectivity of the Electrodes The selectivity coefficients of the PAPA-TPB and the ETHA-TPB electrodes determined against NOSC, LAUD and ATRO are given in Table 3. The order of selectivity for both types of electrodes is ETHA > PAPA > NOSC > ATRO > LAUD. The EHTA-TPB electrodes showed good discrimination of the tested alkaloids.For the selectivity coefficient of the ETHA electrode versus PAPA a log kfj" value of -1.1 was found. This is remarkable as the only chemical difference between ETHA and PAPA is the replacement of the ethoxy groups by methoxy groups (see Fig. 1). The response of the PAPA-TPB electrode to ETHA was about ten times more sensitive than towards PAPA, as indicated by the positive log k!?' value of 1 .O. The slope of this electrode was drastically reduced after single contact with a 1 X 10-3 rnol dm-3 solution of ETHA hydrochloride and could not be recovered by conditioning in buffer or PAPA solutions within 48 h. Therefore, an exchange of PAPA versus ETHA in the ion pair at the membrane surface was assumed. Response Times The response times (90% final signal) of all electrodes were less than 25 s in the range 10-5-10-3 rnol dm-3 and about 35 s at lower concentrations.The final e.m.f. values were attained within 80 s. These response times are typical of ion-selective electrodes which contain bulky PVC membranes of similar thickness (about 0.2 mm). Comparison of the Characteristics of Several Membranes Ten freshly prepared PAPA-TPB electrodes (2.5% m/m PAPA-TPB) were measured in series after conditioning in buffer solution (sodium phosphate, 0.1 rnol dm-3) of pH 5.0 for 30 min. The electrode response to PAPA hydrochloride was determined with decadic concentration steps. The mean value of the derived slopes was 57 mV with an individual deviation of +3 mV. The detection limits of all the electrodes were equal and did not depend on the individual deviations of each membrane.This indicates that membranes of this type are suitable for single-use electrodes without any calibration but with a 30 min conditioning in buffer solution. If a precision better than +3.0% is needed the electrodes must be calibrated individually. Conclusions The electrodes for the opium alkaloids PAPA and ETHA may be successfully applied to the potentiometric determination of these drugs in applications where a detection limit of 2 x 10-6 mol dm-3 is sufficient. Compared with other analyticalANALYST, OCTOBER 1991, VOL. 116 1003 methods such as gas chromatography, HPLC, spectropho- tometry and atomic absorption spectrometry (AAS), which require expensive instrumentation and adequate sample preparation, the proposed electrodes are inexpensive and easy to use.This is important for single-use electrodes, for which the proposed membranes can be used, but they require conditioning in buffer for 30 min. The electrodes are suitable for on-line monitoring of drug concentrations in the manufac- ture of pharmaceuticals. The detection limits in the range 2 x 10-6 mol dm-3 (0.8 pg ml-l)-l x 10-5 mol dm-3 (4.9 pg ml-1) observed for the PAPA-TPB and ETHA-TPB electrodes were poorer than those obtained by conventional analytical methods such as HPLC (2 ng ml-1)14 and indirect determina- tion by AAS (19 ng ml--1)15 but comparable to those obtained by spectrophotometry (4 pg ml-1).16 References Solsky, R., Anal. Chem., 1990, 62,21R.Ma, T. S., and Hassan, S. S. M., Organic Analysis Using Ion-selective Electrodes, Academic Press, London, 1982. Takisawa, N., Denver, G., Evan, W.-J., and Brown, P., J. Chem. SOC., Faraday Trans. 1 , 1988,84,3059. Aubeck, R., Hampp, N., and Brauchle, C., Ber. Bunsen Ges. Phys. Chem., 1988, 92, 1423. 5 6 7 8 9 10 11 12 13 14 15 16 Aubeck, R., Brauchle, C., and Hampp, N., Anal. Chim. Actu, 1990,238,405. Zarechenskii, M., and Gaidukovich, A., Otkrytiya, Izobrer., 1984,42, 130; Chem. Abstr., 1985,102, 197293b. Wullen, H., Stainier, E., and Luyckx, M., J. Pharm. Belg., 1966, 21, 409. Daldrup, T., Busante, F., and Michalke, P., Fresenius 2. Anal. Chem., 1981,308,413. De Smet, M., Peeters, A., Buydens, L., and Massart, D., J. Chromatogr., 1988,457, 25. Szendey, G., Arch. Pharm. Ber. Dtsch. Pharm. Ges., 1968,301, 46. Pungor, E., Toth, K., and Hrabeczy-Pall, A., Pure Appl. Chem., 1979, 51, 1913. Morf, W., The Principles of Ion-Selective Electrodes and of Membrane Transport, Elsevier, Amsterdam, 1981, p. 219. Guilbault, G. G., Ion Sel. Electrode Rev., 1979, 1, 139. Brodie, R., Chasseaud, L., Walmsley, L., Soegtrop, H., Darragh, A., and O’Kelly, D., J. Chromatogr., 1980,182, 379. Nerin, C., Garnica, A., and Cacho, J., Anal. Chem., 1986,58, 2617. Sakai, T., and Ohno, N., Chem. Pharm. Bull., 1979,27,2846. Paper 1f021.566 Received May 8th, 1991 Accepted June 12th, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911601001

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 1005 Nitrate-selective Electrodes Containing Immobilized Ion Exchangers Within a Rubbery Membrane With Controlled Cross-link Density Les Ebdon and Jim Braven Plymouth Analytical Chemistry Research Unit, Department of Environmental Sciences, Polytechnic South West, Drake Circus, Plymouth PL4 8AA, UK Nicholas C. Frampton West Pharmarubber Ltd., Holmbush Industrial Estate, St. Austell, Cornwall PL25 3JL, UK A range of ion-selective electrodes have been produced for the determination of nitrate using polymeric membranes containing immobilized quaternary ammonium salts as the ion exchanger. The use of acrylonitrile-butadiene copolymer with a 19% acrylonitrile content as the polymer matrix produced electrodes with excellent @G3-, cI- values of 1.1 x 10-3 but with short lifetimes.Varying the cross-link density of the membranes from 0.6 x 10-5 to 11.0 x 10-5 mol cross-link cm-3 did not produce any significant difference in electrochemical behaviour. Two novel ion exchangers, triallyloctylammonium bromide and triallyldodecylammonium bromide, were successfully synthesized and fabricated into membranes. Triallyl- octylammonium bromide gave the longest lifetime membranes and the best electrode was fabricated from the acrylonitrile-butadiene copolymer with a 50% acrylonitrile content and 6.5% m/m triallyloctylammonium bromide, 7.2% m/m dicumyl peroxide and 39.7% 2-nitrophenyl octyl ether. The electrode had a lifetime in excess of 500 d and the response to nitrate was Nernstian in the range 1 x 10-1-1 x 10-4 mol dm-3 of nitrate.The limit of detection was 4.5 x 10-5 mol dm-3 and the selectivity coefficient for chloride over nitrate (G%3-,cl-) was 5.3 x 10-3. The immobilized ion exchanger membrane electrode offered superior lifetime and mechanical strength. Keywords: Ion-selective electrode; nitrate determination; immobilized ion exchanger; cross-linked acryloni- trile-butadiene copolymer membrane; extended membrane lifetime In a previous paper' the fabrication of nitrate-selective membranes using quaternary ammonium compounds as the added ion exchanger was described. The ion exchanger incorporated into the membrane mixture was either triallyl- ethylammonium bromide (TAEAB) or triallylbutylammo- nium bromide (TABAB). The immobilization reaction in- volved cross-linking a copolymer of acrylonitrile (ACN) and butadiene with an organic peroxide, initiated by heat, which is commonly termed hot pressing or compression moulding.2 Membranes fabricated with TABAB were found to give the best selectivity coefficients over a range of common anions and the inclusion of 2-nitrophenyl octyl ether (2-NPOE) gave lifetimes in excess of 4 months.In the work described here, significant changes to the membrane composition have been made and two new ion exchangers have been synthesized and the effect on both selectivity and lifetime is evaluated. Firstly, the short chain alkyl substituent (ethyl or butyl) on the ion exchanger was increased in chain length to octyl and dodecyl. Secondly, copolymers with lower ACN contents and correspondingly higher butadiene contents were evaluated.Commercially available copolymers containing ACN levels of 19 and 34% were evaluated and compared with the previously reported 50% ACN copolymer. It was anticipated that the higher butadiene content would result in more cross-links and consequently an increased level of immobilized ion exchanger. Thirdly, electrochemical properties were measured for membranes with cross-link densities ranging from 0.6 x 10-5 to 11.0 x 10-5 mol cross-link (3311-3. It was considered that the difference in selectivity coeffi- cients noted between TAEAB and TABAB indicated that the synthesis of longer chain triallylalkylammonium bromides may offer the opportunity to improve the selectivity, mG3-, x- values further. The literature refers only to the successful synthesis of the hexyl derivative3 and it is known that attempts to produce longer chain triallylalkylammonium bromides have been unsuccessful .4 Studies have been conducted5 wherein the amount of organic peroxide added to a rubber compound has shown a good correlation with cross-link density.In this work, the comparison of cross-link density with electrochemical properties of the membrane is described. Experimental Triallylbutylammonium bromide was synthesized as pre- viously described.4 The synthesis of triallyloctylammonium bromide (TAOAB) and triallyldodecylammonium bromide (TADDAB) required slight modifications to the procedure previously reported. Diallyloctylamine was successfully synthesized by the reaction of diallylamine (Aldrich , Gilling- ham, Dorset, UK) with a 2.5 molar excess of bromooctane (Aldrich) .Diallyldodecylamine was synthesized by the equi- molar reaction of diallylamine and dodecyl bromide (Aldrich) with the reflux period extended from 4 to 28 h. After extensive purification, the tertiary amine was reacted with ally1 bromide (Aldrich) to give the quaternary ammonium salt (QAS). The identity of the compounds was confirmed by analysis of both the tertiary amine and the QAS using gas chromatography- mass spectrometry, infrared spectrometry and carbon, hydrogen and nitrogen analysis. Triallyloctylammonium bro- mide formed a solid, however, TADDAB was a thick oily liquid and both were found to be highly hygroscopic and were stored over calcium chloride. Analytical-reagent grade chemicals and AnalaR water [Merck (formerly BDH), Poole, Dorset, UK] were used unless stated otherwise.Tetrahydrofuran (THF) (Merck) was refluxed over potassium metal (Aldrich) and freshly distilled prior to use. The copolymers of acrylonitrile and butadiene, Krynac 19.65, 34.80 and 50.75 (Polysar UK) with 19, 34 and 50% ACN content termed low, medium and high ACN, respectively, were purified by dissolving 15 g in 75 ml of THF and reprecipitating in 200 ml of cold methanol (Merck, SpectrosoL grade). Two solvent mediators, dibutyl phthalate (DBP) (Merck) and 2-NPOE (Fluka, Glossop, Derbyshire, UK) were used as received, as was dicumyl peroxide (DCP) (Merck, laboratory-reagent grade).1006 ANALYST, OCTOBER 1991, VOL. 116 Preparation of Membranes Membranes were fabricated as previously described1 by dissolving the polymer in THF and adding the DCP and QAS with a small volume of methanol together with the solvent mediator if included in the composition.The solvent was removed under vacuum prior to hot pressing between Melanex film (ICI Films Division, Dumfries, UK) at an elevated temperature in a steel mould. A hydraulic press was utilized to press the membranes to a constant thickness. Electrode Evaluation Discs were punched from the pressed membrane and con- ditioned for 7 d in a 0.1 rnol dm-3 potassium nitrate solution to effect replacement of Br- with NO3- anions and confer nitrate selectivity. The membrane was assembled into the tip of a commercially available electrode body, IS 560 (Philips Analytical, Cambridge, UK) and the internal filling solution was a mixture of 0.1 mol dm-3 potassium nitrate and potassium chloride solutions (1 + 1).The e.m.f. measure- ments were made by a digital voltmeter (Model PW 9409, Philips Analytical) and the electrochemical cell was completed by a double junction reference electrode (Model 90-02, Orion Research, Cambridge, MA, USA) with 0.04 mol dm-3 ammonium sulphate as the outer filling solution. Potassium nitrate standards were prepared using AnalaR water with the addition of 1 x 10-2 mol dm-3 potassium hydrogen phosphate solution as an ionic strength adjuster, held at 25 k 0.5 "C and stirred magnetically during the e.m.f. measurement. Selectivity coefficients =A3-, x- were determined for a number of membranes using 1 x 10-1 mol dm-3 potassium chloride and the sodium or potassium salt of a range of other common anions (X-).The 18 mV method with mixed solutions was used to calculate the selectivity coefficient.6 Cross-link Density Determination The cross-link density expressed as rnol cross-links cm-3 was determined at room temperature for a range of membranes based upon the classical work of Flory and Rehner7 and Flory.8 