ISSN:0003-2654    

DOI:10.1039/AN98409FX009

出版商:RSC    

年代:1984

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98409BX011

出版商:RSC    

年代:1984

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98409BP017

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST, MARCH 1984, VOL. 109 189 EDITORIAL SAC 83-Edinburgh, Scotland, July 17-23,1983 As a consequence of the publication of the SAC 83 Confer- ence Handbook as the June 1983 issue of Analytical Proceed- ings (1983,20,223-340) and the appearance of the abstracts of conference papers in that issue, it was decided that to avoid duplication, publication of the usual extended summaries of conference papers in Analytical Proceedings would not be adopted. Instead, authors were invited to submit the full text of their papers presented at the conference for consideration for publication in a Special SAC 83 Issue of The Analyst. This current, “bumper,” issue of The Analyst is the result, and the Analytical Editorial Board hopes you will find it informa- tive and a valuable insight into the wide range of topics and new developments presented at the conference. Of the 210 papers presented in Edinburgh, 67 were submitted for publication in The Analyst and 46 of these appear in this issue.A number have been rejected by the normal refereeing procedure of The Analyst, and a few, which were submitted late or have been delayed, will appear in later issues of the journal. Most of the delays are due to the time taken by authors to modify their manuscripts to comply with referees’ comments and are quite normal with a proportion of submitted manuscripts in any primary journal. The papers in this issue include the four Plenary Lectures presented at SAC 83, each of which was very well received by delegates and contributed an excellent and balanced overview of developments in specific fields of analytical research.These papers: “Recent Developments in Fluorescence and Chemilu- minescence Analysis,” by J . N. Miller (p. 191); Capillary Separation Methods; a Key to High Efficiency and Improved Detection Capabilities,” by M. V. Novotny (p. 199); “Design and Application of Neutral Carrier-based Ion-selective Elec- trodes,” by W. Simon et al. (p. 207); and “Continuum Source Atomic-absorption Spectrometry: Past, Present and Future Prospects,” by T. C. O’Haver (p. 211), offer a mini-review of their respective topics and we believe they make a substantial contribution to the literature, equivalent to the major contribution made by the authors’ lectures to the success of the conference. The Board is particularly grateful to the Plenary Lecturers for their willingness and efforts in produc- ing their manuscripts promptly for this issue of the journal.The SAC 83 conference was sponsored both by IUPAC, the International Union of Pure and Applied Chemistry, and FECS, The Federation of European Chemical Societies. IUPAC sponsorship normally carries with it the requirement for publication of the Plenary Lectures in the IUPAC journal, Pure and Applied Chemistry. However, on this occasion IUPAC waived their right to publish the Plenary Lectures and the Conference Organisers are very grateful for this helpful decision, and for the general support given by both IUPAC and FECS. Of the large number of contributed papers presented both orally and as posters at the conference, this issue contains approximately one fifth.In view of the fact that many conference papers are not suitable for publication in the same format, and that a number of workers will always prefer to publish their work elsewhere, this proportion is roughly in line with that anticipated by the Editorial Board when it initiated this venture. The Board would like to thank all authors who have submitted their manuscripts to The Analyst, and in most instances would also thank them for their efforts to meet the deadlines essential for the publication of a Special Issue of a journal. We hope our subscribers find the publication of this Special Issue a valuable contribution to the wider dissemination of knowledge and research presented in Edinburgh last July. Non-subscribers may like to know that copies of this issue are also available for personal purchase from the Royal Society of Chemistry Distribution Centre, Blackhorse Road, Letch- worth, Hertfordshire, SG6 1HN (price $35).Many individuals have contributed to the rapid publication of this issue, but the Board is particularly conscious of the extra effort required from the editorial staff in the production of such a large issue of the journal. This is the second Special Issue of The Analyst to be published and follows that in February 1983 which covered papers presented at the First Biennial National Atomic Spectroscopy Symposium. A num- ber of others are already planned by the Analytical Editorial Board, and these will appear in future issues. The Executive Committee of SAC 86, to be held in July 1986 at the University of Bristol, has recently met for the first time. At this stage it seems very likely that they will adopt the same policy with regard to the Conference Handbook and publication of papers as that adopted for SAC 83, and we look forward to the publication of an even larger Special Issue on that occasion. J . M . Ottaway Chairman, Analytical Editorial Board

