Fabrication, Characterization and Computational ... - IEEE Xplore

686

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 3, JUNE 2006

Fabrication, Characterization, and Computational Modeling of a Piezoelectrically Actuated Microvalve for Liquid Flow Control Choonsup Lee, Eui-Hyeok Yang, Member, IEEE, S. Mahdi Saeidi, Student Member, ASME, and Jay M. Khodadadi, Member, ASME

Abstract—Liquid-compatible piezoelectric microvalves have been modeled, fabricated, and characterized. The microvalve was designed for proportional flow control of liquid propellant for integrated spacecraft micropropulsion. The microvalve consists of a custom-designed piezoelectric stack actuator bonded onto silicon valve components with the entire assembly contained within a stainless steel housing. The valve seat configuration includes narrow-edge seating rings and tensile-stressed silicon tethers that enable the normally closed and leak-tight operation. A concentric series of narrow rings simulates a “knife-edge” seal by greatly reducing the valve contact area, thereby increasing the seating pressure and consequently reducing leak. Leak testing of the microvalve, conducted using a Helium leak detector, showed 10 6 scc/s for Helium gas. a leak rate of approximately 3 During operation, the valve flow rate was measured using an external Mass Flow Meter (MFM) with a measurement resolution of approximately 10 2 scc/s. The measured forward flow rate for deionized (DI) water is approximately 64 mg/min at an inlet pressure of 20 psi and an applied voltage of 50 V. The mechanical resonance frequency of the microvalve structure was measured at 11.1 kHz. The measured dynamic power consumption of the microvalve is approximately 60 mW when operated at 50 Hz. The measured static power consumption is approximately 2.5 mW at 20 V. Computational modeling of liquid flow within the piezoelectrically actuated microvalve has also been performed. The commercial computational fluid dynamics (CFD) code FLUENT was utilized for solving the continuity and momentum equations. The pressure drop between the inlet and outlet ports was determined as a function of the inlet mass flow rate, and a pressure drop coefficient was determined for each valve plate deflection value. The model-predicted values were compared to the experimental data, and confirmed the sensitivity of the results to the value of the deflection. [1583] Index Terms—Computational fluid dynamics (CFD), liquidcompatible, liquid flow, low-power consumption, piezoelectric, proportional flow control, microfluidics, microvalve, modeling.

I. INTRODUCTION EDUCTIONS in the mass and size of a space instrument or subsystem result in a nearly exponential savings in launch costs as well as making possible a significant increase in

R

Manuscript received April 22, 2005; revisedAugust 24, 2005. This work was supported by NASA’s Code R Enabling Concepts and Technologies (ECT) Program. Subject Editor N. de Rooij. C. Lee and E.-H. Yang are with the Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109 USA (e-mail: [email protected]). S. M. Saeidi is with Intel Corporation, Assembly Technology Development (ATD), Phoenix, AZ USA. J. M. Khodadadi is with the Mechanical Engineering Department, Auburn University, Auburn, AL 36849-5341 USA. Digital Object Identifier 10.1109/JMEMS.2006.876783

mission duration. Nowhere is this more true than for future “mikg total mass), wherein each subsystem crospacecraft” ( will have to maintain essentially the same macroscale capability while requiring extensive miniaturization to fit within the spacecraft size and mass envelope. Also, given the severely limited power constraints on the overall micropropulsion system, microvalves with a zero power, “normally closed” state, and with low-power consumption during opening or closing are needed. Solenoid valves represent the state of the art, havin been refined by decades of technology development. However, these valves suffer from fundamental limitations such as relatively large mass and volume, and high power consumption. The mass limitation comes from the need for several turns of copper wire wound around a high-permeability core material for generating the requisite actuation force. Furthermore, these valves are not capable of producing the low thrust levels and smaller impulse bits required for microspaceraft. Although significant progress has been made in the miniaturization of solenoid valves, these miniaturized valves perform marginally with regard to valve actuation time [1]–[8] or seating force [9]–[13]. Slow valve actuation leads to long thruster on-times and large impulse bits. An alternative, thermally actuated valve technology suffers from the risk of un-commanded valve opening due to changes in the environment, i.e., ambient heating or cooling resulting in uncontrolled initiation of the actuation mechanism. In this work, the JPL authors have experimentally demonstrated, reliable and reproducible microvalve operation [14]. Exsccm were demonstrated tremely low leak rates of for an inlet pressure of 800 psi. The gas and liquid compatible microvalve was developed in response to the requirements for a JPL-initiated development of 1-kg-class microspacecraft test platforms. As outlined above, these microspacecraft require leak-tight, low-power, microvalves. In this particular instance, the requirement was for liquid-compatible valves capable of providing precisely controlled, low propellant flow from a pressurized liquid propellant tank. Due to the microspacecraft’s severely limited propellant and power resources, it was critical that the microvalves exhibit leak-tight and low-power operation (see Table I). Previously developed liquid-compatible microvalves using either electromagnetic and thermal actuators were found to be unacceptable for the microspacecraft application because of high-power consumption [15], [16]. Alternatively, other, liquid-compatible microvalves using either electrostatic or piezoelectric actuators did not meet the demanding requirements for low leak rates [17], [18]. Therefore, in this paper,

