We are routinely fabricating many superconducting digital devices using three types of Nb-based fabrication process. One of them consists of nine Nb, one Mo and one Nb/AlOx/Nb Josephson junction (JJ) layers. Controllability in the device parameters, such as the critical current density (Jc) of JJ, has been improved. Jc in the Nb 9-layer process was controlled within ±10% of the 10 kA/cm2 target value for all 18 wafers fabricated in 2016. Uniformity of the critical current satisfied our criteria for integrated circuit applications up to a JJ size of 0.5 μm2. Reliability of the Nb 9-layer process reached the 100,000 JJ circuit level. The digital circuit processes will be improved by introducing advanced fabrication machines and/or integrating new materials. Three-dimensional, analog, analog-digital monolithic and quantum computing devices are being developed based on these digital device processes.
The recent trend and perspectives for high-performance microprocessors based on superconductor single-fluxquantum (SFQ) logic families are described. The rapid single-flux-quantum (RSFQ) and its energy-efficient derivatives are promising as a next-generation digital circuit technology for very-large-scale integration in the post-Moore's era because of the capability of ultrahigh-frequency operation over 100 GHz and low energy consumption. Several ongoing research projects as well as results reported, including several demonstrations of SFQ-based microprocessors and their components, are reviewed.
We report recent progress on adiabatic quantum-flux-parametron (AQFP) logic, which is an energy-efficient superconductor logic based on quantum flux parametron (QFP). The switching energy of an AQFP gate can be arbitrarily reduced via adiabatic switching, in which the shape of the potential energy gradually varies between single-well and double-well shapes as the result of using ac excitation magnetic fluxes. First, we show the operation principle of AQFP and explain adiabatic switching in light of potential energy shapes. We also show calculated switching energy and bit error rates to demonstrate that the switching energy of AQFP can be significantly reduced while keeping a low error rate. Then, we report recently developed AQFP subsystems, which include an 8-bit carry look-ahead adder, a register file, and a long-interconnect driver. Our demonstration reveals that AQFP technology is ready to be applied in microprocessor design.
Superconducting nanowire single-photon detectors (SSPDs) with a system detection efficiency (SDE) of over 80% are now commercially available and being used in compact Gifford-McMahon refrigerators. However, not only high SDE, but also total performance including a low dark count rate, high maximum count rate and/or low timing jitter are important for many applications. A multi-pixel SSPD combined with signal processing using a single-flux-quantum (SFQ) circuit is a very promising approach to improve the system performance of SSPDs. In this review, we introduce our recent activities on cryogenic signal processing using SFQ circuits to realize high-performance multi-pixel SSPDs.
We have been developing superconducting time-of-flight mass spectrometry (TOF-MS) systems that utilize a superconducting strip particle detector (SSPD) and a single-flux-quantum (SFQ) time-to-digital converter (TDC). The SSPD has high time resolution and constant detection efficiency even for high-mass molecules. The SFQ TDC can measure the time intervals between multiple input signals with extremely high time resolution and directly convert them into binary data. In this paper, we review recent developments on the superconducting TOF-MS systems in our group.
We derive theoretical expressions for output errors that occur in two types of superconducting level sensors. One is a conventional level sensor composed of a thin superconducting wire located vertically inside a cryostat. The other is a parallel level sensor with both superconducting and non-superconducting wires, which has been proposed to improve the output error in level sensors for liquid hydrogen using an MgB2 wire. The theoretical expressions are validated by comparing them with output errors estimated using numerical simulations based on a one-dimensional heat balance equation. It is assumed that the superconducting wire has an MgB2 monofilament surrounded by a metal sheath of stainless steel or cupronickel, whereas the non-superconducting wire is an as-drawn wire before superconducting heat treatment or a non-composite solid wire identical to the metal sheath in the superconducting wire. The level sensors are calibrated for the condition where the liquid hydrogen is spontaneously evaporated inside the cryostat or the cryostat is slowly refilled with liquid hydrogen. The output error is also estimated for three different cases of rapidly refilling with liquid hydrogen and discharging liquid hydrogen from the cryostat by internal pressurization using each gaseous hydrogen and helium.