Chemical Production Rates of Intermediate Hydrocarbons in a Methane/Air Diffusion Flame

Twenty-first Symposium(International) on Combustion/TheCombustion Institute, 1986/pp. 1057-1065

CHEMICAL PRODUCTION RATES OF INTERMEDIATE HYDROCARBONS IN A METHANE/AIR DIFFUSION FLAME

J. HOUSTON MILLER Department of Chemistry George Washington University Washington, DC 20052

W. GARY MALLARD AND KERMIT C. SMYTH Center for Fire Research National Bureau of Standards Gaithersburg, MD 20899

The net production and destruction rates for acetylene, diacetylene, butadiene, and benzene have been determined from measurements of species concentration, temperature, and velocity in a methane/air diffusion flame. In addition, profile measurements of the methyl radical are used to compute profiles of hydrogen atoms. These results provide an initial evaluation of proposed models of hydrocarbon chemical growth routes in diffusion flames, and indicate that the vinyl radical plays a key role in the formation of benzene. Evidence is also presented for acetylene participation in surface growth processes on small soot particles.

I. Introduction composition is not sufficient to predict the concentrations of intermediate hydrocarbons, The identification of the formation mecha- such as acetylene and benzene, and soot nism of large hydrocarbons produced during particles 17. Therefore, in this paper an analysis combustion is one of the most challenging is undertaken combining detailed species pro- problems in high temperature chemistry. De- file measurements, temperature, and convec- spite the complexity of chemical growth reactive velocity measurements to calculate produc- tions in flames, there have been significant tion and destruction rates for intermediate recent gains in our understanding. Extensive hydrocarbons. The experimental values for the profile measurements have led to a detailed net production/destruction rates are compared description of the chemical structure of pre- with estimated rates of chemical growth reac- mixed flames 1-13. These results have been tions in order to evaluate possible formation combined with the time-temperature history in routes of aromatic hydrocarbon species. one-dimensional flame systems to produce simple models for molecular growth. Modelling efforts have also been undertaken using shock II. Experimental results tube data 14-16. Most practical combustion systems are diffu- Detailed species profiles have been measured sion flames. Here, in contrast to shock tubes in an atmospheric pressure methane/air diffu- and premixed flames, our knowledge of the sion flame using both mass spectrometric and soot formation process is much less complete. optical techniques 17. A Wolfhard-Parker burn- This is due in part to the lack of data which are er was utilized, consisting of a central fuel slot required for a detailed modelling of diffusion and two adjacent slots for air. Two flame sheets flames and in part to the inherently more were formed at the fuel/air interfaces, and all complex interactions between transport and species profiles were symmetric about the chemical processes in these systems. burner centerline. Profiles of stable flame Recently, concentration measurements have species were collected with a direct-sampling been reported for molecular species, including mass spectrometer equipped with a quartz intermediate hydrocarbons, in. laminar meth- microprobe (orifice diameter ~140 ~.m). Cali- ane/air diffusion flames 17'18. Our work has bration of the major species (methane, oxygen, shown that a knowledge of the local chemical nitrogen, carbon dioxide, carbon monoxide,