1057-7157/$20.00 © 2006 IEEE

LEE et al.: MODELING OF A PIEZOELECTRICALLY ACTUATED MICROVALVE FOR LIQUID FLOW CONTROL

687

TABLE I MICROVALVE SPECIFICATIONS FOR 1-kg-CLASS MICROSPACECRAFT PROPULSION

we present the results of the successful demonstration of our approach toward the development of a leak-tight, low-power, fastactuation microvalve for proportional flow control of liquid propellant. In parallel with the experimental development at JPL, the authors at Auburn University undertook a computational modeling study in order to elucidate the complex features of the liquid flow within the various components of the JPL-developed microvalve. II. MICROVALVE DESIGN The JPL microvalve is actuated using a custom-designed piezoelectric stack actuator, which is bonded onto silicon-based components consisting of a valve seat, a lower boss, and an upper boss, as shown in Fig. 1. The piezoelectric stack actuator produces a block-force of approximately 1000 N, resulting in a valve-opening pressure far greater than the seating pressure made up of a combination of inlet gas pressure and the initial seating pressure arising from the tensile-stressed silicon tether and membrane suspension [14]. The segmented piezoelectric stack actuator consists of two “actuation” zones surrounding a central inactive zone. Application of a voltage to the piezoelectric stack causes the actuation zones to expand vertically, thereby lifting the boss center plate (bonded to the inactive zone), away from the valve seat. This actuation opens up a channel between the inlet and outlet ports, permitting the passage of fluids as shown in Fig. 2. Since the piezoelectric actuator is essentially a stacked capacitor, it consumes extremely low power, allowing for a near zero-power operation for the microvalve. Unlike in our previous microvalve design for high-pressure-gas applications [14], the microvalve reported in this paper has a silicon membrane called the upper-boss, for isolating the piezoelectric actuator from the liquid effluent. Because the liquid effluent remains inside the silicon chamber, it does not cause an electrical short within the piezoelectric actuator. The upper boss layer is compression bonded to the lower boss. The bonded boss stack is subsequently bonded to the valve seat. Finally, the custom-designed piezoelectric actuator is epoxy bonded to the top of the upper boss layer. On the seat plate, the concentric series of narrow rings simulate a “knife-edge” seal as described above. On the boss plate, the center portion has a 2- m-thick silicon dioxide layer for a hard seat coating material. The oxide layer generates tensile stress in the silicon tether suspension as well as in the upper boss membrane. Fig. 2 shows the microvalve operation principle. The microvalve is in the normally closed (“off” state), as shown in

Fig. 1. Structure of the liquid-compatible microvalve. All silicon components are metal-to-metal thermocompression bonded and the custom-designed piezoelectric stack is bonded on top of the upper-boss wafer.

Fig. 2(a). The normally closed condition is achieved by applying an initial seating pressure on the valve seat during the bonding of the piezoelectric actuator to the silicon assembly [14]. Application of a voltage to the piezoelectric actuator makes the active zones of the piezoelectric actuator expand vertically, lifting the lower-boss center plate up from the valve seat, as shown in Fig. 2(b). This action opens up a flow path between the inlet and outlet ports. Proportional flow control of the fluid is achieved by controlling the extent of the upward displacement of the piezoelectric actuator by changing the applied voltage. III. MICROVALVE FABRICATION Three silicon wafers (300 m thick) used for the valve seat and boss plates are thermally oxidized (0.5 m oxide). The seating rings are first lithographically defined on the valve seat wafer. The silicon dioxide layer is then selectively removed in a 10:1 Buffered Oxide Etchant (BOE). The remaining oxide forms the masking layer for the deep reactive ion etching step (DRIE), in order to generate 10 m deep seating ring structures.. Following the metallization step, a central outlet port is formed using DRIE. Fabrication of the lower boss plate is initiated by growing and subsequently patterning a 2- m-thickplasma-enhanced chemical vapor deposition (PECVD) silicon dioxide layer. Once again, the oxide layer is etched in BOE (10:1). Also, as in the previous case, the bonding metals (Cr/Pt/Au) are formed. The boss wafer is then patterned to define the boss (or valve flap and tethers), which is released in a final DRIE process. On the upper boss plate, the 150- m-thick silicon membrane is defined using DRIE,

