Label‐Free Femtomolar Protein Detection and Secondary‐Structure Analysis Enabled by Co‐Optimized Parameter‐Decoupled Metasurfaces

ABSTRACT Trace‐level protein detection and secondary‐structure analysis are vital in life sciences and biomedical research. Surface‐enhanced infrared absorption (SEIRA) directly probes intrinsic molecular vibrations, enabling truly label‐free detection without conformational perturbation and thus ensuring accurate structural characterization. Despite these advantages, the performance of SEIRA is often hindered by inefficient molecule‐metasurface coupling and the resulting weak light‐matter interactions. To address this challenge, we introduce a loss parameter co‐optimization strategy based on temporal coupled‐mode theory (TCMT), which jointly tunes the metasurface loss rate and the molecule‐metasurface coupling coefficient to maximize interaction and signal enhancement. Guided by this strategy, we developed a parameter‐decoupled metasurface that independently controls the loss parameters and detects intrinsic protein vibrations at concentrations as low as 1 fM. The device also resolves temperature‐dependent changes in the amide I profile of Aβ1‐40 peptides, revealing a redistribution of conformational components during thermal treatment. Furthermore, we developed a label‐free protein identification framework that integrates secondary‐structure fingerprints with machine‐learning analysis, achieving 100% classification accuracy for nine proteins at concentrations down to 1 pM. These advances establish a versatile platform for label‐free, secondary‐structure‐sensitive analysis and high‐throughput sensing of trace‐level proteins.