Investigate Methane Role On Pah And Soot

Combustion and Flame 134 (2003) 249 –260

Investigating the role of methane on the growth of aromatic hydrocarbons and soot in fundamental combustion processes J. F. Roeslera,*, S. Martinota, C.S. McEnallyb, L.D. Pfefferleb, J.-L. Delfauc, C. Vovellec a

IFP, 1 et 4 av. de Bois-Pre´au, 92852 Rueil-Malmaison Cedex, France Department of Chemical Engineering and Center for Combustion Studies, Yale University, New Haven, CT 06520-8286 USA c Laboratoire de Combustion et Syste`mes Re´actifs, CNRS, 45701 Orle´ans Cedex 2, France

b

Received 30 November 2002; received in revised form 24 March 2003; accepted 1 April 2003

Abstract Experimental results are presented on the effect of methane content in a non-aromatic fuel mixture on the formation of aromatic hydrocarbons and soot in various fundamental combustion configurations. The systems considered consist of a laminar flow reactor, a laminar co-flow diffusion flame burner, and a laminar, premixed flame burner, all of which operate at atmospheric pressure. In the flow reactor, the experiments are performed at 1430 K, constant C-atom flow rates, 98% nitrogen dilution, C/O ratio ⫽ 2, and with fuel mixtures consisting of ethylene and methane. The diffusion flames are performed with fuel mixtures of methane and ethylene diluted in nitrogen to maintain a constant adiabatic flame temperature. The premixed flame experiments are performed with n-heptane and methane mixtures at a C/O ratio ⫽ 0.67 with nitrogen-impoverished air. The results show the existence of synergistic chemical effects between methane and other alkanes in the production of aromatics, despite reduced acetylene concentrations. This effect is attributable to the ability of methane to enhance the production of methyl radicals that will then promote production channels of aromatics that rely on odd-carbonnumbered species. Benzene, naphthalene, and pyrene show the strongest sensitivity to the presence of added methane. This synergy on aromatics trickles down to soot via enhanced inception and surface growth rates by polycyclic aromatic hydrocarbon condensation, but the overall effects on soot volume-fraction are smaller due to a compensating reduction in surface growth from acetylene. These results are observed under the very fuel-rich environments existing in the flow reactor and diffusion flames. In the premixed flames, however, instabilities did not permit investigation of conditions with sufficiently high equivalence ratios to perturb the aromatic and soot-growth regions. © 2003 The Combustion Institute. All rights reserved. Keywords: Soot; Aromatic hydrocarbons; PAH; Methane; Combustion; Flow reactor; Diffusion flame; Premixed flame

1. Introduction The molecular growth-processes leading to polycyclic aromatic hydrocarbons (PAH) and soot forma* Corresponding author. Tel: ⫹(33)1-47-52-5811; fax: ⫹(33)1-47-52-7069. E-mail address: [email protected] (J.F. Roesler).

tion are complex. Early mechanistic models [1,2] suggested that the first aromatic ring was formed by the addition of acetylene to C4 radicals and that the key growth process toward increasingly larger PAH occurred by continuous addition of C2H2. Bi-aryl formation by aromatic-ring addition was considered feasible only in the case of aromatic fuels. This type of pioneering model was able to reproduce most of

0010-2180/03/$ – see front matter © 2003 The Combustion Institute. All rights reserved. doi:10.1016/S0010-2180(03)00093-2

250

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

the major PAH and soot trends observed experimentally in combustion processes [1,3,4]. However, it neglected other reaction pathways that are increasingly shown to be of importance. In 1990, Miller and Mellius [5] published work that suggested that benzene could be formed, in large part, from the re-combination of the odd-carbonnumbered, resonantly-stabilized propargyl radicals. While this specific reaction is now well-recognized and has been implemented in many recent kinetic models [6-9], there is growing evidence that many other such radical species are involved in the production of at least the smaller PAH. This evidence is found in various theoretical analyses of reaction channels and in experimental flame and modeling studies [10 –14]. If these reactions are indeed important, then adding methane to the PAH and soot-forming regions in combustion systems should, under some conditions, promote their growth. The premixed flame results of Senkan and Castaldi [15] that show a methane, laminar, premixed, flat flame to produce more benzene and PAH than a similar ethane flame confirm this line of thought. More direct evidence of this effect of methane was obtained in flow-reactor studies [16,17]. The addition of methane was found to enhance the formation of aromatics and soot in n-heptane oxidative pyrolysis. Detailed reaction mechanisms from the literature were found to reproduce these effects, provided PAH formation was appropriately described with odd-carbon-numbered radical species. The present work further investigates experimentally the influence of methane as a parameter to be considered in the formation of soot and PAH. Our goal was to further quantify the effect, under flowreactor conditions, and to determine whether or not these were applicable to classic, laminar, premixed and diffusion flame conditions.

