Waters Symposium: Atomic Emission Spectroscopy
Atomic Emission Spectroscopy—It Lasts and Lasts and Lasts Gary M. Hieftje Department of Chemistry, Indiana University, Bloomington, IN 47405;
[email protected] Elemental analysis is one of the most common procedures used to characterize a sample. It is useful in most areas of human endeavor and is commonly applied to samples of importance in toxicology, geology, forensic science, prospecting, health, nanotechnology, materials science, industrial process control, semiconductor processing, automobile manufacturing, and others. Because of this importance and range of application, a broad suite of methods for elemental analysis has been devised. These methods range from long-established procedures such as those based on precipitation and titration to more modern ones that rely as much on physical principles as on chemistry. Background Prominent among the physical-based techniques are the methods known collectively as atomic spectroscopy. In this group of methods, the sample or specimen of interest is decomposed as quantitatively and reproducibly as possible into its constituent atoms. Those atoms, in the gas phase, then undergo valence-electronic transitions free from the effects of chemical bonding. As a result, the transitions are highly selective, characterized by narrow spectral lines, and are indicative of the element or elements being determined. Moreover, the strength of each transition is proportional to the number of atoms that are present and, indirectly, to the elemental concentration in the original sample. The field of atomic spectrometry includes the procedures of atomic emission spectroscopy (AES), atomic absorption spectroscopy (AAS), atomic fluorescence spectroscopy (AFS), and, when atomic ions are generated and can be measured, atomic mass spectrometry (AMS). To be sure, additional atomic methods have been devised but these are in less common use. In this latter set are found techniques such as coherent forward scattering spectrometry, opto-galvanic effect spectroscopy, and thermal-lens spectrometry. Atomic emission spectroscopy is perhaps unique in this listing. It was among the first techniques ever used for qualitative and quantitative elemental analysis, yet it remains one of the most widely used today. Of course, the instrumentation, sample-handling approaches, and methodology of AES have evolved over time. Still, it is one of the most resilient of analytical methods. The reasons for this resiliency can perhaps be understood by revisiting the seven stages of an analytical method set down by Velmer Fassel (1). Fassel’s listing, adapted from ref 1, is reproduced below. Seven Stages of an Analytical Method 1. 2. 3. 4. 5. 6. 7.
Conception of idea Design and construction of first operating apparatus Successful demonstration of idea; first publication Improvement of instrumentation; figures of merit Maturity; general acceptance; automation Improved understanding of fundamental principles Old age and senescence
According to Fassel, the natural evolution of any analytical method progresses from its conception through instrument design and testing to maturity, general acceptance, and automation. After this period of considerable success, the method is characterized more fully in a fundamental sense, after which it reaches its period of old age and senescence. I disagree with portions of this analysis. To me, it seems inappropriate that a method would be characterized fundamentally only after it reaches success and that the fundamental understanding would yield little practical benefit. Accordingly, I would modify Fassel’s seven stages as follows: Seven Stages of an Analytical Method (Modified) 1. Conception of idea 2. Successful demonstration and publication of idea 3. Improvement of instrumentation; figures of merit 4. Maturity; general acceptance; automation 5. Improved understanding of fundamental principles; introduction of new instrumentation 6. Iteration of steps 3–5 7. Old age and senescence
Here the first two stages are much the same as those in Fassel’s original listing. After a new idea is conceived, it obviously must be demonstrated in order to be successful. However, once the successful demonstration has been published, it seems likely that the instrumentation will be improved and figures of merit (detection limits, precision, accuracy, interferences, etc.) compiled to underscore its capability. Once this level of performance has been accepted widely, additional funds will no doubt be invested to make the method easier to use; automation will follow, and the technique will reach what might be considered to be a mature stage. At the same time this technique is maturing, it seems inevitable that scientists will begin characterizing it on a more fundamental level, if that fundamental underpinning did not exist previously. Ideally, the deeper level of fundamental understanding that results from these studies will lead to an improvement of instrumentation, better figures of merit, still more widespread acceptance, and an even fuller understanding. These three steps—improvement of instrumentation and figures of merit, broader acceptance, and deeper understanding— will then continue to cycle until the point at which either no further improvement in the method is possible or competitive techniques prove to be superior. Only when these limits have been reached will the method enter its period of old age and senescence. If the modified sequence can be accepted, it should be apparent why AES has long remained a widely used, extremely powerful technique. Instrumentation for AES has continued to improve, the understanding of emission sources and emission instrumentation has progressively evolved, and figures of merit have gotten steadily better. Indeed, as will be argued later in this discussion, AES appears likely to remain competitive for a long time to come.
