JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 69 New Developments and Final Frontiers in Inductively Coupled Plasma Spectrometry* Plenary Lecture Gary M. Hieftje P. J. Galley M. Glick and D. S. Hanselman Indiana University Department of Chemistry Bioomington IN 4 7405 USA A number of areas of study are indicated which are important in furthering the application and capability of the inductively coupled plasma. Featured especially are the importance of overcoming matrix and inter-element interferences the development of diagnostics to monitor instrument operation the use of adaptive computer- based feedback control to optimize instruments and the introduction of intelligent sample processing. Examples are taken from the author’s laboratory of ways in which new developments in instrumentation and understanding are permitting these frontiers to be conquered.Keywords inductively coupled plasma; matrix and inter-element interference; instrument operation; intelligent sample processing It has been just about 20 years since I first became aware of the inductively coupled plasma (ICP) and its potential power in analytical atomic spectrometry.’ Since that time the ICP has become perhaps the world’s most powerful tool for routine multi-element analysis. Applied mostly as a source for atomic emission spectrometry the ICP is used to analyse samples of interest in geology biomedicine petro- leum engineering the chemical industry forensic science environmental monitoring and others. Such an emission system is capable of providing literally thousands of elemental determinations each working hour most at the part per billion level.Somewhat more recently the ICP has also been used effectively as an ion source for elemental mass spectro- metry. In this role the source retains its high-throughput capability but offers also part per trillion detection limits excellent semi-quantitative determinations isotope analy- sis and isotope dilution capability and virtually complete elemental coverage across the Periodic Table. Considering these capabilities the widespread applica- tion of the ICP and the power of modern computer- controlled commercial ICP instrumentation it is appropri- ate to ask at the present time what important developments might take place that could appreciably enhance the power of this versatile source.Are there developments that could make the source even more beneficial? Are there important unsolved questions to which workers in ICP spectrometry should devote their attention? An examination of recent journal articles and a perusal of lectures and posters being presented at the conference at which the present paper was given2 dictates that these questions all be answered in the affirmative. In the following sections of this paper a number of new developments and remaining frontiers in ICP spectrometry will be described. Some of the challenges are to overcome remaining shortcomings of the source and of the instrumentation that utilizes it; others involve improving the utility convenience and reliability of the systems; and still others will emphasize taking greater advantage of the flood of data that the ICP is capable of producing. These ‘final frontiers* can in some instances be explored with current knowledge and technology.In other situations new developments will be needed. Of course in a document of reasonable length it is impossible to detail or even describe the many investiga- *Presented at the XXVII Colloquium Spectroscopicum Interna- tionale (CSI) Bergen Norway June 9- 14 I99 l . tions and improvements which might profitably be pursued in ICP spectrometry. To narrow the scope somewhat attention will be focused on ICP atomic emission spectro- metry. Firstly a brief overview of a number of the remaining frontiers in ICP emission spectrometry will be offered. Next only a few of those frontiers will be selected for amplification.It can be hoped that future manuscripts or other workers will cover the remaining topics in greater detail. The Final Frontiers A few of the ‘final frontiers* that remain in ICP emission spectrometry are compiled in Table 1. Although many readers will feel that other topics should have been included in the list few would dispute the importance of those that are cited. Interestingly many of the topics are related; efforts made to attack one of the ‘frontiers* will be likely to have an impact on others also. For example adding diagnostics to current instrumentation might help reduce the severity or incidence of matrix interferences especially if the operating characteristics of such instruments are adaptively controlled by an on-line computer.In an ideal embodiment of such a system the ICP itself might be monitored by for example a low-cost video camera equipped with spectral-selection capability. The spatial pattern of emission so monitored could then be used by an appropriately programmed control computer to adjust operating conditions of the plasma in a manner to alleviate inter-element interferences. More details will be provided on this topic later. Similarly an instrument that can adapt its