Proceedings of the Combustion Institute, Volume 29, 2002/pp.

901–907

MODELING OF HIGH-TEMPERATURE MICROBURNERS

S. RAIMONDEAU,1 D. NORTON,2 D. G. VLACHOS2 and R. I. MASEL3

1
Department of Chemical Engineering
University of Massachusetts
Amherst, MA 01003-3110, USA
2
Department of Chemical Engineering and
Center for Catalytic Science and Technology
University of Delaware
Newark, DE 19716-3110, USA
3
Department of Chemical Engineering
University of Illinois
Urbana, IL 61801, USA

Flame propagation in microchannels is modeled using two-dimensional parabolic simulations with de-
tailed multicomponent transport, gas-phase chemistry, heat loss through the wall, radical recombination
at walls, and possible temperature discontinuity at the wall due to lack of thermal accommodation. We
show that under certain conditions of preheating and insulation, methane/air flames are able to propagate
in microchannels, providing a possible explanation for recent experimental observations. It is found that
in very small reactors, radial gradients and temperature discontinuity at the wall are negligible but become
significant as the diameter is increased. On the other hand, the near-entrance heat loss and radical quench-
ing at the wall are key issues in controlling flame propagation in microchannels. Finally, a brief comparison
between elliptic and parabolic simulations is presented. Although ignition distances are shorter in elliptic
simulations due to heat propagation upstream of the flame front, similar trends are observed.

Introduction the small dimensions encountered in these micro-

devices render spatially resolved measurements
The rapid advances in microfabrication techniques within a microreactor difficult, detailed mathemati-
have made possible the development of micro- cal modeling is invaluable in elucidating the mech-
reactors and power micro electromechanical anisms of combustion and assisting in the design of
systems such as microburners. High-temperature high-temperature microchemical systems.
(ⱖ1000 ⬚C) microchemical systems can exhibit ma- In an attempt to address the physics involved in
jor advantages compared to large-scale reactors, in- these devices, in this paper, we model oxidation of
cluding higher rates and selectivities for chemical methane with air in microchannels using two-di-
production due to enhanced heat and mass transfer mensional simulations with detailed gas chemistry
coefficients, higher energy densities compared to and multicomponent transport with variable prop-
batteries, higher equilibrium constants for endo- erties. Due to the large surface-to-volume ratio, em-
thermic reactions, abatement of pollutants for en- phasis is placed on interfacial phenomena, which
ergy production [1], and elimination of large-scale could be critically important. The effects of micro-
plant accidents [2]. However, the design of high- chemical system dimensions and operating condi-
temperature microchemical systems, including mi- tions, such as preheating of the reactants and wall
croburners and hydrogen production for fuel cell material, are investigated and a comparison with an
applications, demands an understanding of funda- elliptic model is presented. Finally, a bifurcation
mentals in these devices. map is created to guide the design of high-tempera-
Continuum theory predicts that a critical diameter ture microchemical systems.
(in the millimeter length scale) is necessary for prop-
agation of an exothermic reaction [3–5]. These the-
ories, which compare fairly well with experimental Two-Dimensional Model
data, focus on the competition between heat loss A straight tubular microchannel is modeled using
from the flame to the cold boundary and heat gen- two-dimensional simulations with detailed gas-phase
eration from the chemical reaction [6]. In contrast, chemistry and multicomponent transport. Fig. 1a
recent experiments [7,8] clearly indicate that prop- depicts the geometry studied. The gas-phase chem-
agation of flames in microchannels is feasible. Since istry for methane oxidation in air is taken from Ref.

