Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Ali Bozbey has a total of 40 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Hybrid synaptic structure for spiking neural network realization

    by S Razmkhah, M A Karamuftuoglu, A Bozbey

    Neural networks and neuromorphic computing represent fundamental paradigms as alternative approaches to Von-Neumann-based implementations, advancing in the applications of deep learning and machine vision. Nonetheless, conventional semiconductor circuits encounter challenges in achieving ultra-fast processing speed and low power consumption due to their dissipative properties. Conversely, single flux quantum circuits exhibit inherent spiking behavior, showcasing their characteristics as a promising candidate for spiking neural networks (SNNs). In this work, we present a compact hybrid synapse circuit to mimic the biological interconnect functionality, enabling the weighting operations for excitatory and inhibitory impulses. Additionally, the proposed structure facilitates input accumulation, which is performed before the activation function. In the experiments, our synaptic structure interfaces with a soma circuit fabricated using a commercial Nb process, underscoring its compatibility and supporting its potential for integration into efficient neural network architectures. The weight value on the synapse is configurable by utilizing cryo-CMOS circuits, providing adaptability to the inference networks. We’ve successfully designed, fabricated, and partially tested the JJ-Synapse within our cryocooler system, enabling high-speed inference implementation for SNNs.


  • Open Access

    Efficient Superconductor Arithmetic Logic Unit for Ultra-Fast Computing

    by Sasan Razmkhah, Ali Bozbey

    We present a 4-bit Arithmetic Logic Unit (ALU) utilizing superconductor technology. The ALU serves as the central processing unit of a processor, performing crucial arithmetic and logical operations. We have adopted a bit-parallel architecture to ensure an efficient and streamlined design with minimal fanin/fanout and optimal latency. In terms of fabrication, the ALU has been fabricated using a standard commercial process. It operates at an impressive clock frequency exceeding 30 GHz while consuming a mere 4.75 mW of power, including applied reverse current, encompassing static and dynamic components. The ALU contains over 9000 Josephson junctions, with approximately 7000 JJs dedicated to wiring, delay lines, and path balancing, and it has over 18% bias margin. Designed as a co-processor, this arithmetic logic unit will work with external CMOS memory and processors via interface circuits. Thorough testing and validation of the ALU's functionality have been conducted with digital and analog simulations, and all the components were fabricated and measured within a 4K pulse-tube cryocooler. Experimental verification has confirmed the successful operation of both the arithmetic and logic units. These results have been analyzed and are presented alongside the experimental data to provide comprehensive insights into the ALU's behavior and capabilities.

  • Implementation of Neuro-Inspired Arithmetic and Logic Circuits

    by Mustafa Altay Karamuftuoglu, Ali Bozbey, Murat Ozbayoglu

    Artificial neurons provide a new way of computation for neuro-inspired algorithms, and the abilities may efficiently solve the challenges. We propose implementations of logic gates (AND, OR, XOR, Majority), full adder, full subtractor, even parity generator, and 2-bit multiplier circuit formed by JJ-Soma and standard RSFQ digital library cells. The designed circuits execute ultra-high-speed operations without a clock signal, and they are capable of processing parallel or time-sliced operations. The combination of JJ-Soma cells and SFQ cells creates the potential models for the arithmetic logic unit devices with a small on-chip area, high operating speed, and pipeline structures for microprocessors. In this study, the previously optimized JJ-Soma circuits have low power consumption and high computational speed where the firing rates for two pulse and three pulse threshold circuits were designed to be 50 and 15 GHz with about 10 $^{-19}$ J/spike energy level. The proposed circuits, fabricated with a commercial foundry service, have been implemented and demonstrated experimentally.


  • JJ-Soma: Toward a Spiking Neuromorphic Processor Architecture

    by Mustafa Altay Karamuftuoglu, Ali Bozbey, Sasan Razmkhah

    Neuromorphic computing and artificial neurons have been shown to improve the solution for some of the complex problems for conventional computers. We present a spiking soma Josephson junction-based soma (JJ-Soma) circuit that consists of a double-junction SQUID interfered with a resistor (threshold loop), a decaying superconductor loop cut by a resistor, which is coupled to the SQUIDlike structure. The proposed soma has three main properties: 1) ultrahigh-speed operation with minimal power consumption; 2) compatibility with standard foundry processes that allow fabrication with the available infrastructure; and 3) compatibility with conventional single flux quantum (SFQ) logic gates, which enables the design and implementation of complicated networks. Each soma circuit covers 40 µm × 80 µm area on-chip with different activation functions. The circuits are fabricated in a commercial foundry for superconductors and they have been implemented and demonstrated experimentally.


  • Efficiency enhancement in photoelectrochemical water splitting: Defect passivation and boosted charge transfer kinetics of zinc oxide nanostructures via chalcopyrite/chalcogenide mix sensitization

    by Cigdem Tuc Altaf, Nazrin Abdullayeva, Ozlem Coskun, Alihan Kumtepe, İpEk Deniz Yildirim, Emre Erdem, Maochang Liu, Ali Bozbey, Ertan Agar, Mehmet Sankir, et al.

