SOLDER-ASSEMBLED LARGE MEMS FLAPS FOR FLUID MIXING. We describe surface-micromachined thermal actuator-based MEMS flaps with a length scale of 1,000

Transcription

1 Final Manuscript for IEEE Transactions on Advanced Packaging SOLDER-ASSEMBLED LARGE MEMS FLAPS FOR FLUID MIXING Zhichun Ma 1*, Elizabeth Bradley 1, Thomas Peacock 2, Jean R. Hertzberg 1, and Yung-Cheng Lee 1 1 University of Colorado, Boulder, CO Massachusetts Institute of Technology, Cambridge, MA ABSTRACT We describe surface-micromachined thermal actuator-based MEMS flaps with a length scale of 1,000 µm. These flaps were developed for the enhancement of fluid mixing in an experimental study of a planar air jet. The scales of the flow physics required the actuators to be much larger than the typical MEMS scales, and the experiment required an array of 10 flaps (1 cm in length) to be soldered onto a ceramic substrate with high precision. These unusual requirements made it difficult to achieve good assembly yields that could provide large deflections. To improve the yields and deflections, we modified the initial flap design by reducing the size of the solder pads, removing sharp corners, changing the number and the width of the actuator s hot arms, and strengthening the support beams. In addition, we developed a controllable assembly process to reflow tin/lead solder on gold pads, and we used numerical simulations extensively in the design process to improve the performance. The experimental measurements corroborated these simulations: the deflection was increased from 6.4 to 11.1 µm under a 5 V applied voltage. Flow velocity measurements showed that these MEMS flaps amplified the natural instabilities of the planar jet. * Zhichun Ma; Department of Mechanical Engineering Campus Box 0427, University of Colorado, Boulder, CO Phone: (720) ; FAX: (303) ; maz@colorado.edu 1

2 Index Terms: MEMS, actuator, solder, flip-chip, packaging, modeling, flow control. I. INTRODUCTION The accurate prediction and control of fluid flows have long been major research areas. The ability to control flow, particularly in aerospace and combustion applications, can bring significant economic and environmental benefits. Most experimental flow-control studies have been performed using conventionally machined actuators [1-3], but there has been strong interest in using MEMS based technology for flow control because it offers several advantages over conventional machining, such as batch processing of devices on the µm scale. A review of the use of MEMS devices for flow control is given by Ho and Tai [4]. Many different geometries have been evaluated for flow applications, including micro fans and various types of flaps [5-8]. Most studies on MEMS/fluids interactions have to date focused on the MEMS actuator s effects on the fluid flow [7-8]. To the best of our knowledge, there are not many studies that address the design and fabrication of MEMS specifically for the purpose of fluid mixing. In this paper, we describe the design, fabrication, assembly, and testing of thermal-actuator based MEMS flaps for the purpose of forcing a planar jet of air. A schematic of the experiment is presented in Figure 1. The MEMS flaps are mounted on either side of the jet along the exit slit, providing a disturbance that is naturally amplified as it travels downstream. The planar jet is unstable when the Reynolds number, based on the half-width and peak velocity of the jet, exceeds a critical value [9]. Small-scale velocity disturbances in the boundary layer are amplified as they progress downstream, generating coherent structures that dominate the flow field. The anti-symmetric instability has been excited using acoustic forcing [10]. Using MEMS micro-actuators, we have been able to selectively excite both the anti-symmetric and the symmetric instabilities of the planar jet at low Reynolds numbers [11]. In addition, an asymmetric 2

3 mode in which only one side of the jet is forced has been studied [11]. This was possible because the micro-actuators allowed us to provide localized disturbances, with an arbitrary phase relationship, on either side of the experimental jet. This is contrast to acoustic techniques, for which the entire flow is subject to forcing of uniform phase. The effect of the micro-actuators is significant, as shown by the images in Figure 2. In the unforced jet, Figure 2(a), the MEMS flaps are motionless, the flow is laminar, and there is little entrainment of the surrounding air into the jet. In Figure 2(b), we show an image of the same jet, subject to asymmetric forcing at 6Hz by MEMS flaps. The instability excited by the flaps gives rise to large-scale vortices. These vortices provide enhanced mixing of the jet with the ambient air. This phenomenon is known to have a profound effect on combustion processes. In order to investigate the effect of different forms of forcing on the nonlinear development of the flow, it is very important to study the design and fabrication of MEMS flaps in depth. The driving mechanism for the flaps was electro-thermal. This technology is supported by current commercial foundry processes and provides enough deflection and force to disturb the flow [15]. The polysilicon flaps were mounted on a ceramic substrate using a flip-chip solder assembly technique. This assembly approach allowed multiple actuators to be accurately positioned on a ceramic substrate, as the solder surface tension aligns the flaps on the substrate during the solder re-flow process. The flap assembly was then mounted next to the exit of the jet. The initial design for the MEMS flaps is shown in Figure 3. Views are referenced by the final assembly position. The deflection of the flap was upward. These actuators were intended for a feasibility study, which was highly successful in demonstrating the ability to influence a jet flow. Improvements were needed, however, as the original design was far too fragile and inefficient to survive laboratory conditions. Moreover, experimental study of the planar jet required a 1 cm-long array of flaps to be 3