operating conditions to different incoming samples would benefit from and also could be employed in schemes to automate sample preparation. Again more will be offered on this topic later. One of the challenges ,in Table 1 that is now being addressed but for which an optimum solution seems still Table 1 Final frontiers in atomic spectrometry Overcoming matrix interferences Automating and minimizing sample preparation Adding diagnostics to instrumentation Developing self-adaptive instrumentation Achieving total-spectrum simultaneous read-out Enhancing information extraction and display Improving precision70 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 Atom Source .) detection * Sample processing * not to be available is the rapid electronic and error-free recording of a complete emission spectrum from the ICP. The immediate availability of a full high-resolution spec- trum such as that recorded and processed much more slowly on a photographic film would permit far more reliable background correction the use of multiple or alternative elemental emission lines and would provide a diagnostic tool in itself useful for adaptive control of the instrument by an associated computer.Furthermore if the wavelength registration of such a full spectrum were sufficiently accurate it would enable new signal-processing options to be explored. Such options would include but would not be limited to spectral stripping the use of multilinear regression for calibration application of neural networks for sample recognition and characterization and noteworthy enhancements in information extraction and display. This last topic is a worthy one for examination in itself. New statistical algorithms of increased power are continu- ously being introduced that could enhance data treatment from an ICP spectrometer in important ways.New visual- display approaches in particular could enable a scientist or technician more rapidly to grasp the meaning of subtleties in a data set. Human beings are capable of perceiving amazingly subtle nuances in projected images. Consider for example our ability to recognize the mood even of a stranger and how difficult it would be to describe what minutiae of facial expression led us to the conclusion. In the same way large datasets involving elemental concentrations or even emission spectra themselves could perhaps be displayed to permit visual recognition of errors in plasma performance the presence of unexpected species the class to which a sample belongs or even the answer to an underlying question that the elemental analysis was intended to answer.For example the answer which an elemental analysis is expected to provide is ultimately not really ‘What are the elemental concentrations?’ but rather ‘Is the patient healthy?’ ‘Is the engine lubricant still useful? or ‘Is the site from which this geological sample was taken an attractive target for further prospecting? It should be possible to devise data-display approaches that would enable a human operator to answer such questions directly and immediately. As a further indication of how the goals listed in Table 1 are inter-related consider the functional block diagram of an atomic spectrometric instrument displayed in Fig. 1. In this diagram the sample processing unit includes all of those devices which might be required to extract dissolve dilute or introduce samples into the spectrometric source (here an ICP).It might involve or require the use of electronic balances microwave-digestion systems auto- matic dilutors chromatographic columns possibly a laser- ablation device nebulizers spray chambers or others. The processed sample generated by this first block is then introduced into the source (the ICP). The ICP has the function of converting the introduced sample into the form of free atoms or ions and for the present discussion of exciting those atoms or ions in a reliable reproducible fashion. Only if these operations are carried out appropri- ately can the resulting emission spectrum be related unambiguously to elemental concentrations in the original sample. The atom detection module might be a direct-reader a slew-scan monochromator a linear or two-dimensional 1 Signal processing Fig.1 Functional block diagram of an atomic spectrometric instrument. See text for discussion detector-array spectrometer a Fourier-transform spectro- meter or another interferometric device. Whatever its character its purpose is to process the emitted radiation from the ICP so that the resulting display can again be related clearly to the elemental composition of the sample. Although perhaps the least developed of all the compo- nents in Fig. 1 the signal processing unit is in many ways the most important. At present signal processing is ordinar- ily limited to such mundane functions as spectral-line identification background subtraction the preparation of working curves and the calculation of concentrations.As indicated briefly above future signal-processing schemes will be far more powerful and complex. To improve this conventional instrument dramatically will require feedback from the