Response Mechanisms of Thermionic Detectors with Enhanced

Anal. Chem. 2001, 73, 5698-5703

Response Mechanisms of Thermionic Detectors with Enhanced Nitrogen Selectivity Ha˚kan Carlsson,*,†,‡ George Robertsson,† and Anders Colmsjo 1†

Department of Analytical Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden, and Swedish Defence Research Agency, FOI, Weapons and Protection Division Grindsjo¨n, SE-147 25 Tumba, Sweden

The response mechanisms of a thermionic detector with enhanced nitrogen selectivity operating in an inert gas environment were investigated. According to accepted theory, the analyte has to contain electronegative functional groups in order for negative ions to be formed by the extraction of electrons from the thermionic source. This leads to a selective detector response for compounds containing nitro groups or multiple halogens. However, in the tests described here, polycyclic aromatic nitrogen hydrocarbons (PANHs), acridines, and carbazoles were used as reference substances. These compounds contain no electronegative functional groups. None of the investigated acridines exhibited any response from the detector, but carbazoles generated a strong structure-related detector response. By examining partial charges for all hydrogens of all individual carbazoles and acridine, it was demonstrated that the acidic hydrogen atom attached to the nitrogen heteroatom of the carbazoles has a strong influence on the detector response. Ionization of carbazoles may occur by dissociation of the nitrogen-hydrogen bond during contact with the thermionic surface. Support for this theory was provided by the linear relationship between the relative detector response and the deprotonization energy of the carbazoles (coefficients of determination of 0.90 and 0.98 for linear and quadratic models, respectively, were obtained). Further, there appeared to be no linear relationship between the detector response and electron affinity of the carbazoles, (R2 value, 0.32). Thus, the mechanism involved in ionization of the carbazoles is probably not direct electron transfer from the thermionic surface to the carbazoles. Principal component analysis (PCA) showed that the thermal conductivity of chemically inert detector gases also has an influence on the detector response. The investigated gases were helium, neon, nitrogen, carbon dioxide, and argon. It was found that thermal conductivity can be used to rank the detector response for the carbazoles, and there was no discernible response when helium, which has the highest thermal conductivity, was used as the detector gas. In 1974, Kolb and Bischoff described the first nitrogen- and phosphorus-specific thermionic detector (TID), which was devel* Corresponding author. E-mail: [email protected]. † Stockholm University. ‡ Swedish Defence Research Agency.

5698 Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

oped from the alkali flame ionization detector (AFID).1 The term TID principally refers to the nature of the ionization process, because the sample molecules are converted to negative ions in the detector by extracting electrons emitted from a hot solid surface. Nitrogen-phosphorus detector (NPD) is the most common name for this type of device, but it is misleading, because thermionic detectors have several operating modes and the term NPD should be used only when the TID is specific for nitrogenand phosphorus-containing compounds.2 When a TID operates in NP mode, it requires a detector gas environment containing both hydrogen and air. When the hydrogen flow rate is set to 2-6 mL/min, no self-sustaining flame can be established, as in the flame ionization detector (FID). The latter operates at a hydrogen flow rate of 20-30 mL/min. The main part of an NPD is the electrically heated thermionic source, consisting of an alkali salt in an inorganic ceramic cement matrix. Kolb and Bischoff used rubidium in the thermionic source of the original NPD, but nowadays, cesium is used in many NPD models. The operating principle of the NPD has not yet been completely established, and the detailed mechanism of the specific ionization process (i.e., how the degradation products of nitrogenand phosphorus-containing compounds can become negatively charged ions by extracting electrons from the hot solid source containing an alkali salt) is not fully understood. Two different theories for the ionization process have been suggested: a gasphase ionization theory proposed by Kolb and Bischoff1,4 and a surface ionization theory put forward by Patterson5 and Olah et al.6 According to the gas phase ionization theory, the alkali salt has to be vaporized at a probe temperature of 600-800 °C. However, because the source has a relatively long lifetime, the alkali ions would have to be drawn back to the charged source surface and recycled. Because of this difficulty, the surface ionization theory seems more likely to be true. According to this theory, the reactive radicals H, O, and OH are generated from the detector gases, H2 and O2. Similar to events that occur in the FID, this is a gas-phase process, and the formation takes place in the hot boundary layer in the vicinity of the thermionic source. Nitrogen- and phosphorus-containing compounds are thermochemically decomposed in the hot, active boundary layer contain(1) Kolb, B.; Bischoff, J. J. Chromatogr. Sci. 1974, 12, 625. (2) Patterson, P. L. Detect. Capillary Chromatogr.; 1992, 121, 139. (3) Douglas, B.; McDaniel, D. H.; Alexander, J. J. Concepts and Models of Inorganic Chemistry, 2nd ed; John Wiley & Sons, Inc.: New York, 1983. (4) Kolb, B.; Auer, M.; Pospisil, P. J. Chromatogr. Sci. 1977, 15, 53. (5) Patterson, P. L. J. Chromatogr. 1979, 167, 381. (6) Olah, K.; Szoke, A.; Vatja, Zs. J. Chromatogr. Sci. 1979, 17, 497. 10.1021/ac010204d CCC: $20.00

