Supercurrent in van der Waals Josephson junction

Supercurrent flow between two superconductors with different order parameters, a phenomenon known as the Josephson effect, can be achieved by inserting a non-superconducting material between two superconductors to decouple their wavefunctions. These Josephson junctions have been employed in fields ranging from digital to quantum electronics, yet their functionality is limited by the interface quality and use of non-superconducting material. Here we show that by exfoliating a layered dichalcogenide (NbSe2) superconductor, the van der Waals (vdW) contact between the cleaved surfaces can instead be used to construct a Josephson junction. This is made possible by recent advances in vdW heterostructure technology, with an atomically flat vdW interface free of oxidation and inter-diffusion achieved by eliminating all heat treatment during junction preparation. Here we demonstrate that this artificially created vdW interface provides sufficient decoupling of the wavefunctions of the two NbSe2 crystals, with the vdW Josephson junction exhibiting a high supercurrent transparency.

The surface of NbSe 2 was evaluated through scanning transmission electron microscopy (STEM) and energy dispersive X-ray spectroscopy (EDX). Supplementary Figure 1 prresents the STEM images and EDX profile comparing a non-baked and baked NbSe 2 surface. The EDX profile for Oxygen (O), Selenium (Se), and Niobium (Nb) are plotted.
Here, the non-baked surface is produced by the same device fabrication procedure used to create the Josephson junctions, but the surface is capped with carbon in place of another NbSe 2 crystal. In contrast, the baked surface is produced by spin coating the surface of NbSe 2 with a polymethyl methacrylate (PMMA) resist (without a capping layer), followed by baking at 180 °C for 30 min in a convection oven, and, thereafter, removing the PMMA resist using acetone. The EDX results clearly show a difference in the degree of surface oxidation between these samples, with a ~3-nm-thick surface oxide layer formed on the baked NbSe 2 sample. This oxide layer was confirmed through X-ray photoelectron spectroscopy (XPS) measurement to be Nb 2 O 5 . Oxidation of the non-baked NbSe 2 sample, on the other hand, is greatly reduced to below the resolution of EDX measurement.

Supplementary Note 2: XPS analysis of a cleaved NbSe 2 surface
The results of the X-ray photoelectron spectroscopy ( More than 20 NbSe 2 /NbSe 2 van der Waals (vdW) junction devices were fabricated to determine the nature of junction, and the data obtained from some of these devices are presented in Supplementary Figure 4(a,b,c) and 5(a,b). As can be seen from these figures, both the Josephson critical current and the hysteresis in the current-voltage (I-V) curve vary from device to device, even though the device fabrication conditions were the same.
The Josephson effect is always present when the normal state junction resistance-area product RA is sufficiently low (i.e., less than ~300 μm 2 ); these devices are shown in Supplementary Figure 4(a,b,c). Conversely, with a high resistance junction such as that shown in Supplementary Figure 5(a,b), only a transport property of tunnel barrier was observed without zero bias Josephson current. Thus, it is crucial to reduce the junction resistance to observe the Josephson effect in a vdW junction. For fabricating such low resistance vdW junctions, the use of room-temperature device fabrication methods including exfoliation, dry-transfer, lithography, and metal evaporation play a key role.
Since there is no metal-oxide present at the vdW junction, we believe that the variation in resistance that was observed between the different devices is caused by a difference in the separation of the two NbSe 2 flakes at the vdW junction. This variation is therefore the limit of the present fabrication method, as it involves only the contact between two flakes of NbSe 2 . It is likely that this separation depends on the quality of the fleshly cleaved NbSe 2 surface, the amount of surface adsorption on NbSe 2 , and the force applied during transfer of the NbSe 2 flake. Although the junction resistance of the devices showed some variation despite using the same fabrication procedure, the Josephson effect is always present when the resultant vdW junction has a low normal-state resistance.
In addition, it was found that the junction resistance could be lowered by proper annealing, with the resultant low-resistance junction exhibiting the Josephson effect. the behavior of I 1 shows good agreement with Ambegaokar-Baratoff (AB) theory for a symmetric junction, as discussed in the main text, I 2 exhibits a significantly different temperature dependence. However, when the temperature dependence of I 2 is compared with the critical temperature required to break the superconductivity in bulk NbSe 2 , the results are found to be in good agreement. It is therefore believed that I 1 is related to the junction, whereas I 2 is a contribution from the bulk NbSe 2 . Note that bulk NbSe 2 does not usually experience hysteresis under the application of current, but as a long Josephson junction can produce an inhomogeneous current flow around it, some heating or phase slip can occur in the bulk material close to the junction area. Consequently, hysteresis can occur even in the bulk crystal.