Wenwen Zheng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Wenwen Zheng has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Mario Lanza
Department of Materials Science and Engineering National University of Singapore Singapore Singapore
5 shared publications - Yaqing Shen
Physical Science and Engineering Division, King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia
3 shared publications - Kaichen Zhu
Institute of Optoelectronics, Fudan University, Shanghai, China
2 shared publications - Sebastian Pazos
Department of Materials Science and Engineering, National University of Singapore, Singapore, Singapore
2 shared publications Shiqin Liu
School of Computer Science, China University of Geosciences, Wuhan, China
2 shared publications- Shiyuan Feng
School of Computer Science, China University of Geosciences, Wuhan, China
2 shared publications
Latest Publications
- 20 May 2026
- Published by Springer Nature in Nature Sensors
- Vol. 1 (6) , 479-480
Self‑powered analog sensing‑as‑computation with diffusive and drift memristors
- 20 May 2026
- Published by Springer Nature in Nature Sensors
- Vol. 1 (6) , 479-480
Self‑powered analog sensing‑as‑computation with diffusive and drift memristors
- Open Access
- 15 August 2025
- Published by Elsevier in Applied Surface Science Advances
- Vol. 29
High-yield photolithography protocol to pattern metallic electrodes on 2D materials without adhesive metallic layers
- 15 August 2025
- Published by Elsevier in Applied Surface Science Advances
- Vol. 29
Open AccessHigh-yield photolithography protocol to pattern metallic electrodes on 2D materials without adhesive metallic layers
- 14 August 2025
- Published by Wiley in Advanced Materials
- Vol. 37 (43) , e10501
Scalable Production of Highly‐Reliable Graphene‐Based Microchips
Graphene is a gapless 2D material that could be used to fabricate superior electronic devices and circuits, particularly useful in the fields of telecommunication and sensing. While promising performance has been demonstrated at the laboratory scale, graphene integrated circuits at the wafer level suffer from poor reliability due to native defects, especially at interfaces with dielectrics and electrodes. Here, we show the fabrication of highly reliable graphene‐based microchips, containing transistors and frequency doublers, on 200 mm wafers through a multi‐project wafer tape‐out. Our transistors use multilayer hexagonal boron nitride (hBN) as gate dielectric, and they exhibit record performance in terms of reliability. In particular, our hBN/graphene transistors show ultra‐low hysteresis below 20 mV and negligible shifts of the on‐state current and the charge neutrality point even after 2100 cycles. The ultra‐stable response of our hBN/graphene transistors contrasts with that of devices using metal‐oxide gate dielectrics (HfO 2 , Al 2 O 3 ), which exhibit severe degradation after a few dozens of cycles. These results, consistent across multiple devices, show low variability and demonstrate a scalable process for mass production of graphene‐based microchips.
- 14 August 2025
- Published by Wiley in Advanced Materials
- Vol. 37 (43) , e10501
Scalable Production of Highly‐Reliable Graphene‐Based Microchips
Graphene is a gapless 2D material that could be used to fabricate superior electronic devices and circuits, particularly useful in the fields of telecommunication and sensing. While promising performance has been demonstrated at the laboratory scale, graphene integrated circuits at the wafer level suffer from poor reliability due to native defects, especially at interfaces with dielectrics and electrodes. Here, we show the fabrication of highly reliable graphene‐based microchips, containing transistors and frequency doublers, on 200 mm wafers through a multi‐project wafer tape‐out. Our transistors use multilayer hexagonal boron nitride (hBN) as gate dielectric, and they exhibit record performance in terms of reliability. In particular, our hBN/graphene transistors show ultra‐low hysteresis below 20 mV and negligible shifts of the on‐state current and the charge neutrality point even after 2100 cycles. The ultra‐stable response of our hBN/graphene transistors contrasts with that of devices using metal‐oxide gate dielectrics (HfO 2 , Al 2 O 3 ), which exhibit severe degradation after a few dozens of cycles. These results, consistent across multiple devices, show low variability and demonstrate a scalable process for mass production of graphene‐based microchips.