© 2001 American Chemical Society Published on Web 11/03/2001

ing the reactive radicals, and electronegative decomposition products (CN, PO, and PO2) are thereby formed. The electronegative species are converted to negative ions by extracting electrons from the hot surface of the ceramic source.7 For most NPDs, a polarization voltage of 3-5 V is applied between the thermionic source and the positive charge collector electrode. In surface ionization theory, an important parameter of the thermionic source is the surface work function, which is a measure of the amount of energy required to emit an electron from the source surface. By adding an alkali salt to the ceramic matrix, the work function is decreased, and the detector’s response to electronegative species produced in the gas-phase is increased. The surface ionization efficiency is given by the equation

IE )

()

1 g0 W - EA exp 1 + gkbT

where IE is the ionization efficiency; (g0/g-) is the ratio of statistical weights of neutral and ionic species, respectively; W is the surface work function; EA is the electron affinity of the sample; kb is Boltzmann’s constant; and T is the surface temperature of the thermionic source. According to the equation, if the work function is high (i.e., the content of the alkali salt in the source is low) the operating temperature of the thermionic source must also be high, which increases the background current and can shorten the life of the thermionic source.2 NPD is a type of detector that is both sensitive to mass flow rate and element-specific. The selectivity is defined as the ratio of detector responses to nitrogen or phosphorus compounds, as compared to hydrocarbons. A response ratio of more than 105 is obtained for most NP detectors. An explanation of the selectivity of NPDs, in comparison to FID systems, is that their operating temperatures are substantially cooler than the ca. 1000-2000 °C used in FIDs. Sample compounds are, therefore, not completely decomposed in the NPD environment, which enhances its selectivity.8 Most NPDs exhibit ∼10 times greater sensitivity toward phosphorus compounds than toward nitrogen compounds. The signal-to-mass response of most NPDs is linear up to a ratio of 105. The responses of thermionic detectors in the NP mode are only weakly affected by chemical structure, that is, the way the nitrogen or phosphorus atom is bound in the molecule has only a weak influence on the detector’s response. The nitrogen or phosphorus content in the sample molecule is the most important response parameter. This is probably due to the efficient decomposition of the sample compounds by the detector gases. For nitrogen-containing compounds, small deviations in detector responses will occur for nitro and amide substituted compounds, which exhibit lower responses than other nitrogen-containing compounds. The highest detector responses are obtained for compounds that easily decompose and yield CN radicals.9 In 1982, Patterson and co-workers developed a thermionic detector, subsequently manufactured and supplied by DETector (7) Patterson, P. L.; Gatten, R. A.; Ontiveros, C. J. Chromatogr. Sci. 1982, 20, 97. (8) DET Report; DETector Engineering & Technology, Inc.: Walnut Creek, CA; no. 25, 1994. (9) Hartigan, M. J.; Purcell, J. E.; Novotny, M.; McConnell, M. L.; Lee, M. L. J. Chromatogr. 1974, 99, 339.

