Huimin Zhang
Huimin Zhang has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
Nan Zhang
East China Normal University,School of Integrated Circuits,Shanghai,China
6 shared publicationsXiaofeng Zhou
East China Normal University,School of Integrated Circuits,Shanghai,China
5 shared publicationsZhourui Liu
East China Normal University,School of Integrated Circuits,Shanghai,China
4 shared publications- Xiaofeng Zhou
School of Integrated Circuits, East China Normal University 3 , Shanghai 200241
3 shared publications Huanxi Zheng
2 shared publicationsWanghuai Xu
2 shared publications
Latest Publications
- Open Access
- 24 September 2026
- Published by MDPI AG in AI for Engineering
- Vol. 1 (3) , 12
VMEMSTNet: Thermodynamics-Inspired Cross-Modal Translation for Robust Ultra-Low-Resolution Thermal Recognition
Ultra-low-resolution (ULR) thermal array sensors (e.g., ) offer a privacy-preserving and cost-effective paradigm for intelligent edge perception. However, extracting robust spatial features is severely hindered by extreme spatial aliasing, background thermal interference, and environmental temperature drift. In this paper, we propose VMEMSTNet, an ultra-lightweight framework that tackles the ULR perception bottleneck through a thermodynamics-inspired cross-modal learning strategy. Rather than treating ULR recognition purely as a computer vision task, our approach integrates the underlying sensing physics. To bridge the resolution gap, we leverage a high-resolution visual modality strictly as offline cross-modal guidance. Specifically, a YOLOv8n-seg teacher provides spatial shape priors via Gaussian thermal diffusion—explicitly mimicking physical heat dissipation for spatial knowledge translation—while a MobileNetV2 teacher guides semantic probability distillation. Furthermore, a thermodynamics-inspired preprocessing pipeline fuses dynamic quantile-based normalization with absolute temperature difference () to maintain contrast consistency across environmental drifts. Using hand gesture recognition as a representative spatial classification task, experimental results demonstrate that VMEMSTNet achieves 95.7% accuracy. Remarkably, it outperforms heavier generic edge models while utilizing only 20.0 k parameters and 0.135 MFLOPs. By effectively decoupling spatial semantics from thermodynamic noise, this work establishes a robust, hardware-aware deployment pathway for ULR thermal perception on resource-constrained microcontrollers.- 24 September 2026
- Published by MDPI AG in AI for Engineering
- Vol. 1 (3) , 12
Open AccessVMEMSTNet: Thermodynamics-Inspired Cross-Modal Translation for Robust Ultra-Low-Resolution Thermal Recognition
Ultra-low-resolution (ULR) thermal array sensors (e.g., ) offer a privacy-preserving and cost-effective paradigm for intelligent edge perception. However, extracting robust spatial features is severely hindered by extreme spatial aliasing, background thermal interference, and environmental temperature drift. In this paper, we propose VMEMSTNet, an ultra-lightweight framework that tackles the ULR perception bottleneck through a thermodynamics-inspired cross-modal learning strategy. Rather than treating ULR recognition purely as a computer vision task, our approach integrates the underlying sensing physics. To bridge the resolution gap, we leverage a high-resolution visual modality strictly as offline cross-modal guidance. Specifically, a YOLOv8n-seg teacher provides spatial shape priors via Gaussian thermal diffusion—explicitly mimicking physical heat dissipation for spatial knowledge translation—while a MobileNetV2 teacher guides semantic probability distillation. Furthermore, a thermodynamics-inspired preprocessing pipeline fuses dynamic quantile-based normalization with absolute temperature difference () to maintain contrast consistency across environmental drifts. Using hand gesture recognition as a representative spatial classification task, experimental results demonstrate that VMEMSTNet achieves 95.7% accuracy. Remarkably, it outperforms heavier generic edge models while utilizing only 20.0 k parameters and 0.135 MFLOPs. By effectively decoupling spatial semantics from thermodynamic noise, this work establishes a robust, hardware-aware deployment pathway for ULR thermal perception on resource-constrained microcontrollers. - 2 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Performance Optimization of Piezoelectric MEMS Speaker with Cantilever Diaphragm Array
This paper presents the concept to optimize the performance of a piezoelectric MEMS speaker with a cantilever diaphragm array by delicate structure design. The optimization focuses on the dimensions design of each cantilever unit of the array to balance the sound pressure level and frequency range under the condition of a limited area. The optimized design was first proposed based on the simulation and followingly a proof of experiment is conducted by fabricating a piezoelectric cantilever diaphragm array on a SOI wafer. As a result, measurements in the free field illustrate that the maximum SPL of the proposed microspeaker reaches 99 dB in the range of 200 Hz – 20 kHz with 3.0 V p-p driving voltage.
