Frequency pulling and line-shape broadening in graphene Raman spectra by resonant Stokes surface plasmon polaritons

This Rapid Communication reports on the modulation of the position and linewidth of the $G$ and $2D$ Raman bands for monolayer graphene through coupling to surface plasmon polaritons (SPPs). It is shown that gold nanoresonators tuned to the graphene Stokes emission frequency, not the laser pump, broaden the $G$ and $2D$ peaks and shift them to lower wave numbers, in addition to enhancing the Raman scattering cross section. The variation of the band displacement and broadening with the SPPs’ frequency demonstrates a strongly dispersive phenomenon, ruling out chemical, mechanical, and thermal mechanisms. It is believed that strong graphene-SPP coupling is responsible for the frequency pulling and line-shape broadening.

Figure 1

Schematic representation of enhanced Stokes scattering through resonant coupling to a surface plasmon polariton (SSP) nanoresonator at ${\omega }_{\mathrm{spp}}\cong {\omega }_{\mathrm{Stokes}}$. A pump photon at ${\omega }_{\mathrm{laser}}$ is absorbed by graphene and a Stokes surface plasmon polariton (SSPP) at ${\omega }_{\mathrm{spp}}$ is created in the plasmonic resonator. The resonator, in turn, radiates SSPPs as photons into the free space through its induced electric dipole moment.

Figure 2

(a) Schematic representation of a half-wavelength SPP nanoresonator. (b) SEM image of a nanoresonator array. The inset shows an individual nanoresonator of the array. (c) Resonant wavelength vs length for 22-nm-thick gold nanoresonators of width 40 nm (blue) and 55 nm (red) as obtained from three-dimensional full-wave electromagnetic simulation. Dashed horizontal lines show the $G$ (magenta) and $2D$ (green) Stokes wavelengths for a pump laser at 633 nm. The circles and squares mark the resonator lengths associated with the maximum measured enhancement factors.

Figure 3

Effective enhancement of Raman scattering cross section per nanoresonators as a function of their resonant wavelength. The dashed vertical lines mark the laser line, $G$ peak, and $2D$ peak wavelengths, respectively, from left to right.

Figure 4

Evolution of the (a) $G$ and (b) $2D$ peaks with the SPP resonant wavelength associated with 55-nm-wide nanostructures. The spectra were normalized to have equal heights. The resonant wavelengths (on the left) are related to the nanoresonators’ physical lengths (on the right) via the computations of Fig. 2. The dashed-line curve at the bottom corresponds to a control graphene region with no nanostructures in the vicinity. The vertical dashed line represents the control region peak position (intrinsic Stokes wavelengths in graphene).

Figure 5

Variation of the $G$-peak (a) center and (b) full width at half maximum (FWHM) as a function of the SPP resonant wavelength supported by the nanoresonators. Solid lines are Lorentzian curves used to fit the experimental data.

Figure 6

Variation of the $2D$-peak (a) center and (b) full width at half maximum (FWHM) as a function of the SPP resonant wavelength supported by the nanoresonators. Solid lines are Lorentzian curves used to fit the experimental data.