Engineering & Technology, Inc., with a detection mode called TID-1-N2.7 This detector has a much lower work function, that is, a higher alkali salt content (in this case, a cesium salt) in the ceramic source, as compared to most other NP detectors. The general rationale is that no gas-phase reactions will occur, because the detector operates in an inert nitrogen gas environment and at a lower probe temperature (400-600 °C) than thermionic detectors in NP mode. The sample molecules are intact or partially decomposed by catalytic or thermally induced reactions on the hot solid surface of the thermionic source.8 According to this theory, the formation of electronegative decomposition products, such as CN, PO, and PO2, will not occur. Rather, the sample itself has to contain electronegative functional groups in order to yield negative ions by extracting electrons from the thermionic source. Thus, the detector has a selective response for compounds containing nitro groups and multiple halogens. Furthermore, the response is structure-dependent, because the positions of the electronegative groups in the molecular structure have a strong influence on the magnitude of the detector response.10 CE Instruments has recently introduced a detector that operates on a principle similar to the TID-1-N2 with respect to work function, detector gases, source temperature, and selectivity toward electronegative groups. This detector is denoted TS 1 and exhibits enhanced nitrogen selectivity.11 An investigation of this thermionic source with respect to the detector response for polycyclic aromatic nitrogen hydrocarbons (PANHs) is described in this paper. PANHs exhibit strong aromatic properties. Pyridine and pyrrole have slightly higher resonance stabilization energies than benzene, but an addition of fused benzene rings lowers these energies.12 The investigated PANHs are all planar molecules, except for dibenzo[c,g]carbazole. Molecular mechanics force field calculations, solved using a molecular modeling computer program, show that dibenzo[c,g]carbazole favors a slightly nonplanar configuration as a result of the steric effects of the two outermost benzene rings. The aromatic character arises from each carbon atom and the nitrogen contributing one electron to the delocalized π electrons. Nitrogen is more electronegative than carbon, so the pyridine ring is electron-deficient. Thus, a partial negative charge is located on the nitrogen atom, with a dipole moment of 2.26 D. Furthermore, the unshared electron pair on the nitrogen atom is available for covalent bonding. Thus, acridines may be protonated and may exhibit slightly basic properties. Acridines have pKa values of ∼9, which means that they are weak bases and less basic than aliphatic amines. This is due to the sp2 hybridization of the nitrogen atom in the pyridine ring. An sp2 hybrid orbital has more of an s character than an sp3 hybrid orbital, so the unshared electrons are more tightly held. This favors stabilization of the neutral molecule rather than the cation.12,13 Addition of fused rings to pyridine affects the basic strength only to a minor extent.14 (10) Patterson, P. L. Chromatographia 1982, 16, 107. (11) Verga, G. R.; Bedini, F. 20th International Symposium on Capillary Chromatography 1998. K 16. Sandra, P., Rackstraw, A. J., Eds.; CD Programming and Production: Naxos. Schriesheim, Germany. (12) Later, D. W. In Handbook of Polycyclic Aromatic Hydrocarbons; Bjo¨rseth, A., Ramdahl, T., Eds.; Marcel Dekker: New York, 1985; Vol. 2, p 265. (13) Fessenden & Fessenden, Organic Chemistry, 3rd ed.; Brooks/Cole Publishing Company: Monterey, CA, 1974. (14) Albert, A.; Goldacre, R.; Phillips, J. J. Chem. Soc. 1948, 2240.

Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

5699

Carbazoles, which contain a pyrrole ring, have a pentagonal structure and a hydrogen bonded to the nitrogen heteroatom. Their aromatic properties are due to the nitrogen atom’s contributing to the π electrons. In contrast to pyridine, pyrrole is electronrich because there are six π electrons in the five-membered ring, so in this case, a partial positive charge is located on the nitrogen atom, leading to a dipole moment of 1.81 D.12,13 Because the electrons of the nitrogen atom are used in bonding, carbazoles cannot be protonated on the nitrogen atom without losing their aromatic character. Pyrrole has no resonance stabilization energy for the cation and, therefore, exhibits no basic properties. In contrast, resonance-contributing forms will help to stabilize the anion, which explains why carbazoles act as weak acids.16 The aim of the study presented here was to extract a relationship that can explain why carbazoles with no electronegative functional groups exhibit responses from the thermionic source (TS1), in contrast to acridines, and to suggest mechanisms for the structure-dependent response for carbazoles. We also investigated whether different types of chemically inert gas environment influence the detector response. EXPERIMENTAL SECTION Chemicals. The acridine type PANH reference substances, acridine, benzo[h]quinoline, phenanthridine, benza[a]acridine, benz[c]acridine, 10-azabenzo[a]pyrene, dibenz[c,h]acridine, dibenz[a,h]acridine, dibenz[a,j]acridine, and dibenz[a,i]acridine, were purchased from Promochem (Wesel, Germany). The carbazole-type PANH reference substances, carbazole, benzo[def]carbazole, benzo[a]carbazole, benzo[b]carbazole, benzo[c]carbazole, dibenzo[a,g]carbazole, dibenzo[a,i]carbazole, dibenzo[c,g]carbazole, dibenzo[c,h]carbazole, and 9-methylcarbazole, were kindly provided by Professor M. Zander (Figure 1). Instrumentation. GC analysis of the PANH reference substances was performed using a CE-Instruments 8000 Top gas chromatograph equipped with a DB-5 column (30 m, 0.25-mm i.d., 0.10-µm film thickness; J&W Scientific) and both a thermionic detector with enhanced nitrogen selectivity (TS1) and a nitrogenphosphorus thermionic detector (TS2). When using the TS1 and investigating structurally dependent responses, nitrogen was used as the carrier, makeup, and detector gas. The detector gas flow was set to 100 mL/min. Helium, neon, argon, and carbon dioxide were used as detector gases to investigate the effects of varying the chemically inert gas environment on the detector response. The detector gas flow was varied in the range 25-600 mL/min in these experiments, and helium was used as carrier and makeup gas. When performing analysis with TS2, nitrogen was used as carrier gas and makeup gas, and hydrogen and air were used as (15) Kice, J. L.; Marvell, E. N. Modern Principles of Organic Chemistry 2nd ed.; Collier Macmillan Publisher: London, U.K., 1974. (16) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V.; Montgomery, J. A., Jr.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, B. B.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, Rev. A.3; Gaussian, Inc.: Pittsburgh, PA, 1998.


Figure 1. Structures of the analytes.

detector gases. For both detectors, the column oven was held at 35 °C for 2 min, then raised by 10 °C/min to 300 °C. Samples were injected in the on-column mode. The detector temperatures were set to 300 °C. A PC-based laboratory data system (ELDS Win Pro, Chromatography Data System AB, Svartsjo¨, Sweden) was used for acquiring and processing the detector signals. Quantum Mechanical Calculations. The program Gaussian 98, Revision A.3, run under Linux on a Pentium II platform was used for quantum mechanical ab initio calculations.16 B3LYP, a density functional method, was used together with the 6-31G(d,p) basis set.17 The molecular geometry was optimized and the energy, the dipole vector, and the polarizibility tensor were calculated for each carbazole and acridine. For the carbazoles, the geometry of (17) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.

