JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 13N Atomic Absorption Spectroscopy-Present and Future Aspects Ralph E. Sturgeon Institute for Environmental Chemistry National Research Council of Canada Ottawa Ontario KIA OR9 Canada This XXVII Colloquium Spectroscop- icum Internationale has chosen to honour Sir Alan Walsh with its first CSI-Award for major scientific contri- butions to analytical spectroscopy. It is equally an honour and a great privilege to participate in this event and I extend my congratulations to Sir Alan the recognized ‘Father of AAS’. Although Sir Alan has published widely in areas of atomic infrared and Raman spectroscopy he is universally noted for his pioneering contributions in atomic absorption spectrometry (AAS). Since the initial inception of the concept as revealed in his land- mark paper of 1955,’ Walsh perceived the significance of this technique which would revolutionize analytical atomic spectroscopy and pursued it with the conviction necessary to estab- lish it as an acceptable methodology.2 As an instrumental method it has had until recently few equals in popu- larity as of 1986 AAS was ranked the most significant advance to occur in analytical chemistry in the past 50 years.3 Recent market projections4 suggest a 3.8% sales increase for AAS spectrometers in 199 1 to be surpassed for the first time by ICP-AES pur- chases.What does the future hold for AAS? This question has been ad- dressed numerous times in the p a ~ t * . ~ - ~ ~ and I will attempt to examine this perspective once again. This exer- cise is most often indulged in with techniques which have been deemed to have evolved through at least several of the latter ‘seven stages of instru- ment development’1° and may be pre- sently residing in the age of senes- cence.Having already celebrated its ‘silver jubilee’ more than a decade ago,” it certainly merits broad accep- tance. The major challenge facing the fu- ture of AAS stems from increased competition from newly emerging or ‘rediscovered’ (e.g. glow discharge) spectroscopic techniques. Its con- tinued viability will only be assured through successful evolution and ad- aptation of more productive instru- mentation. In comparison with other popular state-of-the-art spectrochemical tech- niques AAS proffers a number of attractive and unique features for rou- tine analyses as well as fundamental investigation,*J2 not the least of which are its competitive cost per analysis high detection power [for electrother- mal AAS (ETAAS)] and simplistic instrumentation.It is well recognized however that the major shortcoming is sample throughput AAS is by de- sign a ‘single-element at a time’ tech- nique. This can be attributed primarily to the source-detector arrangement and to some extent to the atomizer. If this problem is to be addressed major changes/modifications will have to be implemented to have any impact on these areas. Current international re- search efforts have identified and ex- plored a number of options and these will be the target of discussion of this text. Flame AAS (FAAS) suffers most acutely from competition with induc- tively coupled plasma atomic emission spectrometry (ICP-AES).The latters’ multi-element capability (fast sequen- tial and simultaneous) and detection power that rivals or surpasses FAAS especially for the refractory elements have placed FAAS in a vulnerable position. Instrumentation for FAAS has altered little since Walsh’s first description of the method.’ Since the introduction of the dinitrogen oxide- acetylene flame by Willis in 1965 there has been no subsequent major advance in flame methods ‘which appear to have reached a plateau of de~elopment’.’~ Although this state- ment was made by Walsh more than a decade ago it remains valid today. A limited amount of research continues to be invested in the design of more efficient burner heads14 but the major-14N JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 ity of interest in the fabrication and characterization of new nebulizers and spray chambers appears to lie with ICP users. Automation of commercial FAAS instruments is currently well devel- oped permitting some 500 determina- tions per hour. However as has been noted,I5 such instruments can be de- scribed as multi-sample rather than multi-element in that they repetitively determine the same element in all