4 mounted along the edge of the jet. As each MEMS flap was 1mm long, this required mounting 10 flaps in close proximity on a ceramic substrate, with very high precision (see Figure 4). Such an assembly is two orders of magnitude larger than a typical MEMS device, providing a significant challenge for fabrication. In addition, the flaps must be uniformly aligned on the substrate, so as to provide a uniform disturbance to the jet. This requirement, plus efficiency requirements, demanded major advances in the MEMS flap technology. Significant efforts in design, simulation, characterization, fabrication, and assembly were needed. We describe our solutions to these challenges in the following sections. In section II, we present a new design of the MEMS actuators, and describe structural changes that we made to the original design in order to improve performance. Then, in section III, we discuss the design process and detailed numerical simulations. Next, in section IV, we describe the assembly process that was refined to enable consistent fabrication of the actuators. Finally, in section V, we present experimental results on the performance of the devices, showing good agreement with the predictions of the numerical simulations. We close with novel experimental flow mixing results that were attained with these improved MEMS flap assemblies. II. DESIGN The first generation of MEMS flaps, shown in Figure 3, worked well for initial feasibility demonstrations, but they had several problems: 1. Incomplete release: it took more than twelve minutes to remove the oxide layer under the solder pad for each flap, and it was difficult to control such a long process for complete release. 2. Broken hot arms: the hot arms were weak and easily broken during assembly or operation. 3. Small movements and flimsy structure: the flap's vertical movement was small and strongly affected by the moving support arms. 4

5 The release problem was solved by using multiple bonding pads. The single large solder pad shown in the original design (Figure 3) was replaced by two pads, as shown in Figure 5. Between the pads, etch holes were added to enhance the release process. The difficulties with the hot arms were solved by decreasing the number of arms, while increasing their width. As shown in Figures 5 and 6, the hot-arm width was increased from 5 to 12 µm, and the number of arms for each flap was reduced from 21 to 7. The intention here was to increase the total electrical resistance of the hot arms on each flap. Furthermore, the first-generation MEMS flaps had sharp corners on each arm near the edge of support beams and flap plate, which was structurally unsound (see Figure 7). By placing the hot arms below, when viewed from bottom (referenced by the final assembly position), the support beams and attaching them at the bottom of the flap plate, these corners were eliminated, as shown in Figure 8, and the flap deflection was now downward. The movement problem was solved by strengthening the support beams and shortening the hot arms. As shown in Figure 3, each support beam in the original design was 420 µm long and 45 µm wide. This was too long and compliant. By adding a triangular corner plate and increasing the beam width from 45 to 55 µm, we enhanced the stiffness in the vertical direction significantly. The tip of the hot arm was moved inward from the edge of the flap by 75 µm, allowing the hot arms to push the flap much further downward, as the flap deflection beyond the tip of the hot arm was proportional to the slope of the deformation. Although these design changes appear to eliminate the problems with the first-generation design, it is impossible to verify that conjecture a priori. For example, even with the same current density, the widened hot arm might not generate the same bending power because the structure could have different temperature distributions resulting from different heat transfer characteristics. Such nonlinear 5