- 31 July 2025
- Published by Springer Nature in Nature Nanotechnology
- Vol. 20 (11) , 1596-1604
On-chip direct synthesis of boron nitride memristors
Two-dimensional materials hold promise for advanced complementary metal–oxide–semiconductor (CMOS) and beyond-CMOS electronics, including neuromorphic and in-memory computing. Hexagonal boron nitride (hBN) is particularly attractive for non-volatile resistive-switching devices (that is, memristors) due to its outstanding electronic, mechanical and chemical stability. However, integrating hBN memristors with Si-CMOS electronics faces challenges as it requires either high-temperature synthesis (exceeding thermal budgets) or transfer methods that introduce defects, impacting device performance and reliability. Here we introduce the synthesis of hBN films at CMOS-compatible temperatures (16 states) and low-frequency noise performance with minimal random telegraph noise. Furthermore, we directly integrate memristive devices on industrial CMOS test vehicles to demonstrate excellent endurance, achieving millions of programming cycles with a high technology readiness level. This represents an important step towards the wafer-scale CMOS integration of hBN-memristor-based electronics.
- 31 July 2025
- Published by Springer Nature in Nature Nanotechnology
- Vol. 20 (11) , 1596-1604
On-chip direct synthesis of boron nitride memristors
Two-dimensional materials hold promise for advanced complementary metal–oxide–semiconductor (CMOS) and beyond-CMOS electronics, including neuromorphic and in-memory computing. Hexagonal boron nitride (hBN) is particularly attractive for non-volatile resistive-switching devices (that is, memristors) due to its outstanding electronic, mechanical and chemical stability. However, integrating hBN memristors with Si-CMOS electronics faces challenges as it requires either high-temperature synthesis (exceeding thermal budgets) or transfer methods that introduce defects, impacting device performance and reliability. Here we introduce the synthesis of hBN films at CMOS-compatible temperatures (16 states) and low-frequency noise performance with minimal random telegraph noise. Furthermore, we directly integrate memristive devices on industrial CMOS test vehicles to demonstrate excellent endurance, achieving millions of programming cycles with a high technology readiness level. This represents an important step towards the wafer-scale CMOS integration of hBN-memristor-based electronics.
- Open Access
- 15 May 2023
- Published by AIP Publishing in Applied Physics Letters
- Vol. 122 (20)
Conductance quantization in h-BN memristors
Memristive devices made of multilayer hexagonal boron nitride are attracting much attention for information storage, computation, encryption, and communication. Generating multilevel operation in these memristors would be an important added value for the field of neuromorphic computation, but the main phenomenon producing it (i.e., conductance quantization) is not well understood for this material. Here, we analyze hexagonal boron nitride memristive devices with metal electrodes operating in their quasistatic and transient regimes. Conductance quantization effects were observed both in I-V curves measured under ramped voltage stress and in the reset processes driven by voltage pulse trains. Polar and time lag plots are employed to shed light on the physics behind the device resistive switching. A model is proposed to reproduce the electrical characteristics of progressive reset processes and the discretization of the device conductance linked to quantum effects.
- 15 May 2023
- Published by AIP Publishing in Applied Physics Letters
- Vol. 122 (20)
Open AccessConductance quantization in h-BN memristors
Memristive devices made of multilayer hexagonal boron nitride are attracting much attention for information storage, computation, encryption, and communication. Generating multilevel operation in these memristors would be an important added value for the field of neuromorphic computation, but the main phenomenon producing it (i.e., conductance quantization) is not well understood for this material. Here, we analyze hexagonal boron nitride memristive devices with metal electrodes operating in their quasistatic and transient regimes. Conductance quantization effects were observed both in I-V curves measured under ramped voltage stress and in the reset processes driven by voltage pulse trains. Polar and time lag plots are employed to shed light on the physics behind the device resistive switching. A model is proposed to reproduce the electrical characteristics of progressive reset processes and the discretization of the device conductance linked to quantum effects.
- 1 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1215-1221
A User Behavior-Based Random Distribution Scheme for Adversarial Example Generated CAPTCHA
The widespread and mature application of deep learning technology on human behavior analysis leads to the possibility that the Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA) may not be able to accurately determine whether or not the user is human, which makes the distributed denial-of-service (DDoS) attacks launched by deep learning-based malicious automatic computer programs become possible. The CAPTCHA generated by adversarial examples can defend against the automatic identification by the deep learning-based method, however, if the deployed CAPTCHAs are all adversarial examples solely, the adversary has the chance to collect sufficient adversarial examples, which also causes security issues. In this paper, a user behavior-based random distribution scheme for adversarial example generated CAPTCHA is proposed to tackle the above sole distribution problem. Specifically, we generated two kinds of CAPTCHAs, one is normal and the other is generated by the fast gradient signed method (FGSM), i.e., the adversarial examples. Meanwhile, user behaviors are analyzed and associated with a reasonable probability model. According to the probability selection, the normal CAPTCHAs or strong CAPTCHAs will be delivered to the users. The experimental results illustrate that our scheme has an excellent ability to distinguish computers from humans. Thus, it can protect the CAPTCHA systems from DDoS attacks.
