Answers are grounded in this paper and cite their sources.
Ferroelectric Tunnel Junctions (FTJs) are a candidate for the hardware realization of synapses in artificial neural networks. The fabrication process for a 784 × 100 crossbar array of 500 nm large FTJs, exhibiting effective On/Off currents ratio in the range 50–100, is presented. First, the epitaxial 4 nm-BiFeO 3 /Ca 0.96 Ce 0.04 MnO 3 //YAlO 3 is combined with Ni electrodes. The oxidation of Ni during the processing affects the polarity of the FTJ and the On/Off ratio, which becomes comparable to that of CMOS-compatible HfZrO 4 junctions. The latter have a wider coercive field distribution: consequently, in test crossbar arrays, BiFeO 3 exhibits a smaller cross-talk than HfZrO 4 . Furthermore, the relatively larger threshold for ferroelectric switching in BiFeO 3 allows the use application of half-programming schemes for supervised and unsupervised learning. Second, the heterostructure is combined with W and Pt electrodes. The design is optimized for the controlled collapse chip connection to neuromorphic circuits.
No takes yet. Share an insight, caveat, or question.
Halter et al. (2023) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: