909

Proceedings of the Combustion Institute, Volume 29, 2002/pp. 909–916

DEVELOPMENT OF A MICROREACTOR AS A THERMAL SOURCE FORMICROELECTROMECHANICAL SYSTEMS POWER GENERATION

J. VICAN,1 B. F. GAJDECZKO,1 F. L. DRYER,1 D. L. MILIUS,2 I. A. AKSAY2 and R. A. YETTER3

1Department of Mechanical and Aerospace Engineering2Department of Chemical Engineering

Princeton UniversityPrinceton, NJ 08544, USA

3Department of Mechanical and Nuclear EngineeringPropulsion Engineering and Research Center

The Pennsylvania State UniversityUniversity Park, PA 16802, USA

An alumina ceramic 12.5 � 12.5 � 5.0 mm microreactor was constructed using a modified stereolith-ography process. The design was based on a ‘‘Swiss roll’’ concept of double spiral-shaped channels tofacilitate a high level of heat transfer between the reactants and combustion products and wall surfacecontact of the flow through the microreactor body. Self-sustained combustion of hydrogen and air mixtureswas demonstrated over a wide range of fuel/air mixtures and flow rates for equivalence ratios from 0.2 to1.0 and chemical energy inputs from 2 to 16 W. Depositing platinum on gamma alumina on the internalwalls enabled catalytic ignition at or near room temperature and self-sustained operation at temperaturesto 300 �C. Catalyst degradation was observed at higher operating temperatures and reignition capabilitieswere lost. However, sustained operation could be obtained at wall temperatures in excess of 300 �C,apparently stabilized by a combination of surface and gas-phase reaction phenomena. A global energybalance model was developed to analyze overall reactor performance characteristics. The reactor designand operating temperature range have potential applications as a heat source for thermoelectric and py-roelectric power generation at small scales compatible with microelectromechanical systems applications.

Introduction

With the successful development of microelectro-mechanical systems (MEMS) devices such as micro-sensors and actuators, there has been considerableinterest in developing chemically fueled micropowergeneration systems that could produce 300 mW orless, with footprints sufficiently small to colocatecommunications, sensors, actuators, and power sys-tems all on the same chip. Chip level integration thatincludes power generation would eliminate inter-connects and benefits applications involving multi-point distributed sensing, communications, and re-sponse. Depending on the battery technologyassessment source [1,2], liquid hydrocarbons arefrom 12 to more than 50 times more energy-dense(by volume) than advanced electrochemical sources,provided that ambient air is available as the oxidizingagent. Thus, overall conversion efficiencies to equiv-alent electrical output as low as 10% would suggestchemical energy conversion could be preferred interms of useful life.

Conventional gasoline or diesel fuels are not likelyto be useable in MEMS chemical conversion sys-tems, because of fuel chemical instability (property

degradation) in long-term storage, resulting in par-ticulates and deposits. However, liquefied gaseousfuels (propane, butane, and their conjugate olefins)or hydrocarbon liquids designed specifically for longshelf life (and even higher energy densities) are rea-sonable candidates for MEMS chemical energy con-version devices. Chemically driven power-genera-tion modes may also be advantageous for specificapplications such as complementing battery sourcesfor high-power output or power peaking.

A myriad of chemical-to-electrical conversion ap-proaches are, in principal, open to consideration atthe MEMS level. Current research efforts using me-chanical motion to generate power include micro-gas turbines, micro-Wankel engines, and linear, mi-cropulsating systems based on oscillating electricalfield and piezoelectric deflections/oscillations [3].Direct energy conversion concepts without movingparts that are being considered include fuel cells andthermal reactors coupled to thermoelectric, photo-voltaic, and pyroelectric devices. Highly efficientfuel cells presently operate only on pure hydrogen.However, hydrogen storage technologies producingvolumetric energy densities approaching liquefiedgaseous fuels do not yet exist. On-board hydrogen

910 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

Fig. 1. A conceptual power supply with an excess en-thalpy (‘‘Swiss roll’’) catalytic reactor.

