Integrated Microreactor System for Gas-Phase Catalytic Reactions. 1

DuPont Company, Central Research & DeVelopment, Experimental Station, Wilmington, ... microreactor that is based upon a multilayer laminate microelect...
0 downloads 0 Views 1MB Size
8292

Ind. Eng. Chem. Res. 2007, 46, 8292-8305

Integrated Microreactor System for Gas-Phase Catalytic Reactions. 1. Scale-up Microreactor Design and Fabrication D. J. Quiram and K. F. Jensen Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

M. A. Schmidt Department of Electrical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

P. L. Mills,* J. F. Ryley, M. D. Wetzel, and D. J. Kraus DuPont Company, Central Research & DeVelopment, Experimental Station, Wilmington, Delaware 19880-0304

The design and fabrication of a gas-phase microreactor that is based upon a multilayer laminate microelectronics structure is described. The reactor is a key component in an integrated system whose platform utilizes a commercial computer chassis with modular boards to perform the required process functions. The design combines knowledge from earlier laminate microstructures with new prototyping concepts for incorporation of various on-board devices. A 3-D finite element simulation model was used to identify various design improvements. The final device contains two parallel reaction channels on a chiplike die in which a 1 µm platinum film catalyst is deposited on the underside of a silicon nitride membrane. Seven platinum heaters and temperature sensors are evenly distributed along the top side of the silicon nitride membrane. Electrical contacts for the on-board control and sensing devices are achieved through various pins that are distributed around the reactor die. The experience and knowledge gained in developing the final reactor device is utilized in Part 2 of this series for reactor packaging and development of the integrated system modules. 1. Introduction Microreactor technology (MRT) has evolved over the past decade into an established discipline that has applications ranging from accelerated catalyst screening for chemicals,1 petroleum intermediates,2 fine chemicals and intermediates,3 and pharmaceuticals4 to small-scale manufacturing of agricultural chemicals, specialties, and other classes of organic chemicals.4-8 In addition to these more traditional chemistry and chemical engineering technologies, MRT has been recently applied to other emerging technologies, such as hydrogen production from hydrocarbons9 and development of microscale prototype fuel processors.10,11 MRT and microsystem-based processes are also being used in materials science to accelerate the synthesis of new liquid crystals,12 in biotechnology to study complex Baeyer-Villiger oxidation kinetics,13,14 and in nanotechnology to generate novel nanoparticles.15 Miniaturization is also playing a vital role in improving technologies that are accelerating discovery research or elucidating new phenomena, such as microscale analytical separations,16 microparallel chromatography,17 combinatorial microfluidics,18 multiphase-flow microhydrodynamics,19 microscale adhesion mechanics,20 and multidimensional microflow imaging.21 To construct MRT components, new special-purpose materials of construction, fabrication tools, and fabrication techniques are also being created, such as those for fabrication of microcomponents consisting of specialty glass,22 novel five-axes micromilling machines,23 and micro ultrasonic welding tools for the joining of tiny polymer-based components.24 Miniaturization of process devices has also lead to the invention of various special-purpose * Corresponding author. Current address: Department of Chemical and Natural Gas Engineering, Texas A&M UniversitysKingsville, Kingsville, TX 78362-8202. E-mail: [email protected]. Tel.: (+1) 361 593-4827. Fax (+1) 361 593-2106.

process components, such as microstructured mixers,9 ceramic heat exchangers,25 microscale gas-liquid reactors,26,27 and fluid-fluid separators.28-30 More detailed descriptions of these and other microscale process components is provided in several recent reviews5-7 and monographs31,32 on the subject. The evolution of MRT over the past decade has been largely captured through conference proceedings associated with the annual International Conference on Microreaction Technology (IMRET) that was initiated with IMRET-1.33 Since this first gathering in Europe, seven IMRET conferences have been organized on an alternate basis by the AIChE in North America34,36,38,40 and by DECHEMA in Europe,35,37,39 with the latest one being IMRET-9.41 In addition to these conferences, recent progress on microcomponents, microprocesses, and mathematical modeling is described in a number of excellent review articles,42-44 various monographs,31,32,42,45,46 and in the proceedings of dedicated workshops.47-49 It is also generally agreed that microscale systems represent a definitive methodology for process intensification.27,50 The latter is also consistent with the often-cited advantages of microreaction systems when compared to conventional reactor systems, which include (i) safer operation; (ii) simplified scale-up; (iii) improved heat and mass transfer characteristics; and (iv) application flexibility.31,43,51 The utility of MRT for safe operation of hazardous reactions has been a common theme in recent industrial applications of the technology.5-8,52,53 Development of prototype microreaction systems for industrial applications was originally motivated by safety considerations, since it was surmised that hazardous or unstable chemicals could potentially be manufactured upon demand, thereby eliminating or minimizing the need for either storage or transportation.54 Target molecules were inorganic or organic compounds that contained functional groups needed for synthesis of application-driven high-valued added molecules,

10.1021/ie070107w CCC: $37.00 © 2007 American Chemical Society Published on Web 08/23/2007

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8293

such as agricultural chemicals. Examples of these target molecules included hydrogen cyanide, methyl isocyanate, butyl isocyanate, cyclohexyl isocyanate, and phosgene.55 The synthesis reactions generally involved the use of solid heterogeneous catalysts or fast acid-base chemistries, were highly energetic (e.g., ∆Hr > -50 kcal/mol), occurred in the presence of corrosive reaction media, and could sometimes produce explosive mixtures because of the presence of a hydrocarbon and O2. For these reasons, early microreactor design efforts considered various methods for on-board reactant mixing, thermal energy management and heat integration, techniques for micromachining methods using various materials of construction, methods for controlling reaction light-off, and methodologies for incorporation of solid catalytic materials.54 To meet the above process requirements in a compact microreactor package, various system components were integrated through creation of composite layers that were held in close contact by fusion bonding. The design concept is described in several patents that teach the use of a multilayered laminate structure for conducting chemistry in a chip.56,57 However, these efforts did not address the problem of connecting arrays of reactors to increase overall throughput or integration of microreactors into other system components to create an automated, functioning system.58 The primary objective of this work is to design and demonstrate the operability of a stand-alone gas-phase catalytic microreactor system whose key process functions are integrated to form a packaged system. A gas-phase system was selected to eliminate certain hardware and operational complexities associated with gas-liquid mixtures. However, the basic design concepts could be extended in future work to handle these and other multiphase flow systems. The effort described here was actually part of a coordinated program whose objective was to provide the technology foundation needed to develop future military systems requiring chemical or biological synthesis and analysis as part of their operation.59 If successful, these capabilities could potentially provide the modern soldier and the associated military support operations with programmable micromachines that could perform hundreds of fluid-based process sequences to meet dynamic requirements in chemical and biological analysis as well as chemical synthesis. It was also surmised that demonstration of these capabilities could also provide new knowledge needed for other emerging applications where microreactor systems could have an impact, such as those involving synthesis of novel nanomaterials, biotechnology, alternate energy processes, pharmaceutical discovery and manufacture, and specialty chemicals. For these reasons, the emphasis here was placed on demonstrating the operability of a prototype microelectromechanical system (MEMS) from first principles that had the same functionalities that are present in a more conventional laboratory-scale system versus developing a new automated microreactor system for studying gas-phase solidcatalyzed reactions.60 This work addresses the key challenges in microreactor systems design, packaging, integration, automation, and handling of process hazards. Incorporation of various sensors and devices for process monitoring using a high degree of process automation and control is also achieved to demonstrate safe operation of potentially explosive reaction mixtures involving hydrocarbons or NH3 and O2. A unique feature of the microreactor system described here is that it integrates on/off gas flow valves, gas flow rate control, pressure sensors, temperature sensors, microreactors, process control electronics, and other key process functions by using modular components that fit inside a commercially available

