The Seeding of Methane Oxidation M. B. DAVIS and L. D. SCHMIDT* Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, MN 55455 USA Mixtures of light alkanes and oxygen can be converted into useful products by catalytic partial oxidation in excess fuel at moderate temperature. The relative importance of homogeneous chemistry in the reactions is still a subject of active debate, and, in this paper, the gas-phase partial oxidation of methane was simulated using detailed chemical reaction mechanisms. The sensitivity of homogeneous chemistry to the addition of trace amounts of radical and molecular species was analyzed to evaluate both the possibility of heterogeneous assistance to gas-phase reactions and the effectiveness of different species which might be added as promoters of particular product selectivities. Seeding of reaction mixtures with 2% of radicals reduced the homogeneous ignition delay time by only 50% at 1400 K. The delay was fairly insensitive to particular radical seeds because the radical pool adjusts rapidly. Seeding does not alter the product selectivity to CO and H 2 except at low conversion. 1999 by The Combustion Institute INTRODUCTION Mixtures of methane and oxygen can be converted to syngas (H 2 and CO) with high selectivity over rhodium-coated foam monoliths at millisecond contact times, while over platinum these mixtures are converted more to coupling products such as ethylene and acetylene [1, 2]. Ethane and higher alkanes are converted primarily to olefins over Pt under similar conditions [3, 4], and, on a single platinum-10% rhodium gauze, butane can be converted primarily to oxygenates [5]. In contrast, homogeneous reactions produce more H 2 O and CO 2 than these processes, even though these are not equilibrium products. All of these reactions occur at temperatures of 1100 K to 1700 K where the chemistry is rapid and complex. The role of homogeneous chemistry in these processes is currently not clear because of strong coupling between the gas phase and surface chemistry. In this paper, we explore the sensitivity of gas-phase combustion chemistry by seeding the oxidation reactions with active compounds. Seeding can aid in finding particular selectivity-enhancing additives and the role of heterogeneous assistance to homogeneous reactions through radical generation. *Corresponding author: E-mail: schmi001@maroon.tc.umn. edu 0010-2180/99/$ see front matter PII S0010-2180(99)00042-5 PARTIAL OXIDATION PROCESSES Hickman and Schmidt demonstrated a 95% carbon selectivity to CO and 90% hydrogen selectivity to H 2 at a 1 msec contact time by reacting methane and oxygen (1.7/1 ratio) over rhodium-coated foam monoliths at approximately 1300 K [1]. Hohn et al. obtained a 20% selectivity to C 2 compounds (acetylene and ethylene) with similar feeds using a platinumcovered monolith at high flow rates [4]. Huff and Schmidt found that ethane could be oxidatively dehydrogenated to ethylene (70% selectivity) by reacting ethane with oxygen over a platinum-coated foam monolith [3]. Finally, a process for converting butane and oxygen to oxygenates (primarily formaldehyde and acetaldehyde) using a platinum-10% rhodium gauze was demonstrated by Goetsch et al. Selectivities of 40% to oxygenated products and 35% to olefins were achieved [5]. In each of these processes, premixed fuel and oxygen near 300 K pass through a catalyst which is at a moderately high temperature (1100 to 1700 K). Gases are heated to these temperatures within a few microseconds, and the gas in the boundary layer experiences very steep gradients. At these temperatures, considerable gasphase reaction could be occurring, although homogeneous reactions require an induction period where radicals increase until the reaction can proceed at a significant rate [6]. The catalytic surface could assist or quench gas-phase COMBUSTION AND FLAME 119:182 188 (1999) 1999 by The Combustion Institute Published by Elsevier Science Inc.