The solvent chosen was ethylmethylketone (Merck) because the solvent-polymer interaction coefficient for copolymers of acrylonitrile and butadiene with varying ACN levels was available from the literature.5 Results and Discussion Low and Medium Acrylonitrile Content The relatively short lifetime reported previously1 with mem- branes fabricated with the 50% ACN polymer was considered to be associated with the loss of the ion exchanger.Mem- branes were fabricated using both low and medium ACN content copolymers with 19 and 34% ACN and the corre- sponding level of butadiene. It was envisaged that the greater level of 'rubbery' butadiene segments with associated unsatu- ration within the polymer would lead to an increased cross-link density and a reduced tendency for water absorp- tion. Both these factors were expected to play an important role in determining the lifetime of membranes. The use of the 19% ACN copolymer and 65% m/m [235 pph of rubber (parts per hundred of rubber)] solvent mediator resulted in membranes with poor physical strength and a marked tendency to either split or tear when in use. Reduction of the solvent mediator to 30% m/m (83.8 pph of rubber) improved the physical strength of the membranes to allow normal handling.The composition of the low and medium ACN copolymer membranes are shown in Tables 1 and 2. The membranes were fabricated using the hot pressing method and, after conditioning in a 0.1 mol dm-3 potassium Table 1 Composition of membranes fabricated with a 19% ACN copolymer. Values in parentheses are in % m/m TAEAB TABAB DCP Membrane (pph of (pph of (pph of Solvent No. rubber)* rubber) rubber) mediator 14 - 11.9 (9.3) 15.5 (12.2) None 15 - 11.9 (5.6) 15.5 (7.3) DBP 17 10.2 (3.1) - 15.5 (12.3) None 18 10.2 (4.9) - 15.5 (7.4) DBP 19 10.2 (4.9) - 15.5 (7.4) 2-NPOE - 15.5 (13.4) None 20 21 - 11.9 (6.6) 15.5 (8.5) DBPt 22 - 11.9 (6.6) 15.5 (8.5) 2-NPOET 15.5 (8.6) DBPt 24 10.2 (5.7) - 15.5 (8.6) 2-NPOEt 16 - 11.9 (5.6) 15.5 (7.3) 2-NPOE - 23 10.2 (5.7) - * pph of rubber = parts per hundred of rubber.t Reduced level of solvent mediator. ~~~~~~ Table 2 Composition of membranes fabricated with a 34% ACN copolymer. Values in parentheses are in % m/m TAEAB TABAB DCP Membrane (pph of (pph of (pph of Solvent No. of rubber)" of rubber) of rubber) mediator 25 - 11.9 (9.3) 15.5 (12.2) None 26 - 11.9 (5.6) 15.5 (7.3) DBP 28 - - 15.5 (13.4) None 29 10.2 (3.1) - 15.5 (12.3) None 30 10.2 (4.9) - 15.5 (7.4) DBP 27 - 11.9 (5.6) 15.5 (7.3) 2-NPOE 31 10.2 (4.9) - 15.5 (7.4) 2-NPOE * pph of rubber = parts per hundred of rubber. Table 3 Electrode response of membranes with 19 and 34% ACN copolymers Membrane No. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Slope/mV decade- 1 -9.6 -58.6 -58.1 -33.0 -57.0 -59.2 -38.2 -59.2 -26.6 -59.0 -33.6 -57.7 -57.5 -8.3 -12.6 -56.3 -56.6 Unresponsive Limit of detection/ rnol dm-3 - 4 x 10-5 2 x 10-5 7.9 x 10-5 6.3 x 10-5 2 x 10-5 2 x 10-5 3 x 10-5 4 x 10-5 4 x 10-5 - - - - - - - 1 x 10-4 G&-.c1- 1.3 x 10-3 1.1 x 10-3 2 x 10-3 1.3 x 10-3 1.9 x 10-3 2.1 x 10-3 3.2 x 10-3 3.2 x 10-3 - - - - - - - - 2 x 10-3 4.2 X 10-3 nitrate solution, were evaluated for response to nitrate ions. The results are summarized in Table 3 together with the selectivity coefficient KEA3-, cI-, calculated using the mixed solutions method with a potassium chloride concentration of 0.1 mol dm-3. The omission of a plasticizer from the membrane mixture for both the low and medium ACN copolymer membranes (membranes 14, 17, 25 and 29) resulted in sub-Nernstian responses in the range 9.6-33 mV decade-'.The reduction of DBP to 30% m/m in membranes 21 and 23 also resulted in sub-Nernstian responses whereas membranes with 65% m/m of DBP gave a Nernstian response. For membranes fabricated using 2-NPOE, there was no effect upon the Nernstian response or KrA3-,cI- value by reduction in the solvent mediator.ANALYST, OCTOBER 1991, VOL. 116 Table 4 Lifetime studies for membranes with 19 and 34% ACN copolymers 1007 Membrane No. Lifetime measurement Original Slope/mV decade-' Limit of detection/mol dm-3 Slope/mV decade-' Limit of detection/mol dm-3 Slope/mV decade- Limit of detection/mol dm-3 SlopdmV decade-1 Limit of detection/mol dm-3 Cross-link density/mol cross-link cm-3 21 Days 58 Days 105 Days 16 -58.1 2 x 10-5 -57.6 4.5 x 10-5 -44.2 1.3 x 10-4 -40.3 4.0 x 3.8 x 10-5 27 -58.8 4 x 10-5 -49.4 1 x 10-3 -47.8 1.6 x 10-4 -46.9 5.0 x 1.8 x 10-5 32 -58.8 5 x 10-5 6.3 x 10-5 3.2 x 10-4 1.0 x 10-3 0.4 x 10-5 -58.7 -47.4 -34.1 38 -58.1 2 x 10-5 -59.1 6.3 x -44.5 2.5 x 10-4 -37.8 7.9 x 10-4 8.7 x 10-5 Table 5 Selectivity data for membranes with 19 and 34% ACN copolymers Selectivity coefficient, @$A3-, x- Anion (X-) F- c1- Br- I- SCN- c104- Mn04- HC03- N02- s o p Concentration/ mol dm-3 1 x 10-2 1 x 10-1 1 x 10-2 1 x 10-2 1 x 10-4 1 x 10-4" 1 x 10-4 1 x 10-6 1 x 10-2 1 x 10-1 Membrane 16 4 x 10-3 1.1 x 10-3 5.4 x 10-2 7 6.3 7.9 12.6 2 7.0 x 10-2 3.2 x 10-3 Membrane 27 3.2 x 10-3 5.0 x 10-2 4 7.9 14.1 0.5 9.0 x 10-2 7 x 10-3 126 1.6 x 10-3 Membrane 32 6.3 x 8.0 x 10-2 5 15.8 15.8 100 10 2.2 x 10-3 7.9 x 10-2 4.0 x 10-3 Membrane 38 6.3 x 10-3 1.0 x 10-3 9.0 x 10-2 7.9 6.3 22.4 40 5 7.9 x 10-2 2.0 x 10-3 * Intercept method.Overall, there is little difference between membranes fabricated with the 19 o r 34% ACN copolymer. However, the fl;A3-, ~ 1 - value achieved with these membranes is superior to that for the 50% ACN copolymer previously reported and the Philips nitrate poly(viny1 chloride) (PVC) membrane electrode with a quoted Kc&3-,C1-, value of 1 x 10-2 and a Kr&-.cl- value measured in our laboratory of 8.0 X 10-3. The best fabricated membrane was membrane 16, which, using 19% ACN copolymer and TABAB as the added ion exchanger with 2-NPOE as the solvent mediator, achieved a flG&3-,cI- value of 1.1 x 10-3. Membranes 16 and 27, fabricated with the 34% ACN copolymer, were evaluated for lifetime characteristics and measured for P$A3-, x- values for a range of common ions with the results shown in Tables 4 and 5.The lifetime data were obtained by measuring the slope and limit of detection6 in potassium nitrate standards containing 1 x 10-2 mol dm-3 KH2P04 which can be added to analyte solutions as an ionic strength adjuster. After less than 2 months, sub-Nernstian responses were recorded for both membranes. The selectivity coefficients were measured using the mixed solutions method with the interferent (X-) present as either the sodium or potassium salt. There is little difference between the two membranes, although 16 is superior when the large Mn04- ion was the interferent. This may be related to the difference in cross-link density for the two membranes, with membrane 16 having a higher cross-link density than membrane 27. It is possible that large anions are restricted from entering the polymer matrix because of the increased number of cross-links.Calibration graphs for membrane 16 are shown in Figs. 1 and 2 for the lifetime studies and selectivity data, respectively. 6 5 4 3 2 1 pN03- Fig. 1 Lifetime data for membrane 16 tested in nitrate standards containing 1 x 10-2mol dm-3, KH2P04. A, Original; B, 21 d; C, 58 d; and D , 105 d Effect of Cross-link Density A range of membranes were fabricated based on membrane 16 but with varying levels of DCP, the free radical initiator for the cross-linking reaction.The cross-link density of these mem- branes and the electrode response was determined, and the level of DCP added is shown in Table 6. No relationship between cross-link density and the electrode response of the membranes was established. Within the range evaluated, (0.61008 ANALYST, OCTOBER 1991, VOL. 116 Table 6 Electrode response of membranes with varying levels of dicumyl peroxide (DCP). Values in parentheses are in % m/m Limit of Cross-link Membrane DCP (pph Slope/mV detection/ densit y/mol No. of rubber)* decade-' mol dm-3 *A3, cI- cross-link ~ r n - ~ 31 32 33 34 35 16 36 37 38 39 5.7 (2.8) 9.4 (4.6) 11.3 (5.4) 13.2 (6.3) 15.5 (7.3) 17.0 (8.0) 18.9 (8.8) 20.8 (9.6) 22.6 (10.4) 7.5 (3.7) * pph of rubber = parts per hundred of rubber.-56.1 -58.8 -59.2 -59.1 -59.1 -58.1 -57.1 -58.7 -58.1 -57.1 1.1 x 10-4 2.2 x 10-3 0.6 x 10-5 5 x 10-5 2.2 x 10-3 0.4 x 10-5 7 x 10-5 1.4 x 10-3 1.7 x 10-5 4.2 x 10-5 2.5 x 10-3 2.5 x 10-5 2.5 x 10-5 2.5 x 10-3 3.0 x 10-5 2.0 x 10-5 1.1 x 10-3 3.8 x 10-5 6.3 x 10-5 2.0 x 10-3 4.6 x 10-5 2.0 x 10-5 1.0 x 10-3 8.7 x 10-5 7.1 x 10-5 1.3 x 10-3 11.0 x 10-5 6.6 X 10-5 2.0 x 10-3 6.2 x 10-5 Table 7 Carbon, hydrogen and nitrogen results for the new ion exchangers Expected (%) Found (% ) Compound C H N C H N Diallyloctylamine 80.38 12.92 6.70 80.42 13.23 7.02 TAOAB 61.82 9.70 4.24 61.61 10.06 4.45 Diallyldodecylamine 81.51 13.21 5.28 81.45 13.49 5.60 TADDAB 65.3 10.40 3.60 64.01 10.62 3.85 r 6 5 4 3 2 1 pN03- Fig. 2 Calibration and selectivity data for membrane 16 tested in nitrate standards containing: A, 1 X rnol dm-3 KHzPO,; B, 1 x 10-2 rnol dm-3 F-; C, 1 x 10-1 rnol dm-3 C1-; D, 1 x 10-4 rnol dm-3 I-; and E, 1 x 10-2 mol dm-3 Br- X 10-5-ll.0 X mol cross-links cm-3) a change in the cross-link density of the membrane did not affect the Nernstian response, selectivity coefficient, cl-, or limit of detection.By plotting the cross-link density against the amount of DCP added to the membrane, a relationship was shown to exist between the two properties. The use of the 19% ACN copolymer resulted, as predicted, in an increase in the cross-link density compared with the same membrane fabricated with the 50% ACN copolymer.' This is confirmed by comparison of the cross-link density values for the 50% ACN (1.7 x 10-5 mol cross-links (3111-3) with the 19% ACN copolymer membrane (3.8 X 10-5 rnol cross-links cm-3).The higher level of rubbery butadiene segments in the 19% ACN copolymer are expected to account for the difference in-cross-link density. If the increase in cross-linking had resulted in a higher level of immobilized ion exchanger, an extended lifetime might have been anticipated. Lifetime studies were completed for membranes 32 and 38 to represent high and low cross-link density values. Results are shown in Table 4 and confirm the results for membrane 16, with sub-Nernstian responses occurring within 2 months. This suggests that the increased number of cross-links achieved with membrane 38 has not resulted in increased levels of immobilized ion exchanger. It is postulated therefore that the improvement in cross-link density has been achieved by butadiene-butadiene cross-links and not by butadiene-ion exchanger-butadiene cross-links. Interestingly, there was only a marginal difference between membranes 32 and 38 and, therefore, the presence of cross-links themselves, within the limits studied, do not affect the electrochemical properties of the membrane.For a response mechanism relying on mobile ion exchanger groups, such an observation could not have been anticipated as the increasing level of cross-links would have been expected to reduce mobility and, therefore, affect the response. The selectizty data, shown in Table 5 , confirm that the level of cross-links within the membrane have not dramatically affected the KrA3-,x- values, although, when Mn04- was the interferent, the membrane having the highest cross-link density was more selective.The selectivity coefficient for the smallest anion evaluated, F-, demonstrated a significant difference between the two membranes. Reducing the level of cross-links within the membrane increased the level of interference from F-, which may have resulted from the relative ease with which the small anion could penetrate the surface of the low cross-link density membrane. Longer Alkyl Chain QAS Ion Exchangers Two longer chain triallylalkylammonium bromides were synthesized by modification of the previously reported method.4 In the past, difficulties were experienced in syn- thesizing derivatives with an alkyl chain longer than hexyl.374 The carbon, hydrogen and nitrogen results in Table 7 for the two tertiary amines, diallyloctylamine and diallyldodecylam- ine, and the QASs, TAOAB and TADDAB, confirm their successful synthesis. Membranes were fabricated as before and hot pressed using the standard method; the composition of the membranes are shown in Table 8.The TADDAB was very difficult to handle being both an oily liquid and highly hygroscopic and gave rise to membranes which were distinctly greasy on the surface, which may be owing to the migration of the non-immobilized ion exchanger to the surface. Electrochemical responses for the membranes fabricated with TAOAB and TADDAB are shown in Table 9. The use of TAOAB as the added ion exchanger produced membranes that gave a Nernstian response and the P$A3-,cI- values obtained indicate that decreasing the ACN content of the copolymer produces the best selectivity coefficient.In all instances, the GA3-, cI-, value was superior to the value for commercially available PVC