Piyush Kumar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Piyush Kumar has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
- Azad Naeemi
School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA
13 shared publications - Siri Narla
School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA
5 shared publications - Ann Franchesca Laguna
Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, IN, USA
2 shared publications - Da Eun Shim
School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA
2 shared publications - Daniel C. Ralph
Department of Physics, Cornell University, Ithaca, NY, USA
2 shared publications - Michael Niemier
University of Notre Dame
2 shared publications
Latest Publications
- 30 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 73 (6) , 3565-3571
Modeling and Optimization of Two-Terminal Spin-Orbit-Torque MRAM
This article presents physical modeling and benchmarking for two-terminal spin-orbit-torque magnetic random access memory (2T SOT-MRAM). The results indicate that the common spin-orbit torque (SOT) materials that provide only in-plane torque can provide little to no improvement over spin-transfer-torque (STT) MRAM in terms of write energy. However, emerging SOT materials that provide out-of-plane torques with efficiencies as small as 0.1 can result in significant improvements in the write energy for such two-terminal devices, especially when the magnet lateral dimensions are scaled down to 30 or 20 nm. Additionally, a novel 2T SOT-MRAM device is proposed that can increase the path electrons passing through the SOT layer, hence increasing the generated spin current and the energy efficiency of the device. Our benchmarking results indicate that an out-of-plane SOT efficiency of 0.051 for 20-nm-wide devices can result in write energies approaching SRAM at the 7-nm technology node.
- 30 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 73 (6) , 3565-3571
Modeling and Optimization of Two-Terminal Spin-Orbit-Torque MRAM
This article presents physical modeling and benchmarking for two-terminal spin-orbit-torque magnetic random access memory (2T SOT-MRAM). The results indicate that the common spin-orbit torque (SOT) materials that provide only in-plane torque can provide little to no improvement over spin-transfer-torque (STT) MRAM in terms of write energy. However, emerging SOT materials that provide out-of-plane torques with efficiencies as small as 0.1 can result in significant improvements in the write energy for such two-terminal devices, especially when the magnet lateral dimensions are scaled down to 30 or 20 nm. Additionally, a novel 2T SOT-MRAM device is proposed that can increase the path electrons passing through the SOT layer, hence increasing the generated spin current and the energy efficiency of the device. Our benchmarking results indicate that an out-of-plane SOT efficiency of 0.051 for 20-nm-wide devices can result in write energies approaching SRAM at the 7-nm technology node.
- 5 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 47 (2) , 387-390
Transition Metal Dichalcogenide Thin-Film Resistors for Content Addressable Memories With Improved Search Resolution
Despite the parallel in-memory search capabilities of content addressable memories (CAMs), their use in applications is constrained by their limited resolution that worsens as they are scaled to larger arrays or advanced nodes. In this work we present experimental and simulated results for a novel transition metal dichalcogenide-based back-end-of-line compatible reference resistor that can significantly improve the search resolution of CAMs implemented with CMOS and beyond- CMOS technologies to $\le 5$ -bits.
- 5 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 47 (2) , 387-390
Transition Metal Dichalcogenide Thin-Film Resistors for Content Addressable Memories With Improved Search Resolution
Despite the parallel in-memory search capabilities of content addressable memories (CAMs), their use in applications is constrained by their limited resolution that worsens as they are scaled to larger arrays or advanced nodes. In this work we present experimental and simulated results for a novel transition metal dichalcogenide-based back-end-of-line compatible reference resistor that can significantly improve the search resolution of CAMs implemented with CMOS and beyond- CMOS technologies to $\le 5$ -bits.
