Integrating superconductors with semiconductors has long been thought to be essential to overcome the current limitations of electronic devices, but has been challenging to achieve. In a paper in Nature, Yan et al.1 report their use of a technique known as molecular beam epitaxy to grow layers of semiconductors on top of a superconductor. The resulting device has potentially useful electronic properties that hint at future applications for semiconductor–superconductor interfaces.
The development of increasingly sophisticated electronic devices is aided by efforts to make new combinations of materials — or, more specifically, new interfaces between materials, at which potentially useful electronic effects can occur. The credo underlying this concept is that “the interface is the device”2. This is particularly true for interfaces involving superconductors.
For example, Josephson junctions consist of two superconductors separated by a thin barrier, such as an insulator or a non-superconducting metal. Cooper pairs of electrons — the bound electron pairs that are responsible for superconductivity — can tunnel across the barrier in a fascinating physical process that has led to the development of devices such as those that mix or emit light at terahertz frequencies3. Interfacing superconductors with semiconductors4 such as indium arsenide (an arsenic-based material) can trigger Andreev reflection processes in which a normal electric current becomes a superconducting current. And if a ferromagnetic material (a material that exhibits the type of magnetism associated with iron) is used as the barrier in a Josephson junction, even more opportunities emerge for the manipulation of controllable electronic states5.
Yan and colleagues now report the synthesis of interfaces formed between two nitrides (nitrogen-containing materials), one a superconductor and the other a semiconductor. Nitride semiconductors6 are non-toxic, which makes them much more desirable for most applications than toxic arsenic-containing semiconductors. They can be synthesized in well-established procedures using molecular beam epitaxy — a technique in which atomized elements are deposited on a substrate in a vacuum to form thin films of single crystals. Nitride superconductors are also non-toxic, and, more importantly, are highly stable, particularly in ambient conditions (unlike many superconductors). The authors demonstrate that they can fabricate interfaces between a nitride superconductor and devices known as high-electron-mobility transistors7 (HEMTs) made from nitride semiconductors. HEMTs are widely used in communications infrastructures.
One problem that Yan and colleagues had to contend with is the fact that their nitride semiconductors have hexagonal crystal lattices, whereas the superconductor (niobium nitride) is cubic (Fig. 1). This means that the crystallographic symmetry of their devices is broken at the interface between the cubic superconductor and the hexagonal semiconductor. Such broken symmetries can cause unwanted effects at interfaces, and therefore in devices.
This is where the orientation of the superconductor comes into play. Yan et al. grew a layer of the cubic superconductor on a substrate so that its lattice was oriented in a way that makes it look hexagonal. To picture this, imagine looking at a dice at an angle in which the diagonally opposite corners are aligned. What you see is a hexagon, even though the dice is cubic.
The same is true of the cubic superconductor on the substrate: a hexagonal arrangement of atoms is exposed on the surface, and the hexagonal semiconductor (aluminium nitride) aligns with this when it forms on top of the superconductor. As a result, the aluminium nitride is not perturbed by broken crystallographic symmetry at the interface, and forms an undistorted layer, as needed for the growth of an HEMT structure. Indeed, the authors observed the formation of certain quantum oscillations in their device; the presence of these oscillations is considered a benchmark of high crystal quality.
Yan et al. went on to measure the current–voltage profile of their superconductor–HEMT structure. They observed that this profile of the HEMT is modified by a superconductor-to-metal transition in niobium nitride, and generates a negative differential resistance (NDR) — a property that can be used to increase the power of electrical signals. NDR devices have been known since the end of the nineteenth century88 and include the Gunn diode9, which is widely used to generate microwaves in sensors and measuring instruments. Such devices are of great value for electronic systems that use high-frequency, high-power signals — exactly what is needed in telecommunications networks. In Yan and colleagues’ device, NDR can be switched on or off simply by making the temperature lower or higher than the critical temperature for superconductivity (the temperature below which superconductivity occurs).
Combining materials that have different electronic properties without breaking the crystallographic symmetry at the interface is a remarkable feat. However, the mobility of electrons in the device is currently rather low; much higher mobilities can be achieved in devices that use indium arsenide. Achieving mobilities comparable to those of indium arsenide will be extremely challenging. Moreover, the separation between the superconductor and the 2D electron gas — free electrons that are confined to move in only two dimensions — generated in the device will need to be reduced to enable promising quantum effects.
A future goal could be to use the authors’ system to generate and observe Majorana fermions10 — a type of quasiparticle that would be useful for quantum computing — at the superconductor–semiconductor interface11. Charge carriers in electronic devices can be scattered (for example, by crystal defects), and the average time between scattering events needs to be long to stabilize these quasiparticles. Yan et al. calculate that the charge-carrier scattering time in their devices is impressively long (66 femtoseconds; 1 fs is 10–15 s), but the scattering times will need to be at least 100 times longer, similar to the scattering time in indium arsenide12, to stabilize Majorana fermions. It remains to be seen whether this can be achieved in the authors’ devices.
Ultimately, Yan and colleagues’ work will inspire and accelerate efforts to grow nitride superconductors and nitride semiconductors that enable the ultra-high operating efficiency, structural perfection and opportunities for manipulating electronic properties that have already been achieved in interfaces involving indium arsenide. Because, at the end of the day, the interface is the device.
Nature 555, 172-173 (2018)