ISSN:0003-2654    

DOI:10.1039/AN9840900189

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST. MARCH 1984 VOL. 104 191 Recent Developments in Fluorescence and Chemiluminescence Analysis Plenary Lecture James N. Miller Department of Chemistry Loughborough University of Technology Loughborough Leicestershire, L E I I 3TU UK Recent developments in photoluminescence and chemiluminescence spectroscopy are summarised with particular reference to methods for improving the selectivity of luminescence analyses. The acquisition and manipulation of contour spectra and the problems and benefits of generally available corrected fluorescence spectra are discussed. Fluorescence and chemiluminescence immunoassays are evaluated and some probable future developments in these areas are indicated. Keywords Fluorescence analysis; chemiluminescence analysis This paper presents a brief review of the recent advances in fluorescence and chemiluminescence analysis methods.This dynamic area of analytical science is in the midst of a period of considerable expansion new instrumentation and techniques complement new luminescent reagents and reactions to produce many dividends in terms of enhanced sensitivity, selectivity and convenience. The present survey inevitably represents a personal selection of those advances which seem to be of the greatest theoretical interest or practical promise. Many of the areas which receive scant treatment here have been well reviewed elsewhere and appropriate references to such reviews are given. Fluorescence Spectra To an analyst the outstanding feature of luminescence methods is their sensitivity.With appropriate compounds picogram concentrations can be detected using fluorimetric and phosphorimetric methods and much lower levels still are detectable using chemiluminescence and bioluminescence methods.' The principal applications areas of all these methods are in biochemical and environmental analyses, where such sensitivity is especially important. The basis of this sensitivity is well known (Fig. 1). In the analysis of a very dilute solution absorptiometry involves detecting a small change in a large light intensity a very difficult thing to do; in contrast using the characteristic and still almost universal right-angled optics familiar in fluorimetry the small fluores-cence signal is detected against a background which is (ideally) zero. This difference produces a vast improvement in limits of detection even though the fluorescence spectrometer may in practice collect only a small fraction of the emitted fluores-cence .2 In chemiluminescence spectrometers which funda-mentally consist of a cuvette holder and an adjacent photo-multiplier the light gathering is very efficient and the sensitivity correspondingly greater.The availability of two monochromators in most fluores-cence spectrometers permits the collection of two simple types of spectra. Scanning the excitation monochromator at a constant emission wavelength yields the excitation spectrum; scanning the emission monochromator at a fixed excitation wavelength provides the fluorescence or emission spectrum, which occurs at longer wavelengths than the excitation spectrum (other types of fluorescence spectrum will be discussed below).Fig. 2 shows the excitation and fluorescence spectra of Lucifer Yellow VS a new and very promising fluorescent dye molecule developed by Stewart.3 The most noticeable feature of the spectra and indeed of the spectra of most organic molecules is their band width; they cover 100 nm or more. One result of this is that the apparently additional selectivity of fluorimetry is largely illusory. In theory two species with similar emission wavelengths might have different excitation wavelengths (and vice versa). In practice the spectral band widths are so large that even apparently distinct spectra overlap strongly so discrimination is difficult. This is shown very clearly by comparing Lucifer Yellow VS with ( a ) +---+n---*o-@ Detector Source Dispersion Sample # / Samde R \ Source Dispersion .'I '\* Dispersion 1 0 Detector Fig.1. Optical layouts in (a) absorption and ( b ) fluorescence spectroscopy Wavelengthhm Fig. 2. A Excitation = 429 nm) and B emission spectra (Af = 535 nm) of Lucifer Yellow VS (I) in water. In this compound R = CGHSS02CH=CH 192 ANALYST MARCH 1984 VOL. 109 T -Table 1. Corrected emission spectra. Quinine sulphate in 0.1 M perchloric acid (Acx = 350 nm) W avelengt hi nm 400 410 420 430 440 450 460 470 480 490 500 Spectrum * 1 0.170 0.359 0.586 0.792 0.940 0.999 0.982 0.897 0.782 0.659 0.541 2 0.177 0.376 0.608 0.814 0.957 0.999 0.982 0.924 0.808 0.675 0.551 * Spectrum 1 was obtained at the National Bureau of Standards (reference 5 ) and spectrum 2 in the author‘s laboratory.400 450 500 550 600 650 Wavelengthhm Fig. 3. A Excitation and B emission spectra o f (a) fluorescein and (6) Lucifer Yellow VS in water. Fluorescein he = 465 nrn; h = 535 nm. Lucifer Yellow VS hex = 429 nrn; h = 535 nm fluorescein (Fig. 3) the two dyes have similar emission wavelengths but very different excitation wavelengths. It is nonetheless very hard to choose a pair of excitation and emission wavelengths that would permit the observation of one of them without interference by the other. In practice, these spectral overlaps also have a serious effect on the limits of detection attainable in fluorimetry.In studies of a “real” multi-component sample the background signal is very appreciable because of overlapping spectra; hence the main theoretical advantage of fluorescence spectrometry is largely lost and limits of detection suffer accordingly. It is very evident that good methods of reducing spectral band widths or otherwise improving spectral selectivity will be of great benefit in fluorimetric analysis. Corrected Spectra and their Uses Before considering that point further however a practical problem related to fluorescence spectra must be discussed. Because almost all fluorescence spectrometers are single-beam instruments excitation and emission spectra will reflect both instrumental and sample properties unless appropriate corrections are made.4 Thus an “uncorrected” emission spectrum will reflect not only the sample properties but also the wavelength dependence of the emission grating efficiency and of the photomultiplier sensitivity.One of the great benefits of the microelectronics revolution is that the correc-tion factors needed to eliminate such effects can easily and automatically be incorporated into the spectrometer using a microprocessor or microcomputer. In principle therefore the time is approaching where all fluorescence spectrometers should give the same spectrum for a particular sample examined under comparable conditions. Complacency about the benefits that this will produce must be tempered by the performance of the correction procedures in practice. Table 1 shows a fragment of the corrected emission spectrum of quinine sulphate (this compound is regaining favour as a primary fluorescence standard) as obtained on two different instruments.One is a standard spectrometer in the NBS laboratory in the USAS; the other is a commercially available “corrected” instrument in the author’s laboratory. It is seen that after normalisation the agreement between the spectra is fairly good but it is by no means perfect. Such results have to be borne in mind in considering the possible benefits of corrected spectra. Two particular benefits are apparent. Firstly if all spectro-meters yield the same spectra it is possible in principle to build up a library of spectra (similar to those available in other branches of spectroscopy) and use them for comparison and identification purposes.For example it is now feasible with the aid of a diode-array spectrometer or similar device to determine the fluorescence spectrum of a discrete fraction separated by HPLC “on the fly.” Comparison of this spectrum with a computer-stored reference spectrum will allow rapid identification of the s0lute.6~7 It is important to remember however that the fluorescence properties of a compound both spectral distribution and intensity are solvent dependent to a far greater extent than is true in for example absorption spectroscopy. To be really useful a library collection of spectra must therefore cover a variety of commonly used solvent systems. In the author’s laboratory, such a collection of spectra is being accumulated (previous collections have contained uncorrected spectra) and it is hoped that this library will eventually become generally available in hard-copy and floppy-disc forms.The second application of a collection of universally applicable spectra is more fundamental. In the past a major deterrent to the systematic use of fluorescence spectroscopy has been the paucity of adequate information on structure -spectra correlations. In other words which structural factors cause a molecule to have a high fluorescence quantum yield?; which factors affect the wavelength and intensity of this fluorescence?; and what are the exact effects of one or more substituent groups on the fluorescence properties of organic and organometallic compounds? Although some general rules are well known ,X much detailed information is still lacking and it remains hard to predict the luminescence properties of even the simplest compounds.Once again there is no doubt that progress in this direction has been hampered by the simple fact that until recently different spectrometers have given differ-ent spectral results. However the author’s laboratory has recently embarked on a programme that will apply pattern recognition methods9 to the library of corrected spectra so as to identify structural features responsible for particular spectroscopic properties. While this topic is too complex to consider in detail it is pertinent to note that a necessary preliminary is the coding of spectroscopic properties into a purely numerical form. The coding method must produce sufficiently distinctive data for different compounds but must also be simple enough to allow reasonably rapid calculations to be made.10 (This second requirement arises because the same method of coding spectra will be used in pattern recognition studies and in the comparisons of experimental and library spectra already discussed.) Table 2 lists the photoluminescence properties that we might wish to incorpor-ate in a complete “profile” of a compound; note that i ANALYST MARCH 1984 VOL.109 193 Table 2. Profile of a photoluminescent compound Solvent Maxima and minima in excitation spectrum* Maxima and minima in fluorescence spectrum* Maxima and minima in phosphorescence spectrum" Quantum yield of fluorescence Quantum yield of phosphorescence Lifetime of fluorescence Lifetime of phosphorescence * In these instances several spectral features may be involved the excitation fluorescence and phosphorescence maxima will normally be recorded but the method of moments may be used instead (see Fig.4). A __+ X Fig. 4. Calculation of the six moments of a hypothetical spectrum. The moments are a (the mean value of x). U2+, S and K where UZ4 = C(x - u ) * ~ v,iCy,; S = U7/(U2)3:2; and K = U,/(Li,)* - 3. Further details are gi;en in reference 11 Table 3. Fluorescence profile of anthracene t " Feature Data Notes Solvent . . . . . . 00 1 357 Excitation maxima . . f 376 340 Emissionmaxima . . 000 Quantum yield . . 360 Lifetime . . . . . . 049 Cyclohexane Maxima at 357,376 and 340 nm in order of decreasing intensity Maxima at 400 and 379 nm in order of decreasing intensity; the code 000 is used where the feature is absent or not measured i.e. 0.36 i.e. 4.9 ns includes lifetime and quantum yield data and both phosphor-escence and fluorescence spectra. For many compounds the available data may be incomplete for a number of reasons. The pattern recognition and spectral comparison methods used must take this into account. Several methods of expressing spectral data numerically can be considered. A simple approach involves expressing each characteristic of the compound as a three-digit code. An example of this approach is given in Table 3; note that the code 000 is used wherever the information is not available or where a spectral feature does not occur.Of course each piece of information is subject to experimental error perhaps k2 nm with characteristic wavelengths and so on. A satisfactory comparison program must make allowance for this but must not make so much allowance that the comparison method loses its discriminating power. Another interesting method of describing spectra has recently been suggested. * I It involves using moments to express spectral band shapes. Moments are useful in describing any distribution function particularly where the mathematical form of that function is not known. They have been used in infrared spectrometry12 and in other areas of chemistry.13 As is shown in Fig. 4 the first four moments are readily calculated and two additional para-meters the skewness (S) and the kurtosis ( K ) are readily determined from them.Tests on model compounds" have shown that the combination of these six parameters gives a high degree of discrimination between fluorescence spectra, whether or not the latter are distinguished by vibrational bands. Moreover the data are easily collected and the moments determined in real time and the moments are remarkably insensitive to spectral band width and to smooth-ing procedures. Possible disadvantages are that the system applies only to the spectra themselves and not to other data such as lifetimes and quantum yields and that the higher moments are very sensitive to base-line errors and to small changes at the extremes of a spectrum such as might be caused by scattered light signals or contaminating fluorophores.It is apparent that much more work needs to be done to compare the efficiencies of different methods of expressing fluorescence spectra digitally. Within a short time however, +I nm Lf 300 400 400 h 300 Fig. 5. Projected three-dimensional spectra of a crude oil sample in cyclohexane (concentration 4 pg ml ~ 1). Computer software allows the spectrum to be examined from the high or low excitation wavelength directions. Individual traces represent 2-nm excitation wavelength increments extended collections of corrected fluorescence spectra will be available in both digital and analogue form and they will certainly find many applications. Synchronous and Contour Spectra This section considers the further question of selectivity in fluorescence spectroscopy and in particular the possibility of obtaining characteristic spectra with narrow band widths.It is first necessary to recall that conventional excitation and emission spectra are not complete descriptions of the spectral distribution of a compound. The excitation spectrum is obtained at only a single emission wavelength and the emission spectrum is obtained using only one excitation wavelength. A complete description requires a three-dimensional spectrum,'4 in which one axis is the excitation wavelength scale a second axis represents the emission wavelength and the third axis is the intensity axis. Such spectra are known also as excitation - emission matrix spectra contour spectra and total luminescence spectra.The simpler way of expressing such results is shown in Fig. 5. This spectrum is a projection containing (in this instance) 30 individual spectra 194 ANALYST MARCH 1984 VOL. 109 superimposed with the aid of a “hidden line rer.qova1” routine. The latter simplifies and clarifies the spectra and does not sacrifice information because as can be seen it is also possible (using software only) to reverse the direction of the excitation axis and hence see “behind the hill.” These spectra take about 1 h to accumulate (unless a diode-array spectrometer is available) most of this time being taken up by the individual spectral scans. An interfaced microcomputer stores and presents the spectra. The projected spectra give excellent general pictures of the properties of samples but it is more difficult to extract detailed information e.g.fluorescence intensities at any pair of wavelengths from them. The second method of acquiring “total luminescence” spectra is illustrated in Fig. 6. Here the two normal axes represent the emission and excitation wavelengths while the intensities are expressed as a series of contours. In the figure contours are drawn at go% 70% 50% etc. of the peak intensity but clearly contours at other levels should also be available. It is obvious that this method is much more informative than the former especially where quantitative data are needed but contour spectra are also more difficult to obtain. The computer-interfaced spectrometer must first accumulate all the necessary data and then calculate the contours with the aid of an interpolation program that estimates the points of equal fluorescence intensity.Contour-ing programs are available for some commonly used comput-ers,15 but the process remains time consuming. The process of accumulating contour data can sometimes be accelerated with the help of Vavilov’s law,16 which proposes that the closely related processes of excited state vibrational relaxation and internal conversion which rapidly (ca. 10-12 s) take an excited molecule to the lowest vibrational level of the first singlet excited state should occur with a quantum yield of unity. This means that the spectral distribution of the emission spectrum should be independent of the excitation wavelengths (and vice versa). Fig. 7 shows this principle in operation for Lucifer Yellow VS.The emission spectra excited at 429 and 388 nm are of equivalent shape; the latter has half the maximum intensity of the former because 388 nm is the half-maximum wavelength of the excitation spectrum. It is evident that an extension of this principle would allow the calculation of a complete contour spectrum of a compound from a knowledge of its excitation and emission spectra. Fig. 8 shows that for Lucifer Yellow at least the principle holds good. Remembering the interpolation processes involved in producing contour spectra it is apparent that the experimen-tal contour spectra obtained directly by using all possible excitation and emission wavelengths agree very well with the contours calculated with the aid of the Vavilov’s law method.This method has two important limitations. Firstly it is applicable only to pure compounds not to mixtures; and secondly a number of important exceptions to Vavilov’s law itself have been identified. These include many benzene derivatives and aromatic amino acids. 17 Nonetheless it may be useful to be able to calculate the contour spectrum of a single analyte and thus to compare it with the known contour spectrum of a common sample matrix. Further a comparison between the calculated and experimental contour spectra is a good test of the validity of Vavilov’s law. Additionally recent methods permitting quantitative evaluation of complex con-tour spectra may assume that Vavilov’s law applies,l8.19 so the method discussed is of more than academic interest.One of the most important merits of contour spectra is that sections through them give very useful information. Thus a vertical section ie. at constant emission wavelength is equivalent to an excitation spectrum and a horizontal section is equivalent to a conventional fluorescence spectrum. A 4.5” section with both monochromators set to the same wavelength corresponds to first-order Rayleigh scattered light emission. A 4.5” section with a (fixed) wavelength difference between the two monochromators produces a “synchronous” spectrum __+ hf Fig. 6. Contour spectrum of a hypothetical fluorophore. The dark spot represents the wavelengths of maximum excitation and emission, and the contours join points with fluorescence intensities 90% 70%, etc.of this maximum 400 500 600 Wavelengthlnrn Fig. 7. Application of Vavilov’s law in the fluorescence of Lucifer Yellow VS. The emission spectrum obtained at an excitation wavelength of 384 nm (at which the excitation spectrum is at half its maximum intensity) is the same shape as that obtained at the maximum excitation wavelength (429 nm) but has half the intensity 380 -480 520 560 600 640 hfin m Fig. 8. Contour spectra of Lucifer Yellow VS. Experimental contours are continuous lines; broken lines are contours calculated from conventional excitation and emission spectra using Vavilov’s law (Fig. 9). It was shown over 10 years ago by Lloyd2O that such spectra are simpler narrower and hence more characteristic than conventional spectra. This extra selectivity is illustrated in Fig.10. Three oil samples (a new sample of a light lubricating oil a used sample of the same oil and a third unrelated oil) have conventional excitation and emission spectra that are remarkably similar. However the synchro-nous spectra show clearly which oil is the odd one out ANALYST MARCH 1984 VOL. 109 195 Af -Fig. 9. a contour spectrum Representation of a synchronous spectrum as a 45" section of (a) 1' = 388 nm t -Fig. 11. Use of variable-angle synchronous scanning spectrometry to study a mixture of three fluorophores. The analyte is shown as continuous contours; interferences have broken contours. The scan shown narrowly fails to pass through the fluorescence maximum of the analyte but achieves a high degree of discrimination against the interferences 240 280 320 360 h,,/nm 290 300 A, Inm 250 260 270 280 287 310 333 356 379 402 kf/nrn Fig.12. Simultaneous analysis of phenylalanine (Phe) and trypto-phan (Trp) by variable-angle synchronous scanning spectrometry. Excitation and emission monochromators had starting wavelengths of 245 and 275 nm respectively and were scanned at rates of 0.40 and 0.93 nm s-1 respectively. Solutions contained 1,50 pg ml-1 Phe; 2,50 p ml-1 Phe + 0.5 pg ml-1 Trp; 3,50 pg ml-1 Phe + 1 pg m1-l Trp; 4, 5 8 pg ml- 1 Phe + 1.5 pg ml- 1 Trp; and 5 2 pg ml- I Trp 380 420 460 500 540 Mnm 250 273 350 373 450 h,,lnm 473 hfhm Fig. 10. Use of synchronous scanning spectrometry to characterise oil samples. Three oils (A-C) have similar (a) excitation and ( b ) emission spectra,.but oil C has (c) a characteristic synchronous spectrum. The synchronous spectrum of B is