Bhaswar Chakrabarti
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Bhaswar Chakrabarti has a total of 96 co-authors. Here are the top co-authors by number of shared publications:
Konrad Seidel
Center Nanoelectronic Technologies, Fraunhofer-Institutfur Photonische Mikrosysteme IPMS, Dresden, Germany
7 shared publications- Maximilian Lederer
Center Nanoelectronic Technologies (CNT), Fraunhofer IPMS, Dresden, Germany
7 shared publications - Sourav De
National Tsing Hua University
7 shared publications Luca Pirro
GLOBALFOUNDRIES, Dresden, Germany
5 shared publicationsYannick Raffel
5 shared publicationsSourodeep Roy
Department of Electrical Engineering
4 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 73 (11) , 7174-7177
Impact of γ -Radiation on Volatile ZnO Memristor-Based Reservoir Computing
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 73 (11) , 7174-7177
Impact of γ -Radiation on Volatile ZnO Memristor-Based Reservoir Computing
- Open Access
- 18 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1) , 6836
NeoHebbian synapses to accelerate online training of neuromorphic hardware
Neuromorphic systems that employ advanced synaptic learning rules, such as the three-factor learning rule, require synaptic devices of increased complexity. Herein, a novel neoHebbian artificial synapse utilizing ReRAM devices has been proposed and experimentally validated to meet this demand. This synapse features two distinct state variables: a neuron coupling weight and an “eligibility trace” that dictates synaptic weight updates. The coupling weight is encoded in the ReRAM conductance, while the “eligibility trace” is encoded in the local temperature of the ReRAM and is modulated by applying voltage pulses to a physically co-located resistive heating element. The utility of the proposed synapse has been investigated using two representative tasks: first, temporal signal classification using Recurrent Spiking Neural Networks (RSNNs) employing the e-prop algorithm, and second, Reinforcement Learning (RL) for path planning tasks in feedforward networks using a modified version of the same learning rule. System-level simulations, accounting for various device and system-level non-idealities, confirm that these synapses offer a robust solution for the fast, compact, and energy-efficient implementation of advanced learning rules in neuromorphic hardware.
- 18 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1) , 6836
Open AccessNeoHebbian synapses to accelerate online training of neuromorphic hardware
Neuromorphic systems that employ advanced synaptic learning rules, such as the three-factor learning rule, require synaptic devices of increased complexity. Herein, a novel neoHebbian artificial synapse utilizing ReRAM devices has been proposed and experimentally validated to meet this demand. This synapse features two distinct state variables: a neuron coupling weight and an “eligibility trace” that dictates synaptic weight updates. The coupling weight is encoded in the ReRAM conductance, while the “eligibility trace” is encoded in the local temperature of the ReRAM and is modulated by applying voltage pulses to a physically co-located resistive heating element. The utility of the proposed synapse has been investigated using two representative tasks: first, temporal signal classification using Recurrent Spiking Neural Networks (RSNNs) employing the e-prop algorithm, and second, Reinforcement Learning (RL) for path planning tasks in feedforward networks using a modified version of the same learning rule. System-level simulations, accounting for various device and system-level non-idealities, confirm that these synapses offer a robust solution for the fast, compact, and energy-efficient implementation of advanced learning rules in neuromorphic hardware.
- 31 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 47 (3) , 645-648
A Tunable Capacitance ReRAM for Improvement of Dynamic Range in CMOS Image Sensors
work reports a tunable-capacitance resistive memory (TCReRAM) device. The TCReRAM uses a binary metal-oxide active-layer and exhibits strong voltage-dependent capacitance (VDC) under appropriate operating conditions. The observed VDC dynamics is characterized experimentally, and its physical origin is investigated through comprehensive multiscale simulations. The developed TCReRAM device, integrated at the floating diffusion node of a CMOS image sensor (CIS) pixel, introduces VDC that helps mitigate saturation under high illumination conditions. Pixel-level simulations further demonstrate that incorporating the TCReRAM significantly improves the dynamic range by nearly 25% compared to the existing reported Hybrid CIS, effectively addressing performance limitations across varying illumination levels.