901
902 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

It is only recently that radical quenching has been

modeled as a discontinuity in species concentration
at the wall with proper mass conservation [18]. Phys-
ically, radicals adsorb on the wall where they recom-
bine by surface reactions through surface diffusion
to form stable molecules that could desorb into the
fluid.
The rate of radical removal is herein modeled us-
ing the flux from the kinetic theory of ideal gases
reduced by the sticking coefficient, describing the
probability of an impinging molecule to stick to the
surface and the probability of a radical encountering
an empty active wall site, as described in Ref. [18].
Adsorbed species on the surface follow Langmuir-
Hinshelwood surface kinetics, forming stable mole-
cules. Possible surface reactions are discussed later.
The limiting case of zero sticking coefficient corre-
sponds to an inert wall, whereas a sticking coefficient
of one corresponds to the fastest possible removal
rate of radicals. We should remark that previous
work on stagnation flow has found that while radical
removal retards combustion under isothermal con-
Fig. 1. (a) The reactor configuration used in the simu-
ditions and can cause extinction [5,18], radical re-
lations. (b) Possible discontinuity at the wall according to
combination is exothermic and its released heat can
Knudsen’s experiments.
have a net stabilizing effect on the flame under cer-
tain conditions [18]. It is therefore important to de-
[9] using 46 reactions and 16 species. The parabolic lineate the thermal versus kinetic contributions of
governing equations are solved, that is, the axial dif- radical quenching on flame propagation in micro-
fusion is neglected compared to convection, whereas channels.
in the radial direction only diffusion of the conserved The other discontinuity concerns energy transfer
quantities is considered. The parabolic assumption and has been first discussed by Knudsen [19]. It ac-
is valid for sufficiently fast flows (high Peclet num- counts for the fact that molecules may not become
bers), as known from scaling analysis and demon- thermally accommodated with a wall, that is, the gas
strated with recent simulations [10,11]. Mixture av- temperature next to the wall may differ from that of
erage formulas are used to locally compute the solid, as shown in Fig. 1b. The energy boundary
diffusivities, thermal conductivities, and specific heat condition at the surface should then take into ac-
within the reactor, as outlined in Ref. [12]. The re- count heat loss through the wall of the microchannel
sulting conservation equations can be found else- and the possible temperature discontinuity at the
where [13,14]. wall. This discontinuity, that has not been previously
The inlet conditions are a fixed composition (a taken into account in modeling of flame propagation,
stoichiometric mixture) of methane in air, uniform results in reduction of heat loss through the wall and
temperature, and parabolic flow velocity. Because of could be responsible for flame propagation in mi-
axial symmetry, only half of the microreactor is mod- croburners seen recently experimentally [8]. From
eled. Due to the large surface-area–to–volume ratio rarefied transport phenomena, the temperature dis-
of microreactors, interfacial phenomena between continuity is modeled using the following expression
the gas and solid phases can be critical for flame [20]:
stability and performance. In particular, the en-
hanced heat and mass transfer coefficients in micro- cⳭ 1
reactors imply that radical and thermal quenching 1 ⳮ ah 冢 4c
Pr 冣 ⳵T
could play a key role in flame stability. For this rea-
son, we have placed emphasis on the wall boundary
Twall
s ⳱ Twall
g Ⳮ
ah
2k 冢 ⳵r 冣r⳱R
conditions discussed next. (1)
Here, we consider two possible interfacial discon-
tinuities that may be important in microreactors. where Twall
s is the temperature of the solid, Tgwall is
One discontinuity concerns mass transfer and ac- the gas temperature next to the wall, ah is the ther-
counts for radical quenching at the walls, that is, the mal accommodation coefficient, c is the ratio of spe-
gaseous and surface concentrations of each radical cific heats, k is the mean free path, Pr is the Prandtl
at the gas-solid interface differ (e.g., Refs. [15–17]). number, and (⳵T/⳵r)r⳱R is the radial temperature
MODELING OF HIGH-TEMPERATURE MICROBURNERS 903

difference scheme with adaptive mesh in order to

distribute more nodes where larger gradients are en-
countered. An implicit, first-order Euler method is
employed to solve the resulting set of differential-
algebraic equations using an adaptive length step,
which depends implicitly on the reactivity (stiffness)
of the system. At each length step, the nonlinear
algebraic equations are solved using Newton’s
method.

