Foreword: Glow discharge spectrometry

Glow discharges (GD) are used for a variety of technological, physical and analytical applications, ranging from plasma etching and deposition systems in the micro-electronics industry, to lasers or even plasma monitors. For analytical applications glow discharges have a long and splendid history as sources for emission and mass spectroscopy. Although during the course of the last 35 years direct current glow discharges have matured,¹ as can be seen from a number of published books,^2–4 they are still not outdated because progress is still possible in many areas of research and application. It is therefore the aim of the Guest Editors to highlight current hot topics in glow discharge spectroscopy (GDS) in this special issue.

There are many reasons to compile a special issue on analytical glow discharges. Firstly, the Guest Editors believe it is timely. Secondly, about three years ago a European Network on glow discharge spectroscopy was established and was financially supported by the European Commission to assist in the further development of analytical methods using glow discharge sources both with optical emission and with mass spectrometry, including an extension of their range of application. A final report of this European Network by Edward Steers from North London University is included in this special issue.

This Network activity was supported by JAAS via a special glow discharge virtual section which was publicly available on the JAAS home-page for a couple of years. The Network stimulated European-wide research and co-operation on analytical glow discharges and a number of scientific research papers were presented at the final meeting which took place in Wiener Neustadt, Austria, last year. Some of these papers are now the basis of this special issue. But GDS is not only a European activity, it is a worldwide concern, demonstrated by contributions from many well-known experts.

The scope of the papers presented here can be summarized as follows. They describe how

to find new application niches, for instance at atmospheric pressures or for the direct analysis of liquids

to extend the experiences from continuous dc glow discharges to pulsed and rf discharges

to develop more reliable and controllable power supplies for rf

to explore the fundamentals by measuring or modelling different species in the plasma

to extend the use from inorganic to organic applications.

Fundamental research

Although glow discharge spectroscopy is an established and mature technique, fundamental studies are still needed to understand processes taking place to improve the sources applied.

Although the physical processes taking place in GD sources have been investigated by many different authors⁵ there is still a need to measure fundamental parameters such as temperatures or particle number densities for comparison with or refinement of theoretical models. For this purpose Thomson and Rayleigh scattering is a powerful tool because photons are used to probe the plasma of interest and the plasma is not disturbed, as is, for instance, not the case with Langmuir probes.⁶ Gamez et al. (see page 680) have therefore developed a laser scattering instrument to probe a GD plasma. They have detected and quantified three different groups of electrons in agreement with the findings of other experiments and have also measured the gas temperature at about 1100 K. Thus, in one experiment all the important parameters, such as densities and temperatures, can be measured in the same arrangement.

One interesting feature of GD is that gases such as C, O, N or H in metals can be analysed directly. Usually these gases are present in the samples at extremely low levels so that they do not alter the discharge conditions. Especially in modern hard metal coatings or ceramic coatings, metal oxides or nitrides are used. Hydrogen can be present from the coating process at higher levels so that, after dissociation of the mentioned elements, the gases as main components directly influence the sputtering, for instance the crater shape, the excitation and ionisation mechanisms and thus have a severe influence on calibration and quantification. The latter is still a challenge for ceramic layers. This is the reason why there is increasing interest in studies of gas addition to glow discharges operated with Ar. Three different studies of gas addition are featured in this issue.

The work from Smid et al. (see page 549) was performed using a Fourier transform emission spectrometer to record highly resolved emission lines of Ar, N and of selected metals from the cathode material. They show that nitrogen reduces the population of argon metastable atoms, thus affecting emission as well as ionisation processes and therefore complicating quantification.

In the paper from Hodoroaba et al. (see page 521) the “hydrogen effect” is investigated. It is shown that hydrogen in small amounts can also change the sputter conditions, for instance the crater formation and roughness of the surface. Again, changes in emission and ionisation processes are possible which can lead to a falsification or start-up phenomena, especially if water is adsorbed in the source or on the sample surface. It is shown by the authors that cleanness of the source or possibly the addition of hydrogen might be an approach to overcoming these limitations.