Fig. 2. Cross-section of exemplar microreactor designfabricated by stereolithography. Top flow channel (same onbottom) is visible, spiraling into the center of the structure.

generation significantly increases the complexity ofthe system, and as a result, may eliminate the overallsystem efficiency advantage and negatively affect re-liability and operating lifetime.

Thermal microreactors of various types, used withdirect thermal to electrical energy conversion orcoupled with external heat engine concepts offer theadvantages of simplicity, longer-term compatibilitywith various hydrocarbon fuel types, and flexibilityin decoupling system chemical conversion and en-ergy extraction time scales. Of primary importanceto these concepts is a chemical to thermal energyconversion reservoir coupled to methodologies toconvert thermal to electrical energy (Fig. 1). Alongwith us, other groups, for example, Ref. [4], are de-veloping thermal microreactor designs based on theconcept of excess enthalpy combustion [5]. As a re-sult of regeneratively preheating the reactants, com-bustor temperatures can achieve super adiabatictemperatures, increasing combustor efficiency, ex-tending fuel/air mixture limits for which gas-phasekinetics remain explosively chain branched, and re-ducing wall quenching.

Since 1998, we have been developing thermal mi-croreactors on small scales that would achieve op-erating temperatures below 1100 K that could even-tually incorporate excess enthalpy operating

principles, that are compatible with eventual use ofliquefied gaseous hydrocarbons as fuels, and thathave potential for being scaled to even smaller di-mensions. The microreactor design described in thispaper (see Fig. 2) is intended to provide a favorableenvironment for microscale combustion of gaseousfuel and air mixtures. A layer of catalyst (platinum)is deposited on the internal walls of the channels tolower the range of the operating temperatures. Thespiral ‘‘Swiss roll’’ design shown provides a numberof desirable features, including: (1) thermal energytransfer mechanism to preheat the reactants usingthe exhaust heat combustion volume, (2) relativelylarge internal surface/volume ratio needed for theheat transfer through the walls and for effective ac-tion of the catalyst, and (3) large top- and bottom-face surface area for the heat transfer to externaldevices.

In the present paper, we describe the reactor andthe fabrication process, develop an analytical modelto characterize the performance of the reactor, andpresent results on the operation of the reactor withmixtures of hydrogen and air. In addition, the reactoris coupled to commercial thermoelectric modules todemonstrate chemical to electrical energy conver-sion at the length scales of the present reactor. Ad-vancements in reducing the scale of the reactor, im-proving heat management, and implementinghydrocarbon fuels will be discussed in forthcomingpapers.

Microreactor Fabrication

Three-dimensional stereolithography was utilizedto manufacture the microreactors. A three-dimen-sional virtual solid body produced by computer-aided design is ‘‘sliced’’ finely (0.075 mm) along theZ-axis, creating a build file consisting of a stack ofX–Y layers. During the build, sublayers are succes-sively added to the part by raster scanning an UVlaser across its X–Y surface, wetted by a photocura-ble bath of liquid resin [6]. The photocured, solidi-fied layer is moved down by one layer thickness in-side the liquid bath, and the patterned curingprocess is repeated. The resin used is a highly con-centrated colloidal dispersion prepared by dispers-ing alumina powder in an aqueous solution of ultra-violet curable polymers. The ceramic green bodymanufactured is subsequently dried, the photocur-able binder is burned out, and the part is sintered tofull density.

The ceramic microreactor part (Figs. 2 and 3) canbe generated in about 4 hr and is an alumina mon-olith with two distinct approximately 800 lm2 chan-nels (one each side) that spiral inward, joining nearthe center of the body. The approximate length ofthe internal path was 0.1 m. A larger central hole isadded to the structure to accommodate the posi-tioning of an electrical resistance heating wire

MICROREACTOR FOR MEMS POWER GENERATION 911

Fig. 3. Stereolithography process with examples of fab-ricated parts. The finished reactor, showing inlet and outletholes on the front surface, is shown in the center.

through the center of the body to provide sufficientheat to initiate catalytic reactions of a premixed fuel/air mixture flowing through the spiral passages.

