Anal. Chem. 2001, 73, 2903-2910
An Atmospheric Pressure Glow Discharge Optical Emission Source for the Direct Sampling of Liquid Media R. Kenneth Marcus* and W. Clay Davis
Department of Chemistry, Howard L. Hunter Chemical Laboratory, Clemson University, Clemson, South Carolina 29634-0973
A glow discharge optical emission spectroscopy (GD-OES) source that operates at atmospheric pressure is described. This device utilizes an electrolytic solution containing the analyte specimen as one of the discharge electrodes. The passage of electrical current (either electrons or positive ions) across the solution/gas phase interface causes local heating and the volatilization of the analyte species. Collisions in the discharge region immediately above the solution surface result in optical emission that is characteristic of the analyte elements. Operation of this device with the analyte solution acting as either the cathode or anode is demonstrated. Currentvoltage (i-V) plots reveal abnormal glow discharge characteristics, with operating parameters being dependent on the electrolyte concentration (i.e., solution conductivity) and the gap between the solution surface and the counterelectrode. Typical conditions include discharge currents of 30-60 mA, and potentials of 500-900 V. Electrolyte solutions having pH, pNa, or pLi values of 0.5-2 and interelectrode gaps of 0.5-3 mm produce stable plasmas in which the analyte solutions flow at rates of up to 3.0 mL/min. Preliminary limits of detection are determined for the elements Na, Fe, and Pb to be in the range of 11-14 ppm (∼60 ng) for 5-µL sample volumes. Glow discharge (GD) plasmas have been used as spectrochemical (i.e., optical emission) sources for well over 100 years, dating back to the early studies of atomic structure.1,2 The lowpressure, low-power plasmas are easily controlled and yield emission spectra that are principally atomic in nature. The combination of cathodic sputtering as a means of introducing atoms from bulk solids and the relatively simple optical spectra lead to the implementation of hollow cathode GD devices as line sources for atomic absorption spectrophotometry.3 The development of the Grimm-type glow discharge geometry led to the use of glow discharge optical emission spectroscopy (GD-OES) as a tool for both bulk solid and depth-resolved analysis of metals and alloys.4-6 The subsequent introduction of radio frequency (rf) * To whom correspondence should be addressed. (1) Pashen, F. Ann. Phys. 1916, 50, 901-940. (2) Shuler, H. Z. Phys. 1929, 59, 149-153. (3) Walsh, A. Spectrochim. Acta 1955, 7, 108-117. (4) Grimm, W. Spectrochim. Acta 1968, 23B, 443-454. 10.1021/ac010158h CCC: $20.00 Published on Web 05/25/2001
© 2001 American Chemical Society
powering schemes opened up the scope of application further to nonconductive materials and coatings.7 One of the strongest features of standard glow discharge devices is the fact that they operate in inert environments and are, thus, free from atmospheric contaminants. Although the cathodic sputtering event entails sufficient energy to release neutral atoms and molecules from solid matrixes, the discharge’s gas-phase temperature is insufficient to affect desolvation of analytes introduced in water vapor, as is typical in atmospheric pressure flames and plasmas. As such, a good deal of effort has been devoted to developing strategies of getting liquid-originating analytes into the discharge environment.8-13 The most common method involves drying an aliquot of analyte-containing solution on an inert target that is subsequently introduced as the cathode of the GD source, and the dried residue is sputtered from its surface.8,9 In this way, solvent vapors are excluded from the discharge volume, and the plasma is operated much in its normal manner. Although effective, this approach is laborious and is not amenable to what would ideally be the analysis of flowing streams, such as liquid chromatograph eluents. To address this shortcoming, transport-type liquid chromatography-mass spectrometry (LC-MS) interfaces, such as the moving belt and the particle beam, have been used to introduce dried analytes to the plasmas in a continuous fashion.10-12 Schroeder and Horlick have also attempted to introduce nebulized solutions directly into a hollow cathode emission source with some level of success.13 Over 40 years ago, Couch and Brenner described a phenomenon by which a glow discharge plasma was sustained at atmospheric pressure between a tungsten anode and an electrolyte solution that acted as the cathode.14 Solutions containing Cu and (5) Broekaert, J. A. C. Atomic Emission Spectrometry in Glow Discharge Spectroscopies; Marcus, R. K., Ed.; Plenum: New York, 1993; Chapter 4. (6) Payling, R., Jones, D. G., Bengtson, A., Eds. Glow Discharge Optical Emission Spectrometry, John Wiley: Chichester, 1997. (7) Marcus, R. K.; Harville, T. R.; Mei, Y.; Shick, C. R., Jr. Anal. Chem. 1994, 66, 902A-911A. (8) Harrison, W. W.; Prakash, N. J. Anal. Chim. Acta 1970, 49, 151-159. (9) Ryu, J. Y.; Davis, R. L.; Williams, J. C.; Williams, J. C., Jr. Appl. Spectrosc. 1988, 42, 1379-1387. (10) Brackett, J. M.; Vickers, T. J. Spectrochim. Acta 1983, 38B, 979-985. (11) You, J.; Dempster, M. A.; Marcus, R. K. Anal. Chem. 1997, 69, 34193426. (12) Gibeau, R. E.; Marcus, R. K. Anal. Chem. 2000, 72, 3833-3840. (13) Schroeder, S. G.; Horlick, G. Spectrochim. Acta 1994, 49B, 1759-1773. (14) Couch, D. E.; Brenner, A. J. Electrochem. Soc. 1959, 106, 628-629.