SEEDING OF METHANE OXIDATION 183 chemistry by acting as a source or a sink of gas-phase radical species in the catalyst boundary layer. In this paper, we analyze the possibility of surface-assisted homogeneous reaction in a simpler configuration by seeding the reaction mixtures with radicals to examine how seeding might alter the homogeneous ignition delay time and change the reaction selectivity. Sensitivities to particular radicals and other compounds may suggest how partial oxidation processes may be tuned to promote or eliminate certain products. Seeding involves the introduction of small amounts of species, such as radicals and chain branching agents, to methane oxygen feed mixtures. Seeding is investigated for methane oxidation because, with its higher C-H bond strength [7], methane should be more sensitive to seeding than higher hydrocarbons. Hydrogen abstraction from methane is more difficult than from higher hydrocarbons and would therefore benefit most from seeding with respect to ignition delay time. Numerical Simulation of Homogeneous Chemistry To study the homogeneous chemistry that may exist in partial oxidation reactors, we simulated an isothermal plug flow reactor (PFR). As shown in Fig. 1a, this reactor simulates a single channel (typically 0.1 to 1 mm in diameter and 1 cm long) in a multichannel catalyst. A temperature step is introduced at the channel entrance in agreement with experimental temperature measurements. Homogeneous chemistry is simulated by two detailed reaction mechanisms. The first is the methane combustion mechanism, GRI-Mech v. 1.2, which contains 32 species and 157 reversible reactions [8]. The second mechanism was developed for methane oxidative coupling in excess fuel by Mims and Dean [9] and contains approximately 115 species and 450 reversible reactions. In this mechanism, the reactions and reaction rates were taken from the published mechanism, and the thermodynamic data was either taken from the CHEMKIN Thermodynamic database or estimated using group additivity [10 12]. The chemistry for each mechanism was modeled with CHEMKIN II [10]. The system of ordinary differential equations (ODEs) was solved using VODE, an implicit, variable stepsize ODE solver [13]. Homogeneous Reactions in the Methane-to- Syngas Process The feed conditions in the homogeneous reactor model were matched to those of the catalytic methane-to-syngas experiments [1]. Because the catalyst temperature in a syngas reactor is nearly constant in the axial direction, an isothermal reactor was simulated. A temperature of 1273 K was used, the methane-to-oxygen feed ratio was 1.7, and the reactor pressure was 1 atmosphere. The results of this simulation, using both the GRI and the Mims Dean mechanisms, are shown in Fig. 1. This figure shows the typical behavior of a homogeneous methane oxidation reaction at moderate to high temperatures. Early in the reaction, before ignition, very little reaction occurs except for the generation of radical species. Once enough radicals are created, the conversion rate increases rapidly to produce ignition. After the fast chemistry is completed (when either fuel or oxygen is consumed) the remaining products begin a slow approach to equilibrium values. Two points should be noted concerning the results in Fig. 1. First, neither the GRI-Mech results nor the Mims Dean mechanism results indicate significant reaction occurring before 20 milliseconds at 1273 K. Second, the H 2 selectivity in the homogeneous simulations ( 40%) is significantly lower than that of the experimental values ( 90%) [1]. This strongly suggests that pure homogeneous reaction is not the dominant pathway by which methane and oxygen are converted to syngas through partial oxidation. Furthermore, at least according to the PFR isothermal approximation, homogeneous reaction is not a significant participant in the methane-to-syngas partial oxidation process at all because much longer times are required to ignite homogeneous reactions than the 1 5 ms the gases spend in the catalyst. Other mechanistic possibilities exist, however, and one of these is that the chemistry is predominantly heterogeneous but contains
184 M. B. DAVIS AND L. D. SCHMIDT Fig. 1. Reaction mixture composition versus time on a 2.7 moles/s feed basis for the (a, b) Mims Dean mechanism [9] and for (c, d) GRIMech 1.2 [8]. CH 4 /O 2 1.7, T 1273 K. some form of heterogeneously assisted gasphase reaction. To examine this possibility, we studied the effect of radical seeding. SEEDING HOMOGENEOUS REACTIONS A series of calculations was undertaken to investigate systematically the possibility of heterogeneously assisted gas-phase chemistry and to evaluate possible combustion-promoting additives for homogeneous reaction. In these calculations, mixtures of methane and oxygen were seeded with small amounts of radical species such as O,H, and OH and molecular species such as H 2 O 2 and CH 3 OH. In these simulations, a PFR reactor model was used. Seeding was accomplished in two ways: seeds were added initially as reactor feed, and also a flux of radicals from the reactor walls was simulated with a radical generation step inserted into the homogeneous mechanism. These calculations were performed at several temperatures and were run using