Table 1. Detector Responsea and Mulliken Charge for the Four Most Acidic Protonsb and the Least Acidic Protonc for Selected PANHs compd

detector response

1

2

3

4

5

carbazole benzo[a]carbazole benzo[b]carbazole benzo[c]carbazole benzo[def]carbazole dibenzo[a,g]carbazole dibenzo[a,i]carbazole dibenzo[c,g]carbazole dibenzo[c,h]carbazole 9-methylcarbazole acridine

1.0 4.6 2.6 5.0 1.4 14.1 16.2 20.1 12.5 0 0

0.253 0.255 0.254 0.254 0.254 0.256 0.257 0.257 0.254 0.135 0.102

0.084 0.086 0.086 0.088 0.085 0.090 0.086 0.103 0.090 0.135 0.101

0.084 0.085 0.086 0.084 0.085 0.085 0.086 0.103 0.086 0.134 0.094

0.081 0.085 0.084 0.083 0.081 0.085 0.085 0.084 0.085 0.085 0.092

0.078 0.066 0.077 0.078 0.080 0.065 0.066 0.081 0.077 0.075 0.085

a

Figure 2. GC chromatograms obtained with different thermionic sources. TS 2 exhibits a selective NP response and TS 1 has enhanced nitrogen selectivity. The figure shows the structuredependent response for the thermionic source (TS1) for selected carbazoles; an equivalent amount of the individual carbazoles was used for both thermionic sources. Further, the response of the TS1 compared to TS2 was ∼4-7 times higher for the carbazoles with four aromatic rings and ∼21-32 times higher for the carbazoles with five aromatic rings. 1 ) carbazole, 2 ) benzo[def]carbazole, 3 ) benzo[a]carbazole, 4 ) benzo[b]carbazole, 5 ) benzo[c]carbazole, 6 ) dibenzo[a,i]carbazole, and 7 ) dibenzo[c,g]carbazole.

the anions and the radical anions was also individually optimized. In all energy calculations, zero-point energy corrections were included, that is, the effect of molecular vibrations at 0 K. Electron affinity was calculated as the difference between the energy of the radical-anion and the corresponding molecule. Deprotonization energy was calculated as the difference in energy between the anion of the deprotonated molecule and the molecule itself. The CPU processing time required for calculating the carbazole molecule anion and radical-anion parameters was almost a week, and for a selected dibenzocarbazole, the CPU time required was almost a month. Mulliken charges have also been calculated. RESULTS AND DISCUSSION The responses of a detector with a thermionic source (TS1) toward PANH reference substances without electronegative functional groups were evaluated. None of the investigated acridines exhibited any detector response in the range 0.3-3 pg/ s, the calculated threshold level for detecting carbazoles with three to five aromatic rings. Carbazoles evoked a strong structuredetermined detector response, which was stronger than the response they induced from a thermionic source with selective NP (TS2). Figure 2 shows the GC traces of carbazole reference substances with three to five aromatic rings, using the TS1 and TS2 thermionic sources. The response of the TS1 compared to TS2 was ∼4-7 times higher for the carbazoles with four aromatic rings, and ∼21-32 times higher for the carbazoles with five aromatic rings. Molecular Structure-Induced Detector Responses. Mulliken charges are used to quantify the acidity of hydrogen atoms,

Normalized to carbazole. b Columns 1-4. c Column 5.