membranes. In Table 10, the lifetime results for TAOAB membranes are shown and theseANALYST, OCTOBER 1991, VOL. 116 1009 ~~ ~~~~ - Table 8 Composition of membranes with TAOAB and TADDAB. Values in parentheses are in % m/m TAOAB No. rubber)* Membrane (PPh of 40 14.0 (10.8) 41 42 14.0 (12.3) 43 14.0 (6.5) 44 14.0 (6.5) 45 14.0 (6.5) 46 14.0 (6.5) 47 14.0 (6.7) 48 14.0 (6.4) 49 50 - 51 52 53 54 55 56 - - - - - - - - * pph of rubber = parts per hundred of rubber. TADDAB rubber) (PPh of - - - - - - - - - 16.4 (12.4) 16.4 (14.1) 16.4 (7.5) 16.4 (7.5) 16.4 (7.5) 16.4 (7.5) 16.4 (7.9) 16.4 (7.4) DCP rubber) 15.5 (12.0) 15.5 (13.4) 15.5 (7.2) 15.5 (7.2) 15.5 (7.2) 15.5 (7.2) 7.5 (3.6) 20.8 (9.4) 15.5 (11.7) 15.5 (7.1) 15.5 (7.1) 15.5 (7.1) 15.5 (7.1) 7.5 (3.6) 20.8 (9.3) (PPh of - - Solvent mediator None None None DBP 2-NPOE 2-NPOE 2-NPOE 2-NPOE 2-NPOE None None DBP 2-NPOE 2-NPOE 2-NPOE 2-NPOE 2-NPOE ACN content of copolymer 50 50 50 50 50 19 34 19 19 50 50 50 50 19 34 19 19 ("/.I Table 9 Electrode responses for membranes fabricated with TAOAB and TADDAB Membrane No.40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 SlopeImV decade- 55.3 58.4 58.3 58.6 58.6 58.8 59.2 58.6 57.6 55.1 57.8 59.1 56.2 51.4 57.3 55.0 - Limit of detection/ mol dm-3 2.5 x 10-5 2.5 x 10-5 4.5 x 10-5 5.0 x 10-5 2.5 x 10-5 3.5 x 10-5 3.5 x 10-5 2.0 x 10-5 2.0 x 10-5 3.2 x 10-5 4.5 x 10-5 8.9 x 10-5 4.5 x 10-5 7.1 x 10-5 4.5 x 10-5 6.3 x 10-5 - K!s3-, c1- 2.2 x 10-3 5.9 x 10-3 5.3 x 10-3 6.3 x 10-3 2.1 x 10-3 2.1 x 10-3 1.8 x 10-3 1.8 x 10-3 7.1 x 10-3 6.3 x 10-3 8.9 x 10-3 6.3 x 10-3 5.6 x 10-3 3.2 x 10-3 3.5 x 10-3 - 1.0 x 10-2 confirm the extended lifetime predicted for membranes containing immobilized ion exchangers.The use of extended alkyl chains has played a large part in improving membrane lifetime and at this stage the mechanism, whether it be immobilized ion exchanger, reduced mobility or water solubil- ity of the ion exchanger, has not been confirmed. Membrane 43 is clearly superior from a lifetime point of view, with a Nernstian response after more than 500d and a limit of detection of 6.3 X 10-5 mol dm-3 of nitrate.With the extended lifetime achieved by the use of TAOAB, differences between membranes 47 and 48 have become apparent , where the cross-link density has been successfully modified by the amount of DCP added to the membrane. In this instance the higher DCP level has resulted in an extended lifetime for the membrane and could offer an opportunity to extend the membrane lifetime further by optimizing the level of cross-linking agent in the membrane composition. The ACN content of the copolymer can be seen to be a critical feature in determining the membrane lifetime. The membrane with the highest ACN content gave the best results. This supports earlier comments on the likely cross-linking reactions occurring when the level of butadiene is increased.Membranes fabricated using TADDAB, despite being greasy to the touch and the obvious handling restrictions of the TADDAB, gave Nernstian responses except for membrane 54. The Kf$03-, cI-, values were slightly inferior to those achieved, using TAOAB as the added ion exchanger, however, they were again superior to the values obtained from a commercially available PVC membrane. Conclusion As reported in an earlier work,l membranes have again been successfully fabricated with the so-called hot pressing tech- nique. Membranes fabricated with copolymers of acrylonitrile and butadiene with 19 and 34% acrylonitrile content gave selectivity coefficients, Gh3-,c,-, of the order of 1.1 X 10-3. These selectivity coefficients were superior to those obtained using a commercially available PVC membrane, however, the lifetimes were significantly reduced, with sub-Nernstian responses occurring after only 21 d.The best selectivity coefficient, P$A3-, was achieved for membrane 16, prepared using a copolymer .of acrylonitrile and butadiene with an ACN content of 19% and 5.6% m/m TABAB, 7.3% DCP and 39.7% 2-NPOE. Another range of membranes with varying cross-link densities from 0.6 X 10-5 to 11.0 X 10-5 mol cross-link cm-3 were fabricated and the cross-link density determined using the Flory-Rehner equation .7 Within the range evaluated, cross-link density had no effect on the electrochemical response although there was an impact on the selectivity coefficients. Increasing the cross-link density reduced the interference from the large anion Mn04-, whereas reducing the cross-link density increased the inter- ference from small anions such as F-.Two novel ion exchangers , TAOAB and TADDAB were successfully synthesized. Membranes were fabricated with both ion exchangers although TADDAB was found to be incompatible with the membrane composition and gave greasy membranes. The use of TAOAB as the added ion exchanger produced membranes with the longest lifetime achieved so far. Mem- brane 43 had a lifetime in excess of 500 d and confirms the advantages of immobilizing a carefully chosen ion exchanger within a cross-linked polymer network. The composition of this, the longest lifetime membrane, was based on a 50% ACN copolymer with 6.5% m/m TAOB, 7.2% m/m DCP and 39.7% 2-NPOE. The fabrication technique employed offers the advantage, over the method employed for conventional PVC membranes, of greater control over membrane thickness and other physical parameters.In addition, the selection of a commercially available rubbery polymer and the use of the hot pressing technique, which can be controlled reliably, enables1010 ANALYST, OCTOBER 1991, VOL. 116 Table 10 Lifetime studies for TAOAB membranes Membrane No. Parameter Original Slope Limit of detection? 21 Days Slope Limit of detection 58 Days Slope Limit of detection 106 Days Slope Limit of detection 139 Days Slope Limit of detection 369 Days Slope Limit of detection 412 Days Slope Limit of detection 510 Days Slope Limit of detection * Slope measured in mV decade- * . 40 42 55.3 58.4 2.5 2.5 56.1 51.8 7.9 13 55.0 49.7 5.0 8.9 31.2 42.5 5.0 40 t Limit of detection measured in mol dm-3 x 10-5. 43 58.3 4.5 58.2 6.3 58.7 7.0 58.2 4.0 58.2 5.0 58.2 7.0 58.9 6.3 58.2 6.3 44 58.6 5.0 57.5 7.9 59.0 5.0 16.5 16 - - - - - - - - 45 58.6 2.5 58.7 10 58.5 7.0 41.6 32 - - - - - - - - 46 58.8 3.5 57.4 10 54.3 6.3 54.8 6.3 55.9 10 47.6 25 - - - - 47 59.2 3.2 59.2 13 56.5 8.9 42.7 6.3 - - - - - - - - 48 58.6 2.0 59.1 7.9 56.8 8.9 55.8 5.0 51.6 13 33.2 - - - - - the fabrication method to be modified to suit mass production techniques. References 1 Ebdon, L., Braven, J., and Frampton, N. C., Analyst, 1990, 115, 189. 2 Blow, C. M., and Hepburn, C., Rubber Technology and Manufacture, Butterworth Scientific, London, 2nd edn., 1982, p. 342. 3 Vizgerts, R. V., Sendega, R. V., and Zhovnirchuk, V. M., Ukr. Khim. Zh. (Russ. Ed.), 1970, 36, 822. 4 King, B. A., PhD Thesis, Sheffield City Polytechnic, 1985. 5 Neppel, A., Rubber Chem. Technol, 1986,59,46. 6 Bailey, P. L., Analysis with Ion Selective Electrodes, Heyden, London, 2nd edn., 1982, p. 49. 7 Flory, P. G., and Rehner, J., J . Chem. Phys., 1943, 11, 521. 8 Flory, P. J., J. Chem. Phys., 1950, 18, 108. Paper 1 I0041 OG Received January 29th, I991 Accepted June loth, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911601005

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 1011 Plastic Membrane Ion-selective Electrode for the Determination of Denatonium Benzoate (Bitrex)* 0. G. B. Nambiar, Kalpana Gosavi and T. Ravindranathan National Chemical Laboratory, Pune-4 7 1 008, India A poly(viny1 chloride) matrix membrane ion-selective electrode for the determination of the denatonium ion based on the denatonium salt of tetraphenyl borate is described. The response characteristics of the electrode for the denatonium ion and for several quaternary ammonium compounds were studied. The potentiometric determination of denatonium benzoate in rapeseed oil in the range 1-10 ppm agreed to within +5% of the spiked amounts. The application of the electrode to the titrimetric determination of several quaternary ammonium compounds using sodium tetrapheny!borate as the titrant is also described.Keywords: Plastic membrane ion-selective electrode; denatonium salt of tetraphen ylborate; denatonium benzoate (Bitrex) The denatonium salts* (1) and (2) are probably the bitterest substances known to man and are used as denaturants or as aversive agents. Denatonium benzoate (2) is popularly known as Bitrex (Macfarlan Smith). Hence, these salts are incorpor- ated in household products such as disinfectants, detergent formulations and shampoos in order to prevent accidental poisoning, particularly among children. In this laboratory, while studying the possibility of using denatonium compounds as crop protecting agents against animals and birds, a rapid and sensitive method for the determination of these com- pounds was required.It was found that the denatonium salt of tetraphenylborate incorporated in a plastic membrane re- sponded linearly to the denatonium salts down to a concentra- tion level of 1 x 10-7 rnol dm-3. The determination of denatonium compounds by thin-layer chromatography2 and high-performance liquid chromatography (HPLC)3 has been reported, but these techniques are tedious and time consuming. The melting-points of the salts (X = C1, m.p. = 175 "C and X = benzoate, m.p. = 160 "C) agreed with the reported values for the pure compounds. Dioctyl sebacate (DOS), poly(viny1 chloride) (PVC) and sodium tetraphenylborate (NaTPB) were obtained from Fluka. All of the other chemicals used were of analytical-reagent grade. Construction of the Electrode Denatonium tetraphenylborate (DTPB) was precipitated by mixing 30 ml of 1 x 10-2 rnol dm-3 DC with 20 ml of 1 X 10-2 rnol dm-3 NaTPB solution.The precipitate was filtered, washed and dried. The ion-exchange solution was prepared by dissolving 0.1 g of DTPB in 1 g of DOS. The master membrane was fabricated from 0.5 g of the above ion-exchange solution and 0.21 g of PVC. The method of preparing the PVC- immobilized carrier membrane and its assembly into the electrode was as described in the literature4.5. A solution consisting of 1 x 10-2 mol dm-3 KCI (saturated with AgCI) and 1 x 10-3 rnol dm-3 DC was used as the internal filling solution of the ion-selective electrode. An Ag-AgC1 wire was used as the internal reference electrode. Electrodes were connected to measuring instruments with shielded coaxial copper cables.Denatonium salts (1): X = chloride (2): X = benzoate This paper describes the construction and performance characteristics of an ion-selective electrode for the determina- tion of denatonium compounds. Experimental Apparatus An Ingold standard pH electrode was utilized for pH measurements. All electrode potentials were measured at 25 k 1 "C with a digital pH/millivoltmeter in constantly stirred solutions. An Ag-AgC1 type double-junction electrode with 10% NaN03 as the junction solution was used as the reference electrode for potential measurements. Reagents and Materials Denatonium chloride (DC) (1) and denatonium benzoate (DB) (2) were prepared by the procedure described by Hay.' * NCL Communication No.5170. Electrode Calibration Aliquots (25 ml) of 1 x 10-2-1 x 10-7 rnol dm-3 standard DC solutions were transferred into 50 ml beakers and 0.5 ml of 1 rnol dm-3 NaCl was added to the solutions as an ionic strength adjuster. The DTPB electrode, in conjunction with the reference electrode, was immersed in the solutions. The solutions were stirred and the potential of each solution was recorded after stabilization and was plotted on semi-logarith- mic graph paper as a function of the concentration of DC. The graph was used for the subsequent determination of the unknown concentration of the DB sample. Results and Discussion The DTPB electrode, immobilized with PVC and with DOS as the solvent mediator, was examined as an electroactive material responsive to the denatonium ion.The PVC mem- brane electrode was also used for the potentiometric determi- nation of DB by using the following electrochemical cell: Ag-AgCI, 1 x 10-3 mol dm-3 DC, 1 x 10-2 mol dm-3 NaCl I PVC membrane 1 DB in test solution 11 10% NaN03 11 KCI (3 rnol dm-3) I AgCl-Ag. The calibration graph (Fig. l ) , constructed on semi-logarithmic graph paper by plotting1012 ANALYST, OCTOBER 1991, VOL. 116 250 I 200 $150 . v) Q Q, g 100 a 50 0 - 50 I I I lo-' I O - ~ 1 0 - ~ I O - ~ Denatoni u m c h loride/mol dm -3 Fig. 1 Calibration graph showing the response of the DTPB electrode to increasing concentrations of DTPB O t -50 2 0 1 2 3 4 5 6 7 8 9 PH Fig. 2 concentrations of DC: A , 1 x 10-3; B, 1 x 1 x 10-6 rnol dm-3 Effect of pH on the response of the DTPB electrode at various C, 1 x 10-5; and D , measured potentials versus concentration, exhibited a straight line in the concentration range from 1 x 10-2 to 1 x 10-6 rnol dm-3.The slope of the calibration graph was 56 5 1 mV per concentration decade. The