The product samples are quenched and collected along the center-line axis by means of an oil-cooled probe at 180°C. Only 20% of the total flow is sampled; therefore, the measurements correspond to the products along the center-line and not to the bulk mean. At the probe exit, the soot is filtered at 180°C, and condensable species are then trapped. In a first pass, the trap is at ambient temperature, and the low-molecular-weight gas-phase species, up to benzene, are analyzed on-line with a Fourier-transform infrared spectrometer (FTIR). The gas-phase species are then collected for subsequent analysis by gas chromatography (GC) with a flame ionization detector (FID) in order to more accurately quantify the hydrocarbon gas-phase species with up to eight carbon atoms. The species quantified solely by FTIR are CO, CO2, and H2O. The higher-molecular-weight species are collected on a second pass, with the trap at dry-ice temperature (⫺70°C) that is then washed with CH2Cl2. The detailed speciation of the PAH from the trap is obtained by GC-FID and GC-mass spectrometry (MS) analysis. All other wetted surfaces (probe, Teflon liner, and soot filter) are washed ultrasonically with CH2Cl2. Analysis of the collected solutions showed ⬍ 5% loss of any one of the reported PAH to these surfaces. The total masses of condensable species (hereafter referred to as “tar”) of molecular weight, corresponding roughly to masses greater than acenaphthylene, and the mass of soot (non-dichloromethane-soluble, condensable species) are determined gravimetrically. In the experiments presented here, the fuel was initially C2H4 that was gradually substituted with CH4 at a constant total-carbon molar content of 3%. A mixture parameter characterizing the initial fuel composition is defined as:

2. Experimental set-ups

␤⫽

2.1. The flow-reactor system The flow-reactor device and the analytical techniques for quantifying the composition of the sampled products have been described in detail previously [17,18]. The reactor is made of quartz, with a 30-mm inside diameter. It is inserted in a three-zone furnace. The premixed reactants are rapidly heated in the first zone through a coiled 1.5-m-long tube with an inside diameter of 4 mm. The flow enters the larger-diameter test section and is laminarized by passing through a porous quartz disk. The test section is at atmospheric pressure and is isothermal within 5 K throughout a distance of 35 cm.

X CH4 X CH4 ⫹ 2X C2H4

(1)

and represents the fraction of fuel carbon injected as methane. This parameter was varied from 0 to 0.5. The oxygen content was also held constant to maintain a C/O ratio of ⬃2.0, leading to an equivalence ratio varying slightly between 6 and 7 due to added hydrogen in the presence of methane. The experimental conditions investigated were for a temperature of 1430 K, atmospheric pressure, and a total flow rate of 0.54 slpm (standard liters per minute) with nitrogen as the carrier gas. The product samples were collected at a fixed position of 0.25 m downstream of the porous quartz disk, corresponding to a bulk mean residence time of 0.42 s.

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

251

2.2. The laminar co-flow diffusion flame system

Table 1 Volumetric flowrates for the laminar diffusion flames*

The laminar co-flow diffusion flame burner and the analytical techniques have been described in detail previously [19]. The flames are generated with an atmospheric pressure, axisymmetric, co-flowing burner. The fuel mixture flows from an un-cooled 12-mm-diameter vertical brass tube. The air flows between this tube and a surrounding 108-mm insidediameter chimney. Temperatures are measured with an un-coated 125-␮m wire-diameter/260-␮m junction-diameter type R (Pt-Pt/13% Rh) thermo-couple. Further details are provided elsewhere [20]. The relative and absolute uncertainties in the results are estimated to be ⫾5 and ⫾50 K. The temperatures are corrected for radiation heat-transfer effects using procedures described in the paper cited above. The reaction gases are sampled by means of a quartz micro-probe. Analyses are then performed online, with a commercial electron-impact/quadrupole mass spectrometer (EQMS) and a custom-built photo-ionization/time-of-flight mass spectrometer (PTMS) [19]. The calibration mole fractions for methane, acetylene, and benzene are determined by direct comparison with room temperature mixtures of known composition, while the rest are determined by assuming that their PTMS sensitivity is the same as for benzene. The concentrations of C2H4, CO, and N2 cannot be measured due to mutual molecular-mass interference. The relative uncertainties in the hydrocarbon mole fractions are estimated to be ⫾10%, and the absolute uncertainties are estimated to be ⫾20% (methane, acetylene, and benzene) or ⫹100/⫺50% (all others). These uncertainties are acceptable, because we are primarily interested in how the hydrocarbon mole fractions change as the fuel composition is altered. Soot-volume fractions are measured using laserinduced incandescence (LII) The incandescence is excited at 1064 nm, as opposed to a visible wavelength, in order to eliminate interferences from PAH fluorescence. The incandescent signal is collected orthogonal to the excitation beam and is monitored at wavelengths from 400 to 450 nm. This detection range discriminates against C2 fluorescence. The specific details are similar to those provided elsewhere [21]. The relative uncertainty in the measurements is estimated to be ⫾10%. The experimental conditions investigated are summarized in Table 1. In the present investigation, the composition of the fuel consisted of mixtures of CH4, C2H4, Ar, and N2. As with the flow-reactor experiments, the relative concentrations of CH4 and C2H4 were identified by the mixture parameter ␤ that was varied over its full range. The concentration of

