2D transition metal dichalcogenides for beyond silicon logic devices: improving the Metal/MoS2 interface through molecular doping.
2D materials have demonstrated enormous potential for a great number of applications such as sensors, spintronic, superconductors, and (photo)electronic devices. From these 2D materials, semiconducting transition metal dichalcogenides (MX2) are of special interest for electronic logic devices such as Field Effect Transistors (FETs), given their interesting properties such as ultra-thin bodies and high electronic band gap that could enable lower standby power dissipations and further boost the performance of devices. MoS2, a member of the semiconducting MX2 family, is normally used as a representative of this family given its robustness under normal environmental conditions, and its natural or synthetic availability. Nevertheless, several challenges need to be addressed before implementing MX2-based devices. Challenges such as understanding and characterizing such devices, reducing high contact resistance, and controlling the doping of such devices.
Therefore, this work focuses on understanding and improving the Metal / MoS2 contact resistance and the controllable doping of MoS2-based devices. To achieve this, first the differences of MoS2 FETs compared to conventional inversion FETs are established. Then careful experiments comparing different characterization techniques are carried out to establish the most reliable way for device parameter extraction. The results show that contact resistance was dominated by a Schottky barrier (SB) in the Metal / MoS2 interface leading to high contact resistance of the devices. This contact resistance surpasses the channel resistance for channels smaller than 100nm. It is therefore clear that for further channel length scaling, so as to enhance the performance of the FET, reduction of contact resistance is of primordial interest. The observations from these experiments were then used to carefully model MoS2 FET behavior by using a semi-classical model. Further insight on the nature of high contact resistance was gained with this model and two major trajectories for the electron injection from the metal contact to the MoS2 film were identified. First of all, a vertical trajectory in which the electrons are injected from the metal to the MoS2 region directly underneath the metal contact, and which are then driven toward the channel through the MoS2 film. Secondly, a lateral trajectory is identified, in which the electrons are injected directly from the border of the metal contact to the MoS2 film in the channel region. These trajectories depend on the height of the SB, the perpendicular, and lateral fields as well as the MoS2 film thickness. For films thinner than 2 layers, it becomes really difficult to accumulate carriers in the MoS2 film underneath the metal contact, and thus the lateral trajectory for injection prevails. It was clear from the model that the MoS2 FET sheet resistance varies greatly spatially, and that the assumption of the same sheet resistance for the whole film does not hold, rather it can be very different in the channel region and under the metal contact, especially for thin MoS2 devices. This introduces an error on the contact transfer length and the contact resistivity of the device when these parameters are extracted using the conventional transfer length method.
Additionally, the model revealed that one effective way to reduce contact resistance was to dope the region of the MoS2 film immediately adjacent to the metal contact to enhance the lateral trajectory, which was always the most relevant trajectory. Two sets of experiments were conducted to demonstrate the possibility of effective and controllable surface doping without degrading the carrier mobility, by using two different approaches: self-assembled physisorbed molecules and spin-coating of polymers. Oleylamine (OA) was used for the self-assembly approach and doping was effectively demonstrated together with contact resistance reduction. The advantage of the self-assembly approach lies with the easiness of controlling the spatial distribution and density of carriers through the self-assembly of the molecules. However, the polymer approach is more industry-friendly and robust thermally. The polymer (polyvinyl-alcohol) approach also demonstrated the possibility of doping. After doping, the contact resistance was reduced by 30%. Finally, the relation between the MoS2 film thickness and the surface doping approach were explored, and it was concluded that surface doping is optimal for film thicknesses below 5.2nm.
In general, even though an accelerated progress has been observed in MoS2 based devices, still additional work is required for a successful integration in industry. Further reduction of the contact resistance is required and more important, a way to integrate these devices with the additional stages of the manufacturing process, where temperature as high as 400 C are required (i.e. interconnects at the back-end) needs to be addressed.