Innately cognitive devices for brain-like computational devices
The biological world is one of continual change and renewal in response to the changing environment. The human body is an intricate system combining thinking, memory and sensing devices that can outperform modern day computing devices in many important metrics. One of these is the energy consumed in carrying out a unit of computation: the human and other mammalian brains carry out such computations using a million times less energy than a silicon-based FET. This means, for example, that a human brain consumes ~20 watts of power but to carry out comparable computations in a modern-day silicon-based machine would require of the order of 100 kilowatts. It is of course clear that silicon-based computers are capable of carrying out certain operations beyond the capabilities of the human brain but these are typically mathematical and algorithmic in nature. Where the human brain enjoys immense advantages is in cognition, its ability to “think”! The nominee’s recent research activities have focused on novel solid-state devices that exhibit “cognitive” behaviors, i.e. their fundamental structures “mutate” in response to the external environment. In a paper published in Science in March 2013, the nominee demonstrates how electric fields generated using ionic liquid dielectrics can induce the migration of oxygen ions at the interface of a thin film of a correlated oxide, vanadium dioxide, into the liquid, that results in the metallization of the oxide. This process is non- volatile but can be reversed by switching the direction of the electric field. The nominee proposes a new field of research into “liquid electronics” whereby the flow of liquids in channels fabricated at the surface or within oxide thin films are used to control the properties of the oxide materials with or without gate voltages. In the extreme case reconfigurable electronic circuits can be “painted” onto oxide materials by the controlled flow of liquids through micro- or nano-channels.
This proposed area of research will take advantage of microfluidic and nanofluidic technologies but in a novel manner. This potentially groundbreaking research will provide a powerful focus and will take advantage of the world-class research at Halle on polymers, nano-fabrication and oxide interfaces. The primary goal will be the development of component devices and architectures to produce computing systems that can perform more powerful operations than today’s silicon-based devices but use orders of magnitude less power.