A descriptor-oriented reporting framework for lanthanide-doped upconversion nanoparticle–semiconductor heterostructures in NIR-activated photocatalysis: Photophysics, interfaces, and ML-ready data

Abstract

Near-infrared (NIR) photons remain underutilized in heterogeneous photocatalysis because most semiconductor catalysts absorb mainly ultraviolet or visible light. Lanthanide-doped upconversion nanoparticles (UCNPs) can convert NIR excitation into UV-visible emission and thereby activate coupled semiconductors below their intrinsic bandgap. However, performance is governed by an interdependent parameter space spanning host lattice, dopant distribution, multilayer architecture, surface ligand, spectral overlap, interfacial contact, charge-transfer kinetics, reaction medium, irradiation geometry, and stability. This review proposes a descriptor-oriented reporting and mechanism-labeling framework for UCNP–semiconductor heterostructures. It translates synthesis history, photophysics, interfaces, reactors, and catalytic outputs into machine-readable fields; distinguishes radiative reabsorption, resonance energy transfer, direct or defect-assisted charge transfer, and photothermal effects; and defines the controls required for each mechanism label. Because the available literature is too sparse and heterogeneous for a defensible universal predictive model, the framework is intended to guide standardized data collection and future grouped-validation studies rather than to claim a validated descriptor–performance relationship. A minimum reporting checklist and publication-grouped validation strategy are provided for future interpretable machine-learning studies.