A Comparison of Different Methods of Ionizing GC Effluents

Transcription

1 JOURNAL OF CHROMATOGRAPHIC SCIENCE, VOL. 24, NOVEMBER 1986 A Comparison of Different Methods of Ionizing GC Effluents P.L. Patterson Detector Engineering & Technology, Inc., 2212 Brampton Road, Walnut Creek, California Abstract Some of the most widely used, most sensitive detection methods available for gas chromatography (GC) function by directly converting chemical species in the GC effluent to gas-phase ionization. This article reviews the scientific principles that provide the basis of operation for the flame ionization detector, the electron-capture detector, the photoionization detector, and the thermionic ionization detector. These popular detectors are described from the viewpoint of their common characteristics as ionization detection methods, their specific differences in mechanism of operation and performance characteristics, and their most significant design criteria. Introduction The function of a gas chromatography (GC) detector is to produce an electrical signal in response to chemical compounds eluting from a chromatographic column. In addition to signaling the moment of elution of a given compound, the detector provides the means whereby the quantity of the compound can be electronically measured and recorded. Detection devices that function by directly converting incoming chemical species to gasphase ionization have been especially successful as GC detectors. The general category of ionization detectors includes specific types like the flame ionization detector (FID), the electron-capture detector (ECD), the photo-ionization detector (PID), and the thermionic ionization detector (TID). These represent some of the most widely used, most sensitive detection methods available for GC. The nomenclature of a type of GC detector usually describes its scientific principle of operation. There are currently many different types of GC detectors which encompass a very impressive range of different physical and chemical phenomena. A basic understanding of the phenomena active in a given detector can be a valuable guide to the optimum design and use of that detector. The present paper reviews the operating principles of the four previously mentioned detector types: FID, ECD, PID, and TID. These detectors are discussed from the standpoint of their common characteristics as ionization detection methods, their specific differences in mechanism of operation and performance characteristics, and their most significant design criteria. Features Common to All Ionization Detectors lonizaton production Gas-phase ionization is produced within the working volume of a GC detector by the application of energy in one form or another. In an FID, the energy supplied is the chemical and thermal energy of an H 2 /air flame; in an ECD, ionization is produced from the energy of beta particles emanating from a radioactive material; in a PID, the ionization energy is provided by irradiation with ultraviolet light; and in a TID, the energy is the catalytic and thermal energy of a hot, solid surface. For each of these detectors, the ultimate design goal is to choose a geometrical configuration and operating conditions that maximize the ionization signal from the samples of interest while minimizing the ionization signal from the detector background environment. In the FID, PID, and TID, there are usually small background levels of ionization, and the introduction of samples produces an increase in the ionization level. In the ECD, a high level of ionization is always present, even in the absence of sample. The ECD is unique as an ionization detector in the sense that it functions by a process of redistribution of ionization between different types of ionized species rather than by an increase in the overall ionization level. In all these detectors, the ionization ultimately produced is gas-phase ionization, which then must be moved through the prevailing gas environment to a metallic ionization collector electrode. Ionization collection and measurement The overall efficiency of an ionization detector depends on the efficiency of ionization collection as well as the efficiency of ionization production. All ionization detectors contain a collector electrode and a nearby polarization electrode. The voltage difference between the polarization and collector electrodes creates an electric field that causes gas phase ionized species of a given electrical polarity to move toward the collector. Some ionization detectors collect positive ionization, while others collect negative ionization. In all cases, the specific geometical con- 466 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.