1057 1058 COMBUSTION GENERATED POLLUTANTS water, and hydrogen) led to a mass balance 20 ' ' ' ' I .... I ' ' ' ' I .... across the flame (total mole fraction = 1.01 -+ 0.04). In addition, absolute concentrations of acetylene, butadiene, and toluene were determined using a direct calibration procedure. Profiles of methylacetylene (and/or allene), vinylacetylene, diacetylene, triacetylene, benzene, and naphthalene were also measured and calibrated indirectly 17. In a related series of measurements 19, a quartz microprobe was modified so that iodine vapor could be introduced into the inside of the ! J/ t l/: ',ltt, t,'I,H/, ', quartz tube. Sampled methyl radicals react quantitatively in the til~ of the scavenger probe to form methyl iodide z~ which can then be detected in the mass spectrometer. The optical measurements included laser- -0!0''' "?17' ' ' 0 .... ; .... 10 induced fluorescence,, multiphoton ionization, LATERAL POSITION, rnrn photodissociation, and Rayleigh scattering in- vestigations of the methane/air diffusion FIG. 1. Isothermal contours (solid lines) calculated flame 17'23. These studies have produced pro- from uncorrected thermocouple measurements and files of relative OH and methyl radical concen- streamlines (dashed lines) calculated from the velocity trations, as well as larger hydrocarbon species measurements for the methane/air flame. and small soot particles. Temperature profiles were measured with an 1.0 i .... r .... r .... uncoated 125 I~m diameter fine wire Pt/Pt- [CtHt] 10% Rh thermocouple. Corrections due to radia- o [C,H,]xS tion effects were estimated to be less than 7% at o [C=Ht] x3 the highest temperatures in this flame 17, while + lC4H=]xlO catalytic effects are expected to be small 24 and in g the opposite direction. For equilibrium and kinetic calculations discussed in Section IV, the x z original temperature data were corrected for the estimated radiation effect. Both the horizon- ~ 0.5 tal and vertical components of the convective velocity were determined using laser Doppler =, velocimetry. The gradients in concentration and =o temperature in this flame are steepest in the lateral direction. Therefore, profile data points were collected every 0.2 mm horizontally and every 2 mm in the vertical direction. 0.0 Figure 1 illustrates the temperature and -10 -5 0 5 10 velocity fields for this flame. Shown as solid LATERAL POSITION, mm lines are isothermal contours determined from the thermocouple profiles. Also shown are FIG. 2. Mass spectrometric profile measurements streamlines of convective velocity calculated of several minor species: acetylene, benzene, diacety- from the two measured velocity components. lene, and butadiene at a height of 9 mm above the The streamlines exhibit trajectories which origi- burner. nate in the lean region of the flame, cross the high temperature, primary reaction zones, and continue into the fuel-rich regions. III. Production Rate Calculation Figure 2 presents profiles collected for several intermediate hydrocarbons: acetylene, A. Procedure benzene, diacetylene, and butadiene. Peak concentrations for these species are 6200, 800, 570, A primary goal of this work is to apply kinetic and 1 l0 parts per million, respectively, at this modelling to a methane/air diffusion flame; height. Profiles obtained for the other interme- that is, to compare the predictions of models diate hydrocarbons all exhibit concentration which have been used to describe premixed maxima in the same region of the flame. flame and shock tube results with our diffusion PRODUCTION RATES IN DIFFUSION FLAMES 1059

flame data (see Section IV, below). A viable first term in Eq.2 (concentration-driven diffu- model must not only account for the steady sion) dominates the second (thermal diffusion), state concentrations in species involved in and the contributions of thermal diffusion to chemical growth processes, but also for the mass transport can be ignored. The horizontal chemical flux of these species in the flame. To temperature gradients in our flame are steep in make this comparison, the net chemical pro- certain locations (2000-3000 K/cm), and thus dttction and destruction rates for individual thermal diffusion cannot be dismissed. To criti- species have been computed from concentra- cally evaluate the importance of thermal diffu- tion, temperature, and velocity measurements, sion, kr has been calculated as a function of as described below. temperature and composition for a variety of A laminar flame is a steady-state system, and binary mixtures. Our results indicate that the thus the value of any macroscopic variable does ratio kT/xi is insensitive to the concentration of not change with time at a particular spatial the minor component. Further, at high tem- location. Therefore, the net flux for each species peratures (> 1000 K), kr/xi is relatively indepen- into a given volume element due to convective dent of temperature. Therefore, it is possible to and diffusive transport must be balanced by a select a value of kT/xi which represents a "worst corresponding change in the species concentra- case" for the thermal diffusion contribution to tion due to chemical reactions: the calculated production rate for a given species in order to test the hypothesis that Ri = V[Ni (v + Vi)]. (l) thermal diffusion is not significant. (See next section) Here , Ri is the net chemical rate, Ni is the species concentration, v is the convective veloc- B. Production Rate Calculation Results ity, and Vi is the net diffusion velocity of the species into the local mixture. The rate of Figure 3 presents the diffusive (concentra- chemical production is computed by numerical tion-driven and thermal) and convective contri- differentiation of Eq. 1 using profile data at a butions to the calculated production/destruction particular height, H, as well as data from H --2- 2 rate for acetylene at a height of 9 mm above the mm. Due to extreme sensitivity to noise in this burner. Convection transports material across calculation, it is important that the data be the high temperature reaction zone towards the smooth. Thus, all of the input data sets were burner centerline (see Fig. 1). In contrast, diffu- passed several times through a Savitsky-Golay digital filter 25 before differentiation. .... i .... i .... i .... The diffusion velocity is given by26:

Vi = -Di,~ix {Vln(x3 + (kT/xi)Vln(T)} (2)

where xi is the mole fraction of species i, Di.mix is the diffusion coefficient for species i into the ~E local mixture, and kr is the thermal diffusion o ratio. In this work, D, mix was determined from -~ o the contributions of ali molecules present in the co o flame at concentrations greater than 1% (nitrogen, oxygen, water, carbon dioxide, carbon monoxide, hydrogen, and methane) and is given by: ? oo o o o

i o,mlx:{1-xi}/Ij , Jo, ,l (3) , , h , i , , h , i , h L , t .... -10 -5 0 5 10 3~t LATERAL POSITION, mm

where Dij are binary diffusion coefficients calcu- FIG. 3. Contributions to the net flux of acetylene at lated using the method of Fristrom and 9 mm above the burner: A: thermal diffusion, o: Westenberg26. concentration-driven diffusion, Vq: convection, and The calculation of the thermal diffusion ratio +: chemical production and destruction. The actual is difficult for binary mixtures and is much more molar flux due to the mass transport terms into a complicated in multi-component systems 27. If spatial region is opposite in sign to that shown in the the temperature gradients are small enough, the figure. 1060 COMBUSTION GENERATED POLLUTANTS

sion away from the burner centerline brings is diminished by a new destructive feature (see acetylene into the high temperature oxidation arrow in Figure 4). This feature showed a region of the flame where it is destroyed. systematic evolution with height above the For these calculations, a value of 0.18 was used burner: it was absent in profiles below H = 10 for the magnitude of the ratio kT/xi for acetylene mm, it appeared as an inflection on the produc- in methane. This estimate for the thermal tion rate feature at H -- 11 ram, and as a dip in diffusion contribution was the maximum calcu- the 13 mm calculation. The new destruction lated for acetylene interacting with major feature is not an artifact of the data analysis species (nitrogen, water, carbon dioxide, and since its location and appearance are unaf- methane) in regions of production and destruc- fected by data smoothing. tion rate peaks. The sign of the contribution Harris and Weiner have found that acetylene from nitrogen and carbon dioxide interactions is is the primary species responsible for soot opposite to that from methane and water. Thus, particle growth in r-ich, premixed flames 2s-3~ At the overall thermal diffusion concentration short times the surface area available for growth would be smaller than that calculated above. As of incipient soot particles was measured to be Fig. 3 illustrates, even under these worst case 1.0 cm2/cm3, and a specific surface growth rate assumptions, the flux due to thermal diffusion is of -8 • 10 -5 ]g/(cm~" s) was determined in small with respect to convection and concentra- ethylene flames ~'29. Similar values have been tion-driven diffusion terms. reported for ethylene and propane diffusion Figure 4 shows the evolution of the acetylene flames 31-34. In our experiments-the location of production rate with height above the burner the decrease in the acetylene production rate and reveals that the chemical role of acetylene profile coincides with the peak concentrations of changes. Low in the flame the destruction rate small soot particles detected in earlier profile exhibits a maximum value (Ri =-1.9 x 10 5 measurements (shown as a dotted line on the mol/(cm~ s)) near the high temperature, pri- right hand side of Fig. 4) 17. If one assumes that mary reaction zone, and the production rate these incipient soot particles in our methane/air feature (Ri = 3.1 x 10- 5 mol/(cm3 s)) occurs diffusion flame exhibit surface areas and spe- slightly toward the fuel side. This peak in the cific surface growth rates similar to those found production rate is located on the high tempera- in rich premixed flames and diffusion flames of ture side of the observed maximum in the other fuels, then the magnitude of the change in concentration profile (see Fig. 2). Higher in this total soot mass can be estimated to be 8 x 10 -5 flame it appears that the production rate peak g/(cm ~ s) or 3 x 10 -6 acetylene mol/(cm 3 s). This value is consistent with the magnitude of the destruction feature shown in Fig. 4, and thus is

