Anal. Chem. 2005, 77, 2288-2296
Application of Time-of-Flight Mass Spectrometry with Laser-Based Photoionization Methods for Time-Resolved On-Line Analysis of Mainstream Cigarette Smoke Stefan Mitschke,†,‡ Thomas Adam,†,‡ Thorsten Streibel,†,‡ Richard R. Baker,§ and Ralf Zimmermann*,†,‡,|
Analytical Chemistry, Institute of Physics, University of Augsburg, D-86159 Augsburg, Germany, Institute of Ecological Chemistry, GSFsNational Research Centre for Environment and Health, D-85764 Neuherberg, Germany, British American Tobacco, R&D Centre, Southampton SO15 8TL, U.K., and Environmental Chemistry, BIfAsBavarian Institute of Applied Environmental Research and Technology GmbH, D-86167 Augsburg, Germany
The application of soft photoionization mass spectrometry methods (PIMS) for cigarette mainstream smoke analysis is demonstrated. Resonance-enhanced multiphoton ionization (REMPI) at 260 nm and vacuum ultraviolet light single-photon ionization (SPI) at 118 nm were used in combination with time-of-flight mass spectrometry (TOFMS). An optimized smoking machine with reduced memory effects of smoke components was constructed, which in combination with the REMPI/SPI-TOFMS instrument allows PIMS smoke analysis with a time resolution of up to 10 Hz. The complementary character of both PIMS methods is demonstrated. SPI allows the detection of various aliphatic and aromatic compounds in smoke up to ∼120 m/z while REMPI is well suited for aromatic compounds. The capability of the instrument coupled to the novel sampling system for puff-by-puff resolved measurements is demonstrated. The feasibility of using the experimental system for intrapuff smoke measurements is also shown. Two main patterns of puff-by-puff behaviors are observed for different smoke constituents. The first group exhibits a constant increase in smoke constituent yield from the first to the last puff. The second group shows a high yield of the constituent in the first puff, with lower and constant or slowly increasing yields in the following puffs. A third group cannot be clearly classified and is a combination of both observed profiles.
Conventional smoke analysis is mostly done with wellestablished off-line techniques such as gas chromatography (GC) and liquid chromatography (LC) methods, using the smoke * To whom correspondence should be addressed. E-mail:
[email protected]. † University of Augsburg. ‡ GSFsNational Research Centre for Environment and Health. § British American Tobacco. | BIfAsBavarian Institute of Applied Environmental Research and Technology GmbH.
2288 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
collected from all the puffs of the cigarette and often several cigarettes.1 To simplify and standardize the analyses, cigarette smoke is often divided into the particulate and gas phase using various techniques, the most common nowadays being the Cambridge filter.2 The use of the Cambridge filter pad for collecting the particulate phase of cigarette smoke was first described in 1959 by Wartman et al.3 The quartz glass filter pad traps particles larger than 0.1 µm present in the cigarette smoke aerosol with 99.9% efficiency4 while gas-phase smoke components pass through the filter. However, this results in the loss of information about smoke dynamics and formation mechanisms. For their investigation, analytical methods that provide data in a much smaller time scale are required. Time-of-flight mass spectrometry is well known for its high time resolution, up to 100 Hz, but its combination with the conventional electron impact (EI) ionization usually results in intense fragmentation. Therefore, the interpretation of spectra of complex mixtures, such as tobacco smoke containing several thousands of chemical species,2,5 is very difficult or even not possible. However, in contrast to EI, photoionization (PI) methods are soft techniques that do not produce molecular fragments. PI-based mass spectrometric methods have quickened interest once again within the last years. This is in particular true for the vacuum ultraviolet (VUV) lamp-based atmospheric pressure photoionization technique (APPI), which is well suited for HPLCMS methods. The ionization of complex gas mixtures in an APPI ion source, however, suffers from intense ion molecule reactions, impairing the PI selectivity and possibility of quantification. It should be noted that newly developed PI techniques, utilizing novel highly intense and brilliant VUV light sources,6,7 could be (1) Green, C. R.; Rodgman, A. Recent Adv. Tob. Sci. 1996, 22, 131-304. (2) Dube, M. F.; Green, C. R. Recent Adv. Tob. Sci. 1982, 8, 42-102. (3) Wartman, W. B. J.; Cogbill, E. C.; Harlow, E. S. Anal. Chem. 1959, 31, 1705-1709. (4) Baker, R. R. Beitra ¨ ge Tabakforsch. Int. 2002, 20 (1), 23-41. (5) Stedman, R. L. Chem. Rev. 1968, 68, 153-207. (6) Mu ¨ hlberger, F.; et al. Anal. Chem. In press. (7) Ulrich, A.; et al. Ionenquelle bei der UV-VUV-Licht zur Ionisation verwendet wird. in Deutsches Patentamt; Deutschland, 2000. 10.1021/ac050075r CCC: $30.25
© 2005 American Chemical Society Published on Web 03/17/2005
an alternative to laser-based VUV generation for single-photon ionization (SPI)-MS-based gas-phase analysis instruments. Such lamp-based SPI-MS methods are in particular promising for industrial process monitoring and quality control applications. The resonance-enhanced multiphoton ionization (REMPI) technique uses at least two photons for photoionization, which takes place via an optical resonance absorption step. Due to this resonant absorption, the selectivity of UV gas-phase laser spectroscopy is included in the ionization process. Only those molecules that exhibit a suitable electronic transition with a respective excitation wavelength together with an ionization potential (IP) lower than the combined energy of the photons may be ionized by two-photon ionization. The REMPI technique, in particular, is well suited for the on-line analysis of aromatic compounds.8-11 The SPI technique using VUV photons for ionization can be used to detect additional compounds, e.g., aliphatic hydrocarbons or carbonyl compounds.12-14 As with REMPI, SPI generally causes no fragmentation of molecules. However, the selectivity of the SPI process is less compared to REMPI, because all compounds with an IP lower than the photon energy may be ionized. Despite this, most background gases such as nitrogen, oxygen, carbon dioxide, and water are not subject to ionization, because their IPs are higher than the energy of the most commonly used VUV sources ( 2) is similar to the value achieved by Tonokura et al.36 To prove the linearity of the instrument for SPI different concentrations of benzene (1000, 500, 250, 10 ppm) were measured. The corresponding C13 and 2 × C13 peaks were used to extend the limits to 3 orders of magnitude. The results are given in Figure 2. Linearity in the range of 4 orders of magnitude with a comparable SPITOFMS instrument setup is reported by Tonokura et al.36 For the REMPI-TOFMS method, Oser et al. demonstrated the linearity (35) Hafner, K. Untersuchungen zur Bildung brennstoffabha ¨ ngiger Stickoxide bei der Abfallverbrennung mittels on-line analytischer Messmethoden; Technische Universita¨t Mu ¨ nchen: Mu ¨ nchen, 2004. (36) Tonokura, K.; Nakamura, T.; Koshi, M. Anal. Sci. 2003, 19, 1109-1113.
Table 1 (a) SPI Detection Limits of Different Standard Compounds (S/N > 2) av 1 av 3 av 10 benzene toluene xylene propene butadiene isoprene acetaldehyde acetone NO NO2
56 ppb 92 ppb 109 ppb 87 ppb 64 ppb 93 ppb 82 ppb 82 ppb 26.3 ppm 2.0 ppm
46 ppb 66 ppb 82 ppb 45 ppb 37 ppb 59 ppb 74 ppb 78 ppb 15.4 ppm 1.5 ppm
av 100
28 ppb 44 ppb 53 ppb 30 ppb 20 ppb 29 ppb 43 ppb 39 ppb 8.9 ppm 751 ppb
16 ppb 25 ppb 30 ppb 10 ppb 8 ppb 11 ppb 14 ppb 14 ppb 3.0 ppm 282 ppb
(b) REMPI Detection Limits of Different Standard Compounds at Different Wavelengths35 REMPI detection limit (av 10) S/N > 2
benzene aniline indole naphthalene a
224 nm
266 nm
272.5 nm
2.3 ppm a a 40 ppb
670 ppb 7.2 ppb 2.9 ppb 8.4 ppb
a 5.5 ppb 4.8 ppb 22 ppb
Compound is not detectable at selected wavelength.