688


Fig. 2. Operating principle of the microvalve. The piezoelectric stack actuator consists of two “actuation” zones surrounding a central inactive zone. The actuation zones expand vertically as a voltage is applied to the piezoelectric stack, lifting the boss center plate (bonded to the inactive zone), away from the seat plate. This actuation creates a channel between the inlet and outlet ports, permitting the passage of fluids. (a) normally closed “off” state (b) actuated “on” state.

Fig. 3. SEM images of the (a) valve seat, (b) upper boss plate, and (c) valve sealing surface of the lower boss plate and (d) upper boss bonding side of the lower boss plate.

followed by the deposition and patterning of the Cr/Pt/Au bonding layer. Four inlet ports are formed subsequently in each corner of the boss plate using DRIE. The upper boss, lower boss and seat wafers are bonded simultaneously to create a sealed, yet movable structure [14].

An Electronic Visions aligner and thermocompression bonder is used to align and bond the two wafers. The bonder chamber Torr prior to the bonding operation. is pumped down to A piston pressure of 1 MPa is applied at 380 C in the vacuum chamber to provide the necessary thermocompression


Fig. 4. Fully assembled and packaged microvalve.

Fig. 6. Schematic diagram showing the microvalve test setup. Leak rate (off state) testing of the packaged microvalves was conducted using Helium-gas.

Fig. 5. PZT actuator vertical displacement versus applied voltage in the unbonded and bonded states. The stroke of the piezoelectric actuator was measured before and after bonding to silicon components. There is negligible difference in stroke between the two conditions. This implies that the piezoelectric actuator essentially exerts a very high seating force and can provide robust on–off operation for the microvalve. The measured deflections of the piezoelectric actuator are about 4 m and 1 m, respectively, at applied voltages of 50 V and 10 V.

0

689

0

bonding force. Subsequently, the piezoelectric stack actuator is carefully centered (using a specially designed jig) and bonded onto the top of the boss plate using an epoxy (Hysol E/A 9394, cured at room temperature). Finally, the microfabricated valve components are bonded to stainless steel fixtures, which are then hermetically sealed using the same epoxy. Scanning electron microscope (SEM) images of the microfabricated silicon components are shown in Fig. 3. In Fig. 3, the valve seat has an outlet port, concentric “knife-edge” rings (which can also act as a particle filter). On the upper boss plate, there are 4 inlet ports, a thin membrane. On the lower boss plate, the tether, center plate, inlet ports. Fig. 4 shows the fully assembled and packaged microvalve. The inlet gas tube is then connected to the inlet ports in the upper boss plate. Also seen are the outlet port (diameter: 200 m) in the seat wafer and the wires for applying voltage to the piezoelectric actuator. IV. CHARACTERIZATION The stroke of the piezoelectric actuator was measured before and after bonding to the silicon components. As shown in Fig. 5,

Fig. 7. Measured leak rates for six normally-closed (nonactuated) microvalves. The measured leak rates of several microvalves ranged from 3 10 scc/s to 4 10 scc/s at an inlet pressure of 50 psi.

