JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 75 Glow Discharge Considerations as a Versatile Analytical Source* Plenary Lecture W. W. Harrison Department of Chemistry University of Florida Gainesville FL 3261 I USA This paper reviews the advantages and limitations of the glow discharge as an analytical source for atomic spectroscopy. Salient features of the glow discharge are linked to specific analytical techniques that benefit from these characteristics. The direct analysis of samples in the solid state is of particular importance for ultratrace elemental methods for which dissolution of the sample can introduce unacceptable impurities. Keywords Glow discharge; sputter; excitation; ionization; negative glow In the past decade the inductively coupled plasma (ICP) has become a popular analytical spectroscopic source for elemental analysis both for atomic emission'*2 and for mass spectr~metry.~*~ The ICP certainly deserves the recognition it has received particularly for solution samples or those materials easily transformed into the solution state.Increas- ingly ICP techniques are being used for solid samples including those that might more conveniently be analysed directly in the solid state. Solid sampling accessories for the ICP,5 including lasers sparks filaments and furnaces have been used with some success but create a complex sampling train. A simpler integrated souce for the direct analysis of solids would offer some obvious advantages. This report examines the analytical characteristics of the glow discharge that make it increasingly useful as an analysis source for solids in a variety of spectroscopic techniques including atomic absorption (AA) atomic emission (AE) atomic fluorescence (AF) and mass spectrometry (MS).Glow Discharge Characteristics The glow discharge has many strengths that should make it the source of choice for elemental analysis of solids. Many of its advantages arise from the fundamental steps compris- ing its operation.6 As shown in Fig. 1 the voltage applied to the low pressure environment causes breakdown of the discharge gas normally argon whose ions are then acceler- ated across the dark space and impact on the cathode (sample) surface thereby ejecting primarily neutral atoms and electrons although some ions and polyatomic species are also released. In addition to the ions striking the cathode the charge-exchange processes in the dark space create fast atoms that also can cause sputter ablation.Through these sputtering steps the solid sample yields an atom population that diffuses across a thin dark space into the adjacent negative glow where collisions with electrons metastable atoms and other energetic species cause excita- tion and ionization of a fraction of the sample atoms creating species that are analytically useful. More detailed descriptions of these deceptively 'simple' processes are available.'J An intrinsic phenomenon of the glow discharge that distinguishes it from most other spectroscopic sources is a natural separation of the initial sampling step (atomization) from the subsequent analytical processes (e.g.excitation and ionization). Analysts often go to great lengths to create useful separations such as the tandem use of laser atomiza- tion and ICP ex~itation,~ or even ablation by one laser *Presented at the XXVIl Colloquium Spectroscopicurn Interna- tionale (CSI) Bergen Norway June 9-14 1991. I I I e-/q L Negative glow I Fig. 1 Representation of sputtering excitation and ionization processes in the glow discharge. Mo=neutral sample atom; * =excited state followed in space and time by a second laser for ioniza- tion.Io Fig. 2 shows a representation of this principle as manifested in the glow discharge and from which arise many inherent glow discharge applications and advantages. Step I represents sputter ablation as an atom generator that produces a steady-state sample atom population with a density gradient extending from the sample to the source walls.Atoms sputtered from complex sample matrices lose any chemical memory of that environment after their release and in effect undergo matrix conversion into a dilute argon solution where the analytical signals are produced as represented by Step 11. In this way the glow discharge has produced an advantageous transformation of a solid sample into a totally different matrix (low-pressure argon) that normalizes sample-to-sample inconsistencies. Clearly the space-time separation of atomization from emission and ionization zones is not as complete as indicated but the analytical effect as demonstrated in diverse sample matrices is a good virtual approximation.76 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 Fig. 2 Separation of atomization and analytical zones in the glow discharge. Mo= neutral sample atom Analytical Advantages To appreciate fully the growing popularity of glow dis- charges for atomic spectroscopy a brief review of the inherent advantages might be worthwhile. Direct solids analysis The feature that engenders the greatest interest in the glow discharge is its ability to accept an analytical sample directly in a solid form. Little or no sample preparation is required for bulk solids and even powders require only pressing into an appropriate electrode. Conducting ma- terials may be run by a d.c. discharge; non-conductors will require an r.f. dischargeLLIL2 to avoid the addition of a conducting matrix thus taking full advantage of the solid sampling benefits.Time-consuming sample dissolutions particularly of difficult sample matrices