Following stereolithographic fabrication, the en-trained liquids are removed from the unsintered part(green body) and the polymers are burned out of thestructure at low temperatures using a stepwise heat-ing program raising the part temperature to 425 �Cover 6 hr. The green body is then sintered at hightemperatures (1550 �C for 2 hr) to full density. Post-sintering processing includes the addition of a plat-inum catalyst supported on a gamma alumina washcoat deposited within the microreactor channels.The catalyst is added by injecting hydrogen hexa-chloroplatinate into the channels and heating thedried microreactor to 350 �C for 2 hr in air and thenfor 8 hr in hydrogen.

Microreactor Model Analysis

As a prelude to detailed computational fluid dy-namics and thermal stress analyses, a simple energybalance model was utilized to characterize overalloperational characteristics of an uninsulated reactorin a quiescent environment. Lumped heat capaci-tance was assumed since Biot numbers for the struc-ture are estimated to vary from 0.001 to 0.1, de-pending upon their definition. The Biot number wasdefined in several ways so that a range of Biot num-bers was obtained. The external heat transfer coef-ficient was compared to the overall internal heattransfer coefficient (which included convective heattransfer across the channels in the reactor body).Biot numbers were also calculated using the classicaldefinition for the internal conduction heat transfer

rate to the external heat transfer. The chemical re-action was also assumed to be complete within thereactor; this assumption can be globally removed us-ing experimental measurements of the actual chem-ical conversion efficiency. Thermal dissociation ofproducts was assumed negligible and the chemistrywas described by a single overall global reaction. Atsuch low Biot numbers and the heat recovery natureof the ‘‘Swiss roll’’ design, the reactor body and in-ternal gas temperatures were assumed equal anduniform in the model. The lumped heat capacitanceassumption is verified below from thermocouplemeasurements at various locations on the externalsurface. Under these assumptions and steady-stateconditions, the energy balance is

¯mc T � mQ � mc T � hA(T � T )p,r i rp p,p s s �

4 4� r�A(T � T ) � 0s �

where m is the mass flow rate of fuel and oxidizer,is the reactant mixture specific heat, is thec cp,r p,p

product mixture specific heat, Qrp is the lower heatof combustion per unit mixture mass, Ti is the inlettemperature of the reactants, Ts is the reactor bodyand exhaust product temperature, T� is the environ-mental temperature, h is the film coefficient, A isthe reactor surface area, r is the Stefan-Boltzmannconstant, and � is the reactor body emissivity. Ther-modynamic data were obtained from the JANNAFtables, the film coefficient was estimated from theNusselt number correlation of Churchill and Chu [7]for a flat vertical surface and RaL numbers less than109, and � of aluminum oxide was assumed equal to0.69 [8] independent of temperature. The aboveequation was solved for the surface temperature,which is compared to experimental measurementsbelow. Note that as long as heat transfer betweenthe channels and the reactor body remain high, themodel places no restriction on where the heat re-lease occurs (surface catalysis or homogeneously inthe gas phase). Reaction quenching at the surface isnot an issue, as long as energy transfer to and fromthe surface maintains the surface temperature abovethe gas-phase explosion limit.

Figure 4 shows the predicted reactor surface(body) temperature (Ts) and gas residence times (s)as a function of equivalence ratio, �, for a fixed totalvolume flow rate of reactants of 80 cm3/min. Volu-metric flow rates are reported for standard condi-tions (298 K, 1 atm). Predicted surface temperaturesvary from approximately 350 to 530 K for variationsin � from 0.1 to 1, while residence times decreasefrom about 68 to 52 ms. Predicted residence timesexceed the estimate for mass diffusion times to thewalls by at least a factor of 50, indicating sufficientmass turnover for complete reaction. Body tempera-tures below �373 K may initiate water condensationon the reactor channel surfaces. Thus, for the pres-ent reactor, the minimum � for 80 cm3/min flow rateis anticipated to be 0.2.

912 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

Fig. 4. Predicted reactor temperature, residence time,energy loss distributions, and thermal efficiency for a totalvolumetric flow rate of 80 cm3/min.

