Mid-IR hyperspectral imaging with undetected photons

Sensing with undetected photons has become a vibrant, application-driven research domain with a special focus on the mid-infrared (mid-IR) wavelength region. Since the mid-IR contains spectral bands with highly specific and strong molecular absorbance signatures, often referred to as fingerprints, a multitude of different samples and their compositions can be detected and quantified spectroscopically. Enhancing this inherently sample alteration-free spectroscopic method with imaging capabilities leads to a powerful technique for environmental monitoring and biomedical applications that enables automated diagnostics while omitting time-consuming and non-reversible labeling steps. To evade the shortcomings of state-of-the-art instruments for mid-IR hyperspectral microscopy related to cost, complexity, power-consumption, and performance, which are associated with technological challenges for mid-IR cameras and low-noise and broadband mid-IR sources, here, we construct a proof-of-concept nonlinear interferometer in a wide-field imaging arrangement combined with high-resolution spectral acquisition by pixelwise quantum Fourier transform infrared spectroscopy. For the broadband range of 2300-3100 cm$^{-1}$, covering the important CH-stretch band, we perform hyperspectral imaging that simultaneously resolves 3500 spatial modes, each with a spectral resolution of 10 cm$^{-1}$ leveraging in total around $10^5$ spatio-spectrally entangled photon modes. Our image acquisition uses a commercial, megapixel sCMOS camera, while a medium-power and compact c.w. pump laser is the only necessary light source. For a moderate speed of 360 voxel/s yielding a dominantly shot-noise influenced signal-to-noise ratio (SNR) of 50, we demonstrate the practicality of our novel hyperspectral imaging technique for microplastics detection and bio-imaging tasks, and outline engineering solutions to increase its speed by several orders of magnitude. This shows that our quantum imaging technique is highly promising for applications requiring compact, cost-effective label-free analyses.