signal-processing block to all the earlier components (see Fig. 2). For instance raw sample processing could be improved to insure that a sample is completely dissolved or maintained properly in suspension. The signal-processing module could also moni- tor the aerosol character or laser-ablation rate if the sample processor is equipped with suitable monitors or diagnostic devices. With additional diagnostic tools such as the special-purpose television camera cited earlier excitation or ionization conditions in the ICP could also be gauged so that they are either kept constant or adjusted for special- purpose applications or to accommodate specific sample types.The atom detection unit could also be optimized. For example spectral alignment could be continuously ad- justed the strength of internal-standard lines could be monitored and used for feedback purposes or the internal- standard lines themselves could be changed. Many other possibilities exist of course. For this type of future capability to be achieved will require modifications in current instrumentation and in part in the ways we think about such instruments. For example even though most modem ICP emission spectro- meters are equipped with computers the instruments in many instances are not optimized for computer monitoring or control. Many of the control inputs are fashioned after knobs or levers that were meant originally for human control. Also data outputs from such systems are often analogue and ordinarily limited to the response time of a human operator (typically 0.1 s) rather than tailored to the millisecond time scale on which a computer can respond.Moreover as stated before for this scheme to be effective will require the introduction and use of novel diagnostic approaches among which will be a full spectral display. In this spectral display the fidelity of both the horizontal and vertical axes will be critical. Fidelity on the horizontal (wavelength) axis of a spectrum permits more reliable spectral subtraction spectral stripping and the identifica- tion of lines and bands. Fidelity on the vertical axis of a spectrum means that precision will be enhanced and once more that spectral subtraction will not introduce unaccep- table levels of error.An adaptive-feedback computer-control approach such as in Fig. 2 will also require that a great deal more be known about all the blocks in Fig. 1 than is currently the situation. ~~ Control feedback paths Fig. 2 Future atomic spectrometric instruments will be improved by adaptive feedback from an on-board signal-processing unit to the various components that control the performance of the instrumentJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992. VOL. 7 71 For example unless samples are restricted to a particular type (i.e. unless matrices are matched) more will have to be understood about how matrix interferences originate if computer control is to be used to overcome them.In turn the means by which atoms and ions are generated and excited in the ICP will have to be better characterized and more must be learned about how the sampling process takes place. Indeed characterization of the ICP and how it acts on a sample remains one of the most fruitful areas for continuing research in ICP emission spectrometry. Finally new means will have to be developed for intelligent sample processing. Instrumentation will have to be devised that is amenable to control and in which aerosol character solvent load and sample-introduction rate can be monitored and adjusted independently of other instrumen- tal characteristics. Processing units will have to be created that can handle a variety of sample types forms and phases or that can be quickly modified to accept a different sort of sample.From this discussion it is clear that many of the 'final frontiers' are tied together. Let us therefore examine briefly two pairs of the 'frontiers'. We will devote the greatest attention to matrix interferences and the possible use of diagnostic techniques to overcome them. Less attention will be given to the automation of sample preparation and its use in self-adaptive experimentation since those topics are covered in a companion paper in this issue.3 Overcoming Matrix Interferences Inter-element (matrix) interferences in ICP emission spec- trometry arise from several sources and are still only partially characterized. For example it is now well known that the introduction of alkali metals or other easily ionized elements (EIE) causes a spatial shift and/or a change in emission intensity from both neutral atoms and atomic ions in the ICP. Depending on the observation location such changes can result in either positive or negative errors during an analysis.Other elements less easily ionized (non- EIEs) can also cause errors. Most recently it has been shown that intact droplets or the vapour clouds that result from them survive to points high in the ICP and no doubt affect sample atomization ionization and excitati~n.