2 Journal of Chromatographic Science, Vol. 24, November 1986 figuration of the polarization and collector electrodes determines the pattern of electric field lines within the detector volume; and this electric field has a strong influence on the collection efficiency of the detector. An improperly designed electric field can result in some ionization being driven to a grounded detector wall instead of to the collector electrode. This inefficiency of collection usually shows up as a non-linearity in the performance characteristic of the detector. The magnitude of ionization current arriving at the collector electrode in an ionization detector can be very small. Accurate measurement of these small currents requires that there be no extraneous leakage currents from electrical insulating materials used to isolate the collector from adjacent structures. Since GC detectors must be capable of operating at very high temperatures, the electrical leakage requirement severely limits the insulating materials that can be used. In many commerically available ionization detectors, high purity alumina ceramic is used to provide the appropriate electrical insulation at high temperatures. This electrical insulation requirement is one of the most important considerations in the design of an ionization detector. Signal measurement in an ionization detector is usually accomplished with an electrometer. State-of-the-art GC electrometers currently have noise levels in the A range, and can measure currents ranging from to 10-5 A. Recent demands of capillary column gas chromatography have also led to the development of GC electrometers with response times of less than 100 msec. Flame Ionization Detector The FID (1) is the single most important detector in chromatography. Its success can be traced to four basic characteristics: it responds universally to almost all organic compounds; it has relatively high sensitivity; it has a very wide range of linear responses; and it is relatively simple, maintenance free, and capable of providing a stable response over a long period of time. The mechanism of ionization production in the FID is the combustion of organic samples in the high-temperature, chemically active environment of an H 2 /air flame. The FID flame is a classic example of a diffusion flame (2,3). The GC effluent is mixed with H 2 and is propagated through the center of a flame jet structure; air is supplied around the periphery of the jet. The H 2 and air mix by diffusional processes in the flame rather than by premixing before the flame. Diffusion flames are characterized by large spatial variations in both chemical species and temperature. For example, the perimeter of the FID flame is rich in O z ; the flame core is rich in H 2 ; and in between there is a conical region where the H 2 and O 2 mix. In this mixing region the flame temperature is highest, much of the flame chemistry occurs, and much of the flame ionization occurs. The overall chemical reaction occurring in the FID flame is the conversion of H 2 and O 2 gases to water vapor. In the hottest parts of the flame, the temperature is of the order of 2000 C. In addition, the single most important characteristic of the FID flame is that it is a simple method of generating large concentrations of the flame radicals, Η,Ο, and OH. These radical species are very chemically active and, in combination with the high temperature, they are primarily responsible for the complete decomposition of incoming sample compounds. It is because of this complete decomposition that the FID is capable of providing universal response to almost all organic compounds. The main ionization process in the FID flame involves the decomposition of organics to yield the CH radical followed by a chemi-ionization reaction as follows: where e - represents a free electron. The ionization in an FID occurs in a gas-phase process, so equal concentration of positive and negative ions are formed. In a well designed FID, a variation in the polarization voltage from negative to positive values results in a symmetrical curve of ionization current vs. polarization voltage, such as shown in Figure 1. Consequently, some commercially available FIDs collect positive ions, while others collect negative ions with comparable performance specifications. The most common FID configuration consists of a polarized flame jet structure and a cylindrical collector electrode located slightly downstream of the jet. The pattern of electric field linesthat prevails in this configuration is illustrated in Figure 2. The most significant characteristic of this electric field pattern is that the magnitude of the field gets weaker and weaker as the interior of the collector cylinder is penetrated further and further downstream of the jet. This means most of the ionization in an FID is collected at the lowest portion of the collector, where the field is the strongest. One of the most demanding requirements for a given FID design is to maintain linearity of response at very high sample concentrations (4) where the flame changes size. At low sample concentrations, the size of the flame is determined by the flame jet orifice and by the flow rates of H 2, air, and GC effluent. However, as the concentration of organic samples gets larger and larger, the sample itself contributes an additional fuel to the flame, with the result that the flame gets larger with increasing sample concentration. With the FID geometry shown in Figure 2, the larger the FID flame grows, the more it penetrates into the interior of the collector where the electric field gets weaker and weaker with increasing distance from the jet. This means the ionization collection efficiency for large sample concentrations is much more dependent on the magnitude of the Figure 1. Plot of FID response current at positive and negative polarization voltages. 467