In a paper from Menéndez et al. (see page 557) hydrogen addition in large amounts is studied with a dc glow discharge ion source coupled to a TOF-MS. Again, an influence on the sputter process, a reduction of the sputter rates and a change of the crater shape are observed. Concerning the signals from analytes an enhancement has been observed in comparison to a discharge operated in pure argon. From this point of view the addition of hydrogen for GD-MS looks promising, if proton attachment to trace components can be neglected, which otherwise can cause severe spectral interferences.

In the paper by Bogaerts et al. (see page 533) a comprehensive modelling network is applied to calculate the processes taking place in millisecond pulsed glow discharges, a topic which is of particular interest in this special issue. In this model, special attention is paid to the mechanisms which cause an enhancement of some emission line intensities, especially of the highest excited levels, in the afterglow (so-called “afterpeak”), which make pulsed discharges very interesting for analytical applications because time-resolved detection is possible. Using typical discharge conditions described in experiments, the afterpeak is not observed in the calculations directly, but if the electron density is increased by about two orders of magnitude (compared with steady state conditions) in the model, then collisional radiative recombination between Ar⁺ ions and electrons can become a dominating process leading to the effects observed experimentally. Alternatively, dissociative recombination of Ar₂⁺ ions with electrons may be involved as well. So the model suggests new experiments are required in which time-dependent densities of electrons or Ar₂⁺ ions can be measured.

The paper by Lewis et al. (see page 527) describes a fundamental study of the primary excitation mechanisms of a millisecond pulsed rf discharge in comparison with the same source operated with dc. The final finding of the authors is that the differences of both modes (dc and pulsed rf) are negligible.

Potapov et al. (see page 564) describe the operation of a thin walled metallic hollow cathode in a pulsed mode for application in atomic absorption (AAS) and mass spectrometry. In AAS, matrix effects by background absorption could be reduced significantly, similar to the well-known furnace atomic non-thermal emission spectrometry (FANES) technique, in comparison with a Zeeman AAS system, by improved dissociation of molecules. In mass spectrometry the authors compare the results from their own models with the mass spectra and conclude from this that in the afterglow Penning ionisation is the dominating process.

Direct current discharges

GD-OES

Traditionally dc-GD optical emission spectroscopy (GD-OES) is mainly applied in the materials sciences where it is used routinely for bulk and surface analysis. Here the success of glow discharges as spectrochemical sources is strongly dependent on competing methods. For direct solid analysis, GDS competes with X-ray techniques, but it has fewer matrix effects and lower detection limits. In the last couple of years, there is increasing competition from other plasma techniques such as ICP-OES and ICP-MS when these techniques are coupled with adequate sample introduction techniques such as spark or laser ablation. Here, the main argument in favour of GD techniques is that they are less expensive and can be applied for the analysis of technical layers, which cannot be achieved by competing techniques. Also, the excellent depth profile capabilities of GD-OES and the possibility of analysing light elements should be mentioned here.⁷

Concerning bulk analysis a more spectacular application in comparison to routine analysis of metals is presented by Lavoine et al. (see page 572). They describe the operation of a GD-OES instrument in a glove box for the analysis of C, N, O and H in nuclear materials, which is not an easy task if interference from Pu with more than 20000 spectral lines is taken into account. The instrument development is described as well as the line selection in the vacuum ultraviolet region between 120 and 160 nm, which is in itself a challenge. First experimental results presented here demonstrate that the development was successful.

Nowadays more than 90% of all commercial GD-OES instruments are applied to characterise coated materials. The number of applications is increasing from year to year and this was the reason why the first international conference on GD-OES was organised last year in Japan. A conference report is given elsewhere.⁸ Two papers in this special issue are attributed to depth profiling analysis. The second of these is a review by Angeli et al. (see page 670) discussing the special needs and requirements of near surface and thin film analysis to promote the application of GD-OES in those areas where conventional surface analytical techniques such as SIMS or Auger spectroscopy are used.

Much thicker layers are investigated in the paper by Xhoffer and Dillen (see page 576), using GD-OES for a wide variety of coated and uncoated steel products, where most often qualitative results are sufficient to identify problems or to control the production process. Here, in the steel industry, GD-OES is not usually a stand-alone-technique but is complemented by a multi-method analytical approach, so that the excellent performance can be objectively assessed.