samples prior to quantifying the next element. An instrumental approach to multi-element capability has been de- scribedI5 using conventional hollow cathode line sources having the capa- city for an estimated 1200 determina- tions per hour-very competitive with sequential ICP based instrumentation. Although automation of instrumenta- tion via onboard microprocessor con- trol has revolutionized performance and (unattended) sample throughput of FAAS as well as other atomic spectrometric techniques it is hoped that the intelligent instrument of the future will go beyond these narrow confines and actively participate in the implementation of quality assurance/ quality control of the data.16 Although unfortunately still consi- dered within the realm of a laboratory research tool the wavelength modu- lated continuum source based instru- ment conceived by O'Haver and col- leagues provides an attractive and viable approach to true multi-element AAS when coupled to a high-resolu- tion polychr~mator.~~*~~ When inter- faced to photodiode array detectors the optics are simplified by omission of the wavelength m o d ~ l a t i o n ~ ~ (as well as the need for multiple PMTs).Currently this approach is limited in scope to those elements having reso- nance lines lying above 280 nm be- cause of the poor source intensity at shorter wavelengths. This of course precludes application to the measure- ment of many elements of current environmental interest. This limita- tion may be lifted as a result of promising recent studies aimed at pulsing the continuum source to high intensity levels thereby boosting the UV output. Moulton et aLto reported a signal-to-noise ratio (S/N) increase of 1.4-fold at 24 1 nm by using a pulsed continuum source in conjunction with linear photodiode array detection.An improvement factor of 6.7-fold was noted over the case of PMT detection with a non-pulsed lamp. It is expected that significantly greater benefits in S/N will be reaped with larger pulse currents. An alternative approach to achiev- ing the resolution needed for con- tinuum source AAS is by application of interferometry. Although a Fourier transform AA spectrometer provides all of the advantages of other con- tinuum source dispe:rsive systems it can be subject to a multiplex disadvan- tage when a practical free spectral range is desired and. at high resolu- tion suffers from long scan times.21 How will such research impact on the practising analyst and the market for FAAS? The answer of course is. little unless a commercial instrument becomes available. Despite the ac- knowledged benefits of multi-element capability faster throughput compar- able detection power and greater dy- namic range the increased complexity of utilizing a high-resolution polychro- mator offsets the current cost advan- tage cited for FAAS over its ICP-AES competitor. This is not the whole picture however because such an in- strument would also find application with the graphite furnace and this is the area that could revolutionize the use of AAS.A cursory examination of current analytical spectroscopic litera- ture reveals that it is the graphite furnace atomizer that is of prime interest for further development (not only for AAS but as an atomizer for use in a variety of ,atomic and mass spectroscopic instrumentation). The introduction of the graphite furnace by L'vov in 195922 was even- tually to elevate AAS to a foremost position amongst preferred techniques for ultratrace analysis.The reason was clear all of the inherent advantages of AAS were coupled to a highly efficient atomization device.23 Significant im- provements in signal processing fur- nace design and operation have oc- curred in the last decade and many of these are now embodied in the concept of the stabilized temperature platform furnace (STPF) pioneered by Slavin and c o - w o ~ k e r s . ~ ~ ~ ~ ~ The remarkable success of this relatively interference free technique has permitted extension of the procedure to the analysis of solids and and has advanced to the point where the concept of absolute (standardless) analyses may be realistically e n t e ~ t a i n e d .~ ~ . ~ ~ With the recent release of a commercial graphite furnace atomizer (Perkin- Elmer) based on the design of the spatially isothermal cuvette of Frech et al.,29 analysts are expected to have even greater freedomi from matrix in- terferences as well as reduced spectral background and memory effects. Of course matrix effects cannot be com- pletely eliminated even for constant temperature atomize~rs.