very similar to that of A but has an additional high-wavelength component characteristic of a used oil moreover the used oil shows small but very characteristic differences from its unused counterpart. There is increasing interest in a further selective variant of fluorescence spectrometry which we have called variable-wavelength or variable-angle synchronous scanning.21 Fig. 11 shows the same contours as Fig. 6 but superimposed on them now are hypothetical contours of two further compounds. In this instance conventional spectroscopy will clearly be very non-selective as one of the contaminants has the same excitation wavelength and the other contaminant the same emission wavelength as the target compound.It is evident, however that a non-45' section through the matrix might still give a high degree of selectivity. Variable-angle synchronous scans can be obtained either indirectly (by accumulating the whole contour spectrum and using the interfaced computer to cut the desired section) or directly by scanning the two monochromators of the spectrometer at different rates. (In the example given it will evidently be necessary to scan the excitation monochromator faster than the emission mono-chromator.) Clearly the latter approach will be the simpler and in the author's laboratory an instrument has been modified to record such spectra.22 Two different but related types of application of variable-angle scanning are feasible.In addition to the type just mentioned i.e. optimum resolution of overlapping specta there is a second application illustrated in Fig. 12. Examination of the contours of the amino acid 196 ANALYST MARCH 1984 VOL. 109 phenylalanine and tryptophan shows that there is very little overlap between their spectra so little that two entirely separate experiments would normally be necessary to deter-mine both these compounds in a mixture. However using variable-angle scanning both compounds can be studied in a single spectrum and can be determined in the presence of each other. This approach has a number of potentially interesting applications. Fluorescence Immunoassays This section considers some more applied aspects of lumines-cence spectrometry.An area of application of great current interest is the development of immunoassays. Such methods are largely used in clinical medicine at present but they have tremendous potential also in veterinary. food and environ-mental analysis. Apart from the use of fluorescence as a detector in HPLC systems immunoassays probably represent the most important current use for fluorescence methods and the same will probably soon be true in chemiluminescence. Even more important applications in this area present the same problems as are found in many other areas of application of fluorescence. So many of these problems and many of the possible solutions are of general relevance in fluorescence analysis. Fluorescence immunoassays are of two general types.23 Most attention recently has been devoted to the homogeneous type of assay.2j This utilises the principle that the fluorescence properties of a labelled molecule may change when that molecule binds to an appropriate antibody.Antibody-bound and unbound molecules can thus be distinguished without a physical separation step. and the method is simple and easily automated. These homogeneous methods have the intrinsic disadvantage that any native fluorescence from the original samples is still present at the final measurement step of the immunoassay and adversely affects the limits of detection attainable. In contrast heterogeneous assays25 require a separation for the physical separation of antibody-bound and unbound molecules; this is inconvenient especially if automa-tion is envisaged but it at least serves to remove much of the background interference and better limits of detection are consequently available.Perhaps the best known type of homogeneous assay is that based on fluorescence polarisation studies.26 Fluorescence polarisation measurements have been made for at least 60 years and have found several applications apart from immuno-assays.27 These include the reduction of scattered light interference and studies of molecular shape structure and interactions. In practice the only instrumental modification required for polarisation studies involves the insertion of two polarising films immediately before and after the sample in the light path. The excitation polariser is generally oriented with a vertical plane of polarisation.and the emission polariser can be rotated to be oriented either vertically or horizontally. Electro-optic polarisers have recently been used with success. Automatic changing of the polariser orientation and rapid calculation of the polarisation value p is readily accom-plished; p is given by p = Ill - 1,1111 + Zl where Ill and I , represent fluorescence intensities measured between parallel and crossed polarisers respectively. The significance of p is that it is close to zero for low relative molecular mass fluorophores which can tumble freely and lose all trace of their initial orientation during the 10 ns or so of the excited state lifetime. For macromolecules or macromolecular complexes however the tumbling motion is much slower so their fluorescence emission is partially polarised.In an immunoassay this permits a distinction between the unbound fluorescent labelled molecule which has p close to 0 and its antibody-bound counterpart. In the presence of a sample containing the unlabelled X molecule the labelled molecule is displaced from the antibody and its polarisation falls. Fig. 13 0.20 Q 0.1 5 0.10 0 20 40 60 Amount of drugipmol Fig. 13. Fluorescence polarisation imrnunoassay of lysergic acid diethylamide (LSD) using a fluorescein-labelled LSD derivative. Experimental details are given in reference 28 shows the application of this principle to the determination of a fluorescein derivative of LSD.28 The measurements were made using a very simple fluorimeter and the detected changes in the value of p were small but picomolar levels of the drug could be determined.Such results were obtainable in pure solutions of the drug and in urine samples; much poorer limits of detection were obtained in serum samples however, because of the fluorescence background effect already referred to. A closer examination of this background signal is profitable. It is caused in part by scattered light and in part by the fluorescence of endogenous materials. The effects of scattered light are considerable; solutions containing proteins and other macromolecules scatter much more light than solutions of low relative molecular mass species and many of the labels used in immunoassays have small Stokes shifts-fluorescein is the most obvious example-and so are particularly vulnerable to scattered light interference.A great number of methods for the reduction of background signals have been suggested; here two of them the use of adsorbents and the use of fluorescence lifetime measurements are discussed. Blood serum samples contain about 7% by mass of protein. and many of the individual proteins are capable of binding small molecules so it is reasonable to suppose that at least part of the background fluorescence of serum will be associated with the protein fraction. This is certainly true of the most intense fluorescence band of serum with maximum emission at about 3.50 nm. Although this band is very broad and contributes to the background at wavelengths up to at least 4.50 nm (cf.the discussion of bandwidth above) it is less important in practice than the much weaker fluorescence bands occurring in the visible region of the spectrum. There are two or three of the latter and it is astonishing that they are not better characterised. Studies are proceeding on the use of adsorbents to minimise these interferences in a rapid pre-assay column chromatographic step. Thus an immobilised form of Cibacron Blue an active dye well known in preparative biochemistryZq for its strong affinity for human serum albumin and some enzymes is effective. If a sample of serum is passed through a small column of this affinity medium the albumin is almost totally removed; the process takes only a few minutes and does not necessarily involve any serious dilution of the sample.A fluorescence band which occurs at about 440 nm is reduced by about two thirds by the action of the immobilised dye and is therefore probably due to an albumin- or enzyme-bound fluorophore. In contrast a further fluores-cence band centred at about 650 nm is largely unaffected by the affinity column so is presumably not due to albumin associated fluorescence. (It is worth noting that the existence of this latter band which seems not to have been studie ANALYST MARCH 1984 VOL. 109 197 before puts into some doubt the assumption that high-wavelength labels are the most desirable in fluorescence immunoassays.) Suitable adsorbents either singly or in combination. will be very effective in reducing the background fluorescence of serum and they are cheap rapid and simple to use.By removing proteins from the sample they also contribute to a reduction of the scattered light interference.30 The possible use of fluorescence lifetimes in analytical work is based to a great extent on Weider's survey of the fluorescence lifetimes of organic molecules.31 Most organic molecules have fluorescence lifetimes of less than 30 ns. The use of a fluorescent label with a lifetime of 2.50 ns in conjunction with a gated detector that only detects light emitted more than e.g. 50 ns after an exciting light pulse would largely eliminate the background fluorescence and scattering signals. In practice there are two versions of this approach. The first possibility and the one more often attempted up to the present has been to use lanthanide ions as labels.In principle this is a promising approach.32.33 The lanthanides (in practice terbium or europium) have lifetimes of about 1 ms and the necessary time discrimination is readily accessible on instruments with pulsed xenon flash-tube sources the pulse width being a few microseconds. Lanthan-ide fluorescence is also distinguished by very narrow spectral band widths (this will further help background rejection) and although the free ions are only feebly fluorescent many of their chelate complexes can be detected at picogram levels.34 A major difficulty in practice involves the production of complexes that do not dissociate when placed in a biological sample. At present the conclusion must be that this approach is photochemically easy but chemically difficult.The second approach involves the use of a conventional fluorophore with a long lifetime; the group most often suggested is pyrene with a lifetime of 100 ns or more.35 This presents no chemical problems-for example pyrenesulphonyl chloride is commer-cially available-but it requires a very sophisticated instru-ment to measure nanosecond lifetimes. In the future simpler nanosecond spectrometers may be developed and this approach will then be the more promising. There is a further reason encouraging the use of organic labels in fluorescence lifetime immunoassays. The fluores-cence transitions in lanthanide ions arise from energy levels that are well protected from the environment and that are therefore not likely to change much when for example a lanthanide labelled material is bound to an antibody.The chances of developing a homogeneous assay are thus small. The lifetimes of conventional organic molecules are in contrast. very susceptible to minor changes in structure and environment,3h and might well be suitable for the develop-ment of homogeneous assays. One reason for caution is that even pure samples of some compounds may show non-exponential decays.3' Whatever the reason for this phenom-enon it could complicate data handling in lifetime assays. The significance of light scattering interference has already been mentioned a third possible way of reducing it is to have no incident light source at all that is. to use a chemilumines-cence immunoassay. The development of a variety of chemi-luminescence and bioluminescence assays which combine simple equipment with exquisite sensitivity has been one of the major developments in analytical science in recent years.w w So far as immunoassays are concerned most attention has been given to methods involving luminol40 and acridine labels.41 Despite very low quantum yields (often d 1 "/o) and some difficulties in preparing labelled compounds these assays have proved very sensitive but they seem to suffer from non-specific quenching interferences and have yet to make a serious impact on the real world of immunoassay. Also worthy of attention is a different form of chemilumi-nescence assay based on the reaction of an oxalate ester such as bis(2,4,6-trichlorophenyl) oxalate TCPO with hydrogen peroxide.This reaction generates a highly energetic inter-mediate which is capable of exciting the luminescence of many conventional fluorophores. This reaction has been known for at least 16 years42 and its analytical use was first investigated about 8 years ag0,~3 but it has not so far been widely applied44 to the immunoassay area. Its potential advantages are clear. Apart from the elimination of scattered light effects the TCPO reaction has a high quantum yield (typically 20-2.5%) and can be used to excite the fluorescence of conventional and easily handled fluorophores such as fluorescein fluorescam-ine and dansyl. Such labels can be introduced very readily into a variety of analyte molecules. The TCPO reaction can also be studied using simple luminometers and the combination of luminometry and flow injection analysis (FIA) is especially suitable because very reproducible reagent mixing is required in chemiluminescence assays and FIA provides just that.45 It has proved possible to detect about 10 fmol of fluorescein with a relative standard deviation of ca.3% using only a modified filter fluorimeter as detector similarly ca. 10 pmol concentra-tions of fluorescamine derivatives can be detected whereas it is difficult to detect sub-nanomolar concentrations by conven-tional fluorimetry. The TCPO reaction has been used to detect trace amounts of fluorescent labelled molecules after HPLC separation,46 with the TCPO reaction being used in a fairly simple post-column reactor. Limits of detection for intrinsic-ally fluorescent hydrocarbons are apparently not so good but the discovery of the enhanced detection limits with amino derivatives is very interesting and merits further study.47 The use of the TCPO and similar systems in immunoassays reveals two particular difficulties.One is the non-specific quenching problem mentioned above; this seems to be a drawback of chemiluminescent systems and it will be as important to overcome it as it is to reduce background signals in fluorescence assays. Recent work indicates that the extent of quenching by proteins in the TCPO system may be dependent on the relative molecular mass of the proteins.48 A second major problem is that TCPO is not water soluble. Not only is it rapidly hydrolysed in aqueous systems it also forms turbid solutions.Neither this possibility nor the use of mixed organic - aqueous solvent systems is satisfactory in immuno-assays. However recent work in the author's laboratory shows that it is possible to overcome both these problems and to develop a series of highly sensitive assays that are very easy to perform. Conclusions This brief survey has omitted topics such as the use of lasers as light sources in fluorimetry,49 derivative fluorescence spectro-metry,50 the encouraging developments in room-temperature phosphorimetry,51952 the use of matrix isolation methods,s3 the interesting possibilities of fluorimetry in micellar solu-tions,s"55 combinations of luminescence methods and thin-layer chromatography56 and new bioluminescent systems.j7 However it is apparent that with the increasing demand for highly sensitive and selective analyses in many areas of organic and biological chemistry photoluminescence and chemilumi-nescence techniques will continue to provide important and fascinating fields of research.I thank the Science and Engineering Research Council the Medical Research Council The Home Office The Depart-ment of Health and Social Security Perkin-Elmer Ltd. Baird Atomic Ltd. Glaxo Operations UK and The Boots Company for financial support of research in the areas described in the paper. I also thank Professor D. Thorburn Burns of The Queen's University Belfast and Professor J . W. Bridges of the University of Surrey for their long-standing co-operation and many research colleagues for their essential contributions 198 ANALYST MARCH 1984 VOL.109 References 1. 2. 3. 4. Serio M. and Pazzagli M. Editors “Luminescent Assays,” Raven Press New York 1982. Parker C. A. “Photoluminescence of Solutions,” Elsevier, Amsterdam 1968. Stewart W. W. Nature (London) 1981 292 17. Roberts G. C. K. in Miller J . N. Editor “Standards in Fluorescence Spectrometry,” Chapman and Hall London. 1981 p. 49. Velapoldi R. A. and Mielenz K. D. Natl. Bur. Stand. ( U . S . ) Spec. Publ. 1980 No. 260-64. Jadamec J. R Saner W. A. and Talmi Y. Anal. Chem., 1977 49 1316. Pellizzari E. D. andsparacino C. M. Anal. Chem. 1973,45, 378. Schulman S. G. “Fluorescence and Phosphorescence Spectro-scopy Physicochemical Principles and Practice .” Pergamon Press Oxford 1977.Varmuza K . “Pattern Recognition in Chemistry,” Springer-Verlag Berlin 1980. Lyons J. N. Hardesty P. T. Baer C. S. andFaulkner L. R., in Wehry E. L Editor “Modern Fluorescence Spectroscopy,” Volume 3. Plenum Press New York 1981 p. 1. Mulkerrin M. G . and Wampler J . E. Anal. Chem. 1982,54, 1778. Tamura T. Tenabe K. Hirashi. J. and Saeki S Bunseki Kagaku 1979 28 591. Wampler J . E. Mulkerrin M. G. and Rich E. S Clin. Chern. (Winston-Salem NC) 1979 25 1628. Christian G. D. Callis J. B. and Davidson E. R. in Wehry, E. L. Editor “Modern Fluorescence Spectroscopy,” Volume 4 Plenum Press New York 1981 p. 111. Giering L. P. Ind. Res. Dev. 1978 20 134. Birks J. W. “Photophysics of Aromatic Molecules,” Wiley-Interscience London and New York 1970. Tatischeff I.and Klein R . in Birks J . B. Editor “Excited States of Biological Molecules,” Wiley London 1976 p. 375. Vo Dinh T. Anal. Chem. 1978. 50 396. Williams W. P. Murtz N. R. and Rabinovitch E . Photo-chem. Photobiol. 1969 9 455. Lloyd J . B. F. Nature (London) 1971 231 64. Eastwood D. in Wehry E. L. Editor “Modern Fluorescence Spectroscopy,” Volume 4 Plenum Press New York 1981 p. 251. Miller J . N. Ahmad T. A. Bower A. F. and Fell A. F. to be published. Smith D . S. Hassan M. and Nargessi R. D. in Wehry, E. L. Editor “Modern Fluorescence Spectroscopy,” Volume 3 Plenum Press New York 1981 p. 143. Ullman E. F. Bellet N. F. Brinkley J. M. and Zuk R. F., in Nakamura R. M. Dito W. R. and Tucker E. S. Editors, “Immunoassays; Clinical Laboratory Techniques for the 1980s,” Alan R.Liss New York 1980 p. 13. Maggio E. T. in Nakamura R. M. Dito W. R. and Tucker, E. S. Editors “Immunoassays; Clinical Laboratory Tech-niques for the 1980s,” Alan R. Liss New York 1980. p. 1. Dandliker W. B. Kelly R. J. Dandliker J . Farquhar J . , and Levin J. Zmmunochemistry 1973,lO. 219. Weigert F Z. Phys. 1922 23 232. Hubbard A. R. Miller J. N. Law B. Mason P. and Moffat A. C. Anal. Proc. 1983 20. 606. Travis J . Bowen J . Tewksbury D. Johnson D. and Pannell R. Biochem. J . 1976,157,301. Miller J . N. and Gossain. V. to be published. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. Weider I. in Knapp W. Holubar. K. and Wick G.Editors, “Proceedings of the 6th International Conference on Immuno-fluorescence,” Elsevier-North Holland. Amsterdam and New York 1978 p. 67. Soini. E. and Kojola H. Clin. Chem. (Winston-Salem NC), 1983 29 65. Meurman 0. H. Hemmila I. A . Lovgren T. N and Halonen P. E. J . Clin. Microbiol. 1982 16 920. Petterson K. Sutari H Hemmila I. A. Soini E Lovgren, T. Hanninen V. Tanner P. and Stenman U.-H. Clin. Chem. (Winston-Salem NC) 1983 29 60. Spencer R. D. Vaughan W. M. and Weber G. in Lim, E. C. Editor “Molecular Luminescence,” W. A. Benjamin, New York 1969 p. 72. Cline Love L. J. and Shaver L. A. Anal. Chem. 1980 52, 154. Gauduchon P. Donzel. B. and Wahl. P . in Birks J. B., Editor “Excited States of Biological Molecules,” Wiley, London 1976 p.404. 38. Cormier M. J. Hercules. D. M. and Lee J. Editors, “Chemiluminescence and Bioluminescence .” Plenum Press, New York and London 1973. DeLuca M. A. Editor “Bioluminescence and Chemilumines-cence,” (Methods in Enzymology Volume 57) Academic Press New York 1978. 40. Hersh S. L. Vann W. P . and Wilhelm S. A. Anal. Biochem. 1979 93 267. 41. Simpson J. S. A. Campbell A. K Woodhead J . S., Richardson A. Hart R. and McCapra F. in DeLuca M., and McElroy W. D. Editors “Bioluminescence and Chemi-luminescence,” Academic Press New York 1981 p. 673. Rauhut M. M. Bollyky L. J . Roberts B. G. Loy M., Whitman R. H. and Iannotta A. V. J. Am. Chem. SOC., 1967 89 6515. 43. Williams D. C. Huff G. F. and Seitz W. R . Anal. Chem., 1976,48 1003. 44. Mandle R. M. and Wong Y. N. Int. Pat. Appl. No. PCT/US80/01485 1981. 45. Mahant V. K. Miller J. N. and Thar H. Anal. Chim. Acta, 1983 145 203. 46. Kobayashi S. and Imai K . Anal. Chem. 1980 52 424. 47. Sigvardson K. W. and Birks J . W. Anal. Chem. 1983 55, 432. 48. Mahant V. K. PhD Dissertation Loughborough University of Technology 1983. 49. Richardson J. H. in Wehry E. L. Editor “Modern Fluores-cence Spectroscopy,” Volume 4 Plenum Press New York, 1981 p. 1. 50. Miller J. N. Ahmad T. A. and Fell A. F. Anal. Proc. 1982, 19 37. 51. Miller J. N. Trends Anal. Chem. 1981 1 31. 52. Parker R. T. Freedlander R. S. and Dunlap R. B. Anal. Chim. Acta 1980 120 1. 53. Stroupe R. C. Tokousbalides P. Dickinson R. B. Wehry, E. L. and Mamantov G. Anal. Chem. 1977 49 701. 54. Zachariasse K. A. Chem. Phys. Lett. 1978 57 429. 55. Kalyanasundaram K. Gneser F. and Thomas J . K. Chem. Phys. Lett. 1977 51 501. 56. Hurtubise R. J. “Solid Surface Luminescence Analysis,” Marcel Dekker New York 1981. 57. Dunlap J. C. Hastings J. W. and Shimomura O. Proc. Natl. Acad. Sci. USA 1980 77 1394. Paper A31307 Received September 6th 1983 32. 33. 34. 35. 36. 37. 39. 42