slope decreased to about 42 mV per concentration decade between 1 x 10-6 and 1 x 10-7 rnol dm-3. Determination of Selectivity Coefficients The selectivity coefficients for foreign ions were determined by the mixed solution (fixed interferences method).6 A background concentration of foreign ions of 0.1 rnol dm-3 was employed in each instance. The values obtained for different ions were as follows: Na+, 2.4 X Kf, 6.3 X 10-4; NH4+, 1.8 x 10-3; and Ca2+, 4 X 10-5. Response Time and Effect of pH The electrode displayed average response times ranging from 20 s for DC concentrations up to 1 X 10-4 rnol dm-3 to 60 s for lower concentrations.The electrode response for different DC concentrations at various pH values indicated almost constant potentials to within k 2 mV, in the pH range 3-8. The influence of pH on the response of the liquid membrane electrode to different concentrations of DC is shown in Fig. 2. Analytical Applications Potentiometric determination of DB in rapeseed oil The method for the extraction of DB from oils described by Damon and Pettitt3 was modified for the potentiometric Table 1 Titrimetric determination of several quaternary ammonium salts Quaternary ammonium salt Benzyltriethylammonium chloride Benzyltriethylammonium chloride Triethylpropylammonium bromide Triethylpropylammonium bromide Benzyltrimethylammonium bromide Benzyltrimethylammonium bromide Tetramethylammonium bromide Tetramethylammonium bromide * Mean of three determinations.Amount takedmg 60 10 80 15 120 12 90 30 Amount found*/mg 60.8 9.8 79.8 15.3 118.0 12.3 90.3 29.2 determination of this compound. Accordingly, the aqueous extract of DB, after concentration, was analysed by direct potentiometry . The following procedure was adopted for the determination of DB using the proposed DTPB electrode. A stock solution containing 100 ppm of DB in rapeseed oil was prepared. Standard solutions containing 0.1, 1.0 and 10.00 ppm were prepared by serial dilution of the stock solution with rapeseed oil. A 50 ml volume of each of the DB standards was diluted with light petroleum (boiling-point range, 60-80 "C) (50 ml) and DB was extracted with water-methanol (1 + 1) (50 ml).Each of the extracts was concentrated to about 5 ml on a rotary evaporator separately and, after the addition of 1 ml of 1 mol dm-3 NaCl solution as an ionic strength adjuster, was diluted to 50 ml with water. Potentiometric measurements were made for the standard extracts. A calibration graph was constructed by plotting potential versus concentration. Sam- ples of rapeseed oil were extracted with water as described above and the DB content was determined from the calibra- tion graph. Recoveries from samples spiked with known amounts of DB in the range from 1 to 10 pprn agreed to within +5%. This potentiometric method appears to be attractive when compared with the HPLC procedure3 in terms of rapidity and reproducibility. Use of the DTPB electrode as an indicator electrode for the titrimetric determination of quaternary ammonium salts The DTPB electrode was found to respond to the presence of NaTPB.A number of quaternary ammonium salts, which are precipitated from solution by NaTPB solution as phase- transfer catalysts, were determined by using the above reagent as the titrant and the DTPB electrode as the indicator electrode. The titrimetric determination of several quaternary ammonium salts, using NaTPB as the titrant, was carried out. The results obtained are shown in Table 1. The accuracy was found to be 99 k 2%. K. G. is grateful to the Council of Scientific and Industrial Research (CSIR) for the award of a research fellowship. References Hay, J. E., Br. Puts., 955 309 and 955 3110, 1960. Glover, M. J., and Blake, A. J., Analyst, 1972, 97, 891. Damon, C. E., and Pettitt, B. C., J. Chromarogr., 1980, 195, 243. Moody, G. J., Oke, R. B., and Thomas, J. D. R., Analyst, 1970, 95,910. Craggs, A., Moody, G. J., and Thomas, J. D. R., J. Chem. Educ., 1974,51, 541. Kamata, S., Yamasaki, K., Higo, M., Bhale, A., and Fuku- naga, Y., Analyst, 1988, 113,45. Paper 1 I01 901 E Received April 23rd, 1991 Accepted June I9th, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911601011

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 Use of the Hildebrand Grid Nebulizer for Inductively Coupled Atomic Emission Spectrometric Analysis of Foodware Leach Solutions and Rodent Soft Tissues and Femurs 1013 Plasma Susan C. Hight Division of Contaminants Chemistry, US Food and Drug Administration, Washington, DC 20204, USA Jeanne I. Rader Division of Nutrition, US Food and Drug Administration, Washington, DC 20204, USA The Hildebrand grid nebulizer (HGN) was used for the inductively coupled plasma atomic emission spectrometric determination of major, minor and trace elements in perchloric acid digests of rodent femurs, sulphuric acid digests of rodent soft tissues and in solutions leached from foodware using acetic acid. The HGN performed well when the signal-to-background ratio was optimized for each acid solution by adjusting the injector gas flow, solution uptake rate and observation height.Three problems were overcome while using the HGN: ( i ) nebulizer wash-out time was reduced by rinsing at high uptake rate with the solution to be analysed; (ii) clogging of the injector tip of the torch during femur analysis was minimized by extensive rinsing; and (iii) errors due to the suppression of the Cu, Fe, Mn and Zn signal intensities by matrix elements Ca and P in femur digests were eliminated by calibrating the spectrometer with matched matrix standard solutions. Overall, the precision of analysis for the leach and tissue solutions analysed in this study ranged from 0.5 to 2.9% relative standard deviation. Keywords: Inductively coupled argon plasma atomic emission spectrometry; grid nebulizer; wash-out time; matrix effect; bone and biological tissue analysis; leach solution analysis The US Food and Drug Administration routinely uses inductively coupled argon plasma atomic emission spec- trometry (ICP-AES) to analyse foodware leach solutions and rodent tissues and femurs.Leach solutions consist of 4% v/v acetic acid that has been in contact with foodware for 24 h at room temperature.’ Rodent soft tissues are digested with a mixture of boiling nitric, perchloric and sulphuric acids until all the perchloric and nitric acid has been driven off and only sulphuric acid remains in the digest flask.2 Whole rodent femurs are digested in a mixture of boiling nitric and perchloric acids until the reaction is complete and white fumes of perchloric acid are visible in the flask.All tissue digests are diluted with de-ionized water so that final solutions are 10% v/v in sulphuric or perchloric acid.2 The cross-flow nebulizer (Perkin-Elmer Model N058-0358) designed for use with the spectrometer used in the present study performed poorly for these analyses. The solution capillary became clogged and required replacement at least twice per day during the analysis of femur digests containing high levels of Ca and P and moderate levels of K. After 2-3 d of nebulizer use, the solution began to leak through the space between the capillaries and nebulizer body and resulted in poor precision. Similar observations have been reported by others .3y4 Use of the Hadebrand dual grid nebulizer (HGN) was investigated in order to eliminate these problems.This nebulizer, which was designed for use with solutions contain- ing high concentrations of dissolved solids, is constructed from high density polyethylene and platinum and offers more resistance to acid than the cross-flow nebulizer manufactured for the spectrometer used in this study. Solutions are pumped over a platinum grid located in front of a sapphire gas orifice through which argon flows. The second grid is mounted in an adjustable, threaded end cap, positioned downstream in the argon flow. The design of the nebulizer576 and the character- istics of the aerosol it produces from de-ionized water7 have been described by other workers. This paper presents the results of quantitative analyses using the HGN to analyse rodent bones and soft tissues, leach solutions and quality control (QC) solutions.The problems of long wash-out time and suppression of trace element intensity by high levels of Ca and P are addressed. The optimization of the injector gas flow, solution uptake rate and observation height is also discussed. Experimental Instrumentation and Reagents All experiments were performed on a sequential spectrometer equipped with an autosampler. Instrument specifications and optimized operating conditions for the plasma are given in Table 1. Analytical wavelengths and experimental parameters for the spectrometer are given in Table 2. Single-element stock standard solutions, 1 000 or 10 000 yg ml-1, [Johnson Matthey AESAR Group, Seabrook, NH, USA, and National Institute of Standards and Technology (NIST), Gaithersburg, MD 20899, USA, respectively] were used to prepare all working standards and QC solutions.All dilutions, designated v/v, were accomplished using calibrated pipettes and flasks. Analytical-reagent grade glacial acetic acid, diluted to 4% v/v, was used for acetic acid standards. Extra high-purity , concentrated perchloric and sulphuric acids (‘double distilled’ label, GFS Chemical, Columbus, OH, USA) were diluted to 10% v/v in order to obtain working standard solutions for tissue digests. Analytical-reagent grade phosphoric acid, calcium carbonate, sodium carbonate, potas- sium carbonate and magnesium sulphate were used to prepare simulated solutions of digested femurs and matched matrix standards.High purity concentrated nitric, perchloric and sulphuric acids (‘redistilled’ label, GFS Chemical) were used to digest rodent tissues, femurs and the NIST standard reference material (SRM 1577a, Bovine Liver). De-ionized water (318 MQ) was used in all preparations. Concentrations of digests and QC solutions were calculated from emission intensity via the spectrometer software by comparison with the calibration standards. The spectrometer was standardized using two-point calibration (a blank and one standard) for analyses. The blank and standard solution contained 4% v/v acetic, 10% v/v sulphuric and 10% v/v perchloric acids for leach solutions, tissue digests and femur digests, respectively. The leach solution standard contained 1 pg ml-1 of Cd and Mn and 10.0 pg ml-1 of Pb.The tissue1014 ANALYST, OCTOBER 1991, VOL. 116 Table 1 Instrumentation and optimized operating conditions Spectrometer Radiofrequency Torch generator Spray chamber Nebulizer Solution delivery Perkin-Elmer Plasma 2 spectrometer (Norwalk, CT, USA). Two 1 m Ebert scanning monochromators capable of 0.009 and 0.018 nm resolution (monochromators A and B, respectively), operated at 4.0-6.7 Pa vacuum pressure 27.12 MHz R.f. generator with automatic tuning Demountable (PE Model Type 1). Quartz injector tube, 1.46 mm i.d. (PE part No. OO47-3292), was modified to connect with ball joint on spray chamber and held tightly in place by a PTFE insert in the torch assembly Glass, double-pass (Scott type) without water cooling Hildebrand grid nebulizer (Leeman Labs, Lowell, MA, USA).The HGN adapter drain was sealed by attaching a short piece of tubing that was crimped closed. Waste solution was drained at a port located at the opposite end of the spray chamber. Injector gas flow was controlled by a thermostated mass flow controller. Argon gas flows were checked with a mass flow controller (Model 82001 8102-1433-FC, Matheson, Secaucus, NJ, USA) and a wet test meter (Model 63115, Precision Scientific, Chicago, IL, USA). 12 Roller peristaltic pump, with standard poly(viny1 chloride) tubing, 0.76 mm (0.030 in) i.d., and tubing tension adjustment Optimized operating conditions- Forward power Observation height Solution uptake rate Argon gas flow rate: Outer Intermediate Injector (4% v/v acetic acid) (10% v/v perchloric acid) (10% v/v sulphuric acid) 1.1 kW 12 mm 1.0ml min-l 15 1 min-1 11 min-1 0.70 1 min-1 0.75 1 min-1 0.83 1 min-1 digest standard contained 1 pg ml-1 of Cu, Fe, Mn and Zn, 10 pg ml-1 of Ca and Mg, and 25 pg ml-1 of P.The standard used for femur digests contained 1 pg ml-1 of Cu, Fe, Mn and Zn, 20 pg ml-1 of Na and K, 2450 pg ml-1 of Ca and 1225 pg ml-1 of P. Procedure for Instrument Optimization Pump tension and nebulizer end cap position were adjusted before the HGN was installed on the spray chamber by setting the solution uptake rate and injector gas flow rate to 1.0 ml min-1 and 0.8 I min-1, respectively, aspirating water and visually observing the aerosol formed. The end cap was positioned with respect to the nebulizer body so that a V-patterned aerosol, uniform over time, was formed.