Abstract
The superior density of passive analog-grade memristive crossbar circuits enables storing large neural network models directly on specialized neuromorphic chips to avoid costly off-chip communication. To ensure efficient use of such circuits in neuromorphic systems, memristor variations must be substantially lower than those of active memory devices. Here we report a 64 × 64 passive crossbar circuit with ~99% functional nonvolatile metal-oxide memristors. The fabrication technology is based on a foundry-compatible process with etch-down patterning and a low-temperature budget. The achieved <26% coefficient of variance in memristor switching voltages is sufficient for programming a 4K-pixel gray-scale pattern with a <4% relative tuning error on average. Analog properties are also successfully verified via experimental demonstration of a 64 × 10 vector-by-matrix multiplication with an average 1% relative conductance import accuracy to model the MNIST image classification by ex-situ trained single-layer perceptron, and modeling of a large-scale multilayer perceptron classifier based on more advanced conductance tuning algorithm.
Highlights
The superior density of passive analog-grade memristive crossbar circuits enables storing large neural network models directly on specialized neuromorphic chips to avoid costly offchip communication
The bottom and top electrodes and titanium oxide layers are deposited by reactive sputtering, while aluminum oxide is formed with an atomic layer deposition (ALD) technique
The bottom electrode is planarized via a combination of chemical-mechanical polishing (CMP) and etch-back
Summary
The superior density of passive analog-grade memristive crossbar circuits enables storing large neural network models directly on specialized neuromorphic chips to avoid costly offchip communication. 1234567890():,; Analog-grade nonvolatile memories, such as those based on floating-gate transistor[1,2,3], phase-change[4,5,6], ferroelectric[7,8], magnetic[9], solid-state electrolyte[10,11,12,13], organic[14,15], and metal-oxide[16,17,18,19,20,21,22,23,24,25,26,27,28] materials, are enabling components for mixed-signal circuits implementing vector-by-matrix multiplication (VMM), which is the most common operation in any artificial neural network.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