␤

N2

CH4

C2H4

Ar

0.00 0.03 0.05 0.08 0.11 0.14 0.18 0.21 0.25 0.29 0.33 0.38 0.43 0.48 0.54 0.60 0.67 0.74 0.82 0.90 1.00

500 484 468 452 436 417 397 378 358 336 313 290 267 239 211 182 154 124 87 46 0

0 11 23 35 48 61 75 90 105 121 138 156 176 195 216 241 271 309 346 384 421

220 213 206 199 192 183 175 166 158 148 138 128 118 105 93 80 68 54 38 20 0

7.27 7.15 7.04 6.93 6.83 6.68 6.53 6.40 6.28 6.10 5.94 5.79 5.66 5.44 5.25 5.08 4.98 4.92 4.77 4.54 4.25

* (units are cm3/min at STP).

N2 was adjusted to maintain a constant adiabatic flame temperature of 2230 K. Argon was added to the fuel in order to match its concentration in air. It was also used as an internal standard to correct the concentration measurements for changes in the sample gas-flow rate generated by thermal effects and soot deposition. The air flow around the fuel jet was maintained constant at 44.0 slpm. The total fuel jet flow rates were specifically chosen to make the maximum center-line temperature occur at a height of 71.4 mm. The temperature profiles along the center-line axis are then quite similar for the various conditions observed here. Since these are buoyancy-driven flames, the center-line gas residence times are independent of this flow rate. Therefore, the time-temperature profiles are nearly similar for all flames. Center-line profiles were measured for each species in a subset of the flames in Table 1; then for each species, a height above the burner was identified where its concentrations in each profile was within 10% of the maximum; and then measurements of that species were made at that height in all of the flames listed in Table 1. Thus, the final results are the maximum concentrations of each species as a function of ␤. The exception is methane, for which the measurement represents the maximum only for the lowest ␤ values, whereas it represents the residual from the fuel at the higher ␤ values.

252

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

Table 2 Initial conditions for the premixed flames Flame

1

2

n-C7H16 (mole %) CH4 (mole %) O2 (mole %) N2 (mole %) ␾ T1(K) C/O P (atm) U1 (cm/s at 450 K)

5.3 0.0 27.5 67.2 2.12 450 0.67 1 6.3

4.6 3.5 26.7 65.2 2.16 450 0.67 1 6.5

2.3. The laminar, premixed flame system The laminar, premixed flame burner and corresponding analytical techniques have been described in detail elsewhere [22]. The fuel is vaporized and mixed with oxygen and nitrogen in a pre-burner chamber. The burner surface consists of a porous plate made of 0.5-mm-diameter holes uniformly distributed over an area 40 mm in diameter at atmospheric pressure. The initial gas temperature is fixed by heating the burner surface and up-stream feed line to 450 K, a temperature which prevents fuel condensation. Gas samples are collected by means of a quartz micro-probe. The probe moves along the center-line axis of the flame. The sampling orifice of the probe is 0.1 mm, and the down-stream pressure is maintained at 1 kPa. The sampled gases are then stored in glass bottles for subsequent analysis by GC. The hydrocarbon species were quantified with an Al2O3/KCl column coupled to a FID. Quantification of N2, O2, H2, and CO was performed with a molecular-sieve column coupled to a thermal-conductivity detector, while for CO2 and H2O, a Porapak-Q was used. Aromatic species heavier than benzene were not measured with these techniques. Soot-volume fractions were measured by He/Ne laser-light attenuation at 670 nm, with a soot refractive index of 1.57– 0.46i [23]. The experimental conditions investigated are summarized Table 2. Two flames are compared, both at the same pressure and temperature. The difference is that in one case, the fuel is only n-heptane (␤ ⫽ 0), whereas in the second, ⬃ 10% of the total carbon is injected in the form of methane (␤ ⫽ 0.1). Flames with larger fractions of methane were unstable with the given N2/O2 mixture composition. The C/O is maintained constant, which slightly raises the equivalence ratio from 2.12 to 2.16. Oxygen and nitrogen flow rates are also maintained. The effects on species concentration profiles and on soot production are then observed.

Fig. 1. Flow-reactor concentrations of CO, CO2, and H2O measured as a function of the initial mixture parameter ␤. Lines are curve fits to the data.

3. Results and discussion 3.1. The flow-reactor system Figures 1 through 7 show the evolution of species concentration as a function of the fuel-mixture parameter ␤, as measured at a point 25 cm down-stream from the porous quartz disk, corresponding to a bulk mean residence time of 0.42 s. The concentrations of CO, CO2, and H2O are nearly constant (Fig. 1). These species are formed predominantly before the porous disk in the pre-heat

Fig. 2. Flow-reactor concentrations of H2, C2H2, CH4, and pC3H4 measured as a function of the initial mixture parameter ␤. Lines are curve fits to the data.