higher charges corresponding to more acidic protons. Mulliken population analysis partitions the total charge among the atoms in a molecule. To investigate whether the acidic hydrogen atom attached to the nitrogen heteroatom of the carbazoles influences the detector response, the Mulliken charge for all of the protons of each carbazole and acridine was calculated. The Mulliken charges for the nonmethylated carbazoles were ∼0.25 for the hydrogen atom attached to the nitrogen heteroatom, and ∼0.09 for the most acidic hydrogen attached to a carbon atom in the aromatic ring (Table 1). The two most acidic hydrogen atoms of acridine and 9-methylcarbazole have Mulliken charges of 0.135 and ∼0.10, respectively. The higher value of the Mulliken charge for the nonmethylated carbazoles, due to the acidic hydrogen atom attached to the nitrogen heteroatom, is a possible explanation for the difference in detector response between acridines and carbazoles when using the TS1 probe. Ionization of carbazoles may occur by dissociation of the nitrogen-hydrogen bond during contact with the thermionic surface. The total loss of detector response for carbazoles that occurs when the acidic hydrogen atom attached to the heterocyclic nitrogen is replaced by a methyl group provides support for this theory. Furthermore, phenols, which also have acidic properties, have been shown to induce detector responses when using the TS1 probe.18 The Mulliken charge for the most acidic hydrogen atom in phenol was calculated to be 0.35. The relative molar response of the investigated carbazoles is presented in Table 1. The structurally dependent responses vary from 1 to 20 for the nonmethylated carbazoles when the relative response is normalized to the response of the three-ringed carbazole. The response was found to be strongly dependent on the number of aromatic rings. The detector response increased with the number of aromatic rings. When plotting the relative response versus the deprotonization (i.e., MH f M- + H+) energy for the nonmethylated planar carbazoles, coefficients of determination of 0.90 and 0.98 for linear and quadratic models, respectively, were obtained (Figure 3).This confirms that increased deprotonization energy tends to reduce the detector response. Further, the deprotonization energy decreases with the number of aromatic rings; that is, the detector response increases (18) Patterson, P. L. 16th International Symposium on Capillary Chromatography, Riva del Garda, Italy, 1994.


5701

Figure 3. Detector response relative to carbazole plotted against deprotonization energy. Molecules from left to right are dibenzo[a,i]carbazole, dibenzo[a,g]carbazole, dibenzo[c,g]carbazole (nonplanar configuration), dibenzo[c,h]carbazole, benzo[a]carbazole, benzo[c]carbazole, benzo[b]carbazole, benzo[def]carbazole, and carbazole.

significantly with the number of aromatic rings. The compound that generates the highest detector response is dibenzo[c,g]carbazole, which deviates from the linear and quadratic model, possibly because it has a slightly nonplanar configuration, due to steric effects of the two outermost benzene rings. According to theoretical understanding of the mechanisms of thermionic sources operating in an inert nitrogen gas environment, the sample itself has to contain electronegative functional groups in order to yield negative ions by extracting electrons from the heated thermionic source.9 If anions are formed at the ceramic surface, then the following reaction could be expected to occur.

M + e- (surface) f MThis theory implies that the electron affinity of the analytes should be an important factor. However, when plotting the detector responses versus the electron affinity for the nonmethylated carbazoles, a coefficient of determination (R2) of just 0.32 was obtained, and there is clearly no linear relationship between the detector response and electron affinity of the carbazoles. Therefore, the key mechanism whereby the carbazoles are ionized is not direct electron transfer from the thermionic surface to the carbazoles. These results lend further support to the theory that ionization of carbazoles may occur by dissociation of the nitrogenhydrogen bond during contact with the thermionic surface. Our experimental findings for the relationship between response and electron affinity are similar to those reported for benzenes substituted with various electronegative functional groups.19 Detector Responses as a Function of the Inert Gas Environment. The thermionic source (TS1) operates in an inert nitrogen gas environment. To investigate if different chemically inert gases affect the detector response, helium, neon, argon, carbon dioxide, and nitrogen were investigated as detector gases. The examined analytes in this investigation were carbazole, benzo[def]carbazole, benzo[c]carbazole, dibenzo[a,i]carbazole, and dibenzo[c,g]carbazole. The thermionic source was heated with a (19) Jones, C. S.; Grimsrud, E. P. J. Chromatogr. 1987, 387, 171-186.