- Open Access
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 11, 139-147
Comprehensive Device to System Co-Design for SOT-MRAM at the 7 nm Node
This work presents a comprehensive spin-orbit torque (SOT) based random access memory (MRAM) design at the 7nm technology node, spanning from device-level characteristics to system-level power performance area (PPA). At the device-level, we show the trade-offs among the write current, error rate, and time, based on mircomagnetic simulations. Based on ASAP7 PDK design rules, we create the bit-cell and peripheral layouts for SOT-MRAM and design the entire array. In addition, we quantify various array-level trade-offs using full array SPICE circuit simulations based on layout-extracted parasitic netlists. This is then used to design the entire SOT-MRAM system along with a memory controller. Based on place and route, we evaluate the system-level PPA for various memory capacities, demonstrating bit-densities up to 14.8 Mb/mm2 and read bandwidths up to 2.98 GB/s. Our results show that increasing the memory size from 1 Mb to 16 Mb results in a performance degradation of ∼33-38% due to the impact of interconnect delay. As the results show that the performance of SOT-MRAM is limited by the interconnect delay, it is critical to co-optimize the device and interconnect technology to make SOT-MRAM a viable option at the advanced technology nodes. In addition, material discovery for field-free perpendicular magnetization switching in SOT devices based on out-of-plane spin torque is necessary to achieve SRAM level write energies.
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 11, 139-147
Open AccessComprehensive Device to System Co-Design for SOT-MRAM at the 7 nm Node
This work presents a comprehensive spin-orbit torque (SOT) based random access memory (MRAM) design at the 7nm technology node, spanning from device-level characteristics to system-level power performance area (PPA). At the device-level, we show the trade-offs among the write current, error rate, and time, based on mircomagnetic simulations. Based on ASAP7 PDK design rules, we create the bit-cell and peripheral layouts for SOT-MRAM and design the entire array. In addition, we quantify various array-level trade-offs using full array SPICE circuit simulations based on layout-extracted parasitic netlists. This is then used to design the entire SOT-MRAM system along with a memory controller. Based on place and route, we evaluate the system-level PPA for various memory capacities, demonstrating bit-densities up to 14.8 Mb/mm2 and read bandwidths up to 2.98 GB/s. Our results show that increasing the memory size from 1 Mb to 16 Mb results in a performance degradation of ∼33-38% due to the impact of interconnect delay. As the results show that the performance of SOT-MRAM is limited by the interconnect delay, it is critical to co-optimize the device and interconnect technology to make SOT-MRAM a viable option at the advanced technology nodes. In addition, material discovery for field-free perpendicular magnetization switching in SOT devices based on out-of-plane spin torque is necessary to achieve SRAM level write energies.
- 21 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Magnetics
- Vol. 61 (10) , 1-9
A 5T-2MTJ STT-Assisted Spin-Orbit-Torque-Based Ternary Content Addressable Memory for Hardware Accelerators
In this work, we present a novel non-volatile spin transfer torque (STT)-assisted spin–orbit torque (SOT)-based ternary content addressable memory (TCAM) with five transistors and two magnetic tunnel junctions (MTJs). We perform a comprehensive study of the proposed design from the device level to the application level. At the device level, various write characteristics such as the write error rate, time, and current have been obtained using micromagnetic simulations. The array-level search and write performance have been evaluated based on SPICE circuit simulations with layout extracted parasitics for bit-cells while also accounting for the impact of interconnect parasitics at the 7 nm technology node. A search error rate (SER) of $3.9 \times 10 {^{-}11 }$ is projected for exact search while accounting for various sources of variation in the design. In addition, the resolution of the search operation is quantified under various scenarios to understand the achievable quality of the approximate search operations. Application-level performance and accuracy of the proposed design have been evaluated and benchmarked against other state-of-the-art CAM designs in the context of a CAM-based recommendation system.
- 21 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Magnetics
- Vol. 61 (10) , 1-9
A 5T-2MTJ STT-Assisted Spin-Orbit-Torque-Based Ternary Content Addressable Memory for Hardware Accelerators
In this work, we present a novel non-volatile spin transfer torque (STT)-assisted spin–orbit torque (SOT)-based ternary content addressable memory (TCAM) with five transistors and two magnetic tunnel junctions (MTJs). We perform a comprehensive study of the proposed design from the device level to the application level. At the device level, various write characteristics such as the write error rate, time, and current have been obtained using micromagnetic simulations. The array-level search and write performance have been evaluated based on SPICE circuit simulations with layout extracted parasitics for bit-cells while also accounting for the impact of interconnect parasitics at the 7 nm technology node. A search error rate (SER) of $3.9 \times 10 {^{-}11 }$ is projected for exact search while accounting for various sources of variation in the design. In addition, the resolution of the search operation is quantified under various scenarios to understand the achievable quality of the approximate search operations. Application-level performance and accuracy of the proposed design have been evaluated and benchmarked against other state-of-the-art CAM designs in the context of a CAM-based recommendation system.