2

2

there is negligible difference in stroke between the two conditions. This implies that the piezoelectric actuator essentially exerts a very high seating force and the stroke is largely unaffected by the compliance of the silicon tethers and membranes. The stroke could be affected under fluid flow conditions (See the Section V-C.), however stroke measurements of the packaged microvalve under fluid flow conditions was not possible in this work. The blocking force of the stacked multilayer piezoelectric actuator used in the microvalve is estimated at approximately 1000 N. The measured deflections of the piezoelectric actuator m, respectively, for applied voltages are about 4 m and V. of 50 V and Leak rate (off state) and flow rate (on state) testing of the packaged microvalves were conducted using Helium-gas and deionized (DI) water, respectively. The leak test block diagram is shown in Fig. 6. Helium leak rates were measured using a Helium leak detector. The measured leak rates of several miscc/sec to scc/sec crovalves ranged from at an inlet pressure of 50 psi, as shown in Fig. 7. The low leak rates are attributable to the combination of the smooth hard-seat surfaces (RMS surface roughness is 0.3 nm) and the pre-loaded seating configuration [14]. For the liquid flow rate measurement, we measured the mass of the DI water flowing out from the valve for a fixed period as described in Fig. 8. The measured forward flow rates at various inlet pressures for deionized DI water are

690


Fig. 10. Measured flow rates for a microvalve actuated with voltage pulses with varying duty ratios for various inlet pressures. An 18-V square pulse was applied. The measured dynamic flow rate is 30 mg/min at a 90% pulse width for a 15-psi inlet pressure.

Fig. 8. Test setup for water flow rate. The water flow rate (on state) testing of the packaged microvalves was conducted using DI water flowing out from the outlet for a limited period.

Fig. 9. Measured flow rates for an actuated microvalve at various inlet pressures. As the voltage applied to the piezoelectric actuator increases, the vertical deflection increases. This action increases the flow of the DI water, thereby providing the proportional flow control. The measured flow rate is approximately 60 mg/min at an inlet pressure of 20 psi (for an applied voltage of 40 V).

Fig. 11. Power consumption versus operating frequency for a microvalve sinusoidal signals. The dynamic power operated with 10 V consumption was measured by taking the phase delay into account. The , 50 Hz dynamic power consumption is about 60 mW for 10 V actuation. This value is very low compared to the power consumed by either thermally actuated or magnetically actuated microvalves. This power consumption is further reduced to a few milliwatts when the microvalve is operated in a static proportional flow control mode.

Fig. 12. Measured frequency response of the microvalve. The estimated transient response time is approximately 30 s.

shown in Fig. 9. As the voltage applied to the piezoelectric actuator increases, the vertical deflection also increases. The increased deflection increases the flow of the DI water, thereby


691

Fig. 13. Isometric views of the microvalve geometry used for the computer modeling. (a) One quarter of the microvalve structure was used for modeling as shown. Various components are clearly shown. The solid parts are shown in shades of gray, whereas the fluid is shown in red. (b) The complete microvalve constructed by combining the four quarter sections.

providing proportional flow control operation. The measured flow rate is approximately 60 mg/min at an inlet pressure of 20 psi (for an applied voltage of 40 V). As shown in Fig. 10, we measured the DI water flow rate versus the various duty ratios of the pulsed voltage signal for different inlet pressures. The measured dynamic flow rate is 30 mg/min at a 90% pulse width for a 15-psi inlet pressure. The static power consumption is primarily due to the leakage current in the piezoelectric actuator and was measured at 2.5 mW at 20 V. The dynamic power consumption was measured by taking the phase delay into account. Fig. 11 shows the dynamic power consumption measurement results. For instance, the dy, namic power consumption is about 60 mW for a 10 50 Hz actuation. This value is very low compared to the power consumed by either thermally actuated or magnetically-actu-

ated microvalves. This power consumption is further reduced to a few milli watts when the microvalve is operated in a static proportional flow control mode. We have also measured the mechanical resonance frequency of the piezoelectric actuator bonded to silicon microvalve using a laser doppler vibrometer, as shown in Fig. 12. The measured value is approximately 11.1 kHz, which shows that the microvalve has sufficient bandwidth for providing fast transient response. V. COMPUTATIONAL MODELING AND DISCUSSIONS A. Geometry of the Microvalve Model An isometric view of a one-quarter section of the microvalve is shown in Fig. 13(a). The solid components are depicted in shades of grey, whereas the fluid is shown in red. The flow enters

692


Fig. 14. Isometric views of the three-dimensional computational mesh. (a) The complete computational mesh. The green lines identify the locations where the various blocks meet. (b) A close-up view showing the detail of one quarter of the valve seat rings.

