Machine learning-enabled sensor array using noble metal-functionalized WS2 nanoflakes for drastic selective gas sensing

Abstract

Noble-metal nanoparticle (NP) decoration has emerged as an effective strategy to enhance the performance of gas sensors, yet the underlying principles governing selectivity tuning remain poorly understood. In this study, we decorated two-dimensional (2D) tungsten disulfide (WS2) flakes with noble-metal NPs using a solution-based process without reducing agents, leading to a dramatic shift in selectivity. The decorated WS2 flakes enabled clear discrimination among nitrogen dioxide (NO2), ammonia (NH3), and hydrogen (H2), with a level of selectivity rarely observed in conventional three-dimensional metal-oxide nanostructures. This distinctive behavior originates from the synergy between the layered, edge-enriched architecture of WS2 and the catalytic/electronic effects of the noble-metal NPs. By integrating machine learning algorithms, we further demonstrate robust analyte classification and predictive capability across multiple gases. We propose a selective sensing mechanism based on charge-transfer interactions between target gas molecules and the functionalized sensing surface, which underpins the observed discriminative response. Our findings provide fundamental insights into the design of next-generation sensor arrays based on 2D materials, offering a strategic pathway for superior gas selectivity.