274 DETECTORS AND DATA HANDLING 5.2.9 Practical Operating Hints The following guidelines apply to all detectors. Refer to sections on specific detectors for information pertinent to particular detectors. 1. To avoid temperature fluctuations in the detector (and in the oven) situate the gas chromatograph away from drafts and heating and air conditioning vents. Locations near poorly insulated exterior walls or in direct sunlight should be avoided. 2. Greater thermal stability is left on. 3. Oxygen scrubbers, moisture traps, and the like are worthwhile insurance against a contaminated bottle of gas. However, they must be replaced when exhausted or else they can add impurities to the carrier gas instead of removing them. 4. Always check for leaks in the gas lines. While carrier gas is flowing out, air is diffusing in. chieved when the instrument covers are 5. Never condition a column while it is connected to the detector. 6. Check the detector exit tubes for condensation and plugging after running high-concentration or “dirty” samples and clean whenever necessary. 7. Bring the detector up to operating temperature before turning on the oven when starting up the gas chromatograph. This will prevent condensation in the detector, since the oven heats up more rapidly than the detector. 8. For a stable baseline in the morning, leave the detector at operating temperature overnight. 5.3 FLAME IONIZATION DETECTOR 5.3.1 Background The flame ionization detector (FID) was introduced in 1958 by McWilliam and Dewar (4) in Australia and by Harley et al. (5) in South Africa. It has become the most commonly used detector in GC for several reasons: 1. It responds with high sensitivity to virtually all organic compounds. 2. Modest changes in flow, pressure, or temperature have a minimal effect on its response characteristics. 3. In normal operating mode it does not respond to common carrier gas impurities such as water or carbon dioxide. 4. When properly installed, it has a stable baseline. . It has a wide linear range, about 10°. wFLAME IONIZATION DETECTOR 275 Detector Tower- Ins Collector Electrode Insulator Flame. Ignitor and Jet Polarizing Electrode Insulator Air Diffuser Heated ‘Air 2 Makeup Gas Capillary Column FIGURE 5.5 Schematic diagram of a flame ionization detector. 6. It is easy to adjust. The detector consists of a small hydrogen-air diffusion flame burning at the end of a jet (Figure 5.5). When organic compounds are introduced into the flame from the column effluent, electrically charged species ate formed. These are collected at an electrode and produce an increase in current Proportional to the amount of carbon in the flame. The resulting current is amplified by an electrometer. Although this detector has been widely used for over thirty years, the mechanism of its response is still not completely understood and is still under investigation. 5.3.2 Flame Chemistry The reaction that produces ionization in a flame detector was confirmed in the early 1960s. Most of this work was done by combustion researchers who obtained the direct evidence from mass spectroscopic examination of the interior of flames. Summaries of flame ionization processes have been published by Miller (6) and also Bocek and Janak (7). The mechanisms of the diffusion flame of the flame ionization detector begin at the tip of the jet (Figure 5.6). The mixture of carrier gas, hydrogen, and makeup gas (if used) flows from the restriction of the jet and expands outward. The air flows at a lower velocity around the outside of the jet gases. As the jet flow approaches the reaction zone (dotted line in Figure276 DETECTORS AND DATA HANDLING AFTERGLOW ; REGION INTERDIFFUSION ne ‘ ACROSS REACTION. : ZONE PREHEATING, CRACKING AND STRIPPING 4 AIR H2 &N2 FIGURE 5.6 Schematic diagram of the basic flame processes in the FID. 5.6), it is preheated by back diffusion of the heat energy produced there. Organic materials eluting from the column undergo degradation reactions in this hydrogen-rich region of the flame and form a group of single carbon species. When the two gas flows mix together at the reaction zone and oxygen becomes available, the following chemi-ionization reaction occurs: CH +0-CHO* +e (6.1) The unstable CHO* ions react rapidly with the water produced in the flame to generate hydroxonium ions according to the reaction CHO’ +H,0>H,0* + CO (5.2) These ions and their polymerized form (H,O),H" are the primary positive charge carrying species. This extremely important reaction occurs once for every 100,000 carbon atoms introduced in the flame. It is essentially a quantitative counter of carbon atoms being burned. Therefore, the response of the flame ionization detector is proportional to the number of carbon atoms, rather than the compound weight or moles.FLAME IONIZATION DETECTOR 277 The process by which the organic compounds are broken down into single carbon species is less well understood and several schemes have been proposed (8-11), There must be a good balance among all the flows and diffusions for the flame processes to be efficient. Flame turbulence, whether caused by too high a flow or the wrong diameter jet, must be avoided for proper operation of the detector. 