Figure 3. Comparison of the smoking profiles of the original Borgwaldt single-port smoking machine and the novel sampling system, shown for the fourth and fifth puffs of 44 m/z (acetaldehyde) of a 2R4F research cigarette. The cigarette was removed after the main peaks and five cleaning puffs were taken with the old machine, while two were taken with the newly designed system.
Figure 2. Proof of linearity of the SPI-TOFMS by measuring standards (1000, 500, 250, 10 ppm) and the corresponding C13 and 2 × C13 peaks.
for the detection of benzene to be in the range of 6 37 and of naphthalene in the range of 5 38 orders of magnitude. The 2R4F University of Kentucky reference cigarette is a standardized, wellcharacterized39 cigarette, which is available from the University of Kentucky, Kentucky Tobacco Research & Development Center (KTRDC). The cigarette gives “tar”, nicotine, and carbon monoxide yields of 8.9, 0.75, and 12 mg, respectively, when smoked under standard International Organization of Standardization (ISO) smoking machine conditions.39 The “Light” brand used in this study is available commercially in the United States. The cigarette gives tar, nicotine, and carbon monoxide yields of 8, 0.8, and 10 (37) Oser, H.; et al. Jet-REMPI for Process Control in Incineration, In Combustion Diagnostics; Tacke, M., Stricker, W., Eds.; 1997: pp 21-29. (38) Oser, H.; Thanner, R.; Grotheer, H.-H. Combust. Sci. Technol. 1996, 116117, 567-582. (39) Chen, P. X.; Moldoveanu, S. C. Beitra ¨ ge Tabakforsch. Int. 2003, 20 (7), 448-458.
mg, respectively, when smoked under standard ISO smoking machine conditions. The cigarettes were stored for several days under controlled conditions of 60% relative air humidity and 22 °C and smoked according to the ISO conditions (35-mL puff volume, 2-s puff duration, one puff every 60 s).4 Cigarettes were lit with a Borgwaldt electric lighter. The smoking procedure was carried out using a custom-made smoking machine based on a Borgwaldt single-port smoking machine. Customization became necessary because of the lack of on-line capabilities of the conventional smoking machine, where the smoke is first aspirated into a glass tube and then blown out by a piston. During initial measurements with the commercial system, significant memory effects from volatile and semivolatile smoke compounds were observed (Figure 3). The contamination during the first cleaning puff may be explained by smoke, which is still present in the dead volume of the cigarette holder, and the further presence of compounds strongly indicates condensation and adsorption effects inside the machine. Mixing of the smoke in the glass tube also results in the complete loss of the subpuff time resolution. Therefore, the original Borgwaldt valve was connected directly to a modified and heated Swagelok branch connection, where the sampling capillary could be inserted directly into the sample flow. A membrane sampling pump (KNF-Neuberger) was installed and adjusted to a continuous flow rate of 1.05 L/min, which equals a volume of 35 mL/2 s. A conventional particle filter was installed to protect the pump section of contamination with unfiltered smoke. The continuous ambient air flow between the cigarette Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2291
Figure 4. Comparison of REMPI and SPI mass spectra of filtered and unfiltered MSS of a 2R4F Kentucky research cigarette.