‘Spit- ting’ occurred (large drops) when the cap was too far from the nebulizer body and ‘skipping’ occurred (gaps appeared at about 0.5-2 s intervals) when the cap was too close. The nebulizer was then attached to the spray chamber. Combinations of injector gas flow-and solution uptake rate that produced good (visually dense, non-skipping) aerosols were determined by aspirating the acid solution of interest and visually observing the aerosol formed in the spray chamber. Injector gas flow rates of 0.6-0.8,0.6-0.8 and 0.8-0.9 1 min-1 were found for 4% v/v acetic, 10% v/v perchloric and 10% v/v sulphuric acids, respectively. Solution uptake rates of 0.8-1 .O, 1.0 and 1.0 ml min-1 were found for 4% v/v acetic, 10% v/v perchloric and 10% v/v sulphuric acids, respectively.Table 2 Analytical wavelengths and experimental parameters for the spectrometer Background Wavelength/ Sampling correction/ Element nm PMT*N timet/ms nm Monochromator A (survey2 and peak9 windows = 0.050 and 0.025 nm , respectively)- 228.802 850 500 k0.020 Cd I c u I 324.754 800 200 k0.020 Fe I1 259.940 800 200 k0.020 Mn I1 257.610 750 250 k0.020 Ni I1 221.647 850 200 f0.020 P I 213.618 750 250 Off Zn I 213.856 800 500 k0.020 Monochromator B (survey and peak windows = 0.075 and 0.035 nm, respectively)- Ca I1 393.366 500 200 Off K I 766.490 650 250 Off Mg I1 279.806 600 100 Off Na I 589.592 650 250 Off Pb I1 220.353 850 lo00 0.030 * Photomultiplier tube. t Time used per grating step to generate survey and peak profile 2 Total wavelength range over which the spectral profile is taken.§ Width surrounding analytical line which is used for curve fitting of data. intensity measurement. The signal-to-background ratio was determined for a representative element at 5 4 different observation heights (10-20 mm above the load coil) for each flow combination that produced a good aerosol. Analyte emission was obtained from a solution containing 1 pg ml-1 of Mn in the acid of interest. Background emission was obtained from the appropriate concentration of the same acid. Emission data were generated using the monochromator parameters listed in Table 2 with the exception of background correction, which was ‘turned off‘. The signal-to-background ratio was calculated as follows: S/B = (EMS - EMb)/EMb where EM, is the emission due to the analyte and EMb is the emission due to the background.Monitoring Analytical Performance The spectrometer was calibrated with standard solutions once every 40-60 min in order to compensate for instrumental drift. Quality control solutions were analysed and the results were used to calculate precision because the volume of leach and tissue solution available for analysis was limited. The QC solutions contained acid concentrations equal to, and element levels similar to the unknown solutions. The QC solutions were analysed immediately after calibration of the instrument (before the analysis of leach, tissue or femur digests) and periodically during leach, tissue or femur analysis (QC solution in every sixth and eleventh tube following leach, tissue and femur tubes).When QC results deviated more than k 5 % from known levels, all results obtained before the unacceptable QC analysis were discarded and the solutions re-analysed. Over-all method performance for soft tissue analysis was monitored by digesting and analysing 26 portions of NIST SRM 1577a (Bovine Liver) at regular intervals during the study. Results and Discussion Wash-out Time Long wash-out times for the HGN have been reported by other workers698 who eliminated this problem with extraANALYST, OCTOBER 1991, VOL. 116 1015 rinsing. It was decided not to use an extended rinse time in the present study because the time required to eliminate cany- over, when the rinse solution was pumped at the analytical rate, more than doubled the analysis time compared with other cross-flow nebulizers (Perkin-Elmer Model N058-0358 and Thermo Jarrell Ash Model 90-790) used on this spec- trometer.Increased analysis time is undesirable for the analysis of large numbers of samples. In this study, wash-out time was reduced to an acceptably short time by rinsing the nebulizer and spray chamber with the solution to be analysed for 45 s at 4 ml min-1. Wash-out was considered adequate when a blank analysed immediately after an undiluted femur digest gave emissions of S3 x standard deviation (SD) of emissions obtained from a blank analysed before the undi- luted femur digest. Varnesg reported decreased sensitivity in aqueous solutions when high pump rates were used for the HGN. During analysis of leach and tissue solutions, the high solution uptake rate required to rinse adequately in an acceptable time period also depressed trace element intensities.Intensities measured immediately after rinsing ( t = 0) were about 4% less than intensities measured after the aerosol had stabilized (t 3 3 min). In order to save time and reduce the volume of solution consumed, sequential measurements were initially taken 15 s after the uptake rate was resumed, i.e., before the aerosol had stabilized. In order to obtain acceptable analytical precision, however, it was necessary to reproduce precisely (i) the rinse time; (ii) the solution uptake rate during rinsing; and (iii) the time after resuming the analytical uptake rate and before emission intensities were taken. An autosampler and a computer-controlled pump were used in this study to meet these requirements.Clogging of the Torch Injector Tip Analysis of femur digests presented two problems not encountered in the analysis of soft tissue digests and acetic acid leach solutions. For all elements determined, results showing negative error were found for QC solutions analysed immediately after femur digests. The longer the femur digests were nebulized, the lower were the QC results. Visual inspection showed that clogging of the nebulizer did not occur but a white residue was deposited on the injector tip of the torch when low results were obtained. Other researchers have reported clogging of the injector tip,6,7 which was reversed by rinsing with dilute nitric acid.6 In the present work, it was necessary to rinse with 10% perchloric acid because changing solvents (from perchloric acid to nitric acid and back to perchloric acid again) adversely affected precision because of the accompanying change in solvent vapour pressure in the spray chamber.The design of the spray chamber drain on the ICP used in this study requires that the level of liquid in the drain tube be approximately 2.5 cm below the spray chamber and that the contents of the drain tube be changed every time a new solvent is introduced. Suppression of Trace Element Intensity The second problem encountered in femur analysis was suppression of trace element intensities by high levels of Ca and P. When Fe and Zn were determined by calibrating the spectrometer with standard solutions containing 10 pg mt-* of Ca and 50 pg ml-l of P, the results were 83 and 77%, respectively, of the levels established for a laboratory control material (rabbit femurg).However, when analysed by the method of standard additions, levels found in the control material digest were 97 and 98%, respectively, of the established levels. The analysis of simulated femur solutions, containing 1 pg ml-1 of Cu, Fe, Mn and Zn and increasing concentrations of Ca and P, confirmed that trace metal intensities were suppressed by up to 10% for Cu and 20% for Fe, Mn and Zn by the matrix elements (Fig. 1). When femur .o 0.95 4- E c 5 0.90 8 2 C cI 0.85 0.80 a* 0.75 0 500 1000 1500 2000 2500 Ca concentration/pg ml-' I I I 1 1 1 1 0 250 500 750 1000 1250 P concentration/pg mi-' Fig.1 Effect of Ca and P concentration on the results for A, Cu; B, Fe; C, Mn; and D, Zn. Analytical solutionscontained 1 pgml-1 of Cu, Fe, Mn and Zn and increasing concentrations of Ca and P in 10% v/v perchloric acid. The spectrometer was standardized with solutions containing 10 pg ml-l of Ca and Mg, 50 pg ml-l of P, and 1 pg ml-l of Cu, Fe, Mn and Zn in 10% v/v perchloric acid Table 3 Analysis of QC solutions (containing 4% acetic acid and known concentrations of elements) and a typical leach solution (containing 4% acetic acid used to leach foodware) Elementlpg ml-1 Parameter Pb Cd Mn QC solution analysed before analysis of leach solutions- Actual 10.0 1.00 1.00 Mean ( n = 30) 9.81 0.998 0.987 SD 0.116 0.00536 0.00658 RSD (%) 1.2 0.5 0.7 Outliers* 0 0 0 Actual 10.0 1 .00 NAt Mean (n = 5) 10.03 1.00 NA SD 0.089 0.012 NA RSD (Yo ) 0.9 1.1 NA Outliers* 0 0 NA Found (n = 1) 14.5 <LOQS NA * Outliers = number of results which deviated more than +5% from t NA indicates solution not analysed for this element.f <LOQ indicates not found above the limit of quantification which QC solution analysed periodically during analysis of leach solutions- Leach solution- the actual level. equals 0.01 pg ml-1 (10 x SD of the blank) for Cd. digests were analysed by calibrating the instrument with standards containing 2500 and 1250 pg ml-1 of Ca and P, respectively, the results were in 100% agreement with results obtained using the method of standard additions. These levels were chosen because a typical femur digest may contain about 2000-2500 pg ml-1 of Ca and 1000-1250 pg ml-1 of P.Matched matrix standard solutions were used for routine analyses because the method of standard additions is too time consuming for the analysis of large numbers of solutions and is not feasible for small volumes of digest such as those produced in animal experiments. Suppression of trace element emission by similar levels of Ca has been reported by Thompson and Ramsey for the analysis of geological materials.'" Analytical Performance Summaries of results obtained while monitoring analytical performance are presented in Tables 3-5 for acetic, sulphuric and perchloric acid solutions, respectively. The data in Tables1016 ANALYST, OCTOBER 1991, VOL. 116 Table 4 Analysis of QC solutions (containing 10% v/v sulphuric acid and known concentrations of elements) and digests of NIST SRM 1577a Bovine Liver (in 10% v/v sulphuric acid) Element/pg ml-1 Parameter c u Fe Mn Zn Actual 1 .oo 10.0 1 .oo 1 .oo Mean ( n = 32) 0.998 9.87 0.982 0.986 SD 0.008 0.088 0.012 0.011 Outliers* 0 0 0 0 Actual 1 .oo 10.0 1 .oo 1.00 Mean (n = 26) 0.995 9.94 0.992 0.990 SD 0.018 0.106 0.01 0.017 RSD (%) 1.8 1.1 1 .o 1.7 Outliers* 0 0 0 1 QC solution analysed before analysis of tissue digests- RSD (%) 0.8 0.9 1.2 1.1 QC solution analysed periodically during analysis of tissue digests- NISTSRM 1577a digestst- Mean (n' = 26)$/yg g- 147 180 9.7 118 RSD (Yo) 8.8 11.0 8.7 8.6 Percentage of certified value 93 93 98 96 * Outliers = number of results which deviated more than +5% from the actual level.t Dilution factor for SRM digests is approximately 0.01 g ml-I.$ n' = Number of individual portions digested. Each digest was analysed once. Ca 5.00 4.97 0.056 1.1 0 5 .oo 4.97 0.15 2.9 1 128 8.3 107 Mg 5.00 4.99 0.05 1 .o 0 5.00 4.99 0.061 1.2 0 595 8.4 99 P 25.0 24.9 0.292 1.2 0 25 .O 24.8 0.527 2.1 2 11300 8.8 102 Table 5 Analysis of QC solutions (containing 10% v/v perchloric acid, known concentrations of trace elements, 2450 yg ml-* of Ca and 1225 yg ml-1 of P) and a typical femur digest (in 10% v/v perchloric acid) Element/pg ml-1 Parameter c u Fe Mn Zn Na K Actual 1 .oo 1 .oo 1 .oo 1 .oo 10.0 10.0 Mean (n = 9) 1.01 0.997 1 .oo 1.01 10.0 9.80 SD 0.011 0.013 0.008 0.017 0.117 0.114 Outliers* 0 0 0 0 0 0 Actual 1 .oo 1.00 1 .oo 1.00 25 .O 25.0 Mean ( n = 73) 0.992 0.974 0.993 0.996 24.6 24.9 SD 0.020 0.014 0.011 0.019 0.51 0.58 RSD (Yo) 2.0 1.5 1.1 1.9 2.1 2.3 Outliers* 2 3 0 0 4 3 Found ( n = l)/pg g-1 1.5 32 <LOQ$ 156 4130 6130 QC solution analysed before analysis of femur digests- RSD (Yo) 1.1 1.3 0.8 1.7 1.2 1.2 QC solution analysed periodically during analysis of femur digests- Femur digest?- * Outliers = number of results which deviated more than +5% from the actual level.t Dilution factor for femur digests is approximately 0.01 g ml-l. $ <LOQ indicates not found above the limit of quantification which equals 1.0 pg g-1 (10 x SD of the blank) for Mn. 3-5 were collected during time periods ranging from 1 h to 10 d and provide an estimate of the over-all precision of analysis for the leach and tissue solutions analysed.The precision obtained using the HGN was good for the solutions analysed in this study. Ranges of relative standard deviation (RSD) for QC solutions analysed before analysis of leach, tissue or femur solutions were 0.5-1.2% for acetic acid, 0.&1.2% for sulphuric acid, and 0.8-1.7% for perchloric acid. The precisions of QC solutions, analysed after 5-10 tissue digests, were 1.