5.3.3 FID Design det. The type of column and the analysis to be performed dictate the size of jet to be used. For packed columns, a jet with a 0.018-in. internal diameter is standard. Capillary columns require maximum detector sensitivity. A jet with an internal diameter of 0.011 in. is used for capillary work, since it produces almost double the sensitivity of the 0.018-in.-i.d. jet (12). If this jet is used with packed columns, however, flame-out and clogging problems may occur. Another packed-column jet with a 0.030-in.-i.d. is used with 0.53-mm capillary columns so that the column can be inserted into the jet to within a few millimeters of the flame. This has been shown to improve peak shape by preventing the loss of peak integrity due to adsorption or catalytic decomposition which can occur when some eluted materials contact hot metal surfaces (13). If the fused silica column is allowed to extend beyond the tip of the jet, the stationary phase will pyrolyze and the polyimide coating of the column will decompose, resulting in noise or spiking. Since the diffusion of the gases at the tip of the jet is so central, the detailed shape of the jet affects detector operation, The tip should be smooth. It should be kept free from solid support particles from the column, pyrolysis products of samples, SiO, deposits from the liquid phase, and other contaminents. Gas Flowrates. The maximum sensitivity of the flame ionization detector is achieved at a particular ratio of carrier gas (or carrier plus makeup gas) to hydrogen (14). The maximum and the ratio depend on the particular gas. Detector response is much greater with a nitrogen-hydrogen flame than with a helium-hydrogen flame. Therefore, sensitivity can be dramatically increased with capillary columns by using nitrogen rather than helium as the makeup gas. Since variations in detector design may require different gas flowrates, the instrument manual is the best reference for setting these flows. In the absence of this information, 30 mL/min of both hydrogen and cartier gas is a good starting point. Because not all the air enters the reactive part of the flame in the diffusion flame, it is necessary to use several times the stoichiometric amount of oxygen. Air flows of 300 to 500 mL/min are typical for most detectors. It is important to have a uniform and laminar flow along all sides278 DETECTORS AND DATA HANDLING of the jet. For this reason, many instruments introduce the air through a porous diffuser located well below the tip of the jet. As with the hydrogen and carrier flows, the instrument manual should be consulted when setting the air flow. Exhaust Flow. The exhaust flow system is usually ignored until problems occur. It removes the gases, heat, soot, and silica from the detector. Soot is formed from the incomplete combustion of chlorinated solvents such as chloroform and methylene chloride in the flame. It can build up in the tower and cause spikes as particles fall back into the flame. Silica, the product of the combustion of silicone-coated column bleed, condenses on the interior of the detector as a fine, white powder. This gives an insulating coating on which an electric charge can build up, changing the electric fields in the detector. Water can also condense if the detector is too cool, causing electrical shorts. Most instruments have some mechanism for isolating the flame from air currents in the vicinity of the instrument. Ambient air should be prevented from backdiffusing into the flame chamber to avoid the detection of solvent vapors or cigarette smoke in the laboratory air. Thermal Control. There are several considerations in setting the tempera- ture of the flame detector. The block should be heated above column temperature to prevent sample components from condensing in the carrier- gas transfer line. It should also be kept above 100°C to prevent the condensation of water. The temperature of the body of the detector should be kept stable, because it has a slight effect on the detection mechanism. It is more important not to let the detector get too hot. If parts of the detector are heated close to the threshold for significant thermionic ionization, a large sample can increase the temperature and cause a positive detector response, even with samples that normally have negligible response. Under these conditions, water, which reduces the flame temperature, might give a negative peak. Another reason not to overheat the detector is to prevent electrical leakage across the insulators. This can cause instability in the detector current. Jon Collection. The geometry of the collector electrode and how it affects the distribution of the electrical field and the flow pattern of the gas stream has been extensively investigated. Most collectors are cylindrical and symmetrical and have a relatively large surface area for ion collection. The collector is placed near the top of the flame so that ion collection takes place near the base where the electric field is greatest. When very large samples are burned, the flame can increase drastically in size and change from a diffusion flame to a hydrocarbon-like flame. lon generation now takes placeFLAME IONIZATION DETECTOR 279 higher in the collector where the electric field is weaker or completely above it. This results in nonlinearity of response and detector saturation. 