26th International Combustion Symposium, Naples, Italy, July-August 1996

Polycyclic Aromatic Hydrocarbon and Soot Formation in Premixed Flames

of CH₃Cl/CH₄ and CH₄

Jiawei Huang and Selim M. Senkan^§

Department of Chemical Engineering, University of California, Los Angeles, California 90095

^§ Tel: (310) 206-4106, e-mail: senkan@seas.ucla.edu

ABSTRACT

The micro-structures of atmospheric-pressure, laminar, premixed flames of CH ₃Cl/CH₄ and CH₄ have been studied under similar fuel-rich (equivalence ratio=2.5) and carbon density (1.9x10^-5 mole/cc) conditions using both heated micro-probe and cold trap sampling, followed by analysis by gas chromatography/ mass spectrometry (GC/MS). The mole fraction profiles of a large number species, including aromatic and polycyclic aromatic hydrocarbons (PAH) has been determined. In addition, the temperature and relative soot profiles were obtained using thermocouples and He-Ne laser light extinction measurements, respectively. These experiments indicated that although the CH₃Cl/CH₄ flame was more sooting, the levels of all the aromatics and PAH were significantly lower, some by as much as a factor of 10 than the CH₄ flame in the post flame zones. This is a surprising result that contradicts the generally held belief that increased soot formation is associated with higher PAH levels. These measurements also provide new information on flame chemistry and suggest that the major effect of chlorine is the rapid incorporation of PAH into soot.

INTRODUCTION

Polycyclic aromatic hydrocarbons (PAH), that are formed at trace levels in combustion and incineration processes, are of considerable practical interest because of the potential toxicities of some PAH isomers [1,2]. In addition, PAH are believed to play a major role in soot formation and growth [1,3-5]. The effects of chlorine on PAH and soot formation in flames is also of practical importance because chlorinated hydrocarbons (CHCs) are major components of many industrial wastes which are treated by combustion/incineration for their ultimate disposal.

It is well known that chlorine or chlorinated hydrocarbons promote soot formation in flames [6]. Since PAHs are believed to be soot precursors, the promotion of soot by chlorine has been attributed to increased PAH formation [7].

Previously, the oxidation and pyrolysis of CH₃Cl was studied in premixed laminar flames, in the presence of CH₄ by Karra and Senkan [8], and more recently by Wang et al. [9], where the emphasis was on the formation CO, CO₂ and C₂ products. Related studies in flow reactors were also conducted [10-14]. Probably the most relevant prior work is the jet stirred tank/plug flow reactor study of Marr et al. [15], where the effects of post flame additions of CH₃Cl on PAH formation in ethylene combustion was studied. These investigators used a water cooled probe to withdraw samples from the plug flow section of the reactor, and determined the amounts of PAH formed by analyzing the solvent extracts. The levels of most PAH formed in the presence of CH₃Cl were generally higher within the conditions investigated, with the exception of cyclopenta[cd]pyrene whose mole fractions were lower.

In this paper we compare the micro-structures of atmospheric pressure, laminar, premixed flat flames of CH₃Cl/CH₄ and CH₄ under similar, fuel rich conditions, with particular emphasis on aromatics and PAH.

EXPERIMENTAL

The atmospheric-pressure, premixed, laminar, flat-flames studied were stabilized over a 50 mm diameter porous bronze burner with an argon shroud [16]. Gases were introduced using high accuracy mass flowmeters (MKS, Burlington, MA). Flame sampling was accomplished using an air-heated microprobe that had 0.150 mm orifice at its tapered tip [16]. A quartz wool filter was also used to remove soot particles. The sampling system, which include the probe, the soot filter, transfer lines, and GC valves was maintained at about 300^oC and at sub-ambient pressures to minimize the condensation and/or adsorption of PAHs on surfaces.

The samples withdrawn from within the flame were then directly analyzed using a gas chromatograph/ mass spectrometer (Hewlett-Packard 5890/ 5972A, GC/MS) without pre-concentration [16].

