Dye-sensitized solar cells under ambient light powering machine learning: towards autonomous smart sensors for the internet of things

Abstract

The field of photovoltaics gives the opportunity to make our buildings ''smart'' and our portable devices "independent", provided effective energy sources can be developed for use in ambient indoor conditions. To address this important issue, ambient light photovoltaic cells were developed to power autonomous Internet of Things (IoT) devices, capable of machine learning, allowing the on-device implementation of artificial intelligence. Through a novel co-sensitization strategy, we tailored dye-sensitized photovoltaic cells based on a copper(ii/i) electrolyte for the generation of power under ambient lighting with an unprecedented conversion efficiency (34%, 103 μW cm^-2 at 1000 lux; 32.7%, 50 μW cm^-2 at 500 lux and 31.4%, 19 μW cm^-2 at 200 lux from a fluorescent lamp). A small array of DSCs with a joint active area of 16 cm² was then used to power machine learning on wireless nodes. The collection of 0.947 mJ or 2.72 × 10¹⁵ photons is needed to compute one inference of a pre-trained artificial neural network for MNIST image classification in the employed set up. The inference accuracy of the network exceeded 90% for standard test images and 80% using camera-acquired printed MNIST-digits. Quantization of the neural network significantly reduced memory requirements with a less than 0.1% loss in accuracy compared to a full-precision network, making machine learning inferences on low-power microcontrollers possible. 152 J or 4.41 × 10²⁰ photons required for training and verification of an artificial neural network were harvested with 64 cm² photovoltaic area in less than 24 hours under 1000 lux illumination. Ambient light harvesters provide a new generation of self-powered and "smart" IoT devices powered through an energy source that is largely untapped.

Fig. 1. Fully autonomous IoT devices powered by harvested ambient light directly convert photons into computational information.

Fig. 2. (A) Working principle of the XY1:L1-sensitized DSC. (B) Photovoltaic performance under simulated sunlight (AM 1.5G, 100 mW cm⁻²) and 10% sunlight, (C) incident photon to current conversion efficiency spectra (IPCE), and (D) photovoltaic performance under 1000 lux fluorescent light (303.1 μW cm⁻²) of XY1, L1 and XY1:L1 co-sensitized DSCs. Corresponding parameters are listed in Table 1.

Fig. 3. Electron lifetime in XY1, L1 and XY1:L1-sensitized solar cells.

Fig. 4. Raman spectra of Cu^II/I(tmby)₂ TFSI₍₂₎ and MgTFSI₂ (powders), DSCs sensitized with the XY1:L1 sensitizer combination as well as (sensitized) DSCs containing amorphous Cu^II/I(tmby)₂ TFSI₍₂₎ hole transport material. The two latter spectra were recorded inside the ‘sandwich’ DSC. Spectra are offset for clarity.

Fig. 5. Structure of the neural network implemented on the self-powered IoT node.

Fig. 6. Atmega328P-based wireless node equipped with transceiver and AVX 6.0 V 0.47 F supercapacitor. The sensor is powered by five serial 3.2 cm² photovoltaic cells (total 16.0 cm²). The sensor was tested in simulated light intervals of 16 hours “day” (1000 lux) and 8 hours darkness (10 pm–6 am, red arrows).