Abstract :
[en] The vast majority of micro-electro-mechanical systems (MEMS) devices rely in their mechanical design on very simple geometrical layouts comprising only a few primitives such as beams, rectangular masses and – more rarely – rings or disk-shaped structures. The beams and rectangular structures are typically combined in a Manhattan grid structure to form a suspension system, resonating structures, the proof mass of inertial sensors, comb drives, pick-off capacitive electrodes, etc. However, practically all MEMS devices having movable structures (physical sensors such as inertial sensors, force sensors, pressure sensors, actuators, as well as biochemical sensors based on resonators or beams exhibiting stress induced static deflections) are compliant devices. In other words, they don’t rely on linear or rotational joints as commonly used in macroscopic mechanical designs. However, for compliant devices, orthogonal designs based only on the combination of rectangles may not be the best design approach. Depending on the parameter of interest for a particular device, much more complex geometrical shapes may result in superior performance. Parameters to be maximized could be the deflection of a proof mass in response to an input (e.g. inertial) force, the linear travel range of an electrostatic actuator for a given input voltage, the force in response to an electrostatic drive (e.g. for a micro-gripper), the pull-in voltage, the linearity of a mechanical suspension system, etc. Such a design approach requires substantially higher degrees of freedom for the shapes of the mechanical geometry, rather than restricting it to very few simple primitives. Using simple geometrical shapes is a reminiscence of early MEMS design: the designer had to intuitively understand the design and being able to model it analytically using linear approximations. With modern design tools such as MEMS+, this requirement is no longer a limiting constraint, as these tools can predict the behavior (including nonlinear effects) of virtually arbitrarily complex shapes and designs by multi-physics FEM (finite element modelling) simulations. This offers the exciting possibility for geometrical shape optimization for certain goal parameters for a wide range of MEMS devices.
The presented work aims to introduce a novel and powerful design methodology for a wide range of MEMS devices relying on compliant mechanism. The proposed optimization relies on a GA that has already been successfully employed for optimizing the electronic parameters of closed-loop, electro-mechanical sigma-delta modulator control systems for inertial sensors [1]. The primary objective of this thesis is to apply, for the first time, a GA to MEMS devices for freeform geometries. The GA based algorithm running within MEMS+ is capable of quickly and efficiently designing high performance, near optimal freeform compliant mechanisms for many MEMS devices. The semi-automated design procedure is expected to result in rather unusual geometrical shapes that have not been seen in MEMS designs to date and is therefore likely to establish an entirely new research area for MEMS. It is expected that important performance indicators for many MEMS devices will be substantially improved.
In the thesis, firstly, a the novel, semi-automated design methodology based on a genetic algorithm (GA) using freeform geometries for MEMS devices is introduced. A detailed description of the optimization process is presented including, mechanical model building, figure of merit, robustness analysis.
As a first demonstrator, a MEMS a MEMS accelerometer comprising a mechanical motion amplifier, is presented to validate the effectiveness of the design approach. Experimental results show an improvement of the product of sensitivity and bandwidth by 100% and a sensitivity improvement by 141% compared to a device designed in a conventional way. Furthermore, excellent immunities to fabrication tolerance and parameter mismatch were achieved.
As a second demonstrator an actuator, specifically a MEMS microgripper, is designed with this novel methodology. The use of freeform geometries significantly improved the performance of the microgripper. Experimental data shows that the designed microgripper has a large displacement ( 91.5 μm) with a low actuation voltage (47.5 V), which agrees well with theoretical and simulation results. The microgripper has a large actuation displacement and can handle micro-objects with a size from 10 μm to 100 μm. A grasping experiment on human hair with a diameter of 77 μm was performed to prove the functionality of the gripper. The result confirmed the superior performance of the new design methodology enabling freeform geometries.