JChemEd.chem.wisc.edu • Vol. 77 No. 5 May 2000 • Journal of Chemical Education
577
Waters Symposium: Atomic Emission Spectroscopy
In this overview, the evolution of AES will be traced from its earliest roots to some of the most recent exciting developments, with a view toward identifying some of the developments that made AES attractive in the face of increasingly stiff competition. Greatest emphasis will be placed on instrumental developments, although a number of fundamental studies of particular significance will be mentioned. Greatest weight will be placed on the development of atomization/ excitation sources for AES, because they have changed most greatly over the years. However, some developments in systems for the detection of the emitted radiation will also be included. Because this overview is meant to provide perspective and not to be exhaustive in its coverage, the original literature will not usually be cited. For more detail, especially about historical developments, the interested reader should consult any of the several excellent texts that are available (2–8). Atomic Emission Sources A listing of noteworthy atomic emission sources is compiled below. Noteworthy Atomic Emission Sources Chemical flame dc, ac arc High-voltage spark ICP Glow discharge (dc, RF) dc plasma MIP, MPT FAPES Other “tandem sources”
Of course, in an overview of reasonable length, it is impossible to discuss the impact of each of these sources in detail. Rather, a selective approach will be used, principal emphasis being placed on the AES sources that have had the greatest practical impact both historically and at the present time. Accordingly, the sources that will be emphasized include the chemical flame, the high-voltage spark, and the microwaveinduced plasma (MIP). Although the inductively coupled plasma (ICP) is clearly a source of both historical and current importance, it is described in considerable detail in one of the accompanying papers in this symposium and will therefore not be stressed here.
The Chemical Flame Flame emission spectrometry (FES) is one of the most familiar atomic methods and is probably the oldest for determining chemical elements by means of their characteristic spectra. The field can be traced (see Fig. 1) at least back to Thomas Melville (5), who in 1752 observed that sea salt and other chemical substances added to an alcohol flame caused a discernible difference in the color of the flame. Later, in 1826, the Scotsman W. H. Fox Talbot employed a simple spectroscope to observe flames that had been doped with a range of different salts. In an 1834 paper he defined more exactly the potential power of flame emission spectrometry. Yet it was Robert Wilhelm Bunsen and Gustav Kirchhoff who in 1859 and 1860 set the method on a sounder theoretical and experimental foundation. Bunsen’s flame, invented only a few years earlier, provided a wonderfully transparent medium 578
against which elemental emission lines could be observed more clearly. At the same time, Kirchhoff recognized that the absorption lines described earlier by Fraunhofer coincided exactly with the emission lines from salts introduced into the flame. Together, Bunsen and Kirchhoff set the stage for modern FES by recognizing that spectral lines emitted by metals occur at wavelengths that are independent of the elements (anions) to which the metals are bound. Yet the technique did not enjoy immediate widespread favor because of the difficulty of introducing samples or sample solutions into the chemical flame. Then, in 1879, L. P. Gouy showed that it was possible to employ a pneumatic nebulizer for introducing sample solutions into flames. A truly modern setup was finally assembled in 1929 by Lundegårdh, who coupled a premixed air–acetylene flame, a pneumatic nebulizer for sample introduction, and a spray chamber for conditioning the nebulized solution. Photographic detection was employed. An examination of Lundegårdh’s experimental system (5) reveals a similarity to even the most recent systems. In a completely automated fashion, his apparatus changed samples, controlled the photographic exposure, developed the photographic film, and produced a microphotometer reading of the recorded spectral lines. With this innovation, FES became more widely used. Still, most instruments had to be specially and locally constructed, limiting the popularity of the method. In 1949, all this changed through the efforts of Paul T. Gilbert of the Beckman Company, who fashioned a flame emission attachment for the popular Beckman DU spectrophotometer. Gilbert’s emission source consisted of a totalconsumption turbulent burner that produced an extremely unstable and noisy (in terms of both signal fluctuation and acoustic output) flame. In the Gilbert total-consumption burner, the fuel (either hydrogen or acetylene) and oxidant (usually oxygen) issued from concentric orifices in a highly turbulent stream. This turbulence mixed the gases so they could be readily and safely combusted without danger of explosion or flashback. Into this turbulent mixture an aerosol of the sample solution of interest was then sprayed directly. The whole arrangement sounded like a small rocket motor in operation and yielded such a coarse spray that many droplets survived intact during their passage through the flame. Milestones in Flame Emission Spectrometry 1850
1950
1960
Present
■ 1860:
Bunsen & Kirchhoff; FES with prism spectroscope
■ 1876:
L. G. Gouy; use of pneumatic nebulizer
■ 1928:
H. Lundegårdh; spray chamber, premixed air–C2H2 flame
■ 1937:
Schuhknecht; Na and K detection by filter photometer
■ 1949:
P. T. Gilbert; total consumption nebulizer-burner
■ 1949:
P. T. Gilbert; introduction of Beckman flame photometer (PMT)
■ 1955:
Attack of the atomic absorbers ■ 1960s: Elucidation of matrix interferences in FAAS and FAES ■ 1966: M. Amos & J. B. Willis; introduction of N2O–C2H2 flame ■ 1970:
Attack of the ICP; relegation to clinical workhorse position
Figure 1. Milestones in flame emission spectrometry. The grayscale bar at the top indicates qualitatively how the field developed over time. A darker shade indicates a higher level of activity in the area.