ISSN:0003-2654    

DOI:10.1039/AN9840900191

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST MARCH 1084 VOL. 109 199 Capillary Separation Methods a Key to High Efficiency and Improved Detection Ca pa bi I ities Plenary Lecture Milos Novotny Department of Chemistry Indiana University Bloomington IN 47405 USA The current status of capillary separation techniques is reviewed. While capillary gas chromatography has reached a degree of maturity with respect to both column technology and instrumental systems capillary separations with the condensed mobile phases still present many challenges to future developments. At present micro-column liquid chromatography offers both chromatographic performance and novel detection techniques owing to the very low flow-rates encountered with this method; some unique applications have already appeared. Capillary supercritical fluid chromatography has also attracted much attention during the last 2 years because of its unique position between gas chromatography and high-performance liquid chromatography.High-voltage capillary electrophoresis shares some instrumental aspects with the remaining capillary techniques but also offers some distinct advantages of its own. Keywords Capillary separation; gas chromatography; capillary supercritical fluid chromatography Throughout the entire history of modern chemistry there has been a continuing (and at certain times intensive) effort to identify and measure various chemical substances in increas-ingly smaller amounts and doing so in extremely complex sample matrices is hardly unique to the current scientific efforts. Let us be reminded of the outstanding work of the past generations of chemists who established for the first time the presence of mammalian hormones vitamins insect phero-mones alkaloids etc.in plant and animal samples that often contained hundreds or thousands of interfering compounds. Likewise the discovery that benzo[a]pyrene is a cancer-causing principle of coal tar (through its isolation structural elucidation and biological testing) was a non-trivial task in view of the complexity of the sample. All this important work of the past generations is even more admirable if we consider the primitive analytical techniques that those pioneers of modern science had at their disposal. While many primary objectives of analysing complex samples today are somewhat similar to those of past genera-tions of chemists i.e.identification and accurate measure-ment of the substances of interest the pace at which this task is accomplished has increased dramatically. Reliable analyses of complex organic mixtures are frequently a key to the solution of many problems facing our technological society. During the last two decades high-resolution chromatography and elec-trophoresis have started to assume a central role in such endeavours. An extensive application of high-resolution separation techniques is also desirable for the further develop-ment of reliable measurement techniques based on modern spectroscopy and electrochemistry. It has rapidly become established that the concept of the open tubular chromatographic column developed in the late 1950s by M.J. E. Golay,’ has been one of the most important developments in modern analytical chemistry. However only after two decades has capillary gas chromatography reached a high degree of maturity through the necessary instrumental developments and highly sensitive detection and ancillary techniques. Although the method is obviously confined to relatively volatile compounds. it has now been applied widely in analytical practice. During recent years the geometrical simplicity of the open tubular column has also invited both theoretical and instru-mental developments with condensed mobile phases. Thus, the methods of capillary liquid chromatography capillary supercritical fluid chromatography and capillary zone electro-phoresis have recently emerged as worthwhile areas of analytical research.While the necessary instrumental devel-opments in these directions are technologically demanding, the same has been true of almost any new analytical development in the past. The goal of high resolution with complex non-volatile mixtures is eminently worthwhile in itself. In addition some unique analytical advantages are offered through this direction. Both fundamental developments and a wealth of new applications have been particularly characteristic of capillary separation techniques during the last decade. This paper attempts to summarise the most important developments in this field and to point out certain directions that are worth pursuing and may be particularly fruitful in the near future. Capillary Gas Chromatography After a relatively slow period during the 1960s a sudden increase of interest in capillary gas chromatography (GC) has been characteristic for the last decade.The main reasons for this were that certain patent restrictions were eased opening the area for a wide-scale commercialisation; that technology of highly efficient and inert glass capillary columns was sufficiently developed; and that several relatively simple procedures for direct sample introduction were devised. The development of flexible fused-silica capillary columns by Dandeneau and Zerenner2 further removed “the psycholog-ical barrier” that numerous chemists have had with respect to the relatively fragile glass capillaries. Further instrumental developments and various forms of professional education in this “technique-orientated” analytical approach have also played a significant role in the acceptance of capillary GC as both a research and routine analytical method.Perhaps the most exciting period in the over-all develop-ment of capillary GC is associated with glass capillary columns. Using such column types Desty‘s initial success3 with petrochemical applications was soon followed by the separations of isotopic molecular species4 and eventually, some extremely complex mixtures such as tobacco smoke.5.6 As the earlier applications of capillary GC employed almost exclusively split injection the biochemical and environmental applications of this method were not feasible (owing to typically limited sample amounts) until new sampling proce-dures became available during the 1970s.With the arrival of such concentration and splitless injection techniques th 200 ANALYST MARCH 1984 VOL. 109 prevailing view that capillary GC was an unattractive method for trace analysis became no longer justified. There is extensive use of this technique in environmental analysis and biochemical research and there are now numerous examples of trace determinations .7-9 The importance of improved sampling methods for quanti-tation in capillary GC can hardly be ovekstated. Direct injection procedures have been developed during the last decade that utilise large aliquots of available samples for analysis while minimising the initial band dispersion at the capillary column inlet. The splitless injection method that employs the so-called “solvent effect” 1 0 the falling-needle technique” 912 and various pre-column methods13-lh are among the most versatile approaches to sample introduction in capillary GC.Some of these sampling techniques have been automated resulting in improved reproducibility. It is felt that the pre-column sampling methods deserve increasing use in analytical work as they frequently serve a double function: removal of solvents or derivatisation agents; and protection of the analytical column from non-volatile impurities contained in the injected samples. The chemical nature of pre-column packing can also be varied to suit a. particular sample type. The principle of “selective sampling,“ i.e. various forms of double-column GC procedures. has been increasingly advo-cated to improve various aspects of capillary GC determina-tion.16.17 Carrier inlet Secondary cooling Fig. 1. Capillary on-column injector including the secondary cooling system. 1 = Microsyringe 2 = valvc lever 3 = valve seal; 4 = stainless-steel rotating valve 5 = column seal; 6 = cooling jacket; 7 = capillary column. Reproduced from reference 19 with permission of Elsevier Publishing Company One of the most important recent advances in capillary GC appears to be the method of direct (on-column) sample deposition developed by Grob and co-workers1831y; in this method a thin syringe needle is inserted directly inside a 0.2-0.4 mm i.d. capillary column (see Fig. l) while depositing a “cold sample” on to the first column segment.While this sampling procedure is obviously technique-orientated and cannot be readily automated for routine work a quantitative sample transfer is achieved under the correct operating conditions. Excellent quantitative results have been repor-tedly to counter a general scepticism following the introduc-tion of this sampling technique. The major advantages of this on-column sampling procedure are a minimum thermal decomposition of labile compounds (observed occasionally with other injector types) as well as the lack of discrimination towards the later eluting components. Gradually improving capillary column technology has also been crucial in quantitative determinations. While the prob-lems of deposition of various stationary phases on the column wall in the form of a uniform film have finally been solved to provide efficient columns increasingly the attention of researchers has now been turned towards column de-activation problems (removal of “residual adsorption” asso-ciated with the column wall).There is a general agreement that the fused-silica capillaries are much easier to deactivate than glass capillaries. Numerous examples of chromatography with “difficult samples,” such as either strongly basic or acidic compounds at low-nanogram amounts are now demonstrated throughout the literature. Preparation of the immobilised films of liquid phases2(&-2-2 has also added to the over-all analytical capability of capillary GC. Immobilised polymeric phases are essential to the correct use of the on-column sampling techniques.With ordinary stationary phases migration of the film easily occurs following the sample injection thus causing adsorption problems at the column inlet and a possible loss of column efficiency. Various aspects of capillary column technology have been reviewed recent 1 y .23 A decrease in the column diameter leads to further increases in the column efficiency as well as a shortening of analysis time. These effects are a straightforward consequence of the chromatographic theory. Desty has already demon-strated the power of this approach in his pioneering work3 on capillary GC. Recently several laboratories have started to pursue this direction further. As calculated by Guiochon,24 the small-diameter columns (around 50-100 pm) could be utilised if certain sampling and detection-volume problems were adequately solved.Although it is quite clear that significant departures from today’s GC instrumental para-meters will be necessary to make this approach attractive to analytical laboratories some preliminary results appear pro-mising. As an example Fig. 2 shows a chromatographic 0 “ 2 4 6 8 10 12 14 Time/mi n Fig. 2. reference 25 with permission of Hiithig Publishers. A capillary GC profile of the silylated components from a uraemic serum. using the splitless injection technique. Reproduced fro ANALYST MARCH 1984. VOL. 109 20 1 -J ---I I 1 I 35 80 100 120 140 170 Temperatu rePC I I I I I I 0 20 40 60 80 100 Time/m in Fig. 3. Capillary chromato rams of volatiles from a 24-h urine of a normal male detected by (3 a flame ionisation detector and ( h ) a nitrogen-sensitive thermionic detector.Reproduced from reference 26 with permission of Elsevier Publishing Company E2 sample L I I I L 15 12 9 6 3 Time/min Fig. 4. Mass-fragmentographic analysis of estradiol from a sample o f female rat plasma. Reproduced from reference 32 with permission of Elsevier Publishing Company profile of body fluid constituents2’ that was obtained (with an 8.5 m x 50 um i.d. capillary column) in an analysis time considerably shorter than is achievable with conventional, 0.25 mm i.d. columns. However with decreased column diameters and film thicknesses column capacity and surface deactivation problems become more apparent. Contemporary GC offers an impressive number of highly sensitive and selective detectors and ancillary tools.Fortun-ately. capillary GC combines effectively with most of these devices and in some instances even strengthens their capabilities. Thus. flame ionisation detectors as well as the selective flame detectors based on the flame-photometric and thermionic principles are now commonly employed with capillary columns. While the detection selectivity is often needed to trace the compounds of interest in complex sample matrices peak by peak comparisons of chromatograms obtained simultaneously with a universal and a selective detector (see for example reference 26 Fig. 3) can often have considerable value in peak identification studies (pro-vided that some other information is available). A parallel operation of several detectors combined with a single capillary column is now technically feasible.Whereas the widely used flame-photometric and thermionic detectors provide the often-needed selectivity towards certain elements (nitrogen phosphorus sulphur etc.) in the separ-ated molecules there is still a need for additional detection capabilities in capillary GC. Thus the search in this area is likely to continue for some time. An example of the adaption of a previously used detection principle to capillary GC is shown in a recent work by Estes et ~ 1 . 2 ~ on the microwave plasma detection of lead compounds in gasoline. Detectors that benefit most from their combination with capillary columns are various concentration-sensitive devices. As the column diameters are small and the volumetric flow-rates are correspondingly reduced.the mass sensitivity of these concentration-sensitive detectors is significantly enhan-ced. Numerous efforts have therefore been made to design such detectors with very small volumes including the electron capture,28 the ultraviolet spectroscopic29 and the photoionisa-tion30 detectors. Significant increases in mass sensitivity appear to make it worthwhile miniaturising the detector cells further. Mass spectrometry has become the most important ancillary method to GC. Advances in both coupling technology and mass spectrometer design have made it now entirely feasible to monitor the capillary column effluent in both low- and high-resolution spectral modes of operation. More recently, Fourier-transform infrared spectroscopy has also been effec-tively combined with capillary GC,31 reducing an earlier gap between sensitivities of mass-spectral and optical methods.Besides their obvious identification power both ancillary methods can also be employed as highly selective GC detectors. An example of this is seen in widely used mass-fragmentographic techniques that revolutionised phar-macological biomedical and environmental monitoring procedures. Impressive sensitivites down to less than pico-gram amounts have now been widely documented using a mass spectrometer as a “selective detector.” A typical demonstration32 of such a progress in high-sensitivity measurements is Fig. 4; here a perfluorinated derivative of the female hormone 17P-estradio1 has been measured selectively from a 120-pl aliquot of rat plasma.Capillary Liquid Chromatography Liquid chromatography (LC) becomes the obvious choice for a separation method if a sample lacks either volatility or chemical stability an effective use of GC thus being preven-ted. The rapid growth of high-performance liquid chromato-graphy (HPLC) during the last decade has made numerous new applications feasible; with this method sample complex-ity problems are typically attacked through the detection selectivity approach. There are however many instances where this approach falls short in providing adequate results without improved chromatographic efficiency. Some attempts have recently been made to provide an equivalent to capillary GC in the area of LC as far as resolving power is concerned.Although theoretical prediction of the separating potential for both GC and LC3I places the latter in a very competitive position LC column efficiencies of the order of 103 theoretical plates were the state of the art until recently 202 ANALYST MARCH 1984 VOL. 109 Since 1976 our research group at Indiana University has been exploring various ways of increasing the resolving power of LC techniques. Firstly the potential of (semi-permeable) packed capillaries34 and open tubular columns35 was briefly assessed. The small values of the solute diffusion coefficients in the mobile phase dictate that the column dimensions should be reduced drastically compared with the situation encoun-tered in capillary GC.Such column dimensions in turn, dictate extensive miniaturisation of the sample introduction and detection technology. While microlitre-volume instrumental components are quite acceptable in conventional HPLC nanolitre volumes are generally required to comply with the small dimensions and extremely low flow-rates associated with LC micro-columns. Approaches toward the miniaturisation of LC equipment have been first described in the pioneering studies of Ishii et a1.36 and Scott and Kucera.37 It appears that effective utilisation of open tubular columns in LC will not be feasible for at least several years because of the enormous technological difficulties associated with the approach. The optimisation studies by Knox and Gilbert38 and $! 100 2 8C $ 6C u 4t 1 / n = O 2 - 0 1 2 3 4 Time/h Fig.5 . Gradient elution of epoxy resin oligomers from a 22 m x 31 pm i.d. open tubular column bonded with an octadecylsilyl stationary phase. Column temperature 44 “C; UV detection at 225 nm. Reproduced from reference 40 with permission of Elsevier Publishing Company 3 I d Open-tubular capillary i.d. 15-50 pm L Stationary phase (a liquid or finely dispersed solid) Jorgenson and Guthrie39 show that the capillary inner diameters must be reduced to 10 pm or less. Without a major technological breakthrough it may be difficult to achieve this “ultimate goal.” Yet significant improvements have already been made through the work of Ishii et al.40 and Tsuda et al. ,4* who successfully explored open tubular columns with inner diameters below 30 pm.Such efforts are exemplified by Fig. 5,“) showing separation of synthetic oligomers on an open tubular column operated in a rc rersed-phase mode. it should be noted that “capillary LC” is not necessarily synonymous with “open tubular LC.” Additional types of columns that fall under this loosely defined category include the above mentioned semi-permeable packed capillaries34-42 as well as slurry-packed columns of capillary dimensions.j”-Js The geometrical charac-teristics of different micro-columns are illustrated in Fig. 6. While separating power i.e. large numbers of theoretical plates was among the chief incentives to research on micro-column HPLC originally realisation of some unique additional capabilities of this method quickly followed.These additional advantages are a direct consequence of drastically reduced flow-rates decreased consumption of expensive or environmentally hazardous mobile phases and the possibility of using “exotic” mobile phases for the sake of either improved separation or detection techniques. In order to receive such benefits of micro-column LC it is essential that the over-all chromatographic system be de-signed to match the geometrical characteristics of the micro-columns. Specifically sampling and detection volumes must be drastically reduced in order to minimise extra-column broadening of the chromatographic zones. Much effort has already gone into designing miniaturised LC systems; however new technological directions are still needed.Fortunately coherent laser beams and micro-electrodes, among other approaches are perfectly in tune with the very small (nanolitre) detection volumes that are required in micro-scale HPLC. While certain conventional detectors have now been miniaturised to volumes below 100 nl,36.4648 outstanding opportunities now exist for new directions in HPLC detection. In contrast with a typical flow-rate of 1 ml min-1 in conventional HPLC capillary micro-columns pass only a few pl min-1 into a detector. These drastically reduced flow-rates make it now feasible to couple micro-columns directly to the flame detectors+-sl or a mass spectrometer.s2 The flame and plasma detection devices in conjunction with micro-columns, may hold a key to the development of a highly desirable (so far largely undeveloped) element-selective detection in HPLC.An example of such flame detection devices is shown At this stage of developmei Packed capillary column Small-bore packed column i.d. 40-80 pm i.d. 0.2-1 mm \’ Particles of adsorbent [or a support with bonded phase (5-20 pmll Particles of adsorbent [may be chemically modified (10-30 pm)l Fig. 6. Types of LC micro-column ANALYST MARCH 1984. VOL. 109 203 Cot lector electrode Rubidium bead Analytical flame Hydrogen and air inlets Primary flame Air inlet Ca pi I la ry Hydrogen and nitrogen inlets Fig. 7. Dual-flame thermionic detector for micro-column LC 0 1 2 3 Time/h Fig. 8. High-efficiency separation of polycyclic aromatics from the neutral fraction of a coal-derived fluid on a 0.85 m X 240 pm i.d.capillary packed with 3 pm ODS particles. Reproduced from reference 44 with permission of Pergamon Press in Fig. 7. This particular detector is used in a thermionic mode500.jl for the selective detection of solutes containing phosphorus or alternatively nitrogen atoms in their struc-tures. Unlike its well known GC counterpart which utilises a single flame and a rubidium source the micro-scale LC detector has two flames the primary flame which causes combustion solute volatilisation and breakdown; and the analytical flame which can be optimised for a maximum response. Performance analytical attributes as well as some applications of this detector have recently been de~cribed.5~ In addition the flame photometric detection principle has also been successfully explored in micro-scale LC.49 Both types of detectors possess sensitivities down to the low nanogram levels; actually their typical sensitivities are only 1-2 orders of magnitude worse than those encountered in GC detection.As certain devices for sampling nanolitre volumes into HPLC micro-columns have gradually been deveI0ped39~46~~1 together with satisfactory column technology the last 2 years have marked the beginning of some unique applications of the rniniaturised LC techniques While development of open tubular HPLC still remains a matter for future studies packed capillary columns of the types shown in Fig. 6 offer currently the best compromise between column performance and sample capacity.With the use of slurry-packed capillary columns44 and miniaturised UV or fluorescence detectors our laboratory has been able to resolve complex non-volatile mixtures such as those derived from coal liquids52 or biological samples.53-55 An example is shown in Fig. 8; a 0.85 m x 240 pm i.d. column, packed with 3 pm ODS particles (generating over 110000 theoretical plates) has been used to resolve an aromatic mixture of up to nine-ring polycyclic ~ o r n p o u n d s . ~ ~ ~ ~ * Com-pounds that do not possess readily detectable features in their molecules can frequently be “tagged” with appropriate structural moieties prior to chromatography. Whereas the post-column derivative formation has proved popular in overcoming certain detection problems in the past there is significant rationale to explore increasingly the formation of suitable derivatives prior to micro-column LC.A certain 1OSS of the solute - solvent interaction selectivity can be compen-sated for by a significant increase in plate numbers and a choice of chemical derivatisation scheme is not restricted to fast reactions only. Recently our laboratory has successfully applied several useful pre-column derivatisation procedures to enhance detection capabilities for complex mixtures of steroids,53-55 prostaglandins,55 bile acids54 and biological ketones .56 With concentration-sensitive detectors mass sensitivity can be significantly enhanced by the use of micro-columns and small-volume detectors.57 This advantage has now been demonstrated for UV spectrofluorimetric and electro-chemical detectors.Outstanding examples of this have been demonstrated in a recent paper of Manz and Simon,58 who employed open tubular LC and a miniaturised ion-selective electrode to achieve femtomole sensitivities and a communi-cation of Knecht et a1.S9 on a micro-voltammetric detector. The use of a laser as an excitation source for fluorescence detection has the predictable effect of signal enhancement because of the high intensity of the incident beam. Several laser-based LC detection systems have been reported in the literature.6(bh3 Among them we find the helium - cadmium laser particularly attractive because of its relatively low cost and simplicity; in our laboratory certain reagents have been synthesised54.55 so that their optimum fluorescent properties are compatible with the 325-nm characteristic emission of the helium - cadmium laser.The use of appropriate laser detection systems in micro-column LC should result in sub-picogram detection possibilities for numerous biologically and environ-mentally important substances. Capillary Supercritical Fluid Chromatography The first attempts at supercritical fluid chromatography (SFC) can be traced to a pioneering study of Klesper et a1.,b4 who achieved the migration of porphyrins using a chromatographic column with a pressurised Freon mobile phase. Interesting but technologically demanding investigations of the 1960s into SFC were largely overshadowed by the advent of a new, powerful and simple method of HPLC. However certain obvious merits of SFC for both analytical and preparative purposes have recently been re-evaluated.In particular the idea of using open tubular columns in SFC65 has revitalised the general approach. When used under appropriate conditions certain super-critical fluids can combine advantages of both gases and liquids. Importantly the values of solute diffusivity and viscosity which are of the utmost importance to a chromato-graphic process are in SFC between those encountered in the gas and liquid phases. However the fluid densities approach those of liquids; gradually increasing densities ar 204 Pump -ANALYST MARCH 1984 VOL. 109 Prog ra m m e r t Pressure transducer important to achieve solvation of large non-volatile mole-cules during pressure-programmed chromatographic runs.A simple control of retention through pressure-induced equili-brium shifts without a change of solvent composition is of much importance to certain spectroscopic techniques (e.g. IR detection). In comparison with liquids some supercritical fluids give only minimum background signals in certain detectors. In addition they are far easier to remove from the measured solute molecules prior to detection in flames or a mass spectrometer.h” An optimisation study67 of capillary SFC indicates that, under reasonable conditions efficiencies in the range of 105-106 theoretical plates can be attained. Such high efficien-cies are now feasible with capillary inner diameters between 50-100 pn. This is now well within the technological possibilities of low-volume sampling and detection as previ-ously developed for micro-column LC.A typical capillary SFC system65 is shown in Fig. 9. A high pressure syringe pump is filled with an appropriate fluid which is subsequently transformed into a supercritical medium in a heated oven. A pressure controller adjusts this fluid to an appropriate density, while the sample is introduced through a low-volume valve. After the sample has separated into its components during the passage through a fused-silica capillary column the solutes are detected in a high-pressure cell followed by decompression at the capillary restrictor end. While many substances could be converted into super-critical fluids only certain mobile phases are of immediate interest. The obvious advantage of SFC is its capability to handle thermally labile molecules without decomposition; for instance.supercritical carbon dioxide can be used as the mobile phase slightly above the ambient temperature to separate labile organics such as peroxides or azo compounds, with efficiences comparable6s to capillary GC. The choice of a supercritical-fluid mobile phase can be made approximately according to the “polarity” scale suggested by Giddings er al. ,(j9 based on values of the Hildebrand solubility parameter. Table 1 lists the critical parameters of several useful mobile phases. An adequate polarity range can be covered while using these supercritical media at very reason-able values of pressure and temperature. As capillary columns typically pass a few microlitres of a fluid per minute it is feasible to employ “exotic” mobile phases.The solute retention in SFC is primarily controlled by the column pressure and correspondingly by the density of solvating molecules. 7o Thus density programming is primarily em-ployed while some selectivity adjustments can also be made with the application of a polar “modifier”.71.7~ restrictor Fig. 9. Capillary supercritical-fluid chromatograph Owing to the relatively large solute diffusion coefficients in C2-C5 hydrocarbons,73 these media are among the best for capillary SFC separations. A typical application using super-critical pentane as the mobile phase is shown in Fig. 10; the aromatic fraction of coal tar has been separated here with a 110 pm i.d. glass capillary column and detected with a spectrofluorimetric detector.74 One of the most interesting aspects of capillary SFC is the fact that some LC as well as GC detection techniques can be used.A wide application of flame-based and plasma detection devices in future investigations is anticipated. Some promising results have already been reported with a combination of capillary SFC and mass spectrometry.66 Recently. the relative transparency of some supercritical fluids in the infrared region has been utilised to couple SFC to a Fourier-transform infrared ~pectrometer;7”7~ this combination is very likely to make any attempts to achieve LC - infrared spectrometry obsolete. Capillary Electrophoresis While electrophoresis is generally considered to be a superior approach to the separation of ionic macromolecules it is also probably the least understood and most underrated of all separation techniques.Flat-bed techniques are considerably more common in electrophoresis than the column techniques: a recent considerble success of the two-dimensional method of O ’ F a ~ - e l l ~ ~ attests to the great versatility of the planar approach in high-resolution analysis of biopolymers. The use of capillary columns in electrophoresis is uncom-mon. A notable exception is the method of isotachophoresis (displacement electrophoresis). In addition zone electro-phoresis in narrow PTFE tubes was investigated by Mikkers et a1.78 in order to reduce the zone dispersion originating from the convection currents. Table 1. Critical parameters of some mobile phases Critical Critical temperature/ pressure/ Fluid “C atm Carbon dioxide .. 31.3 72.9 Dichlorotetra-fluoroethane . . 146.7 33.5 Propan-2-01 . . . . 253.3 47.0 Propane . . . . 96.7 41.9 Butane . . . . . . 152.0 37.5 Pentane . . . . 106.5 33.2 Critical density1 gcm 0.448 0.582 0.273 0.217 0.228 0.237 L 18 18 21 24 27 30 33 2 Pressu reiatm 0 0.5 1 .o 1.5 Timeih 2.0 Fig. 10. Chromatogram o f tluorescent components from the aro-matic fraction of coal tar as recorded by the capillary supercritical-fluid chromatograph. Reproduced from reference 73 with permission of the American Chemical Societ ANALYST MARCH 1984 VOL. 109 205 Fig. 11. Capillarv electropherogram of a hydrolysed protein sample. Reproduced fromreference 80 with permission of Elsevier Publishing Com pan y More recently.Jorgenson and Lukacs7”x0 have described a very promising version of capillary electrophoresis. Noting a theoretical prediction by Giddingssl that the zone breadth is inversely proportional to the applied voltage Jorgenson and Lukacs7” designed a system in which a relatively short capillary attached to the respective reservoirs was subjected to voltages in excess of 30 kV. An application of such high voltages in previous electrophoretic experiments had largely been limited by the heat generated during the process. However in the experimental arrangement of Jorgenson and Lukacs,79 using a narrow capillary tube (75 pm or less) heat was easily dissipated through the wall. Whereas the experimental set-up of capillary electrophore-sis7q is somewhat reminiscent of capillary LC the results with the former are considerably better as long as the substances are ionic in nature.Specifically large ‘.plate numbers’’ can be generated in relatively short analysis times because unlike chromatography any diffusion-controlled processes are to be minimised. Efficiencies in excess of 400 000 theoretical plates have been demonstrated. Fig. 11 shows an example of a very efficient separation; here fluorescamine-labelled peptides from a tryptic digest of chicken ovoalbumin present a very complex pattern .SO Except for employing unusually high voltages the method of capillary electrophoresis is relatively simple. Glass or fused silica capillaries (typically 1-m long) are used while the separated solutes are detected spectroscopically using the “on-column approach.“ Other instrumental aspects have been de~cribed.7~,8() Further developments in capillary electrophoresis are highly desirable from the theoretical as well as practical point of view.Eventually this technique could entirely replace an a 1 y t ic a1 ion -exchange c h r om at ogr a p h y . Conclusions In spite of the early powerful demonstrations of capillary GC techniques applications of this method during the 1960s were infrequent and confined to a few research laboratories. Technological innovations of the following decade have brought capillary GC close to the full realisation of its theoretical potential. Developments in micro-column (capil-lary) LC initiated in the late 1970s have been stimulated both by an improved understanding of the column processes and by m 1 n i at u r i s a t i o n techno 1 o g y .W h i 1 e co 1 u m n efficiency i m p r ov e -ments achieved by micro-scale LC remain important this approach is even more essential to the development of optimum detection devices. Some of the5e directions are further reinforced by the advent of capillary supercritical fluid chromatography. Finally high-voltage capillary electrophore-sis further complements the other existing capillary techniques with respect to highly efficient separations of ionic species. Although there are numerous challenges and experimental difficulties associated with modern capillary separation tech-niques the potential benefits are eminently worthwhile attaining.During the past 5 years research on capillary and micro-column chromatography in the author’s laboratory has been aided by the following grants GM 24349 (National Institute of General Medical Sciences US Public Health Service); DOE DE-AC02-81 ER 60007 (US Department of Energy); N14-82-K-0561 (Office of Naval Research); and NSF CHE 82-00034 (National Science Foundation). 0 10 20 Ti m e h i n 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. 26. 27. 28, 29. 30. 31. References Golzy M. J . E. in Desty D . H. Editor “Gas Chromato-graphy 1958,” Academic Press. New York 1958 p. 36. Dandeneau R. and Zerenner E.H J . High Resolut. Chromatogr. 1979. 2 351. Desty D. H and Goldup A . in Scott R. P. W. Editor “Gas Chromatography 1960,” Butterworths London 1960 p. 162. Bruner F. A . and Cartoni G. P. Anal. Chem. 1964. 36, 1522. Grob K. Helv. Chim. Acta 1968 51 718. Bartle K. D. Bergstedt L. Novotny M. and Widmark G., J . Chromatogr. 1969 45 256. Novotny M. Anal. Chem. 1981. 53 1294A. Jennings W. G. “Applications o f Glass Capillary Gas Chromatography,’’ Marcel Dekker New York. 1981. Novotny M. in Natusch D . F. S. and Hopkc P. K. Editors, “Analytical Aspects o f Environmental Chemistry.” Wiley and Sons New York 1983 p . 61. Grob K. and Grob. K. J r . J . Chromatogr. 1974. 94. 53. Van dcn Berg. M. J . and Cox. T. P. H Chrornatographia, 1972 5 301. Verzele M.Redant G . Qureshi. S . and Sandra P J . Chromalogr. 1980 199. 105. Novotny M. Lee M. L. and Bartle K. D. Chromato-graphia 1974. 7 333. Novotny M. and Farlow R . J . Chromarogr. 1975 103 1 . Vogt W. Jacob. K. Ohnesorge A.-B. and Obwexer H. W J . Chromatogr. 1979. 186 197. Schomburg G . Husmann H . and Weeke F. J . Chronia-togr. 1974. 99 63. Schomburg G . in Kaiser R. E Editor “Proceedings of the 4th International Symposium on Capillary Chromatography,” Huthig Heidelberg. 1981 p. 371. Grob K . and Grob K . Jr. J . Chromutogr. 1978 151. 311. Galli M and Trestianu. S. J . Chromatogr. 1981. 203 193. Grob K. Grob G and Grob K . Jr J . Chrornarogr. 1981, 211. 243. Stark T Wright €3. W Peaden. P. A and Lee. M. L J . Chromutogr. 1982 248 17. Springston S.R Melda K. and Novotny. M. J . Chroma-togr. 1983 267 395. Lee M. L and Wright €3. W J . Chromutogr. 1980,184,235. Guiochon G . Anal. Chem. 1978. 50. 1812. Schutjes C. P. M. Vermeer. E. A . Rijks J . A and Cramers, C. A . in Kaiser R . E Editor “Proceedings of the 4th International Symposium on Capillary Chromatography,” Huthig Heidelberg. 1981 p. 687. Hartigan M. J . Purcell J . E. Novotny. M. McConnell. M. L and Lee M. L J . Chromatogr 1974 99 339. Estes S . A. Uden P. C . and Barnes R. M. J . Chromutogr., 1982 239 181. Grob K . Ctzromatogruphia 1975 8 423. Novotny M. Schwcndc F. J . . Hartigan. M. J . . and Purcell, J . E. Anal. Chem. 1980. 52. 736. Jaramillo J . F and Driscoll. J . N J . High Resolut. Chromatogr. 1979 2 536. Smith S . L. and Adams.G. E. J . Chromutogr. 1983. 279, 623 ANALYST MARCH 1984 VOL. 109 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. Tetsuo M. Eriksson H and Sjovall J . . J. Chromarogr., 1982 239 287. Giddings. J . C. Anal. Chem. 1964 36 1890. Tsuda T and Novotny M. Anal. Chem. 1978. 50 271. Tsuda. T. and Novotny M. Anal. [,'hem. 1978 SO 632. Ishii D. Asai. K Hibi K. Jonokuchi T. and Nagaya M. J. Chromatogr. 1977 144. 157. Scott. R. P. W and Kucera. P J. Chromatogr. 1977. 144, 157. Knox J. H. and Gilbert. M. T. J. Chromatogr. 1979. 186, 40s. Jorgenson J. W and Guthrie E. J . J. Chromafogr. 1983, 255 335. Takeuchi T. and Ishii D. J. Chromatogr. 1983 279 430.Tsuda T. Tsuboi. K. and Nakagawa G . . J . Chromutogr., 1981 214 283. McGuffin V. L. and Novotny M. J. Chromatogr 1983,255, 381. Yang F. J . J. Chromatogr. 1982 236 265. Gluckman J. C. Hirose A. McGuffin V. L. and Novotny, M. Chromatographia 1983. 17 303. Hirata. Y . and Jinno K. J. High Resolut. Chromatogr. 1983, 6 196. Hirata Y. and Novotny M. J. Chromatogr. 1979 186. 521. Hirata. Y . Lin P. T. Novotny M. and Wightman R . M. J . Chromatogr. 1980 181 287. Slais S . and Krejci M. J. Chromatogr. 1982 235 21. McGuffin. V. L. and Novotny M. Anal. Chem. 1981 53, 946. McGuffin V. L. and Novotny M . J . Chromutogr. 1981.218. 179. McGuffin. V. L. and Novotny M. Anal. Chem. 1983 55, 2296. Novotny. M. Hirose. A and Wiesler D. submitted for publication.Novotny M Alasandro M and Konishi M. Anal. Chem., 1983 55 2375. Novotny M. Karlsson K.-E. Konishi. M. and Alasandro, M. J. Chromatogr. in the press. Karlsson K.-E. Alasandro M Wiesler D. and Novotny, M. submitted for publication. Raymer J . Holland M. L. Wiesler D. and Novotny M submitted for publication. Scott R. P. W. and Kucera P . J. Chromatogr. 1979,185.27. Manz. A and Simon W. J. Chromatogr. Sci. 1983.21.326. Knecht. L. Guthrie E. J . and Jorgenson J . W.,Anal. Chem., in the press. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. Folestad S . . Johnson L. Josefsson. €3 and Galle B. Anal. Chem. 1982 54 925. Diebold G. J . and Zare R. N . Science 1977 196 1439.Hershberger L. W Callis J . B and Christian G. D. Anal. Chem. 1979 51 1444. Sepaniak. M. J . and Yeung E. S . . J. Chrornatogr. 1981. 211, 95. Klesper E. Corwin A. H . and Turner. D. A J. Org. C'hrm., 1962 27. 700. Novotny M Springston S. R. Peaden. P. A. Fjeldsted, J . C and Lee. M. L. Anal. Chem. 1981 53 407A. Smith R . D. Felix W. D. Fjeldsted J . C and Lee M. L., Anal. (,'hem. 1982 54 1883. Springston S . R and Novotny M. Chromatographia 1981. 14 679. Lee M. L Fjeldsted J . C. Campbell R. M and Kong, R. C J. Chromarogr. in the press. Giddings J. C. Myers M. N . McLaren L. and Keller. R. A. Science 1968 162 67. Fjeldsted J . C. Jackson W. P. Peaden P. A and Lee, M. L J. Chromatogr. Sci. 1981 21 222. Novotny M Bertsch. W. and Zlatkis A J. Chromutogr 1971 61 17. Gere. D. R. Board R and McManigill D Anal. Chrm., 1982 54 736. Springston S. R. and Novotny M. J. Chromutogr. 1983, 279 417. Peaden P. A . Fjeldsted J . C. Lee M. L. Springston S . R., and Novotny M. Anal. Chem. 1982 54 1090. Shafer K. H. and Griffiths P. R . Anal. Chem. 1983 55, 1939. Olesik S . V. French S. Smith S . L and Novotny M submitted for publication. O'Farell P. H . J . Biol. Chem. 1975 250 4007. Mikkers F. E. P. Everaerts F. M. and Verheggen. P. E. M., J. Chromatogr. 1979 169 11. Jorgenson J. W. and Lukacs K. D. Anal. Chem. 1981 53, 1298. Jorgenson J. W. and Lukacs K. D. J. Chromatogr. 1981, 218 209. Giddings J . C. Sep. Sci. 1969 4 181. Paper A31385 Received November 7th 198

