Modeling and design of a MEMS piezoelectric vibration energy harvester

Abstract

The modeling and design of MEMS-scale piezoelectric-based vibration energy harvesters (MPVEH) are presented. The work is motivated by the need for pervasive and limitless power for wireless sensor nodes that have application in structural health monitoring, homeland security, and infrastructure monitoring. A review of prior milli- to micro-scale harvesters is provided. Common ambient low-level vibration sources are characterized experimentally. Coupled with a dissipative system model and a mechanical damping investigation, a new scale-dependent operating frequency selection scheme is presented. Coupled electromechanical structural models are developed, based on the linear piezoelectric constitutive description, to predict uni-morph and bi-morph cantilever beam harvester performance. Piezoelectric coupling non-intuitively cancels from the power prediction under power-optimal operating conditions, although the voltage and current are still dependent on this property. Piezoelectric material selection and mode of operation ([3-1] vs. [3-3]) therefore have little effect on the maximum power extracted. The model is verified for resonance and off-resonance operation by comparison to new experimental results for a macro-scale harvester. Excellent correlation is obtained away from resonances in the small-strain linear piezoelectric regime. The model consistently underpredicts the response at resonances due to the known non-linear piezoelectric constitutive response (higher strain regime). Applying the model, an optimized single prototype bi-morph MPVEH is designed concurrently with a microfabrication scheme.

(cont.) A low-level (2.5 m/s²), low-frequency (150 Hz) vibration source is targeted for anti-resonance operation, and a power density of 313 [mu]W/cm³ and peak-to-peak voltage of 0.38 V are predicted per harvester. Methodologies for the scalar analysis and optimization of uni-morph and bi-morph harvesters are developed, as well as a scheme for chip-level assembly of harvester clusters to meet different node power requirements.