Flame Propagation in Microburners

A stoichiometric mixture of methane in air at at-
mospheric pressure for a flow rate of 2 m/s is con-
sidered for the present study. In this work, we use a
reference temperature of T0 ⳱ 1273.15 K, which
corresponds to the methane ignition at these con-
ditions. Temperature discontinuity at the wall is
taken into account in all simulations, unless other-
wise stated. We focus first on the characteristics of
combustion for conditions where flame propagation
occurs within small channels.
Figure 2a shows the mole fractions of selected,
major species at the wall along the length of the re-
actor for a radius of 100 lm and an overall heat
Fig. 2. (a) The mole fraction of selected species along the transfer coefficient of U ⳱ 4.187 W/m2 K, which
length of the reactor for R ⳱ 100 lm, U ⳱ 4.187 W/m2 K, corresponds to a relatively insulating material, such
a dimensionless inlet temperature of 1.5, and ah ⳱ 0.01. Com- as a ceramic with additional insulation. Our simula-
plete combustion of methane is obtained. (b), (c) The wall tions indicate that for certain design parameters,
dimensionless temperature profiles along the length of the flame propagation is possible in microchannels, as
reactor for three different radii indicated for (b) U ⳱ has also been observed experimentally [7,8]. Com-
4.187 W/m2 K and (c) U ⳱ 41.87 W/m2 K. As the radius is bustion of methane is completed in a very narrow
increased, flame propagation can occur with more conductive reaction zone. It is observed that the intermediate
materials. combustion products, H2 (not shown) and CO, are
formed, which then react to form H2O (not shown)
and CO2.
gradient at the wall. The thermal accommodation Figure 2b shows the corresponding dimensionless
coefficient indicates the affinity of the gaseous mol- temperature (T/T0) along the length of the reactor
ecules impinging on the wall to reach thermal equi- for different radii, but the same conditions as in
librium with the solid. This coefficient decreases for Fig. 2a. An initial drop in temperature is observed
high temperatures and rough surfaces. A typical due to heat loss through the wall. In all cases, a sharp
range of values can be found in Ref. [21]. Overall, increase in temperature is seen, indicative of gas-
the importance of temperature discontinuity at the phase ignition. Eventually, the temperature drops in
wall is a function of gas molecules, material, oper- the postcombustion zone due to heat loss. The in-
ating conditions, and, as shown below, reactor di- duction distance, the maximum temperature
mensions. A constant value of 0.01 is chosen for our reached, and the cooling of the reactor are clearly
simulations to contrast to the non-discontinuity case. affected by the size of the channel. Note that due to
Finally, the heat loss through the wall is determined preheating, the predicted flame temperatures are
as high for any practical device.
⳵T Figure 2c shows the dimensionless temperature at
qloss ⳱ U(Tswall ⳮ Ta) ⳱ ⳮk 冢 ⳵r 冣r⳱R
(2) the wall along the length of the reactor for U ⳱
41.87 W/m2 K and different radii. The parameters
where U is the overall heat transfer coefficient, for are the same as in Fig. 2a, except for the inlet tem-
the wall and between the outer wall and the sur- perature, that is lower for the larger radii (5% lower
roundings, Ta is the outside device temperature, and for 1 mm and 50% lower for 10 mm) in order to
k is the thermal conductivity of the gas. study the effect of reactant preheating on flame
The partial differential equations are discretized propagation. The observed dip in the wall tempera-
in the radial direction using a second-order finite ture is more pronounced as the overall heat transfer
904 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

their high conductivity and affinity for radicals, one

could use such a conductive material if heat loss is
diminished by an appropriate burner design and rad-
ical loss is reduced using appropriate coatings (see
below). Ronney and coworkers [22,23] have studied
a Swiss-roll-type microdevice using a countercurrent
flow heat exchange between the products and the
reactants. This ‘‘excess enthalpy’’ microburner has
the advantage of decreasing the heat losses and thus
enabling flame propagation.