Identification of species was accomplished by matching the GC retention times to pure components as well as the mass spectral fragmentation patterns to MS libraries. Species mole fractions were then determined using multi level calibration mixtures with an estimated accuracy of about ±15% for major species and ±20% for the remaining ones. For species for which calibration standards were not available the relative ionization cross section method was used [17]. This method has been shown to be accurate within a factor of 2 in our prior experiments [16]. Species profiles were generated by moving the burner relative to the stationary quartz probe with the aid of a vertical translator having a precision of about 0.01 mm.

Temperature profiles were measured by using a 0.075 mm Pt-Pt/13%Rh thermocouple with a bead diameter of about 0.15 mm after the concentration measurements. It was freshly coated by silica and vitrified to minimize catalysis, and was kept in the flame for as short time as possible to minimize soot accumulation. The clean thermocouple was inserted into the flame at a pre-determined location and the temperature reading was recorded promptly. Following the temperature measurement, the thermocouple was withdrawn and the accumulated soot was burned off using a small propane torch. The cleaned thermocouple was then reinserted into the flame for the measurement of the temperature at another position. The temperature profiles reported here were also corrected for radiation losses by assuming an emissivity of 0.2 for the bare and 0.5 for the soot coated thermocouple [18]. Thermocouple measurements were used to ascertain the absence of radial variations in flame properties.

Early in the experiments, we discovered that some of the 2 and 4 ring PAH species have been subject to surface catalyzed chlorinations at 300^oC sampling line temperatures. This problem was addressed by developing and using additional sampling procedures that minimize surface reactions and by modifying the results from the direct, on-line sampling procedure described above. In the first procedure, the temperature of the entire sampling system was reduced to about 120^oC. Although these experiments did not allow for the direct determination of PAH, as they readily adsorb on surfaces at 120^oC, we were able to confirm that the concentrations of major species and one-ring aromatics were not affected by surface reactions at 300^oC sampling temperatures.

In the second procedure, a cold trap system was developed to determine the levels of heavier PAH. In this case, the micro-probe was cooled by circulating room temperature air through it, i.e. about 25^oC. The probe was also connected to a 600 cc steel tank. Flame sampling was accomplished by evacuating the tank to about 10^-3 Torr using a mechanical vacuum pump, and by filling it with samples withdrawn from within the flame through the probe until the pressure in the tank reached 80 Torr. This insured the acquisition of the same amount of sample each time. The sampling time varied between 12-15 s depending on the flame and the distance above the burner. After the experiment, the probe and quartz wool filter (where most of the high molecular weight PAH were retained) were washed with methylene chloride. The liquid extracts were then concentrated to 2.5 cc and 5µl samples were injected to GC/MS for analysis. This process was repeated both for the CH₃Cl/CH₄ and the CH₄ flames. Since there are no complicating surface reactions in sampling CH₄ flames [16], PAH determined by direct gas analysis of this flame were used to calibrate the results of the cold trap experiments.

A helium-neon gas laser (Model 1508-0, Uniphase) was used to determine the relative soot volume fraction profiles in the flames. For this, the laser beam was passed through the center of the flame parallel to the burner surface and the intensity was monitored by a power meter to determine the transmitted light. The burner was then moved up and down to generate light transmission profiles, which are related to soot volume fraction profiles [4,19].

RESULTS AND DISCUSSION

The pre-combustion compositions and other parameters of the CH₃Cl/CH₄/O₂/Ar and the CH₄/O₂/Ar flames studied are listed in Table 1. As can be seen from this table, the two flames had the same equivalence ratio (2.5 by considering H₂O, CO₂ and HCl as the preferred combustion products) and carbon density (1.9x10^-5 mole/cc 1 atm, 298 K). We also tested these flames at the same equivalence ratio and argon dilutions, leading to closer peak temperatures and burning velocities, with similar results and conclusions described below.

