Creating 2D Compounds

A 2D material, by definition, has no surface dangling bonds. A bulk material with plate-like structure, such as graphite, is composed of thin layers with a weakly bonded cleavage plane between.

What this means is a monolayer of graphite will seek to satisfy its exposed dangling bonds by absorbing other materials. A monolayer of graphene, in contrast, is energetically complete without a second layer. If a second layer is placed on top of it, only van der Waals forces, not chemical bonds, hold the two together.

Nonetheless, the absence of chemical bonds does not mean that graphene monolayers are unaffected by their environment. The adjacent carbon atoms in a second layer carry their own complements of electrons and protons, shaping the band structure that controls the movement of free carriers in the graphene film. The details of this structure depend on the relationship between the crystal lattices in the two layers. Any offset or twist between the two produces a moiré pattern, with periodicity that varies with the amount of misalignment. By controlling the misalignment, researchers can achieve a wide range of band structures.

Generally speaking, a larger twist angle leads to a smaller effective lattice constant. Certain “magic” angles facilitate correlated exciton behaviors, such as superconductivity and Mott transitions. “Twistronic” devices — a term proposed by Stephen Carr and colleagues at Harvard and the University of Minnesota — seek to exploit the unusual physics of these so-called “van der Waals heterostructures.”

Even in graphene, a relatively simple material containing only carbon atoms, the large unit cells arising from the superlattice structure make simulations difficult. One approach, Carr explained in a talk at the Materials Research Society Spring Meeting in Phoenix, uses density functional theory (DFT) to model each layer independently, then calculates the relaxation as multiple layers interact. Though this method works, the DFT calculations for each superlattice cell can be enormous and time-consuming. Another view, based on tight-binding calculations, treats each layer as a periodic electron density distribution, rather than a lattice with individual atoms.

The modeling challenge is even more acute in other two-dimensional materials. In transition metal dichalcogenides like MoS₂, for instance, Mei-Yin Chou professor emerita at the Georgia Institute of Technology, noted that even a monolayer is still three atoms thick.

The strength of atomic interactions varies with distance. Such a monolayer creates a “bumpy” potential field. In hexagonal boron nitride on ruthenium, for example, the atomic spacing can vary from 2.2 to 3.7 angstroms, and the local work function will vary along with it. It’s not helpful to talk about the “overall” behavior of a structure when local variations are so large. Interactions between monolayers with different twist angles and potentially different compositions can produce very complex band structures.

It’s too early to say what practical applications might arise from these van der Waals heterostructures. However, Kirill Bolotin, physics professor at Freie Universitaet in Berlin, used the photoluminescence of MoS₂ as a starting point to explore some possibilities. In the Berlin group’s research, applied strain reduced the bandgap, while non-uniform strain acted as a funnel, driving excitons toward the point with the highest strain and smallest bandgap.

The group also investigated electro-optical MoS₂ FETs, which are interesting because their absorption and photoluminescence are tunable. Their photoluminescence quantum yield is low, though, as is photon absorption. To address these limitations, Bolotin proposed a hybrid device, using MoS₂ to drive illumination from a CdSSe quantum dot. CdSSe was chosen because its luminescence is in resonance with the B peak of MoS₂. While MoS₂ absorbs 80% of near-field photons from the quantum dot, applying a gate voltage to the MoS₂ resulted in a 75% increase in luminescence from the dot.

While the lack of dangling bonds is fundamental to the unique properties of these materials, it also poses unique integration challenges. Structures that depend on precise alignment (or misalignment) between two-dimensional layers will require a way to transfer or deposit multiple layers while maintaining the correct registration. Reliable alignment control is essential if two-dimensional materials are to find applications in practical devices, rather than merely serving as a playground for physicists. To that end, Imec is developing a generic integration flow, with tools and processes that can be used for whatever structures might ultimately emerge.

Related Stories
Can Graphene Be Mass Manufactured?
Numerous questions surface about whether 2D materials are the best choice for extreme scaling, and so far there are few answers.
What’s A Mott FET?
Strange physics and future devices.
The Good And Bad Of 2D Materials
Single-layer materials are a long way from displacing or even complementing silicon, but no one is ruling them out.
Materials Knowledge Center
Top stories, blogs, white papers and more on semiconductor materials