~~ Unfortunately graphite furnace (GF) techniques are aho characterized by the single-element-at-a-time AAS syndrome which was further com- pounded by the need to establish indi- vidual atomization programmes op- timized for each element or a few broad classes of elements-a situation which turned out to be more inflexible than that encountered by the choice of two different flames in FAAS.Conven- tional single-channel GF instruments usually cannot achieve more than 20-30 determinations per hour in a multi-sample approach. This rate may be increased up to 100 per hour if the furnace cycle time is reduced to a minimum using such measures as hot injections elimination of the drying and/or char stage and reducing the delay between injections by the auto- sampler.31 Bank et ~ 1 . ~ ~ 9 ~ ~ recently de- monstrated the feasibility of flow in- jection thermospray deposition for ETAAS which boosted potential throughput into the 150-200 per hour range while retaining sensitivity preci- sion and the use of pl volumes. The excellent scope for the correction of high non-specific background absorp- tion offered by Zeeman-effect systems serves to more easily implement such modifications to the thermal pre-treat- ment of the sample.Although commercially available 2 and 4 channel instruments (Hitachi and Thermo Jarrell Ash) could serve to increase the multi-sample efficiency of the technique further increased com- petition from the now well established field of ICP mass spectrometry (ICP- MS) will eventually require that ETAAS adopts a true multi-element approach to survive. In this respect the shortcomings of AAS highlighted by Hieftje,7 i.e. those resident in the hollow cathode light (HCL) source and the atomizer need to be addressed and promising alternatives pursued. Multi-element AAS necessitates a multi-wavelength source and com- patible detector. Systems based on multiple HCLs rapidly become im- practical with more than only a few elements and with ETAAS any at- tempt at rapid sequential measure- ment will fail.Only continuum sources and cheap tunable lasers inherently meet the necessary requirements. Dis- persive rather than interferometric means should be considered as a consequence of the cost and practical limitations associated with the lat- Continuum source based ap- proaches are presently the most attrac- tive. Those employing high-resolution polychromators with PMT detectors and wavelength modulation have been shown to provide exceptional perfor- mance for elements having resonance lines above 280 nm. Here detection limits are generally within a factor of 2-3 of those obtained with line sources background correction is rapid and accurate and dynamic range can be extended to cover 4-6 dec- a d e ~ .~ ~ As noted earlier useful access ter.7,1221JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 15N to resonance lines lying below 250 nm now appears promising using pulsed continuum source^*^^^^ but extensive work will be required to evaluate this comprehensively. Schmidt et ~ 1 . ~ ~ re- cently described such an approach based on an Cchelle spectrometer and a charge coupled linear array device as detector. Electrothermal AAS detec- tion limits achieved for Cd and Pb at 228.8 and 283.3 nm respectively were within a factor of two of their line source counterparts. The coupling of pulsed continuum sources and diode array detectors should also find application in the domain of coherent forward scattering wherein more intense sources should improve performance. This technique has already demonstrated useful multi-element results with commercial atomizers and 4-6 decades of linear range.36 Hergenroder and NiemaxJ7 recently demonstrated the feasibility of multi- element ETAAS using temperature and diode current controlled semicon- ductor diode lasers as sources. With modulated diode laser power (using electro-optic KDP crystals) the signal measured with a photodiode in a non- dispersive system can be subjected to Fourier analysis to separate multiple channels.Rapid square wave diode current modulation permits absorp- tion measurements to be made on/off line for simultaneous background cor- rection or in the wings of the