Journal of Chemical Education • Vol. 77 No. 5 May 2000 • JChemEd.chem.wisc.edu
Waters Symposium: Atomic Emission Spectroscopy
Nevertheless, the Gilbert system was used for quite a number of years, especially for qualitative and semiquantitative flamebased analyses. In many ways, the popularity of Gilbert’s invention hindered the long-term development of FES. The coarse aerosol it produced made it impossible for large particles of sample material to be completely atomized in the flame. As a result, the degree of atomization, and therefore the signal from a given element, depended strongly on the other constituents in a sample. The method therefore became cursed with the reputation for exhibiting serious and difficult-to-avoid matrix interferences. These difficulties were exacerbated when Alan Walsh introduced atomic absorption as an analytical method in 1955. Walsh, and most who followed him, also employed a chemical flame, but as a light-absorbing medium. However, the flame he used was much more like the one originally used by Lundegårdh. The gases were premixed and an aerosol spray chamber was used to eliminate many of the largest droplets in the pneumatically generated spray. As a result, much finer droplets were introduced into a flame that was highly stable. The result was a steadier signal and fewer matrix interferences. Regrettably, many workers at that time attributed the lower level of interferences to the use of atomic absorption rather than to the use of a different setup for producing the flame and introducing the aerosol. The use of FES began to decline accordingly. It was about this same time when additional fundamental work on FES began to be published and widely recognized. To be sure, the basic underpinnings of spectral-line generation had been laid years before. However, investigations aimed specifically at elucidating the origin of matrix interferences in FES were undertaken by C. Th. J. Alkemade in the Netherlands and by others (3, 9). For example, Alkemade showed through a series of brilliant experiments (9) that a mutual interference exists among the alkali metals because they are partially ionized at flame temperatures. If, say, sodium is being determined, a fraction of the sodium will ionize in the flame. Because it is the free, neutral atoms that are ordinarily observed in emission, any factor that affects the fraction of atoms and ions will similarly alter the measured signal. If, for example, potassium is added to the sodium-containing flame, a fraction of the potassium also ionizes. The free electrons liberated by the potassium ionization then serve to shift the equilibrium between sodium atoms and ions back toward the formation of free atoms. The sodium signal rises accordingly. Alkemade then showed that the interference could be overcome quite simply, just by adding an excess of potassium to the sample solution; not only was the influence of other alkali metals (including potassium) thereby avoided, but also the sodium signal (and that of all other alkali metals) was increased. Other experiments during that time and later were aimed at determining the rates and mechanisms of atom formation in a chemical flame. Unfortunately, these experiments were handicapped because conventional nebulizers employed in flame spectrometry produced a broad range of droplet sizes, making it extremely difficult to follow the events that occur to a single droplet. These complications were avoided when it was shown possible to introduce a single droplet or a stream of droplets of known size along a reproducible trajectory into a chemical flame (10–19). Because each aerosol droplet follows
exactly the same path in the flame as all others, and because of the reproducibility of the original droplet size, the progress of each droplet or of multiple droplets in the stream could be followed simply by examining progressively later times in the flame. By means of both physical and spectroscopic measurements, it then became possible to measure the rate of droplet desolvation and, later, of particle vaporization. Determining the dependence of those rates on the properties of the droplet, the solvent, the solute, and the flame then made it possible to deduce the mechanism of droplet desolvation and particle vaporization. While these fundamental developments were taking place, new instrumentation was being invented that would improve both flame emission and atomic absorption spectrometry. Perhaps the most important of these innovations was the introduction by Max Amos and John Willis in 1966 of the nitrous oxide–acetylene flame. Unlike the air–acetylene mixture used earlier, the nitrous oxide–supported flame produced a temperature high enough to volatilize even solutes that contained refractory elements such as zirconium, hafnium, or aluminum. Moreover, although its spectral background was a bit more troublesome than had been experienced with the air–acetylene flame, the N2O–acetylene mixture could be supported safely on a burner only slightly different from the ones then in common use. Because of these newly recognized capabilities, most manufacturers of atomic absorption instrumentation incorporated the ability to perform flame-emission measurements into their systems. FES became a workhorse in clinical analyses, especially for the determination of alkali and alkaline-earth metals. It is still widely used for this purpose, although its broader application has been limited since the 1970s by the widespread acceptance of the inductively coupled plasma as an even more powerful emission source. The history of FES, it can be seen, nicely follows the modified sequence of evolutionary stages of an analytical method outlined earlier. After its original conception and experimental demonstration, the method was improved and its figures of merit were delineated. It was automated and became widely accepted, and its fundamental principles were better understood. This improved understanding led to further developments in instrumentation and the cycle repeated itself. Even the introduction of atomic absorption as a popular method was not enough to quell the further development of FES. Only when a decidedly more capable source (the ICP) came along was FES relegated to a secondary position.
The High-Voltage Spark Figure 2 chronicles the development of high-voltage spark emission spectrometry. As in flame-emission spectrometry, electrical sparks were known long before they were applied to elemental analysis. However, as early as 1873–1874, J. N. Locklear realized that a high-voltage spark exhibits a complex spatial structure. More importantly, Locklear observed that the spatial structure of atomic emission within the spark was a function of the concentration of the emitting element. Along with his colleague Roberts, he also determined that improved quantitation was possible by comparing the spatial structure of the spectral lines of a target element with those of another element in the sample. This approach is, of course, quite similar to the modern method of internal standardization.
JChemEd.chem.wisc.edu • Vol. 77 No. 5 May 2000 • Journal of Chemical Education
579
Waters Symposium: Atomic Emission Spectroscopy
In 1884, another advance in quantitation was established by Hartley at the University of Dublin. He employed a condensed spark to directly excite the spectra of elements in solution. The target of these efforts was to improve the accuracy and precision of quantitative analysis. Traditionally, high-voltage sparks were notoriously unstable. Accordingly, the introduction of true internal standardization by W. Gerlach and E. Schweitzer in 1925 made quantitative determinations far more tractable. Shortly thereafter, quantitation was greatly simplified by Scheibe and Neuhausser through their introduction of the rotating log sector. Because the blackening of a photographic emulsion is logarithmically related to emission intensity, the rotating sector made the length of an image on a photographic film or plate proportional to intensity and therefore to concentration. George Harrison of MIT had an enormous impact on many aspects of spectrochemical analysis. His publication in 1939 of a comprehensive set of wavelength tables greatly aided the identification of spectral lines (20). Shortly thereafter, spark emission spectrometry became widely adopted in the metals industry because of its high speed, modest level of matrix interferences (for samples of relatively constant matrix), and large number of practitioners. Still, users of the high-voltage spark yearned for something considerably more controllable and precise. Although fundamental studies had begun long before (21), in the 1950s and 1960s a deeper understanding was sought of events that occur in the high-voltage spark. Early work by Bardøcz led J. P. Walters and H. V. Malmstadt at the University of Illinois to devise a rotating-mirror spectrograph that enabled the events occurring in the spark to be resolved temporally and spatially. The evolution of processes that occur in the discharge was found to be extremely complex but controllable, depending upon the circuitry used to drive the spark. The original thyratron-triggered high-voltage spark of Walters and Malmstadt gradually evolved in Walters’s lab at the University of Wisconsin into a controllable-waveform high-voltage spark source (22). The device enabled the timing of the spark to be precisely controlled and its waveform to be carefully tailored and extraordinarily reproducible. Either a bidirectional or unipolar pulsating discharge could be achieved, the latter being more desirable. With this greater understanding and improved instrumentation came an unprecedented degree of control over the sampling and excitation processes that occurred in the spark. Later excitation sources built around the concepts that sprang from the Walters group remain the most stable and controllable of all high-voltage spark systems. The same pattern emerges in the development of the high-voltage spark as we saw in flame-emission spectrometry. This pattern, once again apparent in the modified evolutionary sequence, reveals that a technique progresses from the initial phases of study to widespread acceptance. A deeper understanding of the fundamental principles and the development of improved instrumentation then naturally arise from this high level of interest. The improved instrumentation leads to better figures of merit, even broader acceptance, and a cycle that continues to produce improved systems. Even today, the high-voltage spark source is the most widely used tool in the metals industry for rapid, nearly real-time analyses. Only in 580
Milestones in Spark Emission Spectrometry 1940
1850
1950
Present
■ 1873–4:
J. N. Lockyer; Spatial resolution of spark emission
■ 1884:
W. N. Hartley; Condensed-spark excitation of solutions
■ 1925:
W. Gerlach & E. Schweitzer; line pairs for internal standard
■ 1928:
G. Scheibe & A. Neuhausser; rotating log sector for quant.
■ 1939:
G. R. Harrison; publication of MIT Wavelength Tables
■ 1940–55: Rapid expansion and widespread use in metals industry ■ 1964–80: J. P. Walters, et al.; development of stabilized spark ■ 1980–present: Automation
Figure 2. Milestones in spark emission spectrometry. The grayscale bar at the top indicates qualitatively how the field developed over time. A darker shade indicates a higher level of activity in the area.
