Demonstration of molecular beam epitaxy and a semiconducting band structure for I-Mn-V compounds

Our ab initio theory calculations predict a semiconducting band structure of I-Mn-V compounds. We demonstrate on LiMnAs that high-quality materials with group-I alkali metals in the crystal structure can be grown by molecular beam epitaxy. Optical measurements on the LiMnAs epilayers are consistent with the theoretical electronic structure. Our calculations also reproduce earlier reports of high antiferromagnetic ordering temperature and predict large, spin-orbit-coupling-induced magnetic anisotropy effects. We propose a strategy for employing antiferromagnetic semiconductors in high-temperature semiconductor spintronics.

Figure 1

Two branches of the closest relatives of silicon emerging from the “proton transfer” rule. The left-hand branch is obtained by imagining one or two proton transfer from the first to the second atom of the primitive cell; the right-hand branch assumes that one proton is transferred into an empty interstitial space in the lattice. Common representatives of these basic semiconductor groups have similar crystal structures as depicted in the figure. Owing to its isovalent nature, Mn naturally extends the elements from the group-II zinc column. Prior to this work, epitaxial growth of the right-hand branch compounds have not been reported and I-Mn-V compounds have not been considered as candidate semiconductors.

Figure 2

(a) Illustration of the competition between the band gap and the FM exchange splitting, which results in the Fermi energy inside the band, i.e., metal character, for large enough exchange splittings. The absence of this competition in AFMs makes strong AFM semiconductors more likely to occur than strong FM semiconductors. (b) Illustration of the survey of magnetic semiconductor counterparts to conventional semiconductors with eight valence electrons per primitive cell. Consistent with (a), most of the magnetic semiconductors are AFMs and the more rare FM semiconductors are found primarily among compounds with more localized (less hybridized) $f$-electron magnetic elements and lighter anions (wider band gaps). Magnetic transition temperatures safely above room temperature are found in the compounds containing the group-I element, and I-Mn-V’s are the simplest representatives of this class of materials.

Figure 3

RHEED images (a) of the LiMnAs film and (b) of the MnAs film after 60 min of MBE growth at identical conditions, except for the Li cell closed in case of the MnAs. (c) Fabry-Pérot interference oscillations of the light back-reflected by the growing LiMnAs film on an InAs substrate plotted as a function of the growth time and wavelength of the detected light. The oscillatory interferences are typical of a semiconductor film. (d) Thickness profile of a LiMnAs wafer across the edges masked by the sample holder during the growth. The total growth time of this wafer was 60 min, resulting in $~200$-nm-thick LiMnAs epilayer.

Figure 4

(a) Full set of (001)-oriented LiMnAs x-ray reflections. Bulk values are denoted by crosses. (b) Azimuthal scans as a function of the wave vector $Q$. The corresponding ${45}^{°}$ in-plane rotation of the LiMnAs unit cell with respect to the InAs substrate is illustrated in the inset. (c) Reciprocal space maps evidencing the vertical alignment of the substrate peak and the peak of the strained LiMnAs film. The black and red crosses in the plot denote the expected positions for the substrate and bulk LiMnAs, respectively. (d) Temperature-dependent remanence at 0.2 mT and magnetic-field-dependent magnetization at 4 K of a 200-nm-thick LiMnAs epilayer (red curve) and of the FM MnAs sample (solid black curve) containing the same amount of Mn. For comparison, we also show the theoretical Brillouin function, which describes magnetization at 4 K of an equal number of Mn atoms represented by uncoupled paramagnetic $S=5/2$ spins in the external magnetic field. (e) Transmissivity of the LiMnAs-InAs sample (red curve) in the near infrared region measured ex situ and compared to the MnAs-InAs sample (black curve), to an unprocessed InAs substrate (green curve), and to a Li-doped InAs substrate (blue curve).

Figure 5

(a) Full-potential LDA calculations of the band dispersions of AFM LiMnAs. Spin-orbit coupling is turned off in this plot for clarity. (b) LDA $+$ U band dispersions of LiMnAs. (c) LDA, (d) LDA $+$ U band dispersions of KMnAs.

Figure 6

Total and element-resolved DOS of LiMnAs calculated in (a) LDA and (b) LDA $+$ U. (c) Complex dielectric function of LiMnAs in the out-of-plane direction calculated in the LDA and LDA $+$ U. (d) Relativistic full-potential LDA calculations of the anisotropy in the DOS with respect to the staggered moment orientation along the [001] and [110] crystal axes. These anisotropies are order of magnitude larger than in the metal Mn-based AFMs and vary strongly near the valence- and conduction-band edges.