Electronic materials — materials useful for their electrical properties — have driven progress in condensed-matter physics by revealing that unprecedented quantum states of matter can exist, ranging from superconductors to topological insulators. Fundamentally, the character of an electronic state is determined by the density and interaction strength of electrons. In 2018, as predicted1, a structure known as magic-angle twisted bilayer graphene (MATBG) was found to have a narrow electron energy band in which electronic interactions are particularly important2. MATBG belongs to an exceptional group of material platforms in which the electron density can be tuned in situ to switch between insulating and superconducting states3. Writing in Nature, Stepanov et al.4 and Arora et al.5 report that the electron interaction strength in MATBG can be tuned at a fixed electron density through tailored design of the dielectric (insulating) environment.
In principle, the density of electrons in a material can be manipulated in many ways, such as by introducing impurities called dopants or by using electrodes known as gates. However, any such modification can affect much more than just the electron density. For example, adding chemical dopants to a material tends to increase its structural disorder. And gating, if achievable at all, often allows the electron density to be changed only slightly. In general, it is difficult to disentangle the different properties of an electronic state in terms of its kinematics and correlations as the density is adjusted.
The strength of the interactions between electrons in a vacuum is characterized by a fundamental physical constant called the fine-structure constant. However, electron interactions in a material are greatly modified by the surrounding electrons, through a process known as screening. Since the development of Fermi-liquid theory6 (a model of how electrons interact), the quantum theory of screening in metals and semiconductors has become one of the most phenomenologically rich and subtle directions for research in correlated-electron physics7. The reason is that aspects of universality and model-specific features intertwine in screened electron interactions.
In most layered conducting materials, the electron density is the only parameter that can be tuned experimentally to alter the interaction strength. The conduction layers are usually composed of electronic orbitals (regions of space in which electrons can be present) that have a spatial extent of only about one nanometre. As a result, the inherent electronic screening, which depends strongly on the electron density, often dominates any effect stemming from the surrounding dielectric environment of the conduction layers.
Stepanov et al. and Arora et al. have shown that the electron interaction strength in MATBG can be tuned at a fixed electron density. MATBG comprises two layers of graphene (2D sheets of carbon atoms) that are stacked with their hexagonal lattices rotated out of alignment by an angle of about 1.1°. The atoms form a periodic structure called a moiré pattern, in which the spatial extent of the unit cell (the smallest repeating unit), and hence of the electronic orbitals associated with the narrow electron energy band, is no more than 15 nm (Fig. 1a). Because these orbitals are much larger than those in usual electronic materials, the dielectric environment of MATBG can strongly affect the electronic screening, and therefore the electron interactions.
The teams took different paths to perform dielectric engineering on MATBG. Stepanov et al. varied the thickness of boron nitride layers that acted as a dielectric spacer between the MATBG and a graphite screening layer (which conducts like a metal). In a related set-up, Arora et al. added a tungsten diselenide layer between the graphene and boron nitride. The dielectric tuning is especially apparent in Stepanov and colleagues’ work. There, the graphite screening can be thought of as inducing mirror charges (charges of opposite sign to those in the graphite) in the MATBG. The screening effect becomes substantial if the thickness of the boron nitride spacer is less than the spatial extent of the moiré-lattice electronic orbitals (Fig. 1b).
In both studies, enhanced screening reduces the electron interaction strength and suppresses the formation of insulating states. As a result, in a phase diagram for MATBG, regions that show an insulating state in untuned MATBG show a superconducting state in dielectrically tuned MATBG. Furthermore, when a magnetic field is applied, these previously insulating regions are associated with an increased propensity to form Landau levels (narrow, field-induced electron energy bands) at remarkably weak field strengths.
Altogether, these findings call into question earlier interpretations of certain observations in MATBG as manifestations of an unconventional form of superconductivity3. Instead, although it is too early to be totally certain, simpler explanations might be more relevant. These include theories centred around an effect known as quantum-Hall orbital ferromagnetism, and also conventional superconductivity mechanisms that result from a coupling between electrons and phonons (lattice vibrations), possibly assisted by electron correlations.
The enormous potential for fundamental progress implicit in these developments, as well as the challenges they imply for understanding the mechanisms involved, should be investigated far beyond the specific material at hand. Experimental observations of MATBG vary substantially from sample to sample, raising the issue of reproducibility8. Addressing this problem will probably become more urgent, because dielectric engineering should lead to even greater sample diversity. The tunability of electronic quantum materials, in terms of interactions and density, is increasing substantially, and is catching up with that of synthetic platforms such as ultracold atomic gases deposited in optical lattices. Therefore, we could soon witness the beginning of a new era of discoveries in tunable electronic quantum matter.
Nature 583, 364-365 (2020)