Novel near-infrared emission from crystal defects in MoS2 multilayer flakes

The structural defects in two-dimensional transition metal dichalcogenides, including point defects, dislocations and grain boundaries, are scarcely considered regarding their potential to manipulate the electrical and optical properties of this class of materials, notwithstanding the significant advances already made. Indeed, impurities and vacancies may influence the exciton population, create disorder-induced localization, as well as modify the electrical behaviour of the material. Here we report on the experimental evidence, confirmed by ab initio calculations, that sulfur vacancies give rise to a novel near-infrared emission peak around 0.75 eV in exfoliated MoS2 flakes. In addition, we demonstrate an excess of sulfur vacancies at the flake's edges by means of cathodoluminescence mapping, aberration-corrected transmission electron microscopy imaging and electron energy loss analyses. Moreover, we show that ripplocations, extended line defects peculiar to this material, broaden and redshift the MoS2 indirect bandgap emission.

The position and the intensity of the peaks confirm the compatibility with the sample beta MoS 2 . There is no evidence of mosaic spread.

Supplementary Figure 14: Raman mapping of E 2g and A 1g vibrational modes
Raman map of the E 2g mode frequency position (left) and of the A 1g mode frequency position (right)

Supplementary Note 1 Additional Cathodoluminescence spectroscopy results
The Gaussian peak fitting in Supplementary Figure 2 reveals that slightly asymmetric peak at 1.25 eV is composed by two different components set at 1.29 eV and 1.23 eV. The peak at 1.29 eV is related to the band-to-band transition of the bulk MoS 2 1 . The 1.23 eV peak is probably due to shallow intra-bandgap levels related to some impurity as cesium 6 or rhenium 7 . The presence of these impurities is due to the geologic nature of molybdenite 8 . In addition the 1.25 eV peak shows a shoulder on the low energy side, reliable with a small components set at 1.14 eV.
The Gaussian peak fitting of Supplementary Figure 1 reveals that the strongly asymmetric peak at 1.07 eV is composed by three different components set at 1.25 eV, 1.10 eV and 0.98 eV. The peaks at 1.25 eV and 1.1 eV are reliable with two components previously found in the pristine molybdenite spectrum, meanwhile the 0.98 eV peak appears only after the mechanical exfoliation process and therefore we can suppose that it is related to the ripplocations formed during this process (see Supplementary Figure 33).
As for the luminescence yield of the 0.75-0.76 eV emission, since the detection system uses a lock in amplifier coupled to a Ge photodetector, it is extremely complicated to determine the quantum yield of a single emission. Therefore appears also at the center of the flake, supporting the hypothesis that this emission is related to an intra-bandgap state.
In addition the peak energy changes depending on the position of the spot mode CL spectra, the emission being peaked at 0.78 eV on the edge and at 0.76 eV in the center of the flake. This effect suggests that increasing the defect concentration the emission peak blue shifts.

Influence of electron beam irradiation on CL spectra
The effect of the electron beam exposure has been carefully considered and studied by CL spectroscopy. In order to

Energy Dispersive X-Ray Microanalysis of Molybdenite
The compositional analysis of Molybdenite is carried out by Energy Dispersed X-Ray spectroscopy (EDX) (Supplementary Figure 9). The analysis reveals that the geological molybdenite is naturally sulfur poor and the exact stoichiometry is MoS X , X=1.9. In addition the EDX analysis reveals the presence of cesium impurities with a concentration of about 1% 2 .

Cathodoluminescence spectroscopy
Supplementary Figure 3 shows the CL spectroscopic analysis of bulk molybenite. It is worth noting that geological molybdenite is used as reference for the light emission properties of bulk MoS 2 . In addition we are able to compare the light emission of pristine (blue line) and cracked (red line) molybenite. The pristine molybdenite CL spectrum presents a peak at 1.25 eV, related to the indirect band-to-band transition of bulk MoS 2 3 . As for the two emissions peaked at 1.14 eV and 1.05 eV, there is no clear attribution in the literature. We can however speculate that the two peaks could be related to intrinsic defects and/or surface states related emissions as for other metal sulfide materials 4,5 . In the case of the cracked molybdenite the 1.25 eV peak intensity decreases drastically with a concurrent broadening of the emission. A new intense emission at 0.79 eV appears in the cracked area.