The distributions of heat loss by convection at thesurface, Qconv, heat loss by radiation, Qrad, and un-recovered energy in the exhaust, all relative to theheat input, Qin, are also shown in Fig. 4. At � � 0.1,�60% of the energy loss occurs by convection, 26%occurs by radiation, and the remaining 14% is re-tained in the exhaust gases. At � � 1.0, energy lossby convection accounts for �54% of the energy in-put, radiative loss accounts for �39%, and the ex-haust contains �7% of the added energy. Also shownin Fig. 4 is the ratio of the reactor temperature tothe adiabatic flame temperature that the body wouldassume if both radiation and convection losses wereeliminated. The adiabatic flame temperature wascalculated using the CEA code [9] and therefore in-cludes dissociation of product species. In the reac-tor’s present operation and configuration, this ratiois an indication of the thermal efficiency. At constanttotal flow rate, leaner mixtures result in reduced op-erating temperatures and, consequently, less heatloss and better efficiencies. However, leanermixtures also imply less power input. For a variationin � at constant input power (i.e., constant fuel flowrate), the results are similar to those presented forconstant total mass flow rate. However, as � is de-creased, the total air mass flow rate becomes signifi-cant, decreasing the residence times in the reactorsubstantially. Short residence times decrease thepercentage of heat loss by convection and radiation,and the majority of the input energy now remains inthe exhaust gas. Thermal efficiencies as indicated byTs/Tad are also higher obtaining a value of 72% at� � 0.1.

The high thermal efficiencies observed at low �,however, come with a tradeoff. In a complete sys-tem, such high mass flow rates will require moreenergy for pumping power, and therefore, the netgains of running extremely lean are not obvious froma systems point of view. Furthermore, short resi-dence times may not be sufficient to obtain complete

reaction in the reactor. Optimization of the aboveparameters will depend on the specific fuel, and itschemical oxidation times, whether they be catalyti-cally produced or principally controlled by gas-phaseoxidation chemistry. As a result of the very nature ofthe present configuration and size of the reactor, itdoes not operate in the excess enthalpy mode as de-scribed above. In such a system, temperature gra-dients would need to be maintained across the re-actor, in contrast to the low Biot number systemstudied here. Future work is needed to isolate re-actor temperature to achieve such gradients at smallscales. One simple methodology is by combinationof several layers of ‘‘thin-sheet’’ layers of similar de-sign to this reactor, with thermoelectric films sand-wiched between the successive layers of the com-posite.

Microreactor Experiments

Experiments were performed in a temperature-controlled laboratory (25.5 � 1 �C) for comparisonwith the simple theory above. Hydrogen was used asthe fuel because of its catalytic reaction character-istics on platinum/gamma alumina catalysts. Hydro-gen and air were supplied from pressurized cylinderswith pressure regulators and a needle valve in eachline to induce the sonic discharge condition for flowcontrol. Flow rates were determined using soap bub-ble meter measurements of each individual flow.The streams were then mixed and delivered to themicroreactor through a 5 cm long section of No. 24stainless-steel hypodermic tubing mated to the ce-ramic microreactor body. An identical arrangementtransferred exhaust from the microreactor to com-position measurement equipment downstream ofthe reactor. The alumina reactor was interfaced withsupply and exhaust tubes using glass welds. The ex-haust is immediately cooled by passing the gasesthrough an ice bath that also removes water fromthe flow to a defined absolute humidity. The cooledgases are then diluted by known amounts of dry ni-trogen such that the oxygen remaining in the streamwill be in a concentration range that can be mea-sured using a Delta F trace oxygen analyzer. Netoxygen consumption was determined to evaluate thechemical conversion efficiency within the micro-reactor. Nickel-chromium/nickel-aluminum (TypeK) and nickel-chromium/copper-nickel (Type E)thermocouples with junction diameters of approxi-mately 25 lm were affixed to the microreactor usingthermally conductive, low-expansion ceramic ce-ment. Thermocouples were located on the face cen-ter and in the middle of each of the two sides of thereactor adjacent to the exhaust/intake port locationat one of the reactor corners. Measured temperaturedifferences were generally within �1.5 �C duringsteady-state operation (even at maximum operating