&2.9% RSD for QC solutions in 10% v/v sulphuric acid and 1.1-2.3% RSD for QC solutions containing high levels of Ca and P in 10% v/v perchloric acid. The over-all precision of the method for soft tissue analysis, calculated from results of 26 portions of SRM 1577a taken through the digestion procedure and analysed once, was 8.3-11% RSD. The number of outliers (results deviating more than &5%0 from the actual level in the QC solution) indicates how seldom the analyst was required to re-analyse tissue and femur digests when using the HGN.When QC solutions were analysed periodically during the analysis of tissue and femur solutions, approximately 16% of the results deviated more than _+5% from the actual level (4 out of 26 for sulphuric acid and 12 out of 73 for perchloric acid). No outliers were obtained, however, when QC solutions were analysed before tissue or femur solutions were introduced. Outliers obtained during the analysis of tissue and femur digests may be due to trace levels of undigested organic material. The number of outliers obtained when QC solutions were analysed periodically during the analysis of leach solutions was the same as when QC solutions were analysed before the leach solutions because the compositions of the QC and leach solutions were not significantly different.Conclusions The HGN provided good precision when injector gas flow and observation height were optimized for 4% v/v acetic acid, 10% perchloric acid and 10% v/v sulphuric acid. The greatest advantage in using the HGN, however, was that perchloric acid solutions of digested femur did not clog the nebulizer.ANALYST, OCTOBER 1991, VOL. 116 Although the problem of nebulizer clogging was eliminated by using the HGN, high salt content in femur digests clogged the injector tip of the torch. Extra rinsing, in addition to the rinsing required to eliminate carryover, was necessary in order to prevent blockage of the injector tip for these digests. The use of matched matrix standard solutions was essential to compensate for severe suppression of trace element intensity in femur digests. The major drawback in using the HGN was the large volume of solution required to eliminate carryover from the previous analysis. The wash-out time was shortened by rinsing with the solution at a high uptake rate, but precise timing of the rinse and analysis was necessary in order to obtain acceptable precision. 1017 References 1 Official Methods of Analysis of the Association of Official Analytical Chemists, ed. Helrich. K . , Association of Official Analytical Chemists, Arlington, VA, 15th edn., 1990, method 973.32, p. 241. 2 3 4 5 6 7 8 9 10 Rader, J. I., Wolnik, K. A., Gaston, C. M., Celeck. E. M., Peeler, J. T., Fox, M. R. S., and Fricke, F. L., J. Nutr., 1984, 114, 1946. Nixon, D. E., and Smith, G. A., Anal. Chern., 1986,58,2886. McLaren, M. A., and Anderau, C., At. Spectrosc., 1989,10,77. Sharp, B. L., Barnett, N. W., Burridge, J. C., andTyson, J . F., J. Anal. At. Spectrorn., 1987, 2, 167R. Brotherton, T., and Caruso, J., J . Anal. At. Spectrorn., 1987,2, 695. Smith, T. R., and Denton, M. B., Appl. Spectrosc., 1990, 44, 21. Varnes. A. W., J. Anal. At. Spectrorn.. 1988, 3 , 803. Wolnik, K. A., Rader, J. I., Gaston, C. M., and Fricke, F. L., Spectrochirn. Acta, Part B, 1985, 40, 245. Thompson, M., and Ramsey, M. H., Analyst, 1985, 110, 1413. Paper 010561 OC Received December 13th, 1990 Accepted May 22nd, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911601013

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, OCTOBER 1991, VOL. 116 1019 Evaluation of a Molecular Recognition Ligand for Performing the Extraction of Palladium, Platinum and Rhodium From lon-charged Solutions and Its Application to Geochemical Exploration Techniques Using Electrothermal Atomic Absorption Spectrometry Mario Bergeron Quebec Geoscience Centre, 2700 Rue Einstein, P. 0. Box 7500, Sainte-Foy, Quebec GI V 4C7, Canada Marc Beaumier, Ministere de I'Energie et des Resources du Quebec, 5700 4ieme Avenue Ouest, Niveau 2, Charlesbourg, Quebec GIH 6R1, Canada Andre Hebert Quebec Geoscience Centre, 2700 Rue Einstein, P.O. Box 7500, Sainte-Fo y, Quebec GI V 4C7, Canada In order t o develop a simple analytical procedure for determining the content of Pd, Pt and Rh in samples of stream and lake sediment, the ability of a molecular recognition ligand (Superlig 202: SL-202) t o extract quantitatively these three elements from solutions loaded with different ions was assessed.The evaluation was based on (1) the use of a 0.1 mol dm-3 HCI solution containing 50 ng of Pd, 250 ng of Pt and 50 ng of Rh, t o which 100 pg of each of the following six metals had been added: Co, Cu, Cr, Ni, Pb and Zn; and (2) a solution resulting from an aqua regia [HCI-HN03 (3+1)] attack on a standard sample of stream sediment the Pd, Pt and Rh content of which had been previously determined by the lead fire assay method. The results indicate complete recovery of Pd, Pt and Rh from the 0.1 mol dm-3 synthetic solution, with no particular interference by the six added metals. For the solution resulting from the aqua regia attack, the data sets for Pd and Pt are indistinguishable from the concentrations measured by lead fire assay, when taking analytical errors into consideration.No Rh was detected in this solution. In order t o show that the loss of Rh was not caused by problems related t o the process of fixing and washing the ligand, but rather t o the solubility of Rh during the aqua regia attack, standard additions of Pd, Pt and Rh (I0 pg kg-1 of Pd, 50 pg kg-1 of Pt and 10 pg kg-1 of Rh) were made directly t o the portion of standard sample t o be attacked by aqua regia. The total recovery of Pd and Pt and the 80% recovery of Rh implies that the effectiveness of the aqua regia attack is the determining factor in the final quantitative determination of Pt and Rh.An analytical approach in three steps for the quantitative determination of Pd, Pt and Rh in stream and lake sediment samples by electrothermal atomic absorption spectrometry is suggested. Keywords: Precious metal; mineral exploration; molecular recognition ligand; extraction; electrothermal atomic absorption spectrometry Geochemical exploration projects based on the sampling of humus and of stream and lake sediment require that thousands of samples be taken. These samples are then analysed for the elements sought. Few projects concentrate on determining the platinum group elements (PGEs: Ir, Os, Pd, Pt, Rh and Ru), because of the requirement that the analysis of different types of geochemical samples for the PGEs be inexpensive and representative, while ensuring sufficiently low detection limits (a few ppb, or pg kg-I).The techniques that best meet these three criteria essentially involve the following procedure: attack on the matrix; extraction of one or more PGEs from the material resulting from the attack on the matrix; and quantitative determination of the PGEs using instrumentation sensitive to these elements. Two analytical approaches are possible, involving either a total or a partial attack on the matrix. In a total attack on the matrix, the matrix destruction and PGE extraction steps can be accomplished by lead fire assay (for a review see references 1 and 2). Lead fire assay is a reliable, inexpensive method which is routinely used in many commercial and governmental laboratories. In its application to PGEs, it allows the extraction of Pd, Pt and Rh from the matrix.In order to obtain detection limits of a few ppb, these elements are usually determined by electrothermal atomic absorption spectrometry (ETAAS)3 or by inductively coupled plasma mass spectrometry.4 It is often preferable to perform a partial attack on the matrix in order to enhance the geochemical signal. Identifica- tion of the phase or phases containing the trace elements concerned (amorphous iron oxides, clays, crystalline lattice of a specific mineral, etc.) is achieved by applying sequential extraction to the samples (for a review see reference 5 ) . Enhancement of geochemical signals can then be obtained by selectively dissolving the phase or phases containing the target elements, attacking the samples with a mixture of specific acids.For the PGEs, the extraction solution contains part of the PGEs contained in the sample and many trace elements, such as Cu, Zn, Ni and Cr. As the instrumental techniques used in measuring PGEs are subject to several sources of interference , particularly during determinations by ETAAS, the PGEs must be extracted from the solution resulting from the attack. This extraction is usually carried out by coprecipi- tation of the PGEs with Te. Extraction by coprecipitation with Te consists of adding a Te solution to the solution resulting from an acid attack on a given geological material; subsequent addition of SKI2 reduces the Te to its metallic form, which causes the PGEs to precipitate out with it. The precipitate is then filtered, dissolved in aqua regia [HCI-HN03 (3 + l)] and analysed by ETAAS.6 However, some major problems substantially reduce the effectiveness of this approach in extracting PGEs. For example, the Te precipitates not only the PGEs, but also many trace elements,' and this can cause interference during analyses by ETAAS.More recently, interference from the presence of Cu in the precipitate has been identified during the determination of Au by ETAAS.8 Given the potential for1020 ANALYST, OCTOBER 1991, VOL. 116 numerous sources of interference during extraction by copre- cipitation with Te, it was decided to consider an approach based on the use of molecular recognition ligands.9 The aim was to evaluate the potential, for extracting PGEs from an acidic solution, of molecular recognition ligands specially designed for this purpose (Superlig, from IBC Advanced Technologies). The ability of these ligands to determine quantitatively PGE content values of a few nanograms which corresponds, when extrapolated to a solid sample of about 5 g, to concentrations at the pg kg-1 level.In order to demonstrate the effectiveness of the Superlig ligands in extracting low PGE concentrations from acidic solutions, a series of recovery tests was performed on: (1) a 0.1 rnol dm-3 synthetic solution of HCI containing 7.1 pg kg-1 of Pd, 36 pg kg-1 of Pt and 7.1 pg kg-1 of Rh, to which had been added 14.3 pg kg-1 of each of the following metals: Cu, Cr, Co, Ni, Pb and Zn; and (2) a solution resulting from an aqua regia attack on a standard sample of a stream sediment which had previously been analysed for Pd, Pt and Rh, the PGEs considered in this work. The analyses had been carried out using an atomic absorption unit equipped with a graphite furnace.Experimental Reagents All of the reagents used were of analytical-reagent grade (ethylenediamine, metallic Pd, metallic Pt, metallic Rh and thiourea) and were obtained from Fluka (Buchs, Switzer- land). Primary standard solutions of Pd and Pt were prepared by dissolving the pure metals in aqua regia, followed by evaporation until nearly dry and redissolution in 25% v/v HCI ( 5 mg kg-1 of Pd and Pt). As Rh in its metallic form is not soluble in aqua regia, an Rh salt [(NH4)3Rh6] was dissolved directly in 25% v/v HCI. Before each operation, all necessary glassware was washed with aqua regia and rinsed in de- mineralized water.The molecular recognition ligands were obtained directly from Superlig (IBC Advanced Technolo- gies, Orem, UT, USA). These ligands are distributed under the names Superlig 201 (SL-201), Superlig 202 (SL-202) and Superlig 203 (SL-203). The size of the particles comprising the ligands, according to the information provided by the sup- plier, varied between 60 and 140 mesh. Instrumentation The Pd, Pt and Rh concentrations were measured with a Varian Spectra AA-30 atomic absorption unit equipped with a Model GTA-96 graphite furnace. Pyrolytic graphite coated graphite tubes were used, also from Varian. The eluent gas was argon-hydrogen (95 + 5 ) . The volume of injected solution was 20 pl.A Varian automatic sampler was used. The deuterium background noise corrector was not employed in the measurements. Choice of Analysis Medium The quantifications were carried out directly in the eluents used to extract Pd, Pt and Rh from the ligands. Hence Pd and Pt were measured in a 2.0 rnol dm-3 thiourea solution, while Rh was determined in a 2.0 rnol dm-3 ethylenediamine solution (adjusted to pH 1 with HCI). Optimization of Operating Conditions for the Graphite Furnace The use of a graphite furnace involves the optimization of the parameters which control the thermal treatment each sample undergoes in the furnace (drying, ashing and atomization). These parameters were optimized by using: for Pd, a 0.1 rnol dm-3 thiourea solution with 50 pg kg-1 of Pd added; for Pt, a 2.0 rnol dm-3 thiourea solution with 250 pg kg-1 of Pt added; and for Rh, a 2.0 rnol dm-3 ethylenediamine solution (pH 1) with 50 pg kg-1 of Rh added.Drying temperature and time The drying temperature is a function of the solvent boiling- point. This temperature is usually set at a few degrees below the boiling-point and the time needed is generally 1 s per microlitre of solution injected. The temperature is then increased to slightly above the boiling-point for a few seconds to ensure that evaporation