Closing the loop in next-generation sensing through shadow sphere lithography, plasmonics, and artificial intelligence

Abstract

The rapid deployment of intelligent energy, health-care and manufacturing platforms is outpacing the capabilities of conventional transducers, demanding sub-percent accuracy, millisecond responses, long-term stabilities and wafer-scale integration. Plasmonic micro- and nano-optical sensors can, in principle, satisfy these metrics, but only if three historically separate research threads converge: (i) physics-guided nanostructure design that realises high-Q hybrid resonances; (ii) fabrication routes that translate these blueprints into low-cost, large-area devices; and (iii) data-centric signal processing and prediction that extracts reliable information from inherently weak, drift-prone optical read-outs. This review (mainly covering the years 2019–2024) provides the first end-to-end account of that convergence. We highlight shadow-sphere lithography (SSL) as a scalable, sub-50 nm patterning strategy; map the resulting structural library onto its plasmonic, lattice and bound-state resonances; and show how physics-aware artificial-intelligence (AI) pipelines denoise spectra, compensate batch variability, enhance the prediction, and even invert the design problem. We close by outlining a closed-loop roadmap—linking SSL, plasmonics, and AI analytics—that targets high refractive-index resolutions within millimetre footprints, while identifying open challenges in wafer-scale 3D patterning inverse design and automated self-assembly, to in-line quality grading, to adaptive signal interpretation.