MICROREACTOR FOR MEMS POWER GENERATION 913

Fig. 5. Experimental microreactor temperature (sym-bols) as a function of input power for three different re-actors at an equivalence ratio of 0.8. Line is the modelprediction.

temperature), experimentally confirming that theoperating reactor was at uniform temperature. Ex-periments were performed to define the range ofoperating parameters over which self-sustained re-action could be accomplished without jeopardizingmechanical integrity of the reactor (through exces-sive operating pressure and temperature). Hydrogenand air flow rates, as well as the steady-state oper-ating temperature of the reactor, were monitoredduring the experiments.

Two types of initiation assistance were employed.For experiments at operating temperatures less than300 �C, the microreactors were enclosed in a Lind-berg clamshell oven, and the monitored oven tem-perature was adjusted using a Vari-AC voltage man-ual adjustment of oven power until reactioninitiation was noted by microreactor temperaturesbeginning to exceed oven temperature. Once reac-tion was stabilized, the oven was turned off andopened, and the reactor reached the reported equi-librium operating temperatures in open ambient air.In a second series of experiments, which investigatedoperating temperatures higher than 300 �C, the mi-croreactor as shown in Fig. 2 was equipped with aKanthal wire heater mounted in the center reactorhole using thermally conducting ceramic cement. Asmall micro-oven was equipped with terminals forthe heating wire, routed through the small hole inthe center of the microreactor. The heating wire waspowered with AC current through an adjustable au-totransformer.

Pressure drop across the reactor was measured asa function of air flow rate. In non-reacting experi-ments, the reactor was heated to 300 �C in the Lind-berg clamshell oven and air was passed through thereactor. At a flow rate of 25 cm3/min, a pressure dropof 6.9 � 103 N-m�2 was measured, while at a flow

rate of 210 cm3/min, a pressure drop of 9.65 � 104

N-m�2 was obtained. Most of the pressure drop isa result of the flow through the No. 24 hypodermictubing, which had an internal diameter of approxi-mately 350 lm. For these flow rates, the pressuredrop is equivalent to a pumping power requirementof approximately 2.8 to 330 mW. As discussed above,lean mixtures and high power may have significantpumping losses.

Two operating regimes became apparent as ex-periments proceeded. Autoinitiation could only beachieved with microreactors that had not experi-enced operating temperatures above about 300 �C.For equilibrium operating conditions below thistemperature boundary, the reaction of H2 and O2cannot occur homogeneously in the gas phase withinthe available reaction residence times, and the re-action must proceed entirely by surface catalysis.However, extensive testing of the catalyst materialsand procedures used (deposited in heated ceramictubes) clearly identifies that the catalyst and supportstructure are degraded by exposure to operatingtemperatures in excess of 300 �C. Without the pres-ence of other stabilizers, the gamma alumina sub-strate reverts to alpha alumina form and the distrib-uted area of the platinum catalyst itself is reducedby sintering and agglomeration of the distributedplatinum catalyst [10]. These observations are con-sistent with literature data. We found that if micro-reactors had been exposed to operating tempera-tures much in excess of 300 �C, room temperatureautoinitiation could no longer be obtained. We alsonoted that catalytic chemical conversion efficiency atlower temperatures was irreversibly degraded. Atoperating temperatures higher than 300 �C, the re-action is apparently sustained within the reactor asa result of both gas-phase chemistry, as well as in-teractions at the microreactor surface. These resultsare consistent with published spontaneous ignitiontemperatures (400 �C) of H2/air mixtures [11]. Suchempirical measurements are, however, strongly af-fected by surface material considerations and surfaceto volume ratio. Experiments show that some hydro-gen conversion can occur on the reactor surfaces ata low temperature even without the presence of cat-alyst. It is also clear that complete conversion of hy-drogen should occur in small passages over somerange of higher gas/wall temperatures that exceedthat required to achieve the critical branching ratiofor chain branched kinetics, inclusive of wall termi-nation effects.