profile for extended linear range.Compact multi channel capability may be realistically achieved using optical fibre techno- logy. The availability of diode lasers covering a wider wavelength range is anticipated. Successful solutions to the optical aspects of the multichannel AAS chal- lenge highlight the remaining deficien- cies resident in the atomizer. Pre- sently separate optimization of ther- mal parameters is generally required for each element due to either pre- atomization losses or incomplete atomization. Three solutions merit consideration the quest for and appli- cation of a ‘universal’ chemical modi- fier; the elimination of the thermal pre- treatment stage and the application of new atomizers. Multi-element analyses may be heavily dependent upon the use of chemical modification techniques in that compromise thermal conditions will be utilized.Reduced palladium appears to be emerging as a potentially universal modifier that will be useful for this purpose.38 It should be noted that the need for chemical modifica- tion techniques should be verified be- fore use on a sample. Manning and S l a ~ i n ~ ’ ” ~ have presented examples of systems in which adequate perform- ance can be achieved by eliminating the char stage and modifier com- pletely. The larger background which normally occurs is then easily handled with a Zeeman-based system. It is easier to conceive of universal com- promise conditions in such circum- stances. ‘If atomic absorption methods are to be substantially improved it seems inescapable that the advances can only result from improved methods of atomization’.This statement was penned by Walsh in 1980.13 With the conventional GF this has included implementation of ideas fostered by L‘vov,~~ i.e. vaporization of samples from a platform placed within the furnace (on which STPF techniques are now based) from a probe subse- quently inserted into an isothermal furnace41 and by rapid heating achieved through capacitive dis- charge.42 The last has proved to be too difficult to utilize routinely because of materials requirements whereas the performance of the probe technique can generally be matched through ap- plication of the simpler STPF con- ~ e p t . ~ ~ Similarly the spatially and tem- porally isothermal two-step atomizer described by Frech and J o n s ~ o n ~ ~ while deemed too complex to justify routine applicati~n,~~ may provide an excellent vehicle for implementation as an atomizer for multi-element M S .This configuration is amenable to automation and lends itself more eas- ily to the establishment of compromise conditions. It has been suggested that use could be made of non-thermal atomizers in particular the rare gas sputter systems which are advantageous in producing a low background and high yield of atomic vapour. A version of the popu- lar glow discharge lamp (Atomsource) has been introduced into the market- place and configured so as to optimize production of ground-state atoms in the analytical volume.*5 In such sys- tems there is a greater conformity between the composition of the sample and the vapour being analysed such that matrix effects may be greatly reduced and samples having a range of compositions can be analysed with the same working curve.‘To date all sput- tering work has been concerned with steady-state systems. Would it not be worthwhile considering the use of ca- thodic sputtering in association with the “total vaporization” method evolved by L‘vov?’. This remark by WalshI3 follows from his early work on sputtering at CSIRO (Australia) and recognizes the significance of L’vov’s approach which has been so successful with the GF. Recent studies by Chak- rabarti et al.46 confirmed the benefits of transient sputter atomization of discrete samples for which absolute detection limits rival those of conven- tional ETAAS. Although an evaluation of the effect of the matrix is awaited it is tempting to speculate that such an approach might present an attractive source for multi-element AAS.Non- conductive samples may likely be sputtered via r.f. discharge^.