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001 2903
In dopants produced optical emission spectra analogous to those obtained in flame emission sources. On the other hand, solutions containing other cationic species (Li, Na, S, and U) did not yield characteristic spectra. That device was actually a modified version of a system originally described by Gubkin15 and later reviewed by Hickling and Linacre16 that was employed for very high yield electrolysis of aqueous solutions of metal salts. Cserfalvi and coworkers reinvestigated this phenomenon as a means of analyzing dissolved metals in electrolytic solutions, coining the term electrolyte-cathode discharge (ELCAD).17-20 In their original apparatus,17 a tungsten electrode (acting as the anode) was mounted 1-5 mm above the electrolyte-containing solution that was in electrical contact with a graphite rod held at the cathodic potential of the discharge circuit. Analyte-containing solution flowed at rates of 2-10 mL/min through a “fountain” to maintain electrical contact with the cathode. A glass frit separated the fountain receiving basin and the cathode rod to eliminate the evolution of H2 gas and a possible explosion. Current-voltage (i-V) plots generated for that device supported the assumption that the devices did, indeed, operate in the so-called “abnormal” glow discharge regime. Both operating voltage and observed analyte emission responses were dependent on the pH of the test solutions, with the authors suggesting that solution conductivity, and more specifically, hydronium ion concentration, is a key aspect of the physical operation of the devices. Detection limits for more or less bulk solutions of metal analytes produced detection limits of ∼0.1-1 ppm, although for total analyte solution volumes of >10 mL. Subsequent studies on the ELCAD source by Mezei, Cserfalvi, and Ja´nossy,19 sought to elucidate the operating mechanism of the device. The authors used a variable pressure cell to study the role of gas phase collision frequency on the operating characteristics. In most cases, increases in gas (atmosphere) pressure from 500 to 1200 mbar yielded greater emission intensities, which the authors ascribed to increased three-body recombination of analyte ions sputtered from the solution surface (i.e., M+ + e + e f Mo + e). Neutralized analytes atoms could then in turn be excited in the plasma region immediately above the solution surface. On the basis of the known field structure in the vicinity of the cathode electrode in a glow discharge, the actual release of a cationic species from the surface seems very unlikely and will, in fact, be refuted by experimental data present herein. The authors subsequently calculated a gas-phase temperature above the cathode surface on the basis of an assumption of the kinetic energy of ions colliding with the liquid surface. A gasphase temperature of ∼7000 K was suggested.20 Hereto, the conclusions drawn are counterintuitive for glow discharge devices and will be refuted in subsequent sections of this manuscript. Kim and co-workers recently described an extension of the studies of Mezei et al. by the use of an ELCAD system wherein argon is introduced as the discharge gas in a psuedo-closed vessel (15) Gubkin, J. Ann. Phys. 1887, 32, 114-115. (16) Davies, R. A.; Hickling, A. J. Chem. Soc. 1952, Part III, 3595-3602. (17) Cserfalvi, T.; Mezei, P.; Apai, P. J. Phys. D 1993, 26, 2184-2188. (18) Cserfalvi, T.; Mezei, P. J. Anal. At. Spectrom. 1994, 9, 345-349. (19) Mezei, P.; Cserfalvi, T.; Janossy, M. J. Anal. At. Spectrom. 1997, 12, 12031208. (20) Mezei, P.; Cserfalvi, T.; Janossy, M. J. Phys. D 1998, 31, L41-L42.
2904
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
system that was purged through a bubbler.21 In their design, a platinum wire anode was placed opposite the analyte fountain having flow rates of 5-10 mL/min. The Ar gas also served to reduce the possibility of explosion, and the high solution flow rates kept the sample solution from boiling. This group performed parametric studies of the sorts described above, finding as well that the plasma i-V characteristics were representative of an abnormal glow discharge with dependencies on both the pH of the solution and the interelectrode gap. Interestingly, the authors observed no emission from the Ar discharge gas species, although in the wavelength range investigated (400-500 nm), only Ar(II) species would be expected to be present. Hereto, a mechanism whereby ions of the analyte metals were sputtered from the solution (cathode) surface, subsequently neutralized in the cathode dark space, and then excited within the plasma was forwarded, with the parametric dependencies indicating that some sort of sputtering threshold must be realized prior to analyte release. Analyte emission intensities were found to come to steadystate conditions following 1 min of introduction at flow rates of 10 mL/min. Once stabilized, analyte signal variability of ∼2.5% RSD was observed. Limits of detection (LOD) were subsequently calculated to be in the 0.001-1 ppm range. Interestingly, although not noted by the authors, the elemental sensitivities roughly follow the relative volatilities of the analytes. Extension of that work to an open-air system yielded similar operating characteristics,22 although the analytical performance was degraded relative to the time required to reach the steady state (2-3 min.) and precision (
