Proceedings of the 5th International Conference on Microreaction Technology (IMRET 5), Strasbourg, France, May 27-30, 2001
A microreactor for in-situ hydrogen production by catalytic methanol reforming Ashish V. Pattekar and Mayuresh V. Kothare∗ Department of Chemical Engineering, Lehigh University, Bethlehem, PA18015, USA.
Sooraj V. Karnik and Miltiadis K. Hatalis Department of Electrical Engineering, Lehigh University, Bethlehem, PA18015, USA.
Abstract A detailed study of the theoretical and experimental issues involved in the design and operation of a silicon-based methanol microreformer is presented in this paper. Thermal simulations for heat loss from the reformer chip and calculations involving the endothermic heat effect of the reforming reaction were carried out to estimate the total power requirement for continuous operation of the reformer. Micromachining technology was utilized to fabricate a prototype silicon microchannel based reformer of channel cross-section 1000µm x 230µm with a copper layer of thickness ~ 33nm as catalyst. The reactor chip was interfaced with the tubing for reactant and product transport using a stainless steel housing machined to exact dimensions and flexible graphite pads to provide gas-tight seals and excellent thermal contact between the chip and the housing for external heating of the reformer. Runs of this prototype microreformer were carried out using a custom made experimental setup for generation of the reactant mixtures, temperature control of the microreactor and online measurement of the product gas composition. Results from test runs of the microreformer are presented and strategies for improving the hydrogen yield are discussed.
1. Introduction The use of microreactors for in-situ and on-demand chemical production is gaining increasing importance as the field of microreaction engineering matures from the stage of being a regarded as a theoretical concept to a technology with significant industrial applications. Various research groups have successfully developed microreactors for chemical processing applications such as partial oxidation of ammonia [10], nitration [2] and chemical detection [4]. The objective of the research effort at the Integrated Microchemical Systems Laboratory at Lehigh University’s Chemical Engineering Department is to demonstrate a working microreaction system for use as a sustained source of hydrogen fuel for proton exchange membrane (PEM) fuel cells through catalytic steam reforming of methanol. The complete reformer-fuel cell unit is proposed as an alternative to ∗
Corresponding author. Phone: (610)-758-6654; Fax: (610)-758-5057; E-mail:
[email protected]
conventional portable sources of electricity such as batteries for laptop computers and mobile phones due to its ability to provide an uninterrupted supply of electricity as long as a supply of methanol and water can be provided. Though considerable work already exists in literature on the catalytic steam reforming of methanol for production of hydrogen using conventional reactors [1,7,8], the use of microreactors for in-situ methanol reforming is a relatively new idea [3,6,11,12]. Literature on the macro-scale steam reforming of methanol includes analysis of the reaction thermodynamics [1] for prediction of optimum reactor temperature and feed compositions, catalyst characterization studies [7], and experimental studies on macro-scale pilot reactors [8]. Results obtained in the study of methanol reforming in these conventional reactors form a good background for the development of microreactors for this purpose. Silicon is considered a good material for fabrication of microreactors due to the high strength of the Si-Si bonds which results in the chemical inertness and thermal stability of silicon. Well established silicon micromachining techniques commonly used in the microelectronics industry facilitate easy fabrication of microchannels and other desired features on silicon substrates thus making silicon the ‘material of choice’ for microreactor fabrication. In the following sections we discuss the theoretical and experimental issues in the development of a prototype silicon chip based microreformer. Preliminary calculations giving an idea of the power required to operate the microreformer are presented followed by a description of the microreformer fabrication procedure. The experimental setup for carrying out test runs of the microreformer is described and results from the test runs are presented.
2. Preliminary Simulations Preliminary simulations on a Silicon chip of size 2cm x 2cm were carried out using the MEMCAD design and simulation software [9] to study the effects of fluid flow and heat loss from a Silicon chip operating at the reforming temperature of 500K. Figure 1a shows the velocity profile (velocity units: µm/sec) of the fluid (water) flowing through microchannels of cross section 1000µm x 500µm. The channel geometry is such that the fluid enters along the negative z axis direction through the two inlets, mixes at the intersection of the two inlets and leaves along the positive z axis direction at the outlet. The boundary condition used for the flow simulation was a constant flow rate of 0.1 cc/sec at each inlet with the outlet being at atmospheric pressure, with no slip at the channel walls. Figure 1b shows the results of a thermal simulation on the same chip with a temperature boundary condition of 500K at the bottom wall of the microchannels in addition to the flow boundary conditions described above. The fluid inlet temperature was taken as the room temperature (300K) and the heat loss from the rest of the chip surface was governed by natural convection to ambient air at 300K.
