Structured illumination microscopy (SIM) has attained high spatiotemporal delineation of subcellular architecture, yet offers limited insight into chemical composition. Here, we present Chem-SIM, a structured-illumination fluorescence detected mid-infrared photothermal microscopy, for super-resolution chemical imaging of microorganisms and mammalian cells. A computational pipeline combining Poisson maximum-likelihood demodulation and spectral normalization across wavenumber is implemented to robustly recover the weak IR-induced fluorescence intensity change under low photon budgets and convert the fluorescence intensity modulation to chemical fingerprints. A photothermal gating scheme further rejects water backgrounds in aqueous samples, while the IR pump maintains cellular activity at near-physiological temperature. Chem-SIM preserves full vibrational fingerprints, achieves SIM-grade lateral resolution, and operates in a high-throughput, camera-based format with minimal modifications and low photothermal load. At the single-bacterium level, Chem-SIM distinguishes stationary phase from log phase cells through chemical content mapping. In ovarian cancer cells, Chem-SIM delivers readouts of lipid chemistry under deuterium fatty acid treatment and resolves lipid droplets dynamics in live cells. Together, Chem-SIM provides an accessible route to super-resolved mapping of organelle chemistry, metabolism, and dynamics.
extends the resolution into three dimensions 25,26 . Nonetheless, shallow modulation of scattered photons limits its sensitivity 27 . Fluorescence-detected MIP (F-MIP) mitigates this issue by using fluorescent dyes as nanoscale thermometers, increasing modulation depth by two orders of magnitude [28][29][30][31][32] . By integrating F-MIP with structured illumination, the technique can surpass the visible diffraction limit 33 . Yet, strong water absorption 34 and limited fluorescence photon budgets 28,31 in living specimens have thus far hindered super-resolution wide-field photothermal imaging in aqueous environments, despite proof-of-concept demonstrations on dried polymer beads 33 .
Here, we present Chem-SIM, a wide-field MIP-modulated structured-illumination fluorescence platform for super-resolution chemical imaging in both fixed and live cells. By integrating sinusoidal illumination with Poisson maximum-likelihood demodulation, Chem-SIM achieves ~2× lateral resolution while boosting sensitivity over previously reported wide-field F-MIP. Our approach preserves full vibrational fingerprints under photon-limited conditions and suppresses water background through photothermal relaxation gating. Notably, the IR pump maintains a nearphysiological temperature during acquisition to help sustain cellular activity. With such capacities, Chem-SIM enables fingerprint-level readout of metabolism and chemistry in single bacterial and single organelles. Chem-SIM requires only minimal modifications to a standard wide-field microscope, and provides a platform for high-throughput mapping of organelle chemistry, metabolism, and dynamics.
Chem-SIM integrates mid-infrared (mid-IR) photothermal excitation with structured-illumination microscopy to deliver super-resolution, wide-field chemical imaging. As shown in Figure 1a, the system adopts an upright wide-field layout optimized for calcium fluoride (CaF2) substrates. A tunable 50 kHz mid-IR quantum cascade laser (QCL) excites molecular vibrations. A nanosecondpulsed visible laser is patterned by a digital micromirror device (DMD) to encode high-spatialfrequency content beyond the diffraction limit. Photothermal-induced fluorescence modulation is acquired in alternating “hot” (IR on) and “cold” (IR off) frames that are pulse-synchronized via electronic gating (Figure 1b). Hundreds of pump-probe cycles are averaged per frame to approach shot-noise-limited fluorescence detection by an sCMOS camera. Full details of the optical and electronic components are provided in the Methods.
For each Chem-SIM acquisition, three pattern orientations and three phase shifts yield nine hot and nine cold images. These raw frames are reconstructed using a joint RL-SIM algorithm, which preserves quantitative modulation contrast (detailed in Supplementary Note 1 and Supplementary Figure 1), and the resulting hot-SIM and cold-SIM images are then subtracted pixel-wise to extract the thermally modulated signal (Figure 1c). For hyperspectral imaging, the mid-IR pump is stepped from 900-1798 cm -1 in 1 cm -1 increments; at each wavenumber we record the full 18-frame sequence at speed of 40 fps with an effective exposure of 23.1 ms per frame, giving a total acquisition time of ~585 s for the entire stack (detailed in Supplementary Note 2). As shown in Figure 1d, cold-SIM exhibits ~20% photobleaching over the scan, whereas hot-SIM additionally encodes wavelength-dependent IR absorption. The modulation-depth spectrum is obtained from the cold-hot subtraction after normalizing by the cold-SIM bleaching curve and it reveals water-vapor absorption lines along the optical path, consistent with the on-sample IR power spectrum (Figure 1e). Finally, a weighted least-squares normalization of this modulationdepth spectrum by the measured on-sample IR power both equalizes wavenumber-dependent pump power and effectively suppresses water-vapor absorption lines, producing the Chem-SIM absorption spectrum (Figure 1f) that closely matches the normalized reference Fourier transform infrared (FTIR) spectrum. Full details of spectrum normalization are provided in the Methods, Supplementary Note 3, and Supplementary Figure 2.
Despite the high-quality area-averaged Chem-SIM spectra (Figure 1f), single-pixel Cold-Hot images remain dominated by noise and contain little usable modulation contrast. Mid-IR absorption induces only a small temperature rise that weakly perturbs the local fluorescence yield (< 10% on average). As shown in Figure 2a, the raw hot and cold frames have sufficient signalto-noise ratio (SNR) to reveal the structured-illumination fringes; however, their cold-hot subtraction is noise-amplifying and fails to recover these patterns. The same holds when hot and cold stacks are reconstructed separately by SIM and then subtracted. The resulting difference image remains noisy and the high-resolution modulation contrast is severely degraded. This behavior is expected, because differencing two Poisson-distri
This content is AI-processed based on open access ArXiv data.