6 interactions cannot be analyzed easily, and so we used numerical simulation tools to simulate the behavior of the actuator in order to characterize the new design prior to fabrication. III. SIMULATION Commercial software, CoventorWare [13], was used to perform numerical simulations of the actuator designs shown in Figures CoventorWare uses the finite element approach, incorporating electrothermal-mechanical analysis. Brick elements of 27 nodes with dimensions 80µm x 80µm x 10µm were used in the simulation. CoventorWare uses an iterative process to calculate the static deflection of the actuator. The voltage-current distribution was solved first, followed by a heat-transfer analysis. The resulting temperature field was used to determine the mechanical deflection. The thermal and mechanical solutions were fed back into the electrical modeling module of the software to correct for temperature effects. This iterative process continued until the solutions converged to within 0.1% tolerance. The simulation results are presented in Figures The boundary conditions were as follows: 1) The regions in contact with the solder pads and the supporting arms at the side were assumed to be rigid, and were therefore neglected in the simulation. 2) The ends of the hot arms and the flexures supporting the flap were assumed to be fixed, and were maintained at room temperature. 3) A known driving voltage (e.g., 5 V) was applied across the hot arms and the support flexures. 4) Convective and radiation heat transfer occurred from all surfaces that were in direct contact with the surrounding air. 6

7 5) Between the hot arms and the flap there was a small air gap of about 0.75µm. The air conduction across this gap was significant, and was modeled as convective to avoid mesh difficulties, with a coefficient derived as 5.2x10 4 w/m 2 k from equation (1): h K d 0.039w / m k 4 = = w / m m air = 2 k (1) Where Kair is thermal conductivity of air, and d is the gap width. The material properties used in the simulation were those of polysilicon. Among these, the most important was the value of electrical conductivity, given by Butler [14], as this was affected by the application of electric currents that generated heat. The temperature-dependent electrical properties of polysilicon used in the simulation are listed in Table 1-3, along with some of the other physical properties [15-18]. The simulation results obtained were very encouraging. Figures 9 and 10 show the comparison of the temperature distributions between the old and the new flap designs. The maximum temperature and the length of the high-temperature regions achieved in the hot arms remained approximately unchanged. Such temperature distributions indicate that the widened hot arms had similar thermal expansions as in the first-generation flap designs. This verified the design choice to reduce the number of hot arms. With thermal expansions similar to those described in the previous paragraph, the maximum tip deflection was improved from 6.4 to 11.1 µm with 5V applied (see Figures 11 and 12). As shown in Figure 12, the flap region beyond the tips of the hot arms in the new design extended the deflection about 5 µm, which is the difference between the displacements of 11 and 6 µm. In addition to the large deflection, the distribution of the deflection became more uniform along the length of the flap, varying by 0.3 µm. 7

8 In contrast, the displacements shown in Figure 11 are non-uniform, with large deflections occurring in the center region of the flap. This warped deflection pattern was the result of the pulling forces from the hot arms. Such non-uniformity was reduced by moving the tips away from the edge of the flap and changing the distribution of the pushing forces from the hot arms. With the new design, a large portion of the deflection was related to the angle at the flap tip, rather than the displacements of that tip. Another concern is the fragility of the MEMS assembly. As shown in Figures 13 and 14, with enlarged cross-sectional area of the hot arms, the maximum von Mises stress of the improved design was only 500 Mpa. This was much smaller than 2000 Mpa of the previous design. Maximum stress occurred at the junction between the solder pads and the support flexures. Because stress solutions were meshdependent, we used the same mesh sizes in both simulations in order to make the comparison reasonable. The new design reduced the stresses significantly, which translates to improved reliability and durability. Given these encouraging numerical simulation results, we submitted the new design for fabrication using the commercial MUMPS process [19], and then assembled them for further characterization and experimental measurements. IV. ASSEMBLY Figure 15 shows the schematic diagram and an SEM image of one of the solder-assembled MEMS flaps from Figure 4. The fabrication and assembly processes were similar to the flip-chip processes reported in previous studies [20], with one major difference: the solder joints, with a pad size of 320 µm x 200 µm, were much larger than in any previous study. As a result, a special technique had to be developed to solder a large size MEMS flap onto the ceramic substrate. 8