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

Fig. 3. Flow-reactor concentrations of C2H4, C4H4, C4H2, and aC3H4 measured as a function of the initial mixture parameter ␤. Lines are curve fits to the data.

253

Fig. 5. Flow-reactor concentrations of the poly-aromatic species that show the strongest rise when the initial mixture parameter ␤ is increased. Lines are curve fits to the data.

regions on a very short time scale compared to that of PAH growth. The slightly varying concentrations express changes in the chemistry in this region as the fuel mixture composition changes. These variations are not very important, as these species have little interaction with the PAH and soot-growth processes. For detailed information regarding the evolution of species concentrations along the reactor axis, the reader is referred to reference [17].

The concentrations of CH4, C2H2, pC3H4, and H2 are displayed in Fig. 2. As would be expected, when the relative amount of methane in the fuel increases, so do the amounts of H2 and CH4 during the reaction. In contrast, the amount of acetylene decreases. As previously noted [17], the propyne in this system varies proportionally with the three former species due to a near partial equilibrium of the global reaction

Fig. 4. Flow-reactor concentrations of single-ring aromatic species as a function of the initial mixture parameter ␤. Lines are curve fits to the data.

Fig. 6. Flow-reactor concentrations of the poly-aromatic species that show a mild rise when the initial mixture parameter ␤ is increased. Lines are curve fits to the data.

254

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

diacetylene, its concentration decreases dramatically, much more so than acetylene, a trend that is coherent with partial equilibrium of the global reaction 2C2H2 º C4H2 ⫹ H2. The evolution of single-aromatic-ring hydrocarbons is shown in Fig. 4. Here, we observe a strong increase in benzene as ␤ increases. The maximum of this species was not reached in the present set of experiments. This increase is attributed to the enhanced production of propyne and allene, which would lead to larger quantities of propargyl radicals and accelerate the production of benzene via: C 3H 3 ⫹ C 3H 3ºC 6H 6

Fig. 7. Flow-reactor concentrations of tar, soot, and total PAH measured by GC analysis as a function of the initial mixture parameter ␤. The concentrations are expressed in terms of their carbon-atom equivalent. Lines are curve fits to the data.

CH 4 ⫹ C 2H 2ºH 2 ⫹ pC 3H 4

(2)

The relation yields: 共 pC 3H 4兲 ⫽

共CH 4兲共C 2H 2兲 K eq 共H 2兲

(3)

The present results confirm again that Keq has a value of 20 at 1430 K. In general, acetylene and hydrogen are present in larger proportions than methane in fuel-rich combustion products. Therefore, addition of methane will generate a larger relative concentration change of itself than of H2 and C2H2. Furthermore, at iso-carbon content, a change in methane is twice that of acetylene. These effects will promote propyne formation. The concentrations of ethylene, allene, vinylacetylene, and diacetylene are displayed in Fig. 3. For ethylene, its concentration increases by a factor of 2, despite the reduced amount injected. This result is due to the fact that, at the extent of reaction observed, ethylene is being produced primarily as a product in the decomposition of the larger quantities of methane. However, at an earlier reaction time, for a position only 1 cm down-stream from the quartz disk, the concentration of C2H4 rises only by 10%, to a value near 1,000 ppm. Thus, in the earlier oxidation region, methane acts more as an inhibitor in the decomposition of ethylene. The allene profile follows the trend of its propyne isomer. Vinylacetylene has a profile that remains rather flat (only a 20% increase) and follows closely a partial-equilibrium tendency with the global reaction C2H4 º C2H2 ⫹ H2. As for

(4)

In contrast, phenylacetylene increases much less and attains a maximum at ␤ ⫽ 0.25. This is easily attributable to the reduced C2H2 and enhanced H2 concentrations that eventually off set the effect of the increased benzene in the growth steps. Toluene, on the other hand, increases by a factor of 5, due to the fact that it further feeds on methyl radicals in its formation process from benzene. The PAH shown in Fig. 5 and Fig. 6 display profiles of two types. The examples of the first type, naphthalene and pyrene, show trends similar to benzene, with an increased growth of about a factor of 2.5. The examples of the second type, acenaphthylene, fluoranthene, phenanthrene, and C18H10 (this curve most likely represents cyclopenta(c,d)pyrene concentrations but possibly contains other indistinguishable minor isomers), show trends more like that of phenylacetylene. This suggests that there are those PAH whose growth depends more directly on the presence of odd-carbon-numbered species and those PAH whose growth depends more on even-carbonnumbered hydrocarbons. Obviously, the cyclopenta aromatics and acetylenated aromatics require the addition of acetylene, and these are found in the PAH category showing lesser growth. It is difficult to assess what the dominant reaction pathways are exactly; however, some conclusions may be reached as to what some pathways are not, particularly regarding acetylene addition. The main path to naphthalene is not likely to be from successive C2H2 addition reactions to phenyl, as in the hydrogen-abstraction-carbon-addition (HACA) mechanism [1]. The concentration of acetylene decreases as ␤ increases, and, assuming that the concentration of phenyl radicals varies proportionally to benzene, it is not possible for such reactions to induce the observed 2.5-fold increase in C10H8. Similar reasoning holds for a pathway involving C4H2 addition to phenyl. This leaves many other possible pathways, such as the addition of C4H4 to phenyl [6,24], the recombination of cyclopentadienyl radicals [10,12],