Figure 1(a): Flow through a simple microchannel
Figure 1(b): Temperature gradients within bulk Si
As can be seen in Figure 1b, most of the chip surface reaches a temperature very close to the maximum of 500K within a short distance from the inlets, which are at 300K due to the incoming fluid being at room temperature. The excellent thermal conductivity of bulk silicon results in close to uniform temperature distributions within the bulk with most of the chip temperature being very close to the maximum (500K for this calculation) when the heat loss from the chip to the surrounding is governed by relatively weaker modes of heat transfer such as natural convection. This conclusion is especially relevant for the operation of high temperature microreactors since this means that Silicon based microreactors for high temperature reactions will have most of their surface at the high reaction temperature resulting in significant heat losses to the surrounding unless sufficient thermal isolation from the ambient is provided. Since the optimum temperature for the methanol reforming reaction is about 250 oC [1], it follows that heat integration and thermal isolation issues will be extremely important for efficient operation of the microreformer. For example, the heat loss from a Silicon chip of size 2cm x 2cm operating at 250 0C is about 4.5 Watt as governed by natural convection to air at 25 0C with a natural convection heat transfer coefficient of 25 W/m2K. A hydrogen flow rate of 0.3721 gmol/hr is needed for a 20-Watt fuel cell assuming 80% hydrogen utilization, which translates into a methanol flow rate of 0.1536 gmol/hr at 85% conversion inside the reformer with a selectivity of 95% towards the reforming reaction. Then the power required for the endothermic reforming reaction (with the reformer operating at 250 oC) at this flow rate is about 2.038 Watt as shown in the calculations below using data on the heat of reaction and specific heats of the components of the reaction mixture[5]. The steam reforming of methanol for hydrogen production in the presence of Cu/ZnO catalyst involves the following reactions at 250 oC:
Primary reactions: CH3OH + H2O ⇔ CO2 + 3H2 CH3OH ⇔ CO + 2H2
(∆H298K = 48.96 KJ/mol) (∆H298K = 90.13 KJ/mol)
(1) (2)
(∆H298K = 41.17 KJ/mol)
(3)
Secondary Reaction: CO2 + H2 ⇔ CO + H2O
Considering only the main reforming reaction, ∆H298K = 48.96 KJ/mol 523
⇒ ∆H250C = ∆H523K = ∆H298K +
∫
∆Cp(T)dT
= 59.17 KJ/mol
298
⇒ Heat of reaction requirement for endothermic reforming reaction = 59.17 x 0.1536 x 0.85 x 0.95 /3600 KJ/sec = 2.038 Watt Thus a large part of the power required to run the reformer is due to heat loss from the reformer chip to the surrounding and can be reduced by effective packaging and insulation of the microreformer.