Milestones in ICP Emission Spectrometry 1942 ■ 1942: ■ 1961: ■ 1964: ■ 1965: ■ 1969: ■ 1974: ■ 1975: ■ 1975–85: ■ 1980: ■ 1980s: ■ 1990s:
1974 1980 Present G. I. Babat; atmospheric-pressure induction plasma T. B. Reed; flowing atmospheric-pressure ICP S. Greenfield; toroidal analytical ICP R. H. Wendt & V. A. Fassel; reduced-power analytical ICP G. W. Dickinson & V. A. Fassel: ICP detection limits First commercial ICP–AES instruments J. Robin & C. Trassy; first use of end-on observation Explosive growth in utilization, instrument development First publication of ICPMS; challenge Characterization of spectral, matrix interferences Settling in; routine usage
Figure 3. Milestones in ICP emission spectrometry. The grayscale bar at the top indicates qualitatively how the field developed over time. A darker shade indicates a higher level of activity in the area.
the area of solution samples has another source, the inductively coupled plasma, proven superior.
The Inductively Coupled Plasma Because the evolution and eventual success of the ICP is documented elsewhere in this symposium, coverage will be restricted here. However, it is useful to display a number of milestones in ICP emission spectrometry just as were given for flame-emission and spark-emission spectrometry. These milestones are compiled in Figure 3. Like the flame and high-voltage spark, the ICP had its roots in fields well removed from analytical chemistry or atomic spectrometry. Also like the other systems, the ICP underwent several cycles of improved fundamental understanding and better instrumentation, although in the case of the ICP, most of the innovations came in the sample-introduction end of the instrument. Even today, work continues in understanding the nature and origin of interelement interferences that afflict ICP emission spectrometry. It is to be hoped that, with a clearer elucidation of those interference effects, they can be brought under control or eliminated entirely. The Microwave-Induced Plasma In contrast to the other sources already discussed, the microwave-induced plasma (MIP) is a relatively recent in-
Journal of Chemical Education • Vol. 77 No. 5 May 2000 • JChemEd.chem.wisc.edu
Waters Symposium: Atomic Emission Spectroscopy
vention. In fact, technology for the routine production of microwave power did not become practicable until the surge of interest in radar instrumentation during World War II. Since that time, specially designed electronic devices such as magnetrons, Gunn diodes, and Klystron tubes have made the generation of microwaves relatively straightforward. Of these devices, the most common for use in sustaining an MIP is the magnetron tube. The most common frequency at which an MIP is sustained is 2.45 GHz. This is the same frequency employed in microwave ovens, so components for an MIP system have become relatively inexpensive. In addition, an MIP is usually operated at power levels considerably below those used in ICP spectroscopy, commonly between 50 and 250 W. Again, a magnetron tube useful for a microwave oven performs well. However, sustaining an MIP is not as easy as operating a kitchen oven. Microwaves lie at frequencies high enough that they can be viewed alternatively as an electrical signal or as a low-frequency light wave. Accordingly, they can be carried alternatively down a coaxial cable or along a resonant structure termed a waveguide. In addition, microwaves can be coupled into a gaseous discharge by a number of approaches. Early designs employed a simple antenna system that focused the microwave energy into a closed discharge cell, an arrangement that proved acceptable but relatively inefficient. Later, resonant cavities based on either quarter-wave (for example the Evenson system) or 3/4-wave (such as the Broida device) designs were introduced. However, these arrangements worked best when the gaseous discharge was at reduced pressure, typically only a few torr. It also was extremely difficult to sustain a helium plasma in such a resonant cavity. An important innovation in microwave cavity design occurred with the introduction of the TM010 cavity by C. I. M. Beenakker in 1976 (23). Unlike the Evenson and Broida designs, which operate on a standing electric field, the Beenakker cavity employed a standing magnetic wave. As a result, the dimensions of the cavity in the direction perpendicular to the magnetic field (the electric-field direction) could be adjusted at will. Accordingly, if the electric-field dimension were made extremely small, the electric field was concentrated in an extremely small region exactly in the center of the cavity. Indeed, the field became great enough in that small zone to sustain atmospheric-pressure plasmas in virtually any gas, including helium. With modification, this basic Beenakker design has formed the basis of the most successful commercial MIP system, offered by Hewlett-Packard for multielement detection of a gas chromatographic effluent. Another innovation in MIP technology occurred with the introduction of the surfatron by Michel Moisan et al. (24). Unlike the Evenson, Broida, and Beenakker devices, the surfatron did not sustain a plasma within a resonant cavity. Rather, it consisted of a wave-shaping structure that launched the microwave field out of the end of a resonant unit. The field traveled along the inner edge of a dielectric material, in most cases a quartz tube that was passed through the resonant and wave-shaping structures. Consequently, the surfatron discharge formed at the interface between the quartz dielectric and the plasma support gas. The discharge was therefore naturally annular in shape, on the inner surface of the quartz discharge tube. Samples could be introduced into the center of the tube without greatly affecting the efficiency with which
