INDUSTRY TRACK: Building AI Hardware with Oxide Quantum Materials and Devices Integrated on Microelectronic Platforms

INDUSTRY TRACK: As the need for energy-efficient computing becomes more imperative, novel semiconductor materials and computing paradigms must be actively explored along with strategies for their seamless integration with CMOS circuitry. In this context, I will discuss promising opportunities utilizing phase transitions in correlated oxide semiconductors as building blocks for non-volatile and multi-state memory devices and neuromorphic information processors. Examples of state changes that will be discussed include magnetic transitions, electrical conductance transitions that can be driven by electric fields and with controllable memory timescales. We will then consider reconfigurable functions using phase transitions, e.g. switching between synapse and neuron functions for instance, that can greatly simplify device design and open implementation of neurogenesis and continual learning algorithms in hardware. Time permitting, I will also discuss how stochastic phase switching in such systems enables realization of probabilistic bits. All of these wonderful properties can be realized at room temperature and with lithographic fabrication processes making them compatible with large scale manufacturing.