Fig. 5: Simpler, accurate, and intelligent spectropolarimetric analysis.

a Illustration of optical rotatory dispersion when light passes through a chiral substance such as glucose. b, c Setups for spectropolarimetric analysis using b a traditional system and c our proposed image-based graded grating system. Our proposed system replaces the second polarizer (analyzer) from the conventional system with a beam splitter and our plasmonic grating. A CCD camera captures images of the plasmonic grating reflection mode, in which the cross-bar patterns appear as a set of dark bars formed from the coupling of light with the grating. d Images of the dark bar patterns under single-peak illumination, for 490, 595, and 660 nm illumination with 0% (DI water), 10%, and 30% glucose solutions. Although difficult for the human eye, the deep-learning algorithm can detect differences in the intensities of the horizontal and vertical dark bars based on the concentration of the sample. e Images of the dark bar patterns produced under double-peak (top row) and triple-peak (bottom row) illumination. When multiple peaks are used in the illumination spectrum, multiple patterns can be produced at once. f, g Predictions of the optical rotation, represented as data points (error bars indicate the standard deviation of measureddatasets), introduced by various glucose solutions for f double-peak and g triple-peak illumination. The solid curves represent the ORD derived from the experiment using conventional methods. Data were collected for 2% (red), 10% (blue), and 30% (purple) glucose concentrations.