microwave energy was coupled into the discharge. The discharge thus became far less sensitive to the presence of sample material and much more tolerant to molecular gases introduced into the discharge. It was possible to sustain a surfatron discharge in a variety of gases, including helium at atmospheric pressure. Although no commercial atomic spectrometric instrumentation based on the surfatron has yet been introduced, the surfatron has become widely studied and used effectively, especially for gas chromatographic detection. The most recent addition to the arsenal of support structures for the MIP is the so-called microwave plasma torch (MPT), introduced by Jin et al. (25). Like the surfatron, the MPT consists of a resonant structure and a wave-launching assembly. However, in the case of the MPT, waves are launched directly into the atmospheric pressure region above the torch. As a consequence, the gaseous discharge that forms in that region serves as its own energy-carrying medium. The result is an annular (hollow) plasma that extends into the open atmosphere. In appearance, the plasma produced by the MPT resembles a miniature inductively coupled plasma. Like the ICP and the plasma formed from a surfatron, the MPT discharge is sustained by power coupled into its periphery. As a consequence, molecular gases or even aerosols can be introduced directly into the center of the discharge without seriously affecting the coupling process. Matrix interferences are accordingly reduced and it becomes easier to apply the MPT to a range of sample types. Moreover, because of the flamelike structure of the MPT discharge, it can be examined in either an end-on or side-on mode and also can be coupled relatively easily to a mass spectrometer if that mode of measurement is desired. Importantly, the MPT is a recent enough innovation that it is still undergoing modification in laboratories throughout the world. However, it is also available commercially in a package intended for the detection of metals and nonmetals in gaseous samples. From this brief overview, MIPs appear still to be in their earliest growth phases, according to the modified “Seven Stages” outlined earlier. The techniques have been conceived and developed, but are still at a relatively early stage in that development. Instrumentation, particularly plasma support structures, is still being devised and tested, and figures of merit are still in the process of being improved. The technique has not yet reached its mature stage and has become widely accepted only as a detector for gas chromatography. If the pattern observed earlier in the evolution of chemical flames, high-voltage sparks, and the inductively coupled plasma applies also to the MIP, that discharge is likely to enjoy continued growth in the future. Systems for Measuring Atomic Emission Spectra Although greatest emphasis in this overview has intentionally been placed on an examination of atomic emission sources, a great deal could also be written about the evolution of systems to decode and detect atomic emission spectra. Milestones in this field are cited in Figure 4.
Dispersion Systems Like instrumentation for atomic emission sources, instrumentation for the dispersion and detection of atomic spectra began to be introduced more than a century ago. Initially,
JChemEd.chem.wisc.edu • Vol. 77 No. 5 May 2000 • Journal of Chemical Education
581
Waters Symposium: Atomic Emission Spectroscopy
spectra were dispersed by means of refraction, and a triangular prism was most commonly employed. Later, diffraction gratings were introduced and underwent the typical cycle of design, understanding, and innovation shown in the modified seven steps. After failings in some of the early mechanically ruled gratings were characterized, interferometric control was introduced. Also, as a better understanding of grating efficiency arose, the concept of blazing gratings was devised. In a blazed grating, each groove has a flat face, positioned at a specific angle, so diffraction of a desired order occurs off that flat face and in a specified direction. The result is highly efficient operation at that particular order, wavelength, and angle. Later still, holographic gratings were conceived that were free of ruling errors altogether and produced extraordinarily low levels of stray light. Means were then found to impart a blaze to holographic gratings also. During all these innovations, a routinely encountered problem was the need to provide both high dispersion (for good spectral resolution) and broad wavelength coverage (so a large number of different elements could be examined at one time). To meet both needs would require an unrealistically large spectrograph. A solution to this problem was found in the concept of crossed dispersion, introduced by George Harrison at MIT. In this mode of operation, high resolution but at extremely high orders of diffraction was achieved by means of an Echelle grating—one ruled very coarsely. Unfortunately, the high orders at which the Echelle grating was used caused severe overlap of the grating orders. To overcome this problem, Harrison employed a second grating or prism that offered a lower level of dispersion but in the perpendicular direction. The combination allowed separation of the orders in the one direction and high resolution in the other. In early work, this two-dimensional spectral display was measured photographically. Later, arrays of discrete detectors (ordinarily photomultiplier tubes) were introduced so a group of individual spectral lines could be measured simultaneously. Other approaches included a roving slit, to enable lines to be examined sequentially, and even a roving detector assembly. Only within the last decade has the true power of such crossdispersion systems been realized, by coupling them with modern charge-transfer detectors, either the charge-coupled device (CCD) or charge-injection device (CID) assembly.
