Photovoltaic Energy Harvesting for Millimeter-Scale Systems

Abstract

The Internet of Things (IoT) based on mm-scale sensors is a transformational technology that opens up new capabilities for biomedical devices, surveillance, micro-robots and industrial monitoring. Energy harvesting approaches to power IoT have traditionally included thermal, vibration and radio frequency. However, the achievement of efficient energy scavenging for IoT at the mm-scale or sub mm-scale has been elusive. In this work, I show that photovoltaic (PV) cells at the mm-scale can be an alternative means of wireless power transfer to mm-scale sensors for IoT, utilizing ambient indoor lighting or intentional irradiation of near-infrared (NIR) LED sources through biological tissue. Single silicon and GaAs photovoltaic cells at the mm-scale can achieve a power conversion efficiency of more than 17 % for silicon and 30 % for GaAs under low-flux NIR irradiation at 850 nm through the optimized device structure and sidewall/surface passivation studies, which guarantees perpetual operation of mm-scale sensors. Furthermore, monolithic single-junction GaAs photovoltaic modules offer a means for series-interconnected cells to provide sufficient voltage (> 5 V) for direct battery charging, and bypassing needs for voltage up-conversion circuitry. However, there is a continuing challenge to miniaturize such PV systems down to the sub mm-scale with minimal optical losses from device isolation and metal interconnects and efficient voltage up-conversion. Vertically stacked dual-junction PV cells and modules are demonstrated to increase the output voltage per cell and minimize area losses for direct powering of miniature devices for IoT and bio-implantable applications with low-irradiance narrowband spectral illumination. Dual-junction PV cells at small dimensions (150 µm x 150 µm) demonstrate power conversion efficiency greater than 22 % with more than 1.2 V output voltage under low-flux 850 nm NIR LED illumination, which is sufficient for batteryless operation of miniaturized CMOS IC chips. The output voltage of dual-junction PV modules with eight series-connected single cells is greater than 10 V while maintaining an efficiency of more than 18 %. Finally, I demonstrate monolithic PV/LED modules at the µm-scale for brain-machine interfaces, enabling two-way optical power and data transfer in a through-tissue configuration. The wafer-level assembly plan for the 3D vertical integration of three different systems including GaAs LED/PV modules, CMOS silicon chips, and neural probes is proposed.