The starting point of this research is motivated by the potential uses of microcantilever arrays for highly parallel scanning probe microscopy and applications such as ultra high density data storage (e.g. the IBM Millipede project). There are two major issues confronting the designer of an arrayed MEMS device

• Dynamical coupling between neighboring devices, this is a particularly acute problem in capacitively actuated devices due to fringe fields.
• Measurements of mechanical variables such as displacement. The most commonly used scheme for AFM, optical levers, is not yet easily scalable to large arrays.

We have designed, built and tested arrays of capacitively actuated microcantilevers to directly address these issues. The picture shown is representative of those arrays. They are tightly packed movable plates forming capacitors with the bottom, rigid, individually addressable plates. There's a large amount of mechanical and electrostatic coupling between neighboring plates. We are using these devices for a "proof of concept" of two new ideas that address the above two questions:

• Use feedback from displacement measurements of each cantilever to electronically compensate for mechanical and electrostatic coupling. This means building an array of coupled feedback controllers to achieve desired properties such as decoupling or perhaps other desired "global behavior". This has the effect of simplifying the mechanical design of the MEMS device, with the additional complexity in designing the feedback controllers.
• To estimate each cantilever displacement, we use the current through each capacitor as an output (the input is the command voltage), and build a dynamical observer to estimate displacement. In some sense, the common scheme of measuring capacitence using a parasitic high frequency signal is a special case of this observer based scheme. We have been able to demonstrate the ability to esitmate displacement without the use of parasitic high frequency signals.