computer-type chassis. The modular components are designed and fabricated from first principles as prototype electromechanical computerlike boards that are configured to perform one or more process-specific functions. Parallel reactor operation is demonstrated by configuring the system for simultaneous operation of four independent gas-phase reaction channels. This was accomplished by etching two parallel rectangular reaction channels into a silicon wafer-type substrate and incorporating on-board electronic devices for independent control of the reactant gas flow rate and local axial temperature of the reaction channel. Another dedicated computer board containing microvalves and mass flow controllers was designed for blending the individual reactant gases from cylinder supplies and directing a controlled flow of the resulting mixture to each microreactor channel. The feed and product gases for the individual reactors were sampled on-line using an array of heated multiport valves according to a user-defined sampling sequence. The subsequent compositional analysis of an on-line gas sample was performed using a multidimensional gas chromatographic method based on capillary and packed columns. A real-time programmable logic controller (PLC) is used to control the microreactor system with a host PC, allowing operator interaction. These systemoriented aspects are described in Parts 3 and 4 of this series. The primary focus of the current work is to describe the design and development of the scale-up microreactor, including the devices and methods used in the fabrication sequence. The use of detailed finite-element models to provide guidance on how various material properties and sensors impact the microreactor performance is also described. This is part of an effort to show how modern design tools can be used to decrease the amount of trial-and-error that might otherwise be involved in the design and prototyping of MEMS-based microreactor devices. Despite using these fairly sophisticated predictive design tools, it will be evident that real-time operating data under hot conditions is still a required ingredient when developing a scale-up microreactor chip. 2. Scale-Up Microreactor Development 2.a. Overview. A microelectronic fabricated device or system (MEMS) is typically designed and fabricated using information derived from engineering analysis and prior art, if any exists. However, two notable design challenges for complex MEMS prototypes include the lack of previous operating experience and the availability of special-purpose computer-aided design tools. In some cases, detailed simulation tools can be used to design a prototype, but they are not universally applicable. This section describes various options that were considered and the approach that was eventually used to guide the design of the microreactor configuration for incorporation into the integrated system package. CoventorWare is an example of one popular software package commonly used for the design of MEMS.61 Early versions included the capability to model moving microstructures, the effect of electromagnetic fields, heat transfer, and fluid flow in a single package. A module for simulation of multiphysics with reactions was added more recently, but no open literature data exist that provide statistics on model predictions versus actual microreactor performance data. An alternative approach for simulating reactive flows on the microscale is to employ software that is commonly used for the design of macroscale reactors and for engineering analysis of complex fluid-flow systems, such as CFD-ACE+ and Fluent.62 However, nonisothermal microscale reactive flow problems are more challenging to solve using these and other related simulators, especially if

8294

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007

Figure 1. Schematic of the T-microreactor: (A) top view; (B) cross section perpendicular to flow; and (C) cross section parallel to the flow.

a high degree of coupling exists between the governing equations for momentum, heat, and mass transfer. These challenges mainly occur due to various nonlinearities as well as the magnitudes of the gradients and boundary layers that can exist between the various dependent variables and model parameters, especially over the small length-scales that are typically involved. However, recent 2-D analysis of novel micromixers demonstrates that codes, such as Fluent, when carefully implemented, can produce reliable simulation results.63,64 Additional experience using these codes on other complex microreactor applications with experimental validation will be required until they can be used with a higher degree of confidence. An alternate simulation approach, which is the one used here, first converts CAD drawings of a particular microreactor configuration into a computational mesh. The mesh is then imported into a special-purpose finite-element based simulation code for 3-D analysis of the microreactor behavior that utilized previous experience with modeling of semiconductor device processes.65 A “T” microreactor design was selected as the starting basis since operating data and simulation results were available from previous work.66,67 After various design modifications were proposed as part of developing a packaged system, detailed numerical simulation tools were used to analyze the microreactor performance before it was fabricated. This approach reduced the number of design iterations that were required to develop a final microreactor design for incorporation into the scaled-up system. It also allowed subsequent development of the electrical and fluidic packaging, since these elements were dependent on the final microreactor design. The methods used for packaging the microreactor and other supporting components would also dictate some key features of the microreactor. This section provides a description of key aspects on the reactor design and summarizes the key findings from simulation studies. 2.b. T-microreactor Description. The original T-microreactor design consisted of two inlet flow channels with flow sensors and a reaction channel with local heating and temperature sensing devices. The top and cross-sectional views of this particular design are illustrated in Figure 1. The branches of the T-structure correspond to the supply channels for two individual reactants. Each channel contains integrated flow-rate sensors. The individual gas feeds first contact each other at the junction of the tee. The gases then undergo mixing and reaction as they traverse the length of the main channel in the catalyst region. The channel dimensions are nominally 0.5 mm wide and 0.5 mm deep. The channel is sealed on the top by a 1 µm

thick silicon nitride membrane that is impermeable to gas flow. The channels are sealed from the bottom by an aluminum plate that is bonded to the reactor die using epoxy. Gas enters and exits the microreactor through holes that are drilled into the aluminum sealing plate. The microreactor is fabricated by starting with a single-crystal silicon wafer and then using photolithography and etching techniques to define and create the various channel structures and metal layers. Figure 1 shows that the reaction channel has platinum (Pt) heaters and temperature sensors on the outer side of the membrane and a Pt catalyst on the reaction channel side. Five heater segments are evenly distributed over the reaction channel, while the two flow sensors are located over the respective gas inlet channels. The heater segments are 3.55 mm in length and 0.3 mm wide, and they are centered in the channel. Each segment has one temperature sensor located on the upstream end and one located on the downstream end. These structures are fabricated using photolithography techniques to define the shape. Physical vapor deposition (PVD) is used to form the Pt layer. 2.c. T-microreactor Modeling. Analysis of the modifications for the scale-up microreactor was based upon a simulation methodology that was previously developed to analyze the performance of the original T-microreactor design. This methodology employed models that are based on the microscopic forms of the conservation equations (total mass, momentum, energy, and species) for single-phase flow. A detailed description of these equations and the boundary conditions used in the problem formulation can be found elsewhere,68,69 so they will not be repeated here. Reactions are typically conducted using flow rates between 10 and 20 cm3/min at standard conditions (sccm) corresponding to mean residence times between a few seconds to tens of seconds. These ranges are not restrictive and can be expanded, depending on the particular application. On the basis of the channel dimensions shown in Figure 1, the Reynolds number can vary from 10 to 50, depending on the properties of the gas at the local reactor temperature. Because of the small channel dimensions of the gas-phase microreactor, the flow regime is laminar so that turbulence models are not required. Moreover, when flow rates on the lower end of the spectrum are utilized, diffusive processes dominate over convective processes. For this reason, the governing equations exhibit strongly elliptic versus hyperbolic behavior. Consequently, the field variables do not typically exhibit shocks or sharp spatial fronts, so special front-tracking methods are not required.