5.3.4 Response Factors The flame ionization detector is mass flow sensitive. Its area response for a compound does not change with small changes in carrier flow like those in temperature-programmed operation. The units for its response factors are coulombs/gram of carbon. It has a high sensitivity to organic carbon containing compounds, most with a response of 0.015 coulombs/g C. The detector is noted for its 10” linear range (Figure 5.4). However, its response is adversely affected by the presence of heteroatoms such as O, $, and halogens. The concept of an effective carbon number, ECN, was introduced to estimate the relative response for any compound. Particular groups of atoms are given a value relative to a reference material, usually an n-paraffin, for which the ECN is simply its carbon number. The set of parameters used to calculate effective carbon numbers is given in Table 5.3 (8, 15, 16). The response for a compound is the total of all its atoms and groups. For example, n-propanol would have an ECN of 2.4, having three aliphatic carbons (3X 1) and one primary alcohol oxygen (—0.6). Likewise, butyric acid would have an ECN of 3.0, having three aliphatic carbons (3 x 1) and one carboxyl group (0). Several workers have tabulated large numbers of experimental relative response factors (17-20). One obvious use of the ECN is in determining the relative response factors of compounds that cannot be secured in sufficient purity to de- TABLE 5.3 Contributions to Effective Carbon Number Effective Carbon Atom Type No. Contribution c Aliphatic 1.0 c Aromatic 10 c Olefinic 0.95 c Acetylenic 1.30 c Carbonyl 0.0 c Nitrile 03 ° Ether -1.0 ° Primary alcohol 0.6 ° Secondary alcohol 0.75 oO Tertiary alcohol, esters 0.25 a Two or more on single aliphatic C 0.12 each cl On olefinic C +0.05 N Inamines Similar to O in corresponding alcohols280 DETECTORS AND DATA HANDLING termine them experimentally. Discrepancies between experimental and theoretical response factors are a good indication of adsorption or decompo- sition of analytes in the chromatographic system, especially in the absence of the usual signs of poor peak shape and tailing. Response factor determi- nation, however, can be seriously impaired by improperly set flow parame- ters, water in the air supply to the detector, and high column bleed. Table 5.4 lists substances that have little or no response in the flame ionization detector. 5.3.5 Modifications of the FID The FID can be modified to respond to inorganic gases by operating it in a hydrogen-rich mode with oxygen to support combustion. This can be accomplished with most commercial FIDs by introducing the oxygen with the carrier gas and using the normal air inlet to introduce the hydrogen fuc! directly into the detector. When modified like this the FID is known as a hydrogen-atmosphere flame-ionization detector (HAFID). Detection limits for some gases are as follows (21): CH, 2x 107" g/sec co, 4x 10-* gisec 4x10 '’g/sec 2x 107" g/sec 5x 107" g/sec 4x 107" g/sec 7x10 ° g/sec 2x 107° g/sec 4x 1077 g/sec 5x10 * g/sec Doping the hydrogen atmosphere with hydrides has enabled enhanced detection of organometallic compounds (22) containing iron, tin, lead, molybdenum, and tungsten. Phosphine was more effective than methane, silane, or germane. Silicon-containing compounds have also been found to give enhanced response in HAFID mode, which on doping the flame with ferrocene, produced a silicon-to-carbon selectivity of 10° (23). TABLE 5.4 Substances With Little or No Response in the FID He N; NO Col, Ar oO, N,0 SiCl, Kr co NO, CH,SiCI, Ne co, N,O, SiF, Xe H,0 NH, SiHC|,THERMAL CONDUCTIVITY DETECTOR 281 5.3.6 Practical Operating Hints 1. One of the major sources of problems with the FID is an incorrect adjustment of the flame. Both the size of the flame and the ratio of its gases are important. When changing the jet, carrier-gas flow, or sample size, always consider the effect this will have on the flame and make adjustments, if necessary. 2. The second source of problems with the FID is contamination. This can be due to impure gases, water shorting out the detector, or soot or silica deposits in the chimney. 