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

the addition of propargyl to toluene [11,19], or the addition of C2H2 to an o-xylyl radical, followed by dehydrogenation. This last pathway is the main source of naphthalene during the oxidation of oxylene [25]. It was found to dominate in methane oxidative pyrolysis at an equivalence ratio of 12 [26], based on a modeling study using the detailed reaction mechanism of Marinov et al. [12]. This pathway possibly becomes important at the higher values of ␤. As with naphthalene, it seems very unlikely for the production pathways of pyrene to involve exclusively HACA reaction steps that describe PAH growth by successive C2H2-addition reactions [1]. This holds, whether the HACA sequence is initiated from benzene, naphthalene, or phenanthrene. Given the decrease in acetylene content, it is difficult to expect the successive HACA steps to yield the observed 3-fold increase. In order to account for the strong increase in pyrene, other reaction pathways thus need to be considered, such as biaryl addition with isomerization [11,27–29] or addition reactions involving the cyclopentadiene moiety [10]. Regarding the formation of tar (Fig. 7), there appears to be either large uncertainty in the data or some contrasting trends as methane is added. The global trend is nonetheless an increase, consistent with the chromatographable PAH. For means of comparison, we have included the trend of the estimated total measured PAH mass obtained by chromatography and by assuming a response factor of unity for all FID unidentified peaks. A similar curve, but with the C12 and lighter species removed, is also shown. These lighter species evaporate during the weighting and are not accounted for in the tar [17]. The large difference between the tar and GC-PAH curves shows that there remains a large portion of unresolved soot-precursor matter, much as observed in premixed flame [30] and flow-reactor studies [31]. Optical analysis [32] reveals this matter to contain a significant fraction of aliphatic structures. This matter is therefore not presently taken into account in current soot-precursor models that consider only condensed PAH. Despite the large increase in tar and PAH, there is no observed enhanced soot formation. This result contrasts with previous experiments that used n-heptane instead of ethylene [17], which gave a 2-fold increase in soot production at ␤ ⫽ 0.1. While the exact reason for these differences has not yet been assessed with certainty, we speculate that they originate from changes in soot induction times that are affected by residual oxygen from the oxidative region. The amount of this oxygen (under 500 ppmv) and the impact of methane on this quantity vary with the initial fuel type. In addition, the oxidative chemistry is fast, and the conditions under which it occurs

255

Fig. 8. Laminar, co-flow diffusion-flame maximum temperature and soot-volume fraction along the center-line axis as a function of the mixture parameter ␤.

are not well-controlled, as it takes place before entrance of the gases into the test section. A more thorough investigation would require gathering sootconcentration profiles along the reactor axes under all conditions. While the present data show no increase in soot, they show no decrease either, a fact that would be anticipated by mere observation of acetylene and hydrogen concentration changes. This result suggests that soot growth under the observed conditions occurs from the combination of enhanced inception and PAH surface addition, compensated by reduced acetylene surface addition. To conclude this section, these flow-reactor data have confirmed earlier results, in which methane was mixed with n-heptane [17] at a fixed ␤ of 0.1. Here, with mixtures of C2H4 and CH4, we observe again that increasing the proportion of methane, with 0 ⬍ ␤ ⬍ 0.5 at constant carbon content, generates a pronounced increase in the formation of aromatic species. The data are rich in information for testing PAH-formation pathways. They show that the formation of PAH up to at least 4 aromatic rings is not always dominated by acetylene addition. Having assessed these effects in ideal flow-reactor conditions, we next investigate their relevance to practical flame conditions. 3.2. Diffusion-flame results The laminar, co-flow, diffusion-flame system represents a system of more practical relevance than the flow reactor. The parameter of prime interest is the maximum soot-volume fraction, which is shown in Fig. 8 as a function of ␤. An overall maximum is reached at ␤ ⫽ 0.4. This variation in soot loading is observed to have the reverse effect on the maximum

256

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

Fig. 9. Laminar, co-flow diffusion-flame maximum species concentrations along the center-line axis as a function of the mixture parameter ␤: C2H2, CH4, and C3H4. Lines are curve fits of data.