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8295

The traditional issues that occur when modeling larger-scale monolith or packed-bed reactors can also occur in microreactor systems. Specifically, the species, momentum, and energy balances are strongly coupled for chemistries involving highly exothermic or endothermic reactions. However, the physical dimensions of microreactors generally introduce new challenges that are infrequently encountered when modeling their largerscale counterparts. For example, when modeling the T-microreactor, the length scales of important features vary widely from ∼0.5 mm (channel width or depth) to 0.0001 mm (Pt layer thickness). Another complexity that occurs when modeling reacting flows for the T-microreactor is the rather large temperature range that can occur inside the reaction channel, e.g., 125-500 °C in the case of ammonia oxidation. This requires a robust kinetic model for the system of interest, which is difficult to identify since most kinetic data is collected using conventional laboratory reactors over a relatively small temperature range. The latter may require special-purpose experiments to collect the required kinetic data needed for intrinsic kinetics model development. The complexities associated with transport-kinetic interactions that can occur within the small physical confines of a microreactor are partly offset, at least in some applications, by the use of thin-film catalysts. Referring to parts b and c of Figure 1, a Pt catalyst is deposited as a film on the underside of the silicon nitride membrane of the T-microreactor. This design feature avoids the need to include microscopic mass and energy balances to describe coupled heat and mass transport effects in porous catalyst particles.70 In the detailed microreactor model, the reaction is confined to the surface of the thin-film catalyst, so it is incorporated into the local mass and energy balances in the form of appropriate boundary conditions. Nevertheless, complex transport-kinetic phenomena, such as ignition-extinction behavior, can still occur in microreactors with thin-film catalysts.65-69 These observations point to the need for using detailed simulation models and to avoid the pitfalls of using simplified approaches that might be tempting to follow because of the small dimensions involved. 2.d. Results and DiscussionsT-microreactor Modeling. Appendix A in the Supporting Information provides details on the methods used to validate the simulation model and subsequent application of the model to analyze the impact of various microreactor physical design modifications on performance. The physical basis for the simulation model and numerical aspects are outlined in Sections A-1-A-3. As explained in Section A-4, the model correctly describes the experimentally measured temperature responses to changes in heater power for a particular heater segment in the absence of reaction. The ability of the model to predict ignition/ extinction phenomena when the oxidation of ammonia was used as test reaction over a Pt-film is summarized in Section A-5. Collectively, these two investigations provide sufficient validation of the model for its subsequent use as a predictive tool for microreactor design modifications. Microreactor design issues were focused on the reactor membrane, the microflow sensor, and the reactor heater segment. Details of this study are described in Appendix B of the Supporting Information. These hardware elements were selected since prior experience with earlier versions of the T-microreactor showed that reactor performance was particularly sensitive to variations in these parameters and that the existing specifications were suboptimal.66,67 Particular aspects that were investigated to improve microreactor performance included the following: (1) the effect of the membrane thickness on the rate of heat removal from the reaction zone; (2) the effect of flow sensor

heater length, heater power input, heater location, gas transport properties, and membrane properties (thickness and material of construction) on sensor response; and (3) the effect of reactor heater geometry, gas flow rate, and magnitude of reaction exothermicity on heater response and reactor thermal behavior. Key findings from these studies are summarized below. 2.d.1. Microreactor Membrane. To assess the role of the membrane properties on performance, observed sensor temperature versus heater power curves were calculated using membranes fabricated from various materials of construction with various thicknesses. The specific goal was to prevent ignitionextinction behavior by improving the rate of heat removal from the microreactors. As outlined in Appendix B-1, the temperature versus heater power response curves suggested that the final membrane design should employ a 2.6 µm silicon-based membrane instead of the original 1 µm silicon nitride membrane. 2.d.2. Flow Anemometers. The flow anemometers, which are located in the side legs of the microreactor, operate by detecting changes in the temperature field that occur in close proximity to a heater element. The temperature changes are induced by a local heat flux from the heater to the surrounding gas. The results of the design study outlined in Appendix B-2 led to the following conclusions: (1) To obtain the greatest temperature sensitivity, the sensor should be placed slightly upstream of the heater. (2) The optimal heater length is determined by sensor sensitivity, thermal response time, and mechanical stresses that are induced on the reactor membrane. Irreversible damage to the membrane can be prevented by using a heater length of ca. 200 µm and then adjusting the heater power so the maximum temperature in the heater segment under a stagnant flow condition is maintained constant at 200 °C. (3) The optimum temperature sensor location is insensitive to changes in the membrane thermal conductivity. However, changing the thermal conductivity of the membrane also affects the shape of the temperature responses because decreasing the resistance to heat transfer in the solid phase decreases the flow sensor’s performance. This same type of temperature response occurs when the membrane thickness is altered, since it has the same effect on the heat transfer behavior, i.e., doubling the membrane thermal conductivity is equivalent to doubling the membrane thickness. (4) Gases having larger values for the thermal diffusivity, such as H2, exhibit the lowest sensor sensitivity, while gases having smaller thermal diffusivities, such as NH3, exhibit the greatest sensor sensitivity. Figures B-1-B-4 in the Supporting Information illustrate the above points to which the reader is referred for details. 2.d.3. Reactor Heaters. The original T-microreactor heater design (see Figure B-5a in the Supporting Information) consisted of a 50 µm Pt line that meandered across the 300 µm heater width down the length of the channel. This design is effective in providing uniform power in the axial direction of the reactor heated zone, but it results in a nonuniform temperature profile across the reactor in the coordinate that is perpendicular to the flow direction. The temperature profile is quasi-parabolic, with the maximum in the upper center zone of the channel. To generate a more uniform or flat temperature profile across the channel, the heater was split into two straight Pt metal segments that run along the length of the channel but are equally spaced across the channel (see Figure B-5b in the Supporting Information). The simulation tools showed that a more uniform temperature profile was obtained across the channel in the absence of heat generation because of reaction up to the

8296

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007

Figure 2. Schematic of the Y scale-up microreactor: (A) top view; (B) cross section perpendicular to flow; and (C) cross section parallel to flow.

Figure 3. Layout of the Y-microreactor heater segment with temperature sensors (electrical leads from sensors and heaters are not shown). The temperature sensor at the left is the upstream sensor. Figure 4. Y-microreactor designed for the scale-up system.

maximum flow rate of 100 sccm. When a reaction was included in the simulation (ammonia oxidation) and it was operated in the ignition regime, the thermal uniformity across the channel was improved since the internal heat generation offsets part of the heat loss through the top of the reactor membrane. The temperature profiles for the old and new heaters are compared in Figures B-6-B-8 in the Supporting Information. Although some nonuniformity still exists with the new design, this could be reduced by the addition of more parallel heaters down the length of the channel and additional optimization of heater parameters, such as heater line width and thickness, number of heaters, and heater spacing. 3. First-Generation Scale-up Microreactor 3.a. Scale-up Microreactor Description. The flow sensor and heater designs discussed in the preceding sections were first incorporated into a Y-shaped microreactor for scale-up, as shown in Figure 2. This microreactor consists of two inlet flow channels and a reaction channel. The inlet flow channels form the branches of the Y and one of the new gas flow sensors is placed on each leg of the Y. The reaction channel contains two heater segments that utilize the split heater design as shown in Figure 3. Two temperature sensors, one upstream and one downstream, are present on each heater segment. For this design, the heaters have two independently operating sides, but the intent was to place these heaters in series and power them with a single circuit. 3.b. Scale-Up Microreactor Design. The simulation tools described above were used to examine the power requirements of the Pt heater lines in the device. The main concern was that the heater segments might require more power than could be

provided by the Pt lines because of electromigration. After performing this study, the length of the heater segment was determined to be 2.8 mm with a 50 µm line width. The dimensions of temperature sensors inside the segment were 0.8 mm long with a 10 µm line width. The heater and temperature sensors span 0.3 mm of the total 0.5 mm channel width. Figure 4 shows a picture of one of the fabricated Y-microreactors. To provide insight into the size of the device, a U.S. penny is also shown for comparison. The microscale size of the reactor is evident. 3.c. Scale-Up Microreactor Fabrication. The primary motivation behind the Y shape and the use of rounded corners was to reduce the stress concentrations in the silicon nitride membrane at the intersection of the feed channels. It was also surmised that this design would provide a membrane that was more robust and be able to withstand higher reaction temperatures for longer periods of time. Because of the shape of the reactor, a KOH etch could not be used to form the channels in the silicon wafer. Instead, a deep reactive ion etch (DRIE) process was used as an alternative fabrication technique. Although this microreactor was successfully fabricated, the DRIE processing step caused considerable difficulty since the boundary between the bulk silicon and the silicon nitride film was often undercut by the etch. This phenomenon is known as footing or notching and is caused by a differential charging of sidewalls and bottom features that occurs when etching a silicon layer over top of a dielectric film.71 The difficulties of DRIE and methods for improving etch performance have since been described in more detail in work conducted at the Microsystems

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8297

Figure 5. Final scale-up microreactor design: (A) top view; (B) cross-sectional view AA; (C) cross-sectional view BB; and (D) enlargement of heater and temperature sensor configuration.

