Analog Resistive Switching Devices for Training Deep Neural Networks with the Novel Tiki-Taka Algorithm

A critical bottleneck for the training of large neural networks (NNs) is communication with off-chip memory. A promising mitigation effort consists of integrating crossbar arrays of analogue memories in the Back-End-Of-Line, to store the NN parameters and efficiently perform the required synaptic operations. The "Tiki-Taka" algorithm was developed to facilitate NN training in the presence of device nonidealities. However, so far, a resistive switching device exhibiting all the fundamental Tiki-Taka requirements, which are many programmable states, a centered symmetry point, and low programming noise, was not yet demonstrated. Here, a complementary metal-oxide semiconductor (CMOS)-compatible resistive random access memory (RRAM), showing more than 30 programmable states with low noise and a symmetry point with only 5% skew from the center, is presented for the first time. These results enable generalization of Tiki-Taka training from small fully connected networks to larger long-/short-term-memory types of NN.

Zeeman- and Orbital-Driven Phase Shifts in Planar Josephson Junctions

We perform supercurrent and tunneling spectroscopy measurements on gate-tunable InAs/Al Josephson junctions (JJs) in an in-plane magnetic field and report on phase shifts in the current-phase relation measured with respect to an absolute phase reference. The impact of orbital effects is investigated by studying multiple devices with different superconducting lead sizes. At low fields, we observe gate-dependent phase shifts of up to φ0 = 0.5π, which are consistent with a Zeeman field coupling to highly transmissive Andreev bound states via Rashba spin-orbit interaction. A distinct phase shift emerges at larger fields, concomitant with a switching current minimum and the closing and reopening of the superconducting gap. These signatures of an induced phase transition, which might resemble a topological transition, scale with the superconducting lead size, demonstrating the crucial role of orbital effects. Our results elucidate the interplay of Zeeman, spin-orbit, and orbital effects in InAs/Al JJs, giving improved understanding of phase transitions in hybrid JJs and their applications in quantum computing and superconducting electronics.

Demonstration of the Nonlocal Josephson Effect in Andreev Molecules

We perform switching current measurements of planar Josephson junctions (JJs) coupled by a common superconducting electrode with independent control over the two superconducting phase differences. We observe an anomalous phase shift in the current-phase relation of a JJ as a function of gate voltage or phase difference in the second JJ. This demonstrates the nonlocal Josephson effect, and the implementation of a φ0-junction which is tunable both electrostatically and magnetically. The anomalous phase shift is larger for shorter distances between the JJs and vanishes for distances much longer than the superconducting coherence length. Results are consistent with the hybridization of Andreev bound states, leading to the formation of an Andreev molecule. Our devices constitute a realization of a tunable superconducting phase source and could enable new coupling schemes for hybrid quantum devices.