ISSN:0003-2654    

DOI:10.1039/AN9840900199

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST MARCH 1984 VOL. 109 207 Design and Application of Neutral Carrier-based Ion-selective Electrodes Plenary Lecture W. Simon,* E. Pretsch W. E. Morf D. Ammann U. Oesch and 0. Dinten Department of Organic Chemistry Swiss Federal Institute of Technology (ETH) CH-8092 Zurich, Switzerland The requirements for neutral complexing agents behaving as ion carriers (ionophores) in membranes of ion-selective electrodes are discussed. Emphasis is placed on the lipophilicity of such compounds on the kinetics (free energy of activation) of the ligand exchange reaction and on the parameters inducing selectivity, e.g. molecular structures and aspects of membrane technology. Keywords /on-selective electrodes; neutral carrier-based Ion carriers (a class of ionophores) are lipophilic complexing agents having the capability to bind ions reversibly and to transport them across organic membranes by carrier trans-location.' Ideally selective ion carriers render a membrane permeable for one given sort of ion I only.If such ionophore-based membranes are used in cell assemblies of the type Sample Internal reference reference 1 ;ohtion 11 Membrane (1 system solution 1) (solution 2) electrode * To whom correspondence should be addressed. YNY and if an electric potential gradient is applied between solutions 1 and 2 an exclusive transport of the ions I across the membrane should result the transport number being then tI = 1. For the same membrane electrode cell the electric potential difference at zero current (e.m.f.) between the external and the internal reference system would depend on the ratio of the activities of the ion I in solutions 1 and 2.If a constant composition of solution 2 is used the activities of the selected ion in solution 1 may therefore be measured potentiometric-ally according to the Nernst equation.* Neutral carriers in particular defined as ionophores that carry no charge when 'N-o Mg2+ 0 ETH1117 c Ca2+ ETH 1001 I nN-ETH 295 I Valinomycin K+ c. -N- bS (O Cd2+ ETH 1062 qs -N-U f N f Fig. 1. Structures of the cation-selective carrier 208 ANALYST MARCH 1984 VOL. 109 not complexed by the transported ion have led to such ion-selective electrodes with a wide range of available selectivities. 1 Analytically relevant ionophores have to meet at least the following three requirements: 1.For a continuous-use lifetime of solvent polymeric membranes3 [ca. 32% mim poly(viny1 chloride) ca. 65"/0 mlm plasticiser 1-3% mim ionophore] of at least 1 year the ionophore has to be very lipophilic. A partition coefficient K , of the carrier between the aqueous sample solution and the membrane phase larger than 10' 5 is necessary.4 Incorporating adequate structural elements (e.g. alkyl groups) into iono-phores. the required lipophilicity may easily be obtained. A reliable estimate for P (partition coefficient of a carrier between water and octan-1-01) and therefore of K may be obtained by thin-layer chromatography (TLC)5 (Figs. 1 and 2). From Fig. 2 it becomes obvious that e.g. the ligand ETH 231 does not exhibit a sufficient lipophilicity for the required membrane lifetime.2. The free energy of activation of the ligand exchange re action rs + s' IS' + s where S and S' are ionophores has to be relatively low. For Zn2f or Cd2+ complexes of ligand ETH 1062 (Fig. 1), free energies of activation of the ligand-exchange reaction of 5 45 kJ mol-1 (in acetonitrile) have been measured.6 Cation permselectivity is indeed observed with CdC12 in the sample solution (see Fig. 3). As CdC1+ is probably the permeating species a slope of the electrode response of approximately 60 mV (25 "C) is obtained.7.8 In systems with a free energy of activation of the ligand exchange reaction of 2 65 kJ mol-1 (in acetonitrile) however the cationic complexes of the iono-phore act as anion exchangers (e.g.complexes with Pt:+ or Pd2+). An electrode containing the PdC12 complex of ETH 1062 in the membrane phase therefore responds to the chloride anions in a sample solution of CdC12 (see Fig. 4). In order to keep the free energy of activation of the ligand exchange reaction sufficiently small the design of ionophores has been focused on non-macrocyclic structures (see Fig. 1). 3. The ionophore has to induce ion-permeability selectivity in membranes. As the ion selectivity of a membrane is related to the free energies of transfer of the ions from the aqueous phase (sample) to the membrane phase the selectivity of neutral carrier-based ion sensors obviously depends on various factors. Such factors are mainly1 (a) the selectivity behaviour of the carrier ligand used which can be completely Valinomycin / - 1 0 1 2 1 Retention [R = log (- - 111 RF Fig.2. Lipophilicities of the ion carriers as determined by their TLC retention RF. The TLC system is calibrated with a set of reference compounds (open circles) of known lipophilicities. TLC plate RP KC 18F; ethanol - water (70 + 30); 23 "C. (See Fig. 1 for the chemical structures of the ion carriers depicted) characterised by complex stability constants (b) the extrac-tion properties of the membrane solvent (plasticiser) (c) the concentration of the free ligand in the membrane phase and (d) the concentration of ionic sites in the membrane. The effects of (b) (c) and (d) can be described to a large extent by model calculations and may be controlled by adequate membrane technology 179,10 The design of ionophores with a given selectivity is more probfematic (a).Although an overwhelming activity in designing host molecules (ionopho-res) for selected guest species (ions) is in evidence,ll-lx only modest use has been made of model calculations describing the host - guest interaction. 12.19-22 Classical electrostatic models are useful for calculating the ion - molecule interactions near the energy minima for Group IAiIIA cations.21 However they require a knowledge of molecular parameters normally not available. It has been shown21 that semi-empirical quantum chemical treatments of ion - ligand interactions often lead to unrealistic results. In contrast ab initio computations give reliable results even if small but well balanced basis sets are chosen.23-26 The application to realistic ionophores is usually prohibited by the extent of the computation even if small basis sets are used.Through ab initio calculations on model complexes and a representation of the interaction energies by pair poten-tial~,~7-31 such large systems are easily made amenable to analysis (see also reference 26). For a discussion of the stability of the complexes the sum of the interaction energies and the conformational energy relative to the most stable conformation of the free ligand is relevant. Unfortunately the computation of such conformational energy changes is still too uncertain . 3 2 3 Using such model calculations. CPK model building and adequate membrane technology it has nevertheless been possible to design neutral carrier-based membranes that show analytically relevant ion selectivities for Li+ Na+ K+ Mg2+, Ca2+ Ba2+ Cd2+ NH4+ U022+ and H30+.Pungor et al. revealed a useful lead selectivity for ligands of the type ETH 32234 (Fig. 1). Although the synthetic ionophores shown in Fig. 1 are non-macrocyclic (see section 2). there are several reports on the successful application of macrocyclic iono-phores (e.g. crown compounds) in ion-selective elec-trode+-36 (for a review see reference 1). Recently. neutral carriers for anions have been unravelled37. Some of the neutral carriers shown in Fig. 1 have found wide application especially in clinical chemistry and physiology. About 200 references for such applications are given in a 'c .' i E ui -6 - 4 - 2 Log aCd2+ Fig.3. E.m.f. response to CdCl? solutions and selectivity coeffi-cients (KCCIM for different cations Mi+) of membrane electrodes based on the carrier ETH 1062. Membrane composition carrier lo/; m'm: (10-hydroxydecy1)butyrate. 65% minz; PVC 34% min ANALYST MARCH 1984 VOL. 109 209 T + E ui -N-A S + 100% PdCI2 mV I ’\ 0 - 1 ,x -2 02 5 -3 -I - 4 -5 -6 I I I I -6 - 4 -2 0 Log acl-Fig. 4. E.m.f. response to CdC1 solutions and selectivity coeffi-cients (Kclo for different anions X-) of membrane electrodes containing th% carrier ETH 1062 and an equimolar amount of PdCI,. Membrane composition carrier + additive 1.8% mim; o-nitrophenyl octyl ether 66% mim PVC 32.2% mim recent review.’ Most recent reports on the use of some selected ionophores can be found in references 38 and 39 (Mg2f - ionophore ETH 1117) 40-42 (H+ - ionophore, tridodecylamine) and 43 (U022+ - ionophore ETH 295).This work was partly supported by the Schweizerischer Nationalfonds zur Forderung der wissenschaftlichen For-schung. 1 . 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 13. References Ammann D . Morf. W. E . Anker P . Meier. P. C. Pretsch, E . . and Simon W. Ion Select. Electrode Rev. 1983 5 3. Nernst W. Z . Phys. Chem. 1889 2 613; 1889 4. 129. Moody G . J . . Oke R . B. and Thomas J . D. R. Analyst, 1970 95 910. Oesch. U. and Simon W. Anal. Chem. 1980 52 692. Oesch. U. Dinten O. Ammann. D . and Simon W. in Kessler M. Editor “Proceedings of the International Sympo-sium on the Theory and Application of Ion-selective Electrodes in Physiology and Medicine Burg Rabenstein FRG 12-15th September 1983,” Springer-Verlag Berlin 1984 in the press.Hofstetter P . . Pretsch E . and Simon W. Helv. Chim. Acra, 1983 66 2103. Hofstetter P. Dissertation ETH No. 7128 ADAG Adminis-tration und Druck AG Zurich 1982. Schneider J. K Hofstetter. P . Pretsch. E Ammann D. and Simon W. Helv. Chim. Acta. 1980. 63 217. Meier P. C . Morf. W. E . Liiubli M. and Simon W. Anal. Chim. Acta in the press. Morf. W. E . and Simon W. in Freiser. H . Editor “Ion Selective Electrodes in Analytical Chemistry,” Plenum Press, New York. London Washington Boston 1978 p. 211. Cram. D. J . and Cram J . M. Science 1974 83 803. Lehn. J.-M Struct.Bonding (Berlin) 1973 16 1. Weber E . . and Vogtle F Chem. Ber. 1976 109. 1803. Pedersen. C. J . . J . Am. Chem. Soc. 1967 89 2495 1967 89, 7017; 1970 92 386; 1970 92 391. 1.5. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41, 42. 43. Prelog V. Pure Appl. Chem. 1978 50 893. Stoddard J . F. Chem. SOC. Rev. 1979 8 8.5. Poonia N. S . and Bajaj. A. V. Chem. Rev. 1979 79 389. Morf W. E. Ammann D . Bissig R . Pretsch E. and Simon, W. in Izatt R. M. and Christensen J . J . Editors “Progress in Macrocyclic Chemistry.” Volume 1 Wiley-Interscience. New York 1979. p. 1. Morf W. E. and Simon W Helv. Chim. Acta 1971,54,2683. Simon W . Morf. W. E. and Meier P.C. Struct. Bonding (Berlin) 1973 16 113. Schuster P . . Jakubetz W. and Marius W. Top. Curr. Chem. 197.5 60 1. Timko J. M. Moore S. S . Walba D . M. Hiberty P. C. and Cram. D. J . J . Am. Chem. Soc. 1977 99. 4207. Pullman A. Berthod. H . and Gresh N. Int. J . Quantum Chem. Symp. 1976 10 59. Kolos W. Theor. Chim. Acta 1980 54 187. Gianolio L. and Clementi E . Gazz. Chim. Ital. 1980 110, 179. Gresh N. and Pullman A. Int. J . Quantum Chem. 1982,22. 709. Clementi E. “Lecture Notes in Chemistry,“ Volume 2. Springer-Verlag Berlin 1976 Vol. 19. Springer-Verlag Ber-lin 1980. Corongiu G . Clementi E . Prctsch E. and Simon W. J . Chem. Phys. 1979 70 1266. Corongiu G . Clementi E. Prctsch E . and Simon W. J . Chem. Phys. 1980 72 3096. Pretsch. E .Bendl J . Portmann P . and Welti M. in Naray-Szabo G . Editor “Proceedings of the Symposium on Steric Effects in Biomolecules Eger Hungary 1981.” Akademiai Kiado. Budapest 1982 p. 85. Pretsch E Gratzl M. Pungor E . and Simon W. in Pungor, E . and Buzas I . Editors “Proceedings of the 3rd Symposium on Ion-Selective Electrodes Matrhfured Hungary October 13-15. 1980,” Akademiai Kiado Budapest 1981 p. 315. Bendl J. and Pretsch E . J . Comput. Chem. 1982 3 580. Uiterwijk J . W. H . M. Harkema S . van de Wad B. W., Gobel F . and Nibelling H. T. M. J . Chem. Soc. Perkin Trans. 2 in the press. Lindner E. Toth K. Pungor E . Behm F Welti D., Ammann D. Oggenfuss P . Morf W. E. Pretsch E. and Simon W. Anal. Chem. submitted for publication. Kimura K. Tamura H. and Shono T.J . Chem. Sac. Chem. Commun. 1983 492. Bussmann W. Lehn J.-M. Oesch U . Plumere P. and Simon W. Helv. Chim. Actu 1981 64. 657. Wuthier U . Pham H . V. Zund R . Welti D. Funck, R . J . J Bezegh A . Ammann D. Pretsch E . and Simon, W. Anal. Chem. accepted for publication. Hess P. Metzger P. and Weingart R . J . Physiol. 1982,333, 173. Alvarez-Leefmans F. J . Gamino S . M. and Rink T. J . . J. Physiol. in the press. Kurkdjian A . C . and Barbier-Brygoo H . Anal. Binchem., 1983 132 96. Kraig R . P Ferreira-Filho C. R . and Nicholson C . , J . Neurophysiol. 1983 49 831. Coray A. and McGuigan J. A. S . . J . Physiol. 1982,336,66P. Bertrand P. A . . Choppin G. R. Rao L. F. and Bunzli J.-C. G . Anal. Chem. 1983 5 5 . 364. Paper A31302 Received September 5th 198

