Nature
Two-dimensional (2D) transition metal dichalcogenides (TMDCs) are materials with fascinating electronic and optical properties. When 2D TMDCs are put in contact with metals, they form metal-semiconductor junctions (MSJs), whose characteristics are fundamental to the electronic and optical properties of TMDC-based devices.
By carefully pairing a metal (say, gold aka Au) with a 2D TMDC (say, molybdenum disulfide aka MoS2) it is possible to control the energy offset between the work function of the metal and the valence or conduction band edges of the semiconductor, known as the Schottky barrier height (SBH). The SBH is a measure of how easy or hard it is for charge (electrons or holes) to cross the interface and is therefore an important factor in determining overall device performance.
So, for a given TMDC, we can use the choice of metal as a sort of tuning knob to control whether it’s easier for electrons to cross the interface (“n-type”, with the metal work function closer to the TMDC conduction band) or easier for holes to make the jump (“p-type”, with the work function closer to the TMDC valence band).
At least that’s how it’s supposed to work. In practice, the metal work function often gets “trapped” in the band gap of the TMDC, and doesn’t change even when we change our metal contact. This process is known as Fermi level pinning and it destroys our ability to tune the SBH by our choice of metal contact.
The exact cause of Fermi level pinning has been debated and has been variously attributed to the presence of defects, metal-induced gap states, and interfacial effects.
In this work we study two common 2D TMDCs, MoS2 and WSe2, in contact with Au, using a combination of density functional theory (DFT) and the GW approximation. We show that, in contrast to most experimentally measured MoS2-based MSJs and previous DFT studies, defect-free Au/MoS2 MSJ are p-type. Crucially, non-local exchange and correlation effects, which are included in the GW approximation but not in DFT, are key to obtaining the correct p-type contact polarity.
This result strongly points to the presence of defects (typically S chalcogen vacancies) as being the culprit for Fermi level pinning. However, unlike the n-type polarity measured for MoS2-based MSJs, WSe2-based MSJs show predominantly p-type polarity (or ambipolarity) in experimental measurements, despite the similar rate of chalcogen vacancy defect formation in both materials. So, if chalcogen (S) vacancies in MoS2 lead to n-type polarity in MoS2-based MSJs, why don’t chalcogen (Se) vacancies in WSe2 also lead to n-type contact polarity?
Studies have shown that Se vacancies in WSe2 are much better at dissociating O2 molecules than S vacancies, leading to oxygen passivation of the Se vacancies in WSe2-based MSJs. Using large model interfaces with defects (see Figure), we show that chalcogen vacancies in both Au/MoS2 and Au/WSe2 MSJs lead to defects states in the TMDC band gap that pin the Fermi level of the metal near the TMDC conduction band, leading to n-type contact polarity. When the Se vacancy is passivated by O, however, the in-gap defect states disappear, leading to a polarity that is very similar to the defect-free case, which, in the case of Au/WSe2, is p-type. In other words, oxygen passivation of the Se vacancy in WSe2 leads to a return to the intrinsic p-type polarity!
DFT band structures of TMDC/metal junctions with common point defects. a Au/MoS2 with S vacancies, and b Au/WSe2 with O-passivated Se vacancies. Projections onto Mo/S and W/Se orbitals are shown as red and blue circles, respectively. The size of the circles is proportional to the weight of the orbital projection. The DFT valence band offsets (VBO) and conduction band offsets (CBO) are indicated. From a, we see that S vacancies produce gap states that pin the Fermi level near the MoS2 conduction band edge. In b, we see that O-passivated Se vacancies in WSe2 do not have in-gap states, leading to a preservation of the intrinsic p-type polarity. The quasiparticle (QP) GW VBO and CBO, obtained from separate GW calculations, are also indicated. Energies are referenced to the Fermi level, EF. Atomic structures are shown as insets: Au (gold); Mo (purple); S (yellow); W (grey); Se (green); O (red). [Credit: npj 2D Materials and Applications]
Our results, then, reconcile the apparently contradictory contact polarity measurements for MoS2- and WSe2-based MSJ: chalcogen vacancies lead to Fermi level pinning that switches the contact polarity from the intrinsic p-type to n-type, while oxygen passivation of the defects (which is expected to occur much more readily in WSe2 than in MoS2) “undoes” the effect of the vacancy and returns the polarity to p-type.
Ultimately, the contact polarity in these systems results from the complex and subtle interplay between the microscopic details of the MSJ, including structure, non-local exchange and correlation effects, as well as the nature of defects present in the TMDCs. Our findings support the idea that, if growth conditions are carefully controlled, the SBH of 2D TMDC-based MSJs can be tuned to provide the desired contact polarity.
Our full results can be found in npj 2D Materials and Applications: https://www.nature.com/articles/s41699-022-00349-x
Please sign in or register for FREE
If you are a registered user on Springer Nature Materials Community, please sign in