is complete. As this step is not a determining factor in the quality of the analysis, attention was focused on qualitative optimization by observing the behav- iour of the solvent and adjusting the drying temperature and time to eliminate spatters. The optimum conditions for drying are summarized in Table 1 (steps 1 and 2). Ashing temperature and time During this stage, a problem was encountered related to the presence of ashes from calcination of the thiourea and ethylenediamine solutions.In order to avoid this undesirable effect, calcination times that were sufficiently long to ensure complete combustion of the organic solvents were chosen. The conditions arrived at for calcination temperatures and times are given in Table 1 for the three elements determined (steps 3, 4 and 5 for Pd and Pt, and steps 3 and 4 for Rh). Atomization temperature and time The final stage in the thermal treatment of the sample involves determination of the minimum temperature that will allow atomization of as much of the element to be determined as possible, in order to avoid premature deterioration of the graphite tubes.The optimum temperatures for atomizing Pd, Pt and Rh are indicated in Table 1. Calibration The graphite furnace was calibrated using, for Pd, a 2.0 rnol dm-3 thiourea solution containing 50 ppb of Pd; for Pt, a Table 1 Optimum parameters related to operation of the graphite furnace in determining Pd, Pt and Rh content. Sample volume, 20 p.1. Lamp currents for Pd, Pt and Rh: 5 , 10 and 5 mA, respectively Step 1 2 3 4 5 6 7 8 * Readout command. Pd TemperaturePC Time/s 95 20 115 5 175 10 250 30 1000 10 2500 1.2" 2500 3* 2500 1 .o Pt TemperaturePC Time/s 95 20 115 5 175 10 250 30 lo00 10 2700 1.3* 2700 2* 2800 1 Rh TemperaturePC 100 130 360 1000 2600 2600 2600 - Time/s 4 70 35 10 1.5* 2.0" 1 .o -ANALYST, OCTOBER 1991, VOL.116 1021 0.40 0.30 0.20 0.10 al 5 0 10 20 30 40 50 60 0.10 1 / i / I 0 100 200 300 Concentration/pg kg- Fig. 1 graph for Pt. The analyses were carried out by ETAAS (a) Calibration graphs for A, Pd; and B, Rh. (b) Calibration 2.0 rnol dm-3 thiourea solution containing 250 ppb of Pt; and for Rh, a 2.0 rnol dm-3 ethylenediamine solution (pH 1) containing 50 pg kg-1 of Rh. The calibration graphs are shown in Fig. 1. General Description of the Molecular Recognition Ligands and Choice of Column Used Three different types of ligand were obtained from IBC Advanced Technologies: SL-201, SL-202 and SL-203. The operating principle of these ligands is based on the selective fixation of certain ions as a function of their ionic radii, the form of their orbitals and their chemical preferences.9 The base metals, the rare earths and the major elements are not fixed on SL-201, SL-202 or SL-203.9 The operating principle of these ligands is based on the permanent fixation of macrocycles on a silica gel by covalent bonding.10.11 The chelation of metallic cations by macrocycles has been known for several years.12 Methods of macrocycle synthesis allowing the design of chelating agents which are very specific to certain cations have been developed. Macro- cyclic compounds have been used to separate metallic ions in mixtures by ion exchange, followed by extraction with solvents and purification on systems of liquid membranes. 13-16 Work carried out by IBC Advanced Technologies, in conjunc- tion with a team of workers from Brigham Young University (Provo, UT, USA), showed that the fixation of macrocycles on a silica gel by covalent bonding makes possible a separation system that could easily be made commercially viable.17-19 Owing to the economic interest in the purification of precious metals, IBC Advanced Technologies has used its expertise to synthesize macrocyles which allow isolation of several of these metals.These ligands have been used successfully in a number of pilot plants, demonstrating their ability to purify precious metals on a large scale.18.19 IBC Advanced Technologies does not wish to divulge the precise structures of the macrocycles used to purify Pd, Pt and Rh in order to protect the proprietary streams and the refining process (reference 18, p.382). Silica gel surface To? -0 \ -0 / - o - s i - o ~ o 0 1) Fig. 2 bonded18 to a silica gel General structure of a macrocycle (18-crown-6) covalently Regarding the separation of less marketable metals, the complete structure of a macrocycle allowing the separation of Ag+, Hg2+, T1+ and Pb2+ has been published.17 As an illustration, the general structure of a macrocycle (18- crown-6) covalently bonded to a silica gel18 is shown in Fig. 2. An explanatory guide supplied by IBC Advanced Techno- logies indicates how to use these ligands in order to isolate Pd, Pt and Rh. The user's guide provided by IBC Advanced Technologies also describes a method for separating Ir, Ru and Au from solutions with an ion load. However, the method for Ir, Ru and Au was not tested in this work.The guide for Pd, Pt and Rh gives three general steps. (1) The SL-201 ligand allows fixation of Pd" + PdIV, Au"' and Ag'. The base metals and other impurities are washed out with 0.1 rnol dm-3 HCI. Various eluents provide for the elution of Ag' and Au"'. Finally, Pd" + PdIV is eluted with a 2.0 rnol dm-3 thiourea solution. (2) The SL-202 ligand fixes Rh"'; the impurities are washed out with 0.1 rnol dm-3 HCI and the RhlL1 is extracted with a 2.0 rnol dm-3 ethylenediamine solution (pH 1). (3) The SL-203 ligand captures Pt" + PtJV, washing with 0.1 mol dm-3 HCl cleans the column of impurities and Pt" + Pt" is recovered by extraction with a 2.0 rnol dm-3 thiourea solution. In order to check the application of this guide to the separation of Pd, Pt and Rh at amounts of the order of 1 ng, preliminary tests were conducted using a synthetic solution containing a mixture of 50 yg kg-1 of Pd (1 ml), 250 pg kg-1 of Pt (1 ml) and 50 pg kg-1 of Rh (1 ml).It quickly became apparent that the SL-201 column was retaining not only the Pd" + PdIV, but also 40% of the PtIV. Moreover, the extraction of Pt" required 10 ml of thiourea, a volume which involved the addition of an evaporation step to obtain adequate sensitivity in the graphite furnace. The SL-202 ligand presented fewer problems, separating not only Rhlll, but also PdIV and PtIV. This behaviour of SL-202 is explained by the fact that fixation of Rhlll, PtIV and PdIV on this ligand depends on the HCl concentration. Fixation of Rh"' is optimum at HCl concentrations greater than 4 rnol dm-3, whereas fixation of Pd" and PttV is at its maximum at an HCI concentration of 0.1 mol dm-3. Hence, depending on the concentration of HCl used in passing through SL-202 and of the eluent, it appeared to be possible to isolate Rh'" in one fraction and PdIV and PtIV in another.As it has been shown that, at a ratio of 1 : 100, Pd : Pt and Pt : Pd do not interfere in graphite furnace measurements,~9 SL-202 was used in this work to extract Pd", Pt'" and Rh"' from an acidic solution with an ion load. With regard to the oxidation states of the complexes fixed by SL-202, these are essentially those that result from the usual attacks performed on lake and stream sediment samples (HN03 or aqua regia). Although the fixation of Pd" and Pt" on SL-202 was not verified, the information provided by IBC Advanced Technologies sug- gests that Pd" and Pt" will also be fixed by SL-202.It is1022 ANALYST, OCTOBER 1991, VOL. 116 important to realize that the SL-202 material is destroyed by alkaline solutions of pH >11, and also by HF and aqua regia. Configuration and Operation of the Column Containing the SL-202 Material The synthetic solution containing 50 yg kg-1 of Pd, 50 yg kg-* of Rh and 250 pg kg-1 of Pt was used to optimize the operation and the parameters related to the diameter, height and amount of SL-202 material in the column for effective extraction of Pd, Pt and Rh. The first experiments were carried out directly in the column provided by IBC Advanced Technologies (diameter, 2 cm; height, 1 cm, SL-202 material, 1 g; rate, about 1 ml s-1).With this arrangement, the recovery of Pd, Pt and Rh was unsatisfactory (about 20%), hence it was abandoned. Experiments were then attempted on columns with much smaller diameters in which 0.8 g of SL-202 was placed. In the experimental assembly used here, the column had an i.d. of 0.8 cm and contained 0.8 g of SL-202 over a height of 50 mm. Two glass wool plugs held the ligand in place (Fig. 3). This configuration allowed the extraction of 100 k 10% of the Pd, Pt and Rh [number of analyses (n) = 61. The procedures developed as a result of the preliminary tests are described below. Washing of the Column The following steps are involved. (1) For new material, rinse the column with a few millilitres of ethanol to clean the SL-202.Wash out trace amounts of ethanol with water. (2) Wash the SL-202 with 10 ml of HCI (0.1 rnol dm-3). In the arrangement used here (Fig. 3), 10 or 50 ml syringes were used to inject the various solutions directly into the column. Extraction of Pd and Pt From the Loaded Solution and Elution The steps are as follows. (3) Inject a 0.1 mol dm-3 solution of HCl* containing various elements, including Pd and Pt, onto the SL-202 (rate, 0.5 ml s-1). (4) Rinse the column with 50 ml of 0.1 rnol dm-3 HC1 (rate, 1 ml s-1) to remove elements other than Pd and Pt. (5) Elute Pd and Pt with two 2 ml injections of c Syringe (10 or 50 ml) Rubber tube (connection) I (-Glass tube Glass wool <!, SL-202 m Test-tube Fig. 3 Assembly used for the extraction of Pd, Pt and Rh (for details see text) * The various chemical components from which the separation of Pd and Pt is desired must be solubilized in 0.1 mol dm-3 HCl; this condition is necessary for complete fixation of Pd and Pt on SL-202. Under Results and Discussion, an attack based on the use of aqua regia is described which, for geological materials, provides a final solution of 0.1 mol dm-3 in HC1 ready for extraction.2.0 mol dm-3 thiourea (rate, 0.5 ml s-1; 5 s between injections), ensuring that the thiourea solution is completely eliminated from the column. Elution of Pd and Pt is now complete. (6) Combine the two thiourea extractions and determine the Pd and Pt content by ETAAS. (7) Wash out trace amounts of thiourea with 30 ml of water. The column is now ready for another sample.Extraction of Rh From the Loaded Solution and Elution The final series of steps are as follows. (8) Acidify the 0.1 rnol dm-3 HCl solution from which Pd and Pt have been extracted with concentrated HCl to obtain a molarity 3 4.0 rnol dm-3 HCl. (9) Prepare the column with an injection of 10 ml of 0.1 mol dm-3 HCl. (10) Inject the solution onto SL-202 (rate, 0.5 ml s-1). (11) Rinse the column with 50 ml of 4.0 rnol dm-3 HCl. (12) Extract Rh with two 2 ml injections of 2.0 mol dm-3 ethylenediamine (pH 1), ensuring that the ethylene- diamine solution is completely eliminated from the column. (13) Combine the two extracts from ethylenediamine and determine the Rh content by ETAAS. Results and Discussion In order to check the procedure described above, a synthetic solution containing Pd, Pt, Rh and various metals, and a standard sample of a stream sediment (SRH-1) that had previously been analysed by different methods for Pd, Pt, Rh and several other trace elements were analysed.In addition, standard additions of Pd, Pt and Rh were introduced directly into the SRH-1 sample. The recovery of these standard additions was evaluated. Verification of the Procedure Using the Synthetic Solution A synthetic solution was prepared of 0.1 mol dm-3 HCI (7 ml) containing 7.1 pg kg-1 of Pd, 36 pg kg-1 of Pt, 7.1 pg kg-1 of Rh and 14.3 mg kg-* of each of Ni, Cr, Cu, Fe, Pb and Zn (individual ratios of 1 : 14000 for Pd and Pt to total metals, and of 1 :2800 for Rh). The results of the application of the procedure described above are given in Table 2.They show that the recoveries of Pd, Pt and Rh from the synthetic solution are, respectively, 104 k 8 (n = 6), 108 k 9 (n = 6) and 98 k 10% (n = 6). No apparent problem was encountered during the operation. The ligand SL-202, therefore, allows complete extraction of Pd, Pt and Rh from the synthetic solution used. However, tests based on the analysis of synthetic solutions are always imperfect, as these solutions do not reflect the geological reality of stream sediment samples. Verification of the Procedure Using a Standard Sample of Stream Sediment In order to check the analytical approach used here on samples that are realistic from a geochemical exploration perspective, a stream sediment sample (SRH-1) which had been previously characterized by lead fire assay for Pd, Pt and Rh content was used.Information on the sampling, prepara- tion, homogeneity and content values for Pd, Pt, Rh, Ir, 0 s and Ru of SRH-1 has previously been reported.20 Additional information on the mineralogy of the ore from which SRH-1 originated is available elsewhere.21 Portions of SRH-1 can be obtained from the Ministhre de I’Energie et des Ressources du Qudbec; SRH-1 is essentially a stream sediment sample taken close to a Cr-PGE-Ni deposit in the ophiolitic belt of the Appalachians in Quebec. Its Pd, Pt and Rh concentrations are given in Table 3, together with those of other trace elements, including Ni and Cr. It should be noted that zones high in Ni and Cr are preferred target areas in geochemical exploration for PGEs.It is therefore not surprising that SRH-1 contains elevated amounts of Ni and Cr.ANALYST, OCTOBER 1991, VOL. 116 1023 Table 2 Recoveries of Pd, Pt and Rh from various solutions using SL-202 Element Pd Pd Pd Pt Pt Pt Rh Rh Rh Solution I t 27 3** 1 2 3 1 2 3 Concentration present*/ kg- 7.1$ 11 f 4)) (50) 24 k 2 t t (6) 36I 101 +- 2411 (50) 130 h 7 (7) 7.1I (6) 9 f 511 (50) l o f t Concentration measured*/ kg- 7.4 k 0.6 (6)O 14 f 2 (6) 26; 23 (2) 38 f 2 (6) 125; 123 (2) 7.0 k 0.7 (6) 8; 8.2 (2) 80 f 7 (7) 0 (6) Recovery* (Yo) 104 k 8 (6) Indistinguishable Indistinguishable 108 * 9 (6) Indistinguishable Indistinguishable 98 f 10 (6) 80; 82 (2) 0 (6) * Mean f 1 standard deviation. t Solution 1: 0.1 rnol dm-3 HCI containing 7.1 pg kg-1 of Pd, 36 pg kg-1 of Pt and 7.1 pg kg-1 of Rh to which has been added 14.3 $ Concentration of synthetic solutions prepared from standards; the error therefore depends on the dilutions.As it is small relative to that 9 The values in parentheses are the number of analyses. f Solution 2: a solution resulting from an attack by aqua regia on 5 g samples of SRH-1 (for details see text). 