- 27 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 72 (4) , 1582-1588
GT3: An Open-Source 3-nm GAAFET PDK and Platform for End-to-End Evaluation of Emerging Technologies
In this article, we present a comprehensive end-to-end evaluation platform for various front-end-of-line (FEOL) and back-end-of-line (BEOL) technology options at the 3-nm technology node. Based on TCAD modeling of FE and BE, we have developed a 3-nm GAAFET-based process design kit (PDK). We have developed a 6-track standard cell library including 65 cells with a library height of 144 nm. Based on TCAD modeling of interconnects, we have evaluated the resistance of the entire BEOL stack using ruthenium for lower metal levels (M0–M3) and copper for higher metal levels (M4–M13). Based on place and route (PnR) studies using our PDK, we have analyzed the impact of high aspect ratio (AR) Ru interconnects at M2 and M3 in terms of performance using benchmark designs. Our results show that using Ru interconnects improves the circuit performance by up to 10.4% compared with Cu interconnects and that increasing the AR generally results in performance degradation due to the increase in capacitance and via resistances. We have observed a 5.9% and 4% degradation in performance for AES and LDPC, respectively, when moving from AR2 to AR6 local interconnects. However, adding an airgap can improve the higher AR Ru interconnect cases and AR4 with airgap case shows the most performance improvement with an overall 19.7% improvement compared with Cu. This case study also serves as an example that shows the importance of an end-to-end evaluation platform.
- 27 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 72 (4) , 1582-1588
GT3: An Open-Source 3-nm GAAFET PDK and Platform for End-to-End Evaluation of Emerging Technologies
In this article, we present a comprehensive end-to-end evaluation platform for various front-end-of-line (FEOL) and back-end-of-line (BEOL) technology options at the 3-nm technology node. Based on TCAD modeling of FE and BE, we have developed a 3-nm GAAFET-based process design kit (PDK). We have developed a 6-track standard cell library including 65 cells with a library height of 144 nm. Based on TCAD modeling of interconnects, we have evaluated the resistance of the entire BEOL stack using ruthenium for lower metal levels (M0–M3) and copper for higher metal levels (M4–M13). Based on place and route (PnR) studies using our PDK, we have analyzed the impact of high aspect ratio (AR) Ru interconnects at M2 and M3 in terms of performance using benchmark designs. Our results show that using Ru interconnects improves the circuit performance by up to 10.4% compared with Cu interconnects and that increasing the AR generally results in performance degradation due to the increase in capacitance and via resistances. We have observed a 5.9% and 4% degradation in performance for AES and LDPC, respectively, when moving from AR2 to AR6 local interconnects. However, adding an airgap can improve the higher AR Ru interconnect cases and AR4 with airgap case shows the most performance improvement with an overall 19.7% improvement compared with Cu. This case study also serves as an example that shows the importance of an end-to-end evaluation platform.
- 4 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 72 (1) , 240-246
Cross-Layer Modeling and Design of Content Addressable Memories in Advanced Technology Nodes for Similarity Search
In this article, we present a comprehensive design and benchmarking study of content addressable memory (CAM) at the 7-nm technology node in the context of similarity search applications. We design CAM cells based on static random access memory (SRAM), spin–orbit torque (SOT), and ferroelectric field effect transistor devices and from their layouts extract cell parasitics using state-of-the-art electronic design automation (EDA) tools. These parasitics are used to develop SPICE netlists to model search operations. We use a CAM-based dataset search and a sequential recommendation system (RS) to highlight the application-level performance degradation due to interconnect parasitics. We propose and evaluate two solutions to mitigate interconnect effects.
- 4 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 72 (1) , 240-246
Cross-Layer Modeling and Design of Content Addressable Memories in Advanced Technology Nodes for Similarity Search
In this article, we present a comprehensive design and benchmarking study of content addressable memory (CAM) at the 7-nm technology node in the context of similarity search applications. We design CAM cells based on static random access memory (SRAM), spin–orbit torque (SOT), and ferroelectric field effect transistor devices and from their layouts extract cell parasitics using state-of-the-art electronic design automation (EDA) tools. These parasitics are used to develop SPICE netlists to model search operations. We use a CAM-based dataset search and a sequential recommendation system (RS) to highlight the application-level performance degradation due to interconnect parasitics. We propose and evaluate two solutions to mitigate interconnect effects.