puffs is used to clean the sampling branch connection; however, it is possible to connect any other gas for that purpose. For further cleaning of the sampling system, two additional blank puffs were taken between the smoking puffs. For that purpose the cigarette was removed carefully from its holder. If any contamination was observed during the cleaning puffs, the signal was added to the puff it succeeded. Both smoking machines were compared by their ratio of total ion signal of a single whole smoke puff to the summed ion signals of the succeeding cleaning puffs. For the original Borgwaldt single-port machine, the average contamination is 88.7% of the original puff area while it is 30.8% for the custom-built machine. The smoking machine and the REMPI/SPI-TOFMS instrument are connected by a heated transfer line (1.5 m, 250 °C), which contains a deactivated silica capillary of 0.32-mm i. d. Reactions inside the capillary have not been observed; the retention time is less than 1 s. Quantitative evaluation is possible by application of calibration gases or diffusion and permeation standards16 which, however, is not the subject of this work. RESULTS AND DISCUSSION Figure 4 shows a comparison of REMPI and SPI spectra measured with and without a Cambridge filter pad, respectively. While SPI, in general, is more suitable for monitoring aliphatic and carbonyl-containing substances, REMPI gives a good overview on aromatic substances. Peak alignment was done using known information such as total yields from previous publications32 together with consideration of the selectivity criteria of the different photoionization methods resulting from the different spectroscopic properties of the substances. The assignments are presented in Table 2. The SPI spectra with and without filtering the smoke using a Cambridge filter pad (Figure 4a, b) look very 2292
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
similar. Differences can be seen in the higher mass region (>100 m/z), where most compounds are located in the particulate phase and therefore are adsorbed on the filter. The REMPI spectrum at a wavelength of 260 nm with the filter pad (Figure 4c) illustrates that larger aromatic species and most other detectable substances except benzene, toluene, and xylene (BTX) are also adsorbed on the filter pad. The REMPI spectra presented in Figure 4c, d only represent a small part of the available data, as the use of different, selected wavelengths can provide further information. Figure 5a illustrates an overview of the measured SPI data for one cigarette. The data are presented in a three-dimensional manner depicting the temporal course of all observed spectra during the whole smoking cycle on a puff-by-puff time scale. The data are clearly shown for each puff, every 60 s in the smoking cycle. The dominating peaks are 44 (acetaldehyde), 58 (acetone), and 68 m/z (isoprene, accompanied by furan). For further clarification, the enlarged third and fourth puffs are presented in Figure 5b, which indicates the clearing puffs taken every 20 s between the main puffs. Minor contaminations in the first cleaning puff can only be observed for the high-concentration constituents while most of the compounds have very low concentrations in the second cleaning puff. Preliminary experiments have proven that a third cleaning puff does not yield additional signal intensity. The remaining memory effects of the smoking machine occur due to the dead volume inside the cigarette holder and the valve. Removing the holder before a cleaning step resulted in a significant decline; however, it has yet to be determined whether this amount also contributes to the smoking puff. In general, the modification of the smoking machine has proven useful in terms of an on-line analysis of cigarette smoke with reduced interpuff memory effects.
