Synapse-Like Phototransistor Enables Persistent, Low-Energy Optical Switching

Researchers at the U.S. Department of Energy’s (DOE) National Renewable Energy Laboratory (NREL) created a phototransistor that allows long-lived persistent photoconductivity (PPC), which is a form of optical memory. The new structure is a heterojunction between perovskite semiconductors and carbon nanotubes. The device could enable optical switching with low energy consumption for potential use in sensors and artificial neural networks and in applications such as self-driving vehicles.

The researchers combined metal-halide perovskite nanocrystals with a network of single-walled carbon nanotubes to make the device. They tested three types of perovskites — formamidinium lead bromide, cesium lead iodide, and cesium lead bromide — and found that each was able to produce PPC.

While PPC typically requires low temperatures and/or high operating voltages and delivers a current spike that lasts for small fractions of a second, the NREL device performed at room temperature and produced an electrical current that flowed for more than an hour after the light shining on the device was switched off.

“What normally would happen is that, after absorbing the light, an electrical current would briefly flow for a short period of time,” said researcher Joseph Luther. “But in this case, the current continued to flow and did not stop for several minutes even when the light was switched off.” The researchers powered the device using low voltages and low light intensities in a demonstration that showed that minimal energy was needed for the device to successfully store memory.

The team’s design could be incorporated into applications for optical memory and neuromorphic computing. Visual perception accounts for most of the input the brain collects, and interconnected neural networks allow the brain to process this input efficiently. The perovskite semiconductor structures could function as artificial synapses and be integrated into image recognition systems to mimic the brain’s ability to process information in an energy-efficient way.

Researcher Jeffrey Blackburn said that NREL originally became interested in the perovskite/carbon nanotube structure for use in photovoltaics, but found that the material combination had several properties that could make it useful in other areas.

“There are many applications where sensor arrays can take in images and apply training and learning algorithms for artificial intelligence and machine-learning-type applications,” he said. “As an example, such systems could potentially improve energy efficiency, performance, and reliability in applications such as self-driving vehicles. In general, these perovskite semiconductors are a really unique functional system with potential benefits for a number of different technologies.

“What we made is only one of the simplest devices you could make from combining these two systems, and we demonstrated a simplistic memory-like operation. To build a neural network requires integrating an array of these junctions into more complex architectures, where more complex memory applications and image processing applications can be emulated.”

The research was supported by the Center for Hybrid Organic-Inorganic Semiconductors for Energy (CHOISE), an Energy Frontier Research Center within the DOE’s Office of Science. The research was published in Science Advances (www.doi.org/10.1126/sciadv.abf1959).