Design and Implementation of a Bistable Switch Using MUMPs Technology

Transcription

1 Engineering Science 494 Final Report Design and Implementation of a Bistable Switch Using MUMPs Technology David Lee Jeffrey Robinson Michael Sjoerdsma Matthew Ward

2 Table of Contents LIST OF FIGURES... III LIST OF TABLES... III OBJECTIVE...1 BISTABLE SWITCHES...1 MEMS TECHNOLOGY...1 SOLUTION...2 GENESIS...2 DESIGN EVOLUTION...3 FINAL DESIGN AND ANALYSIS...5 COMPONENT IMPLEMENTATION...6 RATCHET WHEEL...6 Required Functionality...6 Dimensions...7 RATCHETING SYSTEM...8 Required Functionality...8 Calculations...8 Dimensions...10 Potential Issues...10 HEATUATOR...11 Required Functionality...11 Calculations...11 Dimensions...12 Potential Issues...12 SPRING...13 Required Functionality...13 Calculations...13 Dimensions...14 Potential Issues...14 LEVER...15 Functional Requirements...15 Calculations...16 Dimensions...18 Potential Issues...19 SUMMARY...19 II

3 List of Figures FIGURE 1: MOVEMENT OF PLATFORM...2 FIGURE 2: PRELIMINARY DESIGN OF THE BISTABLE SWITCH...3 FIGURE 3: GEAR DESIGNS...4 FIGURE 4: FINAL GEAR DESIGN...5 FIGURE 5: FINAL CONCEPTUAL DESIGN...5 FIGURE 6: DESIGN 1 OF THE RATCHET WHEEL...7 FIGURE 7: DESIGN 3 OF THE RATCHET WHEEL...7 FIGURE 8: TYPICAL PAWL DESIGN...8 FIGURE 9: A HEATUATOR...11 FIGURE 10: A SPRING...13 FIGURE 11: THE GEAR ARM LEVER DESIGN...15 FIGURE 12: THE FULCRUM ARM LEVER DESIGN...15 FIGURE 13: FIGURE TO AID IN OVERLAP CALCULATION...17 List of Tables TABLE 1: COMPARISON OF RATCHET WHEEL STATISTICS...7 TABLE 2: DIMENSIONS AND CHARACTERISTICS OF PAWLS FOR EACH DESIGN...10 TABLE 3: HEATUATOR STATISTICS FOR EACH DESIGN...12 TABLE 4: SPRING STATISTICS FOR EACH DESIGN...14 TABLE 5: LEVER DESIGN DIMENSIONS...18 III

4 Objective Our principal design objective is to implement a mechanical bistable switch using MicroElectroMechanical Systems (MEMS) technology. Bistable Switches A bistable switch is a switch that toggles between two stable states in response to momentary applications of force. In other words, the switch can provide two stable output levels using a single input action that requires no knowledge of the prior switch state. The property of needing only momentary actuation makes bistable switches useful in many applications. A classic example of a mechanical bistable switch application is that of a moving platform upon which two sets of gears rest. As a bridge between input and output gears, one gear set can make the output gear turn counter to the input gear and the other set can make the output gear turn in the same direction as the input gear. A platform activated by a bistable switch would be able to toggle which gear set is engaged and would not require any applied stimulus (i.e. voltage) once the gears had been engaged. Other possible applications for mechanical bistable switches include: Optical switches Mechanical frequency division Selective refresh digital displays MEMS Technology MEMS are the product of a convergence between miniaturization of electromechanical systems and integrated circuit fabrication techniques. The use of IC fabrication technology means that MEMS devices can designed using standard CAD software under technology specific constraints. This technological accessibility has made MEMS a practical technology for those wishing to miniaturize existing electromechanical systems. Cost, however, remains a significant impediment to prototyping MEMS devices. IC fabrication facilities require sophisticated equipment and stringent environmental controls, and hence are not widely available. A solution to the cost barrier is provided by the Microelectronics Center of North Carolina (MCNC). In conjunction with Defense Advanced Research Projects Agency (DARPA), MCNC developed Multi User MEMS Processes (MUMPs) to provide a multi user fabrication service. This service allows participants to purchase a 1 cm 2 area to create their design, for a relatively low, fixed cost.