Although the ignition phenomenon was found tobe reactor-dependent, steady-state operating char-acteristics were extremely reproducible. Fig. 5 com-pares the measured operating temperatures forthree reactors as functions of chemical energy inputfor H2/air mixtures (� � 0.8). The reproducibilityof the experimental measurements is remarkablyconsistent over a range of power inputs from about

914 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

Fig. 6. Experimental microreactor temperature (sym-bols) as a function of equivalence ratio for three differenttotal reactor flow rates. Lines are model predictions.

Fig. 7. Body temperatures for high input power opera-tion illustrating regimes dominated by catalytic surface re-action and by combined heterogeneous surface and gasphase chemistry.

2 to 6 W. Below approximately 1.8 W power input,steady-state reactor operation was difficult toachieve. The reactor temperature was observed tooscillate, principally due to water condensation inthe reactor channels as a result of microreactor tem-peratures dropping below the condensation tem-perature. Once the channel was filled, reactant flowwould clear the channel of condensate, permittingreignition.

Comparison of experimental results with model-ing results (the lines in Figs. 5 and 6) shows that themodel described above overestimates the resultingsurface temperature for a given power input. Fig. 6reports data for three different flow rates of reactivemixture with varied energy input by changing equiv-alence ratio. While increased flow rate leads to in-creased energy input, residence time within the re-actor is also reduced as discussed above. Theseresults suggest the estimations of heat losses are rea-sonably accurate and that experimental/model dis-parities are primarily due to chemical conversion ef-ficiency variations. The larger differences between

the modeling results and experiments at high flowrates can be attributed to incomplete conversion asa result of shorter residence times. Catalytic conver-sion also may be a function of equivalence ratio asnoted by the nonlinear behavior of the experimentaldata for the highest flow rate and equivalence ratiomeasurements. Measurements of conversion effi-ciency using the trace oxygen analysis approach con-firm these speculations [10]. High-temperature op-erating characteristics were obtained by operatingthe reactor under increased input power levels.Higher input powers continue to extrapolate thebody temperatures to higher values (Fig. 7).

To demonstrate electrical power generation at thelength scales of the present reactor, the microreactorwas sandwiched between two commercial thermo-electric modules (Melcor Hot 2.0-31-F2A) and twocold reservoirs. Each thermoelectric module con-sisted of 31 p- and n-type pairs of doped bismuth-telluride semiconductors connected electrically inseries and thermally in parallel. The two moduleswere connected in series, and power output wasmeasured across a variable potentiometer. Fig. 8shows that the maximum output power varied fromapproximately 30 to 55 mW at a load resistance of�6 X for chemical input powers of 6.8 to 9.1 W. Asshown in Table 1, the cold reservoir temperature wasmaintained at 23 �C, while the hot reservoir tem-perature increased from �100 to 115 �C. These tem-peratures are considerably lower than those shownin Fig. 5 as a result of the increased heat loss acrossthe thermoelectric modules. The resulting conver-sion efficiency varied from �0.44 to 0.57%. Consid-ering that the modules were not optimized for thepresent conditions (e.g., temperature limitations onthermocouple solder connections), plus only 38% ofthe surface was covered with thermocouples, thepresent results are promising. The bismuth-telluridemodules used here had an upper temperature limitof about 225 �C. Although not yet available com-mercially, several classes of materials (Si-Ge alloys,doped Pb-Te, the recently discovered Skudderuditealloys based on CeFe4Sb12, and perhaps, transitionmetal silicides) have potential for thermoelectricconstructs with higher figures of merit and increasedtemperature limits.

Summary

This paper reports a number of important mile-stones: (1) Self-sustained chemical reaction (withand without catalysis) was demonstrated in smallstructures; (2) conversion of fuel to electrical powerhas been demonstrated for commercial thermoelec-tric systems; (3) micropower generation systems inalumina have been microfabricated using ceramicstereolithography; and (4) some capabilities andcritical issues important to the emerging micropower

MICROREACTOR FOR MEMS POWER GENERATION 915

Fig. 8. Power output from commercial thermoelectric modules as a function of load resistance and chemical inputpower.