ISSN:0003-2654    

DOI:10.1039/AN9840900207

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST MARCH 1984 VOL. 109 21 1 Continuum-source Atomic-absorption Spectrometry Past Present and Future Prospects Plenary Lecture Thomas C. O’Haver Department of Chemistry University of Maryland College Park MD 20742 USA The possibility of using a continuum primary source in atomic-absorption spectrometry rather than separate line sources for each element as is the conventional practice has been a recurring theme in the literature of atomic absorption for many years. Past efforts have not been particularly successful. Recent advances in optical technology and data processing techniques have made this approach much more promising. Simultaneous multi-element atomic absorption with flame and electrothermal atomisation is now being carried out routinely with prototype instrumentation built by researchers.This approach may offer a future path for the continued development of atomic absorption. Keywords Atomic-absorption spectrometry; continuum source The Past When Sir Alan Walsh first proposed atomic-absorption spectrometry as a method for chemical analysis in 1955 it was concluded that the most practical approach would be to use narrow spectral line sources rather than the continuum light sources that had been commonly used in molecular absorption spectrometry. This turned out to be an important factor in the ultimate success of atomic absorption for it meant that an instrument of very high effective resolution could be construc-ted using a small low-cost medium-resolution monochroma-tor. Prior to Walsh’s work analytical atomic spectrometry was performed most commonly by arc or spark emission spectro-graphy utilising very large high-resolution spectrometers or by flame emission photometry ordinarily utilising more modest optical equipment.Walsh’s atomic absorption system was not much more complex or expensive than a simple flame photometer yet it offered useful sensitivity for a wider range of elements and it was far less susceptible to spectral interferences. Thus the “image” of atomic absorption was established early on as an effective yet low-cost and easy-to-use method. In spite of the valid reasons for the use of narrow spectral line sources in atomic absorption researchers have for many years sought ways to utilise continuum sources effectively. In the early days of atomic absorption continuum sources were occasionally utilised to survey the absorption lines of elements for which hollow-cathode lamps were not yet available.’ In 1962.Gibson et al.3 suggested the use of a continuum source for analytical purposes primarily because of the convenience of using a single light source for many elements. Early experience with the continuum source used with conventional atomic-absorption instrumentation however showed that the expected disadvantages of poor sensitivity and restricted calibration graph linearity were indeed serious limitations.j.5 Some investigators attempted to improve the sensitivity of measurement by utilising organic solvents ,6 multi-pass opticsh or long-path absorption tubes.7 The difficulty with these studies is that the same measures could just as easily have been applied to line-source measurements with the same expecta-tion of sensitivity improvement.Moreover these techniques may lead to more serious chemical and spectral background interferences and thus cannot be recommended for general analytical work. Since about 1970 work on continuum-source atomic absorption has centred on improving the sensitivity by increasing the effective spectral resolution and on reducing noise by utilising high-intensity light sources and various modulation schemes. Wavelength modulation first applied to continuum-source atomic absorption by SnellemanX in 1968 is effective in compensating for broad-band atomiser back-ground emission and background absorption and for reducing the effects of low-frequency source flicker noise.Elser and Winefordnerg tried a double modulation scheme utilising both source chopping and wavelength modulation which corrected also for analyte atomic emission. Attempts to improve the effective resolution have been based on the use of interferometers resonance monochroma-tors selective line modulation and echelle spectrometers. Nitis et al. 10 and Veillon and Merchant” used Fabry - Perot interferometers in conjunction with medium-resolution monochromators. They were able to achieve characteristic concentrations comparable to line source measurement but the complex optical systems did not prove suitable for routine work. Cochran and Hieftje12 used an interesting selective-line modulation technique that utilises two identical burner systems one for the analytical solution and one for a reference solution containing a high concentration of the sought-for species.A mirrored chopper directs the primary continuum radiation alternately through and around the reference flame. In this way the intensity of the primary beam is intensity-modulated only in the very narrow spectral region of the absorption profile to the analyte. Very high effective resolu-tion is thereby achieved with only a small conventional monochromator. A related approach is the use of a resonance monochromator which has been described by Bower ef al.“ and Blackburn and Winefordner. 14 A comparatively straightforward way to achieve high spectral resolution is simply to use a high-resolution spectro-meter.In the past these have been large and costly but newer designs based on the echelle grating are much more practical, although still larger and more expensive than small medium-resolution monochromators. Keliher and Wohlers’5.lh were the first to apply an echelle spectrometer to continuum-source atomic absorption. They showed that it was indeed possible to obtain characteristic concentrations and calibration graphs comparable to line-source atomic absorption. One of the most significant aspects of this approach is that it is optically and operationally simple at least externally. Although rhe optics of an echelle spectrometer are sophisticated this is an internal complexity. As far as the user is concerned an echelle spectrometer is a commercially available “black box,” much like any other monochromator.‘The operation of an echelle spectrometer is fairly simple and familiar in comparison with interferometers resonance monochromators and selective line modulation devices. The most extensive recent work in continuum-source atomic absorption has been based on the system first described in 1976 by Zander et al. 17 This system which combines the use of a high-resolution echelle spectrometer with Snelleman’s wavelength modulation technique exists in two differen 212 ANALYST MARCH 1984 VOL. 109 configurations a single-element version referred to as a CEWM-AA system and a 16-channel simultaneous multi-element version called SIMAAC. The CEWM-AA and SIMAAC system have now been applied to over 30 elements and to the analysis of large numbers of practical samples and reference materials using both flame and electrothermal at omis at i o n .I * - 40 The Present A schematic diagram of a single-channel continuum-source atomic-absorption spectrometer employing wavelength modulation is shown in Fig. 1. A SIMAAC system is illustrated in Fig. 2. There are five main parts the continuum primary source the atomiser the spectrometer the wave-length modulator and the signal processing electronics or computer. As a primary source we have used xenon short-arc lamps with integral parabolic reflectors. These lamps were initially manufactured by the Eimac division of Varian and are now available from ILC Technology under the name of Cermax. This type of lamp has been extensively applied in various areas of analytical spectrometry where it has been found to offer significant intensity advantages over conventional designs.In its application to continuum-source atomic absorption it is of interest to compare the intensity of the Cermax lamp with that of the hollow-cathode lamps normally used in line-source atomic absorption.22 In Fig. 3 the circles represent the photoanodic currents measured at the resonance lines of various elements when a 300-W Cermax lamp is used in a typical continuum-source atomic-absorption set-up with an ~ I F 1 Function 12F 4 Lock-in 1 1 Digital generator amplifier voltmeter I 1 - 1 Fig. 1. Schematic diagram of a single-channel continuum-source wavelength-modulated atomic-absorption spectrometer with ana-logue signal processing electronics.Reprinted by permission from reference 35 controller EIMAC Furnace Power I Amp c*l Oscilloscope i I I I 'Tq-J7l Computer I IL I Fig. 2. Schematic diagram of a simultaneous multi-element atomic-absorption spectrometer with a continuum source (SIMAAC). Reprinted by permission from reference 27 echelle spectrometer (slit width 25 pm; resolution 0.003 nm at 300 nm). The vertical lines represent the corresponding measurements of commercial hollow-cathode lamps operated at the recommended currents and measured with the same optical system. One can see that the total system response decreases rapidly at short wavelengths but that in spite of that the continuum source is at least as bright as the hollow-cathode lamps even at 193.7 nm and is considerably brighter in the visible region.The atomisers we have used have been conventional, commercially available designs. We have experience with both air - acetylene and dinitrogen oxide - acetylene flames and with various models of tubular carbon furnace atomisers (mostly Perkin-Elmer models) as well as the Instrumentation Laboratory Fastac system. The use of a L'vov platform or other atomisation substrate has often been found valuable. The spectrometer is a Spectraspan I11 (Spectrametrics Inc., Andover MA USA) echelle spectrometer. Wavelength modulation is accomplished by a 3-mm quartz refractor plate mounted on an optical scanner torque motor. The refractor plate is positioned immediately behind the entrance slit.The selection of the slit widths and heights of the echelle spectrometer is a compromise between signal to noise ratio and calibration graph linearity. The most linear calibration graphs are obtained at small slit widths which improve resolution and small slit heights which minimise order-overlap stray light. However the resulting lower light transmission reduces signal to noise ratio and degrades detection limits. Harnly32 has shown experimentally that the best detection limits are obtained with larger slit heights and widths. In general the choice depends on the type of instrument system. In single-channel systems with analogue electronics in which the analytical calibration is ordinarily performed manually small heights and widths have been used in order to improve linearity and thereby simplify the calibration and curve-fitting procedure.On the other hand, when a computer is used for data acquisition and reduction as in the multi-element SIMAAC system the choice has been to optimise detection limits. The multi-element SIMAAC system utilises a multi-channel cassette with 20 exit slits and photomultipliers positioned for 20 pre-selected wavelengths. The selection of the optimum line for each element is comparatively straightforward; in most instances the usual resonance line is used. Because of the spectral simplicity of atomic absorption alternative lines do not have to be used to avoid spectral interferences as is often the case in plasma emission spectrometry. The exit slit heights and widths can be chosen individually.Inasmuch as the dispersion of the order-sorting prism is greater in the ultraviolet than in the visible region it is advantageous to use larger exit slit heights at low wavelengths to increase intensity and smaller slit heights at longer wavelengths to reduce order-overlap strong light. At wavelengths below 250 nm, solar-blind photomultiplier tubes are used to reduce stray light originating from the much more intense visible region of the xenon lamp spectrum. The single-channel CEWM-AA system utilises analogue signal-processing electronics consisting of a sine-wave oscilla-tor to drive the refractor plate torque motor (usually at 100 Hz) a lock-in amplifier to measure the a.c. component of the photosignal and an operational amplifier circuit to convert the lock-in output to absorbance for measurement on a strip-chart recorder.In the multi-element (SIMAAC) system a minicomputer (PDP 11/34) is used to generate the wavelength modulation waveform measure and store intensity data on 16 active channels average intensities compute intensity ratios and absorbances determine peak heights and areas maintain files of analytical signals and experimental conditions perform analytical calibration with linear and non-linear least-squares curve-fitting techniques and print out results ANALYST MARCH 1983 VOL. 109 213 10-6 10-7 10-8 ? > v) c al c Y .-c-' -10-9 10-10 10-1' Cr 0 Na MnO 0 Pb Fe *I Ca Mr?# P Al 200 300 , K Mn Ca Sr 0 Na 0 500 600 Wavelengthhm Fig.3. measured o n a high-resolution k h e l l e spectrometer Comparison of the relative intensity o f a 300-W Cermax xenon-arc lamp (circles) to hollow-cathode lamps (vertical lines) In both systems a real-time display of the absorption spectrum within the modulation interval can be displayed on an x - y oscilloscope by connecting the phototube output to the y-axis and the modulation waveform to the x-axis. This display is very useful for checking for wavelength drift excessive background absorption or unsuspected spectral interferences. Naturally the computerised system provides much greater versatility and convenience than the simpler analogue system. For example the modulation waveform need not be a simple sine or squarc wave but can be customised to enhance the signal to noise ratio and analytical range.28 It is also possible to average many passes through the modulation interval in order to enhance the signal to noise of spectral profile measure-ments.3" Detection Limits Compared with all previously reported continuum-source atomic-absorption systems the CEWM-AA and SIMAAC systems provide significantly lower detection limits on average by about an order of magnitude.This is due we feel, to the combined effect of the high spectral dispersion of the echelle spectrometer the wavelength modulation and the ratiometric data processing which gives the effect of a double-beam measurement. Compared with conventional line-source instrumentation our detection limits are slightly poorer overall but tend to be comparable at wavelengths above 280 nm and poorer below owing to the intensity distribution of the xenon lamp.Tables 1 and 2 show this comparison for flame and electrothermal atomisation respec-tively. Background Correction The background correction capabilities of wavelength modu-lation in atomic absorption have been dealt with at length in previous studies. 19.20 The modulation technique offers many of the advantages of the Zeeman and Smith - Hieftje methods, i. e. only a single light source is used and correction is made very close to the analytical line. Moreover it is well suited to multi-element applications because a single refractor plate behind the entrance slit modulates all the output channels simultaneously. The wavelength modulation background correction system has been shown to be able to correct for up to 3.0 absorbance units of static background absorption.2" Its performance with transient background in electrothermal atomisation is illus-trated in Fig.4 which shows absorbance - time plots for copper in various matrices. Similar experiments have been reported for Pb and Cd.20 Analytical Range One of the problems that must be overcome in any proposed simultaneous multi-element atomic-absorption system is the limited concentration range that is ordinarily considered to be characteristic of atomic-absorption measurement. In our continuum-source atomic-absorption work we have achieved a greatly extended working concentration range by measuring the absorbance at several wavelengths across the profile of a single analytical absorption line.The absorbance at the line centre is utilised at low to medium concentrations. At higher concentrations where the absorbance at the line centre is too high to be useful absorbances measured on the sides of the absorption profile are used. The feasibility of this idea is illustrated by the calibration graphs for sodium by continuum-source atomic absorption shown in Fig. 5. Note that the calibration graphs measured at wavelengths off the line centre extend the useful concentration range all the way to 20 000 yg ml-l in this instance. The detection limit for sodium is less than 0.01 yg ml-1 so the total analytical range is over six orders of magnitude. In a simultaneous multi-element system it is not practical to control the wavelength modulation interval individually for each element in response to its concentration in each sample.The SIMAAC system has been designed to acquire and store 214 ANALYST MARCH 1984 VOL. 109 Table 1. Comparison of detection limits of continuum- and line-source flame atomic absorption Detection limittipg m1.k' Element (flame) Ag . . . . Al(NA) . . Au . . . . Ba(NA) . . Be(NA) . . Bi . . . . Ca(NA) . . Cd . . . . c o . . . . Cr(NA) . . (AA) . . c u . . . . Fe . . . . K . . . . Li . . . . Mg . . . . Mn . . . . Mo(NA) . . Na . . . . Ni . . . . Pb . . . . Pd . . . . Pt . . . . Rh . . . . Sb . . . . Si(NA) . . Sn(AA) . . (NA) . . Sr(NA) . . Te . . . . Ti(NA) . . TI . . .. V(NA) . . Zn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . * . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wavelength/ nm 328.068 396.1/309.31( 242.795 553.548 234.861 223.061 422.673 228.802 240.745 357.869 324.754 248.327 766.49 670.784 285.213 279.482 313.259 588.995 352.454/232.011 283.3/217(( 244.81247.61 I 265.945 343.489 217.58 251.6 286.333 460.733 214.281 365.35 276.789 318.54 213.856 Continuum3 Line4 0.007 0.1 0.17 0.07 0.01 0.3 0.003 0.03 0.07 0.1 0.02 0.01 0.07 0.007 0,003 0.001 0.01 0.3 0.003 0.07 0.1 0.1 1 0.07 1 0.7 1 1 0.02 0.7 0.3 0.1 0.3 0.07 0.009 0.1 0.04 0.09 0.003 0.09 0.003 0.003 0.02 0.1 0.009 0.007 0.02 0.00s 0.005 0.0009 0.004 0.06 0.001 0.1 0.06 0.03 0.3 0.03 0.08 0.5 1 3 0.06 0.1 0.2 0.07 0.2 0.003 * NA = dinitrogen oxide - acetylene; AA = air - acetylene.t Defined as a signal to noise ratio of 3 for a 5-s integration time. $ Background corrected. Calculated for 300-pm slit height and square-wave modulation from data measured with 100-pm slit height and sine-wave modulation.35 5 Not background corrected. Measured on a Perkin-Elmer 5000. I/ First wavelength given is optimum for continuum source second is optimum for line source. Table 2. Graphite furnace instrumental detection limits. Based on 2 0 4 sample size peak-height measurement.signal to noise ratio of 3, atomisation from the tube wall Detection limiting ml 1 Element SIMAAC Perkin-Elmer c o . . . . Cr . . . . c u . . . . Fe . . . . Mn . . . . Mo . . . . Ni . . . . v . . . . . . Zn . . . . 0.9 0.3 0.1 0.7 0.1 0.7 2 0.9 0.4 0.2 0.08 0.2 0.2 0.08 0.2 2 2 0.008 fixed pattern of points across the entire absorption profile of each element as shown in Fig. 6. Sufficient data are taken to allow the calculation of six double-beam background-corrected absorbances of different sensitivities. Thus when the system is calibrated the six absorbances measured for each calibration solution are used to prepare a set of six calibration graphs such as shown in Fig.7. When an unknown sample is measured it also produces six absorbance readings, each of which is read off the corresponding calibration graph 0 C 0 3 C 0, .- 4-f=, Q) 100% 2 0 v) 0.52 0.39 0.26 0.13 0 W 0 c 2 0.13 $ 0.13 g o 10- P Analyte abs. '7 Jz k Background abs. 10 p.p.b. 10 p.p.b. 1560 p.p.b.4.2 p.p.b. Beam blocked -*A-H20 3% NaCl serum urine physically Time -b Fig. 4. Absorbance vs. time graphs for the electrothermal atomisa-tion of Cu in various matrices illustrating the background-correction ability of the wavelength modulation method. Reprinted by permis-sion from reference 20 1 .o 0) g 0.1 f! $ a a 0.01 I 589.6nm 1 10 100 1000 10 000 Concentration/pg ml-1 Fig. 5. Calibration graphs for sodium by continuum-source atomic-absorption spectrometry with flame atomisation.A Measured at the centre of the sodium absorption line; B and C measured on the edges of the line as the indicated distance from the line centre 8-1 3 Wavelength Fig. 6. Intensity measurement points across the absorption profile with a digitally sampled data acquisition svstem. Six types of absorbances calculated [absorbance = log ( I / I ) ~ type 1 I, = 1-5 and 16-20. I = 6-15; type 2 I. = 1 and 20. I = 6 and 15; type 3. I, = 1 and 20. I = 5 and 16; type 4 I, = 1 and 20. I = 4 and 17; type 5 I, = 1 and 20. I = 3 and 18; type 6. I, = 1 and 20. I = 2 and 1 ANALYST MARCH 1984 VOL. 109 215 1 .o al m 0.10 -F 0 n Q 0.01 0.001 - . ' I I 0.1 1 .o 10 100 1000 10000 Concentrationipg ml-l Fig.7. A family of calibration graphs obtained using absorbance data acquired by sampling the absorption profile as in Fig. 6 ( h = 285 nm) l.o r 5, 0.10 9 0 m E 5 0.01 I I 1 I I I 100 1000 10000 0.001 0.1 1 .o 10 Concentratiodpg ml-1 Fig. 8. in Fig. 7 Relative concentration error for the calibration graph shown to give a set of six estimates of the sample concentration. Ideally the six estimates of the concentration should be identical but because of noise and curve-fitting errors they will never be exactly the same. Moreover they will not be equally useful at low sample concentrations the measure-ments taken far from the line centre will be unusable because of poor signal to noise ratio whereas at high sample concentrations measurements taken near the line centre will be unusable because of curve-fitting errors.The selection of the optimum concentration estimate or the optimum combi-nation of concentration estimates can be done in a number of ways. We have based our selection on the observation of relative concentration errors (RCE) which are calculated in the following way. For each sample measured by flame atomisation the integration period (usually 5 s) is divided into 28 equal segments over which each of the six absorbances is calculated. This yields an estimate of the mean and standard deviation of each of the six absorbances. Each of these is converted into an equivalent mean and standard deviation of concentration by utilising the corresponding calibration graph in conjunction with some method of curve fitting or interpola-tion.The relative concentration error (RCE) is the ratio of the standard deviation of concentration to the mean concentra-tion. In our work we have computed the reported concentra-tion for each sample solution on the weighted average of the six concentration estimates where the weighting factor is taken as the reciprocal of the square of the RCE. Thus each of the concentration estimates is utilised to the extent that it can contribute useful information. The use of six calibration graphs seems reasonable when one inspects plots of the calibration graphs themselves; the linear ranges of the graphs overlap over a total concentration range of four to six decades. However the manipulation of six absorbances and six analytical calibrations considerably com-1 .o 0.10 0.01 0.10 V."" 1 0.1 1 .o 10 100 1000 10000 Concentration/pg m 1-1 Fig.9. modulation waveform Simplified calibration graphs obtained using a stairstep plicates the computer software. Moreover a critical inspec-tion of plots of RCE versus concentration shows that not all graphs are really necessary. A typical example is shown in Fig. 8. It can be seen here that the regions of lowest RCE on each graph overlap extensively. This is due to the fact that by utilising appropriate curve-fitting procedures the useful region of each calibration graph can be extended considerably above the upper limit of calibration graph linearity. As a result it is possible to achieve acceptably low values of RCE over a five-decade concentration range by using only two graphs Nos.1 and 4. In our system the data acquisition rate during the wave-length modulation cycle is constant; therefore the distribu-tion of the measured points with respect to wavelength is determined by the shape of the modulation waveform. For line profile measurements we use a triangular waveform, which gives a uniform point distribution. On the other hand, for quantitative analytical applications a bigaussian wave-form17 has been used. This waveform results in a concentra-tion of points near the centre of the line profile which allows the most sensitive (graph 1) absorbance to be calculated from a total of 20 intensity readings resulting in a reduction of random noise (including quantisation noise).The less sensi-tive absorbances however are calculated from only four intensity readings and therefore tend to be noisier. It is for this reason that the RCE for the less sensitive graphs never achieves as low a value as that of the most sensitive graph. In recent work we have solved this problem and also have simplified the calibration algorithms considerably by using a stairstep modulation waveform39 which in essence redistri-butes the measured points across the modulation interval in such a way as to allow the calculation of only two absorbances, similar to Nos. 1 and 4 but with better signal to noise ratio. The results for magnesium are shown in Fig. 9. Even for this very sensitive element measurement of good precision can be made up to 5 000 pg ml-1 using only the 285.213-nm resonance line.It should be pointed out that the technique just described is completely compatible with simultaneous multi-element oper-ation and does not compromise the background correction capabilities or the detection limits of the system. Perhaps the greatest practical benefit of this approach is that it is not necessary to estimate beforehand the approximate concentra-tion of each sample (except of course that the samples must fall within the range of the standards in order to avoid extrapolation errors) 216 ANALYST MARCH 1984 VOL. 109 Spectral Profiles Conventional line-source atomic-absorption spectrometry is in the unique position of being the only kind of modern absorption spectrometry that is ordinarily performed with non-tunable monochromatic light sources.It is therefore impossible to measure the absorption spectrum of a sample. Certainly in other areas of spectroscopy the measurement of the spectrum is considered to be essential. In practical terms, this means that one has no assurance that an absorption signal measured for a given sample is indeed due to the pure atomic absorption of the analyte rather than to some unsuspected problem from excessive or structured background absorption or from some unsuspected spectral interference. Of course, atomic-absorption spectra are fairly simple so the occurrence of spectral interference is rare. Certainly conventional deuterium arc background correctors do work satisfactorily in many instances.However the point is that there is no way to be sure that there are no spectral complications because the spectrum cannot be observed. The potential for such problems can be expected to become greater as more sensitive methods, such as electrothermal atomisation are applied to more complex and unfamiliar materials. With the SIMAAC system the measurement of spectral profiles is facilitated by the ability to perform ensemble averaging over selected time intervals during the atomisation and to plot the resulting profiles on the line printer. The system has been used to measure both flame") and time-resolved furnace37 profiles to study several spectral interfer-ences and to characterise some previously unmeasured absorption lines.30 Table 3. Applications of continuum-source method Applications Table 3 lists some of the samples to which the continuum source method has been applied the elements which have been measured and the mode of atomisation.Many of these are reference materials. In general we find that the accuracy of single-element CEWM-AA measurement in which the analytical conditions are individually optimised for each element is equal to that of line-source atomic-absorption measurements of the same samples. The accuracy of simultan-eous multi-element measurement under compromise analy-tical conditions is generally slightly poorer than the accuracy that can be obtained by optimised single-element measure-ment. This is not unexpected of course. Based on a large number of measurements of many different reference mat-erials over a period of about 5 years it has been found that accuracies of the order of 5-15% can generally be obtained for those elements whose concentrations are well above the detection limit.When electrothermal atomisation is used the best results are generally obtained by using a platform and by basing quantitation on peak-area measurement.4" The prototype of the SIMAAC system is now installed at the Nutrient Composition Laboratory US Department of Agriculture Beltsville MD USA where it is under the supervision of Dr. J . M. Harnly. The system is in daily use both for routine sample analysis and for research purposes. Atomic Emission A significant advantage of the wavelength modulation tech-nique compared to other methods of background correction is that it also provides background-corrected emission measure-ments which is something commercial atomic-absorption systems do not provide.The technique is equally applicable to flame plasma and electrothermal atomisation. Moreover the extended range and profile measurement capabilities also apply to emission measurement. Simultaneous multi-element analysis by carbon furnace atomic emission has been shown to be feasible using the SIMAAC instrumentation.41 Material* Bronze (IPT Standard Brass (IPT Standard 40) Sub-bituminous coal (SRM1635) . . . Limestone (SRM lc) . , Water (SRM 1643a) . . 10A) . . . . . . Bovine liver (SRM 1577) Orchard leaves (SRM1571) . . . . Fly ash (SRM 1633, 1633a) . . . . . . USGS rocks AGV-1 (GSP-1,BCR-1) .. Wheat flour (SRM 1567) Rice flour (SRM 1563) Spinach leaves (SRM1570) . . . . Elements determined Fe Ni Sb Sn Zn Pb Pb Sn Zn c u Ca Ag Cr. Cu Fe Ni Ba Mo Al, V Co Pb Zn Mn Mn Fe Cu Mg, Na K Ca Zn Cu Mn Fe Zn, Ca. K Mn Zn Cu Fe, Mg Ca Na K Mn Zn Cu Fe, Cr Co Mg. Ca, Na K Mn Zn Cu. Fe. Mn Zn Fe Cu, Mn Zn Fe Cu K Mn Cu Fe Zn, Cr Mg Ca Na K Mg K Ca K Mn Zn Fe Cu, Mg K Manganese nodules USGS A1 andP1 . . . . Mn Zn Co. Ni Na, K Mg Fe Cu Orange tomato, pineapple apple, prune juice . . . . Mn Zn Fe Cu, Mg Na Cr. Ni Co Urine (pool sample) . . Cr Blood serum . . . . Na K Mg Ca, Zn Cu Fe At om isat io n Flame Flame Electrothermal Flame Electrothermal Flame Electrothermal Flame Flame Flame Flame Flame Electrothermal Flame Flame Flame Electrothermal Flame * IPT Instituto de Pesquisas Techologicas; SRM National Bureau of Standards Standard Reference Materials; USGS US Geological Survey.The Future The use of a continuum primary source in atomic-absorption measurement is a rational alternative to the conventional line source method. The advantages are clear the convenience of a single source excellent background correction the ability to measure spectral profiles greatly extended analytical range and the proved ability to make useful simultaneous multi-element measurements with either flame or electrothermal atomisation. However the disadvantages are equally clear: the detection limits for several elements in the low-UV region are significantly poorer and the present multi-element system is expensive and complex.Indeed the commercial production of such a system would present many challenging engineering problems. We believe however that this is not an unreason-able future prospect. Atomic-absorption spectrometry may be a mature technique but it is not a static one. The range of commercial equipment now encompasses not only simple low-cost models but also elaborate automated top-of-the-line systems whose cost now approaches that of many plasma atomic-emission systems. The newest sequential multi-element systems are exceedingly complex optically and electromechanically with multiple lamp turrets programm-able power supplies beam splitters and choppers separat ANALYST MARCH 1984 VOL.109 217 background corrector lamps (possibly two deuterium and tungsten) automatically controlled monochromators etc. The continuum-source approach on the other hand is optically simple even elegant. Its complexity lies more in the data acquisition and processing and indeed the computer system and the software required to run SIMAAC is elaborate. However this is one area of modern technology where the performance/cost ratio is increasins iir;imatically and rapidly. There are now comparatively low-cost computer systems available whose memory and computing power far exceed those of our prototype system. It is not likely that that will long remain a limiting factor. In the future development of the continuum-source method, perhaps the most pressing need is to improve the detection limits.In this connection the most serious limitation is the comparatively small slit area of presently available high-resolution spectrometers primarily due to slit-height restric-tions. Harnlyj2 has predicted on the basis of signal to noise calculations that presently available 0.5-m monochromators based on large holographic gratings could offer a 5-fold signal to noise advantage over the present echelle spectrometer for the photon-noise limited case. This would mean that a single-channel continuum-source atomic-absorption system could be constructed now that would provide detection limits better than those at line-source instruments for many ele-ments. Even considering its present limitations however the continuum-source method has proved to be a very useful method capable of solving many analytical problems with convenience and speed.With continued development there is every reason to believe that this approach will also play a role in the future of analytical atomic spectrometry. I . 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References Walsh A . Spectrochim. Acta 1955 7 108. Allan J . E . Spectrochim. Acfa 1962 18 259. Gibson J . H . Grossman W. E. and Cooke W. D. Appl. Spectro.$c. 1962 16 47. Manning. D. C and Kahn H. L. At. Absorpt. Newsl. 1965, 4 224. Frank C. W. Schrenk W. G . and Meloan C . E. Anal. Chem. 1967 39 534 Fassel V. A . Mossotti. V. G . . Grossman W. E. L. and Kniseley R. N . Spectrochim. Acta 1966 22 347. McGee W.W. and Winefordner J. D. Anal. Chim. Acta, 1967 37. 429. Snelleman W. Spectrochim. Acta Part B 1968 23 403. Elser. R. C. and Winefordner J . D. Anal. Chem. 1972 44, 698. Nitis G. J . Svoboda. V and Winefordner J . D. Spectro-chim. Acta Part B 1972 27 345. Veillon. C and Merchant P Jr. Appl. Spectrosc. 1973 27, 361. Cochran R. L. and Hieftje G . M. Anal. Chem. 1978 50, 791. Bower. J . Bradshaw J . and Winefordner. J . Talanta 1979, 26 249. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. Blackburn M. B. and Winefordner J . D. Can. J . Spectrosc., 1982 27 137 Keliher P. N. and Wohlers C. C. Anal. Chem. 1974 46. 682. Keliher.P. N . and Wohlers C. C. Anul. Chem. 1976 48, 140. Zander. A. T. O’Haver. T. C and Keliher. P. M. Anal. Chrm. 1976 48 1166. O’Haver T. C . Harnly J . M. and Zander A. T Anal. Chem. 1977 49 665. Zander A. T. O’Haver T. C .and Keliher P N. Anal. Chem. 1977 49. 838. Harnly. J . M. and O’Haver. T. C Anal. Chem. 1977. 49. 2187. Guthrie B. E. Wolf W. R. and ”tcillon. C. Anal. Chem., 1978 50 1900. O’Haver T. C. Harnly J . M. and Zandcr. A. I Anal. Chem. 1978 50 1918. O’Haver T. C . “Wavelength Modulation Spectroscopy,” in Hercules. D. M. Editor “Contemporary Topics in Analytical and Clinical Chemistry,” Plenum New York 1978 p. 1. Epstein M. S. and Lander A. T. Anal. Chem. 1979,51,915. O’Haver T. C . Anal. Chem. 1979. 51 91A. Veillon C. Wolf W. R. and Guthrie B.E. Anal. Chem., 1979 51. 1022. Harnly J . M. O’Haver T. C. Golden B. and Wolf W. R., Anal. Chem. 1979 51 2007. Harnly J . M and O’Haver T. C. Anal. Chem. 1981 53, 1291. Harnly J . M. Miller-Ihli N. J . and O’Haver T. C J . Autom. Chem. 1982 4 54. Miller-Ihli N. J . O’Haver T. C. and Harnly J . M Anal. Chem. 1982 54. 799. Harnly J . M. Anal. Chem. 1982. 54 876. Harnly J . M. Anal. Chem. 1982 54 1043. Kane J . S. and Harnly J. M. Anal. Chim. Acta 1982 139, 297. Harnly J . M. Kane J . S. and Miller-Ihli N. J . Appl. Spectrosc. 1982 36 637. Messman J . D. Epstein M. S . Rais T. C and O’Haver, T. C . Anal. Chem. 1983. 55 1055. Harnly J . M. Patterson. K. Y. Veillon C. Wolf W. R., Marshall J . Littlejohn D. Ottaway J . M. Miller-Ihli N. J . , and O’Haver T. C. Anal. Chem. 1983 55 1417. Miller-Ihli N. J . O’Haver T. C. and Harnly J . M. Appl. Spectrosc. 1983 37 429. Miller-Ihli N. J . O’Haver T. C . and Harnly J . M. Anal. Chem. in the press. Harnly J . M. Wolf W. R and Miller-Ihli N. J . “Quality Assurance of Analysis of Inorganic Nutrients in Foods,” in Stewart K. Editor “Modern Methods of Food Analysis,“ Proceedings of 7th IFT-IUFOST Basic Symposium 43rd Annual Meeting of International Food Technology New Orleans LA June 1983. Harnly J . M. Miller-Ihli N . J . and O’Haver T. C., Spectrochim. Acta in the press. Marshall J . . Littlejohn D. Ottaway J . M. Harnly J . M., Miller-Ihli N. J . and O’Haver T. C. Analyst 1983 108 178. Harnly J . M Anal. Chem. in the press. Puper A312 70 Received August 19th 198