Technology Laboratory (MTL) at Massachusetts Institute of Technology (MIT).72-75 This problem of footing prevented the realization of a robust silicon nitride membrane for the microreactor. The number of functioning reactors per processed wafer was not particularly good. Furthermore, the DRIE processing step was slow in that only one wafer could be processed at a time and etching required more than 4 h. The tool was also in high demand for other projects, so this created a large bottleneck in microreactor wafer processing. 4. Second-Generation Scale-Up Microreactor 4.a. Scale-Up Microreactor Description. Because of the above experience, the scale-up microreactor was revised by removing the feed gas-mixing region from the die. Instead, premixed feed gas would be prepared externally and then directed to the microreactor. By premixing the feed gas, it was no longer necessary to utilize two intersecting channels, so a single channel would allow premixed feed gas to be introduced and directed to the catalytic zone of the microreactor. This approach would also eliminate the stress concentrations created at the center of the Y structure in the membrane, and the channel etch could again be done with KOH. The second reason was to improve reaction data analysis by direct on-line analysis of both the microreactor feed and product gases. This would allow reactant conversions and product selectivities to be readily evaluated using the known values for gas flow rate and on-line gas analysis data. A description of the second-generation microreactor is provided in this section. 4.b. Scale-Up Microreactor Design. Figure 5 shows a schematic of the second-generation microreactor design that was used as part of the integrated system. The reaction channel is 0.5 mm wide, 0.5 mm deep, and 11 mm long. It has seven distinct zones for heating and temperature sensing that cover a total length of 7 mm. The gas flow sensors present on the feed inlet channels of the Y-microreactor were removed, and only microheaters and microtemperature-sensors were incorporated in this design. The heater and temperature sensor designs were modified, as shown in Figure 5D, to provide higher spatial resolution for controlling the reaction-zone temperature profile. More heater zones were not added because of space limitations for electrical leads on the die. Each heater has an overall length of 0.94 mm, while the temperature sensors have a length of

Figure 6. Scale-up microreactor die showing the reaction channel locations. All indicated drawing units are in millimeters.

0.77 mm. The width of the heater/temperature sensor segments were 0.3 mm, which left a gap of 0.1 mm between the heater and silicon channel walls. Two reaction channels were incorporated on each microreactor die due to a restriction on the die size placed by the packaging method chosen, which is discussed in Part 2 of this series. The required die size, 20.3 mm by 28.2 mm, was large enough that two channels could be readily positioned within the die confines. Figure 6 shows the scale-up microreactor die layout. This particular die has 56 electrical interconnections (not shown in Figure 6) that must be configured so that 14 heaters and 14 temperature sensors can be interfaced to their appropriate devices. 4.c. Scale-Up Microreactor Fabrication. The fabrication process for this microreactor was similar to the process developed for the T-microreactor, but some modifications were introduced. Figure 7 illustrates the main fabrication steps by showing a cross section of the wafer after each step. The majority of the wafer processing took place at the DuPont microfabrication facility at the Experimental Station Laboratory (ESL).76 Low stress silicon nitride depositions were performed either by the MTL at MIT or by ACT MicroDevices.77 4.c.1. Wafer Source and Initial Film. The starting point for the microfabrication process was double-side polished (DSP)

8298

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007

Figure 7. Fabrication sequence for the scale-up microreactor: (A) fresh DSP silicon wafer, 101.6 mm in diameter and 508 µm thick; (B) deposit 1 µm of low-stress silicon nitride on top of 0.1 µm of dry thermal oxide; (C) using a photoresist mask, plasma etch through silicon nitride on backside of wafer to define the channel geometry; (D) using the silicon nitride as a mask, perform a timed KOH etch of the silicon to leave about 25 µm of silicon below the silicon nitride membrane; (E) perform PVD of 0.1 µm Pt metal layer on the front side of the wafer using a lift-off process to define the heater and temperature sensor microstructures; (F) complete the KOH etch of the silicon to the silicon nitride membrane; (G) using a shadow mask, deposit 0.1 µm Pt metal layer on the backside of the silicon nitride membrane to act as a catalyst and then anneal the Pt layer at 400 °C for 1 h; and (H) anodically bond the backside of the wafer to a Pyrex 7740 wafer.

single crystal {100}-oriented silicon wafers obtained from Riotech.78 The wafers were 4 in. in diameter and 0.020 in. in thickness (Figure 7A). A dry silicon oxide film was first grown on the wafers to aid in the anodic bonding step done later in the process. The MIT batch of wafers had 0.1 µm of dry thermal oxide grown. The ACT batch of wafers had the same thickness of silicon dioxide, but it was deposited using low-pressure chemical vapor deposition (LPCVD). A low-stress LPCVD silicon-rich silicon nitride film was then deposited on the silicon oxide layer. The MIT silicon nitride film was 1 µm thick. The ACT silicon nitride film was 0.75 µm thick. Figure 7B shows a cross section of the wafer after this step. 4.c.2. First Photolithography Step. A photolithography step was then performed to define the channels in the silicon wafer. Hexamethyldisilazane (HMDS) from Arch Chemicals was first applied to the wafer surface to aid in photoresist adhesion.79 A spin-coater was used to deposit HiPR 6517 positive photoresist on the wafers. The photoresist was also manufactured by Arch Chemicals. The photoresist was then exposed with a dark-field mask, which is shown in Figure 8, to define the reaction channel geometry. The dark-field mask was manufactured by Photronics.80 After photoresist development and postbaking, the silicon nitride film on the backside of the wafer was plasma etched with the photoresist as the etch mask. Figure 7C shows a cross section of the wafer after this step. 4.c.3. First KOH Etching. The reactor channels were then partially etched using a 30% KOH solution at 70 °C. This anisotropic etch was timed to leave about 25 µm of silicon at the bottom of the channel. Figure 7D shows a cross section of the wafer after this step. The silicon nitride layer was material used for masking in this etch since it is not attacked by KOH. It should be noted that KOH etches preferentially the 〈100〉 direction of silicon and very slowly in the 〈111〉 direction at a ratio of as much as 400-to-1. This leaves a V-shaped groove in the silicon wafer instead of a channel with sidewalls perpendicular to the wafer surface, as illustrated in Figure 9. 4.c.4. Second Photolithography Step. The second photolithography step was then performed to define the Pt metal structures on the microreactor dies. A lift-off process using AZ 5214-E IR negative photoresist from Clariant81 was used to define the Pt metal geometries. A light-field mask, shown in Figure 10, was used to pattern the photoresist. This mask was also manufactured by Photronics.80 After development of the photoresist, a 0.1 µm layer of Pt was deposited onto the wafer surface using an electron-beam PVD process. Thicker layers

are difficult to deposit and remove successfully using a lift-off process. Microstrip from RBP Chemical Corporation82 was used to remove the remaining photoresist along with the Pt layer on top of it to complete this step. Figure 7E shows a cross section of the wafer after this step. Originally, it was hoped that the silicon nitride layer on the bottom of the wafer could be removed using an unmasked plasma etch. The intent here was to provide a better bonding surface for the anodic bonding step at the end of the process. Silicon nitride does not bond well with Pyrex, whereas silicon oxide gives a strong bond.83,84 Unfortunately, it was found that the plasma etch to remove the silicon nitride layer resulted in a poor bonding surface because the silicon nitride layer did not etch uniformly. After several attempts were made at bonding the wafers with the silicon nitride layer removed, it was decided that it would be better to bond directly to the silicon nitride. 4.c.5. Second KOH Etching. The 25 µm of remaining silicon supporting the silicon nitride membrane was then removed using a 10% KOH solution at 80 °C. Figure 7F shows a cross section of the wafer after this step. A different KOH solution formula was used because this offered higher selectivity between etching silicon and silicon oxide. The previous formula has a faster etch rate of silicon. The metallization on the front side of the wafer was protected using a Teflon wafer holder, shown in Figure 11. The holder formed a seal against the front side of the wafer, preventing the aqueous KOH from contacting the Pt metal deposited in the previous step. This holder was previously fabricated in-house at DuPont. 4.c.6. Catalyst Deposition and Annealing. The Pt catalyst was then deposited on the channel side of the silicon nitride membrane using electron-beam physical vapor deposition (PVD). A shadow mask limited the deposition of Pt to only the channel area of the reactors. This mask, shown in Figure 12, was laser machined from stainless steel and was attached to the wafers prior to placement in the e-beam equipment using Scotch tape. A shadow mask was used instead of photolithography because the presence of the loose membranes made the microreactors too fragile for photolithography. The Pt metallization on both sides of the wafer was annealed at 400 °C for 1 h under nitrogen. Figure 7G shows a cross section of the wafer after this step. 4.c.7. Anodic Bonding. A Pyrex 7740 wafer was then anodically bonded to the backside of the microreactor wafer, as illustrated in Figure 7H. This bonding step was performed at 460 °C for 5 h with the applied voltage being stepped up to

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8299

Figure 8. Dark-field mask layout used to define reactor channels in the silicon nitride film. Eight scale-up microreactor dies are produced on a silicon wafer. Four single-channel microreactor dies are also produced for other testing.