measured temperatures, also shown in Fig. 8. As the soot loading increases, the temperature decreases due to thermal radiation losses. The soot profile shows the existence of a synergistic effect of methane, in the sense that for ␤ ⬍ 0.6, the values reached exceed those of either the diluted ethylene flame or the methane flame. As discussed below, this synergistic effect is at least partly attributable to chemical factors that are similar to those observed in the flow reactor. Measurements were also made in similar flames without any nitrogen dilution, and in that case, the soot-volume fractions decreased monotonically from the neat ethylene flame to the neat methane flame. However, the soot-volume fractions in the intermediate flames were still larger than the weighted contribution of the neat flames. This result suggests that the same synergistic effects were present but were outweighed by: 1) the decrease in flame temperature as methane was added; and 2) the very large difference in the sooting tendency of the two fuels. We now look at species concentration variations with ␤. The maximum center-line acetylene concentrations decrease as methane is added, as shown in Fig. 9, because the ethylene is being replaced by a fuel that forms acetylene less readily. In contrast, the C3H4 concentrations increase, because the increase in methane is proportionately larger than the decrease in acetylene. The benzene concentrations also increase (see Fig. 10), which shows the importance of propargyl radicals as benzene precursors, just as in the flow-reactor results. The growth of larger aromatic

Fig. 10. Laminar, co-flow diffusion-flame species maximum concentrations along the center-line axis as a function of the mixture parameter ␤: C6H6, C7H8, and C8H6. Lines are curve fits of data.

species and soot are then also promoted (see Fig. 8, Fig. 10, and Fig. 11). The results further suggest that acetylene addition to phenylacetylene is neither the only nor the dominant pathway for the production of naphthalene. As ␤ increases, we see that all even-carbon-numbered species have decreasing concentrations, and that species requiring an obviously direct addition of acetylene for their production have decreasing concentrations

Fig. 11. Laminar, co-flow diffusion-flame species maximum concentrations along the center-line axis as a function of the mixture parameter ␤: C10H8 and C12H8. Lines are curve fits of data.

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

Fig. 12. Laminar, co-flow diffusion-flame species maximum concentrations along the center-line axis as a function of the mixture parameter ␤: C4H4, C4H2, and C10H6. Lines are curve fits of data.

relative to their parent aromatic. Indeed, phenylacetylene and phenyldiacetylene (C10H6 in Fig. 13) decrease relative to benzene, and acenaphthylene decreases relative to naphthalene. Thus, the HACA mechanism may not be able to fully explain the increasing naphthalene. Analogously, neither can the reaction of vinylacetylene addition to phenyl that was recently suggested and incorporated in a detailed reaction mechanism by Appel et al. [6].

Fig. 13. Laminar, premixed flame species-concentration profiles: C7H16 and C2H4. Lines are curve fits of data. Flame 1: without CH4; flame 2: with CH4.

257

In an earlier study, some of the current authors have shown that propargyl addition to benzyl is a viable channel toward naphthalene production [19]. Here, this channel could well explain the observed growth in naphthalene, since toluene concentrations (Fig. 10) remain relatively flat, while C3H4 concentrations increase up to values of ␤ ⫽ 0.6. These two species should be indicative of the relative amounts of benzyl and propargyl. It is not clear, however, why the toluene profile is so flat. This contrasts strongly from the flow-reactor data, in which toluene displays the strongest increase of all aromatic species and illustrates that kinetic interpretation from these diffusion flames is rather speculative and needs the support of detailed, kinetic modeling or of other reactive systems. Finally, while the cyclopentadienyl pathway [10,12] may exist, the concentration of cyclopentadiene decreases monotonically by a factor 4 as ␤ increases. Such a trend tends to discredit the relevance of the above pathway in these co-flow diffusion flames and corroborates the observations of McEnally and Pfefferle [19]. In summary, the results show that adding methane to ethylene diffusion flames promotes soot formation via a synergistic, chemical mechanism that is similar to the one observed in the flow-reactor experiments. The results also confirm the need to consider PAH growth pathways that include odd-carbon-numbered species. 3.3. Premixed-flame results The premixed-flame experiments investigated two mixture conditions, one with pure n-heptane as fuel (␤ ⫽ 0), and the other consisting of a mixture of n-heptane and methane, with ␤ ⫽ 0.1. The choice of these conditions was motivated by earlier flow-reactor experiments with n-heptane [17]. The fact that ethylene is not used is a secondary consideration, as the present study aims at identifying the role of methane in various combustion configurations. Furthermore, the chemistry in premixed flames is such that n-heptane is rapidly consumed in the low-temperature region to form predominantly ethylene. This is observed in Fig. 13, which shows the rapid conversion of n-heptane into ethylene under both flame conditions. Although not measured specifically for these flames, previous studies [22] indicate that the temperature profile should reach a maximum at the point of ethylene disappearance, i.e., at 4 mm downstream from the burner. The soot measurements, extending to 20 mm down-stream from the burner, are given in Fig. 14. Overall, we see that there is a negligible effect of the presence of methane on the production of soot. Ob-