In each flame, we were able to quantify over 50 species, including CO, CO₂, H₂O, H₂, HCl and C₁-C₁₈ hydrocarbons and their chlorinated analogs, including a large number of aromatics and PAH. These measurements covered a mole fraction range 0.25-1x10^-7. Space limitations prevent us from presenting all the measurements, thus we will emphasize here comparisons of major and trace aromatic and polyaromatic species. However, it is significant to note that regular, i.e. non-chlorinated, polycyclic aromatic hydrocarbons dominated the PAH abundance in the CH₃Cl/CH₄ flame. Although some chlorinated hydrocarbon by-products were formed, the levels of these CHCs were below 500 parts per million [20]. Species mole fraction profiles not reported here due to length restrictions can be obtained upon request from the authors.

A number of practical difficulties exist in the acquisition of accurate flame structure data from atmospheric pressure premixed flames. These difficulties include probe induced distortions in the concentration profiles due to flame attachment and spatial averaging especially in regions of steep concentration gradients, and the possible continuation of reactions in the sampling probe and transfer lines. We have expended considerable time and effort in the past to develop optimal flame sampling techniques and these are discussed in our earlier publications [16,20].

In Fig 1, the temperature and as well laser beam transmission profiles for the two flames are presented. Although we did not quantify the actual soot volume fractions, it is very clear from these data that the CH₃Cl/CH₄ flame was more sooting than the CH₄ flame. As can be seen from Fig 1 also, soot formation began at about 4 mm from the burner surface. In addition, the CH₄ flame was hotter than the CH₃Cl/CH₄ flame and peaked slightly earlier. Lower temperatures in the CH₃Cl/CH₄ flame can be attributed to more soot formationin this mixture.

In Fig 2 the mole fraction profiles for the major products, that include CO, CO₂, HCl, H₂, H₂O and C₂H₂ for the CH₃Cl/CH₄ flame (filled symbols) are compared to the CH₄ flame (open symbols). In this and subsequent figures, lines have been drawn through the data points to indicate trends. It should be noted that data points within 1-2 mm from the burner should be considered unreliable because of probe-burner interactions. The growth rate of soot has been suggested to be directly proportional to C₂H₂ concentration and surface area [21]. As can be seen from Fig 2, the levels of C₂H₂ were only slightly higher in the CH₃Cl/CH₄ post flame zone. The fact that these two flames had similar C₂H₂ concentrations yet different soot levels raises questions pertaining the role acetylene plays in soot and PAH formation.

As evident from Fig 2 also, the levels of H₂ and H₂O were substantially higher in the CH₄ flame than the CH₃Cl/CH₄ flame. This, however, is not surprising as the formation of HCl (ca. 18% in the post flame zone) ties up substantial amounts of hydrogen as it is the preferred combustion product for chlorine.

In Fig 3 and 4, the mole fraction profiles are presented for C₃-C₆ hydrocarbon intermediates that include C₃H₄(CH₂CCH₂), C₄H₂, C₄H₄, C₄H₆(CH₃CCCH), C₅H₆(cyclopentadiene), C₆H₂ (CHCCCCCH), and C₆H₄ (3-hexen-1,5-diene). As can be seen from these figures, the concentrations of species with C/H ratios less than 1 (C₃H₄, C₄H₆ and C₅H₆) were higher in the CH₄ flame, and concentrations of all the species with C/H ratio larger than or equal to 1(C₄H₂, C₄H₄, C₆H₂, and C₆H₄) were higher in the CH₃Cl/ CH₄ flame. These results are consistent with the availability of more hydrogen atoms in the CH₄ flame. C₄H₂ and C₃H₄ were the most abundant species in this group for both flames. As seen in Fig 4, C₄H₄ mole fractions were a factor of 10 lower than those of C₄H₂, and those for C₄H₆ were lower by another factor of about 6.

