Scholarlink

<p>In this study, the structural, dielectric, conductive, and relaxation behavior of neat epoxy (NE) and its composites containing 1 wt.% of various metal oxides (ZnO, TiO₂, Al₂O₃, SiO₂) combined with multi-walled carbon nanotubes (MWCNTs) were investigated. While previous studies have explored nanofiller-doped epoxy systems, they often lacked a detailed understanding of interfacial interactions, dielectric relaxation mechanisms, and predictive modeling for dielectric properties. To address this gap, XRD analysis was performed to confirm the crystalline phases of ZnO, TiO₂, and Al₂O₃, the amorphous nature of SiO₂, and the successful incorporation of MWCNTs into the composites. The dielectric constant (ε′) was measured, revealing the highest value for neat epoxy doped with 1 wt.% of (ZnO + MWCNT). Dielectric loss (ε′′) and electrical modulus analysis were conducted to gain a deeper understanding of relaxation behavior in respective composites. The findings demonstrate the critical role of nanoparticle properties and their synergistic interactions with MWCNTs in tailoring the structural and functional properties of epoxy-based composites. Furthermore, this study evaluates the predictive performance of XGBoost, Extra Trees, and Gradient Boosting regression models for dielectric properties (ε′ and ε′′) across intermediate frequencies, with Extra Trees achieving the best results (<i>R</i><sup>2</sup> = 0.9999 for ε′ and <i>R</i><sup>2</sup> = 1.0000 for ε′′). These findings contribute to the advancement of epoxy-based materials with tuneable electrical properties, making them promising for future electronics and energy storage. Integrating experimental data with machine learning can further enhance composite design for industrial applications, ensuring optimized performance and reliability.</p> Graphical abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-assisted prediction and optimization of dielectric properties in epoxy resin nanocomposites

  • 期刊论文

  • 发表日期：

    2025年5月13日

查看全文
  • Sanketsinh Thakor,
  • Anand Joshi,
  • Hetvi Patel,
  • Prince Jain,
  • Meenu Khan,
  • K. Sruthi,
  • Manansi Soni,
  • Chandan R. Vaja

摘要

In this study, the structural, dielectric, conductive, and relaxation behavior of neat epoxy (NE) and its composites containing 1 wt.% of various metal oxides (ZnO, TiO₂, Al₂O₃, SiO₂) combined with multi-walled carbon nanotubes (MWCNTs) were investigated. While previous studies have explored nanofiller-doped epoxy systems, they often lacked a detailed understanding of interfacial interactions, dielectric relaxation mechanisms, and predictive modeling for dielectric properties. To address this gap, XRD analysis was performed to confirm the crystalline phases of ZnO, TiO₂, and Al₂O₃, the amorphous nature of SiO₂, and the successful incorporation of MWCNTs into the composites. The dielectric constant (ε′) was measured, revealing the highest value for neat epoxy doped with 1 wt.% of (ZnO + MWCNT). Dielectric loss (ε′′) and electrical modulus analysis were conducted to gain a deeper understanding of relaxation behavior in respective composites. The findings demonstrate the critical role of nanoparticle properties and their synergistic interactions with MWCNTs in tailoring the structural and functional properties of epoxy-based composites. Furthermore, this study evaluates the predictive performance of XGBoost, Extra Trees, and Gradient Boosting regression models for dielectric properties (ε′ and ε′′) across intermediate frequencies, with Extra Trees achieving the best results (R2 = 0.9999 for ε′ and R2 = 1.0000 for ε′′). These findings contribute to the advancement of epoxy-based materials with tuneable electrical properties, making them promising for future electronics and energy storage. Integrating experimental data with machine learning can further enhance composite design for industrial applications, ensuring optimized performance and reliability.

Graphical abstract