^' A hurdle to be surmounted in the quest for multi-element AAS is the limited linear range (currently 2-4 decades) which makes it necessary to undertake multiple dilutions in order to cover the variation of elemental concentrations present in a typical sample. Several means have been sug- gested to extend this figure of merit the most promising working concept for commercial instrumentation being use of three-field ax. Zeeman modula- t i ~ n ~ ~ which readily achieves a 5- 10 fold wider dynamic range. Alterna- tively instrumentation based on the ‘Smith-Hieftje’ pulsed HCL back- ground correction approach may take advantage of a corresponding multi- level current pulse to elicit different source profiles and thereby extend linear range.With continuum source based instruments the dynamic range can cover 4-6 decades by making use of information in the line wings. Thus it appears that this obstacle can be easily circumvented with existing tech- nology. The demand for reliable analytical data at ever decreasing concentration levels has outstripped current detec- tion capabilities of ETAAS for many elements. It is for this reason that methodology for enhancing detection power is evolving at a rapid pace. Such approaches include conventional off- line matrix separation and preconcen- tration schemes as well as in situ trapping of volatile forms of elements [e.g.hydrides Ni(C0)4 Pb(CzH!)4J.49 The latter are ultratrace techniques which make specific and profitable use of the GF and are ripe for automation. Throughput even in the single- element mode can be considerably enhanced utilizing flow injection tech- niques to speed both calibration proce- dures and sample analyses. On-line sample preparation systems for matrix management and analyte preconcen- t r a t i ~ n ~ ~ ? ~ ~ will have substantial im- pact on all areas of analytical atomic spectroscopy in the coming years. More difficult samples will be amen- able to quantitation at lower analyte concentrations with greater speed pre- cision and accuracy. As these develop- ments are likely to influence all instru- mentation without prejudice no rela- tive gains can be claimed by AAS proponents.However it is evident from the foregoing that greater atten- tion must be paid to enhancing the16N JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 versatility of GF autosamplers. Soft- ware control should permit the user to program the timing and mechanical events associated with this process. It is clear that it should be possible to provide AAS users with multi- element capability using present tech- nology In this regard current research directions suggest the suitability of an image detector approach in combina- tion with a pulsed continuum source and a GF or sputtering cell. Means will be devised for coping with the volumi- nous amount of data generated by the necessary two-dimensional detectors if a significant wavelength range is de- sired.52 However in the final analysis ‘it is unlikely that AAS would in its over-all capability surpass those tech- niques with which it is now competi- tive’17 (ie.ICP-AES and -MS). Of course it is not necessary perform- ance wise for future AAS instruments to surpass that of other spectrometric techniques in order to remain viable. Quite apart from its sustained contri- butions to fundamental and diagnostic studies of atomic systems AAS will continue to be used in a routine analyt- ical capacity. When methods are com- pared for ultratrace capability the power of detection and accuracy or reliability become the most important criteria.53 Inductively coupled plasma AES is gradually replacing FAAS and this trend will accelerate as the cost of ICP equipment continues to decline and the demand for multi-element capability rises in the face of environ- mental challenge and legislation.Elec- trothermal AAS however is currently ‘holding its own’ and will continue to act in a complimentary rather than redundant manner to enhance the capability of alternative spectrometric techniques. It is the acknowledged ‘benchmark’ against which all other commercial- and laboratory research- based atomic spectrometric tech- niques (including ICP-MS LEAFS LEI CFS FANES and FAPES) are currently and will continue to be gauged. The decline noted in the total number of AAS publications in the last 5 years7 in no way construes the demise of this technique-it only serves to reflect its broad and unchal- lenged acceptance.The remarkable fact that publications continue in this already mature discipline reflects the substantial interest of researchers and users alike. Of course one of the principal ad- vantages of absorption measurements is that they are amenable to undertak- ing absolute analyses. This subject has been raised earlier by RannS4 for the flame and by L‘vovz7 for the furnace. It would be no great surprise to find analysts using this approach in future at least for semiquantitative work. We are indebted to !Sir Alan Walsh for his many contributilons to the birth and growth of this fascinating and ubiquitous analytical tool. The ques- tion posed by him2 some 17 years ago ‘AAS-Stagnant or Pregnant?’ can still be answered with the same re- tort... ‘the subject has not really been stagnant but merely pregnant and has now given birth to new offsprings’. References 1 Walsh A. Spectrochim. Acta 1955 7 108. 2 Walsh A. Anal. Chem. 1974 46 698A. 3 Braun T. Fresenius’ Z. Anal. Chem. 1986,323 105. 4 Howard B. Am. Lab. 1991 Jan. 66. 5 Koirtyohann S. R. Anal. Chem. 1980 52 736A. 6 Slavin W. Trends Anal. Chem. 1987 6 194. 7 Hieftje G. M. J. And. At. Spectrom. 1989 4 11 7. 8 Sturgeon R. E. Fresenius’ 2. Anal. Chem. 1990,337 538. 9 Slavin W. Anal. C‘hem. 1982 54 685A. 10 Koirtyohann S. R. and Kaiser M. L. Anal. Chem. 1982 54 15 1 5A. 11 Boumans P. W. J. M. Spectrochim. Acta Part B 1980 35 637. 12 Hieftje G. M. Fresenius’ 2. Anal. # Chem. 1990,337 5289. 13 Walsh A. Spectrochim. Acta Part B 1980 35 639.14 I Willis J. B. Sturman. B. T. and Frary B. D. J. Anal. At. Spectrom. 1990 5 399. 15 Bernhard A. E. and Kahn H. L. Am. Lab. 1988 June 1261. 16 Hieftje G. M. Spectrochim. Acta 1990 44 (Spec. Suppl.) 11 3. 17 Zander A. T. O’Haver T. C. and Keliher P. N. Anal. Chem. 1976 48 1166. ’ ’18 Harnly J. M. Kane J. S. and Miller- Ihli N. J. Appl. Spelctrosc. 1982 36 637. 19 Jones B. T. Mignardi M. A. Smith B. W. and Winefordner J. D. J. Anal. At. Spectrom. 1989 4 647. 20 Moulton G. P. O’Haver T. C. and Harnly J. M. J. A n d . At. Spectrom. 1989 4 673. 21 Glick M. R. Jones B. T. Smith B. W. and Winefordner J. D. Anal. Chem. 1989,61 1694. 22 L‘vov B. V. Spectrochim. Acta 1961 17 761. 23 Falk H. and Tilch J. J. Anal. At. Spectrom. 1987 2 5.27. 24 Slavin W. Manning D.C. and Carn- rick G. R. At. Spectrosc. 1981 2 137. 25 Slavin W. and Carririck G. R. Am. Lab. 1988 Oct. 88. I 26 Epstein M. S. Carnrick G. R. Slavin W. and Miller-Ihli N. J. Anal. Chem. 1989,61 1414. 27 L‘vov B. V. Spectrochim. Acta Part B 1990 45 633. 28 Slavin W. and Carnrick G. R. Spec- trochim. Acta Part B 1984 39 27 1. 29 Frech W. Baxter D. and Hutsch B. Anal. Chem. 1986 58 1973. 30 Frech W. Cedergren A. Lundberg E. and Siemer D. D. Spectrochim. Acta Part B 1983 38 1435. 31 Slavin W. Manning D. C. and Carn- rick G. R. Spectrochim. Acta Part B 1989,44 1237. 32 Bank P. C. de Loos-Vollebregt M. T. C. and de Galan L. Spectrochim. Acta Part B 1988 43 983. 33 Bank P. C. de Loos-Vollebregt M. T. C. and de Galan L. Spectrochim. Acta Part B 1989 44 571. 34 Harnly J.M. Fresenius’ 2. Anal. Chem. 1986,323 759. 35 Schmidt K. P. Becker-Ross H. and Florek S. Spectrochim. Acta Part B 1990,45 1203. 36 Hermann G. Jung M. Lasnitschka G. Moder R. Scharmann A. and Zhou X . Spectrochim. Acta Part B 1990,45 763. 37 Hergenroder R. and Niemax K. Spectrochim. Acta Part B 1988 43 1443. 38 Schlemmer G. and Welz B. Spectro- chim. Acta Part B 1986 41 1 157. 39 Manning D. C. and Slavin W. Spec- trochim. Acta Part B 1988 43 11 57. 40 L‘vov B. V. Spectrochim. Acta Part B 1978 36 153. 41 Littlejohn D. Cook S. Durie D. and Ottaway J. M. Spectrochim. Acta Part B 1984 39 295. 42 Chang S. B. and Chakrabarti C. L. Prog. Anal. At Spectrosc. 1985 8 83. 43 Wu S. Chakrabarti C. L. and Rogers J. T. Prog. Anal. Spectrosc. 1987 10 111. 44 Frech W. and Jonsson S. Spectro- chim. Acta Part B 1982 37 1021. 45 Lundberg E. Frech W. Baxter D. and Cedergren A. Spectrochim. Acta Part B 1988 43 45 1. 46 Chakrabarti C. L. Headrick K. L. Hutton J. C. and Bertels P. C. Spec- trochim. Acta Part B 1991 46 183. 47 Duckworth D. C. and Marcus R. K. Anal. Chem. 1989 61 1879. 48 de Loos-Vollebregt M. T. C. Koot J. P. and Padmos J. J. Anal. At. Spec- trom. 1989 4 387. 49 Sturgeon R. E. Spectrochim. Acta Part B 1989,44 1209. 50 Karakaya A. and Taylor A. J. Anal. At. Spectrom. 1989 4 261. 51 Fang Z. Sperling M. and Welz B. J. Anal. At. Spectrom. 1990 5 639. 52 Kolczynski J. D. Radspinner D. A. Pomeroy R. S. Baker M. E. Norris J. A and Denton M. B. Am. Lab. 1991 May 48. 53 Tolg G. Analyst 1987 112 365. 54 Rann C. S. Spectrochim. Acta Part B 1968 23 827.