Detectors for Emission Spectroscopy From this brief account, it is clear that the utility of any spectral-dispersion system depended greatly on the kind of detector that was available. The first detector to be used was, of course, the human eye. Its limited dynamic range and sensitivity soon gave way to use of photographic techniques. The photographic emulsion offered a number of highly attractive features: extremely high spatial resolution (limited only by the grain size in the emulsion), direct integration of light intensity, low cost, availability in extremely large formats (useful for measuring entire spectra at one time), and integral storage of a recorded spectrum. However, the emulsion also suffered from a number of important shortcomings, including a limited dynamic range, only moderate sensitivity, extremely nonlinear response, and unreliable long-term archiving of spectra. Also, because emulsions varied considerably in sensitivity from batch to batch, each had to be calibrated individually, a time-consuming and tedious business. 582
Milestones in Spectral Selection & Detection ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■
1860: 1893: 1937: 1930s: 1944: 1944: 1951: 1967: 1970: 1970: 1970s: 1973: 1990:
Bunsen & Kirchhoff; prism spectroscope H. A. Rowland; theory of diffraction gratings Schuhknecht; filter photometer M. Hasler; commercial grating spectrograph R. W. Wood; blazed gratings Direct-reading multichannel spectrometers G. R. Harrison; interferometric control of rulings Jobin-Yvon; Holographic gratings Snelleman, et al.; wavelength modulation methods Margoshes; echelle spectrometers with array detectors Commercial slew-scan systems introduced Horlick, Boumans; linear detector array spectrometers Pilon, Denton; commercial echelle–CTD spectrometer
Figure 4. Milestones in spectral selection and detection.
For these reasons, instrumentation after the 1940s emphasized the use of photoelectric detectors, sometimes vacuum phototubes but more commonly the photomultiplier tube (PMT). Just as the photographic emulsion evolved in sensitivity, utility, and flexibility over time, PMTs soon became available in a variety of sizes, shapes, and levels of performance as more about them was learned. Because of the convenience and simplicity of electrical or electronic readout, it was deemed preferable by many to employ a large number of laboriously aligned PMTs than to use a single photographic film or plate. This direct-reading photoelectric spectrometer became the workhorse of spark-emission spectrometry in the 1950s and remains dominant even to this day. For more than a generation, spectroscopists have sought an electronic equivalent of the photographic emulsion. The first glimmer of hope in achieving this goal was the introduction in the early 1970s of low-cost linear photodiode arrays (PDA) (26 ). Although television cameras had already been available for more than two decades, such devices did not offer the stability, reliability, low cost, convenience, or dynamic range required for spectrochemical applications. In contrast, linear PDAs were modest in cost, boasted an integrated photometric amplifier, and could be coupled in a relatively straightforward way to conventional polychromators. Like the other kinds of detectors or instrumentation employed in AES, linear PDAs evolved from devices originally meant for consumer applications to more sensitive chips tailor-made for spectroscopy. Entire generations of atomicemission spectrometers were specifically designed to take advantage of the capability of the linear PDA (27–29). Again, measurement systems evolved through improved understanding and the availability of better instrumentation. Only when a decidedly superior alternative became available did the PDA lose ground. This superior alternative came in the form of chargetransfer array detectors (30). The two types of such detectors— CCDs and CIDs—offer tremendous capability for AES. Each has its own strengths and weaknesses, but each possesses a broad dynamic range and, when used in the proper configuration, can be used to measure either an entire atomic emission spectrum or critical regions of that spectrum. Over the past decade, a host of new spectrometer configurations based on these devices have appeared commercially.
