Characterizing a rapid thermal annealing process to improve sensor yield

in the production of acceleration sensors, surface micromachining requires the integration of traditional semiconductor processing with a sound, thorough understanding of material properties and mechanics. Basically, these silicon-based precision sensors convert a raw physical input signal into a usable electrical output signal, which can be fed into various control circuits. The linking of the mechanical input signals to electrical output signals is accomplished through resistance and capacitance changes of surface micromachined materials as they undergo mechanical stressing. In the acceleration sensor manufactured by Motorola's MEMS-1 die manufacturing group, three polysilicon plates are fabricated parallel to one another to form a dual, differential capacitive element (see Figure 1). The top and bottom plates are stationary, while the middle plate can move in the direction of the applied force. This center plate is suspended in space by tethers that are under tensile stress in order to prevent buckling. As the center plate moves relative to the top and bottom plates, the capacitance of the adjacent plates changes according to the standard capacitor equation represents the gap between the plates. Frequently, the movement of the sensing elements is modeled as a spring and dashpot mechanism: that is, there is a resistance to displacement from an applied force (spring) and a resistance to the return from that displacement (dashpot). The primary parameter for the resistance to displacement is the traditional spring constant ( ), while the relative difference in capacitance between the two capacitor plates represents the initial offset ( ) in the direction of the applied force. Optimally, the sensor will have zero initial offset and a spring constant that maximizes signal strength while minimizing noise (if the spring constant is too low, the device will be noisy; if it is too high, the device signal will be too small). Figure 1: Schematic of