Abstract
<title>Abstract</title> <p>Optical spectroscopy is a cornerstone of material characterization, chemical sensing, and astronomy. traditional instruments suffer from rigid trade-off between device footprint, spectral range, and resolution. Here, we demonstrate a content-adaptive spectrometer that overcomes these limitations by co-designing dispersive Moiré meta-optics with a recursive sampling algorithm. While Moiré metalenses are typically utilized for varifocal focal-length tuning, the phase-wrapping inherent to their subwavelength metasurface architecture induces severe chromatic aberration. We exploit this hyperchromaticity to establish a deterministic, one-to-one mapping between the mutual rotation angle of the meta-surfaces and the sharply focused wavelength, successfully repurposing the pair into a high-resolution spectral scanner. To optimize data acquisition speed, we implement a content-adaptive, recursive sampling protocol that capitalizes on the structural sparsity of physical spectra. The system executes a rapid, coarse-scale initial sweep, subsequently triggering successive levels of fine-scale angular refinement exclusively within high-information spectral regions. Using a laboratory prototype operating from 400 nm to 800 nm, we demonstrate the high-fidelity reconstruction of diverse spectral profiles—ranging from smooth, broadband spectra to sparse, multi-line laser emissions—while requiring nearly 5× fewer measurements than conventional uniform sampling. Our results establish a framework for intelligent, task-adaptive meta-optical sensors that seamlessly bridge physical dispersion and computational techniques for real-time spectrometry.</p>