w , , attributed to surface growth chemistry on small 3.0 H=13mm soot particles. ,4 The production rates for diacetylene, butadiene, and benzene have also been calculated.

D Figure 5 shows the net chemical rates for o butadiene and benzene. Qualitatively, these E 1.0 profiles are very similar to those for acetylene: there is a destruction peak located near the x -'5 high temperature reaction zone and a production rate feature located between the high ~: -1.0 r temperature reaction zone and the observed z I Ik __9. peak in concentration of the species. Due to the diminished diffusion velocities of these heavier I hydrocarbons relative to that for acetylene, the 0_ -3.0 \ loss of these species by diffusion into the high \ temperature reaction zone is relatively less , = i i i ~ , I L -10 ' -15 -10 -5 important, i.e., their time history more closely LATERAL POSITION, mm follows the streamlines calculated from the convective velocities. FIG. 4. Profiles of the net chemical production rate for acetylene at two heights in the flame (solid lines); only the left hand side of the flame is shown. Also IV. Molecular Growth Chemistry shown is a profile of the small soot particle concentration17 at a height of 12 mm above the The chemistry of methane combustion to burner (dashed line). form carbon dioxide and water is reasonably PRODUCTION RATES IN DIFFUSION FLAMES 1061

o. .... i .... i r .... i that the reaction of C4H 5" with actetylene is the dominant formation pathway to C6H7~ .... "o Benzene is formed at temperatures above 1200 K in our methane/air diffusion flame. It is therefore also appropriate to consider reaction sequences which include the ethynyl radical. E,::; Colket has proposed that the following se- quence of reactions is important above 1500 K: • R7) H- 4- C2H2 = C2H- 4- H2 ~o R8) C2H2 + C2H" = n-C4H3" go R9) C4H3" + C2H2 = C6H5" R10) C6H5" + H2 = C6H6 4- H" Izl 0 C6H6 , 4H A variation on this scheme (R7-R10) leads to the polyacetylenes: ...... , I I t -10 -5 -10 -5 0 LATERAL POSITION, mm RI1) C4H3" = C4H2 4- H" R12) C4H2 4- C2H" = C6H2 4- H" FIG. 5. Production rate profiles for benzene and butadiene at 9 mm above the burner. A common feature to all of the reaction sequences presented above is the lack of oxygen-containing species in the hydrocarbon well established 35. However, the molecular growth chemistry. In our flame the concentra- growth steps which lead to the formation of tion of OH radical decreases rapidly on the polynuclear aromatic hydrocarbons (PAH) and rich side of the high temperature reaction soot particles are much less certain. Tanzawa zone 17. It is likely that reactive species such as and Gardiner 14 examined the mechanism of OH" and O" are important in hydrocarbon acetylene pyrolysis at temperatures ranging radical formation reactions and in the oxida- from 625 to 2500 K. They found that at low tion of hydrocarbons in the high temperature temperatures vinyl acetylene was a major prod- reaction zone I~ but are probably not impor- uct while at higher temperatures diacetylene tant in the growth reactions which eventually was favored. Their analysis indicates that vinyl form aromatic ring compounds. radical (C2H3") was more important at tempera- The mechanisms above may be tested by a tures less than 1250 K, whereas at higher comparison of a model's prediction of produc- temperatures the ethynyl radical, C2H', chemis- tion rates of key intermediates with the ob- try is dominant. Colket studied the pyrolysis of served production rate of benzene. For a acetylene and vinylacetylene in shock tubes pathway to be viable, the production rate of the from 1100 to 2400 K 16, At temperatures below intermediate must be at least as large as the net 1500 K his results indicate that the vinyl radical chemical production rate of benzene deter- is the key species in the formation of benzene mined from the experimental data. The pro- via: posed mechanisms suggest that the ethynyl radical and the vinyl radical are important intermediates. In order to calculate the produc- R1) H-+ C2H2 + M = C2H3" + M tion rate of these species it is necessary to know R2) C2H3" + C2H2 = C4H5" the concentration of hydrogen atoms, which R3) C4H5" = C4H4 + H" can be estimated from profiles of methyl R4) C4H4 + C2H3" = C6H7" radical, hydrogen, and methane if the reaction: R5) C6H7 = c-C6H7" R6) c-C6H 7" = C6H6 + H" R13) CH4 + H" = CH3" + H2