In Fig 5, the mole fraction profiles of 1-ring aromatic hydrocarbons formed are presented. Benzene was the most abundant in this group reaching levels at high as 200-800 ppm, followed by phenylacetylene (20-50 ppm), toluene (3-8 ppm) and styrene (1-5 ppm). The most significant aspect of the results presented in Fig 5 is the fact that above 4 mm from the burner surface, where soot formation also sets in (Fig 1), the concentrations of all of these aromatics in the CH₃Cl/CH₄ flame were significantly lower than those in the CH₄ flame, by at least a factor of two. This is a surprising result because higher levels of aromatics are generally associated with more sooting flames [1,3,4].

The mole fraction profiles of PAH also exhibited trends similar to the 1-ring aromatics described above, and that the PAH levels in the CH₃Cl/CH₄ flame were substantially lower than the CH₄ flame. In Fig 6, the concentration profiles of 2-ring PAH are presented. Naphthalene (C₁₀H₈) was the most abundant species in this group with mole fractions in the CH₄ flame being nearly a factor of 4 higher than the CH₃Cl/CH₄ flame. Naphthalene is followed by acenaphthylene (C₁₂H₈), methyl naphthalene (C₁₁H₁₀), methyl indene (C₁₀H₁₀) and fluorene (C₁₃H₁₀), with similar post-flame ratios between the two flames. It should be noted that for the case of acenaphthylene, its mole fraction profile in the CH₃Cl/CH₄ flame was determined both by direct sampling and analysis (filled triangles) and by the cold trap sampling and solvent extraction procedure (dashed lines). The latter was necessitated by the chlorination of acenaphthylene in heated sampling probe and transfer lines.

In Fig 7, the mole fraction profiles of 3- and 4-ring PAH are presented. These PAH include phenanthrene (C₁₄H₁₀), pyrene (C₁₆H₁₀), fluoranthene (C₁₆H₁₀), cyclopenta[cd]pyrene (C₁₈H₁₀), and benzo[ghi]fluoranthene (C₁₈H₁₀). As evident from this figure, concentration profiles of these PAH also followed the same pattern, with mole fractions in the CH₃Cl/CH₄ flame being significantly lower than the CH₄ flame. In fact the differences in mole fractions between the two flames were as high as a factor of 10 for the case of phenanthrene. When CH₃Cl flame samples were first directly analyzed, neither pyrene nor cyclopenta[cd]pyrene were detected. Instead, fluoranthene was the only C₁₆H₁₀ isomer observed. In addition, significant levels of chlorinated pyrenes were also noted in samples analyzed directly [20]. The analysis of cold trap samples, however, indicated the presence of significantly higher levels of pyrene than fluoranthene, with fluoranthene levels virtually identical to those determined using the heated sampling probe. These results indicate that the heated sampling probe and/or transfer lines must have selectively chlorinated pyrene, but not fluoranthene. The reasons for these intriguing results are not clear and are being investigated. The mole fraction profile of pyrene determined by the cold trap technique is indicated by the dashed line in Figure 7. Chlorine also suppressed the formation of cyclopenta[cd]pyrene, to non-detect levels, a result which is consistent with previous studies [15].

Although we are not in a position to quantitatively predict the surprising result that chlorine reduces PAH levels while promoting soot formation, several explanations can be offered. First, and most likely scenario is that chlorine accelerates PAH conversion to soot. Second, chlorine may inhibit PAH formation in the first place and promote soot formation via reactions that do not involve PAH.