are avoided. Unlike solution techniques in which a solid sample normally undergoes at least a 100-fold dilution during sample preparation no dilution need occur with the glow discharge and the contamination problem associated with dissolution steps is avoided. Minimal matrix efects Solution-based techniques minimize the effects of the original chemical matrix by dissolving a sample in a solution matrix. The glow discharge is able to obtain similar effects while still sampling directly in the solid state primarily owing to the previously described zone separation (Fig. 2). When comparing many different sample matrices differential elemental response is fortunately slight.Rela- tive sensitivity factors determined in steel for example do not differ greatly from those found in aluminium or copper. l 3 Discharge options While a d.c. powered glow discharge is the normal choice and is the only type found in commercial instrumentation other formats have proved useful or are being developed. An r.f. discharge has the considerable advantage of running non-conducting samples such as glasses geological ma- terials or ceramics,** whereas a d.c. discharge requires alteration of the matrix for these samples to be electrically conducting. The d.c. or r.f. discharge also can be run in a pulsed mode with promising time-resolved advantages.15J6 A dual discharge configurationL7 offers an in situ standardi- zation against a known standard. Low power A small inexpensive power supply suffices for the glow discharge. While discharge conditions do vary with applica- tion power levels are often of a low wattage (e.g.1000 V at ? mA). This is in sharp contrast to the ICP which requires kilowatts of power with associated larger costs. Pulsed operation of a glow discharge necessitates a power supply that can accept an external driving pulse. Discharge stability With proper control of the discharge parameters analytical signals have good long term stability without the need for complex discharge monitoring. Perhaps only the glow discharge can produce a steady-state atomic population directly from a solid and continue to sustain that equilib- rium over sample lifetime.Given that actual sample consumption is small these equilibria conditions can be maintained essentially indefinitely (e.g. for hours). Low gas utilization Flow rates of the discharge gas normally argon are very small (e.g. 2-4 cm3 min-I). Laboratory costs from gas tank demurrage charges are often greater than the argon cost ilself. This may be contrasted to flow rates of other discharge methods such as the ICP where the flow is measured in litres per minute with all the attendant gas costs particularly in those countries in which argon is an expensive reagent. Llischarge gas flexibility While argon is the gas used most often for glow discharges many other gases are also suitable and there might be advantages to selecting the best gas for a given application.The normal alternatives are other rare gases. For example the use of neon might eliminate a particular spectral interference caused by an argon product. Different excita- tion and ionization efficiencies are also created by changing the gas. The heavier xenon can be useful for its greater sputter yield for certain elements or helium might be useful when the need is primarily for a high energy negative glow rogion but with little sputtering. Reactive gases such as oxygen nitrogen or even water vapour will easily support a glow discharge although the complexity of the spectra makes such use more specialized. There is also the possibility that a combination of gases (e.g. argon and hydrogen) might give some particular advantage such as a more reducing plasma.Sample adaptability Solids are of course the primary consideration. Bulk conducting samples can be fashioned into sample cathodes; powdered materials are pressed into discs or pins. Non- conducting samples can be mixed with a conducting powder matrix (aluminium copper graphite) or run di- rectly in the pure pressed form using an r.f. discharge. Solution samples can be run by either direct injection of desolvated aliquots or by drying a sample on an electrode for subsequent sputter release of the thin film residue into the plasma. Solutions also can be electroplated onto a conducting cathode.'* Gaseous samples have been analysed b:y bleeding the sample into the negative glow discharge.19 Sdfcleaning of samples A critical process in many trace element procedures is the removal of surface contamination from samples withoutJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 71 the concomitant introduction of other impurities from the cleaning itself. The glow discharge features the unique capability of permitting a final atomic polish by use of the discharge itself to bombard the sample with a high-purity gas thereby stripping away surface layers and exposing the bulk sample for subsequent analysis. This simplifies sample preparation and reduces contamination which is particu- larly important in glow discharge mass spectrometry appli- cations in which sub-ppb concentrations are measured. Matrix bond breaking To those accustomed to high-energy sources the efficiency with which the glow discharge is able to break molecular bonds even tenacious oxides,*O to produce an atomic population directly from a solid sample is often a surprise.The average energy of the argon ions striking the surface is reduced by charge exchange argon collisions in the dark space but sufficient momentum remains to exceed the sputter threshold energy of