In a methane flame another potential source of is equilibrated. the vinyl radical is hydrogen atom abstraction Figure 6 shows the comparison of the ob- from ethylene: served net production rate for benzene with the maximum production rates of vinyl radical RI') H" + C2H4 = C2H3" 4- H2 (through reaction R1) and ethynyl radical (through reaction R7) at a height of 9 mm Cole et al. also propose that C6H7" is a key above the burner. The rate of R1 was calculated intermediate, but their mechanism postulates from the third order rate constant of Payne and 1062 COMBUSTION GENERATED POLLUTANTS

l o-, t .... , .... ,/..- i._t 2) These results indicate that a significant loss mechanism for acetylene involves surface reac- I f ] tions with particles in regions of the flame where soot inception occurs. t /-'J C.H. Production t 3) Production rate analysis is a promising tool 21o-,L ...... J for the evaluation of proposed mechanisms for 0 hydrocarbon growth chemistry. Preliminary results presented here indicate that the vinyl radical is a key intermediate in reaction se- E quences which lead to aromatic ring formation. Further, diacetylene production represents a ul.< 0_ 7 r / parallel condensation pathway and is not in- C2H2+H'~C2H'+Hz// volved directly in benzene formation in our methane diffusion flame. / /

10 -8 .... I .... I .... 5.0 4.0 5.0 6.0 REFERENCES LATERAL POSITION, mm 1. HOMANN, K. H. AND WAGNER, H. GG.: Ber. FIG. 6. Comparison of the experimental net chemi- Bunsenges. Phys. Chem. 69, 20 (1965). cal production rate of benzene (-o-) with the maxi- 2. BONNE, U., HOMANN, K. H., AND WAGNER, H. mum production rates of vinyl radicals (----, RI') GG.: Tenth Symposium (International) on Com- and ethynyl radicals (-., R7) at 9 mm above the bustion, p. 503, The Combustion Institute, 1965. burner. 3. HOMANN, K. H. AND WAGNER, H. GG.: Eleventh Symposium (International) on Combustion, p. 371, The Combustion Institute, 1967. Stief36, the measured acetylene concentration17, 4. CRITTENOEN, B. D. AND LONG, R.: Combustion and the hydrogen atom concentration calcu- and Flame 20, 359 (1973). lated from R13. For the abstraction reaction to 5. BITTNER, J. D. AND HOWARD, J. B.: Eighteenth form ethynyl radical, R7, the rate constant was Symposium (International) on Combustion, p. computed from the experimental rate constant 1105, The Combustion Institute, 1981. of the reverse process 37 and the equilibrium 6. BITTNER, J. D. AND HOWARD, J. B.: Particulate constant. For the calculation of the latter, a Carbon: Formation During Combustion (D.C. value of 127 kcal/mol was assumed for the heat Siegla and G. W. Smith, Eds.), p. 109, Plenum of formation of the ethynyl radical 38. Press, 1981. Figure 6 indicates that the production rate of 7. HOWARD,J.B. AND BITTNER,J.D.: Soot in Combus- vinyl radical exceeds the benzene production tion Systems and its Toxic Properties (J. Lahaye rate throughout flame regions where benzene and G. Prado, Ed), p. 57, Plenum Press, 1981. is formed. In contrast, the ethynyl radical 8. BITTNER, J. D., HOWARD, J. B., AND PALMER, H. formation rate via R7 falls far short of the B.: Soot in Combustion Systems and its Toxic benzene production rate. Since the equilibrium Properties (J. Lahaye and G. Prado, Ed.), p. 95, of R7 favors C2H2 and H relative to C2H and Plenum Press, 1981. H2, the net rate of ethynyl radical formation is 