ISSN:0003-2654    

DOI:10.1039/AN9840900211

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST MARCH 1984 VOL. 109 Corn bined Use of Photoacoustic Spectroscopy and Differential Thermal Analysis in Mineralogical Analysis 219 Malcolm S. Cresser and Neil T. Livesey Department of Soil Science University of Aberdeen Old Aberdeen A69 ZUE UK The ultraviolet - visible and near-infrared absorption spectra of minerals can be conveniently studied by photoacoustic spectroscopy. The spectra show considerable overlap which limits their value for qualitative mineralogical analysis. However the changes in mineral PA spectra on heating which correspond mainly t o dehydration or dehydroxylation reactions changes in bonding of iron or oxidation of Fe(ll) show considerable potential for the identification of the minerals themselves and of the processes that take place during their alteration.The changes are discussed briefly for a wide range of common minerals and the analytical potential of this approach is critically appraised. Keywords Minerals; photoacoustic spectroscopy; thermal analysis Although some interest has been shown in the photoacoustic spectra of minerals over the UV - visible and near-IR regions,l-3 and reflectance spectra have been published for many minerals over the same regions,4-h virtually no use is made of photoacoustic spectroscopy (PAS) or reflectance spectroscopy in routine mineralogical analysis. The reasons for this restricted interest are readily apparent. Most near-IR bands are attributable to overtone or combination bands involving bonding to hydrogen which for most common minerals means to OH bonds; visible spectra are dominated by the oxidation state of iron present and its bonding.Thus, although the PA spectra of different single minerals may exhibit distinct differences there is a high probability of substantial overlap of spectra of the diverse minerals present in samples such as soils or ground rocks. However changes in spectra attributable to specific dehydration dehydroxylation or iron oxidation reactions which occur only after heating at certain specific temperatures and which may be readily identified lend themselves to examination by PAS. The purpose of this investigation was to assess the potential value of changes in mineral PA spectra in the qualitative analysis of minerals and to elucidate the changes that they undergo during heating.Experimental Apparatus The PA spectrometer used was an OAS 400 (EDT Research Ltd. ) controlled by a Commodore 32K minicomputer using the manufacturer's software. The computer was used to produce ratios of all spectra to carbon black reference spectra (Darko G60 dried at 105 "C) to compensate for differences in the responses of the PA detector and reference detector to radiation of different wavelengths and in non-uniformity in the absorption of radiation between leaving the spectrometer beam splitter and reaching the sample or reference cells e.g., from water on silica surfaces. Differential thermal analysis (DTA) and thermogravimetric (TG) traces were run at 10 or 20 K min-1 on Stanton Redcroft Model DTA 673-4 and TG 750 instruments respectively to allow selection of suitable heating temperatures.All mineral samples were run in air apart from a series of ochre samples containing up to 30% of organic matter which were run in oxygen-free nitrogen. Materials Samples of minerals and related materials were obtained from the authors' Departmental collection or from Gregory, Bottley & Lloyd London. All were identified by thermal analysis and/or X-ray diffraction and it was established that they were composed almost exclusively of one mineral except for those reported otherwise. Samples were air dried and ground if necessary to pass a 240-mesh (64-vm) sieve either by hand with an agate pestle and mortar or for hard materials with a Grindex ball-mill using tungsten carbide balls. Except where otherwise stated, the powdered samples were oven-dried overnight at 105 "C, cooled and stored in desiccators over calcium chloride prior to running PA spectra.The sample cell was filled with unpacked powder (ca. 50 mg) and the sample surfaces were levelled with the edge of a microspatula. Results and Discussion Iron Oxides and Oxyhydroxides Common oxides oxyhydroxides and hydroxides particularly those of iron are of great interest because of their importance in distinguishing soil types and because of their role in soil physico-chemical processes. Moreover the capability of PAS for investigation of changes in absorption spectra with depth below particle surfaces via changes in detector phase or in the modulation frequency used is potentially valuable for the study of oxide coatings.For these reasons and because of the frequent difficulty in chemical characterisation of oxides and hydroxides by alternative techniques this group was the first selected for study. Goethite Goethite (a-FeOOH) is the most frequently occurring iron oxide in soils because of its high stability under a wide range of soil conditions.' Heating goethite at about 380 "C causes a mass loss corresponding to dehydroxylation to haematite heat 2a-FeOOH -.+ a-FelO3 + H20 goethite haematite (yellow - brown) (bright red) The mass loss started at 312 "C on the sample used and was complete at 393 "C. The change in the UV - visible PA spectrum is clearly visible the spectrum after heating (in air) at 394 "C resembling closely that of haematite [Fig.1 (a) and (b)] and like that of haematite remaining unchanged on further heating to 1000 "C. The near-IR spectrum of goethite, like the reported reflectance spectrum," is devoid of diagnos-tically useful features ANALYST MARCH 1984 VOL. 109 220 T -m C a m .-a a 394 "C (Haematite) (a) ? Goethite 1000 "C (Haematite) __ y;yoc L e p i doc r o c i t e ' - \\. . 105 "C \I0 '\80 Hz 160 Hz @ = O T 105990 "C (b) Haematite I I (4 I 1 (fl 560 "C (Haematitel 105 7++ "C Magnite 330 "C (Maghemite) 300 500 700 900 500 700 900 Wavelengthinm Fig. 1. temperatures indicated. Names of products are given in parentheses UV - visible PA spectra of iron minerals heated in air to the TemperatureiOC Fig. 2. Variation in the mass magnetic susceptibility.x of lepidocrocite from "ochre ," on heating in air Haematite In soil haematite (a-Fe203) often occurs in association with goethite where it has a strong red pigmenting e f f e ~ t . ~ The mineral is stable over a wide temperature range the small reversible peak observed at around 680 "C8 being of no consequence in this study. As already mentioned the UV -visible PA spectrum [Fig. l(b)] is unchanged as a result of heating up to 1000 "C. The near-IR spectrum is broad and featureless over the range 1.2-2.5 pm like the reported reflectance spectrum ,h and is also not significantly influenced by heating. Lepidocrocite Lepidocrocite (y-FeOOH) is found often associated with goethite only in hydromorphic soils where Fe(I1) is produced at very low oxygen concentrations.7 An ochre sample from a waterlogged site at Peatfold Burn Glenbuchat Grampian Region Scotland was identified as impure lepidocrocite from its DTA trace and from the variation in its mass magnetic susceptibility with heating (see Fig.2). Lepidocrocite dehy-drates to maghemite when heated at 275-410 oC.9,10 The sample heated in air at 210 "C exhibited a substantial increase in susceptibility indicative of maghemite formation. The maghaemite is stable to ca. 625 "C because of impurities,9,11 but then undergoes exothermic conversion to haematite with a corresponding decrease in susceptibility. The UV - visible PA spectra after heating at 105,375 and 1 000 "C are shown in Fig. l(c). The spectrum of the impure lepidocrocite differs slightly from those of goethite and haematite but the difference would be more readily discernible from comparison of the derivative PA spectra as suggested for goethite and haematite.l* For pure lepidocrocite it might be expected that a PA spectrum corresponding to maghemite would be obtained on heating at 375 "C and at 1000 "C a spectrum corresponding to haematite should be observed. On the authors' sample however the PA spectrum after heating the impure lepidocrocite at 375 "C was closer to that of haematite than to that of maghemite which may be more reliably obtained by low-temperature oxidation of magnetite.9 This probably reflects the conversion of maghemite to haematite, at or near particle surfaces to an extent that depends on the nature and distribution of impurities.9 It was therefore decided to investigate the homogeneity of the particles of lepidocrocite after heating in air at 375 "C.Generally the higher the modulation frequency or the smaller the phase lag between the irradiation period and the detector response the more the characteristics of the PA spectrum obtained are dominated by the sample surface. Lower frequencies or a higher phase favour a greater contribution from sub-surface zones of the sample particles. For the sample in question lower phase and/or higher frequency yielded a spectrum increasingly similar to that of haematite [Fig. l ( d ) and ( e ) ] confirming the presence of haematite rather than maghemite at least at the particle surfaces. The spectra in Fig. l(c) were plotted at a modulation frequency of 80 Hz and a phase of 105" conditions found by the authors to give optimum signal to noise ratios for numerous mineral samples the higher frequency reduces the effective sampling depth.The results must however. be interpreted with great caution because for particles in the silt-size range the particle size becomes limiting in this approach. Provided spectra are normalised at one wavelength to allow comparisons of PA spectra to be made easily the authors have found scanning PA spectra at different frequen-cies and/or phase angles to be useful for investigation of particle homogeneity in spite of the change in PA signal with frequency of modulation. Some ochre samples studied for example appear to be completely uniform whereas others show a marked lack of homogeneity.Maghemite Samples of *'pure" maghemite (y-Fe,O,) of natural origin are not readily available. However it may be assumed that the PA spectrum from magnetite after heating in air to 330 "C corresponds to maghemite [Fig. 101. At higher temperatures maghemite is converted into haematite. The near-IR spec-trum was featureless over the range 1.2-2.5 pm. Magnetite The UV - visible PA spectrum of magnetite (Fe304) is clearly very different from the spectra of other iron oxides and hydroxides [Fig. 103 but the near-IR PA spectrum like the reflectance spectrum,6 is completely featureless. According to mull in^,^ magnetite oxidises to maghemite at 15&250 "C but this is a low temperature range for natural samples.8 The authors' sample gave the first DTA peak at 328 "C a more typical value and the features of the UV - visible spectrum after heating at this temperature may be attributed to maghemite which is increasingly oxidised to haematite at higher temperatures depending on purity."Limonite" "Limonite" is now obsolete as a mineral name but in the past was extensively used to describe brown rusty iron oxide accumulations generally containing appreciable amounts of goethite.7 PA spectra of a "limonite" sample have been published elsewhere by the authors. 12 That particular sample, which was shown by X-ray diffraction to be predominantl ANALYST. MARCH 1984. VOL. 109 I Calcite (Near IR) (a) 22 1 Calcite (UV - visible) ( b ) quartz gave when oven dried a UV - visible PA spectrum closer to those of haematite or lepidocrocite than that of goethite."Cryptocrystalline lepidocrocite along with haema-tite and additional water in some form," one suggestion for the nature of limonite,b would certainly fit the PAS data as would the slight dehydration at ca. 335 "C. Unlike goethite the limonite sample showed an absorption band at 2.46 pm which disappeared after heating to 335 "C. Carbonates Hunt and Salisbury' commented that most of the broad bands in the 1-pm region of the reflectance spectra of carbonates are attributable to the presence of iron substituted for the appropriate metal ion in the carbonate. This is true even when the iron is present at relatively low concentrations. The near-IR reflectance spectra also show a series of bands between 1.6 and 2.5 pm attributable to overtone and combination tones of the internal vibrations of the C032-radical or of the C032- radical with vibrations of the entire structure.Calcite The DTA trace for calcite gave a very strong endothermic peak at 962 "C well within the usual range of 860-1010 "C quoted for this mineral.13 The TG showed negligible loss up to ca. 720 "C followed by rapid mass loss of37.5 and 43% for two different samples studied attributable to evolution of C02. The near-IR PA spectra of oven-dried samples were similar to the reflectance spectrum reported by Hunt and Salisbury,s with absorption bands at 1.9-2.0 2.2-2.4 and 2.5 pm [Fig. 3(a)]. Heating samples at 800 or 960 "C gave two similar spectra the main features being a broad band at 1.2-1.6 pm and a sharper band at around 2.25 pm.The magnitude of the former in the carbonate samples studied was related to their I 500 700 - m1.2 1.6 2.0 2.4 2.8 300 Wavelength/pm c (UV - visible) (C) Do I om ite Calcium iron Wavelennthhm 1 1 1 I I I I I 300 500 700 1.2 1.6 2.0 2.4 2.8 Wavelengthhm Wavelengthlpm (UV - visible) 800 "C 105 "C 930 "C 300 500 700 Wavelengt hlnm Fig. 3. PA spectra of carbonate minerals heated in air to the temperatures indicated. Calcium carbonate analytical-reagent grade iron concentration as determined by dissolution and flame atomic-absorption spectroscopy. The dominant feature in the UV - visible spectra of both calcite samples and indeed of all the carbonates studied was the appearance of a broad absorption band at around 300 nm.This band becomes even stronger on heating to high temperatures [Fig. 3(b)] but for calcite the spectrum obtained on heating was different to that obtained on heating chalk marble or dolomite samples [Fig. 3(e)l* Chalk and marble The near-IR PA spectra of chalk and marble dried at 105 "C were similar to that of calcite although there were slight changes in the relative peak heights. The particular marble sample selected had a low iron content and did not give a strong absorption band at around 1.2 pm on heating to 960 "C. The dominant features of the UV - visible spectra have already been summarised. Fig. 3(c) shows the spectra of chalk, dolomite and analytical-reagent grade calcium carbonate, which contained low very low and negligible amounts of iron, respectively.Dolomite Fig. 3(d) shows typical near-IR spectra for a dolomite sample low in iron. The spectrum of the oven-dried sample is similar in most respects to that of calcium carbonate. Although the DTA trace for dolomite showed two distinct strong endother-mic peaks as is invariably the case,13 heating a small finely ground sample for 1 h in a furnace at 800 "C may have been sufficient to remove C02 from both magnesium and calcium carbonates. It is difficult to state with certainty whether any residual carbonate was contributing to the near-IR PA spectrum after heating to this temperature. It seems improb-able bearing in mind that identical spectra were obtained after heating at 800 and 930 "C.Further heating above 800 "C also had only a very minor influence on the UV - visible spectrum of dolomite. After heating to high temperatures the spectra obtained from all carbonates except calcite were very similar to those obtained from heating appropriate oxide samples Bt the same temperature although the features attributable to iron still varied with the iron concentration. Silicates Most useful diagnostic information for silicate identification comes from spectral features corresponding to dehydration and/or dehydroxylation reactions in the near-IR region. The UV - visible PA spectra are generally of less value again being dominated by the effects of iron especially after heating at elevated temperatures. Talc DTA curves for talc [Mg3Si4010(OH)2] are reported to be very consistent with a solitary endothermic peak at 950-1000 "C.14 The TG trace showed that mass loss commenced at ca.950 "C corresponding to dehydroxylation and was complete at >lo00 "C. Two different talc samples gave virtually identical sets of near-IR PA spectra and only one example is therefore included [Fig. 4(a)]. Bands were observed for the oven-dried sample at 1.45 2.14 2.18 2.27, 2.36 2.43 and 2.50 pm but only a band at 2.26 pm was conspicuous after heating at 1040 "C. The spectrum is similar to the reflectance spectrum published by Hunt and Sali~bury,~ although the strong OH band they reported at ca. 1.4 pm is much reduced here presumably as a result of oven drying. The UV - visible PA spectrum of the oven-dried sample was featureless but the heated sample exhibited strong absorption attributable to iron substituted for small amounts of magne-sium [Fig.4(b)]. This spectrum is typical of those obtained o 222 105 "C ANALYST MARCH 1984 VOL. 109 heating most iron-containing silicates and bears some similar-ity to that obtained from strongly heated dolomite [Fig. 3(e)]. Montmorillonite The composition of montmorillonite [(My+ .nH20)(A12,-Mgr.)Si4010(0H)2] is very variable and considerable differ-ences between the near-IR absorption spectra of different samples might be expected and have indeed been reported.4 The PA spectrum shown for an air-dried sample from Japan, is similar to that reported for an air-dried sample from the USA,4 having strong bands at 1.48 1.97 and 2.27 pm.The band at 1.97 pm is relatively much enhanced if an air-dried rather than an oven-dried sample is used. This PA spectrum shows small. but significant shifts in the band positions relative to those for a different sample (Wyoming bentonite) reported by the authors elsewhere. 12 In both of the authors' samples the band at around 1.5 pm was negligible for the oven-dried material because of the low temperature of the first stage of dehydration. Micas There is a considerable difference between the near-IR and UV - visible absorption spectra of muscovite [KA12(Si3A1)-010( OH)2] and biotite [K(Mg,Fe)3( Si3Al)O10(OH)2]. The near-IR PA spectrum of muscovite [Fig. 4(d)] is strikingly similar to the reflectance spectrum,4 with strong absorption bands at 1.47 2.26 2.40 2.44 and 2.48 pm.It also closely resembles a PA absorption spectrum reported elsewhere. The TG trace showed mass loss from 670 "C becoming more rapid at around 850 "C. The strong OH bands are not present in the PA spectrum at these elevated temperatures because of the dehydroxylation. Muscovite absorbs very little light over the range 350-800 nm until strongly heated. At 860 and 1015 "C the PA spectra are similar to that of talc heated to 1040 "C. The near-IR PA spectrum of biotite is very different from that of muscovite in that it 105 "C I I 1.2 1.6 2.0 2.4 Wavelengthipm (Near IR) 4 1.2 1.6 2.0 2.4 Wave I e n g t h ill m \ (UV -visible) hl "C is dominated bythe broad ( b ) Talc (UV - visible) I I ~ ~~ 300 500 700 Wavelengthlnm Muscovite (Near IR) (4 1; r, I \ r" I \v' n I 1.2 1.6 2.0 2.4 Wave I en g t h i 11 r n Biotite (Near IR) I I I 1 I I I 1.2 1.6 2.0 2.4 300 500 700 Wavelengthiyrn Wavelengthin m Fig.4. temperatures indicated PA spectra of sheet silicate minerals heated in air to the absorption band in the 0.8-2.0 pm region caused by Fe(I1) and Fe(II1) ions,4 and exhibits negligible fine structure in this respect closely resembling the reflectance spectrum. The near-IR and UV - visible PA spectra indicate oxidation of Fe(I1) to Fe(II1) at elevated temperature [see Fig. 4(e) and The near-IR spectrum of oven-dried hydrous mica ("illite") shows relatively little fine structure because dehydration invariably commences for this type of mica at a temperature below 100 "C with further dehydroxylation taking place at around 550 "C.14 This second DTA peak occurred at 585 "C for the sample studied. The unresolved series of peaks at around 2.4-2.5 pm were still present after the sample was heated at 420 "C but disappeared after heating at 590 "C [Fig. 5(a)]. An unexpected feature of the DTA trace was a small endothermic peak at 420 "C. Heating at this temperature caused the major change in the UV - visible PA spectrum which was similar to that obtained after heating at 590 "C as shown in Fig. 5(b), presumably owing to a change in oxidation state of iron present as an impurity. 01-Kaolinite The DTA curves of kaolinites [A12Si205(OH)4] are dominated by a strong endothermic dehydroxylation peak at 500-700 "C, and a sharp exothermic peak at around 1000 "C.14 On the samples studied slight mass loss commenced at around 450 "C.The near-IR PA spectrum [Fig. 5(c)] showed strong bands at 1.47,2.23 and 2.26 pm and weaker bands at 1.88,1.96 and 2.43 pm after drying at 105 "C. The bands at 1.47 and 1.96 pm were in fact much more intense on air-dried samples. After dehydroxylation at 600 "C only weak bands at around 1.9 pm and a stronger band at 2.26 vm remained. This spectrum is very similar to the reflectance spectra reported by Hunt and Salisbury.4 Another sample of kaolinite from Japan gave a very similar PA spectrum and behaved in the same way on heating. However for this sample the 2.23-ym band was stronger than that at 2.26 pm. Hydrous mica (,a '\ 105°C 570 "C I I I 1 1.2 1.6 2.0 2.4 Wavelengthiprn 600-1 000 "C I L I 1.2 1.6 2.0 2.4 Wavelengthiyrn Bauxite (Near (el I 101.0 "C (UV - visible) 920-990 "C 1 /\\,'y 1 105-200 "C .- - - -I 1 1-300 500 700 900 Wavelengthinm Allophane (4 (Near Air dried 105°C 1 I _ - - -I ' IR) I j 1.2 1.6 2.0 2.4 Wavelengthipm (Near IR) 3 7 0 3 7 Gypsum .cn Fig.5. PA spectra of various minerals. heated in air to the temperatures indicate ANALYST MARCH 1984 VOL. 109 223 Allophane There are several minerals included under the general name allophanels [mA1203.nSi02.pH20]. The authors’ sample came from Japan and its behaviour is probably typical of a wide range of these materials. The changes in the near-IR PA spectrum on heating this sample occur below normal oven-drying temperature.Thus although there is a considerable change on oven drying an air-dried sample subsequent changes on heating at 600 “C were negligible. On heating at 105 “C peaks at 1.48 2.44 and 2.52 pm were lost and that at 1.99 pm was greatly reduced [Fig. 5 ( d ) ] . Feldspars The feldspars are generally thermally inert the examples studied giving mass losses of <1% over the range 20-1 000 “C. A small endotherm was obtained for oligoclase at 908 “C. The PA spectra in the near-IR region were generally of limited diagnostic value showing only peaks corresponding to liquid-filled vacuoles very similar to those observed for reflectance spectra.4 This was true for microcline orthoclase anortho-clase albite and oligoclase the five samples studied.Changes occurred in the UV - visible PA spectra on heating to around 1000 “C owing to the presence of small amounts of iron. Quartz As might be cxpected quartz shows little of interest in TG or in its near-IR PA spectrum although weak included-water bands were observed at ca. 1.9 2.5 and 2.6 pm for milky quartz. Heating one of the samples studied at 1000 “C produced weak spectral features in the UV - visible region attributable to iron. This is surprising unless the sample was slightly contaminated with iron during milling although a tungsten carbide ball-mill was used and this seems unlikely. Other Minerals Bauxite A sample of bauxite gave a sharp endothermic peak only at 320 “C and was therefore predominantly gibbsite [y-Al(OH)3].The near-IR PA spectrum [Fig. 5(e)] was also similar to the reflectance spectrum reported for a sample of gibbsite from Brazil.6 Dehydration proceeded rapidly from 245 “C with a distinct change of slope in the TG trace at 315 “C. Mass loss was complete only at ca. 600 “C. The effect of heating on the near-IR PA spectra also indicates clearly that dehydroxylation proceeds in at least two stages the formation of y-A100H occurring as the first stage.8 Gypsum The DTA trace for gypsum showed the expected double endotherm the first corresponding to loss of water to give the hemihydrate CaS04. ‘/2Hz0 and the higher temperature endotherm corresponding to dehydration to anhydrite. At ambient temperature gypsum shows very strong absorption bands at 2.02 and 2.30 pm and at 2.50 and 2.57 pm (unresolved).with weaker bands at around 1.5 pm. The PA signal intensity is reduced significantly on oven drying at 105 “C through partial dehydration [Fig. 5 0 1 . Dehydration was still incomplete after heating at 370 “C but complete at 1000 “C. Conclusion As might have been expected the near-IR region of the PA spectra of minerals is dominated by OH with a lesser contribution coming from C032- and SO1’- radicals vibrating as entities in the mineral structure. Therefore. minerals where 0’- is the predominant anionic species the feldspars for example show no fine structure of diagnostic value and little change on heating. In some respects this contributes to improved selectivity for the qualitative analysis of mixtures. However these minerals are also often difficult to identify conclusively by other thermal methods of analysis.The UV - visible PA spectra for the minerals studied are dominated by the presence of iron and its position(s) and oxidation state within the mineral structure. The results for the characterisation of iron oxides oxyhydroxides and hydroxides are encouraging especially bearing in mind that differentiation between some of these substances may be difficult by alternative techniques. One of the strengths of PAS is that it may be used to characterise both surface coatings and the material immediately underlying these coatings to a certain extent. PAS may therefore be a useful tool for investigating surface reactions that occur during DTA or TG. However PA results must sometimes be interpreted with caution.The lepidocrocite sample in this study was shown conclusively from magnetic susceptibility measurements to have been converted to a maghemite on moderate heating, whereas the PA spectrum of the particle surfaces was closer to that of haematite indicating a surface phase change. Thus it should always be remembered that PA spectra identify surface minerals in non-uniform materials. Some hydrated minerals of interest gave particularly strong absorption bands in the near-IR region. It seems probable that near-IR PAS may be useful for the quantitative determination of minerals such as gypsum. It is worth considering briefly the potential usefulness of the technique for qualitative and/or semi-quantitative analysis of relatively simple mixtures of X-ray amorphous minerals.If two or more minerals contain OH groups but undergo dehydroxylation at considerably different temperatures then a careful choice of heating conditions may allow positive identification. The presence of minerals that do not contain OH should not complicate such procedures. Where carbonate is present the possibility exists for improving selectivity by simple acid dissolution. The latter approach may in fact be particularly suitable for confirming the presence of carbonate suspected from a near-IR PA spectrum. The authors are indebted to The Royal Society London for the award of a grant for the purchase of the PA spectrometer and to Anthony Edwards Prof. T. Fujisawa David Grierson, Fiona Mitchell and Hazel Moir for the contributions they have made to this study.1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. References Schmidt R. L. in Stucki J . W. and Banwart W. L. Editors, “Advanced Chemical Methods for Soil and Clay Minerals Research,” Reidel Dordrecht 1980 p. 451. Adams M. J . Beadle B. C. King A. A. and Kirkbright, G. F . Analyst 1976 101 553. Adams M. J . Beadle B. C. and Kirkbright G. F. Analyst, 1977,102.569. Hunt G. R. and Salisbury J . W. Mod. Geol. 1970,l. 283. Hunt G. R. and Salisbury J . W. Mod. Geol. 1970.2,23. Hunt G. R . Salisbury J . W. and Lenhoff C. J . Mod. Geol., 1971 2 195. Schwertmann U . and Taylor R. M. in Dixon J . B. and Weed S. B. Editors “Minerals in Soil Environments,” Soil Science Society of America Madison WI 1977 Chapter 5 p. 145. Mackenzie R. C. and Berggren G. in Mackenzie R. C Editor “Differential Thermal Analysis”. Academic Press, London 1970 Chapter 9 p. 271. Mullins C. E. J . SoilSci. 1977,28,223. Scheffer F . Meyer B. and Babel U . Beitr. Miner. Petrogr., 1959,6,371. Stacey F. D. and Banerjee S . K . “The Physical Principles of Rock Magnetism,” Elsevier. Amsterdam 1973. Livesey. N . T. and Cresser M. S . in Miller B. Editor, “Thermal Analysis.” Volume 1 Wiley Chichester. 1982. p. 325 224 ANALYST MARCH 1984 VOL. 109 13. Webb T. L. and Kruger J. E. in Mackenzie R. C . Editor, “Differential Thermal Analysis,” Academic Press London, 1970. Chapter 10. p. 303. Mackenzie R. C. in Mackenzie R. C. Editor “Differential Thermal Analysis,” Academic Press London 1970. Chapter 18 p. 497. 15. Wada K. in Dixon J . B . and Weed S. B. Editors “Minerals in Soil Environments,” Soil Science Society of America, Madison WI 1977 Chapter 16 p. 603. Paper A31224 Received July 25th 1983 Accepted August 25th 1983 14