sample atoms. High-current arcs high-voltage sparks and high-power discharges use thermal means to tear bonded atoms asunder. Instead the glow discharge uses the atom-by-atom momentum transfer of each accelerated ion impacting on the sample to produce a form of 'three-dimensional billiards' whereby a surface atom is cleanly ejected from its matrix. This process can remove one or two sample atoms for every incoming ion. The fact that ion sputtering can remove atoms from metals and alloys might be more easily understood than the corresponding task of freeing atoms from complex tightly bound matrices.For example geological materials often feature silicate chains that encompass elemental inclusions all in a sample that is a non-conductor. Fig. 3 attempts to illustrate the added difficulty presented by a compacted sample consisting of 10% sample and 90% conducting matrix. The great majority of the bombarding species will strike the matrix particles and not the sample. In addition the non-conducting sample particle is continuously being covered by atoms sputtered from the matrix. Sensitive analyses of such materials by glow discharge techniques have been repeatedly demonstrated even so.z' Much is yet to be learned about the analytical processes that are central to the glow discharge in its interaction with these chemi- cally resistant samples.Uniform sputter response Essentially all elements exhibit sputter yields (the number of atoms released per impacting argon ion) that are relatively uniform within a factor of 3-5 of each other.z2 By comparison many analytical spectrochemical techniques produce sensitivities that vary by many orders of magni- tude among the range of elements examined. Thus the initial sputter step supplies subsequent analytical processes (AA AE AF MS) with a reservoir of sample atoms that will promote uniformity of response. Techniques that are affected by widely ranging transition probabilities exhibit considerable variation in elemental detection limits.On the other hand for techniques such as MS wherein the electron impact or metastable ionization steps are relatively uni- form detection limits are confined within a range generally construed as narrow. Sampling integrity After the discharge reaches stability normally within a few minutes the sputtered atoms presented to the glow dis- charge are representative of the bulk composition of the sample. The initial sputter yields will differ somewhat among elements causing surface enrichment of some elements and depletion of others. However as the sputter Compacted sample 90+ 10 Fig. 3 Considerations of a non-conducting compacted sample ( looh) mixed with a conducting matrix (90%) for glow discharge atomization. M =metal conducting matrix process continues to ablate its way through the sample those elements with lower sputter yields will be compen- sated for by larger surface concentrations shifting the plasma equilibrium towards a population more representa- tive of the sample atom concentrations.Trace element methods place great demands on the glow discharge sampling mode (i) representative atomic populations should be created from elemental concentrations that range over many orders of magnitude and (ii) the plasma populations should remain stable with respect to each other throughout the measurement process. The glow discharge performs well with respect to both aspects. Ease of accessibility Because the glow discharge requires little in the way of optical or other radiation shields or high voltage/current protection the source can be situated conveniently in a multi-windowed chamber (such as a standard six-way cross) that permits easy interaction with external probes or injection devices.While this is not a common feature with commercial instrumentation analysts designing their own sources for potential research applications find this an attractive feature. Atomic absorption or laser fluorescence is easily effected by the introduction of appropriate spectral beams through quartz window ports.' The discharge might be investigated by the insertion of measurement probes (e.g. Langmuir) without the danger of their melting. The ports also permit convenient sample injection or the introduction of special reagents. In such an open configura- tion the glow discharge is a very user-friendly source.Fundamental studies In this paper the glow discharge is being considered as a working analytical source for elemental analysis but analyt- ical chemists are also interested in the fundamental pro- cesses that take place in the discharge. Because many different analytical techniques can be easily interfaced to the glow discharge basic properties can be explored by monitoring sequentially or even simultaneously critical parameters such as atom densities excited state popula- tions and the number of ionic species. Even though local thermal equilibrium does not exist in glow discharges useful estimations can still be made of critical discharge parameters. In this manner the effect of variation in78 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 experimental parameters can be advantageously followed by monitoring a range of species that might be involved in a given reaction ionization mechanism etc. Spectral simplicity The sputter process in glow discharge sources yields primarily atomic species,’ with relatively low abundances of molecules and assorted polyatomic species. The negative glow reactions excite resonance lines and other high probability transitions leading to optical spectra that are essentially