Table 2. Assignment of the Observed Masses to the Corresponding Compoundsa m/z 17 30 34 40 42 43 44 48 52 54 56 57 58 59 66 67 68 69 70 71 72 78 79 80 81 82 84 86 92 93 94 95 96 98
104 106 108 110 112 117 118 120 122 124 128 130 131 132 134 136 142 144 145 146 150 156 158 162 172
compounds NH3 NO H2S propyne propene carbohydrate fragment: C3H7+, C2H3O+ acetaldehyde methanthiol 1-buten-3-yne, 1,3-butadiene, 1-butyne 2-propenal, butene, 2-methylpropene carbohydrate fragment, 2-propen-1amine acetone, propanal 2-propanamine cyclopentadiene pyrrole furan, isoprene, 1,3-pentadiene, cyclopentene pyrroline 2-butenal, methyl vinyl ketone, methylbutene, pentene, butenone, 2-methyl-2-propenal pyrrolidine 2-methylpropenal, 2-butanone, butanal benzene pyridine pyrazine methylpyrrole methylfuran, methylcyclopentene, cyclohexene, 2-cyclopenten-1-one nicotine fragment, cyclopentanone, dimethylbutene, hexene, 3-methyl-3-buten-2-one methylbutanal, 3-methyl-2-butanone, pentanone, 2,3-butanedione toluene aniline, methylpyridine phenol, 2-vinylfuran pyridinol, ethylpyrrol, dimethylpyrrol dimethylfuran, furfural furanmethanol, methyl-2-pentenal, 5-methyl-2(5H)-furanone, 3-methyl-2(5H)-furanone, 1,3-cyclopentanedione styrene, 3-pyridinecarbonitrile xylene, ethylbenzene, benzaldehyde anisole, dimethylpyridine, methylphenol dihydroxybenzene, 2-acetylfuran, methylfurfural acetylcyclopentane, 2-hydroxy-3-methyl-2cyclopenten-1-one indole indane, methylstyrene, benzofuran methylethylbenzene, trimethylbenzene benzoic acid, ethylphenol, hydroxybenzaldehyde dihydroxymethylbenzene, guaiacol naphthalene methylindene 3-Mlethyl-1H-indole methylbenzofuran, 1-indanone, isopropenyltoluene isopropyltoluene limonene, methoxybenzaldehyde, 2-ethyl5-methylphenol methylnaphthalene 5-quinolineamine 2-(4-pyridyl)furan, 2,3-dimethyl1H-indole myosmine 4-vinylguaicol bipyridine, dimethylnaphthalene nicotyrine nicotine, anabasine 3,5-dimethyl-1-phenylpyrazole
refs 5, 44 45, 46 5, 44 5 5, 45 44, 45, 47 5, 48, 49 5, 44 5 5, 45 5, 45, 49, 50 5, 44 5, 44, 48, 4 5 5, 45, 49 5, 44, 45, 49-51 5, 44, 45, 48, 49 5, 44, 45, 49 5, 48, 49 5 5, 44, 48, 49 5, 45, 48, 49, 51 5, 49, 5 5, 44, 45 5, 44, 48-50, 52 5, 44, 45, 49 5, 44, 49 5, 44, 45, 49 5, 45, 49-51, 53, 54 5, 44, 49, 50, 54 44, 45 5, 44, 45, 48-50, 52, 54, 55 44, 45, 49, 50, 52 52, 54, 55
5, 45, 49, 51 5, 49, 50 5, 44, 49, 50, 52, 56 5, 44, 45, 49, 52, 54 44, 49, 53, 54 44, 45, 50, 51 45, 49 5, 49, 50 5, 44, 49, 50, 52, 56 5, 44, 49, 52, 54, 57 44, 45, 49, 50 49 58 5, 49 5 5, 45, 49, 50, 52-54, 56 5, 44, 45, 49, 59 58 58 44, 50 5, 49, 53, 54 58 5, 49, 50, 54 58
a The data are based on the given literature references and spectroscopic considerations.
Figure 5. (a) 3D plot of measured SPI data during the smoking cycle of a Kentucky 2R4F research cigarette. (b) Enlarged third and fourth puffs showing the reduced memory effects of the novel smoking machine.
Figure 6 shows typical smoking profiles of selected compounds, again with the puff taken every 60 s and intermediate puffs taken every 20 s. According to their behavior during the smoking cycle, they can be classified roughly into two different categories. The first group, represented by 58 m/z (acetone), exhibits a steady increase in yield with each puff during the smoking cycle. This is due to a gradual reduction in tobacco rod length as the cigarette is consumed, which results in a decrease in filtration by the tobacco rod for products in the particulate phase Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2293