Most commercial instruments in GD-OES use spectrometers with a pre-selected number of emission line channels. The most prominent or less interfered lines have been selected, whereas most of the remaining lines are neglected. In doing so, up to 99% of the total sensitivity is wasted and not used analytically. Therefore an approach presented by Weiss (see page 584) is a first attempt to use a set of emission lines for calibration by application of a CCD-based simultaneous spectrometer. By this approach he could improve the accuracy obtained for a Fe–Ni–Cr alloy significantly, a result which can stimulate development of novel instrumentation in GD-OES.

Various sample types have been analysed by GD-OES ranging from gases, particles and solids to liquids. An overview is given in a previous review on GDS.⁹ In this issue two studies are presented where liquids are analysed by GD-OES.

In the first study of Jin and Marcus (see page 589) the coupling of a particle beam sample introduction interface to a hollow cathode and detection by emission spectroscopy was used to apply a GD to speciation studies of organoselenium compounds. But what makes this coupling most interesting is the capability of measuring H and C additionally to selenium, so that the empirical formula can be derived directly from the measured signals.

However, even without an interface, liquids can be analysed directly at atmospheric pressures, when applying the atmospheric electrolyte cathode glow discharge cell approach with detection by emission spectroscopy, as described by Cserfalvi and Mezei.¹⁰ In comparison to their previous design a capillary is now used to decrease the cell volume as well as the liquid flow rates, resulting in improved detection limits at ng levels for a number of heavy and toxic metals measured in a flow injection mode (see page 596). By this modification their design is coming closer to something that is well-known in organic mass spectrometry, i.e., the electrospray technique, although it is here in combination with emission spectroscopy.

GD-MS

In comparison to GD-OES the more expensive commercial version of GD-MS is mainly applied for ultra-trace bulk analysis. Here dc-GD is well established, for instance in the production of pure metals and alloys for semiconductor applications, where certificates from glow discharge mass spectroscopy (GD-MS) are obligatory. Even after 20 years these applications are fully dominated by only one commercial instrument, the VG 9000,¹¹ a sector field device which can be operated in a high mass resolution mode. Most analyses run on these instruments are performed in a semi-quantitative mode, which does mean that relative sensitivity factors are used from a library or from a pre-calibration. Although the operational conditions might vary from user to user, the accuracy achieved so far is convincing, as has been demonstrated in a previous report from Venzago et al. of a round robin analysis of 4N Al samples.¹² In the report presented here by Kasik et al. (see page 603) two grades (4N and 5N) of copper standard reference materials (BCR-CRM 075 and 074) have been investigated by GD-MS as well as by GD-OES. Ten different laboratories with 11 different instruments have participated and up to 32 elements have been analysed. However, most of the elements of interest were too close to the quantification limit of GD-OES so mainly GD-MS results were presented. Again, similar to the finding of the Al round robin analysis, a very good uniformity of the data has been achieved by the different laboratories, although different instrumental conditions for the GD (discharge current and voltage) and types of samples (pin type and flat samples) were used. A good uniformity is, however, not so surprising if all candidates have used the same set of sensitivity factors. Therefore, it is a pity that from the results presented here it is not clear whether a semi-quantitative approach or a full calibration approach was used for quantification of the different GD-MS instruments, whereas such information is needed if one wants to assess the accuracy achieved.

Analyses of thick technical layers by use of a VG 9000 instrument have been reported in the past by various authors.^13,14 If the layers are too thin, the scanning speed of this instrument is no longer sufficient to measure transient data. “Time of flight” (TOF) instruments, on the other hand, have the highest scanning speed and therefore are very well suited for this purpose.¹⁵ This was the reason why Peláez et al. (see page 612) have coupled a home-made glow discharge ion source to a TOF instrument and have optimised the glow discharge working conditions for depth profiling. Results of the measured intensity–time profiles for Zn coatings are presented and are validated by results from an rf-GD-OES instrument, demonstrating a good correlation, although the sources and operational conditions used were quite different.

