Phase-sensitive tip-enhanced sum frequency generation spectroscopy using temporally asymmetric pulse for detecting weak vibrational signals

TL;DR

Phase-sensitive TE-SFG addresses the diffraction-limited resolution of conventional SFG by leveraging a nanoscale tip–substrate junction and a temporally asymmetric visible pulse to suppress the non-resonant background (NRB). The method extracts phase information and absolute molecular orientations by interferometric enhancement between vibrational signals and NRB, while simultaneously capturing forward and backward scattering to confirm tip-origin. A significant signal enhancement, on the order of $6.3\times10^6$ to $1.3\times10^7$, is achieved, enabling detection of weak vibrational modes within a nanometer-scale region. This approach promises nanoscale vibrational spectroscopy with potential for time-resolved studies and pump–probe extensions, bridging infrared vibrational signaling to accessible visible detection through gap-mode plasmonics.

Abstract

Vibrational sum frequency generation (SFG) spectroscopy is a powerful technique for investigating molecular structures, orientations, and dynamics at surfaces. However, its spatial resolution is fundamentally restricted to the micrometer scale by the optical diffraction limit. Tip-enhanced SFG (TE-SFG) using a scanning tunneling microscope has been developed to overcome this limitation. The acquired spectra exhibit characteristic dips originating from vibrational responses located within the strong broadband non-resonant background (NRB), which distorts and obscures the molecular signals. By making the second pulse temporally asymmetric and introducing a controlled delay between the first and second laser pulses, the NRB was effectively suppressed, which in turn amplified the vibrational response through interference and facilitated the detection of weak vibrational signals. This interference also made the technique phase-sensitive, enabling the determination of absolute molecular orientations. Furthermore, forward- and backward-scattered signals were simultaneously detected, conclusively confirming that the observed signals originated from tip enhancement rather than far-field contributions. Finally, the signal enhancement factor in TE-SFG was estimated to be $6.3\times 10^6-1.3\times 10^7$, based on the experimental data. This phase-sensitive TE-SFG technique overcomes the optical diffraction limit and enables the investigation of molecular vibrations at surfaces with unprecedented detail.

Figures (17)

• Figure 1: (a) IR and visible laser fields, with a time delay between the peaks of the two pulses. (b) Vibrational polarization generated by excitation with the IR pulse. (c) Second-order polarization induced by interaction with the visible pulse. (d) Energy level diagram illustrating the SFG process. The dashed line corresponds to the virtual state associated with anti-Stokes Raman scattering.
• Figure 2: Temporal profiles of IR ($E_\mathrm{IR}(t)$) and visible ($E_\mathrm{vis}(t;\tau)$) pulses, along with their product ($E_\mathrm{vis}(t;\tau)E_\mathrm{IR}(t)$). Their central frequencies are $\omega_\mathrm{IR}^0=2900\, \mathrm{cm}^{-1}$ and $\omega_\mathrm{vis}^0=9680.5\, \mathrm{cm}^{-1}$ (1033 nm). $E_\mathrm{IR}^0$ in Eq. (\ref{['E_IR']}) and $E_\mathrm{vis}^0$ in Eq. (\ref{['etalon']}) were determined so that the maximum amplitude of $E_\mathrm{IR}(t)$ and $E_\mathrm{vis}^\mathrm{etalon}(t;\tau)$ was equal to 1, respectively. (a) The IR and visible pulses are assumed to be Gaussian, with full widths at half-maximum (FWHMs) of 0.28 and 1.5 ps, respectively, and a time delay $\tau$ = 0.5 ps. (b) The visible pulse is expressed by Eq. (\ref{['etalon']}), with an original input pulse FWHM of 0.28 ps, a central wavelength of 1033 nm, a reflectivity $R=0.93$, and an air gap $d=18.6\,\mu\mathrm{m}$. FWHMs are defined for $|E(t)|^2$, while the displayed profiles correspond to $E(t)$.
• Figure 3: Simulated SFG spectra calculated from Eq. (\ref{['I_SFG2']}), combined with Eqs. (\ref{['P2_3']}), (\ref{['E_IR']}), and (\ref{['etalon']}), with the following parameters: $A_\mathrm{NR}=2.0\times10^4$, $\theta_\mathrm{NR}=90^\circ$, $A_q=-1$, $\omega_q=2900\,\mathrm{cm}^{-1}$, $\Gamma_q=10\,\mathrm{cm}^{-1}$, $\omega_\mathrm{IR}^0=2900\,\mathrm{cm}^{-1}$, $\omega_\mathrm{vis}^0=9680.5\,\mathrm{cm}^{-1}$, $\sigma_\mathrm{IR}=\sigma_\mathrm{vis}=40\,\mathrm{cm}^{-1}$, $R=0.93$, and $d=18.6\,\mu\mathrm{m}$. $\epsilon_0$ was set to 1. $E_\mathrm{IR}^0$ in Eq. (\ref{['E_IR']}) and $E_\mathrm{vis}^0$ in Eq. (\ref{['etalon']}) were determined so that the maximum amplitude of $E_\mathrm{IR}(t)$ and $E_\mathrm{vis}^\mathrm{etalon}(t;\tau)$ was equal to 1, respectively. The time delay $\tau$ was varied between 0 and 1000 fs (0, 200, 300, 400, 500, 600, 700, and 1000 fs). The spectra shown in panel (b) are normalized to their respective maximum values. With increasing $\tau$, the SFG intensity decreases (a), while the dip structure arising from the interference between the vibrational response and the NRB becomes more pronounced (b). At $\tau=1000\,\mathrm{fs}$, the NRB contribution is almost entirely suppressed, and only the resonant vibrational signal remains.
• Figure 4: Simulated SFG spectra calculated from Eq. (\ref{['I_SFG2']}), combined with Eqs. (\ref{['P2_3']}), (\ref{['E_IR']}), and (\ref{['E_vis']}), with the following parameters: $A_\mathrm{NR}=2.0\times10^4$, $\theta_\mathrm{NR}=90^\circ$, $A_q=-1$, $\omega_q=2900\,\mathrm{cm}^{-1}$, $\Gamma_q=10\,\mathrm{cm}^{-1}$, $\omega_\mathrm{IR}^0=2900\,\mathrm{cm}^{-1}$, $\omega_\mathrm{vis}^0=9680.5\,\mathrm{cm}^{-1}$, $\sigma_\mathrm{IR}=40\,\mathrm{cm}^{-1}$, and $\sigma_\mathrm{vis}=5\,\mathrm{cm}^{-1}$. $\epsilon_0$ was set to 1, and the time delay $\tau$ was varied between 0 and 1000 fs (0, 200, 300, 400, 500, 600, 700, and 1000 fs). The spectra shown in panel (b) are normalized to their respective maximum values. With increasing $\tau$, the SFG intensity decreases (a), but the relative depth of the dip remains almost the same (b).
• Figure 5: SEM images at different magnifications of STM tip used in this study. The red dot in panel (b) represents a 50 nm-diameter circle, whose size is comparable to that of the tip apex.