Chemical ordering rather than random alloying in SbAs

The semimetallic group-V elements display a wealth of correlated electron phenomena due to a small indirect band overlap that leads to relatively small, but equal, numbers of holes and electrons at the Fermi energy with high mobility. Their electronic bonding characteristics produce a unique crystal structure, the rhombohedral A7 structure, which accommodates lone pairs on each site. Here, we show via single-crystal and synchrotron x-ray diffraction that antimony arsenide (SbAs) is a compound and the A7 structure can display chemical ordering of Sb and As, which were previously thought to mix randomly. Formation of this compound arises due to differences in electronegativity that are common to IV-VI compounds of average group V such as GeTe, SnS, PbS, and PbTe, and also ordered intraperiod compounds such as CuAu and NiPt. High-temperature diffraction studies reveal an order-disorder transition around 550 K in SbAs, which is in stark contrast to IV-VI compounds GeTe and SnTe that become cubic at elevated temperatures but do not disorder. Transport and infrared reflectivity measurements, along with first-principles calculations, confirm that SbAs is a semimetal, albeit with a direct band separation larger than that of Sb or As. Because even subtle substitutions in the semimetals, notably Bi${}_{1-x}$Sb${}_x$, can open semiconducting energy gaps, a further investigation of the interplay between chemical ordering and electronic structure on the A7 lattice is warranted.

Figure 1

Unit cell of the A7 structure (a) of pure As, Sb, and Bi (and black P at high pressure). Increasing ionicity leads to the ordered GeTe structure in (b).

Figure 2

Rietveld refinement of SbAs using synchrotron x-ray radiation at beamline 11-BM of the APS. The full refinement range is given in (a), with the high-$Q$ region enlarged to show detail. Ordering between Sb/As produces a change in intensity in the $\left\{101\right\}$ peak at $Q=1.88$ Å${}^{-1}$, highlighted and arrowed. This area is enlarged in (b), where the experimental and refined pattern is compared to the model using fully ordered (dotted line) and random solid solution (dashed line). The difference between the two models (O-D difference) is shown at the bottom of (b). The largest intensity variation in the pattern lies on the $\left\{101\right\}$ peak.

Figure 3

Least-squares refinements to the SbAs PDF collected at room temperature using (a), (b) x rays and (c) neutrons. In both cases, fits are excellent. The inset in (a) shows a two-phase mixture of Sb and As (dashed line) that would indicate clustering. This is not present in the data. In the x-ray PDF, the calculated difference between ordered and solid-solution SbAs is larger than the noise. In the neutron case, due to similar scattering lengths, this difference is well below the noise threshold.

Figure 4

St. John–Bloch plot in the style of Littlewood (Refs. 8 and 51) of IV-VI and group-V compounds using bond orbital coordinates calculated from orbital radii of Chelikowsky and Phillips (Ref. 52). Only the low-temperature structures are considered for this plot. The compound SbAs and the random alloy BiSb (both marked as $\times$) lie within the rhombohedral A7 region.

Figure 5

Sequential Rietveld refinements to high-temperature synchrotron x-ray diffraction data (beamline 1-BM, APS) show disordering of Sb/As around 550 K while maintaining low goodness-of-fit $R_{wp}$. In (b), the temperature dependence of the lattice parameters indicates a first-order transition around 550 K. SbAs remains in the A7 phase until melting.

Figure 6

Resistivity (a) of SbAs shows impurity scattering at low temperature and linear temperature dependence, typical of semimetals. Resistivity of an ingot at higher temperatures (b) displays hysteresis, likely due to the phase transition around 550 K. Concurrent measurement of the Seebeck coefficient $S$ (c) shows $p$-type conductivity with a broad maximum around 500 K.

Figure 7

Experimental reflectivity spectrum (a) of SbAs and best-fit calculated reflectivity (solid line) using Eqs. (2) and (3). The Kramers-Kronig obtained Im($\varepsilon$) and the energy-loss function Im($-1/\varepsilon$) are shown in (b). The Drude fit [Eq. (3)] to Im($-1/\varepsilon$) gives the plasma frequency ${\omega }_p$ and complex dielectric function $\varepsilon \left(\omega \right)$.

Figure 8

Electronic structure of SbAs (relativistic and spin-orbit interaction are included): (a) Band structure with a pseudogap along the $\Gamma L$ direction; (b) projected density of states (DOS) showing a finite value at $E_F$. Detailed band structures around $E_F$ are shown below for (c) SbAs, (d) pure Sb, and (e) As.