line dominant with little band structure and exhibiting very low background continua. Mass spectra are atomic in nature and thus relatively simple. The low energy nature of the glow discharge prevents the analyst from being overwhelmed by excessive data. Interpretation of both optical and mass spectra is thus simplified compared with corresponding ICP results.Non-elemen tal applications While the glow discharge is primarily an atomic source it plays an increasing role in alternative applications includ- ing organic samples.23 The sputter process generates some polyatomic and small cluster species that are of interest to physical chemists. Discharge conditions can be adjusted to enhance the formation of such species. In addition the plasma can serve as a simple and convenient detector for chromatographic effluents,24 using either optical or mass spectrometric read-out. The latter approach is used in several commercial instruments for the detection of organic compounds which are injected into the negative glow region and ionized.Glow discharge as a chemical reaction cell The glow discharge can be thought of as a low pressure reaction chamber in which countless chemical reactions are taking place or might be caused to occur by analyst intervention. In its normal rare-gas mode the source is hardly inert. As can be determined particularly by MS myriad reactions occur in an argon discharge that can be useful analytically or that can help in understanding reaction mechanisms. Optimization or alteration of critical reactions can greatly benefit the analyst. In addition the glow discharge can serve as a reaction cell for the injection of beneficial analytical reagents. Reagents may be added as a vapour to an incoming gas stream or by direct sputter injection from a cathodic sample. In some cases a secon- dary reagent electrode may be used for pulsed reagent addition by means of an auxiliary discharge.Glow dis- charges are fascinating in terms of the opportunities they present for chemical modification of plasma chemistry. For example reactive atoms can be added as reagents in a very controlled manner by pulsed sputter addition permitting a pseudo-titration of the host species. Analytical Limitations Primarily applicable to solids Just as the application of the glow discharge to solids is one of its great advantages the fact that solutions require special sample handling is a real limitation given the large number of materials that are either solutions by nature or are conveniently handled in solution form. It would be just as inefficient to use a solid-based technique for solutions as the reverse which is actually now much more prevalent. Spectral interferences The predominantly atomic nature of the glow discharge and the occasional overstatements regarding spectral purity can mislead the inexperienced analyst into failing to take proper precautions for spectral interferences.More spectral inter- ferences are seen at levels that cause problems as greater demands are placed on analyses at extremely low concen- trations. The discharge plasma will support the presence of unusual (and therefore often unanticipated) polyatomic species that can contribute to optical and mass spectra (i.e. molecular band species in optical emission and cluster ions in MS). Plasma susceptibility to contaminants The presence in the discharge gas of even trace amounts of certain contaminants greatly affects the operation of the source.2s The most notorious of these is water vapour which may enter through the discharge gas system leaks or more commonly by simple desorption from the surface of the source housing sample probe and inlet lines.Water affects discharge efficiency in two major ways. Firstly it quenches argon metastable species that are important in excitation and ionization mechanisms. Secondly and per- haps more importantly it yields discharge products such as H+ and H3+ that contribute disproportionately to the ion current striking the sample and yet because of their small mass produce little sputtering of the sample. Vacuum system requirements Glow discharges operate at reduced pressure requiring vacuum technology that is modest by today’s standards but because of the contaminant problem discussed above must be of high quality.In addition the water problem makes a cryogenic attachment to the source highly desirable parti- cularly for those elements at the low mass end of the spectrum. Low energy While the glow discharge sputter step is quite fairly efficient in breaking even tenacious bonds the energetics of the plasma itself are relatively low. Unlike the ICP which features high thermal energies throughout the analytical region of the torch the glow discharge is a low-energy source so that polyatomic species formed within the negative glow may not be completely dissociated. In fairness it should be noted that even the ICP does not fully dissociate all polyatomic species.Standards problem Solid-sampling techniques suffer a particular problem of difficulty in switching frequently between sample and standard something that can easily be done in solution methods. The low-pressure environment of a glow dis- charge exacerbates this problem because of the required passage of samples and standards in and out through vacuum interlocks and the attendant re-equilibration of the system each time. The net result is that internal standards are often used for glow discharge methods particularly in MS and since it is somewhat difficult to add a standard to a metal or alloy sample in the solid state an element intrinsic to the sample is selected often the matrix itself. An added discomfort to analysts in solids analysis is the lack of a real ‘blank for consideration of detection limits.