Radial Gradients in Microburners

Figure 4a and 4b shows contour temperature and
methane plots within a 100 lm radius microreactor.
These contour plots clearly indicate that radial gra-
dients are negligible in microchannels. This result
Fig. 3. Heat generated by gas-phase reactions and heat can also be deduced through a simple scaling argu-
loss through the wall per unit time along the length of the ment of the dimensionless conservation equations.
reactor for R ⳱ 100 lm and two overall heat transfer co- The implications of this result are twofold. First, the
efficients indicated. The other parameters are as in Fig. 2a. effect of temperature discontinuity in reducing the
The near-entrance heat loss is important in determining wall heat loss (Knudsen’s idea) is negligible since
flame propagation within the reactor. (⳵T/⳵r)r⳱R r 0. Therefore, flame propagation is not
a result of temperature discontinuity at the walls.
Second, since no radial mass and energy transfer gra-
coefficient is increased, which can eventually lead to dients exist in such microreactors, they can be used
flame extinction, as seen for the R ⳱ 100 lm mi- to extract kinetic information using relatively simple
croreactor. While using a microreactor with slightly plug flow reactor models that are ordinary instead of
less insulating material does not lead to flame prop- partial differential equations. Furthermore, the pri-
agation in microchannels, larger radii reactors are mary mechanism of heat loss in previous experi-
able to sustain flame propagation, even with less pre- ments on flame propagation was from the flame to
heating. the cold boundary, whereas in microchannels it is
Overall, Fig. 2 indicates that the thermal proper- mainly through the wall due to the absence of radial
ties of the wall and preheating of the reactants influ- gradients. This leads to a different dependence of
ence the critical diameter for flame propagation. To the quenching distance on parameters, such as pres-
further illustrate this fact, Fig. 3 shows the cumula- sure (compared to published theory). Further ex-
tive power along the length of a microchannel for perimental work on microreactors is needed to val-
two overall heat transfer coefficients indicated. idate this aspect.
Three zones can be distinguished: a precombustion In contrast to microreactors, gradients develop in
zone, where heat loss through the wall dominates, channels of larger diameter and/or increased heat
the combustion or reaction zone, where steep gra- transfer coefficients, as shown in the temperature-
dients in heat generation can be observed upon ig- contour plots of Fig. 4c–4f. The increased radial gra-
nition, and a postcombustion zone where nearly dients with increasing radius are a result of the
complete conversion has been reached and the re- longer times that energy and species have to diffuse
actor just cools down. The initial heat loss is crucial from the bulk of the reactor to the wall. Because
for flame propagation in microchannels mainly due gas-phase ignition and combustion occur at the bulk
to the lack of radial gradients (see also below), which of the reactor (higher temperatures observed at the
would create a delay in cooling the bulk of the re- bulk in Fig. 4d–4f), the development of radial gra-
actor. Indeed, for U ⳱ 41.87 W/m2 K, the initial dients in larger channels results in a weaker gas-solid
heat loss is significant when compared to the more interaction and thus flame propagation is less af-
insulating system (U ⳱ 4.187 W/m2 K), and it causes fected by interfacial phenomena. Indeed, as seen in
the temperature to drop below the gas-phase igni- Fig. 2c, more conductive materials can be used for
tion temperature, resulting in lack of propagation. these larger reactors.
Experimentally, microburners are built using sili- So far, we have focused on the possibility of flame
con carbide plates coated with silica/alumina, yield- propagation and the presence of radial gradients in
ing enough insulation for flame propagation in a sin- microburners using the parabolic approximation. In
gle channel [7,8]. While propagation of flames in order to explore the validity of the approximation,
stainless steel microburners is impossible because of Fig. 5 shows temperature profiles along the center
MODELING OF HIGH-TEMPERATURE MICROBURNERS 905

Fig. 4. (a), (b) Contour plots for dimensionless temperature and methane mass fraction, respectively, within the reactor
for the conditions of Fig. 2a. No radial gradients are observed under these conditions. (c), (d) Dimensionless temperature
contours for R ⳱ 1 and 10 mm, respectively, for the same conditions. (e), (f) Dimensionless temperature contours for
R ⳱ 1 and 10 mm, respectively, and U ⳱ 41.87 W/m2 K. Significant gradients appear for larger channels and increased
heat transfer coefficients.

of symmetry and the wall using the parabolic and the

elliptic codes, for a radius of 200 lm, and an over-
all heat transfer coefficient of U ⳱ 41.87 W/m2 K.
Similar radial gradients are observed for the two
models. Furthermore, although the reaction zone is
broader in the elliptic case due to axial mass and heat
transfer, combustion occurs around the same posi-
tion. It is to be noted that for adiabatic conditions
(not studied here), the ignition distance is largely
overpredicted with the parabolic model. Gas-phase
axial diffusion and conduction appear to be less im-
portant under significant wall heat loss. Since the
focus of this study is on interfacial discontinuities
and not ignition distances, the parabolic model is
adequate to draw important semiquantitative con-
clusions (see also below).

Role of Radical Quenching

So far, we have focused on understanding the role
Fig. 5. Dimensionless temperature along the centerline of thermal quenching in flame propagation for rad-
and the wall of the reactor for the elliptic and the parabolic ical quenchless materials (zero sticking coefficient).
models for R ⳱ 200 lm, one-step chemistry, the same inlet We expect that at these small scales, radical quench-
temperature of T/T0 ⳱ 1.2, and U ⳱ 4.187 W/m2 K. No ing at the wall can become crucial in determining
temperature discontinuity is taken into account. The re- flame propagation, especially for walls having high
action zone is broader, and the front stabilizes closer to the radical removing efficiency. This is a result of fast
entrance in the elliptic model. Similar radial gradients are transport of radicals from the bulk of the fluid to-
observed. ward the wall.
906 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