Figure 9. Detail of the cross section of the reaction channel etched in the silicon wafer.

900 Vdc. Typically, anodic bonding of Pyrex is performed between 180 and 500 °C with the applied voltage ranging from 200 to 1000 V.85 These aggressive bonding conditions were necessary because the Pyrex wafer was bonded to a silicon nitride surface instead of silicon or silicon oxide, with the latter being the preferred surface. The Pyrex wafer contained fluidic inlet and outlet holes drilled in it before the bonding step. A coarse alignment between the silicon wafer and the Pyrex wafer was acceptable since the fluidic ports were relatively large features. 4.c.8. Wafer Dicing. The final step was dicing the wafer using a die saw. The saw blade used was 0.008 in. thick, 400 grit, with a 37 µm particle size. Initially, the diesaw blade was 0.006 in. thick, but this was found to be too thin. Figure 13 shows a picture of a completed microreactor. The dual parallel channels are visible on either side of the chip. 4.d. Scale-up Reactor Temperature Sensing. The method used for reactor temperature sensing was based on Pt resistance

thermometry. As explained in the fabrication sequence (Section 4c), the Pt was deposited as thin film using a PVD process. Because this was a prototype process, some variation in the Pt film thicknesses and other film dimensions could conceivably occur that might affect its local electrical resistivity. For this reason, an experimental methodology was developed so that the Pt resistivity could be measured as a function of temperature. The resulting measurements were then correlated so an algebraic expression could be subsequently used in the process automation for determination of the local microreactor temperatures. This section describes the special-purpose experimental methodology that was developed to collect the data and key results from the data analysis. 4.d.1. Pt Resistivity Determination. The dependence of Pt resistivity on temperature (the temperature coefficient of resistivity or R) is used as the basis for the resistive temperature devices (RTDs) in the microreactor. The equation used for determining temperature as a function of the Pt structure resistance is

R ) 1 + R(T - To) Ro

(1)

In eq 1, R is the measured resistance, Ro is the resistance at a reference temperature, T is the temperature of the RTD, and To is the reference temperature. The reference temperature used in this work was the triple point of water (0.01 °C). The above relationship is highly dependent on the deposition process used

8300

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007

Figure 10. Light-field mask used to define the Pt metal layer on the microreactor wafers. The metal lines between microreactor dies are used as guides for dicing the reactor chips.

Figure 11. Teflon holder used to protect the front side of the wafers during the second KOH etch. A Viton O-ring was used for sealing to the wafer.

for the Pt layer. Previous experience showed that Pt deposited under the same process conditions using the same fabrication device, but in successive runs, could exhibit different values for R. Examination of eq 1 shows that the value of R will depend on the reference temperature used. Many literature reports that utilized eq 1 for temperature measurements are not useful, since the reference temperature that was used to calculate the parameter R was not listed. For this reason, it is difficult to compare R values reported in the literature. Furthermore, errors could be easily introduced in the temperature calculation if literature values for R were used without the appropriate matching reference temperature. This would ultimately affect

the quality of the process temperature control. It should also be noted that eq 1 represents a simplification of more accurate equations used for standard Pt resistance thermometers (SPRTs) as specified in the ITS-90 standard.86 However, the accuracy produced by eq 1 is adequate for the microreactor RTDs, since previous experience has shown it provides a good approximation to the experimental temperature versus electrical resistivity data. 4.d.2. Experimental Methods. Determination of the temperature coefficient of resistivity was performed by increasing the temperature of the entire microreactor die and recording the measured resistance. The temperature was varied from 22 °C (room temperature) to approximately 275 °C. Higher temperatures were not used because of limitations of the testing

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8301

Figure 12. Shadow mask used in the deposition of the Pt catalyst.

Figure 13. Microreactor used in the scale-up microreactor system.

apparatus. Although it would have been ideal to use higher temperatures, the value of R should not change provided that the intrinsic structure of the Pt is stable. The upper temperature limit depends on the annealing conditions of the Pt. The resistance data from these experiments were used to determine the R value through linear regression of the data for each group of wafers, where a group is specified as wafers that underwent the Pt deposition at the same time. The equipment used in this experiment was taken from the anodic bonding apparatus used at DuPont in their microfabrication facility.76 Figure 14 is a schematic of the experimental setup. Testing was performed on whole wafers before they had been cut into chips on the die, which facilitated handling the devices. The wafer was placed on a heated cylindrical aluminum block having a diameter of 4 in. The block temperature was monitored using both a 1/16 in. thermocouple and a 0.020 in. thermocouple. Both thermocouples were also inserted into holes drilled into the block. A 1/16 in. type K thermocouple was used for temperature control with a series CN-2010 Omega temperature controller.87 The temperature was also monitored with a

0.020 in. type K thermocouple with an Omega model HH21 microprocessor thermometer. To assess the temperature accuracy of these two devices, the thermocouple and temperature monitor were checked and calibrated against a NIST-calibrated SPRT model 5626 from Hart Scientific.88 The aluminum cylindrical block was heated from below by another cylindrical aluminum block that had four 1/4 in. heating cartridges inserted inside it (not shown in Figure 14). This block also had a diameter of 4 in. The heated blocks and wafer were surrounded by insulating ceramic material to help ensure temperature uniformity. The resistances for both the microreactor heater and temperature sensor structures were measured using a custommanufactured probe card. This device is also shown at the top of Figure 14. Electrical leads from a Tektronix DMM916 multimeter89 were connected to the probe card so the resistances of three structures on the microreactor could be measured at each temperature. Temperatures higher than 275 °C were not used since discoloration of the resin holding the probe card pins was observed, which suggested that the upper temperature limit had been reached. Structure resistances were recorded starting

8302

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007

Figure 14. Schematic of the test setup used for the Pt resistance characterization experiment.

at room temperature (∼22 °C) up to ∼275 °C in 25 °C increments. Resistances of these same structures were recorded again as the wafer was cooled back to room temperature in 50 °C increments to check for hysterisis and potential intrinsic structural changes to the Pt film. 5. Results and Discussion This section focuses on observations that were made during development of the first-generation and second-generation scaled-up microreactors as well as the data collected from the Pt resistance tests. The key findings from the simulation and modeling efforts during the microreactor design stage were discussed earlier in Section 2 for ease of reference. 5.a. Scale-up Microreactor Development. Fabrication of the first-generation scale-up microreactor met with considerable difficulty because of the DRIE step. Most of the fabrication attempts were performed at MIT using an etch-to-membrane process, which was a much simpler procedure than the alternative bonding-based process. This approach resulted in low microfabricated reactor yields. The fabrication difficulties encountered with the DRIE step were the main reasons why the microreactor design was altered to a simpler design. The second-generation reactor eliminated the two feed gas legs and the associated flow sensors. On-board feed gas mixing was also eliminated. Both of these process functions had been demonstrated in off-line prototyping work, so it was surmised that they would have functioned properly if the first-generation Y-microreactor had been successfully fabricated. Simplification of the reactor design allowed most of the wafer processing to be performed at DuPont. In this case, prototyping proceeded at a faster pace since the fabrication facility could be dedicated to producing the required microreactor wafers. Two main batches of wafers were fabricated, one with a MITdeposited silicon nitride film and one with an ACT-deposited film. The MIT batch was processed first, but poor membrane

