Data-driven modelling of host–guest interactions and diffusion-induced swelling in pillar[5]arene-based chemosensors
Journal of Inclusion Phenomena and Macrocyclic Chemistry, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s10847-026-01352-1
- Dergi Adı: Journal of Inclusion Phenomena and Macrocyclic Chemistry
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Chemical Abstracts Core, Compendex, Natural Science Collection (ProQuest), Biological Science Database (ProQuest), Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
- Anahtar Kelimeler: Gaussian Process Regression, Mathematical model, NARX-ANN, QCM, SPR, Thin film sensor
- Uşak Üniversitesi Adresli: Evet
Özet
In this study, we investigate a pillar[5]arene-based thin film sensor through an integrated experimental and data-driven modelling approach, with the aim of elucidating diffusion-induced swelling behavior. Within this framework, diffusion coefficients were calculated using time-dependent response data for three different volatile organic compounds (VOCs) obtained via Surface Plasmon Resonance (SPR) and Quartz Crystal Microbalance (QCM) measurements. These values were determined using Fick’s early-stage diffusion model. The SPR-derived diffusion coefficients were higher than the QCM values, being approximately 3.3 times higher for dichloromethane and 2.8 times higher for chloroform. To capture the nonlinear and dynamic sensor response, Nonlinear Autoregressive Exogenous Artificial Neural Networks (NARX-ANN) and Gaussian Process Regression (GPR) models were developed and validated using experimental datasets. Both approaches accurately reproduced the temporal response profiles and yielded diffusion coefficients in close agreement with those obtained from physics-based calculations. This study presents a comprehensive framework that combines diffusion-based physical analysis for VOC detection using pillar[5]arene with advanced machine learning techniques. The proposed approaches contribute to the understanding of VOC–sensor interactions and may serve as a useful framework for the design and optimization of chemical sensing systems.