In the first case, it is has been recognized that H radicals play a crucial role in activating PAH and soot [3-5] in flames. The basic reactions can be represented by:

PAH + H = PAH* + H₂ (1)

soot + H = soot* + H₂ (2)

where * indicates an activated, i.e. radical, entity. In flames containing chlorine, additional reactions involving Cl radicals must also be considered:

PAH + Cl = PAH* + HCl (3)

soot + Cl = soot* + HCl (4)

The reactions (1)-(4) have equilibrium constants (K_p) in the range 1-10 at flame temperatures, thus the partial pressures of H₂ (P_H2) and HCl (P_HCl) would have a considerable effect on the fraction of PAH and soot that will be activated. Heterogeneous reactions (2) and (4) require special attention; in addition to reaction thermochemistry and kinetics, the rates of diffusion of species to soot surfaces, both external and internal [5], and sticking probabilities [22] must be considered in assessing their importance. Reaction (2) should remain dominant even in the CH₃Cl/CH₄ flame because of the higher diffusivities of H and H₂ relative to Cl and HCl. On the other hand, Cl and HCl would be favored species based on sticking potentials.

Once PAH* and soot* are generated, PAH can be converted to soot via reactions such as:

PAH* + soot_i = soot*_i+PAH (6)

PAH + soot*_i = soot*_i+PAH (7)

PAH* + soot*_i = soot_i+PAH (8)

As can be seen in Fig 2, post-flame P_H2 in the CH₃Cl/CH₄ and CH₄ flames were 0.28 and 0.12 atm, respectively. Consequently, reactions (1) and (2) should proceed to right more in the CH₃Cl/CH₄ flame than in the CH₄ flame. In this regard, it is important to note that chlorine only minimally alters the levels of H radicals in the post flame zone, because the following thermoneutral reaction:

Cl + H₂ = HCl + H (9)

whose K_p=1, is rapidly equilibrated. In addition, reactions (3) and (4) also would be major contributors to PAH and soot activation in the CH₃Cl/CH₄ flame [7].

Chlorinated hydrocarbons also decompose at temperatures lower than the analogous hydrocarbons, because of the relatively weak C-Cl bond [6]. Consequently, they can form a larger number of soot nuclei early in the flame, leading to higher surface area soot aerosols that can more efficiently scavenge the PAH. All of these considerations suggest that PAH are rapidly incorporated into soot in the CH₃Cl/CH₄ flame.

In the second case, CHCs may also suppress PAH formation because they are excellent H radical scavengers [6,23]. For example, the reaction:

CH₃Cl + H = CH₃ + HCl (10)

can effectively compete with reaction (1) or PAH precursors in the cooler, i.e. earlier, parts of the flame, thereby decreasing the rate of formation of PAH. However, if this mechanism is true, then soot growth must occur via reactions that do not involve PAH. The direct addition of smaller hydrocarbons, such as C₂H₂, has been proposed to be such a soot growth species [3-5]. However, as noted before (see Fig. 2), C₂H₂ levels in the CH₃Cl/CH₄ flame were nearly the same in the post-flame zone. Finally, oxidation cannot account for the lower PAH levels observed in the CH₃Cl/CH₄ flame, because of the well known flame inhibition characteristics of CHCs [23].

In summary, the micro-structures of fuel-rich CH₃Cl/CH₄ and CH₄ flames, determined under similar equivalence ratio, carbon density and argon dilution, indicate the formation of significantly lower concentrations of PAH in the CH₃Cl/CH₄ flame than in the CH₄ flame, although the former was more sooting. These results are both surprising and significant because they invalidate the generally held belief that increased soot formation is associated with higher PAH levels. The measurements provide valuable new data for the development and verification of detailed reaction mechanisms, and suggest that the major effect of chlorine is the rapid incorporation of PAH into soot.

References

[1] Longwell, J.P., 19th Symposium (Int'l) on Combustion, p.1339, The Combustion Institute 1982.

[2] Klaassen, C.D., Amdur, M.O., Doull, J., "Casaret and Doull's Toxicology", 2nd Ed., Macmillan, New York 1986.

[3] Frenklach, M., Clary, D., Gardiner, Jr., W.C., and Stein, S.E., 20th Symposium (Int'l) on Combustion, p.887, The Combustion Institute 1984; Frenklach, M., and Warnatz, J., Combust. Sci. Tech., 51, 265 (1987). Kazakov, A., Wang, H., and Frenklach, M., Combust. Flame, 100, 111 (1995).