the microvalve from four inlet ports positioned at the corners of the upper boss plate (Fig. 13(b)) and exits the valve from an outlet port fabricated in the middle of the seat plate. The diameter of the inlet and outlet pipe is 200 m. The length of the inlet pipes is 7100 m whereas the length of the outlet pipe is 600 m. The liquid enters a 10- m high box after passing through the inlet pipe and then flows into a bigger box with 6 mm 5.2 mm 0.56 mm dimensions. There are twelve thin seat rings with different radii that surround the outlet hole. The thickness of the rings is 1 m and their height is 10 m. The diameter of the innermost ring is 110 m and other rings are evenly spaced concentrically at a distance of 150 m from each other. A square-shaped lower boss plate is shown on top of the outlet port with dimensions of 1.8 mm 1.8 mm 0.4 mm. The flow is directed to the outlet pipe after passing through the space between the lower boss plate and the rings. The spacing between the lower boss plate and the seat plate ranges from 11 to 13.5 m, which means the distance between the top of the valve seat rings and lower boss plate varies between 1 to 3.5 m. The model considered three deflections of 1, 2.7, and 3.5 m. B. Problem Formulation A 3-D structured mesh with 330 000 hexahedral cells that represents the major features of the microvalve was generated

for the numerical model. Due to the symmetry of the microvalve about two planes, only a one-quarter section of the microvalve was necessary for the model. The mesh is denser in regions that experience excessive pressure drop (i.e., between the boss and seat plates and where the flow enters the microvalve). Different views of the computational mesh are shown in Fig. 14. The green lines identify where the various blocks meet. Continuumbased momentum and continuity equations are valid for this problem [19]. The flow is laminar, incompressible, and steady. The governing equations of continuity and momentum in index notation form are: (1) (2) Version 6.2 of the commercial code FLUENT was utilized for solving the governing equations. The no-slip boundary condition was chosen for the walls whereas symmetry boundary conditions were used for the planes of symmetry. A second order upwind scheme was chosen to discretize the momentum equation whereas the SIMPLE algorithm was used to couple the pressure and velocities. Two different types of boundary conditions were applied to the inlet and outlet ports. At first, a gage


693

Fig. 15. Isometric views showing pathlines of selected particles released at the inlet plane for a deflection 2.7 m and pressure drop of 2.4 psi (a) The path of particles released at the inlet are colored by the local fluid pressure. The color bar on the left shows the pressure values. Most of the pressure drop happens just below the inlet port and over the rings. Also note that the monitored particles prefer to flow over the tether rather than beneath it. (b) A different view for the same operating conditions is presented.

stagnation pressure was specified at the inlet port and the outlet pressure was set to zero (gage). In order to verify the results, the mass flow rate of the system that was found by applying the above boundary condition was chosen as the inlet boundary condition assuming that the flow at the outlet was fully developed. A total of 15 cases were run for three deflections of 1, 2.7, and 3.5 m and 5 inlet stagnation pressures of 3, 5, 10, 15, and 20 psi with water as the working fluid. C. Results and Discussion of the Modeling Study In Fig. 15, two different views of the lines of fluid flow (tracked by imaginary particles) are shown for the case of 2.7 m deflection and a pressure drop of 2.4 psi. The monitored particles are released at the inlet plane from two rake sets that are placed normal to each other. The monitored “pathlines”

released from these rakes are colored by the local stagnation pressure values. The fluid particles prefer to flow over the tether rather than beneath it due to lower resistance. Pressure is almost constant within the inlet pipe but drops suddenly and markedly where the flow impacts the valve seat plate that is located 10 m away from the inlet port. Once the flow emerges into the microvalve cavity, pressure remains almost constant until the flow passes over the rings in the gap between the seat and the boss plates. The pressure monotonically decreases as the flow passes over each ring. Passage of the flow through this narrow gap causes most of the pressure drop in the system. There is also a lesser pressure drop associated with the small box under the inlet port. Since most of the pressure drop occurs across the rings, especially for large deflections, a 2-D axisymmetric analysis was also performed, taking into account the geometry

694


Fig. 16. Pressure drop versus mass flowrate for a 1 m deflection. Computational results based on 2-D (ring-only) and 3-D (entire microvalve) are compared to the experimental data. The 2-D predictions are consistently below the 3-D results. This is expected since the 2-D data represents the pressure drop over the rings only. For the 3-D runs, two different boundary conditions were used and the predictions for both boundary conditions exhibits a unified trend since the 3-D data follow a single straight line. The experimentally measured values fall in between the 2-D and 3-D predictions.

Fig. 18. 3-D predictions of the pressure drop between the inlet and outlet ports versus mass flow rate for various deflections. As expected, for a given mass flow rate the predicted pressure drop increases as the deflection is lowered. For a given deflection, a linear variation between the pressure drop and mass flow rate is observed supporting the conclusion that this is indeed a proportional microvalve.