the Mims Dean mechanism. Ignition delay times, t ig, for different conditions were determined. In this paper, the ignition delay time is defined as the time in which the limiting reactant, O 2,is50% converted. For the seeding simulations, two different temperatures were investigated most extensively: 1000 K and 1400 K. At these two temperatures, the oxidation of a fuel-rich mixture (CH 4 /O 2 1.7) was simulated at 1 atmosphere. RESULTS Low Temperature (1000 K) The first case studied was seeding at 1000 K. Results of seeding simulations in the rich regime (CH 4 /O 2 1.7) are shown in Figs. 2a and 2b. Clearly, seeding reduces the ignition delay time of homogeneous reactions. With an initial O concentration of about 3.7%, t ig is decreased by about 80%, from 1.1 to 0.2 seconds. All other seeds, both molecular and radical, showed sim-
SEEDING OF METHANE OXIDATION 185 Fig. 2. Calculated ignition delay time versus mole percent seed at CH 4 /O 2 1.7 using the Mims Dean mechanism [9] for (a, b) T 1000 K, (c, d) T 1400 K. ilar results. Having 3.7% of any radical in a system is unlikely. Realistically, we can only expect radical percentages of 0.1% in most cases, and from Figs. 2a and 2b, these percentages affect t ig only slightly. The molecular seeding compounds varied in their seeding effectiveness. Hydrogen peroxide, for instance, reduced ignition delay by 50% at a feed rate of 0.25%. Methanol addition was only half as effective, with a reduction of around 25%. High Temperature (1400 K) The second case involved a reaction temperature of 1400 K and identical concentrations as for the low-temperature case. At 1400 K we are near the conditions where the syngas experiments are performed. In Figs. 2c and 2d, we see that seeding does have some effect at high seeding concentrations. For example, O seeding decreases t ig from 0.015 s to 0.005 s for a seed percentage of about 3.7%. This reduction is only 66%, which is, as expected, lower than the comparable seeding effectiveness in the low-temperature case (80%). We see that seeding with extremely large amounts of radicals only reduces the ignition delay time to 5 msec, which is still greater than the 1 msec contact time seen in the experimental catalytic partial oxidation system. Molecular seeds such as H 2 O 2 and CH 3 OH seemed to have as great an effect as the radical seeds for the fuel-rich case. Effect of Seeding on Selectivity Another interest in seeding is how different seeds affect the overall selectivity of homogeneous reactions to particular products. Note that in these situations, the products are far from equilibrium (H 2 and CO) for times less than 1 second. To study selectivity, simulations were run at 1400 K in the excess fuel, COproducing regime (C/O 2 1.7). The most effective radical and molecular seeds were intro-
186 M. B. DAVIS AND L. D. SCHMIDT Fig. 3. Selectivity versus time at 1400 K for CH 4 /O 2 1.7 with 2% seed flow using Mims Dean [9]. (a) Hydrogen selectivity, (b) CO selectivity. Fig. 4. Radical distribution versus time using the Mims Dean mechanism [9] at 1273 K on a 2.7 moles/s feed basis for CH 4 /O 2 1.7 with (a) no seed (b) 3.7% oxygen seed. duced into the reaction mixture at a level of 2.0%. In Fig. 3, we show the selectivity of CO (carbon selectivity) and of H 2 (hydrogen selectivity) versus time for many different seeds. Selectivities toward CO and H 2 are affected significantly by seeds very early in the reaction. The effect on overall yield, however, is negligible. Adding seeds at the beginning of the reaction seems to perturb the reaction, but in a very short time the reaction returns to its standard pathway. Short Time Radical Distributions One final calculation involved seeding a CH 4 / O 2 1.7 mixture at 1273 K with radicals and observing how the radical pool was affected throughout the reaction. In Figs. 4a and 4b, plots are shown of the CH 4 /O 2 reaction with no seed and with a 4% oxygen seed. Comparing 4b to 4a, we see that the high initial seed concentration disappears very quickly. This high radical concentration is transferred to the entire radical pool, resulting in increased concentrations of all primary radicals (O,H,OH, etc.), and, therefore, decreased ignition delay times. All seeds exhibited similar behavior at short times. Continuous Radical Seeding In the previous simulations, radicals were added to the system as feeds. While this form of seeding is effective in testing mechanism sensitivity, it is somewhat artificial in its simulation of a reactor, where radicals may be generated continuously by the reactor wall. To complement the seeding simulations, we introduced a continuous radical flux into reacting mixtures by adding a first-order radical generation reaction to the chemical mechanism. We found basically the same trends with these runs as with the previous seeding simulations. Oxygen is the most effective seed, and at 1400 K, the amount of radical flux necessary to reduce the ignition delay time significantly is at least 10 21 radicals/s cm 2, a value much greater than any expected desorption rate. At 1000 K, 10 20 radicals/s cm 2 were required while the expected desorption rate constant decreased by a factor of 100. No significant changes in selectivity were observed in either temperature case.