9 The MUMPS devices came with gold pads after fabrication. Indium or indium/lead solder alloys were evaluated first due to their compatibility with gold pads, but their re-flow properties were very poor because of gold-indiumide formation associated with high gold-contents. For effective re-flow, 63Sn/37Pb soldering was chosen. However, Sn/Pb solder has the well-known gold scavenge effect. This effect was managed through a controlled procedure listed below: A. Before flip-chip bonding 1. Clean the ceramic substrate using acetone and methanol. Then put the substrate in argon plasma cleaner for 5 minutes at 25 W. 2. Place ceramic substrate on heating stage at room temperature. 3. Use micro-positioning probes to position solder balls on gold electrode pattern etched onto the ceramic substrate. 4. Enclose the heating stage and pass N 2 over the substrate. 5. Commence heating to 193 degrees centigrade. 6. Introduce formic acid vapor to the N 2 when the temperature reaches 160 degrees centigrade. 7. At 183 degrees centigrade the solder balls begin to melt, wetting the electrodes. Turn off both the heating element and the formic acid vapor, and increase the flow of N Wait for the substrate to reach room temperature before removing the ceramic substrate from the heating chamber. B. After flip-chip bonding 9. Place the assembly in argon plasma cleaner for 5 minutes at 25 W. 10. Repeat steps 2 to 6. 9

10 11. When the temperature reaches 183 degrees centigrade, turn off the formic acid vapor, leave for 3 minutes, then increase the nitrogen flow rate and turn off the heating element. 12. When the substrate is at room temperature, turn off the N 2 and remove the substrate from the heating chamber. V. CHARACTERIZATION The performance of the assembled MEMS flaps was investigated using a ZYGO interferometer microscope that is capable of measuring vertical deflections of the flaps. For comparison, the measured and simulated deflections of the MEMS flap are presented in graphical form in Figure 16 (a). The deflection increased from approximately 0.5 µm to 12 µm as the voltage was increased from 1 to 5 volts. The agreement between experiment and simulation is excellent, confirming both the accuracy of the model and the effectiveness of the design: the amplitude of deflection as a function of voltage is significantly improved over the first-generation MEMS flaps. This new design for the electro-thermal MEMS actuators is more than capable of providing the required disturbance for the jet flow experiment. The dynamic behavior of the MEMS flaps is presented in Figure 16 (b). This was obtained by reflecting a HeNe laser beam bounced off the free end of a vibrating flap and tracking the displacement of the reflected beam. The resonant frequency of a flap is around 25 Hz, as shown in Figure 16 (b). Above this value, the amplitude of oscillation decreased significantly with increasing frequency to a value of about 4 µm at 200 Hz. Using these improved MEMS flaps, we have been able to selectively excite the natural instabilities of the planar air jet. As illustrated in Figure 1(b), filtered air entered the base of a plenum, whose internal cross-section was 400mm by 15mm, and passed through three flow treatment screens constructed using 10

11 24 mesh with 0.19 mm diameter wire. Upon reaching the top, the air entered a matched cubic contraction of area ratio 6 to 1, to further reduce turbulence and produce a more uniform flow. Flow emerged from an exit slit 400mm long and 2.5±0.01mm wide. The system was highly sensitive to mechanical vibrations and motion of the ambient air. To minimize these effects, the entire experiment stood on vibration control mounts and was semi-enclosed by a 1m 3 plexiglas box. External forcing of the jet was provided by arrays of the MEMS flap assembly described in the previous sections, positioned on either side of the exit slit. A TSI 1210 hotwire probe and a TSI 1010 hotwire anemometer were used to measure the velocity profile and velocity fluctuations in the jet. The probe was mounted on translation stages and could be moved in the streamwise and cross-stream directions with an accuracy of 0.025mm. The hotwire was operated by a constant temperature method, and its output digitized using a 12-bit A/D converter. First, the centerline velocity was measured along the length of the exit slit, and was found to be uniform to within 2%, supporting our assumption of a 2D flow near the exit slit. Thereafter, hotwire measurements of both mean and fluctuating velocity profiles were acquired across the center of the exit slit. Compare to the unforced flow, the amplitudes of disturbances forced by the MEMS flaps can be 7~50 times their natural levels. Figure 17 compares the forced (17b) and unforced flow (17a), clearly showing the effects of these improved flaps. VI. CONCLUSIONS The first-generation actuators developed for a feasibility study of MEMS-based flow control had three major problems: incomplete release, broken hot arms, and small deflections. To improve on these flaps, we have reduced the size of solder pads, removed sharp corners, changed the number and the width of the actuator s hot arms, and strengthened the support beams. In addition, we have developed a 11

12 controllable process to re-flow tin/lead solder on gold pads. Numerical simulations using CoventorWare were used to characterize the performance of the new design prior to fabrication. Simulation results indicated that the new design increased flap-tip deflection from 6.5 to 11.1 µm for a 5 V driving voltage, and reduced the maximum von Mises stress from 2000 to 500 Mpa. The simulated deflection-.vs.- voltage relationship has been confirmed by experimental characterizations. Experimental results show that forcing the instabilities of the planar jet with these flaps increases the amplitudes of those instabilities by a factor of 7~50. Ongoing studies include the details of flow around the flaps, and the development of the instabilities in the jet. Acknowledgments This work is supported by the National Science Foundation with award number ACI The authors would like to express their appreciation for the support of Coventor, Inc., and helpful discussions with W. Zhang, P. Kladitis, K. Harsh, J. Negretti, and J. P. Giardino at the University of Colorado. References [1] J. M. Wiltse and A. Glezer, Manipulation of free shear flows using piezoelectric actuators, J. Fluid Mech., vol. 249, pp , [2] J. M. Wiltse and A. Glezer, Vectoring and small-scale motions in free shear flows, Phys. Fluids, vol.10, pp , [3] S. A. Jacobson, and W. C. Reynolds, Active control of stream wise vortices and steaks in boundary layers, J. Fluid Mech., vol. 360, pp ,

13 [4] C-M. Ho, and Y-C. Tai, Micro-electro-mechanical-systems (MEMS) and fluid flows, Ann. Rev. Fluid Mech., vol. 30, pp , [5] Z. Ma, T. Peacock, E. Bradley, and Y-C. Lee, Solder-assembled MEMS flaps to enhance fluid mixing, in Proc. of ASME IMECE (International Mechanical Engineering Congress and Exposition), 2001, pp [6] T. Tsao, C. Liu, Y-C. Tai, and C-M. Ho, Micromachined magnetic actuators for active fluid control, Application of Microfabrication to Fluid Mechanics, ASME, vol. 197, pp ,1994. [7] L. Löfdahl, and M. Gad-el-Hak, MEMS applications in turbulence and flow control, Prog. Aero. Sci., vol. 35, pp , [8] R. Linderman, Development and parametric analysis of the resonant micro fan for fluidic transport and suspended particle collection, Ph.D. dissertation, University of Colorado, [9] E. N. da C. Andrade, The velocity distribution in a liquid-into-liquid jet. Part 2: The plane jet, in Proc. Phys. Soc., 1939, pp [10] G. B. Brown, On vortex motion in gaseous jets and the origin of their sensitivity to sound, in Proc. Phys. Soc., 1935, pp [11] T. Peacock, E. Bradley, J. Hertzberg, and Y. C. Lee, Forcing a planar jet flow using MEMS, Phys. Fluids, to be published. [12] W. Cowan and V. M. Bright, Vertical thermal actuators for micro-opto-electro-mechanical systems, in Proc. SPIE, 1997, vol. 3226, pp [13] Coventor Inc., NC. [Online]. Available: [14] J. T. Butler, Development and packaging of microsystems using foundry services, Ph. D. Dissertation, Air Force Institute of Technology, Jun

14 [15] O. M. Paul, J. Korvink, and H. Balts, Determination of the thermal conductivity of CMOS IC polysilicon, Sensors and Actuators, vol , pp , [16] P. J. Timans, Emissivity of silicon at elevated temperatures, Journal of Applied Phys., vol. 74, pp , [17] W. Riethmuller and W. Benecke, Thermally excited silicon microactuators, IEEE Trans. on Electron Devices, vol. 35, no. 6, pp , [18] Y. Okada and Y. Tokumaru, Precise determination of lattice parameter and thermal expansion coefficient of silicon between 300 and 1500 K, Journal of Applied Phys., vol.56, no.2, pp , [19] JDS Uniphase Corp., CA. MUMPs Process, [Online]. Available: [20] K. Harsh, B. Su, W. Zhang, V. M. Bright, and Y-C. Lee, The realization and design considerations of a flip-chip integrated MEMS tunable capacitor, Sensors and Actuators A, vol. 80, pp , Biographies Zhichun Ma received the B.S. and M.S. degrees in Mechanical Engineering from Tsinghua University, China in 1988 and 1990 respectively. He worked in industry for 10 years before coming to the University of Colorado, Boulder for his Ph.D. study. He received his M.S. degree from the University of Colorado in He is currently working toward his Ph.D. in Mechanical Engineering, University of Colorado. His research interests include: advanced electronics packaging, the design and development of micro-electro-mechanical systems (MEMS), finite element modeling (FEM), and reliability evaluation. Zhichun Ma is currently a research assistant in the Center for Advanced Manufacturing and Packaging for Microwave, Optical, and Digital Electronics (CAMPmode). 14