TABLE 1
Radical quenching mechanism along with kinetic
parameters

Pre-Exponential
(sⳮ1) or
Reactions Sticking Coefficient

1. CH3 Ⳮ * r CH*3 0ⳮ1

2. H Ⳮ * r H* 0ⳮ1
3. OH Ⳮ * r OH* 0ⳮ1
4. O Ⳮ * r O* 0ⳮ1
5. 2CH*3 r C2H6 Ⳮ 2* 1013
6. 2H* r H2 Ⳮ 2* 1013
7. 2OH* r H2O Ⳮ O* Ⳮ * 1013
8. 2O* r O2 Ⳮ 2* 1013
9. CH*3 Ⳮ H* r CH4 Ⳮ2* 1013
10. OH* Ⳮ H* r H2O Ⳮ2* 1013

Fig. 7. Inlet dimensionless temperature at the extinction

limit versus overall heat transfer coefficient for three radii
and quenching affinities indicated. The other parameters
are as in Fig. 2a. For R ⳱ 1 mm, the s ⳱ 10ⳮ3 results (not
visible) are practically the same as those for s ⳱ 0. For
R ⳱ 10 mm, neither radical quenching nor thermal
quenching play any role. For combustion in microchannels,
more insulation and relatively quenchless materials are re-
quired.

most important for flame ignition and extinction.

The sticking coefficients of the adsorption steps are
varied from zero to one to simulate walls having dif-
ferent radical quenching affinity. All surface reaction
and desorption steps are taken as non-activated. This
situation corresponds to the fastest possible rate of
radical quenching.
Figure 6a shows the mole fraction of methane for
a quenchless wall (solid line) and for a fast radical
Fig. 6. Mole fraction of (a) methane and (b) dimension- quenching wall (dashed line) for a 100 lm radius
less temperature along the length of the reactor for microreactor. Fig. 6b shows the corresponding wall
quenchless walls (solid lines) and quenching material temperature. Fig. 6 shows that radical quenching
(dashed lines). The parameters are as in Fig. 2a. Tempera- can result in a lack of flame propagation. We have
ture discontinuity is not taken into account in these simu- found that the rate of heat released by radical re-
lations. Radical quenching cannot be neglected at these combination at the wall is insufficient to compensate
small scales. for the rate of heat loss through the wall. Therefore,
the kinetic mechanism clearly dominates over the
thermal one.
To model radical quenching, a chemically reactive
surface is modeled, as outlined in the modeling sec-
Microburner Design
tion. That is, the important radicals for ignition and
extinction are quenched out by adsorbing them on Our simulations indicate that both the initial heat
the surface and allowing them to recombine to form loss and wall radical quenching are key interfacial
stable gas-phase species that desorb into the fluid phenomena for understanding flame propagation in
phase. Table 1 shows the heterogeneous (surface) microchannels. To summarize such issues, Fig. 7
reactions we consider in simulating radical quench- shows the dimensionless inlet temperature versus
ing along with the corresponding kinetic parameters. the overall heat transfer coefficient for three radii
Previous work [17] has shown through sensitivity and various sticking coefficients indicated, illustrat-
analysis that the radicals depicted in Table 1 are the ing the important competition between heat loss,
MODELING OF HIGH-TEMPERATURE MICROBURNERS 907

