Continuous-wave, high-resolution, ultra-broadband mid-infrared nonlinear spectroscopy with tunable plasmonic nanocavities

Abstract

Vibrational sum- and difference-frequency generation (SFG and DFG) spectroscopy probes the nonlinear response of interfaces at mid-infrared (MIR) wavelengths while detecting upconverted signals in the visible. Recent work has moved from large-area films and colloids to nanoscale structures using dual-resonant plasmonic nanocavities that co-confine light and matter in deep-subwavelength volumes. Here we implement high-resolution ($<1$~cm$^{-1}$), continuous-wave ultrabroadband vSFG, vDFG, and four-wave mixing (FWM) coherent spectroscopy from 860 to 1670~cm$^{-1}$ on dual-resonant antennas under ambient conditions. Using a commercial, broadly tunable quantum-cascade laser and eliminating geometric phase matching simplify acquisition and expand spectral reach. The resulting spectra exhibit coherent interference between resonant (vibrational) and nonresonant (electronic) contributions to the effective $χ^{(2)}$, previously accessible only under fs/ps excitation. Simultaneous measurement of SFG and DFG enables a {ratiometric} analysis that suppresses common-mode drifts and helps reveal vibrational resonances. We demonstrate versatility and reproducibility across several analytes that span distinct relative strengths of vibrational vs. electronic nonlinearities. Together, these capabilities position our approach as a scalable route to multiplexed, high-resolution MIR sensing and a practical basis for chip-level, label-free coherent spectroscopy. It opens a feasible path toward single- and few-molecule optomechanical studies using nanoscale trapping strategies.

Figures (11)

• Figure 1: (a) Experimental scheme. A 780 nm VIS laser (objective NA = 0.9) and a tunable MIR QCL (NA = 0.78) are co-focused onto a nanoparticle-on-slit (NPoS) nanocavity. The analyte layer resides in the nanometer gap between the gold nanoparticle and the slit walls (inset). (b) Energy-diagram representation of photon scattering pathways under simultaneous VIS $(\omega_{\mathrm{VIS}})$ and MIR $(\omega_{\mathrm{IR}})$ excitation. (c) Two-dimensional map of cavity emission during a continuous MIR sweep (890--1650 cm$^{-1}$, 1 cm$^{-1}$ steps; 761 frames; vertical axis) from an individual BPhT-functionalised NPoS (from M1 array, see below). The SFG, DFG, and 2p-FWM peaks are labeled. (d) Representative emission spectra at selected MIR wavelengths (coloured), compared with the MIR-off Raman spectrum (grey), all from the same nanocavity. (e) Zoom-in on the Stokes-sideband region around the DFG signal. After subtracting spontaneous Raman (grey), the net DFG peak is fit with a Gaussian, whose area is $\tilde{I}^{(2)}_{\pm}(\omega_{\mathrm{IR}})$ in eq. (\ref{['eq:normI']}).
• Figure 2: (a) Fourier-transform IR (FTIR) absorption spectra of three metasurface designs (M-1/M-2/M-3, right axis), overlaid with the 780 nm Raman spectrum of BPhT (left axis).(b) Scanning-electron microscope (SEM) images of two representative slit-array MIR metasurfaces: M-2 (left) and M-3 (right). An estimate of the VIS and MIR illumination spot sizes are overlaid on each image. (c–d) Power-normalized DFG (top, light blue) and SFG (bottom, dark blue) spectra from single BPhT-functionalized NPoS cavities fabricated on the M-2 (c) and M-3 (d) metasurfaces. For both designs, the nonlinear signal strength follows the MIR absorption spectrum of the corresponding array (orange curves).
• Figure 3: (a) Single–nanocavity broadband readout. From bottom to top: a representative Stokes Raman spectrum (grey) used as a vibrational reference (invariant across NPoS); power-normalized SFG and DFG spectra (blue dots) from the NPoS in Fig. \ref{['fig:arrays']}d together with their fits (black lines); and the corresponding SFG/DFG ratio (pink dots) from the same cavity. Fit parameters are summarized in Table S2. (b) SFG and DFG spectra (colore coded) from three independent NPoS on two metasurfaces (M-2, M-3), overlaid with a typical SERS spectrum (grey).
• Figure 4: DFT calculations for an Au–S capped molecule (thiol hydrogen replaced by a single Au atom to model chemisorption) provide a baseline for mode positions and for the expected sign of the resonant $\chi^{(2)}_{zzz}$ contribution (bar plot: red/blue denote opposite signs), together with the corresponding DFT Raman and IR spectra (broadened for readability). While positions broadly align, the measured amplitudes, line shapes, and even signs can deviate in a robust, cavity-specific manner, consistent with electromagnetic/chemical environmental effects in the nanogap that are not captured by the single-atom adsorption model.
• Figure 5: (a) 4-NTP on two metasurfaces. Top: SFG spectrum from a single NPoS on M-1. Inset: overlay with the M-1 FTIR spectrum, yellow line. Middle: SFG spectrum from a single NPoS on M-3. Inset: overlay with the M-3 FTIR spectrum, orange line. Bottom: corresponding SERS spectrum. For 4-NTP the VIS power was limited to 1 µ W to avoid photochemistry. (b) cn-BPhT on M-1. Power-normalized DFG (top) and SFG (middle) from a single NPoS, overlaid with the M-1 FTIR spectrum (yellow); bottom: SFG/DFG ratio and co-recorded SERS.