Fig. 1. Fabrication details and material characterization of the MIM L-cavities. (a) Flow diagram of the fabrication process involved in direct laser writing of the MIM L-cavities. (b) Graphic of the final L-cavities used in EM simulations; top, angled side profile; bottom, angled overhead view. (c) SEM image of a fabricated L-cavity with arm lengths of 795 nm and 1324 nm (left). 45° image of similar L-cavities (right). Scale bars represent 400 nm. (d) AFM height image of a similar set of L-cavities. Scale bar represents 1000 nm.

Fig. 2. Modal analysis of the L-cavity resonances. (a), (b) Simulated out-of-plane electric field (Ez) profiles of the modes identified in the polarization-resolved reflectance spectra in (c). The modes are color-coded according to the border color of each plot. (c) Experimental (upper) and simulated (lower) polarization-resolved reflectance spectra. The polarization angle of each measurement is indicated in the legend, and the spectral locations of the modes are outlined by the color-coded dashed lines. The spectra in waterfall plots are separated by increments of 0.8. (d), (e) Polar plots of the differential experimental reflectance of the modes shown in (a)–(c). IAu and IL are the reflected intensities from a planar Au film and L-cavity array with arm lengths of Lx=797  nm and Ly=1324  nm, respectively.

Fig. 3. Vibrational-plasmon strong coupling. (a) Experimental reflectance of modes L, M, and U in the frequency range where VSC to the IP-Dip vibrational mode occurs as the aspect ratio of the L is varied as a function of Lx. The dotted lines are guides to the eye indicating the polariton disperson. (b) Three-coupled-oscillator model where both cavity modes are coupled to each other as well as to a molecular vibration within the polymer. (c) Comparison between the measured location of the upper, middle, and lower polaritons in (a) (black circles) and the coupled three-oscillator model (black lines). The uncoupled Lx (blue) and Ly (red) modes predicted by Eq. (B1) and the vibrational transition (green) are given by the dashed lines. (d) Hopfield coefficents or the eigenvectors for the eigenvalues of the three-coupled-oscillator model in Eq. (B3) as a function of Lx.

Fig. 4. Multimode coupling for various Ly arm lengths. (a)–(c) Experimentally measured location of the upper, middle, and lower polaritons (black circles) and the three-coupled oscillator model (black lines) for three different average Ly: (a) 959 nm, (b) 1126 nm, and (c) 1300 nm. The uncoupled Lx (blue) and Ly (red) modes predicted by Eq. (B1) and the vibrational transition (green) are given by the dashed lines. (d)–(f) Hopfield coefficents of the 3×3 oscillator model in Eq. (B3) as a function of Lx for an average Ly of (d) 959 nm, (e) 1126 nm, and (f) 1300 nm.

Fig. 5. E-field components of mode L in Fig. 2. Electric field mode (Ez, Ex, Ey) plots of mode L, Ez is also shown in Fig. 2(a).

Fig. 6. Simulated reflectance spectra for L-antenna in Fig. 2. The field mode plots of higher-order modes are not characterized (5750  cm−1)/not seen (4400 and 6050  cm−1) in the experimental data.

Fig. 7. Simulated reflectance as a function of Lx: (a) Ly=760  nm and (b) Ly=1300 cases overlaid on the experimental dip locations (black dots) and the polariton model (black lines) from Figs. 3(c) and 4(c).

Fig. 8. Size-dependent modal dispersion of a single arm. (a) Geometry of a single arm (Lx=1165  nm) used to simulate the dispersion in the contour plot of (b). (b) Field mode profile at 2250  cm−1 for Lx=1165  nm. (c) Simulated reflectance spectra of the structures in (a) and (b) shown as a contour plot overlaid with the calculated dispersion of Eq. (B1).

Fig. 9. Modeled L-cavity bonding and antibonding modes without the presence of the IP-Dip vibrational transitions. The L-cavity modes were simulated via COMSOL for various Lx while (a) Ly=956  nm and (b) Ly=1300  nm. The black solid lines are the bonding (lower) and antibonding (upper) modes modeled via a two-oscillator strong coupling model of the two uncoupled Lx and Ly cavity resonances associated with each arm (dashed lines). For this, Eq. (B3) is modified by setting plasmon–CSVT interaction potential to 0, V2=0, and removing the CSVT, ωvib. Thus, the eigenvalue problem is now given by the 2×2 matrix [ωx[x]V1V1ωy][α[x]β[x]]=λ±[α[x]β[x]], where λ± are the eigenvalues of the system. The plasmon–plasmon interaction potential, V1, was kept at 541  cm−1 as was found from Figs. 3 and 4.

Fig. 10. Two-oscillator model prediction of the light–matter interaction. Predicted dispersion of two-oscillator model between the cavity bonding mode and the CSVT when (a) Ly=956  nm and (b) Ly=1300  nm. The dispersion of bonding mode is first modeled as described in Fig. 9. Then the coupling between CSVT and the bonding mode is fit to another two-coupled-oscillator model given by the 2×2 matrix [ωB[x]V2V2ωvib][α[x]β[x]]=λ±[α[x]β[x]], where ωB[x] is the bonding mode dispersion. The resulting interaction potentials are found to be V2=23  cm−1 for (a) and V2=63  cm−1 for (b).

Fig. 11. Vibration-plasmon coupling for various Ly. (a)–(c) Waterfall plots of the experimental reflectance as Lx is varied for three different Ly: (a) 959 nm, (b) 1126 nm, and (c) 1300 nm. Modes L, M, and U are indicated in each plot by the dashed black lines.

Fig. 12. FTIR reflectance spectra of bare IP-Dip on an Au subtrate. The absorbion of the IP-Dip CSVT can be seen at 1732  cm−1 with an FWHM of 32  cm−1.