ISSN:0003-2654    

DOI:10.1039/AN9840900219

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST, MARCH 1984, VOL. 109 225 Determination of the Stoicheiometry of Uranium Dioxide by Differential-pulse Polarography* Pier Luigi Buldini and Donatella Ferri CNR-Lamel, Laboratorio Analisi Chimica dei Materiali, Via dell'ldraulico 17/2, 1-40 127 Bologna, Italy and Ego Pauluzzi and Mario Zambianchi Agip Nucleare, Direzione Laboratori, Via Sabbionara 61 1, 1-40059 Medicina, Italy Uranium dioxide is widely used as a nuclear fuel and usually it exists as a non-stoicheiometric hyperstate U02 + because of the oxygen interstitial arrangement. The proposed method for determining the oxygen to uranium ratio in uranium oxides is based on the dissolution of the nuclear fuel in concentrated phosphoric acid under an inert atmosphere, to preserve the uranium oxidation states. After complete dissolution, sulphuric acid is added in order to obtain a 1.47 M H3P04 - 1.5 M H2S04 supporting electrolyte.Differential- pulse polarographic determination of uranium(V1) directly follows a t -0.09 V versus S.C.E. An aliquot of this solution is then oxidised with an almost equivalent amount of cerium(lV) sulphate solution, converting all uranium(lV) into uranium (Vl); the total uranium content is then determined in the same way. The proposed method permits determinations of uranium(V1) levels as low as 0.2 pg ml-1 with a relative standard deviation of about 2%. The oxygen to uranium ratio is calculated by the equation O/U = 2.000 0 + U(Vl)/total U and a result of 0.001 unit is obtainable with a coefficient of variation of about +0.1%. Keywords: Uranium dioxide; stoicheiometry; differential-pulse polarograph y; nuclear fuels Uranium dioxide is well known as a nuclear fuel; its composition approaches the theoretical ratio of uranium to oxygen of 1 : 2, but the exact stoicheiometry is seldom attained and usually uranium dioxide exists as a hyperstate U02+x because of the oxygen interstitial arrangement.Knowledge of the oxygen to uranium ratio is required in nuclear fuel specification and it is important for fuel behaviour in the reactor cycle. The oxygen to uranium ratio in uranium oxides has been determined by several methods: fusion, ignition. titrimetry and the determination of uranium(V1) in the oxide. The determination of uranium(V1) is the most precise of these methods and perhaps the least susceptible to interference. Oxygen is chemisorbed interstitially in the crystal lattice of uranium dioxide and oxidises uranium(1V) to a higher oxidation state.Thus, the determination of the uranium(V1) serves as a measure of the excess oxygen and, coupled with total uranium determination, as a way to determine the exact composition of the oxide. Several techniques, such as potentiometry,lJ spectrometry3 and coulometry,4~5 have been applied to the determination of uranium(V1) in the oxide. Polarography has also been applied to the determination of the oxygen to uranium ratio,G8 but no advantage was taken either of the use of modern pulsed techniques or of the uranium(V1) and total uranium determi- nation. Modern pulsed techniques permit a highly accurate determination of the stoicheiometry values close to the theoretical one and the determination of U(V1) and total uranium with the same technique under the same conditions makes the calibration step unnecessary because the oxygen to uranium ratio is calculated as a ratio of U(V1) and U(IV + VI) without needing to know the absolute values.The method described here is based on the dissolution of the nuclear fuel in non-oxidising conditions with concentrated phosphoric acid under an argon stream. After complete dissolution, sulphuric acid is added in order to obtain a 1.47 M H3P04 - 1.5 M H2S04 supporting electrolyte. A differential- pulse polarographic determination of uranium(V1) follows. An aliquot of this solution is then oxidised with an almost equivalent amount of cerium(1V) sulphate solution, so con- verting all uranium( IV) into uranium(V1) and determining total uranium in the same way.Experimental Apparatus A Metrohm (Herisau, Switzerland) E 506 Polarecord, equipped with an E 505 stand, was used. A forced-drop time of 2 s was imposed on the EA 1019/1 dropping-mercury electrode (DME). A Metrohm Model E A 285 platinum-wire counter electrode and a Metrohm Model EA 427 silver - silver chloride reference electrode were used. The differential-pulse polarographic conditions used were: DP, damping zero, mm/tdrop 0.5 and Upulse -100 mV. i * Research performed within the framework of the AGIP NUCLEARE - ENEA agreement for thermal reactor fuel development. I , 5crn Steel (AISI 316) mortar. A, Nuclear fuel pellet Fig. 1.226 ANALYST. MARCH 1984, VOL.109 Fig. 2. Pyrex dissolution flask. A , Ar bubbler; capacity 50 ml I I Y 1 - 0 2 4 6 8 [HJ’041IM Fig. 3. Effect of H,PO, concentration. U(V1) concentration, 2 pg ml- I : polarographic conditions as reported under Experimental The solutions were de-aerated with pure nitrogen for 5 min before polarographic analysis. All measurements were carried out at 25.0 k 0.1 “C. Nuclear-fuel pellets were crushed to a coarse powder in a steel mortar (see Fig. 1) and then the sample was directly weighed into the Pyrex dissolution flask shown in Fig. 2. The calibrated flasks were filled with concentrated phos- phoric acid, heated, left overnight and rinsed with doubly distilled water before use. If a set of equipment is kept for uranium determination only, it is sufficient to wash it after use with distilled water until it is required again.Reagents Erbatron electronic-grade reagents were used. The cerium- (IV) sulphate was supplied by Merck (Darmstadt, FRG) and its saturated solution was obtained by dissolving 3.4 g of Ce(S04)2 in 2 M H2S04 in a 50-ml calibrated flask. Normal precautions for trace analysis were taken throughout. Working standards were prepared by diluting a uran- ium(V1) stock solution (1 000 pg ml-1) obtained by dissolving 0.1179 g of U30R (JMC Specpure 765) in 4 ml of hot 1 + 1 nitric acid. The solution was twice evaporated to dryness. After the nitric acid had been completely removed, the residue was dissolved by adding 20 ml of hot 85% H3P04 and the solution was diluted to 100 ml with doubly distilled water.Study of Polarographic Working Conditions The polarographic behaviour of uranium is well known.y.10 In order to couple the highest sensitivity with the best sample dissolution requirements, it was necessary to define the best working conditions with respect to the supporting electrolyte composition, the formation and oxidation kinetics and the influence of the oxidising agent. Fig. 4. Effect of HSO1 concentration. U(V1) concentration, 2 pg ml- 1; polarographic conditions as reported under Experimental: supporting electrolyte. 1.47 M H,PO, Table 1. Interferences due to foreign species Mass excebs Ion over U(V1) ( 1 pg) . . . . . . AI(II1) 5 Cu(I1) 1 Fe(1II) . . . . . . 0.25 Mn(I1) . . . . . . 5 Pb(I1) . . . . . . 5 Si(1V) . . . . . . 5 Ti(1V) . . . . . .0.12 Zn(I1) . . . . . . 5 Cr(V1). . . . . . 0” 25 . . . . . . Mo(V1) . . . . 0.5 Error. O/O -7.5 -8.0 -3.0 -4.0 -5.0 - 10.0 - 10.0 -7.0 +5.0 -7.0 Because hot concentrated phosphoric acid is used to dissolve the nuclear fuel under an inert environment in order not to influence the uranium valency, the same acid was considered as the supporting electrolyte. As shown in Fig. 3, the sensitivity of the peak response in dilute phosphoric acid is better than that in concentrated acid, but it must be pointed out that uranium phosphate gels may be formed when the acid concentration is less than 0.5-1 M . The optimum acid concen- tration seems to be 3 M . Sulphuric acid was added in order to avoid gel formation without using more concentrated phosphoric acid, because it was found that the addition of sulphuric acid does not influence the uranium peak when present in concentrations up to 1.5 M (Fig.4); the best sensitivity was reached when a 1.47 M H3P04 - 1.5 M H2S04 supporting electrolyte was used. When using this supporting electrolyte it was shown that the uranium(V1) response is constant for about 3 h and after this period it decreases. The amount of phosphoric acid required for the supporting electrolyte was found to dissolve up to 200 mg of sample; the dissolution time may vary from 10-15 min to 2 h, depending on the coarse powder size. An aliquot of the sample solution was oxidised with a saturated solution of cerium(1V) sulphate in 2 M sulphuric acid. The presence of a drop of methyl red indicator, changing its colour from red to yellow, indicates an excess of cerium- (IV) and the completion of the uranium(1V) oxidation to uranium(V1).It was shown that this oxidation step is completed within 5 min at room temperature, and that it is also stable when a small excess of oxidising agent is present. The excess is not critical; the peak response does not change even if a 40 : 1 cerium excess over uranium is present. Effect of Foreign Species More than 20 elements including aluminium, bismuth, cad- mium, cerium, chromium(II1) and -(VI), cobalt, copper, lead, manganese, molybdenum, nickel, rare earths, silicon, titan- ium, vanadium, zinc and zirconium were tested. Table 1 shows the effect of the ions that interfere. No interferences were observed in the presence of greater than a 40-fold mass excess of the other elements mentioned above.ANALYST.MARCH 1984. VOL. 109 227 Table 2. Oxygen to uranium ratio of nuclear-fuel pellets Re fe re n ce G r avi me t r i c a n a 1 y si s : Po I a rug r a p h ic an a I ys is : Type value OiU k s.d. (95%) OiU k s.d. (9S%) poi1 ‘82 . . . . 2.10 Po’2i82 . . . . 2.10 Pw2i82iUG . . Po/2/82/UG-C . . Po/YX2/UG-B . . P’58G . . . . 2.02-2.03 Pi58GiB5 . . P:58GiB1 . . Pi58G:CS . . PI58GIC1 . . PiBL . . . . 2.0&2.01 765 Specpure 2.667 UiO, JMC 2. 1 00 t 0.002 2.105 2r 0.002 2.097 5 k 0.002 0 2.1040 2ro.001 2 2. 103 4 t 0.002 2 2.104 0 k 0.001 5 2. 104 2 1 0.002 0 2.021 5 1 0.001 2 2.022 2 k 0.001 6 2.019 5 +_ 0.001 0 2.023 4 2 0.001 4 2.020 2 k 0.001 9 2.008 5 t 0.001 0 2.672 k 0.008 No. o f de t e rm i n at io ns 10 8 8 8 x 10 8 8 8 8 10 20 Polarographic analysisx: OiU k s.d.(95%) 2.096 5 t 0.003 5 2.108 5 +_ 0.005 4 2.106 2 k 0.003 2 2.103 0 t 0.002 0 2.106 4 2 0.002 8 2.021 2 Ik 0.001 8 2.023 0 k 0.002 2 2.020 s k 0.002 4 2.023 0 ? 0.002 0 2.020 4 t 0.002 4 2.009 0 t 0.001 5 2.669 k 0.008 No. of determinations 4 4 4 4 4 4 4 4 4 4 4 4 Procedure Crush the pellet sample to a coarse powder in a steel mortar (Fig. 1) and weigh 50-150 mg of the coarse powder into a Pyrex dissolution flask (Fig. 2 ) . Cover the sample with 5 ml of 85% H3P04 and de-aerate the flask by bubbling argon through it. After 10 min, heat the flask to 210 k 10 “C in order to complete the dissolution of the sample. Keep argon bubbling at a rate that ensures an inert atmosphere is maintained over the solution during both the dissolution step and the cooling stage.Add 12.5 ml of 6 M H2S04, washing the bubbler and the neck of the flask. Dilute to the mark with doubly distilled water with constant shaking in order to avoid gel formation at the interface. Transfer 1 ml of this solution into a 25-ml calibrated flask. Add one drop of 0.1% methyl red indicator and add, dropwise, the saturated solution of cerium(1V) sulphate in 2 M H2SO4, until the colour of the solution changes from red to yellow. Wait for 5 min and add 2.4 ml of 85% H3P04 and 5.75 ml of 6 M H2S04. Dilute to the mark with doubly distilled water. De-aerate both solutions for 5 min and record the polaro- grams from +0.4 to -0.5 V versus S.C.E. The absolute value of the uranium content can be calculated by means of a calibration graph or standard additions method.The uranium(V1) content of the sample may be calculated from the first polarogram and, after uranium(1V) oxidation, total uranium is found from the second polarogram. Results The described method was tested on certified U308 and compared with reference values, results from gravimetric analysis. when possible, and polarography.8 Some results are shown in Table 2. The polarographic data are substantially coincident. The method described by Papei et a1.8 is more time consuming and less precise, probably owing to the errors occurring when pipetting the Ce(SO& - K2Cr207 titrant and the determina- tion of the equivalence point of amperometric titration of total uranium. Several pellets were analysed and then spiked with uran- ium(V1) standard solution, to provide an amount of several milligrams of uranium(VI), and re-analysed.The added amount of uranium was recovered without noting any uranium(1V) oxidation. The proposed method allows good accuracy for samples covering a wide range of oxygen to uranium ratios. When the prescribed conditions are used, the method permits the determination of uranium at concentra- tions as low as 0.2 pg ml-1 of U(VI), with a relative standard deviation of about 2%. The calibration graph is linear and reproducible up to The oxygen to uranium ratio can be calculated by the uranium concentrations of at least 0.35 mg ml-1 of U(V1). following equation: CIVl fu - _ u(V1) = 2.000 0 + - * - O - 2.0000 + U(IV + VI) fu c2v2 U h = 2.0000 + 25h2 where C1 and C2 are the U(V1) and U(1V + VI) contents (mg ml-l), M the mass of the sample, h , is the height of the uranium(V1) polarographic peak and h2 is the height of the total uranium peak.The proposed method is very simple and rapid. It needs only one dissolution to determine both uranium(V1) and total uranium. Excluding the dissolution step, a few minutes are necessary for both determinations. The presence of other oxides andi’or sintering additives in the nuclear-fuel pellets did not influence the uranium determi- nation. In order to evaluate the influence of atmospheric oxygen on the uranium(1V) oxidation, crushing and weighing steps were performed under an inert atmosphere in a glove-box and then both deoxygenated sulphuric acid solution and doubly dis- tilled water were added to the sample solution. No difference was noted, even when analysing nuclear fuels having an oxygen to uranium ratio close to 2, in comparison with performing normal manipulations as previously described. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. References Herczynska, E., Nukleonica, 1976, 21, 285. Kuvik, V., Krtil, J., and Moravec, A , , Radiochem. Radioanal. Lett., 1982, 54, 209. Khun, E., Baumgaertel, G . , Schmieder, H., and Goergenyi, T., Fresenius Z. Anal. Chem., 1973, 267, 103. Schaefer, E. A . , and Hibbits, J. O., Anal. Chem.. 1969, 41. 254. Bartscher. W., Fresenius Z . Anal. Chem., 1982, 310, 413. Sipos, L.. and Branica, M., J . Polarogr. Soc.. 1968, 14, 3. Cheng-I Wu, Fu-Chung Chang and Yu-Chai Yeh, Nucl. Sci. J., 1981, 18, 108. Papei. V., Veternik, J . , and Krtil, J . , Rudiochum. Radioanal. Lett., 1982, 55. 141. Szefer, P., Mikrochim. Acta, 1979. I , 463. Sinyakova, S. I . , in Ryabchikov, D. I., and Senyavin, M. M., Editors, “Analytical Chemistry of Uranium,” IPST, Jerusalem, 1963, pp. 134-176. Paper A311 87 Received June 27th, 1983 Accepted September 12th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9840900225

出版商:RSC    

年代:1984

数据来源: RSC