15 Elizabeth Bradley received the S.B., S.M., and Ph.D. degrees from the Massachusetts Institute of Technology in 1983, 1986, and 1992, respectively, including a one-year leave of absence to compete in the 1988 Olympic Games, and has been with the Department of Computer Science at the University of Colorado at Boulder since January of Her research interests include nonlinear dynamics, control theory, scientific computation, and artificial intelligence. Bradley is a member of the external faculty of the Santa Fe Institute and the recipient of a National Young Investigator award, a Packard Fellowship, and the 1999 University of Colorado College of Engineering "innovation in teaching" award. Thomas Peacock received the B.S. degree from Manchester University in 1993, and the Ph.D. degree from Oxford University in Now he teaches nonlinear dynamics II: continuum mechanics at Applied Mathematics at M.I.T.. His research and teaching interests concern experimental, numerical and theoretical fluid dynamics. Present and previous research projects include micro-active control of a turbulent jet and the transition to turbulence in a liquid crystal. Jean Hertzberg has been an Associate Professor of Mechanical Engineering at the University of Colorado since January Her research interests center around pulsatile, vortex dominated flows with applications in both combustion and cardiac fluid dynamics. Three dimensional flows and passive and active flow control techniques are of particular interest. Dr. Hertzberg teaches graduate and undergraduate courses in measurement techniques, fluid mechanics, thermodynamics, heat transfer, design and numerical methods. Y. C. Lee is a Professor of Mechanical Engineering and a faculty member of the NSF Center for Advanced Manufacturing and Packaging of Microwave, Optical and Digital Electronics (CAMPmode), University of Colorado at Boulder. Prior to joining the University in 1989, he was a Member of Technical Staff at AT&T Bell Laboratories, Murray Hill, New Jersey. Professor Lee s research activities include low-cost prototyping and thermal management of multichip modules, 3-D packaging, self- 15

16 aligning soldering, fluxless or solderless flip-chip connections, optoelectronics packaging, process control using fuzzy-logic models, microelectromechanical systems and protein integration for microsystems. Dr. Lee received the following awards: 1) Presidential Young Investigator (National Science Foundation, 1990), 2) Outstanding Young Manufacturing Engineer Award (SME, 1992). Dr. Lee was the General Chair of ASME InterPACK'01. He is the Associate Editor of ASME Journal of Electronic Packaging and a Guest Editor for IEEE CPMT Advanced Packaging: Special Issue on MEMS/NEMS Packaging. Dr. Lee edited two books on Optoelectronic Packaging (John Wiley and Sons, 1997) and Manufacturing Challenges in Electronic Packaging (Chapman & Hall, 1997). 16

17 Figure 1: Schematic of a planar air jet. (a) Cross-section. Flaps are placed on either side of the jet, providing a transverse disturbance that amplifies as it travels downstream. (b) The sideview. Figure 2: Images of the jet. (a) Without forcing, the flow is laminar and featureless. (b) With 6Hz forcing, large-scale vortices are generated, entraining air into the jet. The scale of these images is approximately 10cm wide and 15cm high. Figure 3: Schematic of the first generation MEMS flap, bottom view. Figure 4: 1-cm, 10-flap array of MEMS assembly needed for a two-dimensional jet, top view. Figure 5: Improved design with multiple solder pads, bottom view. Figure 6: Improved design with widened hot arms, top view. Figure 7: SEM images of the first-generation MEMS flap. (a) Bottom view (referenced to the final assembly position). (b) the sharp corner between the hot arm and support was structurally unsound. Figure 8: SEM of improved design. (a) Bottom view of the structure view (referenced to the final assembly position). (b) Widened hot arm without the sharp corner. This was accomplished by placing the hot arm below the flap plate. Figure 9: Configuration and temperature contour of the first-generation design with 5 V applied. Flaps viewed from below (referenced to the final assembly position); resultant deflection is upward. Figure 10: Configuration and temperature contour of the improved design with 5 V applied. Flaps viewed from above (referenced to the final assembly position); resultant deflection is downward. Figure 11: Displacement contour of the first-generation design with 5 V applied. Figure 12: Displacement contour of the improved design with 5 V applied. Figure 13: Von Mises stress contour of the first-generation design with 5 V applied. 17