11 See Table 3 and reference 11. ** Solution 3: solution 2 with the addition of 10 pg kg-1 of Pd, 50 pg kg-1 of Pt and 10 pg kg-1 of Rh. tt Concentration and error corresponding to the sum of the concentration measured in solution 2 and the standard addition. Example for Pd: mg kg-1 of each of the following metals: Co, Cr, Cu, Ni, Pb and Zn. from ETAAS measurements, it is not indicated. 24 f 2 comes from (14 f 2) + 10.The standard sample SRH-1 was first attacked by aqua regia. Portions (5 g) of SRH-1 were weighed in 600 ml beakers To these were added 30 ml of concentrated HC1, followed by 2 ml additions of 70% v/v HN03 up to 10 ml. The mixtures were agitated under heat (70°C) for 90 min and then evaporated until nearly dry. The sides of the beaker were washed with a few millilitres of de-mineralized water. The mixtures were filtered under gravity using Whatman 44 filter-paper and the filtrates were evaporated until nearly dry. The residues were dissolved in 10 ml of concentrated HCl, evaporated until nearly dry and redissolved in 10 ml of 10% v/v HCl. Finally, the solutions were diluted with de-mineral- ized water to obtain an HCI molarity between 0.1 and 0.2 rnol dm-3 (adztion of about 250 ml of de-mineralized water).The solutions were then passed through the column containing SL-202. The results for the recovery of Pd, Pt and Rh from SRH-1 using SL-202 are given in Table 2. The results for Pd and Pt are similar to those obtained from the synthetic solution. Two 2 ml injections (0.5 ml s-1,5 s between injections) of 2.0 mol dm-3 thiourea provide complete elution of the Pd and Pt fixed in the column. A comparison of the average content values obtained using the proposed method (Pd: 14 k 2 pg kg-1, n = 6; Pt: 80 k 7 pg kg-1, n = 7) with those obtained by lead fire assay (Pd: 11 k 4 pg kg-1, n = 50; Pt: 101 k 24 pg kg-1, n = 50), taking analytical errors into account, showed that the data sets were indistinguishable. The presence of Rh was not detected in the extractions from ethylenediamine.The standard sample SRH-1 contains 9 pg kg-1 of Rh (Table 3), and the detection limits achieved by ETAAS (see below) make its determination feasible. The inability to detect Rh raises some doubts about the effectiveness of aqua regia attacks or about the recovery of Rh using SL-202. In the first situation, an aqua regia attack would not allow the solubilization of Rh. It would therefore be normal to find no Rh in the extraction solution. In the second situation, problems related to the processes of fixing Rh on SL-202 in a realistic context would cause a total loss of this element. In order to investigate these two possibilities, standard additions were made directly to SRH-1 at the very start of operations.A 2 ml volume of a 0.1 mol dm-3 solution of HCI containing 10 pg kg-1 of Pd, 50 pg kg-1 of Pt and 10 pg kg-1 of Rh was therefore added to two 5 g portions of SRH-1 and the aqua regia attack performed. For the samples combining SRH-1 and the synthetic additions, taking ana- lytical errors into consideration, the data sets for Pd and Pt are indistinguishable. For Rh, the recovery is 80%. Table 3 Content of Pd, Pt, Rh, other elements and loss on ignition in SRH-1 (reference 10) Element Method As INAAt Ba ICP-AESI Cr ICP-AES Fe ICP-AES Mn ICP-AES Ni ICP-AES CO ICP-AES CU ICP-AES K ICP-AES MO ICP-AES P ICP-AES Pb ICP-AES Pd PbO-ICP-AESO Pt PbO-ICP-AES Rh PbO-ICP-AES Zn ICP-AES Loss on ignition n* 20 20 20 20 20 20 20 20 20 20 20 20 50 50 50 20 20 Mean * 1 standard deviation 8.2 k 0.6 mg kg-l 47 k 1 mg kg-I 67 h 2 mg kg-l 472 k 19 mg kg-l 9.5 k 0.5 mg kg-I 2.4 h 0.07% m/m 0.08 k 0.005% m/m 1136 f 116mgkg-l 14 f 1 mg kg-l 861 k 20 mg kg-I 405 f 13 mg kg-l 29 _+ 2 mg kg-l 11 f 4 pg kg-1 101 f 24 pg kg-I 9 k 5 pg kg-l 67 f 2 pg kg-l 10 f 2% m/m * n = number of analyses.t INAA = instrumental neutron-activation analysis. $ ICP-AES = analysis by inductively coupled plasma atomic emission spectrometry of the solution resulting from a hot HN03 attack on SRH-1. 9 PbO-ICP-AES = lead fire assay followed by inductively coupled plasma atomic emission spectrometry. The ligand SL-202 therefore allows the complete extraction of Pd and Pt from loaded solutions, provided that the attack used on the geological material causes these elements to dissolve.It is more difficult to ascertain the recovery of Rh with SL-202 from loaded solutions, in view of the solubility problems of Rh in the acids used (aqua regia) in this study. In spite of these dissolution problems, the recovery of Rh with SL-202 from loaded solutions appears to be at least 80%. The detection limits for Pd, Pt and Rh were evaluated using three synthetic solutions: one of 2.0 mol dm-3 thiourea with 1 pg kg-1 of Pd added, a second of 2.0 mol dm-3 thiourea with 10 pg kg-1 of Pt added and a third of 2.0 mol dm-3 ethylenediamine with 3 pg kg-1 of Rh added, all concentra- tions being near the expected detection limits. Each of these solutions was analysed ten times for their Pd, Pt and Rh content using injections of 20 pl.The detection limits (solution value) were calculated by taking 3SD (standard deviation) of the absorbances obtained and converting these into concen-1024 ANALYST, OCTOBER 1991, VOL. 116 trations. The detection limits are 0.9,3 and 1.5 pg kg-1 for Pd, Pt and Rh, respectively. By relating these content values to the mass of the sample analysed (5 g) and the amount of eluent needed for extraction (4 ml), detection limits (solid sample data) of 0.7, 2.4 and 1.2 pg kg-1 for Pd, Pt and Rh, respectively, were obtained. The total capacity of SL-202 for the retention of PGEs was not determined. According to Ruckman,g it is 0.08 g per gram of Superlig material for a given element. This appears to be adequate considering that, as in this work, approximately 0.6 g of SL-202 material is used to fix PGE amounts of the order of nanograms.When the experiments had been completed, one of the ligands used had been subjected to about 30 cycles of fixation, elution and washing without any noticeable reduction in effectiveness. Ruckman9 has reported that the Superlig materials can withstand 200 cycles of fixation, elution and washing without losing their effectiveness. The operations performed here using SL-202 indicated that this type of extraction is rapid, simple and inexpensive. Compared with the usual lead fire assay technique, the main advantage of a technique based on acid attack is its versatility. Indeed, various acid mixtures can be used to release Pd, Pt and Rh from specific geochemical phases (amorphous oxides, organic materials and crystalline lattices).Conclusions The recovery, using SL-202, of Pd, Pt and Rh from a 0.1 mol dm-3 solution of HC1 containing 7.1 pg kg-1 of Pd, 36 pg kg-1 of Pt and 7.1 pg kg-1 of Rh, to which six different metals have been added (14.3 mg kg-1 each of Cu, Cr, Co, Ni, Pb and Zn), is 100%. The elements Pd and Pt are recovered simultaneously in 2.0 mol dm-3 thiourea (two 2 ml injections), while Rh is recovered in 2.0 mol dm-3 ethylenediamine (two 2 ml injections). When the standard sample SRH-1 is attacked by aqua regia, taking analytical errors into account, the data sets for Pd and Pt are indistinguishable. No trace of Rh could be detected following the attack by aqua regia on SRH-1. In order to demonstrate that the incomplete recovery of Rh depends on the effectiveness of the aqua regia attack and not on the fixing and elution processes related to the use of SL-202, standard additions were made to SRH-l(10 pg kg-1 of Pd, 50 pg kg- * of Pt and 10 pg kg-1 of Rh).For Pd and Pt, data indicate complete recovery of the additions made and of the amounts released from SRH-1 by the aqua regia attack. For Rh, the recovery is 80%. In view of the ability of SL-202 to recover Pd, Pt and Rh, it would appear feasible to use acid attacks on stream sediment samples for determining the content of Pd, Pt and Rh, which are extracted from the solution by SL-202. Subsequent analysis by ETAAS for Pd, Pt and Rh achieves sufficiently low detection limits (0.7, 2.4 and 1.2 ppb for Pd, Pt and Rh, respectively) in samples of 5 g to justify the promotion of Pd, Pt and Rh content values in samples of stream and lake sediment as a geochemical exploration tool in the search for PGEs.However, it is strongly recommended that, at the start of any exploration project, an evaluation be carried out of the effectiveness of the chosen acid attack in dissolving Pd, Pt and Rh. This appears to be particularly important for Rh. We thank Normand TassC for his critical revision of an earlier version of this manusFript. Financial support was provided by the Ministkre de 1’Energie et des Ressources du QuCbec (Section de la Gkochimie et de la GCophysique) and by the Natural Sciences and Engineering Council of Canada (operat- ing grant to M. B.). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 References Potts, P.J., A Handbook of Silicate Rocks Analysis, Blackie, Glasgow, 1987, p. 622. Bergeron, M., CIM Bull., 1990, 83, 108. Page, N. J., Myers, J. S., Haffty, J., Simon, F. O., and Aruscavage, P. J., Econ. Geol., 1980, 75,907. Date, A. R., Davis, A. E., and Cheung, Y. Y., Analyst, 1987, 112, 1217. Forstner, U., Hydrobiologia, 1982, 91, 269. Fryer, B. J., and Kerrich, R., At. Absorpt. Newsl., 1978,17,4. Signolfi, G . P., Gordini, C., and Mohamed, A. H., Geostand. Newsl., 1984, 8, 25. Hall, G. E. M., Pelchat, J. C., and Dunn, C. E., J. Geochem. Explor., 1990,37, 1. Ruckman, J. H., in Proceedings of the Thirteenth International Precious Metals Institute Conference, Montrkal, Canada, ed. Harris, B., International Precious Metals Institute, New York, 1989, p. 381. Bradshaw, J. S., Bruening, R. L., Krakowiak, K. E., Tarbet, B. J., Bruening, M. L., Izatt, R. M., and Christensen, J. J., J. Chem. SOC., Chem Commun., 1988, 812. Bradshaw, J. S., Izatt, R. M., Christensen, J. J., and Bruening, R. L., US Pat. Appl. 093 543, September, 1987. Izatt, R. M., Bradshaw, J. S., Nielsen, S. A., Lamb, J. D., and Christensen, J. J., Chem. Rev., 1985, 85,271. Izatt, R. M., Clark, G. A., Bradshaw, J. S., Lamb, J. D., and Christensen, J. J., Sep. Purif Methods 1986, 15, 21. Dulyea, L. M., Fyles, T. M., and Whitfield, D. M., Can. 1. Chem., 1984,62, 498. Bartsch, R. A., Czech, B. F., Kang, S. I., Stewart, L. E., Walkowiak, W., Charewicz, W. A., Heo, G. S., and Son, B., J. Am. Chem. SOC., 1985, 107,4997. Izatt, R. M., Lindh, G. C., Clark, G. A., Nakasutsuji, Y., Bradshaw, J. S., Lamb, J. D., and Christensen, J. J., J. Membr. Sci., 1987, 31, 1. Izatt, R. M., Bruening, R. L., Bruening, M. L., Tarbet, B. J., Krakowiak, K. E., Bradshaw, J. S., and Christensen, J. J., Anal. Chem., 1988, 60, 1826. Bruening, R. L., Izatt, S. R., and Griffin, D. L., in Precious Metals 1990, Proceedings of the 14th International Precious Metals Conference and Exhibition, Sun Diego, USA, ed. Corrigan, D. A. , International Precious Metals Institute, New York, 1990, p. 97. Haines, J., and Rob&, R. V. D., S. Afr. Tydskr. Chem., 1984, 37, 121. Bergeron, M., and Beaumier, M., Geostand. Newsl., 1990, 14, 461. Gauthier, M., Corrivaux, L., Trottier, L. J., Cabri, J., Laflamme, J. H. G., and Bergeron, M., Miner. Deposita, 1990, 25, 169. Paper 0f03795H Received August 21st, I990 Accepted May 13th, 1991

ISSN:0003-2654    

DOI:10.1039/AN9911601019

出版商:RSC    

年代:1991

数据来源: RSC