- Open Access
- 23 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 10, 13-21
Impact of Technology Scaling and Back-End-of-the-Line Technology Solutions on Magnetic Random-Access Memories
While magnetic random-access memories (MRAMs) are promising thanks to their non-volatility, relatively fast speeds, and high endurance, there are major challenges in adopting them for the advanced technology nodes. One of the major challenges in scaling MRAM devices is caused by the ever increasing resistances of interconnects. In this paper, we first study the impact of shrunk interconnect dimensions on MRAM performance at various technology nodes. Then, we investigate the impact of various potential back-end-of-the-line (BEOL) technology solutions at the 7nm node. Based on interconnect resistance values from TCAD simulations and MRAM device characteristics from experimentally validated/calibrated physical models, we quantify the potential array-level performance of MRAM using SPICE simulations. We project that some potential BEOL technology solutions can reduce the write energy by up to 34.6% with spin-orbit torque (SOT) MRAM, and 29.0% with spin-transfer torque (STT) MRAM. We also observe up to 21.4% reduction in the read energy of the SOT-MRAM arrays.
- 23 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 10, 13-21
Open AccessImpact of Technology Scaling and Back-End-of-the-Line Technology Solutions on Magnetic Random-Access Memories
While magnetic random-access memories (MRAMs) are promising thanks to their non-volatility, relatively fast speeds, and high endurance, there are major challenges in adopting them for the advanced technology nodes. One of the major challenges in scaling MRAM devices is caused by the ever increasing resistances of interconnects. In this paper, we first study the impact of shrunk interconnect dimensions on MRAM performance at various technology nodes. Then, we investigate the impact of various potential back-end-of-the-line (BEOL) technology solutions at the 7nm node. Based on interconnect resistance values from TCAD simulations and MRAM device characteristics from experimentally validated/calibrated physical models, we quantify the potential array-level performance of MRAM using SPICE simulations. We project that some potential BEOL technology solutions can reduce the write energy by up to 34.6% with spin-orbit torque (SOT) MRAM, and 29.0% with spin-transfer torque (STT) MRAM. We also observe up to 21.4% reduction in the read energy of the SOT-MRAM arrays.
- 29 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 70 (2) , 506-513
Design of a Compact Spin-Orbit-Torque-Based Ternary Content Addressable Memory
This article presents the design of a novel and compact spin-orbit torque (SOT)-based ternary content addressable memory (TCAM). Experimentally validated/calibrated micromagnetic and macrospin simulations have been used to quantify various tradeoffs regarding the write operation, such as write energy, error rate, and retention time. SPICE simulations incorporating various sources of variability are used to evaluate search operations, optimize the proposed novel TCAM cell based on SOT magnetic random access memory (SOT-MRAM), and benchmark it against static random access memory (SRAM)- and FeFET-based TCAMs. We show low search error rates (SERs) (< 10−4) while considering various sources of variability for four transistors- and two magnetic tunnel junctions (MTJs)-based TCAM array.
- 29 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 70 (2) , 506-513
Design of a Compact Spin-Orbit-Torque-Based Ternary Content Addressable Memory
This article presents the design of a novel and compact spin-orbit torque (SOT)-based ternary content addressable memory (TCAM). Experimentally validated/calibrated micromagnetic and macrospin simulations have been used to quantify various tradeoffs regarding the write operation, such as write energy, error rate, and retention time. SPICE simulations incorporating various sources of variability are used to evaluate search operations, optimize the proposed novel TCAM cell based on SOT magnetic random access memory (SOT-MRAM), and benchmark it against static random access memory (SRAM)- and FeFET-based TCAMs. We show low search error rates (SERs) (< 10−4) while considering various sources of variability for four transistors- and two magnetic tunnel junctions (MTJs)-based TCAM array.