Fig. 19. Pressure drop coefficient versus deflection. The slopes of the pressure drop versus flowrate data from Fig. 18 were computed and they exhibit an exponential dependence on the deflection. Given that uncertainty in the value of deflection inherent in the system, the discrepancies between the measurements and computational predictions are attributed to such errors in value of deflection. Fig. 17. Pressure drop versus mass flowrate for a 3.5 m deflection. Computational results based on 2-D (ring-only) and 3-D (entire microvalve) are compared to the experimental data. The 2-D predictions are consistently below the 3-D results. The measured pressure drop values for a given flow rate are consistently above the 2-D and 3-D predictions.

of the rings. The pressure drop over the rings determined from the 2-D analysis verifies the accuracy of the 3-D microvalve model results, which are consistently higher than the 2-D predictions. Pressure drop versus mass flowrate results are compared between the numerical (2-D and 3-D) and experimental data for a 1- m deflection in Fig. 16. The observed trend for the numerical results is a straight line that passes through the origin. This is consistent with observations for fluid systems operating at low Reynolds numbers (maximum of 6 in this problem). The experimental results lie between the 2-D and 3-D results. The experimental data are close to 2-D results for low inlet pressure cases but for the cases with high inlet pressures, the experimental data tend toward 3-D results. A similar comparison is demonstrated in Fig. 17 for

a deflection of 3.5 m. In contrast to the 1- m deflection case, the experimental results are very different from numerical results in this case, although the difference between 2-D and 3-D numerical simulations is still reasonable. In Fig. 18, the 3-D numerical simulations for pressure drop between the inlet and outlet ports versus mass flowrate are shown for deflections of 1, 2.7, and 3.5 m. For a constant pressure difference, the mass flow rate increases for greater values of deflection. The values of the pressure drop coefficient, K versus deflection are shown in Fig. 19. The pressure drop coefficient is essentially the slope of each line in Fig. 18 (3) Because the pressure drop changes linearly with respect to change in mass flowrate, K is a constant number. K is exponentially dependent on deflection, and is therefore extremely sensitive to changes in deflection. The empirical dependence


obtained by plotting a trendline through the points in Fig. 19 is as follows:

(4) is the preswhere def is the value of deflection in microns, sure drop (the difference between the inlet and outlet stagnation is the mass flowrate of the system in pressures) in psi and mg/sec. The exponential trend exhibited in Fig. 19 can explain the large difference between experimental and numerical results for high deflection cases. The other reason could be the error in determining the deflection precisely. The measured deflections of the PZT stack actuator did not incorporate any changes in deflection during liquid flow, which could give rise to approximately 1 m level of error in the value of the deflection. This uncertainty in the value of deflection can lead to the big changes observed in the K value.

VI. CONCLUSION We have successfully demonstrated piezoelectricallyactuated, liquid-compatible silicon microvalves. The microvalve incorporates a custom-designed piezoelectric stack actuator to provide the actuation forces necessary for high-pressure operation. A hard seat configuration using a series of narrow concentric seating rings contributes to the enhanced leak-tight microvalve operation. The measured leak rate of a microvalve scc/s. The measured static power using Helium gas is consumption is approximately 2.5 mW at 20 V. The mass flow rate varies linearly with pressure difference. This is expected because the flow has low Reynolds numbers, with the highest Re value being 6. The results indicate extreme sensitivity of the pressure drop coefficient to changes in deflection. Therefore, the discrepancies between the numerical and experimental results can be explained in terms of the error in the measured value of the deflection. The microvalve is capable of proportional flow control of liquid propellant for integrated micropropulsion applications. Potential applications for this microvalve technology include low impulse-bit thruster modules for use in very small spacecraft as well as for providing precise attitude control functions for larger spacecraft.

ACKNOWLEDGMENT The authors would like to acknowledge Dr. Thomas George for his support and advice throughout this research project. The authors would also like to thank Mr. Christopher Johnson at Auburn University for his technical contributions on the computational modeling. The experimental phase of the research described in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology under a contract with the National Aeronautics and Space Administration.