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 79 Mass spectrometry + Atomic emission fluoresc Atomic absorption :ence -A ...,.,.lP- Laser RIMS Sample Fig. 4 Analytical techniques that have been demonstrated using a glow discharge RIMS resonance-ionization mass spectrometry; Po unattenuated radiant power; and P attenuated radiant power Analytical Versatility of the Glow Discharge These strengths and limitations of the glow discharge have led analysts and commercial instrument manufacturers to focus on the opportunities presented in today’s competitive analytical market-place. It is not a source that will necessa- rily replace highly active existing sources but there exist sufficient advantages to make it a complementary valuable approach.For example a recently introduced instrumentz6 has combined a glow discharge and an ICP in a single mass spectrometer permitting selection of the best source for the application at hand a possible harbinger of future develop- ments. Other have explored the instrumenta- tion and applications of broad glow discharge use in a variety of spectroscopic Settings. These are based on the generation of species that permit the analyst to select an ultimate measurement approach. The basic discharge processes in the glow discharge permit a variety of techniques to be employed. The initial sputter step primarily creates atoms so the analyst can take advantage of this process to conduct AA analysis. As the atoms diffuse into the glow discharge they are subjected to excitation interactions the relaxation from which serves as the basis for AE.Those atoms that encounter sufficiently energetic species suffer complete loss of an electron yielding ions representative of the sample easily employed for MS. Fig. 4 represents some of the analytical processes that can result. The ‘intrinsic’ selfcontained analytical phenomena should be distinguished from those that require external stimulation. The glow discharge is a complete source for AE and MS in that the negative glow produces photons and ions than can be sampled directly to provide analytical information. On the other hand AA requires a line source probe (normally from a hollow cathode lamp) to measure atomic populations.Similarly AF requires a laser probe to stimulate emission for the analytical measure- ment. The laser alsopermits resonant ionization techniques for high specificity. To varying degrees each of these processes finds increasing use in trace element analysis. Those techniques that have benefitted from commercially developed instrumentation are not surprisingly finding the greatest use. Glow discharge techniques came through a considerable gestation period long awaited withe interest by instrument manufacturers and are now in the midst of a maturing process. How long they remain a competitive player among the numerous elemental techniques depends on the per- ceived advantages and the development of some new source(s) that diverts attention and resources. Presently however the glow discharge exhibits advantages for solids analysis that offer promise of a substantial analytical lifetime.The author expresses appreciation to the Department of Energy Basic Energy Sciences for support. References 1 humans P. W. J. M. Inductively Coupled Plasma Emission Spectrometry Part I Wiley New York 1987. 2 Mayer G. A. Anal. Chem. 1987 59 1345A. 3 Houk R. S. and Thompson J. J. Mass. Spec. Rev. 1988 7 425. 4 Hieftje G. M. and Vickers G. H. Anal. Chim. Acta 1989 216 1. 5 Denoyer E. R. Fredeen K. J. and Hager J. W. Anal. Chem. 199 l,63,445A. 6 Chapman B. Glow Discharge Processes Wiley New York 1980. 7 Westwood W. D. Prog. Surf Sci. 1976 7 71. 8 Hamson W. W. and Bentz B. L. Prog. Anal. At. Spectrosc. 1988 11 53. 9 Arrowsmith P. in h e r s and Mass Spectrometry ed.Lub- man D. M. Oxford University Press New York 1990 p. 179. 10 Becker C. H. in Lasers and Mass Spectrometry ed. Lubman D. M. Oxford University Press New York 1990 p. 84. 1 1 Coburn J. W. Taglauer E. and Kay E. J. Appl. Phys. 1974 45 1779. 12 Winchester M. R. and Marcus R. K. J. Anal. At. Spectrom. 1990 5 575. 13 Sanderson N. E. Hall E. Clark J. Charalambous P. and Hall D. Mikrochim. Acta 1987 I 275. 14 Winchester M. R. Lazik C. and Marcus R. K. Spectrochim. Acta Part B 199 1 46 483. 15 Klingler J. A. Savickas P. J. and Hamson W. W. J. Am. Sm. Mass Spectrom. 1990 1 138. 16 Klingler J. A. Ph.D. Dissertation University of Florida Gainesville FL USA 199 1. 17 Klingler J. A. and Harrison W. W. Anal. Chem. 1991 63 2982. 18 Ahearn A. J. J. Appl. Phys. 1961 32 1197. 19 Strange C. M. and Marcus R. K. Spectrochim. Acta Part B 1991,46 517. 20 Coburn J. W. Taglauer E. and Kay E. Jpn. J. Appl. Phys. Suppl. 1974 2 501. 21 Brenner 1. B. Laqua K. and Dvorachek M . J. Anal. At. Spectrom. 1987 2 623. 22 Laegreid N. and Wehner G. K. J. Appl. Phys. 1961,32,365. 23 McLuckey S. A. Glish G. L. and Asano K. G. Anal. Chim. Acta 1989 225 25. 24 Pfasma Discharge LC-MS Technical Bulletin Kratos Analyti- cal Manchester UK. 25 Mei Y. and Hamson W. W. Spectrochim. Acta Part B 199 1 46 175. 26 TS SOLA Technical Bulletin Turner Scientific Wamngton UK. 27 Harrison W. W. Barshick C. M. Klinger J. A. Ratliff P. H. and Mei Y. Anal. Chem. 1990 62 943A. 28 King F. L. and Hamson W. W. Mass Spectrom. Rev. 1990 9 285. Paper 1 /04520B Received August 29 I991 Accepted November 11 I991