~*~ It would be surprising if droplet-related events were not linked in some way to matrix interferences. It is not yet clear how a concomitant species (EIE or non- EIE) influences the emission strength or pattern of an analyte element. For example an added element or the electrons it releases upon ionization might alter any of several things (i) the manner by which energy is coupled into the plasma; (ii) the rate of propagation of such energy throughout the plasma and therefore the plasma structure itself; and (iii) the coupling of the plasma energy with a sample aerosol to atomize it with sample atoms to ionize or excite them or with sample ions to excite them.Importantly no matter which of these events is affected by a concomitant element it would be tied in some way to one or more of three fundamental parameters in the plasma the electron number density the electron distribu- tion and the gas temperature. Consequently much can be learned about the ICP and how it interacts with a sample by characterizing these three parameters.Of course because the ICP is highly heterogeneous the three variables must be mapped spatially. Also although the ICP is certainly a steady-state source,6 additional information could be gleaned by its response to a transient pert~rbation.~-l~ Thus it would be desirable if the three parameters could be monitored temporally as well. Laser-light scattering has been found to be among the most powerful of possible methods for providing the necessary information. One such technique Thomson A Pulsed Electron - motion (e7- Thomson Fig. 3 Schematic diagram of the Thomson-scattering process. Radiation from a pulsed Nd:YAG laser (532 nm) is scattered from a rapidly moving electron (e) in a plasma causing a substantial Doppler shift in the scattered light.As a result the velocity (energy) distribution of a collection of electrons can be deduced from the spectrum of the scattering and the local number density of the electrons can be calculated by integrating the intensity of the scattered light over the entire spectrum scattering yields temporally and spatially resolved knowl- edge about electron energy distributions and number densities. Another method Rayleigh scattering can be performed at the same time and with the same apparatus as Thomson scattering and offers temporal and spatial values for gas-kinetic temperatures. Thomson scattering has been described in detail else- ~ h e r e l ~ - ~ ~ but its nature can be appreciated from the simple drawing of Fig.3. Simplified Thomson scattering is the quasi-elastic scattering of light from free electrons in a discharge. Because there is an unambiguous intersection between the incident laser beam and the observation direction the measurement of Thomson scattering provides point-by-point spatial information. Furthermore because a pulsed laser is utilized temporal resolution is also possible. Naturally electrons in a high-temperature plasma such as the ICP move at extremely high velocities (approaching 1 x lo6 m s-l); consequently the light they scatter is greatly Doppler shifted Indeed the Doppler shift is so significant that it can be measured with a conventional optical spectrometer. Because the electrons in an ICP travel at different speeds and in different directions with respect to both the incident laser and observation system a distribu- tion of Doppler shifts is registered.With the proper instrumental configuration and for an electron-velocity distribution that is Maxwellian (behaviour that has recently been confirmed20) the Doppler shifted scattered light assumes a Gaussian pattern. From the width ofthe Gaussian peak an electron temperature can be ascertained.22 Furthermore the wavelength-integrated intensity of the Doppler shifted scattered light indicates the total number of electrons that are present in the viewing volume. Therefore from the shape and integrated amplitude of the Thomson- scattering spectra both the electron energy distribution (temperature) and localized electron concentration can be found. At the same time the Doppler shifted Thomson spectrum is determined the Rayleigh-scattering intensity can be measured.Because Rayleigh scattering occurs from large particles principally argon atoms in the ICP it is not noticeably Doppler shifted and thus appears as an ex- tremely large peak centred at the laser wavelength on top of the Thomson-scattering spectrum. Because the contribu- tion of Thomson scattering at the laser wavelength is small compared with that of Rayleigh scattering the two can bc gauged separately.72 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 Of course the intensity of Rayleigh scattering is propor- tional to the number of scattering species (argon atoms) in the viewing volume. In turn the number of argon atoms in the viewing volume is inversely proportional to the gas- kinetic temperature in accordance with the gas law,19*22*23 that is as the gas temperature goes up the gas number density decreases proportionally and the Rayleigh scatter- ing drops.With a calibrated spectrometer the Rayleigh scattering can therefore be used to ascertain gas-kinetic temperature. Rayleigh- and Thomson-scattering maps have now been developed for the ICP under a wide variety of operating