3 Journal of Chromatographic Science, Vol. 24, November 1985 polarization voltage than is the collection for small sample concentrations. This effect is illustrated by the data in Figure 3. The sample chromatographed for Figure 3 consisted of three minor hydrocarbon components dissolved in a fourth hydrocarbon solvent. The electrometer sensitivity was adjusted between the solvent elution time and the elution times of the minor compounds to display all four peak maxima. The data of Figure 3 shows the signals from the minor components were insensitive to the change in polarization voltage, but the solvent response increased substantially with increased voltage and with increased penetration of the electric field into the interior of the collector electrode. Most commercially available FIDs have hydrocarbon sensitivities in the range of 10 to 30 mc/g of carbon, and detectivities in the g of carbon/sec range. The outstanding characteristic of the FID is that approximately these same responses are achieved for wide classes of organic compounds regardless of their original molecular structure. The linear range of an FID is often specified as 10 7 or greater, but the FID user should be aware that FID responses at high sample concentrations may be sample-dependent because the sample is contributing an additional fuel to the flame. Also, the use of N 2 rather than helium as the GC carrier gas or capillary make-up gas usually provides the best sensitivity and linearity for most commercially available FIDs (5). Electron Capture Detector electron. The ECD has been especially popular for the specific detection of trace levels of halogenated compounds. As mentioned earlier, a high level of ionization is continuously maintained in the ECD by irradiation of the gas within the detector volume by beta particles emanating from a radioactive surface. The most widely used radioactive material is nickel-63 because it can emit beta particles for long periods of time even when operated at 400 C. The beta irradiation ionizes the detector background gas even in the absence of sample compounds. As a result, an equilibrium concentration of ionization, consisting of positive ions, free electrons, and sometimes negative ions is maintained in the ECD. The ECD functions by sensing changes in the concentration of free electrons within the detector volume. When an electronegative sample is introduced into the detector, it attaches to a free electron to form a negative ion species, producing a net decrease in the detector's electron concentration. Since the concentration of electrons is very large in the absence of sample, and since the concentration of sample compound is usually very small, the ECD usually senses very small changes in electron concentration superimposed on an otherwise very large current of free electrons. Hence, more than any other detector discussed in this article, the ECD is adversely affected by contaminants in the detector gases used, by contamination from excessive GC column or septa bleed, or by contamination from complex sample matrices. Such contamination causes decreased ECD sensitivity and increased baseline instabilities. The ECD (6) is one of the most sensitive detectors available for gas chromatography. It is a specific detector that responds only to compounds with molecular structures conducive to the formation of negative ions by the process of attachment of an Figure 2. Illustration of the pattern of electric field lines in an FID. Figure 3. Chromatograms of 1-μg each amounts of C 1 4, C 1 5, and C 1 6 norma alkanes in an iso-octane solvent for two different magnitudes of FID polanza tion voltage. 468

4 Journal of Chromatographic Science, Vol. 24, November 1986 In order to function properly, an ECD must collect negative polarity ionization in the form of electrons, but must not collect ionization in the form of negative ion species. This is the single most important constraint in designing the geometrical configuration, the gas flow path, and the electrical field pattern of an ECD cell. If a given ECD design does not adequately discriminate between the collection of electrons vs. negative ions, then that ECD design will usually exhibit a non-linear performance characteristic. A primary difference affecting the collection of electrons vs. negative ion species is that electrons move much more quickly than negative ions in a given electric field. As a result, the direction of gas flow through the ECD has a much greater effect on the motion of ion species than on the motion of electrons. Figure 4 illustrates two different geometrical configurations used in commercially available ECD cells. In the configuration described as displaced coaxial cylinders (7), the collector electrode is located entirely outside and upstream of the volume of gas ionized by beta particles from the radioactive material. In this configuration, electrons must move against the gas flow in order to reach the collector, and negative ions tend to be swept away from the collector. In the ECD configuration described as concentric cylinders, the collector electrode is a rod extending through the center of the ionized volume, and collected electrons are moved in a direction transverse to the gas flow. This configuration usually requires a more delicate optimization of electric field