9. BITTNER, J. D.: D. Sc. Dissertation, Massachusetts much lower than that shown in Fig. 6 and can Institute of Technology, 1982. not account for the observed formation of 10. BITTNER, J. D., AND HOWARD, J. B.: Nineteenth benzene. This result indicates that the produc- Symposium (International) on Combustion, p. tion of diacetylene by reactions R7 and Rll 211, The Combustion Institute, 1981. represents a parallel hydrocarbon condensation 11. WARNATZ, J., BOCKHORN, H., MOSER, A., AND pathway in our flame, one that does not lead to WENZ, H. W.: Nineteenth Symposium (Interna- the formation of aromatic rings. tional) on Combustion, p. 197, The Combustion Institute, 1982. 12. BOCKHORN, H. FETTING, F., AND WENZ, H. W.: VI. Conclusions Ber. Bunsenges. Phys. Chem. 87, 1067 (1983). 13. COLE, J.A., BITTNER, J.D., LONGWELL,J.P., AND 1) Experimental measurements of species HOWARD, J.B.: Combustion and Flame 56, 51 concentrations, temperature, and velocity have (1984). been used to calculate the net chemical rate of 14. TANZAWA,T. AND GARDINER,W. C.,JR. :J. Phys. formation and destruction of intermediate hy- Chem. 84, 236 (1980). drocarbons in a methane/air diffusion flame. 15. FRENKLACH, M., CLARY, D.W., GARDINER, W.C., PRODUCTION RATES IN DIFFUSION FLAMES 1063

JR., AND STEIN, S.E." Twentieth Symposium R.B.: Molecular Theory of Gases and Liquids, (International) on Combustion, p. 887, The Chapter 8, John Wiley, 1964. Combustion Institute, 1984. 28. HARRIS, S.J., AND WEINER, A.M.: Combustion 16. COLKET, M.B. III: Twenty First Symposium Science and Technology 31, 155 (1983). (International) on Combustion, The Combustion 29. HARRIS, S.J., AND WEINER, A.M.: Combustion Institute, (this volume). Science and Technology 32, 267 (1983). 17. SMYTH, K.C., MILLER, J.H., DORFMAN, R.C., 30. HARRIS, S. J., AND WEINER, A.M.: Combustion MALLARD, W.G., and SANTORO, R.J.: Combustion Science and Technology 38, 75 (1984). and Flame 62, 157 (1985). 31. KENT, J.H., AND WAGNER, H.GG.:Combustion 18. SAITO, K., GORDON, A.S., AND WILLIAMS, F.A.: and Flame 47, 53 (1982), analyzed in Ref. 29. Trans. ASME 108, 640 (1986). 32. VANDSBERGER, U., KENNEDY, I., AND GLASSMAN, 19. MILLER, J.H., AND TAYLOR, P.H., Combustion I.: Combustion Science and Technology 39, 263 Science and Technology, in press. (1984). 20. FRISTROM, R.M.: Science 140, 297 (1963). 33. FLOWER, W.L., AND BOWMAN, C.T.: Chemical and 21. FRISTROM, R.M.: Ninth Symposium (Interna- Physical Processes in Combustion, 16-1, Novem- tional) on Combustion, p. 560, Academic Press, ber, 1985. 1963. 34. HURA, H.S., AND GLASSMAN, I.: Chemical and 22. FRISTROM, R.M.:Combustion and Flame 50, 239 Physical Processes in Combustion, 50~1, Novem- (1983)~ ber, 1985. 23. SMYTH, K.C., AND TAYLOR, P.H.: Chem. Phys. 35. WARNATZ, J.: Eighteenth Symposium (Interna- Lett. 122, 518 (1985). tional) on Combustion, p. 369, The Combustion 24. SCHOENING, S.M. AND HANSON, R.K.: Combus- Institute, 1981. tion Science and Technology 24,227 (1981). 36. PAYNEW.A., AND STIEF, L.JA J. Chem. Phys. 64, 25. SAvrrsv,,Y, A., AND GOLAY, M.J.E.: Anal. Chem. 1150 (1976). 36, 1627 (1964). 37. BROWN, R.L., AND LAUFER, A.H: J. Phys. Chem. 26. FRISTROM, R.M., AND WESTENBERG, A.A.: Flame 85, 3826 (1981). Structure, Chapter V, McGraw-Hill, 1965. 38. OKABE, H., and DIBELER, V.H.: J. Chem. Phys. 27. HIRSCHFELDER, J.O., CURTISS, C.F., AND BIRD, 59, 2430 (1973).

