Scientists develop new synapse-like phototransistor for use as self-driving car sensors

According to foreign media reports, researchers at the U.S. Department of Energy's National Renewable Energy Laboratory (NREL) have made a breakthrough in energy-saving phototransistors. The devices they developed can help computers process visual information like the human brain and can be used as sensors for self-driving cars.

(Image source: cleantechnica.com)

The structure relies on a new class of semiconductors, metal-halide perovskites, which have proven highly efficient at converting sunlight into electricity and show great promise in a range of other technologies. "Perovskite semiconductors in general are really unique functional systems with potential benefits for many different technologies," said Jeffrey Blackburn, a senior scientist at NREL. "NREL is beginning to get interested in this material system for photovoltaics, but many of its properties can be applied to completely different areas of science."

The researchers combined perovskite nanocrystals with a network of single-walled carbon nanotubes to create a material combination that could have interesting properties for photovoltaics or detectors. When the researchers shone a laser on it, they discovered a surprising electrical response. "Normally, after absorbing light, the current only flows for a short period of time," said senior scientist Joseph Luther. "But in this case, the current continued to flow and did not stop even for several minutes after the light was turned off."

This behavior is called "persistent photoconductivity" and is a form of "optical memory" in which light energy shining on the device can be stored in the memory in the form of electric current. This phenomenon can also simulate the synapses in the brain used to store memories. However, persistent photoconductivity usually requires low temperature or high operating voltage, and the current peak only lasts for less than a second. In this new discovery, this persistent photoconductivity effect produces current at room temperature and the current lasts for more than an hour after the lights are turned off. In addition, only low voltage and low light intensity are required, showing that storing memories only requires low energy.

Other scientists have been working on optical memory and neuromorphic computing, which mimics the way the human brain stores information. The brain uses neural networks of neurons that interact with many other neurons through synapses. This highly interconnected network is one of the main reasons why the brain can process information in such an efficient way, so scientists use strong power to create artificial neural networks to mimic the brain's functions.

The research provides previously lacking design principles that can be incorporated into optical storage and neuromorphic computing applications. Visual perception accounts for the vast majority of the input information the brain gathers about the world, and these artificial synapses can be integrated into image recognition systems. "In many applications, sensor arrays can receive images and apply training and learning algorithms to AI and machine learning type applications," Blackburn said. "For example, such systems have the potential to improve energy efficiency, performance and reliability in applications such as self-driving cars."

The researchers tried three different types of perovskites: formamidinium lead bromide, cesium lead iodide and cesium lead bromide, and found that each produced persistent photoconductivity. "All we did was combine these two systems to create one of the simplest devices, and we demonstrated simple memory-like operations," Blackburn said. "Building neural networks requires integrating arrays of these nodes into more complex architectures to simulate more complex memory applications and image processing applications."