18 Figure 14: Von Mises stress contour of the improved design with 5 V applied. Figure 15: View of one of the solder-assembled MEMS flaps, taken from Figure 4. (a) Schematic view. (b) SEM image. Figure 16: The response characteristics of the MEMS flaps. (a) Measured and simulated tip deflection as a function of applied voltage at steady state. (b) Measured deflection vs frequency for an applied a.c. voltage of 5 volts. Above 25 Hz, the amplitude of displacement decreases significantly. Figure 17: The velocity profile of the jet. (a) Unforced. (b) 4Hz asymmetric forcing. Table 1. Co-efficient of thermal expansion of polysilicon. Table 2. Electrical conductivity of polysilicon. Table 3. Other properties of polysilicon. 18

19 Contraction Air jet exits Exit slit Instabilities are sensitive here Plenum MEMS flaps Air (a) Air (b) Figure 1 : Schematic of a planar air jet. (a) Cross-section. Flaps are placed on either side of the jet, providing a transverse disturbance that amplifies as it travels downstream. (b) Side view. (a) (b) Figure 2: Images of the jet. (a) Without forcing, the flow is laminar and featureless. (b) With 6Hz forcing, large-scale vortices are generated, entraining air into the jet. The scale of these images is approximately 10cm wide and 15cm high. 19

20 Flap Hot arms Support Solder Pads Figure 3: Schematic of the first generation MEMS flap, bottom view. Flaps Solder pads Ceramic substrate 1mm 1 cm Figure 4: 1-cm, 10-flap array of MEMS assembly needed for a two-dimensional jet, top view. 20

21 Flap Hot arms Support Solder Pads Figure 5: Improved design with multiple solder pads, bottom view. Etch hole Via Flap Hot arms Figure 6: Improved design with widened hot arms, top view. 21

22 Flap Hot arms (b) Sharp corner Support Solder pads (a) Hot arm Sharp corner Support (b) Figure 7: SEM images of the first generation MEMS flap. (a) Bottom view (referenced to the final assembly position). (b) the sharp corner between the hot arm and support was structurally unsound. 22

23 Flap (b) Improved structure of hot arms Support Solder pads (a) Flap Hot arm without the sharp corner (b) Figure 8: SEM of improved design. (a) Bottom view of the structure view (referenced to the final assembly position). (b) Widened hot arm without the sharp corner. This was accomplished by placing the hot arm below the flap plate. 23

24 Figure 9: Configuration and temperature contour of the first generation design with 5 V applied, flaps viewed from below (referenced to the final assembly position), resultant deflection is upward. Figure 10: Configuration and temperature contour of the improved design with 5 V applied, flaps viewed from above (referenced to the final assembly position), resultant deflection is downward. 24

25 Figure 11: Displacement contour of the first generation design with 5 V applied. Figure 12: Displacement contour of the improved design with 5 V applied. 25

26 Figure 13: Von Mises stress contour of the first generation design with 5 V applied. Figure 14: Von Mises stress contour of the improved design with 5 V applied. 26

27 Support Hot arms Flap Solder Joints Ceramic substrates (a) Flap Hot arms Support Solder pads (b) Figure 15: View of one of the improved solder-assembled MEMS flaps, taken from Figure 4. (a) Schematic view. (b) SEM image. 27

28 Max. tip deflection (microns) measured Simulated Applied voltage (volts) (a) 14 Max. tip deflection (microns) Frequecy (hz) (b) Figure 16: The response characteristics of the MEMS flaps. (a) Measured and simulated tip deflection as a function of applied voltage at steady state. (b) Measured deflection vs frequency for an applied a.c. voltage of 5 volts. Above 25 Hz, the amplitude of displacement decreases significantly. 28

29 (a) (b) Figure 17: The velocity profile of the jet. (a) Unforced. (b) 4Hz asymmetric forcing. 29

30 Table 1. Co-efficient of thermal expansion of polysilicon Temperature Co-efficient of thermal expansion (K) (1/K) x x x x x x x x x x x x x

31 Table 2. Electrical conductivity of polysilicon Temperature Electrical conductivity (K) (pa 2. s 4 /Kg-µm 3 ) x x x x x x x x x x x x x Table 3. Other properties of polysilicon Property Value Young s modulus 169 Gpa Thermal conductivity 30 W/m-K Poison ratio 0.22 Specific heat 754 J/Kg-K 31