[4] Bockhorn, H., (ed), Soot Formation in Combustion: Mechanisms and Models, Springer Verlag, Berlin, 1994.

[5] Mauss, F., Schafer, T., and Bockhorn, H., Combust. Flame, 99, 697 (1994).

[6]Senkan, S.M., Robinson, J.M., and Gupta, A.K., Combust. Flame, 49, 305 (1983).

[7] Mckinnon, J.T., and Howard, J.B., Combust. Sci. Tech.,74, 175 (1990);Frenklach, M, Combust. Sci. Tech., 74, 283 (1990).

[8]Karra, S.B., Gutman, D., and Senkan, S.M., Combust. Sci. Tech.,54, 333 (1987).

[9] Wang, L., Jalvy,P., and Barat,R.B., Combust. Sci. Tech., 97, 13 (1994).

[10] Mitchell, T.J., Benson, S.W., and Karra, S.B., Combust. Sci. Tech., 107, 223 (1995); Weissman, M., and Benson, S.W., Prog. Energy Combust. Sci., 15, 273 (1989); Weissman, M., and Benson, S.W., Int. J.Chem. Kinetics, 16, 307 (1984).

[11] Granada, A., Karra, S.B., and Senkan, S.M., Ind. Eng. Chem. Research, 26, 1901 (1987).

[12] Fisher, E.M. and Koshland,C.P., Combust. Flame, 90, 185 (1992).

[13] Ho, W.P., Yu,Q.R., and Bozzelli, J.W., Combustion Sci. Tech., 85, 23 (1992).

[14] Hung, S., and L. D. Pfefferle, Combust. Sci. Tech.,87, 91 (1993).

[15]Marr, J.A., Allison, D.M., Giovane, L.M., Yerkey, L.A., Monchamp, P., Longwell, J.P., and Howard, J.B.,Combust. Sci. Tech.,85, 65 (1992).

[16] Castaldi, M.J., Vincitore, A., and Senkan, S.M., Combust. Sci. Tech., 107, 1 (1995).

[17] Fitch, W.L., and Sauter, A.D., Anal. Chem., 55, 832 (1983).

[18] Bradley, D., and Matthews, K.J., J. Mech. Eng. Sci., 10, 299 (1968).

[19]Prado, G., J. Jagoda, K. Neoh, and J. Lahaye, 18th Symposium (Intl) on Combustion, p.1127, The Combustion Institute, 1981.

[20] Huang, J.W., and Senkan, S.M., Enviro. Sci. Tech., submitted.

[21] Harris, S.J., and Weiner, A., Combust. Sci. Tech.,72, 67 (1990).

[22] Boudart, M., and Djega-Maridossou, G., Kinetics of Heterogeneous Catalytic Reactions, Princeton University Press, New Jersey 1984.

[23] Valeiras, H., Gupta, A.K., and Senkan, S.M., Combust. Sci. Tech., 36, 123 (1984).

Acknowledgements

This research was supported, in part, by the National Science Foundation CTS-9311848, the US Environmental Protection Agency Grant No:R819178-01, and the UCLA Center for Clean Technology. We also thank I. Gargurevich for useful comments.

1. Relative transmittance and temperature profiles along the flames. Dashed lines are temperature profiles corrected for radiation.

2. Comparison of mole fraction profiles for major products.

3. Comparison of mole fraction profiles for C₄ aliphatic species.

4. Comparison of mole fraction profiles for C₃, C₅ and C₆ aliphatic species.

5. Comparison of mole fraction profiles for 1-ring aromatic species.

6. Comparison of mole fraction profiles for 2-ring aromatic species. Dashed line is the acenaphtylene profile determined by the cold-trap sampling procedure.

7. Comparison of mole fraction profiles for 3- and 4-ring aromatic species. Dashed line is the pyrene profile determined by the cold-trap sampling procedure.