Optical properties of transition metal dichalcogenide monolayers: a new class of 2D semiconductors

 (See the article written in french for the journal of the "Société française de physique" here )



Since the discovery of graphene in 2004, many layered materials have been thinned down to a monolayer and proven to be stable in ambient conditions. Among these, monolayers of the MX2 family, where M is a transition metal atom (typically Mo, or W) and where X is a chalcogen atom  (S, Se, Te), have been shown to be direct band-gap semiconductors with emission in the visible-near infrared range. This makes them promising candidates for ultrathin electronic and opto-electronic devices. In addition, due to the particular bandstructure and broken inversion symmetry in these atomically thin crystals, the optical selection rules are chiral and electron-hole pairs can be selectively created in one of the two non-equivallent K valleys of the Brillouin zone. This could allow to store and process information via the valley index of carriers in these ideal 2D systems.


               Photoluminescence of TMD monolayers deposited onto SiO2 at cryogenic temperatures


As shown in the above Figure, the optical properties of these materials are dominated by excitons: bound electron-hole pairs with strong binding energy due to quantum confinement and weak screening of the Coulomb interaction. For all the members of the TMDC family, the photoluminescence spectrum at low temperatures exhibits well resolved neutral excitons (X) and charged excitons (T) peaks. However, for MoS2, which is particularly interesting due to it's abundance in nature (molybdenite), the spectrum which is usually reported in the litterature consistes of a single, very broad peak (50 meV linewidth) at around 1.9 eV that has been attributed before to the neutral exciton. This is in stark contrast with the spectrum shown  in the above figure, which suggests that the spectrum of monolayer MoS2 is actually similar to the ones observed for the other members of TMDC family, with well defined X and T peaks. Where does this discrepancy comes from?. As will be discussed below, it comes from laser induced-doping and degradation of the optical quality of MoS2 monolayers due to interactions with the SiO2 substrate.


What is the optical spectrum of monolayer molybdenum disulfide (MoS2) ?


- Laser induced changes in optical properties of monolayer molybdenum dichalcogenides

While studying the optical properties of monolayer MoS2 and MoSe2 deposited onto SiO2 substrates, me and my colleagues at LPCNO (Toulouse) have discovered that laser exposure changes the photoluminescence spectrum of monolayers in a non-reversible way. This happens at power densities that so far have been considered to be non-destructive (fractions of µW with a laser spot size of about 1 µm). We have shown that after several minutes of laser exposure, the trion (charged exciton) emission increases while the neutral exciton emission decreases, suggesting a laser induced doping of the monolayers. This is shown in the top pannels of the following figure; the spectra as a function of excitation power reveals an increase of the trion emission in both MoS2 and MoSe2 monolayers. When going back to the lowest power excitation, the PL shape has significantly changed (bottom pannels).



Interestingly, in MoS2 monolayers the neutral exciton emission can be totally quenched when exposed to a laser of enough power. When this happens, the photoluminescence spectrum of MoS2 at low temperature consists of a broad peak at 1.9 eV followed by a small low energy emission (see blue curve below). This is the most common PL spectrum of MoS2 found in the litterature. This broad peak has been attributed before to the neutral exciton emission. In reality, it lies at a much lower energy than that of the exciton, which is found to be around 1.96 eV previous to high laser exposure. The figure below shows the PL spectrum of a MoS2 monolayer before (red curve) and after (blue curve) being exposed to a pulsed laser with an average power of several tens of µW. Note that both spectra have been obtained for the same monolayer under identical conditions, revealing that the optical properties of this monolayer has been changed in a non reversible way.



The exact microscopical mechanism by which laser exposure changes the optical properties of TMD monolayers still needs to be clarified, but it seems clear that the SiO2 substrate plays a key role. Indeed, by inserting a buffer layer of hexagonal boron nitride (h-BN) between the silicon substrate and the monolayer, the effects of laser exposure are dramatically reduced. The most immediate consequence of these findings is that in order to study the intrinsic properties of excitons in MoS2 (and other members of the TMD family), very low excitation power densities should be employed if the monolayers are directly in contact with SiO2. In many cases, experiments are performed with pulsed lasers, in which case the peak power densities can be easily larger than the ones required to change the optical response of these materials.


See the full article here (published in 2D Materials)




- Valley coherence and valley polarization in MoS2 monolayers

It has been shown recently that treating MoS2 monolayers with an organic superacid (TFSI) can increase the photoluminescence quantum yield up to nearly 100% at room temperature (see here). We have employed this technique and performed photoluminescence spectroscopy at T= 4 K on MoS2. We show that this treatment reduces the localized emission induced by defects on the monolayer, and together with the use of very low excitation power, we were able to obtain very well defined neutral exciton (X) and trion (T) peaks at 1.96 eV and 1.93 eV, respectively. Since the majority of published MoS2 spectra consist of a single broad peak at 1.9 eV, we have performed polarization resolved photoluminescence in order to confirm the correct identification of both peaks.

When excited with circularly polarized light, only one of the two non-equivallent K valleys of the Brillouin zone will be occupied, and at steady state both excitonic complexes exhibit partially circularly polarized emission (see figure below). The very sharp peak at 1.88 eV correspond to a Raman peak (the laser energy is 2.1 eV in these experiments). Note that the low energy emission, possibly related to localized excitons, is unpolarized.



When excited with linearly polarized light, which is a mixture of right and left circularly polarized photons, a coherent superposition of excitons is created in both valleys (valley coherence). As a consequence, the neutral exciton emission is linearly polarized in the same axis of that of the incident laser (see figure below). As expected, the trion complex presents no linear polarization.


See the full article here (published in Applied Physics Letters)



- Approaching the homogeneous linewdith in MoS2 based Van der Waals heterostructures

Recently, we have shown that encapsulating TMDC monolayers with few-layer hexagonal boron nitride (h-BN) allows to access the intrinsic, high optical quality of these 2D crystals. Indeed, we observe an excitonic linewidth as low as 2 meV at low temperatures (see figure below), which is more than one order of magnitude narrower than the typically broad peak that has been attributed to the neutral exciton in the past. In addition, no signature of charged excitons is visible in the photoluminescence spectra. This spectacular improvement can be due to several reasons:  the capping layers act as a barrier to charge transfer from the SiO2 substrate to the monolayers, and it also avoids any transfer of the substrate's roughness to the MoS2.



As shown in the figure above, excitation with a circularly-polarized laser 25 meV above the exciton resonance  results in a high degree of circular polarization at steady state (35%), whereas using a linearly polarized laser results in robust valley coherence as evidenced from  the high degree of linear polarization of the exciton peak (55%).

Applying a perpendicular magnetic-field results in a Valley Zeeman effect and a rotation of the linear polarization of the luminescence, which is a key result in what concerns the possibility of manipulating a coherent superposition of valley states in these promising 2D systems.


See the full article here (published in Physical Review X)





PhD thesis: Spin-dependent electron transport in semiconductors.


I did my PhD in the Electrons-Photons-Surfaces group at the Condensed matter physics laboratory of the Ecole polyechnique,

under the supervision of Daniel Paget and Alistair Rowe. 

My thesis work was funded by the Gaspard Monge international grant.


To get the manuscript of my PhD thesis, click here.