conditions solvent loads and with different combinations of elements added. Unfortunately the resulting maps have led to a data ‘flood’. At first attempts were made directly to compare values for electron temperature gas temperature and argon-ionization temperatures which would be derived from localized number density data.I3-I9 Similarly values from different plasmas or different operating conditions were correlated.The impossibility of managing the result- ing torrent of data led to the use of contour maps.22 However meaningful conclusions were difficult to derive because of the complexity of the resulting patterns and the unsurprising difficulty of detecting subtle shifts in them. Very recently more sophisticated image-development software has been used (Spyglass Dicer Transform View Format Spyglass Champaign IL USA) and it has been found that the resulting displays are far more easy to manipulate appreciate and interpret. For example Fig. 4 shows a side-by-side comparison of spatial maps taken at a variety of applied radiofrequency (r.f.) power levels of an ICP in the presence and absence of introduced solvent (water) vapour.The images in Fig. 4 show a wealth of information; five dimensions are displayed in ways that can be manipulated rapidly and easily. In this instance the five dimensions are radial position in the discharge height above the load coil (ALC) r.f. power level electron concentration (by the colour) and the influence of added solvent. Importantly each of the false-colour images in Fig. 4 resides in the computer as a four-dimensional array. As a result variables can be displayed in different ways ‘cuts’ can be shown through volume elements as desired and outer levels in a displayed volume can be rendered transparent so the underlying structure can be examined. False-colour images similar to those in Fig.4 for electron number density are displayed in Figs. 5 and 6 for electron temperature and gas temperature respectively. A striking example of the degree to which the ICP departs from local thermal equilibrium (LTE) is apparent by comparing Figs. 5 and 6. No matter whether a solvent is present or not the gas temperature in an ICP is in most locations and at every r.f. power level considerably lower than the electron temperature. Immediately obvious also in the images is the startling and somewhat surprising effect that water vapour has on the discharge. For example in Fig. 4 the introduction of water vapour appears to reduce electron number density in most plasma zones when the r.f. power level is low. In contrast the presence of water vapour has the effect of elevating electron concentrations when the r.f.power is high. Of course many other observations concerning the images in Figs. 4-6 could be made. However they represent only a small fraction of the data which have now been collected and which are better left for interpretation in a later paper. Therefore the mapping of other plasma features will now be considered. Although electron temperatures electron number densi- ties and gas temperatures are undeniably important in unravelling matrix interferences in the ICP of significance also are excited-state and ground-state maps of analyte and concomitant species and those intrinsic to the ICP itself. To obtain such maps a device termed a ‘monochromatic imaging spectrometer’ (MIS)24 and computed tomography have been ~ t i l i z e d .~ ~ ~ ~ The MIS employs a two-dimensional charge-coupled device (CCD) detector and is capable of recording electronically a full two-dimensional spatial image of the ICP at a selected wavelength. If the ICP is assumed to be cylindrically symmetrical each horizontal row in the CCD image could be Abel inverted to yield a radially resolved map. However it was found that the ICP is more often than not asymmetrical and computed tomo- graphy was used instead to unravel the true three- dimensional structure of the discharge. Considered simply computed tomography requires view- ing the plasma from a number of angles much as a human observer would do in order to discern the shape of a solid object. To perform this task the apparatus has the ICP placed on a rotational stage so that the MIS can view the plasma from 100 angles over a 180” range.The resulting array of images can then be reconfigured using computed tomography into the radially resolved equivalent. Further details about the tomographic set-up are provided else- here.