Yoshinao MIZUGAKI (水柿 義直)

Abstract

This paper presents a new method for implementing an All-Digital Phase-Locked Loop (ADPLL) using rapid single flux quantum (RSFQ) circuit technology. The ADPLL consists of a digitally controlled oscillator (DCO), a time-to-digital converter (TDC), a proportional-differential (PD) controller, a DCO controller, and a frequency divider. The DCO is based on a ring oscillator with newly designed tunable delay Josephson transmission line (TDJTL) cells in series. Verilog simulations of the ADPLL demonstrate improved performance with the integration of the TDC and PD controller. In comparison with the configuration without the TDC, the new ADPLL configuration realizes a 54% reduction in the variation of the DCO oscillation period. This reduction in frequency variation makes the ADPLL particularly well-suited for applications requiring highly accurate frequency sources, such as transmon qubit control.

The propagation of acoustic phonons between mesoscopic superconducting devices on a Si substrate with a SiO2 surface layer was studied in terms of emission-detection measurements of phonons of an energy approximately 350μeV. The emitter and detector of the phonons were composed of mesoscopic Al/AlOx/Al junctions, and the recombination phonons in the Al electrode were used for propagation measurement. It was observed that the phonons propagated laterally from the emitter to the detector as well as the vertical propagation into the interior of the substrate. The results of a comparative measurement using a device with an etched groove in the SiO2 layer between the emitter and the detector indicate that the lateral propagation path to the detector is almost completely through the SiO2 surface layer (>97%), and the propagation fashion is leaky two-dimensional propagation with the leakage due to their transmission across the SiO2–Si interface to the Si substrate. A leaky decay factor F(d) for the lateral propagation was experimentally derived as a function of the distance d between the emitter and the detector, and a simple model for F(d) was formulated on the basis of the acoustic mismatch model of phonon transmission across the interface. The devised model well explained the observed behavior of F(d). The model also indicates that the lateral propagation of phonons is mostly induced by those emitted from the emitter at nearly the critical angle for perfect reflection at the SiO2–Si interface.

Abstract

We demonstrate the fabrication and performance evaluation of physical reservoir computing (PRC) using a random network of gold-nanoparticles (GNPs). We fabricated two random arrays of GNPs with six electrodes on one sample. Electrical measurements for each array and two parallelized arrays were conducted at temperatures of 4.2 K, 77 K, and 294 K. The random connections of GNPs brought varied tunneling resistances, resulting in various output voltages. The computational performance was assessed using the second-order nonlinear autoregression moving average task. The parallelized reservoir configuration achieved the normalized mean square error as small as 0.03 at 294 K. This higher performance was evaluated through information processing capability and was attributed to the increased second-order capacity at 294 K. Although PRC with GNPs was traditionally regarded to rely on the Coulomb-blockade-induced nonlinearity, nonlinear dynamics possibly due to thermal noise and non-uniform tunnel barriers were effective in reservoir calculations even at room temperature.

Abstract

We fabricated a random network of gold nanoparticles (RN-GNPs) over 12 NiCr/Au electrodes by using a multi-step immersion method, where a sample was immersed in a gold colloid solution three times. 
Nonlinear current-voltage characteristics due to the Coulomb blockade were observed at 77 K. 
For demonstration of physical reservoir applications, input-output characteristics of the RN-GNPs were also measured in a one-input, nine-output terminal configuration. 
Distorted output voltage waveforms were obtained for a sinusoidal voltage input of 100 Hz. 
The higher-order harmonic components were confirmed in the frequency spectra of the outputs. 
The waveform reconstruction task and short-term storage capacity estimation were performed by an echo state network model with ridge regression and linear regression, respectively.