Our background IP and previous calculations established the proof of concept for electric field switching in a bilayer two-dimensional material with the insertion of a middle spacer layer (1–4). The spacer layer not only acts as an interfacial stabiliser to hold the two separated atomically thin films, but also serves as a decoupling layer that drives the electronic structure towards a degenerate, monolayer-like form. As a result, the Stark effect induced by the external electric field becomes more effective because of reduced interlayer screening, leading to an enhanced field response of the transistor channel.

We have also shown that this working principle is general for two-dimensional heterostructures that form a Type-I (straddling) band alignment upon contact. i.e. The valence band maximum (VBM) and conduction band minimum (CBM) of the middle spacer should enclose those of the active thin-film material. Here, we use MoS₂ and hBN, two widely available materials in the field, to form a heterostructure channel and experimentally verify this concept.

In the following, we demonstrate the successful delivery of the first working device incorporating a MoS₂/hBN/MoS₂ channel, whose performance is significantly better than that of a conventional MoS₂ bilayer. Advanced characterisation has confirmed the morphology and stability of the channel. These observations are consistent with our previous density functional theory (DFT) calculations and support the proposed mechanism behind the enhanced electric field response.

Device Fabrication

The metal electrodes (bottom gate – 15nm, source-drain contacts – 20nm, top gate – 50nm) were firstly fabricated on sacrificed SiO₂/Si substrates using standard photolithography, electron-beam evaporation (Au) and lift off process. A bottom gate electrode was picked up and laminated onto a precleaned SiO₂/Si substrate by the PC-assisted dry transfer method, after which a ~20-nm-thick hBN flake was transferred on top via the same technique to form the bottom-gate architecture. Following an annealing step (300 °C in Ar ambient for 2 hours), exfoliated monolayer MoS₂ (bilayer MoS₂ or monolayer MoS₂/monolayer hBN/monolayer MoS₂) were transferred onto the gating area by the PC method. The stack was then annealed at 200 °C in an Ar atmosphere for 2 hours. Finally, a ~20nm hBN and top gate was transferred on the target region to complete the dual gated field-effect transistor. Here it should be noted that we van der Waals integrated all layers including the metal electrodes, which optimised the contact performance and reduced the gate leakage risks, resulting in a higher device yield.

Device Performance

To evaluate Stark-effect bandgap engineering in a decoupled 2D semiconducting system, we compared three dual-gated architectures: monolayer MoS₂ (Fig. 1a), bilayer MoS₂ (Fig. 1c), and a monolayer MoS₂/few-layer hBN/monolayer MoS₂ heterostructure (Fig. 1e). Independent top and bottom gates enable decoupled control over both the carrier density and the out-of-plane electric field across the active channel. We first confirmed the independent functionality of each gate by sweeping one while grounding the other. Next, we mapped the source-drain current across the full - bias space. In the monolayer device (Fig. 1b), the constant-current contours maintained a fixed slope across both subthreshold and high-current (linear) regimes, as expected when current depends solely on the net carrier density. Conversely, for bilayer MoS₂, the contour slopes slightly decreased under high out-of-plane electric fields in the subthreshold regime. This slope deviation became significantly more pronounced in the monolayer MoS₂/thin hBN/monolayer MoS₂ heterostructure under high-field conditions. We attribute this phenomenon to Stark-effect-induced bandgap modulation (). Inserting the thin dielectric spacer (such as few-layer hBN) strongly decouples the two semiconductor monolayers, suppressing channel quantum capacitance screening and allowing the out-of-plane field to efficiently shift the bands, thereby altering the bandgap engineering.

Characterisation of the Heterostructure

To evaluate the interfacial quality of the dual-gate MoS₂ devices, we used cross-section tunnelling electron microscope (TEM) to take high-resolution images. The device photo is shown in Fig. 2a, in which the channel of MoS₂/hBN/MoS₂ is sandwiched by two multi-layer hBN/Au gates. The black line highlights the area of interest, which has been lifted out using a dual-beam SEM (scanning electron microscope, Thermoscientific FEI Helios 5) with a Ga source focused ion beam. The stackings of 2D materials has been sketched in Fig. 2b. The low-resolution cross-section image was taken using the scanning electron tunnelling microscope (STEM) function in Helios 5 (with 30 kV accelerate voltage). The sharp interface across the whole device is shown in Fig. 2c. We further show the zoom-in plots in Fig. 2d, in which one can see the near comparable thickness of top- and back-dielectric hBN layers. The high-resolution photo (Fig. 2e) was taken using the FEI Talos TEM, running at 200 kV and in bright-field mode. The spacer layer in this device is 3 layers of hBN, ~ 1.3 nm.

Materials Modelling

We have also performed DFT calculations to understand the electronic structure of the MoS₂/hBN/MoS₂ heterostructure and its response to an external electric field. As shown in Fig. 3a, the band gap of MoS₂ with an hBN spacer is significantly more sensitive to the external electric field compared with monolayer and bilayer MoS₂.

This enhanced response is also reflected in the maximum switching figure of merit (FOM, ) for the three systems, as shown in Fig. 3b. The FOM is a dimensionless metric that quantifies and normalises the electric field response across the channel, defined as , where is the band gap and is the channel thickness. A higher indicates a stronger band-gap response to the electric field for a given channel thickness.

We can also observe clear band separation and shifts in MoS₂ with the hBN spacer under the external electric field, as shown in Fig. 3c-g. These results are consistent with our experimental observations and complete the proof-of-concept demonstration and validation of the proposed heterostructure design for enhanced electric field response.

Next Steps

1. Ferroelectric incorporation and device modelling. The next step is to incorporate ferroelectric materials into the heterostructure channel to achieve non-volatile switching. We will also perform device modelling to predict the performance of devices incorporating ferroelectric materials.
2. Engagement with new collaborators. We are currently engaging with new collaborators to expand the scope of the R&D program, particularly in large-scale crystal growth and device fabrication.