3. Microreformer Fabrication Fabrication of reactor microchannels of cross sectional dimensions 1000µm x 230µm on the prototype silicon reformer chips was carried out using standard silicon micromachining techniques such as Photolithography and KOH etching. We used a single mask process to get four identical microreformers from a single 100mm silicon wafer polished on both the sides. Silicon nitride was deposited by plasma enhanced chemical vapor deposition (PECVD) on both sides of the wafer. Photolithography was done on the back side of the wafer to obtain the microchannel pattern. Silicon nitride on the back side was then etched by plasma etch to get pattened nitride which acts as an etch stop during KOH etching of silicon. Bulk silicon was etched in KOH from the back side to obtain microchannels of desired depths. About 33 nm of copper (catalyst layer) was then sputter deposited. Figure 2 illustrates the major steps involved in the fabrication of the microreformer chips. As will be discussed in the section on the experimental set-up and external interfacing of the microreformer, one end of the channel needs to have a hole in the silicon substrate for the dosing of reactants into the channels. This hole was made after fabrication of the channels by etching through the wafer with KOH. A protective coating of black wax prevented etching of the rest of the wafer. This technique exposes only one end of the channel (an area of
approximately 1mm2) to the etchant, which results in the desired hole after completion of the etching process. This step is illustrated in Figure 2(d). Figure 3 shows a schematic of the final reformer chip. Figure 2: Microreformer fabrication steps ( a) Si wafer with nitride ( b) Plasma etch to expose Si
( c) KOH etching of bulk Si
( d) Formation of inlet hole Unprotected channel end
Inlet Hole
KOH etch Removal of black wax Black Wax
( e) Copper deposition by sputtering Cu layer
Figure 3: The fabricated microreformer 36 mm
Inlet hole 36 mm Channel width 1000 µm, depth 230 µm, Copper layer 33 nm
4. External Interfacing The interfacing of the microreformer chip to tubing for reactant and product gas transport was done using a custom made stainless steel housing shown in Figures 4 and 5. The steel housing blocks have ridges of appropriate length connecting the central bore to appropriate points on the reformer. The ridge on one of the steel blocks was machined to face the reactant inlet hole on the reformer chip when the chip was properly aligned with the housing. Similarly, the product gases exit from the other end of the channel through a similar ridge in the housing block machined to face the other end of the channel when aligned correctly. Figure 4: Schematic of microreformer housing Flexible graphite pads with holes aligned to microreformer inlet/outlet Ridges for gas flow in housing
Microreformer inlet Swagelok connector
Exhaust tubing (To gas analyzer)
Microchannels
Feed tubing
Microreformer outlet
Reactor housing Microreformer
Leak-proof sealing between the reformer chip and the steel housing was ensured by using flexible graphite pads acting as gaskets with holes drilled at the right positions for reactant and product transport. Thus the reformer chip was covered from either side by a graphite covering pad to provide a leak-proof conduit for reactant and product gas transport between the housing and the reformer. The tubing for gas transport was connected to both sides of the housing via standard Swagelok connectors so that the reactants entered the housing-microreformer assembly from one side, passed to the microreformer through the hole in the silicon substrate and exited the reformer through a similar hole in the covering graphite pad at the other end of the microchannel after flowing over the deposited catalyst. The advantage of using flexible graphite pads for sealing purposes was that graphite being a good conductor of heat, the reactor temperature could be maintained at 250 oC by heating the steel housing using heating tapes and suitable temperature controllers. This allowed testing of the reformer chip without having to fabricate on-chip resistive heaters and temperature sensors, which will be integrated with the microreformer in future.
Figure 5: Microreformer housing Reactor housing
Reformer Chip
5. Experimental set-up The operation of the microreformer involved: 1. Setting up of the methanol vapor and steam source 2. Interfacing of the microreformer chip to external tubing for transport of reactants and products as discussed in Section 4, and 3. Connection of the reactor exhaust tubing to the gas analysis equipment (a quadrupole mass spectrometer) for online analysis of the product gas composition. A gas-tight sample cylinder containing a liquid methanol-water mixture immersed in a constant temperature hot water bath maintained at temperatures in the range 80 oC – 95 oC was used for generation of methanol vapor-steam mixtures of desired compositions by manipulating the composition of the liquid mixture in the sample cylinder on the basis of methanol-water vapor-liquid equilibrium (VLE) data. Analysis of the VLE data for the mixture and pressure drop calculations from simulations using MEMCAD confirmed that enough pressures could be generated in this set-up to drive the flow of the reactants through the reformer. The product gas tubing coming from the microreformer housing connected to the mass spectrometer, which gave an online analysis of the composition of the product gases. Figure 6 shows a schematic of this set-up. Electric heating tapes and simple on-off controllers with thermocouple probes were used to maintain the temperature of the connecting tubing above 100 oC to avoid condensation of the vapors during flow. A pressure transducer-strain gage meter combination was used to continuously monitor the pressure in the sample cylinder. The reactor housing was also maintained at the temperature required for the reforming reaction (around 250 oC) using electric heating tapes and an autotuning PID controller with a thermocouple sensor which maintained the reactor temperature to within 1 oC of the setpoint.
