Berkeley Lab

Materials and Devices for Brain-like Computing

By Alberto Salleo

The “Materials and Devices for Brain-like computing” explored technologies for next generation computing -widely believed to benefit from taking inspiration from biology- as well as technologies that mimic nature or can integrate with living matter. Dr. Talin (Sandia National Lab) and Prof. Strukov (UC Santa Barbara) described new devices that can be used to make neural nets in hardware, such as crossbars. Importantly, Prof. Kuzum (UC San Diego) made a strong point that devices and algorithms should be co-designed in order to avoid putting extremely challenging constraints on materials and device characteristics. Neural nets are meant to mimic the computing processes in the brain, with a large degree of abstraction however. Dr. Gkoupidenis (Max Planck Institute, Mainz) on the other hand showed how arrays of polymer-based electrochemical transistors can be used to emulate actual natural processes occurring in the brain, such as coherent oscillations. Going one step further, Prof. Lee (Seoul National University) used organic electrochemical devices to transduce external stimuli (e.g. pressure) into contractions in resected insect legs, thus demonstrating an electronically functional interface to living matter. In a similar vein, Dr. Tang (Stanford University) explained how blind mice can react to optical signals thanks to the exceptional optoelectronic properties of TiO2 nanowires. Integrating foreign materials with living matter however is extremely challenging, as shown by Dr. Santoro (Italian Institute of Technology, Naples). Indeed, Dr. Santoro strikingly demonstrated that cells bind very differently to different surface features, thus suggesting that electronic devices that couple to living matter will need to have non-conventional 3D surface topographies.