^^-^^ In the tomographic investigations as in the application of Thornson and Rayleigh scattering it was found that the use of image-display software was indispensable. For example Fig. 7 contains a false-colour slice of an ICP and reveals immediately its lack of cylindrical symmetry. This particu- lar image is for the argon emission line at 430 nm in a dry argon plasma and not surprisingly shows a strong off-axis rnaximum low in the discharge in the toroidal region. Multiple slices through the ICP disclose additional details.Fig. 8 shows both the horizontal slice of Fig. 7 and a superimposed vertical cross-section. This mode of display reveals that argon emission is weakest in the centre of the plasma throughout its entire vertical length. Multiple horizontal slices (Fig. 9) or vertical cross-sections (Fig. 10) are also possible. Regrettably a detailed discussion of these images is beyond the scope of the present coverage as with the Thornson- and Rayleigh-scattering data. Importantly however the images of Figs. 7-10 demon- strate immediately the diagnostic value of visual informa- tion. In most situations where only the diagnostic power of an image is desired it would not be necessary to perform a full tomographic reconstruction or even an Abel inversion of the collected laterally resolved data.Instead the lateral images could be used directly to indicate the status of the discharge. Useful lateral presentations would include back- ground emission maps the intensity of plasma gas (argon) emission excitation temperatures throughout the plasma t?fc. Such displays would be of immense diagnostic value iind a great benefit to computer-control schemes. Intelligent Sample Processing and Adaptive Control In a fully adaptive instrument incoming data from a sample would be combined with diagnostic input about instrumental components and used to generate feedback- control signals which could alter conditions in the plasma the sample-processing unit or the spectrometric system. In .that way each sample would be analysed under conditions that were optimum for it under which interferences would be minimized and figures of merit maximized.However the liability of such an arrangement is that each sample would be analysed under different conditions so that an earlier calibration would probably not be valid. For this scheme to be attractive it will therefore be necessary either to perform emission measurements on an absolute (stan- dardless) basis or to provide a means for rapid recalibration of the instrument. Importantly rapid recalibration using a single standard isFig. 4 False colour maps ofclectron density in an ICP opcrated at Fig. 5 False colour maps of electron tcmpcraturc in an 1('P different r.C powcr lcvcls and in the presence and absence of opcrated at dil'ferent r.f.powcr levels arid in the prrscnc-e mid introduced water vapour. Data were obtaincd b> means of absence of introduced water vapour. Data wrc obtaincd by mcans Thomson scattering. Thc K'P was opcrated with an outer-gas of Thornson scattering and under the samc operating wilditions its flow of 14.0 I min-I. an intcrmediatc-gas flow of 0.8 I min-' and used for Fig. 4 an inner-gas flow of 0.5 I min-I. Solvcnt vapour was added using a bubble (gas-dispersion) nebulizer at a total solvent load of 17 mg min-' Fig. 6 False colour maps of kinetic temperature of the gas in an ICP operated at different r.f. power levels and in the presence and absence of introduced water vapour. Data were obtained by means of Rayleigh scattering at the same time and under the samc operating conditions as used for Fig.4 [to face page 721Fig. 7 False colour map of Ar I emission at 430 nm obtained by means of computed tomography. The 40.68 MHz ICP contained no solvent or solvent vapour and was operated at an r.f. power of 1.25 kW an outer-gas flow of 14.0 I min-I an intermediate-gas flow of 1.0 1 min-' and an inner-gas flow of 0.8 1 min-' Fig. 8 As for Fig. 7 but with an added vertical segment of the Ar 430 nm emission. The lack of cylindrical symmetry is apparent Fig. 9 Multiple horizontal slices taken along the vertical axis of an ICP indicating not only the conical shape of the discharge but also how the plasma becomes radially more homogeneous in its upper zones taken under the same conditions as in Fig. 7 Fig. 10 Multiple vertical cross-sectional cuts of the Ar 430 nm emission providing a useful and intuitively satisfying view of the spatial features of the ICP taken under the same conditions as in Fig.7 (to facepage 731JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 73 now possible using a modem high-performance liquid chromatography (HPLC) pumps3 Such pumps are readily available and are intended to produce pulse-free concentra- tion gradients useful for gradient-elution HPLC. Ordinar- ily the pumps are equipped with computer-controlled valves and have at least two pumping systems. As a result it is a simple matter to change standard solutions for use with different elements and to generate concentration gradients at any desired sample flow rate. Such pumps have many potential uses some of which are documented in a companion paper in this issue.3 For example a range of standard concentrations from a single stock solution can be generated by the pump in rapid sequence to produce a calibration curve under newly optimized instrumental conditions.Also it should be possible to use a pseudo-null-point approach2* for calibra- tion in situations where the dynamic range of an instrument is limited. In a null-point scheme the signal from a sample solution would be measured and the HPLC pump pro- grammed to vary the concentration of the introduced standard solution until the resulting signal matched that from the original sample. The sample concentration would therefore be