Journal of Chemical Education • Vol. 77 No. 5 May 2000 • JChemEd.chem.wisc.edu
Waters Symposium: Atomic Emission Spectroscopy
Recently, however, the success achieved with two-dimensional CCD and CID chips has spawned another cycle in thinking. Recognizing the stability and utility of earlier systems based on linear PDA systems, instrument manufacturers have begun considering the use of linear CCD or CID chips instead of the more costly two-dimensional arrays. Not long ago, two manufacturers announced spectrometric systems similar in concept to earlier designs that employed groups of singlechannel detectors made from individual exit slits and photomultiplier tubes. However, for each of these single-channel assemblies the new designs have substituted a linear CCD array, available now at extremely low cost because of their use in consumer products. With a large group of such arrays it therefore becomes possible to cover the entire spectral range of interest in atomic spectrometry. Also, because each array possesses a large number of detector elements (pixels), it can be positioned with much less care than would be necessary with a narrow exit slit; calibration can occur after assembly under software control simply by noting the pixels on which particular spectral lines fall. As in the case of sources and dispersion devices, it seems the cycle of detector development follows the typical pattern: conception, implementation, cyclic development, and competition. Conclusion From the foregoing account, it is apparent that atomic emission spectrometry has had a long and illustrious period of contribution to science and technology. Its components— sources, dispersion systems, and detectors—have undergone continuous revision, development, and characterization while at the same time new and competitive devices were being introduced. Each one of the components was then evaluated in the face of this competition; some continued to improve and succeed, whereas others became less important. With this repeating cycle of renewal, introduction, innovation, and competition, it is perhaps not surprising that AES began as a powerful means of elemental analysis and continues to this day to be the most widely used. Atomic emission spectrometry— it lasts and lasts and lasts. Acknowledgments Supported in part by the U.S. Department of Energy through grant DE-FG02-98ER14890 and by ICI Technology. Literature Cited 1. Fassel, V. A. Fresenius’ Z. Anal. Chem. 1986, 324, 511. 2. Ahrens, L. H.; Taylor, S. R. Spectrochemical Analysis; AddisonWesley: Reading, MA, 1961.
3. Alkemade, C. T. J.; Herrmann, R. Fundamentals of Analytical Flame Spectroscopy; Wiley: New York, 1979. 4. Grove, E. L. Analytical Emission Spectroscopy, Part I; Dekker: New York, 1971. 5. Laitinen, H. A.; Ewing, G. W. A History of Analytical Chemistry; Maple: York, PA, 1977. 6. Moore, F. J. A History of Chemistry; McGraw-Hill: New York, 1939. 7. Schrenk, W. G. Analytical Atomic Spectroscopy; Plenum: New York, 1975. 8. Slavin, M. Emission Spectrochemical Analysis; WileyInterscience: New York, 1971. 9. Alkemade, C. T. J. A Contribution to the Development and Understanding of Flame Photometry; Ph.D. Dissertation, University of Utrecht, Utrecht, The Netherlands, 1954. 10. Hieftje, G. M.; Malmstadt, H. V. Anal. Chem. 1968, 40, 1860–1867. 11. Bastiaans, G. J.; Hieftje, G. M. Anal. Chem. 1974, 46, 901– 910. 12. Bleasdell, B. D.; Wittig, E. P.; Hieftje, G. M. Spectrochim. Acta 1981, 36B, 205–213. 13. Boss, C. B.; Hieftje, G. M. Anal. Chem. 1979, 51, 1897–1905. 14. Boss, C. B.; Hieftje, G. M. Anal. Chem. 1979, 51, 895–901. 15. Childers, A. G.; Hieftje, G. M. Appl. Spectrosc. 1986, 40, 939–944. 16. Childers, A. G.; Hieftje, G. M. Anal. Chem. 1993, 65, 2753–2760. 17. Childers, A. G.; Hieftje, G. M. Anal. Chem. 1993, 65, 2761–2765. 18. Clampitt, N. C.; Hieftje, G. M. Anal. Chem. 1972, 44, 1211–1219. 19. Pak, Y.; Hieftje, G. M. Spectrochim. Acta 1985, 40B, 209–216. 20. Harrison, G. R. MIT Wavelength Tables; M.I.T. Press: Cambridge MA, 1969. 21. Kaiser, H.; Wallraff, A. Ann. Phys. 1939, 34, 297. 22. Coleman, D. M.; Walters, J. P. Spectrochim. Acta, Part B 1976, 31, 547–587. 23. Beenakker, C. I. M. Spectrochim. Acta, Part B 1976, 31, 485. 24. Moisan, M.; Beaudry, C.; Leprince, P. IEEE Trans. Plasma Sci. 1975, PS-3, 55. 25. Jin, Q.; Zhu, C.; Borer, M. W.; Hieftje, G. M. Spectrochim. Acta 1991, 46B, 417–430. 26. Multichannel Image Detectors; Talmi, Y., Ed.; American Chemical Society: Washington DC, 1979. 27. Multichannel Image Detectors, Vol. 2; Talmi, Y., Ed.; American Chemical Society: Washington DC, 1983. 28. Brushwyler, K. R.; Furuta, N.; Hieftje, G. M. Talanta 1990, 37, 23–32. 29. Brushwyler, K. R.; Furuta, N.; Hieftje, G. M. Spectrochim. Acta 1991, 46B, 85–98. 30. Charge-Transfer Devices in Spectroscopy; Sweedler, J. V.; Ratzlaff, K. L.; Denton, M. B., Eds.; VCH: New York, 1994.
JChemEd.chem.wisc.edu • Vol. 77 No. 5 May 2000 • Journal of Chemical Education
583