中红外单光子上转换光谱技术进展(特邀)

The Mid-Infrared (MIR) spectral region (2.5~25 μm) encompasses the fundamental vibrational and rotational absorption bands of a wide variety of molecules and is therefore widely recognized as the“molecular fingerprint”region. Owing to its direct access to intrinsic molecular transitions, MIR spectroscopy offers superior chemical specificity and selectivity compared with visible and near-infrared techniques, and has become an indispensable tool in atmospheric remote sensing, environmental monitoring, industrial process control, biomedical diagnostics, and materials and cultural-heritage analysis. As practical applications increasingly demand trace-level sensitivity, real-time response, and field-deployable operation, MIR spectroscopy is confronted with stringent requirements arising from extremely weak signal levels, strong thermal background radiation, and the need for high temporal resolution. Achieving highly sensitive, low-noise, and fast MIR spectral measurements at room temperature has therefore emerged as a central scientific and technological challenge. Conventional MIR spectroscopic systems rely primarily on narrow-bandgap semiconductor detectors combined with dispersive or Fourier-transform spectral analysis schemes. While these approaches provide broad wavelength coverage and high spectral resolution, they suffer from intrinsic limitations in room-temperature noise performance, detection sensitivity, and measurement speed. In recent years, frequency upconversion spectroscopy based on nonlinear optical processes has attracted growing attention as a powerful strategy to overcome these constraints. By coherently transferring MIR spectral information to the visible or near-infrared domain via sum-frequency or difference-frequency generation, upconversion techniques enable the use of mature silicon or InGaAs detectors characterized by low dark current, low noise, high quantum efficiency, and fast response. As a result, system-level noise can be drastically reduced at room temperature, enabling high-sensitivity MIR spectroscopy under conditions that are otherwise inaccessible to direct detection. Notably, in the photon-starved regime, frequency upconversion effectively circumvents the thermal-noise-dominated detection limit of the MIR, extending spectroscopic measurements toward the single-photon level. This review provides a comprehensive overview of the physical principles, major technical routes, and representative advances in MIR upconversion spectroscopy, with particular emphasis on schemes based on second-order nonlinear optical processes. Three major classes of upconversion spectroscopic architectures are systematically discussed and compared. Spatially dispersive upconversion spectroscopy exploits angular or spatial mapping to achieve broadband, high-throughput parallel spectral acquisition and is well suited for real-time measurements. Temporally dispersive approaches, notably time-stretch upconversion spectroscopy, map spectral information into the time domain via group-velocity dispersion, enabling megahertz-level acquisition rates and making them particularly attractive for ultrafast dynamics and transient phenomena. In addition, nonlinear-interferometric and quantum-correlated schemes leverage induced coherence and photon-pair correlations to reconstruct MIR spectral information without directly detecting MIR photons, thereby achieving ultrahigh sensitivity down to the single-photon regime and opening new paradigms for MIR spectroscopy. From an application perspective, the review summarizes recent progress in MIR upconversion spectroscopy across a broad range of fields, including atmospheric and long-range remote sensing, combustion diagnostics, trace-gas detection, non-destructive materials imaging, quantum sensing, and ultrafast spectroscopy. These examples highlight the unique advantages of upconversion techniques in scenarios involving weak signals, strong backgrounds, and high-speed dynamics. At the same time, key challenges that currently limit further development are critically analyzed, including the trade-off between conversion efficiency and spectral bandwidth, restricted access to long-wavelength and fingerprint MIR regions, system complexity and stability, as well as limited levels of engineering integration and scalability. Looking forward, continued advances in nonlinear materials with large second-order coefficients and broad transparency windows, together with improved quasi-phase-matching, cavity enhancement, and pump-engineering strategies, are expected to further enhance conversion efficiency and spectral coverage. Moreover, the integration of upconversion spectroscopy with photonic integrated platforms and artificial-intelligence-assisted spectral analysis is anticipated to significantly improve system compactness, robustness, and functionality. Through the parallel development and convergence of multiple technical routes, MIR upconversion spectroscopy is poised to evolve from a laboratory-based research tool into a high-performance spectroscopic platform, enabling impactful applications in environmental monitoring, medical diagnostics, industrial inspection, and security, and playing an increasingly important role in the future landscape of MIR spectroscopy and applied photonics.