Table 3. Assignment of the Observed Masses to the Different Puff-by-Puff Smoking Patterns increasing yield with puff number compound
m/z
high yield in first puff m/z
not clearly defined/ intermediate
compound
m/z
compound ammonia furan, isoprene, 1,3-pentadiene, cyclopentene benzene
30 34
nitric oxide hydrogen sulfide
40 52
propyne 1-buten-3-yne,
17 68
42
propene
54
78
43 44
carbohydrate fragment: C3H7+, C2H3O+ acetaldehyde
66
1,3-butadiene, 1-butyne cyclopentadiene
48 56 57 58 59 67 69 70
methanthiol 2-propenal, butene, 2-methylpropene carbohydrate fragment, 2-propen-1-amine acetone, propanal 2-propanamine pyrrole pyrroline 2-butenal, methyl vinyl ketone, methylbutene, pentene, butenone, 2-methyl-2-propenal pyrrolidine 2-methylpropenal, 2-butanone, butanal pyridine pyrazine methylpyrrole methylfuran, methylcyclopentene, cyclohexene, 2-cyclopenten-1-one nicotine fragment, cyclopentanone, dimethylbutene, hexene, 3-methyl-3-buten-2-one methylbutanal, 3-methyl-2-butanone, pentanone, 2,3-butanedione toluene aniline, methylpyridine phenol, 2-vinylfuran pyridinol, ethylpyrrol, dimethylpyrrole dimethylfuran, furfural furanmethanol, methyl-2-pentenal, 5-methyl-2(5H)-furanone, 3-methyl-2(5H)-furanone, 1,3-cyclopentanedione xylene, ethylbenzene, benzaldehyde anisole, dimethylpyridine, methylphenol dihydroxybenzene, 2-acetylfuran, methylfurfural
71 72 79 80 81 82 84 86 92 93 94 95 96 98
106 108 110
104 112
styrene, 3-pyridinecarbonitrile acetylcyclopentane, 2-hydroxy-3-methyl-2cyclopenten-1-one
of smoke or a decrease in air dilution and outward gaseous diffusion for products in the gaseous phase.40,41 The second group, represented by 40 m/z (propyne), has a high yield in the first puff while exhibiting a lower, but slowly increasing level during the following puffs. Similar behavior can be observed for 54
(mainly butadiene), 66 (cylcopentadiene), and 52 m/z (1-buten3-yne). These unsaturated hydrocarbons are well known to appear in the early stages of combustion processes.42 This phenomenon has also observed by Baren et al. for ethylene,28 Parrish and Harward for formaldehyde,26 and Li et al. for benzo[a]pyrene.43
Figure 6. SPI time profiles of 40 (propyne), 58 (acetone), and 68 m/z (isoprene) recorded during the smoking cycle of a Kentucky 2R4F research cigarette.
(40) Baker, R. R.; Robinson, D. P. Recent Adv. Tob. Sci. 1990, 16, 3-101. (41) Baker, R. R.; Crellin, R. A. Beitra ¨ ge Tabakforsch. Int. 1977, 9 (3), 131140. (42) Glassman, I. Combustion, 2nd ed.; Jovanovich, Harcourt Brace, Academic Press: Orlando, 1987. (43) Li, S.; et al. Combust. Flame 2002, 128, 314-319. (44) Scheijen, M. A.; Boer, B. B.-d.; Boon, J. Beitra ¨ ge Tabakforsch. Int. 1989, 14 (5), 261-282. (45) Shin, E.-J.; Hajaligol, M. R.; Rasouli, F. J. Anal. Appl. Pyrolysis 2003, 6869, 213-229. (46) Im, H.; Rasouli, F.; Hajaligol, M. J. Anal. Appl. Pyrolysis 2003, 51, 73667372. (47) Baker, R. R.; Kilburn, K. D. Beitr. Tabakforsch. Int. 1973, 7 (2), 79-87. (48) Seeman, J. I.; Dixon, M.; Haussmann, H.-J. Chem. Res. Toxicol. 2002, 15 (11), 1331-1350. (49) Schlotzhauer, W. S.; Chortyk, O. T. J. Anal. Appl. Pyrolysis 1987, 12, 193222. (50) Baker, R. R. J. Anal. Appl. Pyrolysis 1987, 11, 555-573. (51) Schmeltz, I.; Schlotzhauer, W. S.; Higman, E. B. Beitr. Tabakforsch. Int. 1972, 6 (3), 134-138. (52) Schlotzhauer, W. S.; Arrendale, R. F.; Chortyk, O. T. Beitr. Tabakforsch. Int. 1985, 13 (2), 74-80.
2294
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
Figure 7. Comparison of a 2R4F Kentucky research cigarette and a conventional, commercial Light cigarette by SPI-TOFMS: (a) 40 (propyne) and (b) 58 m/z (acetone).