- 2 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Performance Optimization of Piezoelectric MEMS Speaker with Cantilever Diaphragm Array
This paper presents the concept to optimize the performance of a piezoelectric MEMS speaker with a cantilever diaphragm array by delicate structure design. The optimization focuses on the dimensions design of each cantilever unit of the array to balance the sound pressure level and frequency range under the condition of a limited area. The optimized design was first proposed based on the simulation and followingly a proof of experiment is conducted by fabricating a piezoelectric cantilever diaphragm array on a SOI wafer. As a result, measurements in the free field illustrate that the maximum SPL of the proposed microspeaker reaches 99 dB in the range of 200 Hz – 20 kHz with 3.0 V p-p driving voltage.
- 2 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Spice Modeling of a Transistor-Like Droplet-Based Electricity Generator (DEG)
Recently, droplet-based electricity generators (DEG) emerged as a promising solution to harvest clean and sustainable energy. The DEG consists of a top electrode, a dielectric layer, and a bottom electrode, which effectively transforms randomly distributed water kinetic energy into electrical energy. In this paper, a SPICE model of the DEG is constructed to fully simulate this device from a circuit perspective. Followingly, a proof of experiment is conducted by measuring the real-time output of the DEG with various types of droplets. The proposed SPICE model is well verified in agreement with experimental results. The presented model is expected to serve as an important guidance for DEGs’ applications.
- 2 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Spice Modeling of a Transistor-Like Droplet-Based Electricity Generator (DEG)
Recently, droplet-based electricity generators (DEG) emerged as a promising solution to harvest clean and sustainable energy. The DEG consists of a top electrode, a dielectric layer, and a bottom electrode, which effectively transforms randomly distributed water kinetic energy into electrical energy. In this paper, a SPICE model of the DEG is constructed to fully simulate this device from a circuit perspective. Followingly, a proof of experiment is conducted by measuring the real-time output of the DEG with various types of droplets. The proposed SPICE model is well verified in agreement with experimental results. The presented model is expected to serve as an important guidance for DEGs’ applications.
- 25 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 45 (5) , 833-836
8F² Ternary Content Addressable Memory Array Utilizing Interface Passivated Ge Memory-Diodes With 2 × 10⁵ Self-Rectifying Ratio
This work presents the first demonstration of the transistor-free memory-diode (MD) array fabricated on Ge-on-Insulator (GeOI), which can work as an ultra-compact 8 F 2 ternary content addressable memory (TCAM). Through the wafer bonding and thinning techniques, the Ge layer in the GeOI structure maintains the monocrystalline quality originated from bulk Ge to ensure the Fermi-level pinning effect, forming an effective Schottky barrier with low reverse-bias leakage currents. Furthermore, an ultra-thin Y-doped GeO x layer is introduced to suppress the surface-state current, enabling a higher rectifying ratio. Thanks to the ultra-high rectifying (> 2 × 10 5 ) and sufficient on/off ratios (> 500), the 4-bits MD-TCAM array shows over 2 orders differences between the match and mismatch currents, which experimentally verifies the feasibility and accuracy of parallel data search.
