Traditional Semiconductors in the Two-Dimensional Limit

Interest in two-dimensional materials has exploded in recent years. Not only are they studied due to their novel electronic properties, such as the emergent Dirac fermion in graphene, but also as a new paradigm in which stacking layers of distinct two-dimensional materials may enable different functionality or devices. Here, through first-principles theory, we reveal a large new class of two-dimensional materials which are derived from traditional III-V, II-VI, and I-VII semiconductors. It is found that in the ultrathin limit the great majority of traditional binary semiconductors studied (a series of 28 semiconductors) are not only kinetically stable in a two-dimensional double layer honeycomb structure, but more energetically stable than the truncated wurtzite or zinc-blende structures associated with three dimensional bulk. These findings both greatly increase the landscape of two-dimensional materials and also demonstrate that in the double layer honeycomb form, even ordinary semiconductors, such as GaAs, can exhibit exotic topological properties.

Figure 1

(a) Top and side views of bilayer-thick TB and (b) DLHC GaAs. Red arrows in (a) indicate atomic displacements to form DLHC from TB. Charged atoms are denoted schematically by the (+) and (−) signs. (c),(d) The corresponding ELFs with contour values ranging from 0 (blue) to 0.8 (red).

Figure 2

Charge density difference between DLHC and SLHCs for GaAs in (a) top and (b) side view, respectively. The displayed isosurface value is $3\times {10}^{-3}\mathrm{e}/{\AA }^3$ with the light brown being positive and green being negative. (c) Formation energy per formula unit of the calculated DLHCs (relative to infinite periodic bulk) as a function of Phillips ionicity. The dotted line is a two parameter least-squares fit to the data.

Figure 3

Formation energies of DLHCs (open circles) and TBs (solid lines) as a function of layer thickness for (a) GaAs, (b) HgSe, and (c) AgI. The thickness $n$ corresponds to the number of stacked DLHCs and is 4 times the number of atomic layers. In (a), the formation energy of SLHC and a ${(\mathrm{GaAs})}_8$ cluster is also shown.

Figure 4

Atomic and band structures of GaAs by HSD. (a) SLHC. (b) Character ${(c_S)}_n$ of the $n$th wave function of SLHC and parity $P_{n^{\prime }}$ of the $n^{\prime }$th wave function of DLHC at Γ (see main text for definition). (c) DLHC. Inset at the bottom of (b) is an enlargement of the framed area in (c) showing band inversion. Energy zero is at the valence band maximum.

Figure 5

Atomic and band structure of GaAs. Left is DLHC and right is 2-DLHC. In going from DLHC to 2-DLHC, splitting occurs in both the VBM and CBM. Both states, marked with (−) parity, split into a higher lying (−) state and a lower lying (+) state. This leads to a second band inversion (indicated by the crossing between the thickened red and blue lines) which makes the GaAs 2-DLHC a TI.