COMMENTS

A. C. Eckbreth, United Technologies Research Center, REFERENCES USA.: Flames produced on Wolfhard-Parker burn- ers are generally quite fragile and easily perturbed. 1. SMYTH, K. C. AND TAYLOR, P. H.: Chem. Phys. Lett. Have you checked your probe data against any laser 122, 518 (1985). measurement techniques to ascertain whether there 2. MILLER, J. H. AND TAYLOR, P. H.: Combust. Science are any probe-induced profile shifts or distortions? and Tech. in press.

Author's Reply. We have been concerned with the possibility of flame structure perturbations by the quartz microprobe. Recently the methyl radical, M. B. Colket, United Technologies Research Center, CH3", was detected with both a non-intrusive optical USA.: The calculation of production rates for species measurement (multiphoton ionization) 1 and in an are dependent strongly on the measured, spatial application of the scavenger probe technique with gradients. If species profiles are broadened due to mass spectrometric detection 2. When methyl radical the sampling process, how much would your calcula- profiles from the two measurements are compared, it tions of production rates be effected by probe is seen that the mass spectrometric profile is slightly perturbations? broadened with respect to the optical measurements, as expected from the larger effective sampling vol- Author's Reply. This is an important question, for ume diameter for the quartz microprobe. However, which no single answer is appropriate for all circum- there is no significant difference in the location of the stances. We have estimated that the quartz micro- maximum CH3 concentration or in the shape of the probe has an effective sampling diameter of - 0.7 profile peak, indicating that flame disturbances are mm [Ref. 17]. Therefore, the measured concentra- small--at least in the region of the flame where the tion profiles are broadened by microprobe sampling, methyl radical comparison can be made. and the peak concentrations are lower than actual 1064 COMBUSTION GENERATED POLLUTANTS flame concentrations. This broadening has a more pronounced effect on production rate calculations for radicals such as CH3", whose spatial distribution is P. R. Westmoreland, University of Massachusetts, USA. relatively narrow, than for intermediate hydrocar- (1) Your conclusion that C2H3 could be an impor- bons such as acetylene and benzene, whose concen- tant intermediate in forming benzene is reasonable, trations profiles are broad relative to the effective although its importance may be in the C4 formation probe sampling diameter. Calculations have been step. Simple addition of C2H3 to acetylenic carbon in carried out for acetylene in which the measured a C4 species gives a linear C6 radical species that must concentration profile was broadened by a binomial rearrange before cyclizing. smoothing function equivalent to a 0.8 mm effective (2) Your net formation rates (e.g., of CH3) show a sampling diameter. The additional broadening of the small formation peak on the oxidation side. Is this a concentration profile leads to a decrease in the smoothing artifact, as we observe in similar analysis of acetylene production rate of approximately 10%. The flat flames? overall uncertainty in the production rate calculations is estimated to be 50%, which includes errors intro- Author's Reply. duced in the calibration and temperature determina- (1) It is not possible for us to identify the dominant tions as well as the reproducibility of the experimen- C4 species involved in benzene formation in our tal measurements. However, these uncertainties do methane/air diffusion flame. Our estimates of radical not invalidate the key conclusion made in this paper concentrations, other than methyl radical which was concerning the importance of vinyl radical chemical experimentally determined, require a high level of growth reactions to the formation of benzene. confidence that the formation reaction is equilibrated, i.e. the characteristic times for the forward and the reverse chemical reactions are short with respect to the