695

REFERENCES [1] M. J. Zdeblick, R. Anderson, J. Jankowski, B. Kline-Schoder, L. Christel, R. Miles, and W. Weber, “Thermopneumatically actuated microvalves and integrated electro-fluidic circuits,” in Proc. Solid-State Sensor and Actuator Workshop, Hilton Head, SC, 1994, pp. 251–255. [2] P. W. Barth, “Silicon microvalves for gas flow control,” in Proc. Transducers’95, Stockholm, Sweden, Jun. 25–29, 1995, p. 276. [3] X. Yang, C. Grosjean, Y. C. Tai, and C. M. Ho, “A MEMS thermopneumatic silicone membrane valve,” in Proc. MEMS’97, Nagoya, Japan, 1997. [4] H. Jerman, “Electrically-activated micromachined diaphragm valves,” in Solid State Sensor and Actuator Workshop, Hilton Head, SC, June 4–7, 1990, p. 67. [5] J. Franz, H. Baumann, and H. Trah, “A silicon microvalve with integrated flow sensor,” in Proc. Transducers’95, Stockholm, Sweden, Jun. 25–29, 1995. [6] C. Ray, C. Sloan, D. Johnson, J. Busch, and B. Petty, “A silicon-based shape memory alloy microvalve,” in Mat. Res. Soc. Symp. Proc., vol. 276, 1992, pp. 161–166. [7] J. Busch and D. Johnson, “Prototype of micro-valve actuator,” in IEEE Proc., CH2832-4/90, 1990, pp. 40–41. [8] D. Johnson and M. Bokaie, “Valves for instrumentation and propulsion systems in microspacecraft,” in Proc. 9th Adv. Space Propulsion Workshop. Pasadena, CA, Mar. 11–13, 1998, JPL Internal Document D-15671,. [9] M. Huff, J. Gilbert, and M. Schmidt, “Flow characteristics of a pressurebalanced microvalve,” in Proc. Transducers’ 93, Yokohama, Japan, Jun. 7–10, 1993. [10] M. Huff and M. Schmidt, “Fabrication, packaging, and testing of a wafer-bonded microvalve,” in Solid State Sensor and Actuator Workshop, Hilton Head, SC, Jun. 22–25, 1992, pp. 194–197. [11] M. Shikida, K. Sato, S. Tanaka, Y. Kawamura, and Y. Fujisaki, “Electrostatically-actuated gas valve with large conductance,” in Transducers’93, Yokohama, Japan, June 7–10, 1993. [12] T. Ohnstein, T. Fukiura, J. Ridley, and U. Bonne, “Micromachined silicon microvalve,” IEEE Proc., CH2832-4/90, pp. 95–98, 1990. [13] S. Kluge, G. Klink, and P. Woias, “A fast-switching, low-power pneumatic microvalve with electrostatic actuation made by silicon micromaching,” Amer. Lab., pp. 17–18, Mar. 1998. [14] E. H. Yang, C. S. Lee, J. Mueller, and T. George, “Leak-tight piezoelectric microvalve for high-pressure gas micropropulsion,” J. Microelectromech. Syst., vol. 13, no. 5, Oct. 2004. [15] M. Capanu, J. G. Boyd, and P. J. Hesketh, “Design, fabrication, and testing of a bistable electromagnetically actuated microvalve,” J. Microelectromech. Syst., vol. 9, pp. 181–189, Jun. 2000. [16] H. Kawada, H. Yoshida, M. Kamakura, K. Yoshida, M. Saitou, M. Kawahito, and S. Tomonari, “Thermally driven microactuator containing thermal isolation structure with polyimide and it’s applications to microvalve,” Proc. Transducer’03, pp. 1935–1938, 2003. [17] S. Messner, J. Schaible, J. Vollmer, H. Sandmaier, and R. Zengerle, “Electrostatic driven 3-way silicon microvalve for pneumatic applications,” Proc. IEEE MEMS’03, pp. 88–91, Jan. 2003. [18] S. Kluge, G. Neumayer, U. Schaber, M. Wackerle, M. Maichl, P. Post, M. Weinmann, and R. Wanner, “Pneumatic silicon microvalves with piezoelectric actuation,” Proc. Transducers’01, pp. 924–927, 2002. [19] G. E. Karniadakis and A. Beskok, Micro Flows: Fundamentals and Simulation. New York: Springer, 2001.

