Micro electromechanical systems (MEMS) provide many devices in sub-millimeter size for sensing and actuation. However, the lack of size-compatible power supplies prohibits entire systems to be within the same scale. The same problem of battery scaling exists for micro electronic devices. Reported in this dissertation is a novel way of micro power generation with radioisotopes. Due to the high energy densities and long half-lives of selected radioisotopes, high energy density power sources with extremely long operation time are possible.

Conversion of direct charge collection to mechanical actuation is the main achievement. A cantilever with a conductive collector collects the emitted electrons from a Ni-63 beta source. Due to charge conservation, positive charges are left in the radioactive source. The resulting electrostatic force moves the cantilever toward the source. When the cantilever contacts the source, charges are neutralized and the spring force pulls the cantilever back to its initial position. This cycle repeats itself as long as the radioactive source is active. Therefore a self-reciprocating cantilever is realized. An electromechanical model is developed to characterize the cantilever and verified with experimental results. The factors that limit the energy conversion efficiency are discussed. Further, radio frequency (RF) pulse generation at the end of the reciprocation cycle is achieved using a dielectric cantilever with metal electrodes, due to the excitation of dielectric waveguide mode. This RF pulse could be used for self-powered remote sensing and wireless communication. To generate electricity, a piezoelectric unimorph replaces the cantilever. At the end of the reciprocation, the sudden release of the unimorph excites its mechanical vibration, thereby generating electricity through the piezoelectric element.

The radioisotope-powered self-reciprocating cantilever provides a single platform for mechanical actuation, RF pulse generation and electrical power generation. Integration of all these functions holds great potential to enable self-powered autonomous systems.