Figure 6: Schematic of experimental set-up Pressure sensor
Flow control valve
Microreformer housing
Reactor exhaust
o
250 C using heating tape
Release valve
o
Hot water bath (90 C) Sample cylinder Mass Spectrometer
The gas analyzer used for analysis of the product gas composition was a quadrupole ion-trap mass spectrometer based residual gas analyzer (RGA) with a pressure-reducing inlet allowing the sampling of gases at atmospheric pressure. The pressure reducing inlet capillary of the analyzer was also maintained at above 100 oC using heating tapes to avoid any condensation before the pressure was sufficiently reduced to obviate condensation of water and methanol at room temperature inside the capillary. The mass spectra were collected at short intervals and the peak amplitudes were converted to mole fraction data for conversion calculations.
6. Results and Discussion I. Test Runs Test runs of the microreformer were carried out to verify hydrogen production capability inside the microreformer and to get an idea of the conversion possible using this set-up. Though the reformer design wasn’t optimized from the point of view of conversion or hydrogen yield, the preliminary runs gave important insights into the operation of the microreformer in this set-up. For the test run reported here a 50:50 mixture of methanol and water was fed into the sample cylinder maintained at 85 oC using the hot water bath shown in Figure 6. Using methanolwater VLE data at 85 oC obtained using the UNIFAC method in Aspen Plus, the vapor mixture at equilibrium with this liquid will have methanol vapor mole fraction (Ymethanol) of about 0.76. The methanol vapor-steam mixtures generated in this manner were passed through the reformer maintained at 250 oC and the product gas mass spectra were obtained at regular intervals during the run. The mass spectra were converted to mole fraction data using the library for molecule fragmentation inside the mass spectrometer ionizer. A plot of the exit gas composition (mole fraction) v/s time (sec.) for a microreformer test run of 500 sec is shown in Figure 7. As can be seen from the plot, sufficient amount of hydrogen was detected at the reactor exit to confirm that reforming was indeed taking place in the microreactor.
Figure 7: Microreformer product gas composition (mole fraction) v/s time (sec)
Looking at Average values of the mole fraction of each component in the product gas for this run can be used to get a crude estimate of the hydrogen yield. These values were 0.755 for methanol, 0.21 for water, 0.070 for hydrogen, 0.00129 for carbon dioxide and 0.017 for carbon monoxide. Though the hydrogen yield obtained in this prototype unoptimized microreactor was not significant (approximately about 0.092 moles H2 produced per mole CH3OH fed into the reformer with an average flow rate of about 2.36 x 10-3 gm/sec of methanol-steam mixture at the microreformer inlet with inlet methanol mole fraction of 0.76) these results demonstrate the capability to carry out methanol reforming in a microreactor and to test the microreactor operation at different operating conditions. Future work will involve attempts to improve the hydrogen yield by increasing the catalyst contact area per unit channel length and optimizing microchannel geometries for more efficient utilization of available total chip area. Experimental runs with reformer chips having the composite copper-zinc oxide catalyst instead of using only a copper layer will also be carried out to study the effect of catalyst properties on reaction extent. II. Catalyst Deactivation Referring to the chemical reactions (equations 1, 2 and 3 in Section 2) it should be noted that the amount of CO and CO2 produced in the test runs as obtained from the mass spectra was less than what would be expected from the reaction stoichiometry for the observed production of H2. One possible reason for this could be the deposition of minute quantities of carbon during the reaction in the microchannels which would reduce the amount of CO and CO2 present in the exit gases [1]. Carbon deposition on the catalyst surface can have several undesirable
effects affecting the product purity. It was also observed that the copper layer in the microreformer degraded and turned blue (as shown in Figure 8) after several test runs probably due to the formation of some compound of copper in the presence of the reaction mixture at the high reforming temperature of 250 oC. These issues will be important in refining the design of the microreformer since catalyst deactivation will lead to lower yields and some form of catalyst regeneration will have to be incorporated for restoring good efficiency of the reformer operation. Figure 8: Degradation of copper layer in the microreformer
Inlet hole Channel width 1000 µm, depth 230 µm
Degraded copper layer (blue)
III. Conclusion The main contribution of this work is the demonstration of a microchannel based methanol reformer for small-scale hydrogen production and the development of a set-up for carrying out experimental runs of the microreformer. A prototype silicon based microreformer was successfully fabricated and tested using a setup designed for easy operation and replacement of microreformer chips and testing of different microreformer designs. Preliminary results from the test runs confirmed the presence of hydrogen at the reformer exit and revealed the need for extensive optimization of the reformer design for obtaining good hydrogen yields. Catalyst degradation in the microchannels was also observed. The optimization of the reformer design and methods to deal with catalyst deactivation issues will be an important part of future work in this project. Acknowledgment: Financial support for this project from the U.S. National Science Foundation under the ‘XYZ-on-a-chip’ grant CTS-9980781, the Pittsburgh Digital Greenhouse and Sandia National Laboratories is gratefully acknowledged. We also acknowledge the assistance of undergraduate Chemical Engineering student Joshua Tilghman in machining of the microreactor housing and setting up of the experimental apparatus for testing the microreformer.