SEEDING OF METHANE OXIDATION 187 DISCUSSION This seeding study had two objectives. One was to evaluate the possibility of heterogeneously assisted homogeneous reaction in partial oxidation processes. The second was to search for potential promoters of homogeneous reactions that could be introduced into the monolith catalytic reactor to change the relative importance of homogeneous and heterogeneous mechanisms. Heterogeneous Homogeneous Mechanism From the data shown, seeding obviously affects the ignition delay time and selectivity of the reaction in the rich regime. One question, however, is whether the effect is large enough that the heterogeneous production of radicals could assist in the initiation of predominantly gasphase reaction mechanisms at conditions similar to those of catalytic partial oxidation processes. The ignition delay reduction at the optimum syngas forming ratio (CH 4 /O 2 1.7) for even the most effective of seeds is only 50% at 1400 K with an unrealistically high seed concentration of 2%. This would decrease the ignition delay time from 15 msec to 7 msec, which is still significantly longer than the observed reaction time in the monolith reactor. While heterogeneous assistance via seeding at higher temperatures may be insignificant, this phenomenon may be significant at even lower temperatures ( 1000 K), although the ignition delay times become long and the expected radical desorption rates become low. The observed selectivities also seem to point to directions other than a heterogeneously assisted gas-phase mechanism. The reaction pathway seems quite resistant to the addition of large amounts of radical and molecular seeds at high conversions, making a seeding initiation ineffective in modifying the overall selectivity of homogeneous oxidation reactions. Seeding was not able to modify the reaction to achieve carbon selectivities of 95% to CO and hydrogen selectivities of 90% to hydrogen, which are observed in the syngas experiments [1]. This resistance of gas-phase chemistry to seeding is most likely rooted in the fact that any seed added in large amounts causes an increase in the concentration of all primary radicals at short times, as indicated in Fig. 4. Because of this uniform increase in primary radical concentrations, seeding can significantly decrease induction times but it will not significantly change the reaction pathway to alter selectivity. Molecular Homogeneous Promoters The second goal of this work was to evaluate potential promoters of homogeneous reaction. Promoters studied were hydrogen peroxide, methanol, and various C 2 species. From the results it seems that hydrogen peroxide is an effective seed at most conditions studied. The promoters definitely change the behavior of homogeneous reactions in the model, and we are exploring these effects on the catalytic monolith system experimentally by seeding the reaction with molecular seeds such as methanol. CONCLUSIONS The simulations undertaken in this work suggest the following points: 1. Unassisted homogeneous reaction plays an insignificant role in the partial oxidation of methane at very short contact times at 1 atmosphere, especially for methane-to-syngas. 2. The addition of large amounts of any primary radical at the beginning of a reaction increases the concentration of all primary radicals in a very short time; thus seeding has very little effect on overall selectivity. 3. Homogeneous ignition delay time and selectivity are insensitive to small amounts of seed concentration at high temperatures. This insensitivity seems to suggest that heterogeneous assistance via radical generation is not important at catalytic partial oxidation conditions. 4. Molecular seeds such as hydrogen peroxide and methanol have a significant effect on the ignition delay time of the simulated gasphase reactions, almost as significant as adding radicals directly. Because these seeds can enhance homogeneous reaction, addition of methanol and H 2 O 2 could aid in determining how homogeneous reaction will affect partial
188 M. B. DAVIS AND L. D. SCHMIDT oxidation processes such as syngas production, oxidative coupling, and oxidative dehydrogenation. The computer simulations performed here do not account for every possible seeding situation, even in this simple PFR geometry. Combination seed feeds were not evaluated systematically, although several combinations were tried without any obvious effect. We also did not investigate the possibility of the catalyst surface acting as a sink for radicals. This could potentially lengthen reaction times and could alter selectivities to particular products in homogeneous chemistry. Although many other homogeneous seeding situations are possible, we believe that the ones tested strongly indicate that surface-assisted homogeneous chemistry via radical generation is unlikely to significantly alter partial oxidation processes at high temperatures above 1000 K. REFERENCES 1. Hickman, D., and Schmidt, L. D., J. Catal. 138:267 (1992). 2. Huff, M., and Schmidt, L. D., J. Phys. Chem. 97:11815 (1993). 3. Huff, M., and Schmidt, L. D., J. Catal. 149:127 (1994). 4. Hohn, K., Davis, M., Witt, P., and Schmidt, L. D., submitted for publication. 5. Goetsch, D., Witt, P., and Schmidt, L. D., Heterogeneous Hydrocarbon Oxidation ACS Symposium Series 638:124 (1996). 6. Steinfeld, J., Francisco, J., and Hase, W., Chemical Kinetics and Dynamics, Prentice Hall, NJ, 1989. 7. Handbook of Chemistry and Physics, CRC Press, Cleveland, Ohio 1997. 8. Frenklach, M., Wang, H., Goldenberg, M., Smith, G., Golden, D., Bowman, C., Hanson, R., Gardinier, W., and Lissianski, V. (1995) GRI-Mech An Optimized Detailed Chemical Reaction Mechanism for Methane Combustion, Gas Research Institute Rept. No. GRI- 95/0058. 9. Mims, C., Mauti, R., Dean, A., and Rose, K., J. Phys. Chem. 98:13357 (1994). 10. Kee, R., Rupley, F., and Miller, J. (1989) Chemkin II: A Fortran Package for the Analysis of Gas Phase Chemical Kinetics. Sandia National Laboratories Rept. No. SAND89-8009, Livermore, CA. 11. Benson, S., Thermochemical Kinetics: Methods for the Estimation of Thermochemical Data and Rate Parameters, Wiley, New York, 1976. 12. Ritter, E., and Bozzelli, J., Int. J. Chem. Kinet. 23:767 (1991). 13. Brown, P., and Hindmarsh, A., J. Appl. Math. Comput. 31:40 (1989). Received 5 May 1998; revised 8 March 1999; accepted 15 March 1999