directly determined and in a way that did not depend on instrumental linearity.Calibration based on standard additions is also simplified with the HPLC gradient-calibration m e t h ~ d . ~ Here the sample solution itself serves as a carrier and different flows of a standard solution are introduced into it; the combined flow is programmed to be held constant. In this way a standard additions calibration curve could be constructed quickly and a number of potential matrix interferences overcome. Because the gradient-calibration scheme enables any desired sample flow rate over a wide range to be chosen it permits a nebulizer to be starved or over-fed a capability that in turn permits the resulting aerosol character to be modified in an on-line fashion. Finally because the HPLC pump is capable of delivering sample or standard solutions at extremely high pressure it facilitates the use of unusual nebulizer types such as the jet- impact nebulizer (JIN) developed in this laboratory a number of years In the JIN the sample or standard solution is forced from a small capillary to produce a fine jet of liquid.Ordinarily such a jet would break up into random-sized droplets because of natural stochastic distur- bances within the jet. However the jet can be forced to break up into far smaller droplets simply by placing a solid impact surface just before the point where the jet begins to disintegrate by itself. The result is an extremely fine aerosol which is generated in a fashion independent of any auxiliary gas flows enabling both aerosol characteristics and sample-carrier flows to be optimized independently.Clearly the JIN would benefit greatly from the gradient- calibration scheme and is now being utilized in our laboratories. Conclusion It should be clear from the foregoing discussion that many exciting frontiers remain in ICP emission spectrometry. Of course even more challenging and unexplored frontiers exist in the combination of ICP and mass spectrometry as well. Perhaps the most important and intriguing develop- ments will arise from an enhanced understanding of the ICP itself from improvements in instrumentation that incorporate it and in our ability to control through computer-based adaptive feedback the characteristics of components in such instrumentation. Despite its maturity the ICP remains an important object of study development and characterization.This work was supported by the National Science Founda- tion through grant CHE 90-20631 and by the Leco Corporation. 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 References Fassel V. A Electrical ‘Flame’ Spectroscopy Colloquium Spectroscopicum Internationale XVI Plenary Lectures and Reports Heidelberg Germany October 4-9 197 1 Adam Hilger London 1972 p. 63. Hieftje G. M. Recent Developments and Final Frontiers in ICP Spectrometry Colloquium Spectroscopicum Internationale X X VII Bergen Norway June 9- 14 paper C-PL2. Starn T. and Hieftje G. M. J. Anal. At. Spectrom. 1992 7 335. Olesik J. W. and Fister J. C. 111 Spectrochim. Acta Part B 1991,46 851. Fister J. C. 111 and Olesik J. W. Spectrochim Acta Part B 199 1,46,869.Rayson G. D. and Hieftje G. M. Spectrochim. Acta Part B 1986 41 683. Ensman R. E. Carr J. W. and Hieftje G. M. Appl. Spectrosc. 1983 37 57 1. Cam J. W. and Hieftje G. M. Determination of Energy Transport Rates in the Inductively Coupled Plasma Collo- quium Spectroscopicum Internationale XXIII Amsterdam The Netherlands June 26-July 1 1983 paper 143. Farnsworth P. B. Appl. Spectrosc. 1985 39 1078. Parisi A. F. and Hieftje G. M. Appl. Spectrosc. 1986 40 181. Farnsworth P. B. Rodham D. A. and Ririe D. W. Spectrochim. Acta Part B 1987 42 393. Fey F. H. A. G. Stoffels W. W. van der Mullen J. A. M. van der Sijde B. and Schram D. C. Spectrochim. Acta. Part B 1991,46 885. Marshall K. A. Huang M. and Hieftje G. M. Spectrochim. Acta Part B 1985 40 121 1. Huang M. and Hieftje G. M. Spectrochim. Acta Part B 1985,40 1387. Huang M. Marshall K. A. and Hieftje G. M. Anal. Chem. 1986 58 207. Marshall K. A. and Hieftje G. M. Specrochim. Acta Part B 1988 43 841. Marshall K. A. and Hieftje G. M. Spectrochim. Acta Part B 1988 43 85 1. Huang M. and Hieftje G. M. Spectrochim. Acta Part B 1989,44 291. Huang M. and Hieftje G. M. Spectrochim. Acta Part B 1989,44 773. Huang M. Yang P. Hanselman D. S. Monnig C. A. and Hieftje G. M. Spectrochim. Acta Part B 1990 45 51 1. Huang M. Hanselman D. S. Jin,’ Q. and Hieftje G. M. Spectrochim. Acta Part B 1990 45 1339. Huang M. Hanselman D. S. Yang P. and Hieftje G. M. Spectrochim. Acta Part B in the press. Marshall K. A. and Hieftje G. M. J. Anal. At. Spectrom. 1987 2 567. Olesik J. W. and Hieftje G. M. Anal. Chem. 1985,57,2049. Monnig C. A. Marshall K. A. Rayson G. D. and Hieftje G. M. Spectrochim. Acta Part B 1988 43 1217. Monnig C. A. Gebhart B. D. Marshall K. A. and Hieftje G. M. Spectrochim. Acta Part B 1990 45 261. Hieftje G. M. Spectrochim. Acta Part B 1992 47 3. Bastiaans G. J. and Hieftje G. M. Anal. Chem. 1973 45 1994. Doherty M. P. and Hieftje G. M. Appl. Spectrosc. 1984 38 405. Paper 1/03645I Received July 17 1991 Accepted September 2 1991