References [1] Amphlett, J., M. Evans, R. Jones, R Mann and R. Weir. Hydrogen production by the catalytic steam reforming of methanol Part 1: The thermodynamics. Can. J. Chem. Engng. 59:720-727, 1981. [2] Antes, J., T. Tuercke, E. Marioth, K. Schmid, H. Krause and S. Loebbecke. Use of microreactors for nitration processes. Proceedings of the Fourth International Conference on Microreaction Technology. 194-200. Atlanta, GA, 2000. [3] Fitzgerald, S., R. Wegeng, A. Tonkovich, Y. Wang, H. Freeman, J. Marco, G. Roberts and D. VanderWeil. A compact steam reforming reactor for use in an automotive fuel processor. Proceedings of the Fourth International Conference on Microreaction Technology. 358-363. Atlanta, GA, 2000. [4] Floyd, T., K. Jensen and M. Schmidt. Towards integration of chemical detection for liquid phase microchannel reactors. Proceedings of the Fourth International Conference on Microreaction Technology. 461-466. Atlanta, GA, 2000. [5] Perry, R., and C. Chilton. Chemical Engineers’ Handbook. Mc.Graw-Hill Book Company. 1973. [6] Pfeifer, P., M. Fichtner, K. Schubert, M. Liauw and G. Emig. Microstructured catalysts for methanol steam reforming. Proceedings of the Third International Conference on Microreaction Technology. 372-382. Frankfurt, Germany, 1999. [7] Raphel, O., and N. Bakhshi. Production of Hydrogen from Methanol. 1. Catalyst characterization studies. Ind. Eng. Chem. Res. 33:2047-2055, 1994. [8] Raphel, O., and N. Bakhshi. Production of Hydrogen from Methanol. 2. Experimental studies. Ind. Eng. Chem. Res. 33:2056-2065, 1994. [9] Senturia, S., R. Harris, B. Johnson, S. Kim, K. Nabors, M. Shulman and J. White. A computer-aided design system for microelectromechanical systems (MEMCAD). Journal of Microelectromechanical Systems. 1:3-13, 1992. [10] Srinivasan, R., I. Hsing, P. Berger, K. Jensen, S. Firebaugh, M. Schmidt, M. Harold, J. Lerou and J. Ryley. Micromachined reactors for catalytic partial oxidation reactions. AIChE Journal. 43:3059-3069, 1997. [11] Tonkovich, A., S. Fitzgerald, J. Zilka, M. LaMont, Y. Wang, D. Vander-Wiel and R. Wegeng. Microchannel chemical reactors for fuel processing applications. Proceedings of the Third International Conference on Microreaction Technology. 364-371. Frankfurt, Germany, 1999. [12] Zilka-Marco, J., A. Tonkovich, M. LaMont, S. Fitzgerald, D. VanderWiel, Y. Wang and R. Wegeng. Compact microchannel fuel vaporizer for automotive applications. Proceedings of the Fourth International Conference onMicroreaction Technology. 301-307. Atlanta, GA, 2000.
