Cover story: Chen YX, Zhang FY, Dang ZB, He X, Luo CX et al. Chiral detection of biomolecules based on reinforcement learning. Opto-Electron Sci 2, 220019 (2023).

Chirality is one of most important properties for biomolecules. It is normal that enantiomers which exhibit similar physical properties possess totally different physiologic functions, so chiral detection has become a research focus around the world. The optical chirality of biomolecules usually arises in ultraviolet band and exhibits very low intensity, which brings huge troubles to the optical detection of biochirality. In recent years, many plasmonic researches state that surface plasmons enhance the circular dichroism signals of biomolecules, and chiral metasurfaces, metamaterials and nanoparticles are widely used in chiral detection. Due to the nearfield interaction between chiral biomolecules and metal nanostructures, the far-field spectrum exhibits a frequency shift related to the chirality of the biomolecules. Chiral detections for diverse biomolecules require different nanostructures. In order to find the relationships between biomolecules and corresponding nanostructures, a nanostructure design method based on reinforcement learning is involved to solve the problem. The algorithm utilizes different data generation method from traditional intelligent design and replaces data fitting with explorations of parameter space. It successfully proposed numerous nanostructures with a sharp peak in CD spectra, and the detection chips based on those nanostructures succeeded in differentiation of enantiomers. The difference of resonance wavelength shifts between the enantiomers of glucose reaches 7 nm. The algorithm has the potential for diverse nanophotonic circumstances, and the combinations with other plasmonic nanostructures probably form better chiral detection.