258

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

Fig. 14. Laminar, premixed flame soot-volume fractions. Lines are curve fits of data. Flame 1: without CH4; flame 2: with CH4.

servation of the measured intermediate gas-phase species shows that the concentration profiles of H2 and C2H2 are nearly unchanged by the presence of methane. The final concentration of H2 is, however, slightly higher in flame 2 as would be expected, since the C/O is preserved but the C/H ratio decreases. Figure 15 shows the profiles of methane and benzene. The methane in flame 2 generates a perturbation only in the low-temperature regions of the flame. Eventually, oxidation reduces the methane concentration to nearly the same value as in flame 1. Although the

methane concentration in flame 2 is considerably enhanced in the early flame region, the concentration of propyne is actually slightly reduced. This is explained by modeling analysis that shows propyne to derive predominantly from the decomposition of nheptane in this region. As a consequence, contrary to the diffusion-flame and flow-reactor results, the addition of methane slightly reduces the concentration of benzene. Since the post-flame reaction paths are similar, no large change in other aromatic species or soot concentrations are expected. The results relate well to experimental observations discussed by Glassman [33], which showed sooting limits in premixed flames at a fixed adiabatic temperature to be more related to the number of C-C bonds than to the fuel structure. This conclusion was explained by the fact that much of the fuel characteristics are destroyed in the oxidizing region before the sooting region. Harris and Weiner [34] also reached a similar conclusion regarding the minor role of fuel structure in premixed flames, provided the C/H ratio does not change significantly. For the present data, where the C/O ratios remain close to those of the sooting limits and the C/H ratio changes little, the same reasoning can be applied to explain the weak effect of methane substitution. A more distinct effect may possibly have been observed for mixtures at higher values of ␤, in which case, lessdiluted flames or the use of Ar for dilution would be required for stabilization purposes. Here, we focused on realistic fuel/air flames for practical considerations. Within stability limits of these flames, no enhanced sooting or aromatic-species production was observed.

4. Summary and conclusions

Fig. 15. Laminar, premixed flame species-concentration profiles: CH4 and C6H6. Lines are curve fits of data. Flame 1: without CH4; flame 2: with CH4.

The effect of methane content on the formation of PAH and soot production in a non-aromatic fuel mixture was investigated in three fundamental combustion processes, including a laminar flow reactor, a laminar co-flow diffusion-flame burner, and a laminar, premixed flame burner. The results varied significantly with the combustion system used. In the flow reactor, increasing the content of methane at constant carbon content showed PAH concentrations to increase. This increase reached a factor 2.5 within the range of conditions observed and could have possibly been larger at still higher methane contents. The total mass of heavy hydrocarbons, or tar, also rose, while the soot concentration stayed constant. In the diffusion flames, increasing the methane fraction of fuel carbon while decreasing that of ethylene also generated increases in PAH concentration

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

and soot, if the ethylene was nitrogen-diluted, that could be attributed to synergistic chemical effects. If the ethylene was not diluted with nitrogen, then the soot concentrations decreased monotonically, but synergistic effects were still evident. In contrast to the other two systems, the premixed flame results showed no synergy in the presence of methane. The maximum amount of methane that could be used and the maximum attainable equivalence ratio were limited, due to flame instabilities. For the conditions attained, the detailed flame structures suggested that the methane contained in the fuel was primarily consumed by the oxygen, leading to a virtually unperturbed flame in the PAH and soot-growth regions. The synergy of methane with other alkanes to produce PAH is attributable to the ability of methane to produce methyl radicals that will then promote production channels of aromatics that rely on oddcarbon-numbered species. This is clearly the case for benzene that is formed from the propargyl recombination reaction. As for further aromatic growth, the HACA mechanism could merely be enhanced by the increased formation of benzene. However, the data strongly suggest otherwise. Indeed, in conjunction with decreasing acetylene concentrations, the acetylenated or cyclopenta fused aromatics, which require direct addition of acetylene, show lesser growth than aromatics for which various growth channels involving odd-carbon-numbered species have been speculated in the literature. This is the case for naphthalene, for which proposed production channels have been the recombination of cyclopentadienyl radicals and the addition of benzyl and propargyl. Regarding soot, the synergy must arise from the enhanced PAH that act either as precursors or surface growth species in reactions that compensate for the lesser growth by acetylene addition. These results show the complexity of PAH chemistry and the important role of methyl radicals and of odd-carbon-numbered species in the growth process. These data should provide a good basis for testing detailed reaction mechanisms by specifically sensitizing the reactions with odd-carbon-numbered species. Clearly, the HACA mechanism does not fully describe the molecular growth process in many applications. From a practical point, the results suggest that methane is not an actual soot promoter in flame situations. However, it does interact synergistically with alkane fuels to produce more PAH and soot than would otherwise have been expected. Finally, these effects of methane are more likely to occur in diffusion flames than in premixed flames.