- 1 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1215-1221
A User Behavior-Based Random Distribution Scheme for Adversarial Example Generated CAPTCHA
The widespread and mature application of deep learning technology on human behavior analysis leads to the possibility that the Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA) may not be able to accurately determine whether or not the user is human, which makes the distributed denial-of-service (DDoS) attacks launched by deep learning-based malicious automatic computer programs become possible. The CAPTCHA generated by adversarial examples can defend against the automatic identification by the deep learning-based method, however, if the deployed CAPTCHAs are all adversarial examples solely, the adversary has the chance to collect sufficient adversarial examples, which also causes security issues. In this paper, a user behavior-based random distribution scheme for adversarial example generated CAPTCHA is proposed to tackle the above sole distribution problem. Specifically, we generated two kinds of CAPTCHAs, one is normal and the other is generated by the fast gradient signed method (FGSM), i.e., the adversarial examples. Meanwhile, user behaviors are analyzed and associated with a reasonable probability model. According to the probability selection, the normal CAPTCHAs or strong CAPTCHAs will be delivered to the users. The experimental results illustrate that our scheme has an excellent ability to distinguish computers from humans. Thus, it can protect the CAPTCHA systems from DDoS attacks.
- 1 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 421-428
Exploration of the Influence on Training Deep Learning Models by Watermarked Image Dataset
Deep learning has achieved great success in various applications with the help of a large scale of datasets. As a result, sharing the valuable big data that can be applied to training deep learning models is of essential importance currently. However, how to claim ownership and protect the copyright of image big data during the sharing process is still a vital issue that should be addressed. The application of digital watermarks can protect the copyright of image data, at the same time, it also degrades the image quality at the same time. As for invisible digital watermarks, the higher the watermark embedding intensity causes the greater the host image changes. Therefore, the performance of deep learning models may decline due to using the watermarked training set. In this paper, we evaluate the influences of different embedding intensities of various watermarking algorithms on several mainstream models and conclude how the watermarking intensity affects the model training. Besides, referring to watermarking algorithms that have been proposed, we proposed a novel discrete Fourier transform-based watermarking algorithm that can achieve image copyright protection yet maintain the utility of models.
- 1 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 421-428
Exploration of the Influence on Training Deep Learning Models by Watermarked Image Dataset
Deep learning has achieved great success in various applications with the help of a large scale of datasets. As a result, sharing the valuable big data that can be applied to training deep learning models is of essential importance currently. However, how to claim ownership and protect the copyright of image big data during the sharing process is still a vital issue that should be addressed. The application of digital watermarks can protect the copyright of image data, at the same time, it also degrades the image quality at the same time. As for invisible digital watermarks, the higher the watermark embedding intensity causes the greater the host image changes. Therefore, the performance of deep learning models may decline due to using the watermarked training set. In this paper, we evaluate the influences of different embedding intensities of various watermarking algorithms on several mainstream models and conclude how the watermarking intensity affects the model training. Besides, referring to watermarking algorithms that have been proposed, we proposed a novel discrete Fourier transform-based watermarking algorithm that can achieve image copyright protection yet maintain the utility of models.
- 1 July 2010
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 108 (1) , 013104
Intersubband absorption in modulation doped heterostructures
Intersubband absorption is reported in a new modulation doped structure using strained InGaAs quantum wells (QWs) that support transistor operation. Well defined absorption peaks (1000 to ) from 8 to have been obtained using either -type or -type modulation doped wells. Both and well responses show strong polarization dependence with maximum values at incident angles of 65°–70° and peak positions which are adjusted by the QW parameters. The well shows a double peaked response with a peak separation of about which results from heavy and light hole contributions. The absorption data are compared with a theoretical model developed on the basis of variable conservation and excellent agreement is obtained. It is shown that a mass difference between the upper and lower states is essential to predict a finite spectral width of the response.- 1 July 2010
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 108 (1) , 013104
Intersubband absorption in modulation doped heterostructures
Intersubband absorption is reported in a new modulation doped structure using strained InGaAs quantum wells (QWs) that support transistor operation. Well defined absorption peaks (1000 to ) from 8 to have been obtained using either -type or -type modulation doped wells. Both and well responses show strong polarization dependence with maximum values at incident angles of 65°–70° and peak positions which are adjusted by the QW parameters. The well shows a double peaked response with a peak separation of about which results from heavy and light hole contributions. The absorption data are compared with a theoretical model developed on the basis of variable conservation and excellent agreement is obtained. It is shown that a mass difference between the upper and lower states is essential to predict a finite spectral width of the response.