This project investigates the use of ultra-violet light to modify on-chip voltages. Analog quantities in on-chip long term storage can be used to compensate for fabrication variations and to provide variable synapses in neural networks. The voltage on floating gates in a standard CMOS integrated circuit (IC) can be set by shining uniform ultra-violet light on the chip and arranging the circuit geometry so that nearby active areas have the correct voltages. The use of ultra-violet light to set the weights on synapses in a Hopfield style neural network is being investigated. Research tasks include: (1) refine the basic circuit model through experimentation; (2) measure cycle lifetimes on circuits with a faster settling time; (3) investigate electrical methods of controlling charge on floating gates; (4) fabricate, test and refine a pattern recognition chip with floating node storage of synapse weights; and (5) identify and test circuits suitable for distribution as layout libraries.