- 25 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 45 (5) , 833-836
8F² Ternary Content Addressable Memory Array Utilizing Interface Passivated Ge Memory-Diodes With 2 × 10⁵ Self-Rectifying Ratio
This work presents the first demonstration of the transistor-free memory-diode (MD) array fabricated on Ge-on-Insulator (GeOI), which can work as an ultra-compact 8 F 2 ternary content addressable memory (TCAM). Through the wafer bonding and thinning techniques, the Ge layer in the GeOI structure maintains the monocrystalline quality originated from bulk Ge to ensure the Fermi-level pinning effect, forming an effective Schottky barrier with low reverse-bias leakage currents. Furthermore, an ultra-thin Y-doped GeO x layer is introduced to suppress the surface-state current, enabling a higher rectifying ratio. Thanks to the ultra-high rectifying (> 2 × 10 5 ) and sufficient on/off ratios (> 500), the 4-bits MD-TCAM array shows over 2 orders differences between the match and mismatch currents, which experimentally verifies the feasibility and accuracy of parallel data search.
- 5 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 359-364
Finite element simulation of Cu-SiO2 direct hybrid bonding: impact of via on bonding integration
In this work, the influence of vias design on hybrid bonding is explored by finite element analysis approach. The impact of via shape, width, depth, and pad dishing on the bonding quality are firstly investigated by taking the copper pad with one via into account. The results indicate that the bonding quality is determined by the via width, depth, and pad dishing, while the via shape is of little influence. Followingly, the simulation about one via design is extended to the via array design. An optimized via array arrangement is achieved with a high contact area ratio and uniform interfacial stress at last. The simulation results in this paper are expected to provide guidelines for optimizing the hybrid bonding process.
- 5 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 359-364
Finite element simulation of Cu-SiO2 direct hybrid bonding: impact of via on bonding integration
In this work, the influence of vias design on hybrid bonding is explored by finite element analysis approach. The impact of via shape, width, depth, and pad dishing on the bonding quality are firstly investigated by taking the copper pad with one via into account. The results indicate that the bonding quality is determined by the via width, depth, and pad dishing, while the via shape is of little influence. Followingly, the simulation about one via design is extended to the via array design. An optimized via array arrangement is achieved with a high contact area ratio and uniform interfacial stress at last. The simulation results in this paper are expected to provide guidelines for optimizing the hybrid bonding process.
- 17 December 2022
- Published by Elsevier in Nano Energy
- Vol. 106
Performance transition in droplet-based electricity generator with optimized top electrode arrangements
- 17 December 2022
- Published by Elsevier in Nano Energy
- Vol. 106
Performance transition in droplet-based electricity generator with optimized top electrode arrangements
- Open Access
A droplet‐based electricity generator with ultrahigh instantaneous output and short charging time
The past several years have witnessed the rapid development in effectively transforming randomly distributed water kinetic energy into electrical energy, especially triggered by the emergence of droplet‐based electricity generators (DEG). Despite this, it still suffers from relatively low average power density, which is also achieved at the cost of long charging time, the time to reach stable and saturated surface charge density either through continuous droplet impingement or precharging. Although the harvested energy per droplet in DEG remains as the dominant metric, ultrahigh instantaneous output and short charging time are equally important in some specialized applications such as instantaneous luminescence. Here, we conduct systematical modeling and optimization to build the link between the hydrodynamic and electrical systems, which enables us to determine ultrahigh instantaneous output and short charging time by tailoring parameters such as dielectric layer thickness, droplet ion concentration, and external load. We envision that this strategy in achieving ultrahigh instantaneous output as well as shortening charging time would provide insights and design routes for water energy harvesting.
Open AccessA droplet‐based electricity generator with ultrahigh instantaneous output and short charging time
The past several years have witnessed the rapid development in effectively transforming randomly distributed water kinetic energy into electrical energy, especially triggered by the emergence of droplet‐based electricity generators (DEG). Despite this, it still suffers from relatively low average power density, which is also achieved at the cost of long charging time, the time to reach stable and saturated surface charge density either through continuous droplet impingement or precharging. Although the harvested energy per droplet in DEG remains as the dominant metric, ultrahigh instantaneous output and short charging time are equally important in some specialized applications such as instantaneous luminescence. Here, we conduct systematical modeling and optimization to build the link between the hydrodynamic and electrical systems, which enables us to determine ultrahigh instantaneous output and short charging time by tailoring parameters such as dielectric layer thickness, droplet ion concentration, and external load. We envision that this strategy in achieving ultrahigh instantaneous output as well as shortening charging time would provide insights and design routes for water energy harvesting.