Artificial intelligence is transforming our lives. Even at the earliest stages of commercialization, we have seen the power and potential of AI to support workers, diagnose diseases, and improve national security. Conventional digital computers are not efficient for brain-like (neuromorphic) computing such as artificial neural networks, which creates a tremendous power demands. The next stage of artificial intelligence requires devices and circuits that mimic biological processes. In particular, synaptic devices must be capable of learning in hardware. They must retain distinguishable states, be suitable for parallel computing and operate at high speed and low power. Previous attempts at such devices have been limited by disruption of the synapse memory during read operations. Essentially, the act of remembering can cause the synaptic device to forget. This award supports fundamental research to explore a new synaptic mechanism that circumvents this issue. These devices will be based on the ferroelectric materials, which have a spontaneous polarization that can be reversed by an applied electric field. This polarization can be read nondestructively by light, fulfilling the requirement for a synaptic device. These synaptic devices will be made from perovskites, materials with organic and inorganic components that are loosely bound by van der Waals forces. This approach will enable the realization of high-performance synaptic devices by informing the selection of materials, devices and circuits for neuromorphic computing. The results from this award will thereby benefit the U.S. economy, national security and health by enabling fundamental advances in the science and technology of artificial intelligence. The PI will also introduce and disseminate knowledge on materials and devices to enable AI by outreach to K-12 students.
The research objective of this project is to explore, understand and exploit the synaptic memory property and characteristics in photo-ferroelectric devices based on van der Waals halide perovskites-based for the development of artificial synapses with designed plasticity. The key idea is to exploit the non-volatile photo-ferroelectric switching in high quantum efficiency multifunctional perovskites with devices in crossbar architecture. Van der Waals halide perovskites carry both ferroelectricity and semiconducting property from separate functional groups so one could decouple and optimize both properties. In this project, the PIs will predict and understand the intrinsic/extrinsic properties of materials proposed under external stimuli, and evaluate and understand the device characteristics and performance. The PIs will synthesize thin film high quantum efficiency halide perovskite materials, fabricate two-terminal artificial synapses/circuits, study atomic structures and fundamental physical properties, determine device performance including switching speed, retention time, dynamic window, number of distinguishable states, energy consumption per synaptic event, endurance, and test/optimize the synaptic device and circuit for simulating spike-timing-dependent plasticity. This work will therefore advance knowledge of fundamental material physics, synaptic switching dynamics as well as ideal circuit architectures of novel photo-ferroelectric devices based on layered halide perovskite materials.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