stability resulted in a very low yield of intact channels per wafer. The survival rate was only around 30%. The reason for the poor survival rate was not known, but the membranes that did survive processing satisfied the minimum tensile strength requirements. For this reason, a batch of wafers was sent to ACT for silicon nitride deposition. Originally, the silicon nitride film on the backside of the wafers was removed with a plasma etch, but the nonuniformity in the thickness of the wafer left a very poor surface for bonding. This again resulted in poor reactor yield. Bonding directly to the silicon nitride on the backside circumvented this problem. Although anodic bonding to silicon nitride does not result in the strongest bond, the process was satisfactory for this application. With this change, the fabrication yield increased to about 70% microreactor survival per wafer. Collectively, these fabrication experiences suggest that any prototyping of microreactors requiring multiple MEMS-like fabrication steps (see Figure 7) will likely require significant time and effort, especially when new designs are under development. Special MEMS fabrication hardware and expertise are also required if any chance of success is to be expected. However, once a particular fabrication sequence has been demonstrated and experience is gained with producing a microreactor die, both the cost and time required to generate multiple samples of the same design should be reduced to lower levels. This reduction in prototyping effort is parallel to what might be expected when more conventional microreactors are fabricated from glass22 or from metal or plastics using multipleaxis computer numerical control (CNC)-type milling machines.23 5.b. Scale-up Microreactor Temperature Sensing. The results of the testing to determine the temperature coefficient of resistivity for the Pt thin films are given below. Only the testing results of the ACT wafers are described because these reactors were ultimately the ones used in the AIMS testing. The temperature coefficients of resistivity for the Pt used as the microreactor temperature sensors ranged from 0.00205 to

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8303 Table 1. Temperature Coefficient of Resistivity for the Pt on the ACT Wafers ACT wafer group

R

error

1 2 3

0.002 16 0.002 05 0.002 26

0.000 04 0.000 08 0.000 06

0.00226 where the reference temperature used was the triple point of water (0.01 °C). Table 1 lists the values for R for each of the ACT microreactor groups along with their error value. The R value listed is the one obtained by linear regression of the resistance data and the corrected temperature from the Omega microprocessor thermometer used in the experiment. This value must be corrected since this monitor was calibrated against a NIST SPRT, and this calibration was used to determine the actual measured temperature. Although the temperature monitor was close in calibration to the NIST second source standard, this did make a slight difference in the calculated R values. The error values for R were determined by also calculating its value using the temperatures measured by the Omega temperature controller. It was postulated that the wafer temperature measurement had the largest associated error and that the two temperature values recorded reflected this error. The R values determined are relatively close with respect to the measurement error. This is expected since all of the wafers involved had very similar processing conditions. 6. Conclusions The first-generation Y-microreactor design was not acceptable for the scale-up project because too much emphasis was placed on microfabrication techniques. The focus of this work required that more time and effort be placed on designing the system versus running the microreactors. However, a new microreactor design was necessary to accommodate the packaging methodology, which is discussed in Part 2 of this series. The slot scaleup microreactor design allowed easier fabrication and eventually provided a good yield with the ACT silicon nitride wafers. The temperature coefficient of resistivity of the Pt thin films deposited on the microreactors was found to be slightly different for the three Pt deposition groups (0.00216, 0.00205, and 0.00226). This variation is expected and is partly attributed to the error in the determined R values. The reference temperature used for the RTD equation was 0.01 °C, which is the triple point of water. Characterization of the microreactor heaters and temperature sensors will be discussed in subsequent parts along with the testing of the microreactor packaging. Testing under nonreacting flow conditions will be discussed in Part 3 with some discussion in Part 4. The results of reaction testing will be discussed with the results of the scale-up system testing in Part 4. Supporting Information Available: Text and figures describing the microreactor design simulations and the use of the simulation model as a design tool. This material is available free of charge via the Internet at http://pubs.acs.org. Literature Cited (1) Younes-Metzler, O.; Svagin, J.; Jensen, S.; Christensen, C. H.; Hansen, O.; Quaade, U. Microfabricated High-Temperature Reactor for Catalytic Partial Oxidation of Methane. Appl. Catal., A 2005, 284 (1-2), 5-10.

(2) Huybrechts, W.; Mijoin, J.; Jacobs, P. A.; Martens, J. A. Development of a Fixed-Bed Continuous-Flow High-Throughput Reactor for LongChain n-Alkane Hydroconversion. Appl. Catal., A 2003, 243 (1), 1-13. (3) Chen, B.; Dingerdissen, U.; Krauter, J. G. E.; Lansink Rotgerink, H. G. J.; Mo¨bus, K.; Ostgard, D.; Panster, J.; Riermeier, T. H.; Seebald, S.; Tacke, T.; Trauthwein, H. New Developments in Hydrogenation Catalysis Particularly in Synthesis of Fine and Intermediate Chemicals. Appl. Catal., A 2005, 280 (1), 17-46. (4) Watts, P. Chemical Synthesis in Micro Reactors. Chem.-Ing.-Tech. 2004, 76 (5), 555-559. (5) Hessel, V.; Lo¨we, H. Microchemical Engineering: Components, Plant Concepts, User Acceptance. Part I. Chem. Eng. Technol. 2003, 26 (1), 13-24. (6) Hessel, V.; Lo¨we, H. Microchemical Engineering: Components, Plant Concepts, User Acceptance. Part II. Chem. Eng. Technol. 2003, 26 (4), 391-408. (7) Hessel, V.; Lo¨we, H. Microchemical Engineering: Components, Plant Concepts, User Acceptance. Part III. Chem. Eng. Technol. 2003, 26 (5), 531-544. (8) Hessel, V.; Lo¨we, H. Organic Synthesis with Microstructured Reactors. Chem. Eng. Technol. 2005, 28 (3), 267-284. (9) Pfeifer, P.; Bohn, L.; Go¨rke, O.; Haas-Santo, K.; Schubert, K. Microstructured Components for Hydrogen Production from Various Hydrocarbons. Chem. Eng. Technol. 2005, 28 (4), 474-476. (10) Delsman, E. R.; De Croon, M. H. J. M.; Pierik, A.; Kramer, G. J.; Cobden, P. D.; Hofmann, C.; Cominos, V.; Schouten, J. C.; Design and Operation of a Preferential Oxidation Microdevice for a Portable Fuel Processor. Chem. Eng. Sci. 2004, 59 (22/23), 4795-4802. (11) Shah, K.; Ouyang, X.; Besser, R. S. Microreaction for Microfuel Processing: Challenges and Prospects. Chem. Eng. Technol. 2005, 28 (3), 303-313. (12) Styring, P. Towards Liquid Crystal Synthesis Using High Throughput and Micro Reactor Technologies. Mol. Cryst. Liq. Cryst. 2004, 411, 1059-1070. (13) Doig, S. D.; Pickering, S. C. R.; Lye, G. J.; Woodley, J. M. The Use of Microscale Processing Technologies for Quantification of Biocatalytic Baeyer-Villiger Oxidation Kinetics. Biotechnol. Bioeng. 2002, 80 (1), 42-49. (14) Puskeiler, R.; Kaufmann, K.; Weuster-Botz, D. Development, Parallelization, and Automation of a Gas-Inducing Milliliter-Scale Bioreactor for High-Throughput Bioprocess Design (HTBD). Biotechnol. Bioeng. 2005, 89 (5), 512-523. (15) Shchukin, D. G.; Sukhorukov, G. B. Nanoparticle Synthesis in Engineered Organic Nanoscale Reactors. AdV. Mater. 2004, 16 (8), 671682. (16) Palmer, C.; Remcho, V. T. Microscale Liquid-Phase Separations. Anal. Bioanal. Chem. 2002, 372 (1), 35-36. (17) Nanostream, Inc. Micro Parallel LC with Time-Triggered, Parallel Fraction Collection Accelerates Sample Preparation for MS Assays in Drug Discovery and Development. LC-GC (North America) 2005, 23 (June Supplement), 63. (18) Wiese, M., Schneider, G., Kappe, C. O., Eds. QSAR Comb. Sci. 2005, 24 (6), 695-803. (Special Issue on Combinatorial Fluidics.) (19) Tovbin, Y. K. Gas and Liquid Transfer in Narrow Pores. Theor. Found. Chem. Eng. 2002, 36 (3), 214-221. (20) Zhao, Y.-P.; Wang, L. S.; Yu, T. X. Mechanics of Adhesion in MEMSsA Review. J. Adhes. Sci. Technol. 2003, 17 (4), 519-546. (21) Satake, S.; Kunugi, T.; Sato, K.; Ito, T.; Taniguchi, J. ThreeDimensional Flow Tracking in a Micro Channel with High Time Resolution Using Micro Digital-Holographic Particle Tracking Velocimetry. Opt. ReV. 2005, 12 (6), 442-444. (22) Dietrich, T. R.; Freitag, A.; Scholz, R. Production and Characteristics of Microreactors Made from Glass. Chem. Eng. Technol. 2005, 28 (4), 477-483. (23) Bong-Bang, Y.; Lee, K-M.; Oh, S. 5-Axis Micro Milling Machine for Machining Micro Parts. Int. J. AdV. Manuf. Technol. 2005, 25, 888894. (24) Truckenmuller, R.; Cheng, Y.; Ahrens, R.; Bahrs, H.; Fischer, G.; Lehmann, J. Micro Ultrasonic Welding: Joining of Chemically Inert Polymer Microparts for Single Material Fluidic Components and Systems. Microsyst. Technol. 2006, 12, 1027-1029. (25) Alm, B.; Knitter, R.; Hausselt, J. Development of a Ceramic Micro Heat ExchangersDesign, Construction, and Testing. Chem. Eng. Technol. 2005, 28 (12), 1554-1560. (26) Chambers, R. D.; Fox, M. A.; Holling, D.; Nakano, T. T.; Okazoe, G. Versatile Gas/Liquid Microreactors for Industry. Chem. Eng. Technol. 2005, 28 (3), 344-352.