    ZnO thin films in nanorod (NR) and nanoflower (NF) morphologies were used as photoelectrode scaffolds for efficient visible-light-driven photoelectrochemical (PEC) water splitting process, where their decoration with copper indium gallium sulfide (CIGS) and indium sulfide (In2S3) layers resulted in significant PEC performance enhancement. ZnO NF/CIGS/In2S3 photoelectrodes exhibited a remarkably high PEC efficiency (∼6.0% applied bias photon-to-current efficiency, 83% incident photon-to-current efficiency) due to the negligible dark current, while ZnO NR/CIGS/In2S3 generated a photocurrent density of 30.0mA.cm–2 at 0.4 V (vs Ag/AgCl), being one of the highest performances reported in the literature for copper-based chalcopyrite photoelectrodes so far. The interfacial photoelectrode-electrolyte charge transport dynamics, investigated via intensity-modulated photocurrent spectroscopy, exhibited a sevenfold increase in charge transfer efficiencies with a significant drop in surface recombination kinetics for ZnO NF after CIGS/In2S3 decoration. The obtained results show consistency with numerically modeled electric field distribution profiles and electron paramagnetic resonance results of ZnO NF, rationalizing the enhanced charge transfer rates for decorated samples and confirming the defect passivating nature of CIGS/In2S3.


  • Synchronous SFQ Address Encoder for Superconductor Detector Arrays

    by Ali Bozbey, Eren Can Aydogan, Kubra Usenmez, Sasan Razmkhah, Masamitsu Tanaka, Akira Fujimaki

    With the advances in superconductor detectors and single flux quantum (SFQ) logic circuits, megapixel imagers have become possible. Various detectors such as superconducting stripline detectors (SSLD), kinetic inductance detectors (KID), nanowire single-photon detectors have quite matured in terms of speed and performance. However, scaling of the read-out electronics is still a challenge. We propose an SFQ based synchronous address encoder to read-out the response of an SSLD array. The encoder circuit provides two sets of outputs at each clock period: one is the time-division multiplexed, serial data output, and the other is the synchronized address output associated with the time-multiplexed data. So, the number of connections to the room temperature is substantially reduced. We experimentally demonstrate a 2-bit encoder. Then we present the scalability of the circuit to larger arrays and show the simulation results of a 4-bit encoder which is suitable to read-out a 16x16 detector array.


  • A compact high frequency voltage amplifier for superconductor–semiconductor logic interface

    by Sasan Razmkhah, Ali Bozbey, Pascal Febvre

    The many advantages of cryogenically cooled single-flux quantum (SFQ) circuits imply that the high speed and low voltage output signals must be amplified and interfaced with standard electronics. State-of-the-art low-noise and wide-band amplifiers are required to interface with room temperature electronics. One solution is to place preamplifiers at the cryogenic stage, which requires specific semiconductor design and fabrication. However, a more viable and energy-efficient approach is to integrate the pulsed logic circuit output stages with on-chip superconductor preamplifiers. We designed, fabricated, and tested an on-chip compact voltage multiplier integrated with the output stage of SFQ circuits to increase the voltage amplitude of SFQ pulses. The circuit is designed with the same technology as the logic circuit hence its noise level is lower, and it works at higher frequencies compared to complementary metal-oxide semi-conductor amplifiers and due to quantized nature of it there is no added noise. The fabricated circuit has a compact size of 160 µ m × 320 µ m and provides about 10 dB gain with measured 600 µ V output voltage at frequencies up to ∼25 GHz in simulations. By stacking more levels, over 20 dB gain at circuit level is achievable as shown in simulations. Moreover the gain of the superconductor voltage amplifier is quantized and programmable.


  • SFQ Parallel Encoders Promising for Video Imaging With Superconductor Stripline Detectors

    by Masamitsu Tanaka, Misaki Kozaka, Kyohei Kamiya, Kubra Usenmez, Eren Can Aydogan, Sasan Razmkhah, Ali Bozbey, Akira Fujimaki

    We report demonstration of a readout circuit with high-throughput single-flux-quantum (SFQ)-based encoders for a large-scale video imaging system with a conventional frame rate. The encoders are fully parallel and integrated on the same chip as linear arrays of serially connected superconductor stripline detectors (SSLDs), superconducting quantum interference device (SQUID)-based comparators, and SFQ synchronizers. Mutual coupling of parallel SSLDs and SQUID-based comparators allows the use of only one DC bias line; it is possible to use thousands of SSLDs without increasing the bias current or number of bias ports. To achieve crosstalk-free readout, the serially connected SSLDs are separated by low-pass filters or large inductors. We experimentally demonstrated crosstalk-free readout including SFQ parallel encoders from one- or two-dimensional serially connected SSLD arrays.


  • Design and Implementation of a Single Flux Quantum Logic-Based Memory Controller for Josephson-CMOS Hybrid Memory Systems

    by Seda Demirhan, Ali Bozbey

    Single Flux Quantum (SFQ) digital circuits have shown the potential for high speed/low power computation applications. Unfortunately, because of the low integration density and low driving capability of SFQ circuits, realizing large-scale memories by using only SFQ circuits is still a challenge. Josephson-CMOS hybrid memory, hybridizing high-speed, low power SFQ circuits and high-density CMOS memories is already proposed as a solution to the large-scale memory problem in SFQ digital systems. In this study, an SFQ based memory controller which works up to 10 GHz clock frequencies is designed for Josephson-CMOS hybrid memory systems. The memory controller acts as a SFQ/CMOS interface between the SFQ circuit and SRAM module and generates the required waveforms for SRAM read and write operations. The circuit which has 4-bit data and 2-bit address lines is fabricated with AIST 2.5 kA/cm2 STP2 process and operations of the circuits are verified. To demonstrate the scalability of the circuit, memory controller was scaled to 8-bit data and 13-bit address to control of the 64kb SRAM. Operations of the scaled memory controller are verified with analog simulations. Scaled circuit consumes 0.76 mW power in an area of 1.64 mm x 1.60 mm.