5 MUMPs provides designers with a three-layer polysilicon surface micromachining fabrication process. Seven film layers used in the fabrication process: one low-stress silicon nitride isolation layer two sacrificial oxide layers three polysilicon layers one metal layer A total of eight photolithography levels are used to pattern these layers. MUMPs provides designers with a very accommodating design environment, however there are a number of significant contraints to design flexibility. These constraints are summarized in the design rules section of the MUMPs Design Handbook. For us, the most onerous design rule was the 2 µm feature spacing, especially since heatuators typically provide just 8 10µm of movement. Solution Genesis The criterion for success of our bistable switch design is to be able to move a platform as shown in Figure 1. Figure 1: Movement of Platform As described previously, the position of the platform determines the rotational direction of the output wheel by varying the number of idle gears between the input and output stages. The design of our bistable switch concentrates on creating two states such that a platform could be moved into one of two stable positions. For the purposes of a proof of concept design, we have simplified Figure 1 so that only the platform remains; that is, the gears have been removed to simplify fabrication. The implementation of the bistable switch is heavily dependent on the MUMPs technology being used to fabricate the system. Due to the many known limitations 2

6 and uncertain outcomes inherent in the process, we decided to keep potential designs similar to previously successful designs. Our initial inspiration for designing a bistable switch was to consult macro designs already present in everyday life. We investigated the construction of bistable switches used in electronic devices and ballpoint pens. The implementation of these device were not suitable for micro machining due to the three dimensional movement of these switches. Note that the micro machined designs are essentially two dimensional, thus movements occurs in only one plane. The initial design we considered for moving the platform was to use a cam that would produce two positions. The difficulty with this design was that the moving the cam would be difficult. Expanding on this idea, we decided that a gear with two different sized teeth would produce two distinct states. Additionally, the gear design would allow for a lever to move it between positions. The problem that now remained was how to drive the gear in only one direction. We decided to make a ratchet system the gear would be the ratchet wheel and we would need a pawl to restrict the movement to one direction of rotation. With the above considerations, we decided that our bistable switch would need the following components to move the platform: a ratchet wheel, a pawl, and a lever arm. Design Evolution Incorporating the design components we obtained a preliminary configuration for the bistable switch as shown in Figure 2. Figure 2: Preliminary Design of the Bistable Switch In the design above the ratchet wheel can only move clockwise because the pawl (right most arm) prevents the wheel from rotating in the other direction. The 3

7 unidirectional movement is accomplished due to the eccentric teeth of the wheel and the shape of the pawl. When moving clockwise the arced shape of the ratchet wheel and pawl allow movement between the two. However, if the ratchet wheel tries to move counterclockwise then, as a result of their shapes, the ratchet wheel and pawl will bind. Similarly, the lever arm (left most arm) will move the ratchet wheel clockwise because the pawl will provide no resistance. However, in the opposite direction the lever arm will flex around the ratchet wheel because of the restorative force from the spring. The configuration shown in Figure 2 changed significantly in order to adhere to the design rules required by the MUMPs technology. The pawl and lever arm remained functionally the same. That is, the lengths and orientations were modified after performing calculations, however, the ideas behind each component did not change significantly from a high level perspective. The major changes to the design were as a result of modifications to the ratchet wheel. The ratchet wheel in Figure 2 does not adhere to the design rule in MUMPs that forbids acute angles. We made modifications to the wheel as shown in Figure 3. Figure 3: Gear Designs The first two gears in Figure 3 (A and B) are a modification of the original ratchet wheel so that the gear teeth ends do not contain acute angles. The difference between gears A and B is the height differential between the short and long gears. Gear C in the figure is similar to gear B except that the acute angles where the gear teeth meet have been removed. Unfortunately, gear C in Figure 3 was also found to be unsuitable for our design. We had to reject this gear due to the MUMPs design rule that requires at least a two micron separation between components. The difficulty with gear C is that the lever arm and pawl have to be fabricated with this minimum spacing. Furthermore, the ratchet wheel s fabrication will also introduce play in the system. Because of the amount of play in the system, we were not confident that the lever arm and pawl would function properly. From a design model we constructed, we concluded that the lever arm and the pawl had a high probability of skipping over the smaller teeth. Even though we had to reject the small tooth large tooth gear design, we realized that some variation on a gear would be the optimal way to create a bistable switch. 4