- 31 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 47 (3) , 645-648
A Tunable Capacitance ReRAM for Improvement of Dynamic Range in CMOS Image Sensors
work reports a tunable-capacitance resistive memory (TCReRAM) device. The TCReRAM uses a binary metal-oxide active-layer and exhibits strong voltage-dependent capacitance (VDC) under appropriate operating conditions. The observed VDC dynamics is characterized experimentally, and its physical origin is investigated through comprehensive multiscale simulations. The developed TCReRAM device, integrated at the floating diffusion node of a CMOS image sensor (CIS) pixel, introduces VDC that helps mitigate saturation under high illumination conditions. Pixel-level simulations further demonstrate that incorporating the TCReRAM significantly improves the dynamic range by nearly 25% compared to the existing reported Hybrid CIS, effectively addressing performance limitations across varying illumination levels.
- Open Access
Interfacial and bulk switching MoS2 memristors for an all-2D reservoir computing framework
In this study, we design a reservoir computing (RC) network by exploiting short- and long-term memory dynamics in Au/Ti/MoS$_2$/Au memristive devices. The temporal dynamics is engineered by controlling the thickness of the Chemical Vapor Deposited (CVD) MoS$_2$ films. Devices with a monolayer (1L)-MoS$_2$ film exhibit volatile (short-term memory) switching dynamics. We also report non-volatile resistance switching with excellent uniformity and analog behavior in conductance tuning for the multilayer (ML) MoS$_2$ memristive devices. We correlate this performance with trap-assisted space-charge limited conduction (SCLC) mechanism, leading to a bulk-limited resistance switching behavior. Four-bit reservoir states are generated using volatile memristors. The readout layer is implemented with an array of nonvolatile synapses. This small RC network achieves 89.56\% precision in a spoken-digit recognition task and is also used to analyze a nonlinear time series equation.
Open AccessInterfacial and bulk switching MoS2 memristors for an all-2D reservoir computing framework
In this study, we design a reservoir computing (RC) network by exploiting short- and long-term memory dynamics in Au/Ti/MoS$_2$/Au memristive devices. The temporal dynamics is engineered by controlling the thickness of the Chemical Vapor Deposited (CVD) MoS$_2$ films. Devices with a monolayer (1L)-MoS$_2$ film exhibit volatile (short-term memory) switching dynamics. We also report non-volatile resistance switching with excellent uniformity and analog behavior in conductance tuning for the multilayer (ML) MoS$_2$ memristive devices. We correlate this performance with trap-assisted space-charge limited conduction (SCLC) mechanism, leading to a bulk-limited resistance switching behavior. Four-bit reservoir states are generated using volatile memristors. The readout layer is implemented with an array of nonvolatile synapses. This small RC network achieves 89.56\% precision in a spoken-digit recognition task and is also used to analyze a nonlinear time series equation.
- 19 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 46 (11) , 2014-2017
Trapping Dynamics and Endurance in HfO 2 -FeFETs: An Insight From Charge Pumping
We investigate endurance degradation in hafnium oxide (HfO₂)-based ferroelectric field effect transistors (FeFETs) by analyzing interface (Nit) and near-interface oxide traps (Nox) using a two-level charge pumping (CP) technique. After 10⁴ program (PG) / erase (ER) cycles at ±5 V, 100 ns, Nit increases by ~2×, and the memory window (MW) reduces from 1.2 V to 0.45 V. Temperature-dependent measurements reveal accelerated trap generation above 50 °C, with decreasing charge-pumping current (ICPmax) due to deeper oxide trap occupation. Pulse parameter tuning shows that longer rise/fall times, and shorter hold durations mitigate Nit accumulation and delay memory window collapse. The extracted average interface trap density (Dit) reaches ~4×10¹² cm⁻²eV⁻¹ after cycling. These results emphasize the role of trap engineering in improving FeFET endurance for scaled non-volatile memory applications.