Choonsup Lee received the B.S. degree in electrical engineering at the Kyoungpook National University in 1996 and the M.S and Ph.D. degrees in electrical engineering at the Korea Advanced Institute and Science and Technology (KAIST) in 1998 and 2002, respectively. In 2002, he joined the MEMS Technology Group at the Jet Propulsion Laboratory (JPL) as a Caltech Postdoctoral Scholar at JPL. He has extensive experience in the design and characterization of microsensors and microactuators such as thermal inkjet printhead, force-balanced tunneling microaccelerometer, infrared detector, silicon microlens, MEMS-based bandpass filter, microvalve, nanochannel, lateral field emission device and other MEMS/NEMS devices. Currently, he is working on microfluidic/nanofluidic devices and nanowire-based sensors. He has published 15 international journal papers and 25 refereed conference papers.

696


Eui-Hyeok Yang (A’95–M’03) received the B.S., M.S., and Ph.D. degrees in the Department of Control and Instrumentation Engineering from Ajou University, Korea, in 1990, 1992, and 1996, respectively. He joined the Fujita MEMS research group at the Institute of Industrial Science, University of Tokyo, Japan, as a Visiting Postdoctoral Researcher in 1996. He received a research fellowship from the Japan Society for the Promotion of Science from 1996 to 1998. Since 1999, he has been employed at JPL, where he initiated the development of MEMS actuator-based adaptive optical devices. He is currently a Senior Member of the Engineering Staff at NASA’s Jet Propulsion Laboratory (JPL) and the task manager for several MEMS technology development projects in the area of micro- and nanotechnologies. He has extensive experience in microactuator, deformable mirror, and optical membrane fabrication. He is leading the development of MEMS-based deformable mirrors and actuators for future large aperture telescopes. He is also leading the development of MEMS-based piezoelectric valves for future microspacecraft applications. He participated in the technical evaluation of MEMS mirror array technologies being developed for the Multi Object Spectrometer (MOS) project for the James Webb Space Telescope (JWST). He has been successful in wiinning extremely competitive major research grants that represents an exceptional achievement and productivity within NASA. He is a technical monitor for a NASA SBIR project, and he is a Research Adviser for the National Research Council (NRC) in the area of microactuators and active mirror technologies. His current research interests include all aspects of microsensors/actuators, microfluidics, adaptive optics, micro/nanoenergy conversion, and nanomanufacturing technologies. He has published about 90 papers in the field of MEMS, and has eight patents issued or pending. Dr. Yang is a member of the Technical Program Committee (TPC) for the IEEE Sensor Conference. He is Topic Organizer of the Micro and Nano Devices Topic, within the MEMS Division of the ASME International Mechanical Engineering Congress and Exposition. He has been serving as a referee fro several journals, international conferences, and proposals. in recognition of his excellence in advancing the use of MEMS-based actuators for space applications, he received the Lew Allen Award for Excellence for 2003 at JPL.

S. Mahdi Saeidi received the B.Sc. degree from the Mechanical Engineering Department of Sharif University of Technology, Tehran, Iran, in 1997 and the M.Sc. degree from the Mechanical Engineering Department of Shiraz University, Shiraz, Iran, in 2000. He was then employed at Iran-Khodro Powertrain Company (IP-CO) as a CFD analyst, Tehran, Iran. In 2002, He joined the Mechanical Engineering Department of Auburn University, USA, where he received the Ph.D. degree in 2005. His research areas of interest are Computational and Analytical Fluid Dynamics and Heat Transfer. His specialties are in Microfluidics, Turbomachinary, Combustion and Turbulence. He has several refereed journal and conference papers. He is currently a Senior Thermal Engineer, Intel Corporation, Assembly Technology Division (ATD), Phoenix, AZ. Dr. Saeidi has been a Student Member of the American Society of Mechanical Engineers (ASME) since 2004.

Jay M. Khodadadi received the B.S., M.S., and Ph.D. degrees in mechanical engineering from the University of Illinois at Urbana-Champaign, IL, in 1980, 1982 and 1986, respectively. Soon thereafter, he joined the Faculty of Mechanical Engineering at Auburn University, Auburn, IL, in 1987. His expertise is in the area of fluid/thermal sciences. He has focused on studying transport phenomena in materials processing. His research to date involve mathematical and physical modeling of tundish flows, mold of continuous casters, thermophysical property determination and droplets under microgravity. In recent years, he has focused on computational modeling of active control of flow and heat transfer in heat exchange systems, microfluidic systems and phase change thermal storage systems. Dr. Khodadadi has been a Member of the American Society of Mechanical Engineers (ASME) since 1987.