8 We eventually decided that instead of varying the size of the tooth that we should vary the depth of the tooth. Figure 4 illustrates our final gear design. Figure 4: Final Gear Design Note that in this design the lever arm and pawl can travel in a like manner over each tooth regardless of depth. Final Design and Analysis The final design for our bistable switch is shown in Figure 5. Figure 5: Final Conceptual Design The design shown above is only a conceptual model of what we created in Cadence. The particulars of each component are discussed in detail in the following section. Due to the gear shape, the ratchet wheel will no longer move the platform when rotated by the lever arm. In Figure 2, the shape of the wheel would have provided force in the correct direction such that the wheel would have automatically moved the 5

9 platform. However, because of the deep-shallow tooth design, our platform requires a protruding bar that rests on the wheel in order to achieve the different heights required. Clearly, this configuration does not allow for the wheel to move freely because the protrusion of the platform will cause the system to bind. To alleviate this problem, the platform will have to be moved using a heatuator. The sequence of activation for our bistable switch is as follows: 1. Pull out the platform from the ratchet wheel. 2. Engage the lever. 3. Disengage the lever. 4. Release the platform. The major advantage to our design for the bistable switch is that it utilizes components from previously successful designs with slight modification. Essentially, our design is a ratchet systems except that the ratchet wheel has been modified so that every second tooth is deeper. Furthermore, our bistable switch is truly bistable in the sense that the same input is applied regardless of the current state. Finally, power need only be applied to the switch when a toggle is desired. Component Implementation Ratchet Wheel Required Functionality The central component of the bistable switch is the ratchet wheel. The two different between gear levels implemented the memory in our system. Many factors influenced its design. First, we wanted to come up with a design that was similar to those that had been successful in the past. Therefore, we decided to make a ratchet wheel that had a radius of approximately 50 microns. The depth of the teeth on the wheel was also very important. Past designs had shown that teeth with a depth of eight microns meshed very well. This would be the maximum tooth depth. From this point, we set the higher between-tooth position to be at a depth of four microns so that a platform resting between teeth would have a significant difference in position between states. Another consideration that must be taken into account is that structures could only be fabricated with a minimum feature spacing of two microns. We made sure to fabricate everything in the Off state. With a difference of four microns between the low and high platform position levels, we therefore have a two micron displacement between the manufactured position of the platform rod and the position of the platform rod when limited by the between-gear structure. The two micron spacing limitation, together with the two micron minimum feature size limitation, was a major design consideration. 6

10 Dimensions Two designs for the ratchet wheel were implemented, namely design 1 and design 3. Design 2 was never used. Table 1 shows a comparison of statistics for the two designs. Figure 6 and Figure 7 show layout results for each of the designs. Table 1: Comparison of Ratchet Wheel Statistics Design 1 Design 3 Number of On or Off States per Rotation of Ratchet Wheel 9 10 Total Number of Teeth Width of On Platform (µm) Width of Off Platform (µm) Depth of On State (µm) 4 6 Depth of Off State (µm) 8 9 Inner Radius of Wheel (µm) Figure 6: Design 1 of the Ratchet Wheel Figure 7: Design 3 of the Ratchet Wheel 7

11 Ratcheting System Required Functionality Our design utilizes a ratcheting system in conjunction with a ratchet wheel. A pawl is used to keep the wheel in place with each successive turn, causing the position or memory of the platform to be mechanically locked in place until further clockwise rotation of the gear. While allowing for clockwise rotation of the gear, the pawl must prevent counter-clockwise rotation. Figure 8 shows an image of a typical pawl used in our various designs. Figure 8: Typical Pawl Design Calculations Theory of Operation The pawl is essentially a single ended cantilever, which is designed to bend in only one direction. One end is anchored to a stationary ground layer. Clockwise rotation of the gear will cause the gear tooth to catch onto the lip of the cantilever and bend the cantilever. The cantilever will bend back sufficiently to allow for clearance of the gear tooth. Upon clearance of the gear tooth, the cantilever will snap back into place and lock the position of the gear. Counter-clockwise motion of the gear will cause the back of the gear tooth to push the cantilever in a direction parallel to its plane, where it cannot bend. This will prevent the gear from rotating in the wrong direction. Analysis Rotation of the ratchet wheel results in a tangential force, which is applied at an angle to the long side of the ratchet. It is the perpendicular component of this tangential 8

