Design and optimization of MEMS piezoelectric energy harvesters for improved efficiency

This paper proposes an automated design and optimization methodology for performance improvement of the MEMS unimorph piezoelectric harvesters based on Genetic Algorithm (GA), which can be conducted with minimum human efforts. In this regard, an analytic equation, which can predict harvested voltage in terms of geometric dimensions, is utilized as an objective function of GA. By using micro-fabrication process, we fabricated the optimized MEMS harvesters. The experimental results demonstrate that the prototyped energy harvester has larger generated voltage magnitude by a factor of 31% in comparison with un-optimized ones, along with device area reduction from 3.1 mm
²
to 1.55 mm
²
. Furthermore, a new T-shaped unimorph MEMS harvester with higher conversion efficiency from mechanical vibration to electricity by offering both bending stress and torsion stress, is proposed and fabricated. Our measurement shows that the T-shaped harvester can generate higher voltage by a factor of 96% with reference to the optimized conventional unimorph piezoelectric harvester, while its occupied area is further reduced by 35% to 1.01 mm
²
.