transport times of the reacting species. This proved to be the case in the equilibrium of methyl radicals and W. C. Gardiner, University of Texas, USA. The molecular hydrogen with methane and atomic hydro- relative fluxes through vinyl- and ethynyl-based gen. Our unpublished calculations suggest that other reaction sequences depend strongly upon tempera- radical equilibria, such as the abstraction reaction of ture, and so models that suggest a primary flux acetylene by H. to give ethynyl radical and molecular through vinyl channels at one position in a flame may hydrogen, do not meet the characteristic time criteria. also suggest that the ethynyl channels dominate in Further progress in understanding C4 and C6 chemis- hotter parts of the same flame. Similar' arguments try in our flame will depend on a more reliable pertain to the effects of H-atom concentrations and estimation scheme for polyatomic hydrocarbon radi- fuel structure) Could you please comment on the cal concentrations or on new quantitative radical anticipated range of validity of your conclusion that concentration measurements. vinyl pathways dominate. (2) As you surmise, the extraneous features on the edge of the production rate profiles are smoothing artifacts. Our smoothing routine is based on a cubic REFERENCE Savitsky-Golay filter, which tends to overshoot if there are not enough data points to define a corner in 1. FRENKLACH, M., CLARY, D.W., GARDINER, W.C., the profile. Hydrocarbon concentrations fall off JR., AND STEIN, S.E., this symposium. rapidly to zero at the flame front, and production rate profiles derived from these concentration pro- Author's Reply. Figure 7 in the text compares the net files often exhibit overshoot features in the oxidation rate of benzene formation with the maximum for- region. It is easy to identify such artifacts by varying ward formation rates of the vinyl and ethynyl the number of passes of a profile data set through the radicals. Temperatures in this region of the flame filter: smoothing artifacts grow and real features vary from less than 1200 K at a lateral position of 3.0 decrease in magnitude with additional passes. mm to more than 1900 K at 6.0 mm. The peak temperature at this flame height occurs at a lateral position of 6.6 mm. As Figure 7 illustrates, benzene formation via the ethynyl channel is never fast K. H. Homann, TH. Darmstadt, West Germany. When enough to account fbr the experimental production discussing the vinyl radical as a precursor for ben- rate over this entire temperature range. The domi- zene, it should not be forgotten that only very little nant role of the vinyl radical in chemical growth benzene is formed when H-atoms are mixed to reactions indicated by our results is likely to hold for acetylene in a discharge flow system at room tempera- hydrocarbon combustion in general, except perhaps ture or a few hundred degrees Celsius. However, the for acetylenic fuels where ethynyl radical concentra- reaction of O-atoms with C2H2 produces some ben- tions are expected to be high. zene. One difference between these two reaction PRODUCTION RATES IN DIFFUSION FLAMES 1065 systems is that methylene and methylene-like species is occurring (see text). If O atoms are important, they (CH2, C3H2...) are formed from O-atom reactions are likely to be involved in radical formation reactions but not from reaction with H-atoms. Thus, one close to the high temperature reaction zone. It would should also consider these species in discussing be interesting to compare an estimate of the rate of benzene formation in flames. methylene formation at the flame front with the observed benzene formation rate as we have done for Author's Reply. It is difficult to comment on the role the vinyl and ethynyl radicals. Unfortunately, quanti- of methylene chemistry in the methane/air diffusion tative O-atom concentration profiles are not yet flame. Our data suggest a limited role for O atoms in available for any hydrocarbon diffusion flame nor flame regions where hydrocarbon growth chemistry can estimates be made with confidence.