3. For maximum sensitivity with capillary columns, use the proper size jet and use nitrogen as the makeup gas. 5.4 THERMAL CONDUCTIVITY DETECTOR 5.4.1. Background ‘The thermal conductivity detector (TCD) is one of the most commonly used detectors in gas chromatography. It measures changes in the thermal conductivity of the carrier gas caused by the presence of eluted substances. Chromatographers began using thermal conductivity because devices for measuring this property had been in use since the 1880s. The major uses of these devices included gas analysis and process control. So it was natural to use thermal conductivity in early experiments with gas-solid chromatog- raphy. The TCD was technically mature by the late 1950s. There was always a general dissatisfaction with the problem of sensitivity despite many attempts to improve it. The more sensitive ionization detectors were used in its place for trace analysis and with capillary columns. The advantage of the TCD lies in the detection of gases such as CS,, COS, H,S, SO), CO, NO, NO), and CO, in gas-solid chromatographic analyses on packed columns. The introduction of the porous layer open tubular (PLOT) columns, however, brought gas-solid chromatography to capillary columns and established a market for the microflow cell TCD. 5.4.2 Principles of Operation The property of a material transmitting heat when subjected to a tempera- ture difference is known as thermal conductivity A. It can be defined in terms of the heat flow Q through a unit thickness of material x of a unit cross section A when the temperatures on each side differ by unity. From Figure 5.7, the heat flow is o- A(T, — T3)A : (5.3)282 = DETECTORS AND DATA HANDLING AREA= A FIGURE 5.7 Thermal conductivity between two planar x surfaces of temperatures T, and T, and cross-sectional area A, separated by a distance x. T2 Th If thermal conductivity is measured in a cylindrical geometry, it is much casier to avoid heat losses other than conduction through the gas. This is accomplished by replacing one planar surface with a cylindrical cell of temperature T, and the other planar surface with a wire filament running through the center of the cell with a temperature of 7;. Replacing the dimensions of the planar surfaces A/x with a geometry factor G in the equation above gives the equation for heat transfer in a cylindrical geome- try: Q=G(T,—T.)A G.4) The thermal conductivity can be determined by supplying heat at a known rate Q and measuring the temperatures of the center conductor wire and the outside cell wall. The center conductor can be a wire of known electrical resistance, R, ohms. The heat flow can then be supplied by an electric current, / amperes: GR: Q=5 (5) where J is Joule’s constant (4.183 W/cal-sec). If a wire whose resistance R, is temperature dependent is used, the temperature of the wire can be measured: R,=R? (1+ a7, (5.6) where a is the temperature coefficient of resistance. The wire now doubles as a heating element and as a resistance thermometer. Use the voltage across the wire V to calculate the resistance (at temperature) V/J and substitute in the above equation to find the temperature T,.THERMAL CONDUCTIVITY DETECTOR 283. The thermal conductivity can be calculated from the temperatures, heat flow, and geometry. However, in the thermal conductivity detector for GC, the absolute value of A for the column effluent is not important. The small change in A when eluents are present in the carrier gas is important because it changes T,, which can be detected by the change in the resistance: AR, =a? AT, (5.7) and (5.8) These equations can be combined to give ee aR§(T,—T.) Aa a) Note that the sensitivity for changes in A is proportional to the temperature difference across the cell. However, there is a limit to the temperature that the filament can be run. For most commercial TC detectors that use a hot wire, a practical upper limit is 450°C for continuous operation and 500°C for short periods of time. This is a result of oxidation of the filament with trace amounts of oxygen in the carrier gas. A variety of other reactions can occur with the sampie. At the maximum filament temperature, sensitivity can be increased by lowering the detector temperature T.. In practice, the best sensitivity can be obtained by setting the detector block temperature as low as possible without condensing eluted substances in the cell and setting the filament temperature as high as possible without burning it out. Besides the thermal conductivity of the gas, there are four other processes that contribute to the loss of heat from the filament: L. Thermal radiation 2. Thermal conductivity of leads and connections (end losses) 3. Forced convection of the gas 4. Free convection and diffusion of the gas Thermal radiation is dependent on the surface area of the element, on its temperature, and on its material quality. Using the type of detector shown in Figure 5.8 operating under the conditions of Table 5.5, the power loss through thermal radiation can be estimated as 15mW. At the highest filament temperature, radiation can transfer a few percent of the power. Although the temperature of the filament is constant over its length, the ends must drop to the cell body temperature. This heat loss is proportional to the difference in temperature between the filament and the body of the cell. For the case cited, this amounts to 45 mW. For the small-diameter wire284 DETECTORS AND DATA HANDLING TUNGSTEN-RHENIU! FILAMENT: LENGTH 19 CM COIL DIA. 0.012 WIRE DIA. 0.002 WIRE LENGTH (extended) 7.0 FIGURE 5.8 Typical geometries of thermal conductivity detector cells. The feedthroughs supporting the axial filament are insulated from the stainless-steel body by alumina ceramic. All dimensions are in centimeters. TABLE 5.5 Typical Operating Conditions for Thermal Conductivity Cells ‘Temperature Body of detector 150°C Filaments 350°C Filaments Material Tungsten—rhenium Temperature coefficient, a 0.0033/°C Resistance, R, (at 0°C) 25.0 Resistance (at 350°C) 550 Electrical (for four-element bridge) Current 03A Voltage 16.5V Power (for each filament) 4.95 W Current 015A Voltage 8.25V Power 1.24W Carrier gas Helium Flowrate 1.0 mL/see ‘Thermal conductivity, A (at 150°C) 4.4 x 107* cal/sec-om-°CTHERMAL CONDUCTIVITY DETECTOR 285 and high-conductivity carrier gases used today, this heat loss mechanism can usually be neglected. The heat loss due to the forced convection of the gas is a product of the mass flowrate of the gas, the specific heat of the gas, and the difference in temperature between the gas flowing out of the cell and the detector temperature 7. For the geometry in the example, the gas temperature exceeds the block temperature by only 2.4% of the 200°C difference across the cell. Assuming uniform axial flow, this gives an approximate heat loss of 7TmW. The larger the detector cell, the more important the free-convection heat transfer becomes. In modern detector cells this effect is negligible. Table 5.6 summarizes all the heat-transfer terms considered. All the effects, except for the thermal conductivity of the gas, contribute only a few percent to the heat transfer. However, if nitrogen were used as the carrier gas, the TC term would be six times lower, and the other terms would be comparably more important. The mass flow term is the cause of the flow dependence of the TC cell and thus contributes to the noise and drift of the detector. 5.4.3 TCD Design Most general-purpose TC detector cells are constructed similar to the one in Figure 5.8. One common variation of this design is to support the filament on posts, both of which are mounted on the same face of the cavity. To obtain higher resistances, coiled filaments are used instead of straight wire. Platinum, platinum alloys, tungsten, tungsten alloys, and nickel can be used for the filament. The choice of filament material depends on mechanical strength and chemical inertness since the resistivity and a of these materials are similar. The thermal conductivity cell is extremely sensitive to fluctuations of physical variables. Some of these can be canceled out if two cells are used (Figure 5.9), one to detect the sample and the other to serve as a reference. A bridge circuit is used to balance the resistance R, of the sample cell TABLE 5.6 Summary of Heat Transfer Effects With Thermal Conductivity Detectors Effect Heat Transfer (mW) Thermal conductivity (G = 3.08) 1130 Radiation 15 End losses 45 Mass flow a Free convection Negligible Total (measured) 1240 Note. Cell of Figure 5.8 and Table 5.5.286 © DETECTORS AND DATA HANDLING CURRENT | —= RL VOLTAGE MEASUREMENT FIGURE 5.9 Bridge circuit used in a two-cell detector, The reference cell Ry ve compensates for the drift in the ana- ANALYTICAL lytical cell R, due to flow temperature COLUMN fluctuations. EFFLUENT REFERENCE FLOW against the resistance R, of the reference cell. Two conventional resistors, R, and R,, comprise the remainder of the bridge. When no substance is being eluted, all the resistors have the same value and there is no voltage difference to be measured. However, when a sample elutes from the column, the resistance R, increases, and a voltage difference is measured. If a change in room temperature affects the resistance of the analytical cell, it will also affect the resistance of the reference cell by nearly the same amount. No significant change in voltage would be measured. This configu- ration can compensate for column bleed by setting the reference flow equal to the analytical flow and passing the reference flow through an identical column. This is particularly useful with temperature programming. One serious problem that the designers of this type of detector had to solve was the effect of temperature on the conventional resistors. Either resistors with very low temperature coefficient could be used, or the resistors could be temperature controlled. Figure 5.10 shows another configuration which replaces the conventional resistors with two more filament cells. The effluent from the analytical column is split into two matched cells of resistance R,. The effluent from the reference column is sent to two matched reference cells of resistance R,. In some detectors the matched cells are simply two filaments mounted in the same cell cavity. One of the benefits of this approach is that the response factor is doubled since two cells are contributing to the change in signal. Two important parts of the TCD electronics are not shown in the figures. Variable resistors must be connected in the bridge to null the output voltage before a chromatogram is run. This is necessary because the four cells of the detector rarely match. These resistors are controlled by the “fine” and “coarse” balance adjustments on the front panel of the detector. If a recorder is used instead of an integrator, the attenuator switch must be used to reduce the response to keep the larger peaks on scale.THERMAL CONDUCTIVITY DETECTOR 287 i— R3 VS 7 ANALYTICAL REFERENCE ‘COLUMN FLOW EFFLUENT FIGURE 5.10 Bridge circuit used in a four-cell detector. This approach gives twice the response of that shown in Figure 5.9. The electrical requirements for the TCD are simpler than those for almost any other gas chromatographic detector. The mechanical require- ments, particularly thermal control, are, by contrast, unusually demanding. The cells of the TCD are mounted closely together and embedded in metal, with the entire assembly meticulously insulated. It is so important to control the temperature of the detector body that this circuit can often provide better thermal stability than that of the chromatographic oven. The oven should not influence the detector temperature. The detector insulation should prevent heat transfer by thermal conduction. But the carrier gas should also not transfer heat to the detector. If it does, variations in the gas flow can cause noise or drift in the detector. Of course, the filaments of the detector must match electrically; but they must also match mechanically because of the effect on the geometry factor. Manufacturers commonly select not only matched sets of filaments, but also cells matched for geometry factor. Another stringent requirement is that the cells be gastight, even at high temperatures. Any diffusion of air into the cell will give drift and noise and burn up the filament. Electrical leakage from the filaments to the detector body can cause noise. High-density ceramics have replaced glass as insulators for the filament mounts to prevent this problem. Another configuration of the TCD is known as a modulated detector. It has only one filament cell. A valve alternates the flow of the analytical column effluent with the reference column effluent. The flow of gases in this cell is illustrated in Figure 5.11. The flows are switched at a rate of ten times per second, faster than any changes due to thermal fluctuations. This design facilitates ease of startup and reduces baseline drift during temperatureAnalytical Flow ..,, Filament, Reference Flow Ve Modulator Flow In ¥¥® vaiye Analytical Mode Analytical Flow) .....Filament,._ | Reterence Flow ‘Channel’: Valve Modulator Flow in Reference Mode V =[G(Tr —Tc)(Aa—As)110 Hz Intermediate Signal (10Hz) Modulated Detector. FIGURE 5.11 Diagram of a commercial TCD. Modulated thermal conductivity detector that uses only one filament cell. The analytical and reference gas flows are switched at a rate of 10 times/sec. This design results in a detector with very low drift. This assembly is further insulated on all sides and mounted in a metal enclosure. 288THERMAL CONDUCTIVITY DETECTOR 289 programming. The cell volume is on the order of SmL, making it suitable for capillary systems. Another variation in the thermal conductivity detector is the use of thermistors instead of filaments. These metal oxide beads used as tem- perature-sensitive resistors have been used in detectors since the 1950s. One of the advantages of thermistors is that their temperature coefficient is larger by an order of magnitude, although it is negative. Another advantage is that their resistance R” is large, usually thousands of ohms. This makes their response factor much larger than that of a hot wire detector. They are also small—as little as 0.25mm in diameter—allowing cells to be made with volumes as small as 5 x 10°* mL. Because the thermistor is a metal oxide glass, it is almost inert to oxidizing conditions. However, thermistors are quite sensitive to reducing conditions. To overcome this problem, thermis- tors with an inert coating have been developed. The major problem with a thermistor detector is that the sensitivity worsens rapidly as the detector temperature is raised above 50°C. At typical operating temperatures for liquid samples, a hot-wire detector is actually more sensitive. The operating conditions for the thermistor detector are more difficult to set. These detectors are used mostly for capillary columns, which require fast response and low detector volumes, and gaseous samples that can be run at low temperatures. The thermal conductivity detector is nearly always operated with helium or hydrogen as the carrier gas. Both these gases have a thermal conductivity much higher than virtually everything else. All other carrier gases share several serious problems: 1. Some compounds give positive peaks, while others give negative peaks. 