MS is very often applied in the semiconductor industry as a process monitor of reactive sputter devices based on dc or rf glow discharges, to trace the species which are formed in the glow discharge by reactive gases. For this application more instruments have been installed worldwide than for direct analysis of solid samples. We are not aware of these applications, because most often the results obtained are not presented in analytical journals, although gas phase chemistry and analytical spectroscopy is needed to optimise the processes involved. From this point of view the paper presented by Snyder et al. (see page 618) is very informative for these kinds of applications. In this paper the sample investigated is not the cathode material directly but the different gas phase molecular species in the glow discharge generated from the cathode material. The authors used GD-MS as well as a residual gas analyser to characterise and optimise the processes involved in an Ar–O₂ dc magnetron sputtering device, if metals such as Ag, Sn or Ti are used as the cathode material.

Pulsed dc discharges

As has been mentioned in the fundamental section, pulsed discharges have a number of advantages in comparison to sources operated with dc and this holds true for mass spectrometry as well as for emission spectroscopy.^16–18 For both methods time gated detection is an important tool to improve the signal to noise ratios. For instance, in emission spectroscopy the signal to background ratio in the “afterpeak” can be improved whereas in mass spectrometry argon related ion intensities fall off more rapidly than those of metal ions. Both effects can be used to improve detection limits. Pulsed discharges are gaining increasing interest in mass spectrometry, especially if they are coupled to TOF instruments.

In the paper of Jackson et al. (see page 665) a pulsed discharge is applied to reduce the sputter rates of critical (toxic or radioactive) samples without losing too much intensity. They have developed a new dc source with additional electrodes to enhance the matrix ion current so that their new source works more like a duo-plasmatron ion source rather than a conventional dc discharge. In this way, an improvement by about one order of magnitude has been reported in comparison with a previous design. A simple sample preparation is described by which non-conducting (oxide) powders are embedded in a metal capillary. Using the same source again with a metal capillary as a leak, organic traces in ambient air can be analysed too, demonstrating the flexibility of this new source.

In the paper of Pisonero et al. (see page 624) a double-pulsed GD was coupled to a TOF mass spectrometer. The idea of such an approach is to use a first pulse mainly for sputtering and the second pulse to improve the ionisation of the sputtered material. Depending on pulsing delay and gas flow rates an enhanced analyte ionisation was observed.

A couple of years ago various research groups observed an interesting effect of glow discharges used as ion sources in mass spectrometry for speciation of elements. In a low current mode intact organic molecules have been detected to measure the molecular mass, whereas for higher currents also the atom ion intensities of the elemental composition could be measured. Due to the fact that such a source can be used to measure both the atomic as well as the molecular information of organic compounds the sources were called “tunable sources”. A special issue about tunable plasma sources in analytical spectroscopy has been published in JAAS¹⁹ previously.

Lewis et al. (see page 629) describe an elegant new method for the same purpose. They have coupled a GC to a microsecond pulsed GD ion source and used a TOF mass spectrometer to detect ions at three different time regimes, again giving information about the molecular and the atomic composition and producing additionally a fragmentation pattern similar to electron impact ionisation. The latter is most important for application of search routines in data base systems which are always related to electron impact fragmentation. A brief description of source optimisation is given and the new approach is applied for determination of aromatic and chlorinated hydrocarbons.

RF discharges

RF-glow discharges are applied in industry on a routine basis, although not all problems for calibration and quantification are solved. For instance, in emission spectroscopy the emission yields are used for calibration. They are normalised to discharge conditions such as sputter rates and discharge power, and the measurement of the latter is not available on a direct read-out as is the case for dc discharges. Indeed, even without plasma displacement currents the source may consume even more power than the discharge itself. For the analysis of layers this problem might become even more pronounced, because the discharge and sputter conditions might be completely different between the layer and bulk material. For this application the reduced sputter rate, which is the sputter rate normalised to the discharge power, is again needed to transform the measured sputter time into a depth scale. Many examples have been shown by various authors at conferences with the conclusion that an exact measurement of the power dissipated by the discharge is essential for calibration and quantification. Therefore, three papers in this issue present novel approaches on how to cope with this problem.

In the paper of Marshall et al. (see page 637) the losses are analysed and assessed and a new rf-design is presented. The blind power (without plasma) is measured and subtracted from the actual power after ignition of the plasma and cable losses are minimized. The new design is evaluated and experimental evidence is collected, presented and discussed but validation by the users in routine applications is needed to assess this new approach.