12 force, which is crucial to calculation and design of the cantilever. We obtained an estimate for this value by successively approximating the force transfer from component to component until the ratchet was reached. These approximations accounted for factors such as the force lost due to rotational friction, and force multiplication via torque. The flow of force transfer is listed below. 1. Force applied from heatuators to lever (10 µn per heatuator) 2. Force transferred from input to output end of lever (accounting for frictional losses from rotation, and force loss due to torque conversion) 3. Force transferred from lever output to edge of bistable gear (including rotational frictional losses) 4. Tangential force transferred from edge of gear onto cantilever (accounting for the perpendicular component of this force which acts onto the cantilever) 5. Perpendicular force (load) on end of cantilever The specific value for the perpendicular load on the cantilever varied from design to design, as the characteristics and dimensions of specific components in each design varied. Our approximation of this value included the consideration of: number of heatuators, force changes from lever/gear size, rotational friction losses, and forces applied to the cantilever at an angle. Including these considerations into our pawl design would help ensure a functional and reliable ratcheting system. The maximum deflection of a cantilever is given by 3 PL Wmax =, (1) 3EI where P is the load on the end of the beam (as deduced from the method described above. L is the length of the beam, E is Young s modulus for polysilicon, and I is the moment of interia of the beam. Also, the moment of inertia (I) for a beam is given by 3 wh I =, (2) 3 where w is the width of the cantilever, and h is the height. Upon insertion of (2) into (1), and re-arrangement, we can obtain the formula for L in terms of P, E, w, h, and W max to be W Ewh P 3 3 max L =. (3) With W max chosen as the height of the gear teeth, E known, P known from the above approximation, w arbitrarily chosen as 2µ, and h chosen as 3.5µ for double thickness construction (using a POLY2 layer deposited on a POLY1 layer), we can deduce a value for the length of the cantilever (L) that adheres to our needs. 9

13 Dimensions Table 2 outlines the specific dimensions and characteristics of the various pawls included in our final layout. Table 2: Dimensions and Characteristics of Pawls for Each Design Design Length (µm) Width (µm) Height (µm) Expected Load (P) (µn) D1H2GA D1H4GA D1H4GA D3H2FA D3H2FA D3H2FAT D3H2FATT D3H2FAW D3H2FAW D3H4FAW Potential Issues Max deflection (µm) The ratcheting system in the layout was designed to be as resilient as possible, while allowing the proper bending for gear tooth clearance. This requires stringent calculation of the beam dimensions based on the expected load on the beam. Unfortunately, our method for deducing this load involves an element of approximation, which may render the cantilever under-optimized. 10

14 Heatuator Required Functionality Electro-thermal actuators (heatuators) are used in our design in order to provide the forces necessary to rotate the bistable gear. Applying an appropriate voltage across their input terminals activates them. The heatuators are hinged to an appropriately designed lever arm, in order to rotate the bistable gear by the proper distance. Figure 9 is a sample image of a heatuator used in our layout. Figure 9: A Heatuator Calculations Theory of Operation The heatuator we used follows the conventional U-shape, where one arm of the U is thinner than the other. Operation of the heatuator is simple, when a voltage is applied between the terminals, a current flows through the device. However, since the two arms have different widths, the current density is unequal in the two arms. This leads to a different rate of Joule heating in the two arms, and thus to different amounts of thermal expansion, causing the heatuator to bend in the direction of the thick arm. A good way to characterize a heatuator s function is to consider it as a cantilever. A cantilever has a no load deflection based on some input parameter (like voltage or current). Its range of motion can be determined by 11

15 d = d o ( a) + CF, (4) where d is the deflection (length), d 0 is the no load deflection (length) dependant on some input parameter a, C is the compliance (length per force), and F is the applied force. Typically, d 0 (0) = 0. Standard Solution Our group decided to use a standard geometry for our heatuator, and then create the rest of our design based on the known behavior of the chosen geometry. The 200 micron length / micron flexure geometry has delivered consistent performance in the past. Mechanical simulation and empirical testing have shown that a heatuator of the above dimensions has a deflection d of 8 to 10 microns with a 5 Volt input. Also with this input, a single heatuator exhibits a force of about 10 micronewtons, with a compliance of about Using this proven geometry allows us to easily integrate heatuators into our design with predictable results. In cases where a higher applied force was needed, additional heatuators are used in parallel as inputs to the appropriate lever. Dimensions Table 3 lists the heatuator statistics for each design present in the final Cadence layout. Table 3: Heatuator Statistics for Each Design Design # of Heatuators Orientation D1H2GA 2 Dual in-line, 1 per column D1H4GA 4 Dual in-line, 2 per column D1H4GA2 4 Dual in-line, 2 per column D3H2FA 2 Dual in-line, 1 per column D3H2FA2 2 Dual in-line, 1 per column D3H2FAT 2 Dual in-line, 1 per column D3H2FATT 2 Dual in-line, 1 per column D3H2FAW 2 Dual in-line, 1 per column D3H2FAW2 2 Dual in-line, 1 per column D3H4FAW 4 Dual in-line, 2 per column Potential Issues Various factors in the fabrication of the heatuator may affect its functionality, and thus its desired range of motion and/or desired application of force may vary slightly from our predictions. This potential discrepancy is not within our control, and is generally minimal due to the stringent fabrication processes involved. 12