2. Response factors are dependent on temperature in unexpected ways, and linearity is often poor, 3. Minimum detectable levels are usually an order of magnitude lower than with helium or hydrogen. 4. In some cases W-shaped peaks are observed because the thermal conductivity changes with sample concentration. 5.4.4 Response Factors When a rather large peak enters the detector certain changes take place. Using the detector of Figure 5.8 and the conditions of Table 5.5, these changes are summarized in Table 5.7. The 0.1 mg peak with a width of 10sec causes a change of voltage on the recorder of 140 mV for the peak height. The response factor is the number of millivolts for the concentration of sample in milligrams per milliliter at the detector. It is sometimes called290 DETECTORS AND DATA HANDLING TABLE 5.7 Changes in Typical TC Detector (see Table 5.5) Due to Sample Sample concentration at peak maximum 0.02 mg/mL. Thermal conductivity 4.4-4.14 x 10 * cal/sec-cm-°C Filament temperature 350-362°C Filament resistance 55-56. Voltage across filament 8.25-8.39V Recorder response 140 mv Detector response factor 7000 mV — mL/mg the DPS number, after Dimbat, Porter, and Stross (24), who first proposed it. One of the best features of thermal conductivity detectors used with helium carrier gas is the ease of quantitative analysis. Relative response factors, using sample weight, are independent of the following: Type of detector (filament or thermistor) Cell and sensor temperatures Sample concentration Helium flowrate YR Yeye Detector current In a series of homologous compounds, the relative response factors change only slightly. The first systematic study of TCD responses in helium was done by Rosie and Grob and is summarized in Reference 25. A tule of thumb is that all compounds have a weight response close to that of isooctane. The three exceptions to this are that compounds con- taining heavy metal atoms have unusually low response factors; halogenated compounds tend to have low response factors; and very light compounds, with molecular weights below 35, tend to have high response factors. Of the 171 compounds in Reference 25 (excluding the three types of exceptions) for which there are relative response factors, 88% are within 20% of the response of isooctane and 96% are within 30%. Response factors have also been tabulated for hydrogen (26) and nitrogen (27) carrier gas. 5.4.5 Practical Operating Hints 1. Keep carrier gas flowing whenever the filament current is on. If air is allowed to diffuse into the detector, the filaments will literally burn up. Some instruments are equipped with sensors to turn off the filament current if the carrier flow is interrupted. 2. Turn off the filament current when performing any maintenance that might introduce air into the carrier line, e.g., when changing columns, septa, or gas cylinders.ELECTRON CAPTURE DETECTOR 291 3. A permanent shift in the baseline following a large peak may be caused by a change in the resistance of the sample filament due to an oxidizing, halogenated, or strongly reducing compound. If this is occurring, reverse the sample and reference sides of the detector periodically to equalize the changes. 4. To extend filament life, turn off the detector current overnight. For the best detector stability in the morning, leave the current on overnight (along with the oven temperature). S. Set the detector temperature just slightly above the highest column temperature when programming. Higher setpoints sacrifice sensitivity. 6. Use the instrument operating manual for setting the filament current Higher current settings give better sensitivity; lower current settings prolong filament life. 7. Since response factors are inversely proportional to column flow, set carrier flow as low as is consistent with good column performance and speed of analy: 5.5 ELECTRON CAPTURE DETECTOR One of the most popular and powerful gas chromatographic detectors in use today is the electron capture detector (ECD). Based on gas-phase electron capture reactions, this approach to detection can provide response to picogram or even femtogram levels of specific substances in complex matrices. This makes it an ideal detector for environmental and biomedical studies. 5.5.1 Background The introduction of the electron capture detector in 1960 (28) resulted from a series of developments begun in 1951 with the invention of the B-ray ionization cross-section detector (Section 5.9). In 1958 Lovelock (29) modified this detector to produce the B-ray argon detector (Section 5.8). He substituted argon as the carrier gas and placed a potential of 1000 V across the electrodes. Argon absorbed £-radiation from a radioactive source and formed a metastable species with sufficient energy to ionize most substances. Because anomalies were observed when halogenated compounds were eluted, Lovelock proposed the theory that electronegative species, func- tionally present in an organic molecule, could capture an electron to form a negatively charged species: CX +e CX +energy (5.10) These entities would then cause a reduction in the standing or background