radical quenching, and heat generation. The inlet 3. Davy, H., Phil. Trans R. Soc. London 107:45 (1817).
temperature, which determines the amount of re- 4. Kuo, K. K., Principles of Combustion, Wiley, New
actant preheating needed, increases significantly as York, 1986.
the radius decreases and the heat transfer coefficient 5. Lewis, B., and von Elbe, G., Combustion, Flames and
increases. Explosions of Gases, Academic Press, Orlando, 1987.
As the reactor radius decreases, the effect of rad- 6. Wohl, K., Proc. Combust. Inst. 4:68–89 (1952).
ical quenching becomes significant due to the ab- 7. Masel, R. I., and Shannon, M., Microcombustor Hav-
sence of radial gradients (fast mass transfer) at these ing Submillimeters Critical Dimensions, U.S. patent
small scales. Furthermore, the system becomes 06193501, 2001.
more sensitive to radical quenching for more insu- 8. Jensen, C., Masel, R. I., Moore, G. V., and Shannon,
lating materials. Interestingly enough, for suffi- M. A., unpublished manuscript, 2001.
ciently large radii, heat and mass transfer to the walls 9. Giovangigli, V., and Smooke, M. D., Combust. Sci.
is so slow that the flame propagates regardless of the Technol. 53:23–49 (1987).
wall material and the inlet enthalpy. This is consis- 10. Raja, L. L., Kee, R. J., Deutschmann, O., Warnatz, J.,
tent with previous stagnation flow simulations where and Schmidt, L. D., Catal. Today 59(1–2):47–60
minimal wall-flame interactions were observed when (2000).
the flame was 4–5 mm away from the surface [1]. 11. Mantzaras, J., Appel, C., Benz, P., Daily, J. W., Andrae,
From the work of Masel and coworkers [8,24], in- J., and Zerkle, D. K., Proc. Combust. Inst. 28:1349–
sulating materials, such as appropriately prepared 1357 (2000).
ceramics, do not strongly adsorb radicals as com- 12. Kee, R. J., Dixon-Lewis, G., Warnatz, J., Coltrin,
pared to conductive materials, such as stainless steel M. E., and Miller, J. A., A FORTRAN Computer Code
or copper. Thus, careful design of microburners, in Package for the Evaluation of Gas Phase Multicom-
terms of materials choice and insulation, is essential ponent Transport Properties, Sandia report SAND86-
for flame propagation. Finally, comparison with an
8246, 1990.
elliptical model shows that heat loss dominates com-
13. Groppi, G., Tronconi, E., Forzatti, P., and Berg, M.,
pared to heat conduction but needs to be included
Catal. Today 59:151–162 (2000).
for quantitative model prediction.
14. Mantzaras, J., and Benz, P., Combust. Flame
119(4):455–472 (1999).
Conclusions 15. Tomlin, A. S., Pilling, M. J., Turanyi, T., Merkin, J. H.,
and Brindley, J., Combust. Flame 91:107–130 (1992).
Modeling of homogeneous microreactors with de- 16. Vlachos, D. G., Schmidt, L. D., and Aris, R., Combust.
tailed transport and gas chemistry has been carried
Flame 95:313–335 (1993).
out. Emphasis has been placed on interfacial discon-
17. Vlachos, D. G., Schmidt, L. D., and Aris, R., AIChE
tinuities of species concentration and temperature.
J. 40(6):1018–1025 (1994).
We have shown that for sufficiently narrow channels,
18. Aghalayam, P., Bui, P.-A., and Vlachos, D. G., Com-
radial gradients and temperature discontinuities at
bust. Theory Modeling 2:515–530 (1998).
the wall are negligible. On the other hand, the near-
19. Knudsen, M., Ann. der Physik 34:593–656 (1911).
entrance heat loss and radical wall quenching are key
20. Rosner, D. E., and Papadopoulos, D. H., Ind. Eng.
players for flame propagation in these microdevices.
The choice of the wall material is therefore crucial, Chem. Res. 35:3210–3222 (1996).
in terms of insulating and radical quenching prop- 21. Saxena, S. C., Thermal Accommodation and Adsorp-
erties, when designing such microreactors. tion Coefficients of Gases, Hemisphere, Washington,
DC, 1989.
Acknowledgment 22. Sitzki, L., Borer, K., Schuster, E., Ronney, P. D., and
Wussow, S., ‘‘Combustion in Microscale Heat-Recir-
This work was supported by the Army Research Office culating Burners,’’ Third Asia-Pacific Conference on
under contract DAAD19-01-1-0582. Any opinions, find- Combustion, Seoul, Korea, June 24–27, 2001.
ings, and conclusions or recommendations expressed in 23. Maruta, M., Takeda, K., Sitzki, L., Borer, K., Ronney,
this publication are those of the authors and do not nec- P. D., Wussow, S., and Deutschmann, O., ‘‘Catalytic
essarily reflect the views of the Army Research Office. Combustion in Microchannel for MEMS Power Gen-
eration,’’ Third Asia-Pacific Conference on Combus-
tion, Seoul, Korea, June 24–27, 2001.
REFERENCES
24. Masel, R. I., in Principles of Adsorption and Reaction
1. Aghalayam, P., and Vlachos, D. G., AIChE J. on Solid Surfaces, Wiley, New York, 1996, pp. 356–
44(9):2025–2034 (1998). 374.
2. Srinivasan, R., Hsing, I., Berger, P., Jensen, K. F., Fi- 25. Norton, D., and Vlachos, D. G., unpublished results,
rebaugh, S., Schmidt, M., Harold, M., Lerou, J., and 2002.
Ryley, J., AIChE J. 43(11):3059–3069 (1997). 26. Ronney, P. D., personal communication, 2002.