16 Spring Required Functionality Springs are used in some of our designs to ensure that heatuators and levers are restored to initial positions upon rotation of the ratchet wheel. One end of the spring was anchored to the wafer, and the other end attached to the lever such that the spring would be stretched upon deflection by the heatuator. An example of the spring we used is shown below in Figure 10. Figure 10: A Spring Calculations Theory of Operation The spring is implemented as a number of bars connected on alternating ends. As At the top and bottom of the spring is a "half-bar" so that the spring can be connected along its axis. As the ends of the spring are pulled apart the bars angle between adjacent bars increases and this deflection results in increased resistance against further stretching. Analysis The springs in our layout were designed to match the force of the opposing heatuator upon full deflection. This would cause the heatuator to move back to its relaxed state 13

17 upon reaching full deflection, and in turn, move the lever back. The compliance of the general form of a spring shown in Figure 10 given by 3 L N 1 C = +, (5) EI where C is the compliance of the spring, L is the length of the bars, E is Young s Modulus for the spring material, I is the 2nd moment of the bars, and N is the number of bars (excluding the two half-bars at the top and bottom of the spring). The compliance of the spring is related to its spring constant k by C = 1. (6) k Thus, we can apply Hooke s Law to match the forces of the spring and heatuator, d dei F heatuator = kd = =. (7) C 3 N 1 L The force of the heatuator (F heatuator ) is known, as are d, I, and E. If we arbitrarily select a value for L, we can deduce the number of bars (N) that would result in a spring matching the force of the heatuator. Dimensions Table 4 outlines the characteristics of each spring used in our layout. Note that not all designs included springs. Table 4: Spring Statistics for Each Design Design Compliance (m/n) Length (L) (µm) Number of Bars (N) D1H2GA D1H4GA D1H4GA D3H2FA Potential Issues Although we are able to predict the function of the spring accurately with the above formulae, various factors in the fabrication of the spring may affect its functionality. This potential discrepancy is not within our control, and is minimal due to the stringent fabrication processes involved. 14

18 Lever Functional Requirements The role of the lever is to use the input displacement from the heatuator assembly to rotate the gear through the angular displacement of a single tooth, such that the pawl will engage before the heatuator is fully displaced. The heatuator assembly assures us of an 8 µm input displacement. With this displacement the lever must move the gear through 18 degrees of angular displacement, plus an additional range of movement to allow the ratchet to engage. The lever must also be capable of returning to its original position after the heatuators have been deactivated. This entails the output lever arm sliding across a tooth of the gear held stationary by the ratchet. The two main lever designs incorporated in our layout are shown below as Figure 11 and Figure 12. Figure 11: The Gear Arm Lever Design Figure 12: The Fulcrum Arm Lever Design 15

19 Calculations Model Properties To calculate the required dimensions of the lever assembly, we first create a model to describe the system. A model is required, because the MUMPs process introduces numerous extraneous parameters (mostly due to feature spacing) that render an analysis of the true geometry impracticable. Our model is base on the following premises: The lever rotates about a single fulcrum The output contact of the lever starts 2 µm away from the tooth contact surface The ratchet contact is fabricated 2 µm away from the tooth contact surface Resistance to forward (clockwise) motion of the gear is negligible Once engaged by the pawl, resistance to the reverse (counter clockwise) motion of the gear is very large The gear rotates about a single contact point The gear teeth have been designed to accommodate any insertion of the lever contact required Arbitrary output lever geometries may be modeled as a straight segment between fulcrum and output contact for purposes of geometric calculations Input / Output Arm Relationship The most important property of the lever is that is provides a mechanism for trading force for displacement. The relationship between the input and output displacements of a lever is r input θ = r θ (8) input output Required Angular Displacement of Gear In order to change the state of the switch the gear must be advanced by 18 degrees in a clockwise direction. Also, the gear must be advanced such that the ratchet is able to engage a gear tooth. This requires that the rotation of the gear overcome the 2 µm feature spacing between the ratchet and gear tooth. Further, the output contact of the lever must be fabricated 2 µm away from contact with the gear tooth. Using the simple relationship relating arc length, radius and angle of displacement (in radians), output α = r θ, (9) and approximating length of the arc to be equal to the 2 µm feature separation, we are able to determine the total angular displacement of the gear required. 16