259

Acknowledgments Co-authors C.S. McEnally and L.D. Pfefferle gratefully acknowledge financial support from the National Science Foundation and the US Environmental Protection Agency.

References [1] M. Frenklach, D.W. Clary, W.C. Gardiner, Jr, S.E. Stein, Twentieth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1994, p. 887. [2] M. Frenklach, Twenty-Second Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 1075. [3] M. Frenklach, J. Warnatz, Combust. Sci. Technol. 51 (1987) 265. [4] H. Bockhorn, Soot Formation in Combustion—Mechanisms and Models. Springer Series in Chemical Physics 59, Springer-Verlag, Heidelberg, 1994. [5] J.A. Miller, C.F. Mellius, Combust. Flame 91 (1992) 21. [6] J. Appel, H. Bockhorn, M. Frenklach, Combust. Flame 1221 (2000) 122. [7] N.M. Marinov, W.J. Pitz, C.K. Westbrook, A.E. Lutz, A.M. Vincitore, S.M. Senkan, Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 881. [8] H. Richter, T.G. Benish, O.A. Mazyar, W.H. Green, J.B. Howard, Proceedings of the Combustion Institute, Vol. 28, The Combustion Institute, Pittsburgh, 2000, p. 2609. [9] A. D’Anna, A. Violi, A. D’Alessio, Combust. Flame 121 (2000) 418. [10] C.F. Melius, M.E. Colvin, N.M. Marinov, W.J. Pitz, S.M. Senkan, Twenty-Sixth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1996, p. 685. [11] M.B. Colket, D.J. Seery, Twenty-Fifth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1994, p. 883. [12] N.M. Marinov, W.J. Pitz, C.K. Westbrook, M.J. Castaldi, S.M. Senkan, Combust. Sci. Technol. 211 (1996) 116 –117. [13] N.M. Marinov, W.J. Pitz, C.K. Westbrook, A.E. Lutz, A.M. Vincitore, S.M. Senkan, Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 605. [14] A. D’Anna, A. Violi, Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 425. [15] S.M. Senkan, M. Castaldi, Combust. Flame 107 (1996) 141. [16] J.F. Roesler, X. Montagne, M. Auphan de Tessan, Chemosphere 42 (2001) 823. [17] J.F. Roesler, M. Auphan, de Tessan, Combust. Sci. Technol. 161 (2000) 245.

260

J.F. Roesler et al. / Combustion and Flame 134 (2003) 249 –260

[18] J.F. Roesler, Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 287. [19] C.S. McEnally, L.D. Pfefferle, Combust. Sci. Technol. 128 (1997) 257. ¨ .O ¨ . Ko¨ ylu¨ , L.D. Pfefferle, D.E. Ros[20] C.S. McEnally, U ner, Combust. Flame 109 (1997) 701. [21] C.S. McEnally, L.D. Pfefferle, A.G. Robinson, T.S. Zwier, Combust. Flame 123 (2000) 344. [22] A.E. Bakali, J.L. Delfau, C. Vovelle, Combust. Sci. Technol. 140 (1998) 69. [23] W.H. Dalzell, A.F. Sarofim, J. Heat, Transfer Trans ASME Ser. C 91 (1969) 100. [24] N.W. Moriarty, M. Frenklach, Proceedings of the Combustion Institute, Vol. 28, The Combustion Institute, Pittsburgh, 2000, p. 2563. [25] J. Emdee, K. Brezinsky, I. Glassman, Twenty-Third Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1990, p. 77. [26] J.F. Roesler, M. Auphan-de-Tessan, Twenty-Seventh Symposium (International) on Combustion, University of Colorado at Boulder, Boulder, CO, 1998, Work-inProgress Poster W4A05. [27] Sarofim A.F., Longwell J.P., Wornat M.J., Mukherjee J. In: Soot Formation in Combustion—Mechanisms

[28]

[29] [30]

[31] [32]

[33]

[34]

and Models, H. Bockhorn (ed.), Springer Series in Chemical Physics 59, Springer-Verlag, Heidelberg, 1994, p. 485. M.J. Wornat, A.F. Sarofim, A.L. Lafleur, TwentyFourth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1992, p. 955. J. Mukherjee, A.F. Sarofim, J.P. Longwell, Combust. Flame 96 (1994) 191. A. Ciajolo, A. D’Anna, A. Barbella, A. Tregossi, A. Violi, Twenty-Sixth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1996, p. 2327. J.A. Marr, L.M. Giovane, J.P. Longwell, J.B. Howard, A.L. Lafleur, Combust. Sci. Technol. 101 (1994) 301. A. Ciajolo, R. Barbella, A. Tregossi, L. Bonfanti, Twenty-Seventh Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 1481. I. Glassman, Twenty-Second Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1998, p. 295. S.J. Harris, A.M. Weiner, Combust. Sci. Technol. 31 (1983) 155.