In a second paper a novel approach is presented by Wilken et al. (see page 646) to measure directly the correct plasma voltage and current in the source and by knowledge of these to calculate the power dissipated only by the plasma. The authors found out that higher harmonics can be generated easily, which can cause severe problems in a correct measurement, if not compensated by adequate electric low pass filters. From the experience of this team manufacturers can learn how to improve their instrumentation and make rf-discharges as reliable and applicable as dc-sources.

In the last paper of this section, presented by Payling et al. (see page 656), all components of an RF-powered GD are analysed and measured to produce equivalent circuits which then are used to calculate the plasma resistance. By the knowledge of the plasma behaviour the emission yields in a multi-matrix calibration are corrected, and are compared to those obtained by corrections applied in a dc mode.

Even nowadays GD techniques are very attractive in many routine applications for bulk and surface analysis. Concerning research, the interest has changed from dc applications to pulsed dc or rf applications. Mass spectrometry, especially by use of TOF instrumentation, is becoming more and more the basis for novel inorganic and organic applications.

For the future, GD are becoming smaller and are operated at higher pressures at even lower powers. Because of this they become ideal plasma sources for miniaturized analytical devices, as will be discussed soon in JAAS in a review article by Franzke et al.²⁰

Norbert Jakubowski

Volker Hoffmann

Annemie Bogaerts

References

1. H. Grimm, Spectrochim. Acta, Part B, 1968, 23, 443.
2. Glow Discharge Optical Emission Spectrometry, eds. R. Payling, D. Jones and A. Bengtson, John Wiley & Sons, Chichester, 1997.
3. Glow Discharge Spectroscopies, ed. R. K. Marcus, Plenum Press, New York, 1993.
4. Glow Discharge Plasmas in Analytical Spectroscopy, eds. R. K. Marcus and J. A. C. Broekaert, John Wiley & Sons, Chichester, 2003.
5. J. A. C. Broekaert, Appl. Spectrosc., 1995, 49, 12A.
6. A. Bogaerts, A. Quentmeier, N. Jakubowski and R. Gijbels, Spectrochim. Acta, Part B, 1995, 50, 1337.
7. A. Bengtson, Spectrochim. Acta, Part B, 1994, 49, 411.
8. K. Shimizu and N. Jakubowski, J. Anal. At. Spectrom., 2003, 18, 2N.
9. S. Baude, J. A. C. Broekaert, D. Delfosse, N. Jakubowski, N. G. Orellana-Velado, R. Pereiro and A. Sanz-Medel, J. Anal. At. Spectrom., 2000, 15, 1516.
10. T. Cserfalvi, P. Mezei and P. Apai, J. Phys. D, 1993, 26, 2184.
11. K. Robinson and R. Naylor, Eur. Spectrosc. News, 1986, 68, 18.
12. C. Venzago, L. Ohanessian-Pierrard, M. Kasik, U. Collisi and S. Baude, J. Anal. At. Spectrom., 1998, 13, 189.
13. A. Raith, R. C. Hutton and J. C. Huneke, J. Anal. At. Spectrom., 1993, 8, 867.
14. C. Venzago and M. Weigert, Fresenius’ J. Anal. Chem., 1994, 350, 303.
15. M. J. Heintz, D. P. Myers, P. P. Mahoney, G. Li and G. M. Hieftje, Appl. Spectrosc., 1995, 49, 945.
16. X. Yan, W. Hang, B. W. Smith, J. D. Winefordner and W. W. Harrison, J. Anal. At. Spectrom., 1998, 13, 1033.
17. A. Bengtson, C. Yang and W. W. Harrison, J. Anal. At. Spectrom., 2000, 15, 1279.
18. C. Yang, M. Mohill and W. W. Harrison, J. Anal. At. Spectrom., 2000, 15, 1255.
19. R. K. Marcus, E. H. Evans and J. A. Caruso, J. Anal. At. Spectrom., 2000, 15, 1.
20. J. Franzke, K. Kunze, M. Miclea and K. Niemax, J. Anal. At. Spectrom., 2003 10.1039/B300193H.

---