8304

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007

(27) Commenge, J.-M.; Falk, L.; Corriou, J.-P.; Matlosz, M. Analysis of Microstructured Reactor Characteristics for Process Miniaturization and Intensification. Chem. Eng. Technol. 2005, 28 (4), 446-458. (28) Benz, K.; Ja¨ckel, K.-P.; Regenauer, K.-J.; Schiewe, J.; Drese, K.; Ehrfeld, W.; Hessel, V.; Lo¨we, H. Utilization of Micromixers for Extraction Processes. Chem. Eng. Technol. 2001, 24 (1), 11-17. (29) Majors, R. E. Miniaturized Approaches to Conventional LiquidLiquid Extraction. LC/GC (North America) 2006, 24 (2), 118-130. (30) Sato, M.; Goto, M. Note: Gas Absorption in Water with Microchannel Devices. Sep. Sci. Technol. 2004, 39, 3163-3167. (31) Ehrfeld, W.; Hessel, V.; Lo¨we, H. Microreactors: New Technology for Modern Chemistry, 1st ed.; Wiley-VCH Publishers: Berlin, 2000. (32) Hessel, V.; Hardt, S.; Lo¨we, H.; Mu¨ller, A.; Kolb, G. Chemical Micro Process Engineering; John Wiley & Sons Publishers: New York, 2005. (33) Ehrfeld, W., Ed. Microreaction Technology: Proceedings of the 1st International Conference on Microreaction Technology; Springer-Verlag Publishers: Berlin, 1998. (34) Ehrfeld, W., Rinard, H., Wegeng, R. S., Eds. Process Miniaturization: 2nd International Conference on Microreaction Technology, Topical Conference Preprints; AIChE Press: New York, 1999. (35) Ehrfeld, W. Microreaction Technology: 3rd International Conference on Microreaction Technology; Springer-Verlag Publishers: Berlin, 2000. (36) Ehrfeld, W., Eul, U., Wegeng, R. S., Eds. 4th International Conference on Microreaction Technology, Topical Conference Proceedings; AIChE Press: New York, 2000. (37) Matlosz, M., Ehrfeld, W., Baselt, J. P., Eds. Microreaction Technology: IMRET 5: Proceedings of the Fifth International Conference on Microreaction Technology; Springer Publishers: Berlin, 2002. (38) Ehrfeld, W., Eul, U., Wegeng, R. S., Eds. 6th International Conference on Microreaction Technology, Topical Conference Proceedings; AIChE Press: New York, 2002. (39) Renken, A., Baselt, J. P., Matlosz, M., Eds. 7th International Conference on Microreaction Technology, Special Issue. Chem. Eng. J. 2004, 101 (1/3), 1-484. (40) Ehrfeld, W., Ed. 8th International Conference on Microreaction Technology, Topical Conference Proceedings; AIChE Press: New York, 2005. (41) Schu¨tte, R.; Matlosz, M.; Renken, A.; Liauw, M.; Langer, O.-U. (organizers) The Ninth International Conference on Microreaction Technology (IMRET 9), Potsdam/Berlin, Germany, Sept 6-9, 2006. (42) Behr, A.; Ebbers, W.; Wiese, N. MiniplantssEin Beitrag zur inha¨renten Sicherheit? Chem.-Ing.-Tech. 2000, 72 (10), 1157-1166. (43) Chow, A. W. Lab-on-a-Chip: Opportunities for Chemical Engineering. AIChE J. 2002, 48 (8), 1590-1595. (44) Kirschneck, D.; Kober, M.; Marr, R. New Scopes in Process Design Using Microstructured Devices. Chem. Eng. Technol. 2005, 28 (3), 314317. (45) Kockmann, N., Brand, O., Fedder, G. K., Hierold, C., Korvink, J. G., Tabata, O., Eds. Micro Process Engineering: Fundamentals, DeVices, Fabrication, and Applications AdVanced Micro and Nanosystems; John Wiley & Sons Publishers: New York, 2006. (46) Stankiewicz, A., Ed. Re-engineering the Chemical Processing Plant; Marcel-Dekker Publishers: New York, 2003. (47) Ehrfeld, W.; Hessel, V.; Mo¨bius, H.; Richter, Th.; Russow, K. Potentials and Realization of Microreactors. Presented at The Workshop on Microsystem Technology, Mainz, Germany, Feb 20-21, 1995. In Microsystem Technology for Chemical and Biological Microreactors; DECHEMA Monograph Series 132; Verlag Chemie Publishers: Weinheim, Germany, 1996; pp 1-28. (48) Ehrfeld, W.; Hessel, V.; Lehr, H. Microreactors for Chemical Synthesis and BiotechnologysCurrent Developments and Future Applications. Presented at The Workshop on Microsystem Technology, Mainz, Germany, Feb 20-21, 1995. In Microsystem Technology for Chemical and Biological Microreactors; DECHEMA Monograph Series 132; Verlag Chemie Publishers: Weinheim, Germany, 1996; pp 233-252. (49) Hasebe, S. (Chairman). The 5th International Workshop on Micro Chemical Plants, Kyoto University, Nishikyo-ku, Kyoto, Japan, January 29-30, 2007. (Meeting program available at http://www.mcpt.jp/english/ 5THMCPT_Program_index_070110.pdf). (50) Charpentier, J.-C. Process Intensification by Miniaturization. Chem. Eng. Technol. 2005, 28 (3), 255-258. (51) Jensen, K. J. Microchemical Systems: Status, Challenges, and Opportunities. AIChE J. 1999, 45 (10), 2051-2054. (52) Vanden Bussche, K. M. The Direct Synthesis of Hydrogen PeroxidesFundamentals and Conceptual Designs. Presented at the Symposium on Modeling of Complex Processes, Texas A&M University, College Station, TX, Mar 1, 2005.