- Open Access
- 20 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 8 (2) , 185-193
High-Density Spin–Orbit Torque Magnetic Random Access Memory With Voltage-Controlled Magnetic Anisotropy/Spin-Transfer Torque Assist
This paper explores an area saving scheme for spin-orbit torque (SOT) magnetic random-access memory (MRAM) by sharing the SOT channel and write transistor among multiple magnetic tunnel junctions (MTJs). We use two write mechanisms to selectively write the MTJs; voltage-controlled magnetic anisotropy (VCMA) assisted write in the presence of an external magnetic field and field-free spin-transfer torque (STT) assisted write. Using micromagnetic simulations that are augmented by the rare-event enhancement, we study various tradeoffs among write current, time and energy, write error rate (WER), and the number of MTJs on an SOT channel. We quantify the issue of IR drop on the SOT channel as a function of the SOT layer thickness and number of MTJs. Our results show having more than 4 MTJs on an SOT channel poses major challenges in terms of IR drop and WER. In addition, we evaluate the impact of the proposed scheme on read performance.
- 20 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
- Vol. 8 (2) , 185-193
Open AccessHigh-Density Spin–Orbit Torque Magnetic Random Access Memory With Voltage-Controlled Magnetic Anisotropy/Spin-Transfer Torque Assist
This paper explores an area saving scheme for spin-orbit torque (SOT) magnetic random-access memory (MRAM) by sharing the SOT channel and write transistor among multiple magnetic tunnel junctions (MTJs). We use two write mechanisms to selectively write the MTJs; voltage-controlled magnetic anisotropy (VCMA) assisted write in the presence of an external magnetic field and field-free spin-transfer torque (STT) assisted write. Using micromagnetic simulations that are augmented by the rare-event enhancement, we study various tradeoffs among write current, time and energy, write error rate (WER), and the number of MTJs on an SOT channel. We quantify the issue of IR drop on the SOT channel as a function of the SOT layer thickness and number of MTJs. Our results show having more than 4 MTJs on an SOT channel poses major challenges in terms of IR drop and WER. In addition, we evaluate the impact of the proposed scheme on read performance.
- 16 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 70 (2) , 789-795
A Drift-Diffusion Based Modeling and Optimization Framework for Nanoscale Spin-Orbit Torque Devices
We present a comprehensive set of experimentally validated/calibrated models that capture the physics of the nanoscale spin-orbit torque (SOT) devices. We consider various effects that are prominent at nanoscale including incomplete current redistribution, interface spin mixing, and nonuniform resistivity that were ignored in the prior modeling efforts. We develop a formalism based on drift-diffusion equations and the transfer matrix method to accurately estimate spin current distribution. We utilize finite element simulations to accurately calculate electric current density and field, and see considerable differences to the results from the simplified lumped model commonly used. For example, we calculate ${\sim}\text{20}\%$ smaller spin current in a 15-nm-wide ferromagnet with a 4-nm-thick $\beta$ -W SOT channel. We further account for the effect of interface scattering on resistivity. Finally, we quantify the optimal SOT-layer thickness that minimizes the write energy as a function of spin diffusion length and conductivity of SOT materials. We show that the optimal thickness increases with the spin diffusion length and decreases with the conductivity.
- 16 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Electron Devices
- Vol. 70 (2) , 789-795
A Drift-Diffusion Based Modeling and Optimization Framework for Nanoscale Spin-Orbit Torque Devices
We present a comprehensive set of experimentally validated/calibrated models that capture the physics of the nanoscale spin-orbit torque (SOT) devices. We consider various effects that are prominent at nanoscale including incomplete current redistribution, interface spin mixing, and nonuniform resistivity that were ignored in the prior modeling efforts. We develop a formalism based on drift-diffusion equations and the transfer matrix method to accurately estimate spin current distribution. We utilize finite element simulations to accurately calculate electric current density and field, and see considerable differences to the results from the simplified lumped model commonly used. For example, we calculate ${\sim}\text{20}\%$ smaller spin current in a 15-nm-wide ferromagnet with a 4-nm-thick $\beta$ -W SOT channel. We further account for the effect of interface scattering on resistivity. Finally, we quantify the optimal SOT-layer thickness that minimizes the write energy as a function of spin diffusion length and conductivity of SOT materials. We show that the optimal thickness increases with the spin diffusion length and decreases with the conductivity.