- 19 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 46 (11) , 2014-2017
Trapping Dynamics and Endurance in HfO 2 -FeFETs: An Insight From Charge Pumping
We investigate endurance degradation in hafnium oxide (HfO₂)-based ferroelectric field effect transistors (FeFETs) by analyzing interface (Nit) and near-interface oxide traps (Nox) using a two-level charge pumping (CP) technique. After 10⁴ program (PG) / erase (ER) cycles at ±5 V, 100 ns, Nit increases by ~2×, and the memory window (MW) reduces from 1.2 V to 0.45 V. Temperature-dependent measurements reveal accelerated trap generation above 50 °C, with decreasing charge-pumping current (ICPmax) due to deeper oxide trap occupation. Pulse parameter tuning shows that longer rise/fall times, and shorter hold durations mitigate Nit accumulation and delay memory window collapse. The extracted average interface trap density (Dit) reaches ~4×10¹² cm⁻²eV⁻¹ after cycling. These results emphasize the role of trap engineering in improving FeFET endurance for scaled non-volatile memory applications.
- Open Access
- 1 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of the Electron Devices Society
- Vol. 13, 762-768
Spike-Timing Dependent Learning Dynamics in Silicon-Doped Hafnium-Oxide-Based Ferroelectric Field Effect Transistors
Brain-inspired computing, with its potential for energy-efficient spatio-temporal data processing, has spurred significant interest in spiking neural networks and their hardware implementations. Leveraging their non-volatile memory and analog tunability, Ferroelectric field-effect transistors have emerged as promising candidates for realizing low-power synaptic devices within spiking neural networks. However, previous ferroelectric field-effect transistor-based implementations of spike-timing-dependent plasticity, a crucial learning mechanism in spiking neural networks, have often relied on complex circuit topologies or suffered from high energy consumption. Here, we report a comprehensive study of spike-timing-dependent plasticity learning dynamics in silicon-doped hafnium oxide-based ferroelectric field effect transistors, demonstrating precise control of synaptic weight modulation using various spike shapes and timings. We investigate the impact of different spike waveforms on energy consumption and find that triangular spikes achieve a 20% reduction in energy consumption compared to rectangular spikes, a significant improvement for large-scale spiking neural network implementations. Our results highlight the potential of single-device ferroelectric field-effect transistor synapses for realizing energy-efficient and scalable spiking neural networks, paving the way for next-generation neuromorphic computing.
- 1 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of the Electron Devices Society
- Vol. 13, 762-768
Open AccessSpike-Timing Dependent Learning Dynamics in Silicon-Doped Hafnium-Oxide-Based Ferroelectric Field Effect Transistors
Brain-inspired computing, with its potential for energy-efficient spatio-temporal data processing, has spurred significant interest in spiking neural networks and their hardware implementations. Leveraging their non-volatile memory and analog tunability, Ferroelectric field-effect transistors have emerged as promising candidates for realizing low-power synaptic devices within spiking neural networks. However, previous ferroelectric field-effect transistor-based implementations of spike-timing-dependent plasticity, a crucial learning mechanism in spiking neural networks, have often relied on complex circuit topologies or suffered from high energy consumption. Here, we report a comprehensive study of spike-timing-dependent plasticity learning dynamics in silicon-doped hafnium oxide-based ferroelectric field effect transistors, demonstrating precise control of synaptic weight modulation using various spike shapes and timings. We investigate the impact of different spike waveforms on energy consumption and find that triangular spikes achieve a 20% reduction in energy consumption compared to rectangular spikes, a significant improvement for large-scale spiking neural network implementations. Our results highlight the potential of single-device ferroelectric field-effect transistor synapses for realizing energy-efficient and scalable spiking neural networks, paving the way for next-generation neuromorphic computing.