magnitude and gas flow in order to prevent negative ions from also reaching the collector. Among ionization detectors, the ECD is distinguished by having the most novel method of electronic operation. Most commercially available ECDs are now operated according to a constant-current, pulse-modulated method of operation (8). Instead of simply connecting the ECD to a dc polarization voltage and to an electrometer, the ECD cell is an integral part of an electronic feedback circuit. Polarization of the ECD cell is accomplished by applying periodic pulses of voltage rather than a constant voltage. The average ionization current that emanates from the ECD cell is electronically compared with a preset reference current, and the frequency of the voltage pulses is automatically adjusted to always maintain an ionization current magnitude equal to the constant reference current. The average ionization current from the ECD cell is proportional to the concentration of free electrons in the cell multiplied by the frequency of the applied voltage pulses. Hence, when there is no sample in the cell, the electron concentration is relatively high and the pulse frequency is low (Figure 5). When an electronegative sample enters the cell, the electron concentration decreases and the pulse frequency increases in order to keep the average ionization current constant. The actual detection signal of a state-of-the-art ECD is, therefore, a change in pulse frequency rather than a change in measured ionization current. This frequency change is electronically transformed to an output signal voltage that is proportional to the concentration of electronegative sample in the ECD cell. From the schematic diagram in Figure 5 of the pulse train that corresponds to the presence of a sample in the ECD cell, it is clear that the maximum possible frequency coincides with the point where the pulses begin to overlap. Hence, in order to obtain dynamic range extending to very large sample concentrations, the pulse width must be very narrow so that the maximum frequency can extend to very large values. Typically, the pulse-modulated method uses pulse widths that are only a fraction of a microsecond in duration. In the absence of sample, the time period between voltage pulses is typically 10 3 to 10 4 times longer than the width of each pulse. Hence, the ECD cell is effectively not polarized for most of the time. This allows the concentration of free electrons to build to its highest possible equilibrium level without the continuous electron removal process that existed in older dc methods of operating ECD cells. Consequently, the constantcurrent, pulse-modulated method of ECD operation provides greater sensitivity and a wider range of linear responses in comparison to dc methods of operation. Most commercially available ECDs are able to detect halogenated and other electronegative compounds at 0.1 to 1.0 pg levels, and provide linear response signals over a range of 10 3 to 10 4 in sample amount. However, the ECD is a non-destructive detector and its response depends strongly on the electronegative character of the sample's molecular structure. Hence, ECD responses can vary over a wide range depending on the compounds being detected. Figure 4. Schematic diagrams of two commonly used ECD cell configurations. Figure 5. Schematic diagram of the polarization voltage pulses applied to the ECD cell in the constant-current, pulse-modulated method of operation. 469

5 JOURNAL OF CHROMATOGRAPHIC SCIENCE, VOL. 24, NOVEMBER Photo Ionization Detector Ionization is produced in the PID by irradiating sample compounds with ultraviolet light. Whether or not a given sample compound is ionized depends on the ionization potential of the sample and the energy of the irradiating photons. Generally, compounds are ionized if their ionization potentials are less than or equal to the photon energy. Since both organic and inorganic samples can satisfy this criterion, the PID has applicability to a much wider range of substances than a detector like the FID. The PID also provides the possibility of substantially varying the selectivity of detector response by varying the energy of the photons. Hence, commercially available PIDs can be operated in either universal or selective modes of detection. As in the FID, the ionization process in the PID occurs in the gas phase. Consequently, equal concentrations of positive and negative ionized species are formed. Some PIDs collect positive ionization while others collect negative ionization. Since stray UV photons can also eject photoelectrons from metallic surfaces within the ionization chamber, the collection of positive ionization generally presents less of a design problem in suppressing collection of these extraneous electrons than does negative ionization collection. The key component of the PID is the source of ultraviolet light. In recent years, the most widely used PID designs have employed a sealed ultraviolet light source coupled to the ionization chamber (9). This arrangement has allowed the operating parameters of the ionization chamber and the light source to be independently optimized, and it has especially facilitated changes in PID selectively by interchange of different ultraviolet light sources. The two important characteristics of the ultraviolet sources are photon energy and photon intensity. While the photon energy determines PID selectively, the photon intensity determines the PID sensitivity. A large body of PID data have been obtained (10,11) with a 10.2 ev light source because it provides a relatively high proton intensity and