20 The total angular displacement required, in degrees, is thus θ req = , (10) r π where r gear is specified in micrometers. Minimum Overlap of Contacts Knowing the required angular displacement of the gear at full lever displacement, we can find the minimum insertion of the lever contact into the gear teeth such that the lever arm maintains contact with the gear through the required range of motion. Given the length of the output lever arm and the radius of the gear (center to tooth tip) we can construct two right triangles as shown in Figure 13. gear L h r θ Y 1 Y 2 Figure 13: Figure to Aid in Overlap Calculation Using Figure 13 to deduce variable definitions, the overlap required at fabrication can be calculated by h = r sinθ, (11) y 2 = r cosθ, (12) y l h =, and (13) 2 2 ( y 1 y ) overlap = l + r +. (14) 2 Beam Bending on Return to Natural Position After the heatuator is deactivated, the ratchet locks the gear in position, and the lever arm must return to its original position. In order for this to occur, the lever arm must 17

21 be able to slide around the gear tooth. An examination of the lever geometry yields the immediate conclusion that the lever arm must flex in order for this to occur. In some lever designs, the only consideration given to this requirement is that we make the lever arm as thin as possible. In other cases we have experimented with different output arm geometries in order to capitalize on the orthogonality of the forces (tangential force applied by the lever, radial contact displacement in return path). Through application of the beam bending (cantilever) formulas mentioned above for our ratcheting system, we were able to deduce the optimal beam dimensions to ensure proper flex. In order to accomplish this without interfering with the proper gear displacements, we varied the thickness of the beams, rather that vary its length. The various dimensions for our lever systems are listed in the following section. Dimensions In our bistable switch designs we used variations on two different lever designs. In the first design the input and output arms are fixed to a disc that rotates about a pin. This design is straightforward and relies on a predictable mechanism for operation. The second design uses an anchored tab, joined to a straight bar at a narrow junction. When an input force is applied to the input side of the bar, the joint flexes and the output side of the bar moves opposite to the direction of the input force. To accommodate the lever moving back over the next tooth, either a straight or bent flexing section was added to the lever as an interface to the ratchet wheel. Table 5 shows statistics relating to the design of the lever section of each of the ten designs in our final submission. Table 5: Lever Design Dimensions Design Gear radius (µm) Required Angular Displacement Contact Insertion (µm) Input arm length (µm) Output arm length (µm) Lever Mechanism D1H2GA o Pin rotation D1H4GA o Pin rotation D1H4GA o Pin rotation D3H2FA o Straight anchored tab D3H2FA o Straight anchored tab D3H2FAT o Straight anchored tab D3H2FATT o Straight anchored tab D3H2FAW o Bent anchored tab D3H2FAW o Bent anchored tab D3H4FAW o Bent anchored tab 18

22 Potential Issues Though we expect our analysis to yield a functional design, it is possible that some features of our model have been overlooked. In the first design, the feature spacing required between the anchor pin and the rotating disc facilitates some measure of displacement in, and hence a reduction the rotation of the lever. Since our design dimensions are near the scale of the feature spacing (as dictated by the limitations of the heatuators), relatively small deviations from the model can be catastrophic. In the second design the joined tab mechanism will suffer some displacement. Since we were unable to quantify this displacement (owing to the irregular geometry of the tab) we cannot be sure that our model adequately represents reality. Further, both designs suffer may suffer from displacement of the gear itself, since it too must adhere to feature spacing design rules. Summary This report has outlined the goals, design solution, and MUMPs implementation of a bistable switch. From our objective of creating a truly bistable switch, that requires state independent activation and provides a useful bistable output, ten unique designs have emerged. Based on a common concept, each design illustrates different approaches to shared problems. Wherever possible, we have used proven component designs and incorporated reasonable error margins. We are hopeful that our careful analysis and use of proven MUMPs designs have conquered the uncertainties and constraints of the MUMPs technology. Regardless of the outcome, we have realized an appreciation for the issues involved in MEMS design; even for the relatively simple structure of a bistable switch. 19