Skip to main navigation Skip to search Skip to main content

Molecular network engineering of photocrosslinked ferroelectric polymers for tunable synaptic plasticity

  • Young Seok Song
  • , Qun Gao Chen
  • , Seongbo Jang
  • , Seungyeon Kim
  • , Dae Hong Kim
  • , Bhise Sneha
  • , Jaewon Yeom
  • , Sukang Bae
  • , Byung Joon Moon
  • , Seoung Ki Lee
  • , Beomjin Jeong
  • , Wen Ya Lee*
  • , Tae Wook Kim*
  • *Corresponding author for this work
  • Jeonbuk National University
  • National Taipei University of Technology
  • Pusan National University
  • Korea Institute of Science and Technology
  • National Chengchi University

Research output: Contribution to journal › Journal article › peer-review

Abstract

Precise control of ferroelectric polarization in organic materials is essential for optimizing neuromorphic device operation. In this work, we introduce a photocrosslinking strategy to modulate the ferroelectric and synaptic properties of poly(vinylidene fluoride–trifluoroethylene) [P(VDF-TrFE)] by incorporating controlled amounts of a bis-azide crosslinker, bis-FB-N3. Upon UV activation, nitrene intermediates react with the polymer backbone to form a covalently crosslinked network. Structural analyses show that increasing the crosslinker content reduces β-phase crystallinity and reinforces molecular confinement within the ferroelectric matrix. These changes suppress spontaneous polarization and lead to a narrower hysteresis window, enabling direct control over dipole alignment and switching behavior. Transistors incorporating photocrosslinked P(VDF-TrFE) exhibit tunable long-term potentiation and depression, supporting analog modulation of synaptic weights. Higher crosslinker concentrations reduce dynamic range due to polarization suppression but improve the linearity and symmetry of conductance updates, which are critical for stable neural network training. A trade-off between dynamic range and energy consumption is also observed; the device containing 22.2 wt% of crosslinker achieved the lowest energy usage of 22 nJ per spike while maintaining clear synaptic responses. Deep neural network simulations based on experimentally extracted synaptic parameters confirm the system-level impact of this molecular tuning. The device containing 22.2 wt% of the photocrosslinker achieved a classification accuracy of 80.67% on the Fashion-MNIST dataset, reflecting improved learning stability enabled by more linear and symmetric weight updates. These results demonstrate that molecular-level photocrosslinking provides an effective route to engineer ferroelectric polarization and enhance the performance and energy efficiency of organic neuromorphic computing platforms.

Original languageEnglish
Article number177305
JournalChemical Engineering Journal
Volume541
DOIs
StatePublished - 2026.08.1

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

  1. SDG 7 - Affordable and Clean Energy
    SDG 7 Affordable and Clean Energy

Keywords

  • bis-FB-N
  • Organic neuromorphic transistors
  • P(VDF-TrFE)
  • Tunable ferroelectricity
  • Tunable neuromorphic computing

Fingerprint

Dive into the research topics of 'Molecular network engineering of photocrosslinked ferroelectric polymers for tunable synaptic plasticity'. Together they form a unique fingerprint.

Cite this