Figure 4. Score vectors for five molecules with neon, argon, nitrogen, and carbon dioxide (marked ×, +, /, and o, respectively) as detector gases. The molecules are, from left to right, carbazole, benzo[def]carbazole, benzo[c]carbazole, dibenzo[c,g]carbazole, and dibenzo[a,i]carbazole.

constant current and reached a temperature of ∼400-600 °C. When gas flows over the hot surface of the thermionic source, a gaseous boundary layer in proximity to the hot surface is obtained.20 The gaseous boundary layer of the thermionic source (TS1) is not chemically active, but it plays an important role in the heat transfer from the source to the sample and in the mass transfer of the sample molecules. The detector response using different inert gases at various flowrates was investigated for five different analytes. The obtained detector responses were arranged into a matrix (analyte × flow rate), one such matrix for each inert gas. A response matrix was formed, with a row of responses for each of the analytes tested and a column for each of the different flow rates (25-600 mL/ min). The obtained matrixes were decomposed by principal component analysis (PCA) for evaluation and interpretation.21 Prior to model calculations, data was pretreated, and the calculations were conducted in the Matlab program using in-house algorithms.22 One principal component was found to be sufficient to describe the variation in the matrixes, respectively. The variance in the data explained by one PCA factor was 96, 98, 98, and 99% for neon, argon, nitrogen, and carbon dioxide, respectively. This means that each response matrix was approximated by the product of a column vector (score vector) and a row vector (loadings vector). The score vectors, from each PCA model, of the detector gases are presented in Figure 4. The scores for individual compounds show the detector response as a function of the detector gas type. The highest score was obtained for argon, followed by nitrogen, carbon dioxide, and neon. The differences in detector response may be due to differences in the thermal conductivity of the gases. The thermal conductivity at 500 K for helium, neon, nitrogen, carbon dioxide, and argon are 222.3, 69.6, 38.3, 33.5, and 26.8 mW/mK, respectively.23 The thermal conductivity of the gases can be used to rank the detector responses for (20) Coulson, J. M. Chemical Engineering, 6th ed.;; Oxford Pergamon: Oxford, 1999. (21) Wold, S.; Esbensen, K.; Geladi, P. Chemom. Intell. Lab. Syst. 1987, 2 (1-3), 37-52. (22) The MathWorks Inc., MA.

the carbazoles, with the exception of nitrogen and carbon dioxide, which appear in reversed order. There was no discernible detector response when helium, which has the highest thermal conductivity, was applied as the detector gas. Detector gases with high thermal conductivity reduce the temperature of the thermionic source. This is presumably why the response to carbazoles was increased when gases with relatively low thermal conductivity were applied. Carbon dioxide and nitrogen were anomalous in this respect, because the carbon dioxide has lower thermal conductivity but generates a weaker detector response. This could be because carbon dioxide has a larger collisional cross section than nitrogen and, thus, provides slower mass transport. The loading vectors showing the detector response as a function of detector gas flow rate for neon, argon, nitrogen, and carbon dioxide are presented in Figure 5. Note that the mass flow in the detector of the carbazoles remained constant when the detector gas flow rate was changed. When the gas flow rate was increased, the thickness of the boundary layer was reduced. Thus, the mass transfer of the sample molecules to the thermionic surface increased, but the temperature of the thermionic source decreased. These two parameters, heat transfer and mass transfer, counteract each other; therefore, to maximize the detector response, an optimal balance between these two parameters should be sought. This is in agreement with the observed loading vectors of the detector gases shown in Figure 5. CONCLUSIONS The thermionic detector with enhanced nitrogen selectivity operating in an inert gas environment exhibits at least one additional response mechanism besides the one described by established theory. Analytes with acidic protons induce strong structure-related detector responses, as confirmed by the linear (23) Linde, D. R. Handbook of Chemistry and Physics; CRC Press: Cleveland, OH, 1996.

Figure 5. The loadings for neon, argon, nitrogen, and carbon dioxide (marked ×, +, /, and o, respectively) as a function of detector gas flow rate.

relationship found between relative response and deprotonization energy. Analytes with more acidic protons exhibit stronger detector responses. Further, the nature of the chemically inert gas environment influences the detector response, and gases with low thermal conductivity induce stronger detector responses for analytes with acidic protons. ACKNOWLEDGMENT The authors wish to thank Ludvig Moberg for valuable discussions. Received for review February 19, 2001. Accepted September 17, 2001. AC010204D


5703

Response Mechanisms of Thermionic Detectors with Enhanced

Recommend Documents