Li et al.37 deduced that the excess benzo[a]pyrene in the first puff was from soot particles covered in polynuclear aromatic hydrocarbons, produced in the incomplete combustion of the yellow lighting flame. In a separate study, Li et al.20 deduced that the high levels of formaldehyde in the first puff were due predominantly (∼90%) to the tobacco rod in the region of ∼4 mm downstream of the burning zone not having been preheated in the previous puff. The preheating needed to be done to at least 250 °C in order to eliminate the high levels of formaldehyde in the first puff. A minor contribution (∼10%) was due to particulate matter condensed in the tobacco rod behind the burning zone acting as a filtering medium for formaldehyde by somehow binding to it. Other components in the present study that follow this pattern of high yield in the first puff are, for example, 94 (phenol) and 56 (acroleine und butene), 42 (mainly propene), 44 (acetaldehyde), and 30 m/z (nitric oxide). However, the puff-by-puff behavior of some compounds is between the two extremes, as illustrated by the behavior of 68 (53) Wynder, E. L.; Hoffmann, D. Tobacco and tobacco smoke. Studies in experimental carcinogenesis; Academic Press: New York, 1967; p 730. (54) Halket, J. M. J. Anal. Appl. Pyrolysis 1985, 8, 547-560. (55) Schulten, H.-J. J. Anal. Appl. Pyrolysis 1984, 6, 251-272. (56) Stotesbury, S. S. Beitr. Tabakforsch. Int. 1999, 18 (4), 147-163. (57) Haider, K. J. Anal. Appl. Pyrolysis 1985, 8, 317-331. (58) Zha, Q.; Moldoveanu, S. C. Beitr. Tabakforsch. Int. 2004, 21 (3), 184191. (59) Severson, R. F.; et al. Beitr. Tabakforsch. Int. 1977, 9 (1), 23-37.
m/z (mainly isoprene and furan). This mass exhibits a slow decrease in the first three puffs and then a very slow increase in the succeeding puffs (Figure 5). An assignment of observed masses to the different puff-by-puff patterns is presented in Table 3. A more detailed statistical analysis to investigate these findings, especially the first-puff deviation, will be presented in a future paper. In Figure 7, single puffs of a commercially available Light cigarette and a Kentucky 2R4F research cigarette are compared qualitatively by the averaged summed signal intensity of every puff for 40 (propyne; Figure 7a) and 58 m/z (acetone; Figure 7b) of three cigarettes together with the respective standard deviation. For both cigarette types, the typical continuous increase in signal intensity (i.e., yield) as is expected for acetone can be observed. During the smoking cycle, the Light cigarette shows lower concentration levels for acetone than the 2R4F research cigarette. Both cigarettes show enhanced levels of 40 m/z (propyne) in the first puff and an increase from the second to the last puff. The differences in the succeeding puffs are less than for 58 m/z (acetone). However, these effects are not necessarily directly correlated to the different tobacco compositions used in the cigarette but may also result from different cigarette design parameters such as filter ventilation and filter design. The 2R4F cigarette, on average, also yields one puff more than the Light brand. This has to be taken into consideration, when comparing total amounts of different cigarette types. Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2295
puff is rectangular while the concentration of benzene increases constantly during the puff and drops rapidly at the end of the puff. This visualizes the dynamics of pyrosynthesis and release of smoke constituents within a single puff. The high time resolution is well suited for an in-depth analysis of smoke component formation mechanisms and subpuff behavior.
Figure 8. Illustration of the time-concentration profile of a single puff, shown for 78 m/z (benzene) during the third puff of a Kentucky 2R4F research cigarette. At each measuring point, a whole mass spectrum was recorded.
Figure 8 shows an enlarged view of the concentration profile of 78 m/z (benzene) during the third puff of a 2R4F research cigarette measured with a resolution of 10 Hz. About 20 full mass spectra are recorded during a single puff. The flow profile of the
2296 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
CONCLUSION The coupling of soft and selective photoionization methods (SPI and REMPI) to time-of-flight mass spectrometry offers a powerful tool for the on-line real-time detection of various trace species in cigarette MSS. The time resolution provided by the instrument and the newly developed smoking machine are well suited for the investigation of puff-by-puff and intrapuff behavior of MSS as well as SSS, with significantly reduced memory effects. Furthermore, combination of REMPI and SPI-TOFMS provides comprehensive information about a large number of compounds in cigarette smoke. The instrumental setup allows further investigation of formation mechanisms and smoke dynamic behavior down to the subpuff level. Received for review January 14, 2005. Accepted March 3, 2005. AC050075R