TABLE 1Power generation with coupled micro reactor and commercial thermoelectric (TE) coolers

Chemicalpower (W)

H2 flowrate

(cc/min)

Totalflow rate(cc/min) �

Opencircuit

voltage (V)

Maximumpower

out (mW)THOT

(�C)TCOLD

(�C)Open circuitDT (�C) g (%)

gc (%)adjusted forTE-coveredsurface area

6.83 50.1 148 1.21 0.876 30.1 101.2 23 78.3 0.44 1.27.14 146 146 1.01 0.936 35.7 98.5 23 75.5 0.50 1.38.59 176 176 1.02 1.041 44.8 108.2 23 85.2 0.52 1.49.10 188 188 1.06 1.136 52.2 115.5 23 92.5 0.57 1.5

MEMS knowledge base have been identified anddemonstrated.

The present study has shown that because of thesmall length scales and flow rates involved in themicroreactor, a nearly uniform reactor body and gasflow temperature are obtained. In a quiescent en-vironment with an uninsulated reactor, significantheat losses from the surface of the reactor via con-vection and radiation occur. Methods to reduce heatlosses are now under study. However, it should be

noted that even if convection losses were minimizedthrough methods such as vacuum insulation, the ra-diation losses would continue to play a significantrole, unless radiation-shielding approaches are pur-sued. In the present reactor, the catalyst was distrib-uted on the internal channels throughout the reactorbody. Future designs are under development to limitcatalyst location to the center regions of the reactor,benefiting some of the possible approaches to re-ducing heat losses.

916 NEW CONCEPTS IN COMBUSTION TECHNOLOGY—Micro-Power Generation

Acknowledgments

This research was supported by grants from ARO/MURIand DARPA (through subcontract to Dyncorp, Inc.).

REFERENCES

1. Koeneman, P. B., Bush-Vishniac, I. J., and Wood,K. L., J. MEMS Systems 6(4):355 (1997).

2. Nowak, R., ‘‘Batteries and Fuel Cells,’’ MEMS BasedMicroscale Power Generation and Energy ConversionConcepts and Systems, DARPA Workshop, Arlington,VA, February 27–28, 1998.

3. Fernandez-Pello, C., Proc. Combust. Inst. 29:883–899(2002).

4. Sitzki, L., Borer, K., Wussow, S., Schuster, E., Maruta,K., and Ronney, P. D., ‘‘Combustion and Power Gen-eration in Microscale Excess Enthalpy Burners,’’ paper111, U.S. Sections Second Joint Meeting of the Com-bustion Institute, Oakland, CA, March 26–29, 2001.

5. Weinberg, F., Combust. Sci. Technol. 121:3 (1996).

6. Garg, R., Prud’homme, R. K., and Aksay, I. A., J. Opt.Soc. Am. A 15(4):932 (1998).

7. Churchill, S. W., and Chu, H. H. S., Int. J. Heat MassTransfer 18:1323 (1975).

8. Incropera, F. P., and DeWitt, D. P., Fundamentals ofHeat and Mass Transfer, 4th ed., Wiley, New York,1996.

9. McBride, B. J., and Gordon, S., “Computer Programfor Calculation of Complex Equilibrium Compositionsand Applications,” NASA reference publication 1311,Lewis Research Center, Cleveland, OH, 1996.

10. Shinjoh, H., Muraki, H., and Fujitani, Y., in Studies inSurface Science and Catalysis 71: Catalysis and Au-tomotive Pollution Control II (A. Cruzq (ed.), ElsevierScience, Amsterdam, 1991, p. 617.

11. Zabetakis, M., U.S. Bureau of Mines Bulletin 627,1965.

12. Vican, J., Operational Characterization of a CatalyticMicro-Combustor for MEMS Scale Electrical PowerProduction, Fall Independent Work Report, Mechan-ical and Aerospace Engineering, Princeton University,Princeton, NJ.