Proximity-Induced “Magic” Raman Bands in TERS Spectra of MoS₂ and WS₂ Deposited on the 1L h-BN-Capped Gold

Over the last decade the tip enhanced Raman scattering (TERS) imaging has become an important nanoscale spectroscopic technique capable of identifying structural peculiarities and defects in 2D semiconductors and their vertical and lateral heterostructures with routine spatial resolution of 10-20nm in ambient. Particularly advantageous implementation of sample geometry for TERS imaging is the so-called gap mode when thin sample is sandwiched between a plasmonic tip and plasmonic substrate. At the same time, the direct contact of 2D semiconductors with metallic substrate makes observation of tip enhanced photoluminescence (TEPL) very difficult if possible at all due to non-radiative decay of excitons in adjacent metal.<br/>To resolve the above problem we suggested to use the monolayer h-BN – capped gold substrates as an ideal platform for the gap mode TERS and TEPL imaging, that on the one hand, should preserve strong gap mode enhancement of Raman signal due to small thickness (0.3 nm) of the dielectric h-BN layer, and on the other hand preserve strong TEPL response due to de-coupling of 2D semiconductors from the metallic substrate. Data collected on mono- and a few-layer-thick crystals of MoS₂ and WS₂ show both the TERS and TEPL response, confirming the validity of the proposed approach.<br/>In addition to the enhancement of both the PL and Raman signal, in the course of assessment of TERS/TEPL response of mono- and a few-layer-thick crystals of MoS₂ and WS₂ deposited on 1L h-BN-capped gold we observed in TERS spectra, completely unexpectedly, appearance of Raman bands at about 796 cm^-1and 76 cm^-1 which are not normally observed in regular Raman spectra of h-BN or WS₂/MoS₂. We can safely state that these “magic” bands belong to h-BN as they appear at the same spectral position in TERS spectra of both the monolayer MoS₂ and WS₂ deposited on the monolayer h-BN capped gold, moreover, the 796 cm^-1 band often was the strongest band observed in TERS spectra, even stronger than A’ mode from WS₂ or MoS₂. Presence of the transition metal dichalcogenide (TMD) monolayer is mandatory for the appearance of these “magic” bands as they are absent outside of the monolayer TMDs in these samples. Literature search showed that similar (but not identical) phenomenon was observed earlier in h-BN encapsulated WSe₂ [1,2], MoSe₂[2] and WS₂[3]. There have been several significant differences between our data and the earlier reported one: in our case we have not been able to observe the “magic bands” in MoSe₂ and WSe₂ @ 1L h-BN@Au, while WS₂ monolayers deposited on the same substrate as WSe₂, showed expected response. More importantly, the excitation laser wavelength dependence in our case was completely different from what was reported earlier: in WS₂-based samples we observed strong “magic” bands with excitation at 830 nm, 785nm, 594nm, but not 633nm, the wavelength closest to the A exciton in this material. This excitation profile is remarkably reminiscent of the excitation profile of the monolayer WS₂ in intimate contact with silver where we observed strong dip of the intensity of main A’ mode in TERS spectra at 633nm excitation wavelength.<br/>We will argue that intricate interaction between the tip-substrate gap plasmon, TMD excitons and most probably, normally mid-IR-active phonons in h-BN is responsible for the appearance of observed “magic” bands. We’ll briefly discuss the next steps in investigation of this fascinating phenomenon such as the use of MoS_2xSe_2(1-x) alloys and MoSSe/MoSeS Janus monolayers on 1L h-BN-capped gold or silver.<br/>REFERENCES.<br/>Jin, C., Kim, J., Suh, J. et al. <i>Nature Phys</i> <b>13</b>, 127–131 (2017)<br/>Jacob J. S. Viner, Liam P. McDonnell, et.al. <i>Phys. Rev. B</i> 104, 165404<br/>Luojun Du, Yanchong Zhao, Zhiyan Jia, et.al. <i>Phys. Rev. B</i> 99, 205410