- Open Access
NeoHebbian Synapses to Accelerate Online Training of Neuromorphic Hardware
Neuromorphic systems that employ advanced synaptic learning rules, such as the three-factor learning rule, require synaptic devices of increased complexity. Herein, a novel neoHebbian artificial synapse utilizing ReRAM devices has been proposed and experimentally validated to meet this demand. This synapse features two distinct state variables: a neuron coupling weight and an "eligibility trace" that dictates synaptic weight updates. The coupling weight is encoded in the ReRAM conductance, while the "eligibility trace" is encoded in the local temperature of the ReRAM and is modulated by applying voltage pulses to a physically co-located resistive heating element. The utility of the proposed synapse has been investigated using two representative tasks: first, temporal signal classification using Recurrent Spiking Neural Networks (RSNNs) employing the e-prop algorithm, and second, Reinforcement Learning (RL) for path planning tasks in feedforward networks using a modified version of the same learning rule. System-level simulations, accounting for various device and system-level non-idealities, confirm that these synapses offer a robust solution for the fast, compact, and energy-efficient implementation of advanced learning rules in neuromorphic hardware.
Open AccessNeoHebbian Synapses to Accelerate Online Training of Neuromorphic Hardware
Neuromorphic systems that employ advanced synaptic learning rules, such as the three-factor learning rule, require synaptic devices of increased complexity. Herein, a novel neoHebbian artificial synapse utilizing ReRAM devices has been proposed and experimentally validated to meet this demand. This synapse features two distinct state variables: a neuron coupling weight and an "eligibility trace" that dictates synaptic weight updates. The coupling weight is encoded in the ReRAM conductance, while the "eligibility trace" is encoded in the local temperature of the ReRAM and is modulated by applying voltage pulses to a physically co-located resistive heating element. The utility of the proposed synapse has been investigated using two representative tasks: first, temporal signal classification using Recurrent Spiking Neural Networks (RSNNs) employing the e-prop algorithm, and second, Reinforcement Learning (RL) for path planning tasks in feedforward networks using a modified version of the same learning rule. System-level simulations, accounting for various device and system-level non-idealities, confirm that these synapses offer a robust solution for the fast, compact, and energy-efficient implementation of advanced learning rules in neuromorphic hardware.
- 11 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 71 (9) , 5775-5778
High-Performance TiN/TaOx/TiN Selectors With Short-Term Memory Characteristics
We present TiN/TaOx/TiN device structure showing varistor and short-term memory (STM) characteristics that could be useful for selector applications and temporal processing in one-selector one-RRAM (1S1R) architectures. The devices show high selectivity ( $10^{{3}}$ ) and excellent endurance (one million cycles) and can sustain high-voltage stress ( $10^{{3}}$ s). The varistor behavior persists with device scaling down to $2.25~\mu $ m2 and thickness as thin as 2 nm. Moreover, the device exhibits a volatile switching behavior under pulse programming that could offer an additional analog output from single-digital input feed into the integrated 1S1R arrays.
- 11 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 71 (9) , 5775-5778
High-Performance TiN/TaOx/TiN Selectors With Short-Term Memory Characteristics
We present TiN/TaOx/TiN device structure showing varistor and short-term memory (STM) characteristics that could be useful for selector applications and temporal processing in one-selector one-RRAM (1S1R) architectures. The devices show high selectivity ( $10^{{3}}$ ) and excellent endurance (one million cycles) and can sustain high-voltage stress ( $10^{{3}}$ s). The varistor behavior persists with device scaling down to $2.25~\mu $ m2 and thickness as thin as 2 nm. Moreover, the device exhibits a volatile switching behavior under pulse programming that could offer an additional analog output from single-digital input feed into the integrated 1S1R arrays.
- 31 January 2024
- Published by American Chemical Society (ACS) in ACS Applied Electronic Materials
- Vol. 6 (2) , 1299-1307
Memristive Dendritic Device with a Chemical Vapor-Deposited Monolayer MoS2 Film
- 31 January 2024
- Published by American Chemical Society (ACS) in ACS Applied Electronic Materials
- Vol. 6 (2) , 1299-1307
Memristive Dendritic Device with a Chemical Vapor-Deposited Monolayer MoS2 Film