a photon energy sufficient to ionize many organic and inorganic samples. This 10.2 ev source, therefore, provides a universal mode of detection for the PID, which has excellent sensitivity. Other light sources that have been used in the PID are 11.7 ev, 9.5 ev, and 8.3 ev lamps. The 11.7 ev lamp extends the universal response of the PID to additional compounds, while the 9.5 and 8.3 ev lamps provide selective PID responses to a more limited number of compounds. A particular advantage of the PID is that it is a universal detector that requires only a single inert detector gas, such as helium or N 2, unlike an FID, which requires H 2 and air to support a flame. However, response factors for organic compounds generally exhibit a greater compound-to-compound variance for the PID (10) than for an FID. This difference can be attributed to the complete decomposition of organics in the H 2 /air flame of an FID vs. the photo-ionization of intact organic molecules in the PID. In general, any GC detector which functions by decomposing samples to their elemental constituents will usually produce more uniform detector responses that are independent of the original molecular structure of the sample. The fact that a PID is non-destructive to sample compounds is another advantage that allows the PID to be used in series combinations with other GC detectors (11,12). The PID has been combined with the FID, the ECD, the nitrogen-phosphorus detector (NPD), and the Hall electrolytic conductivity detector (HECD). Such series combinations can be very valuable because they simultaneously produce two different types of detector responses for each sample eluting from the GC column One disadvantage of the PID in comparison to an FID is that the PID is a concentration-sensitive detector, whereas the FlD is a mass flow-rate sensitive detector. Consequently, a design constraint for the PID is the construction of an ionization chamber with as small an internal volume as possible in order to obtain high detection sensitivity with low gas flows without adverse tailing of chromatographic peaks. Commercially avail able PIDs now have internal volumes less than 50 μl. The PID is an especially good detector for the measurement of aromatic hydrocarbons. For these compounds the PID is approximately 50 times more sensitive than an FID, and has a 10 7 linear dynamic range. Other important PID applications have included the detection of vinyl chloride, organics in drinking water, polyaromatic hydrocarbons, sulfur compounds, tetraethyllead in work atmospheres, and many other compounds (11) Thermionic Ionization Detector The TID is based upon a principle of surface ionization in which gas phase sample compounds are ionized by their interaction with a heated, appropriately activated solid surface. The most widely known use of the TID is for the specific detection of compounds containing nitrogen or phosphorus atoms. However, as is true for the PID, the basic TID equipment can be used to generate many different modes of detection through simple interchanges in the type of solid surface or its operating conditions. The different modes of TID detection have been most extensively developed according to principles of negative ion chemistry (13). All modern TIDs employ a solid surface composed of a ceramic or glass matrix molded onto an electical heater wire. An alkali metal compound is a common additive to the ceramic or glass material in order to provide a surface with an appropriately lowered electronic work function. Such a surface is biased at a negative voltage with respect to a nearby collector electrode, and is heated to a surface temperature in the range of 400 to 800 C, with the actual operating temperature dependent upon the mode of detection. Under such conditons, the solid surface functions as a reservoir of negative charge, which can be transferred from the surface to gas-phase sample compounds that are electronegative in character. Such a thermionic emission process produces gas-phase negative ion species, which are collected and which constitute the TID detection signal. The ionization produced in the TID is the result of a gassolid interaction rather than a gas-gas interaction, so there is no condition that equal concentrations of positive and negative ion species are formed, as in the case of the FID and PID. Instead, a plot of measured TID ionization current vs. polarization voltage is quite asymmetrical with respect to negative and positive polarities, as is shown in Figure 6. There are three key parameters in the TID that determine its response: the electronic work function of the thermionic surface, which is determined by the chemical composition of that surface; the operating temperature of the thermionic surface, and the chemcial composition of the gas environment immdiately surrounding the thermionic surface. Different modes of TID operation are realized by variations in one or more of these key parameters. 470

6 journal of Chromatographic Science, Vol. 24, November 1986 The simplest mode of TID operation is obtained when the thermionic emission surface is operated in an inert gas like N 2. In this mode, sample compounds eluting from the GC column interact directly with the hot thermionic surface with no intervening gas-phase chemistry. This mode of detection provides very specific ionization of sample compounds that contain electronegative functional groups, such as NO 2, multiple halogen atoms, SH, and others. Optimum sensitivity and specificity for this basic TID mode require a thermionic surface with a very low work function, and this has been achieved by embedding a high concentration of a cesium compound in a ceramic matrix (13). This mode of detection has produced sub-picogram detectivity for some nitro-compounds with an extraordinary specificity factor of 10 8 vs. alkane hydrocarbons. However, the response is very dependent on the detailed electronegative character of the sample's molecular structure, so compound-to-compound response differences can be substantial. Other modes of TID response are obtained by systematic introduction of detector gases, which are more reactive. For example, when air or O 2 are used as the detector gas environment, the presence of the electronegative O 2 molecule alters the transfer of negative charge in a manner that enhances the specific ionization of halogenated compounds. With N 2, air, or O 2 gas environments, the TID responses are most specific when the thermionic surface temperature is relatively low (i.e. 400 to 500 C). As the surface temperature is increased further, the TID begins responding to wider classes of compounds in a manner resembling what occurs in the PID as photon energy increases. It is interesting that even at high surface temperatures, the TID produces negligible ionization for aromatic hydrocarbons. This is in sharp contrast to the PID, which is an exceptional detector for aromatics. The well-known nitrogen/phosphorus-specific mode of TID response is obtained with a detector gas environment comprising air and a small flow of H 2. The nitrogen/phosphorus response turns on when the thermionic emission surface is hot enough to ignite a dissociated H 2 /air gaseous boundary layer in the immediate vicinity of the hot surface. The active gas-phase chemistry in this boundary layer decomposes sample compounds, and nitrogen or phosphorus compounds preferentially form electronegative decomposition products that are selectively ionized by the thermionic surface. A key operating condition for this nitrogen/phosphorus-specifc mode is that the H 2 supplied is not sufficient to support a self-sustaining flame if the heating current to the thermionic surface is turned off. Also, the nitrogen/phosphorus-specific mode requires a thermionic surface of higher work function (i.e., of less alkali additive) than the surface used in the N 2 or air gas environments. Commercially available NPDs typically provide detectivities in the 1- to 10-pg range and linear responses over a range of 10 4 to Another distinct mode of TID operation is realized by preceeding the TID with a self-sustaining H 2 /air flame and by separating the TID and the flame sufficiently so that any ionization produced in the flame is dissipated before it reaches the TID. The TID then re-ionizes the neutral combustion products of the flame. This mode of detection provides specificity for nitrogen or halogen compounds, and the pre-combustion of samples in the flame promotes uniform responses regardless of the original molecular structure of the sample compound. The surface ionization principle of the TID and the gas phase ionization principle of the FID have been combined in a universal detection mode called the catalytic flame ionization detector (13). This is accomplished by inserting a thermionic surface directly into the center of a self-sustaining H 2 /air flame. In this mode of detection, the optimum thermionic surface has a high work function in order to avoid excessive background currents. The gas-phase chemi-ionization processes of the flame provide FID-type ionization of organic compounds, and ionization processes on the thermionic surface provide a secondary means of ionizing compounds containing heteroatoms like halogens and phosphorus. By judiciously adjusting the electrical heating of the thermionic surface, the response factors for many heteroatom compounds can be made comparable with the response factors for hydrocarbons. Summary GC detectors function according to very specific physical and chemical principles, and a basic understanding of these principles provides a logical basis to explain why particular detectors behave the way they do. This article has stressed the mechanisms active in the FID, ECD, PID, and TID and has attempted to show how these mechanisms relate to the proper design and operation of these detectors. The multiple-mode aspects of the PID and TID especially demonstrate the significance of following a course of detector development based on sound scientific and engineering principles. References Figure 6. Plot of TID response and background current at positive and negative polarization voltages for the nitrogen/phosphorus mode of detection. 1. I.G. McWilliam and R.A. Dewar. Nature 181: 760 (1958). 2. I. Glassman. Combustion Academic Press, New York, A.G. Gaydon and H.G. Wolfhard. Flames, Fourth Edition. Chapman and Hall, London E.R. Colson. Anal. Chem. 58: 337 (1986). 5. AT. Blades. J. Chromatogr. Sci. 14: 45 (1976). 6. J.E. Lovelock. Anal. Chem. 33: 163 (1961). 7. P.L. Patterson. J. Chromatogr. 134: 25 (1977). 471

7 Journal of Chromatographic Science, Vol. 24, November R.J. Maggs, P.L. Joynes, A.L. Davies, and J.E. Lovelock. Anal. Chem. 43: 1966 (1971). 9. J.N. Driscoll and F.F. Spaziani. Res./Dev. 27: 50 (1976). 10. M.L. Langhorst. J. Chromatogr. Sci. 19: 98 (1981). 11. J.N. Driscoll. J. Chromatogr. Sci. 23: 488 (1985). 12. W.A. McKinley, R.J. Anderson, and P.W. Thiede. Presented at the Pittsburgh Conference on Analytical Chemistry and Applie Spectroscopy, Atlantic City, NJ, March 1982, Abstract # P.L. Patterson. J. Chromatogr. Sci. 24: 41 (1986). Manuscript received August 11, 1986; revision received September 16,