(53) Li, L.; Vanden Bussche, K. M. Micro Technology for Large Scale Application: Direct Synthesis of Hydrogen Peroxide. Presented at the 2006 AIChE Spring Meeting, Orlando, FL, Apr 24, 2006; Paper 71N. (54) Ryley, J. F. Microchemical System ApplicationssDuPont Experience. Presented at the Army Research Office/MIT Workshop on Microchemical Systems and Their Applications, Reston, VA, June 16-18, 1999. (55) Anonymous. DuPont researchers receive award for improving chemical safety, http://www.scienceblog.com/community/older/2003/C/ 2003411.html (56) Ashmead, J. W.; Blaisdell, C. T.; Johnson, M. H.; Nyquist, J. K.; Perrotto, J. A.; Ryley, J. F., Jr. (inventors) Integrated Chemical Processing Apparatus and Processes for the Preparation Thereof. U.S. Patent 5,534,328 (assigned to E. I. du Pont de Nemours), July 9, 1996. (57) Ashmead, J. W.; Blaisdell, C. T.; Johnson, M. H.; Nyquist, J. K.; Perrotto, J. A.; Ryley, J. F., Jr. (inventors) Integrated Chemical Processing Apparatus and Processes for the Preparation Thereof. U.S. Patent 5,690,763 (assigned to E. I. du Pont de Nemours), Nov 25, 1997. (58) Lohf, A.; Ehrfeld, W.; Hessel, V.; Lo¨we, H. A Standardized Modular Microreactor System. Presented at the 4th International Conference on Microreaction Technology, Atlanta, GA, 2000. In Topical Conference Proceedings; AIChE Press: New York, 2000; pp 441-451. (59) Lee, A. Microfluidic Molecular Systems Program at DARPA. Presentation given at the NIST Labs, Microanalytical Lab/MEMS Seminar Series, Gaithersburg, MD, Oct 28, 1999. (60) Mills, P. L.; Nicole, J. F. Multiple Automated Reactor System (MARS). 1. A Novel Reactor System for Detailed Testing of Gas-Phase Heterogeneous Oxidation Catalysts. Ind. Eng. Chem. Res. 2005, 44, 64356452. (61) CoventorWare is the new name for the software that replaces MEMCAD and FlumeCAD. CoventorWare is a product of Coventor, Inc., 4001 Weston Parkway, Cary, NC 27513, (919) 854-7500. (62) CFD-ACE+ is a product of CFD Research Corporation, 215 Wynn Dr., Huntsville, AL 35805, (256) 726-4800. Fluent is a product of Fluent, Inc., 10 Cavendish Ct., Centerra Park Lebanon, NH 03766, (603) 643-2600. (63) Deshmukh, S. R.; Vlachos, D. G. Novel Micromixers Driven by Flow Instabilities: Application to Post-Reactors. AIChE J. 2005, 51 (12), 3193-3203. (64) Deshmukh, S. R.; Vlachos, D. G. The Power of Microreactors. Fluent News 2005, (Summer), 15. (65) Hsing, I.-M. Simulation Strategies for Microfabricated Chemical Systems. Ph.D. Dissertation, Massachusetts Institute of Technology, Cambridge, MA, 1998. (66) Srinivasan, R.; Hsing, I.-M.; Berger, P. E.; Jensen, K. F.; Firebaugh, S. L.; Schmidt, M. A.; Harold, M. P.; Lerou, J. J.; Ryley, J. F. Micromachined Reactors for Catalytic Partial Oxidation Reactions. AIChE J. 1997, 43 (11), 3059-3069. (67) Srinivasan, R. Microfabricated Reactors for Partial Oxidation Reactions. Sc.D. Dissertation, Massachusetts Institute of Technology, Cambridge, MA, 1998. (68) Hsing, I.-M.; Srinivasan, R.; Harold, M. P.; Jensen, K. F.; Schmidt, M. A. Simulations of Micromachined Chemical Reactors for Heterogeneous Partial Oxidation Reactions. Chem. Eng. Sci. 2000, 55 (1), 3-13. (69) Hsing, I-M.; Srinivasan, R.; Harold, M. P.; Jensen, K. F.; Schmidt, M. A. Finite Element Simulation Strategies for Microfluidic Devices with Chemical Reactions. In Technical Digest, International Conference On Solid State Sensors and Actuators; IEEE: New York, 1997; pp 1015-1018. (70) Aris, R. The Mathematical Theory of Diffusion and Reaction in Permeable Catalysts; Oxford University Press: Oxford, U.K., 1975; Vol. 1 and Vol. 2. (71) Arnold, J. C.; Sawin, H. H. Charging of Pattern Features during Plasma Etching. J. Appl. Phys. 1991, 70 (10), 5314-5317. (72) Ayo´n, A.A.; Zhang, X.; Khanna, R. Ultra Deep Anisotropic Silicon Trenches Using Deep Reactive Ion Etching (DRIE). Solid-State Sensor and Actuator Workshop, Hilton Head, SC, June 4-8, 2000; Transducer Research Foundation: Cleveland, OH, 2000; pp 339-342. (73) Ayo´n, A. A.; Ishihara, K.; Braff, R. A.; Sawin, H. H.; Schmidt, M. A. Microfabrication and Testing of Suspended Structures Compatible with Silicon-on-Insulator Technology. J. Vac. Sci. Technol., B 1999, 17 (4), 1589-1593. (74) Ayo´n, A. A.; Braff, R. A.; Bayt, R.; Sawin, H. H.; Schmidt, M. A. Influence of Coil Power on the Etching Characteristics in a High Density Plasma Etcher. J. Electrochem. Soc. 1999, 146 (7), 2730-2736. (75) Ayo´n, A. A.; Braff, R. A.; Lin, C. C.; Sawin, H. H.; Schmidt, M. A. Characterization of a Time Multiplexed Inductively Coupled Plasma Etcher. J. Electrochem. Soc. 1999, 146 (1), 339-349. (76) DuPont Experimental Station, Building E-357, Wilmington, DE 19880-0357. (77) ACT MicroDevices, Inc., 7586 Peppers Ferry Loop, Radford, VA 24141, (540) 639-1986.

Ind. Eng. Chem. Res., Vol. 46, No. 25, 2007 8305 (78) Riotech, Inc., 3369-T White Mountain Blvd., Lakeside, AZ 85929, (520) 368-5013. (79) Arch Chemicals, Inc., 501 Merritt 7, P.O. Box 5204, Norwalk, CT 06856, (800) 797-1629. (80) Photronics, 15 Secor Rd., Brookfield, CT 06804, (203) 775-9000. (81) Clariant Corporation, Business Unit Electronic Materials, 70 Meister Ave., Somerville, NJ 08876, (908) 429-3500. (82) RBP Chemical Corporation, 150 S. 118th St., Milwaukee, WI 53214, (800) 558-0747. (83) Weichel, S.; de Reus, R.; Bouaidat, S.; Rasmussen, P. A.; Hansen, O.; Birkelund, K.; Dirac, H. Low-temperature Anodic Bonding to Silicon Nitride. Sens. Actuators, A 2000, 82A, 249-253. (84) Plaza, J. A.; Esteve, J.; Lora-Tamayo, E. Effect of Silicon Oxide, Silicon Nitride and Polysilicon Layers on the Electrostatic Pressure during Anodic Bonding. Sens. Actuators, A 1998, 67A, 181-184.

(85) Madou, M. Fundamentals of Microfabrication; CRC Press: New York, 1997. (86) Preston-Thomas, H. The International Temperature Scale of (ITS90). Metrologia 1990, 27 (1), 3-10. (87) Omega Engineering, Inc., One Omega Drive, Stamford, CT 069070047, (800) 848-4286. (88) Hart Scientific, Inc., 799 E. Utah Valley Drive, American Fork, UT 84003-9775, (800) 438-4278. (89) Tektronix, Inc., 14200 SW Karl Braun Dr., Beaverton, OR 97077, (800) 835-9433.

ReceiVed for reView January 16, 2007 ReVised manuscript receiVed May 1, 2007 Accepted May 3, 2007 IE070107W