Epitaxial growth of (100)-oriented SmN directly on (100)Si substrates

We demonstrate growth of epitaxial (100)SmN thin films directly on (100)Si surfaces. By using physical vapor deposition of Sm metal in an ${\mathrm{N}}_2$ atmosphere we show that careful control of substrate temperature, ${\mathrm{N}}_2$ pressure, and postannealing steps leads to epitaxial SmN without the formation of samarium silicide impurity phases. While rare-earth silicide formation competes with and is favored over nitride formation at high growth temperatures, we find that low-temperature grown SmN seed layers are stable against high-temperature annealing, and thus allow for subsequent high-temperature growth of SmN with a clear epitaxial relationship to the Si substrate. The relatively low lattice mismatch of SmN with (100)Si, compared to other commonly available substrates, coupled with the low cost and maturity of Si processing technology provide a promising route for further studies of the fundamental properties of SmN and other isostructural members of the rare-earth nitride series. Because SmN is a ferromagnetic semiconductor, which also becomes superconducting close to 4 K under sufficient doping, integration with Si technology presents new opportunities for spin-transport devices.

Figure 1

In situ STM image of the (100)Si substrate after deposition of 50 nm homoepitaxial Si.

Figure 2

RHEED patterns along the $\langle 110\rangle$ Si azimuths. (a) (100)Si substrate, the arrows mark the additional lines from the ($2\times 1$) surface reconstruction, (b) 4 nm SmN seed layer, and (c) 30 nm SmN main layer, each taken at high temperature after their respective annealing step.

Figure 3

RHEED patterns along the $\langle 100\rangle$ Si azimuths. (a) 4 nm SmN seed layer and (b) 30 nm SmN main layer, each taken at high temperature after their respective annealing step.

Figure 4

(a) XRD of a 5 nm SmN seed layer after annealing at $600{}^{\circ }\mathrm{C}$, capped with 30 nm amorphous Si. (b) XRD pattern of a 52 nm SmN film. Additionally to the SmN and Si reflections, the positions of the most prominent ${\mathrm{SmSi}}_2$ reflections are marked, proving there is no detectable amount of silicide in this sample. The inset shows the SmN 200 rocking curve with a FWHM of $1.89{}^{\circ }$.

Figure 5

$\varphi$ scan of the 111 reflections of (a) the SmN layer and (b) the Si substrate of the same sample. The $90{}^{\circ }$ spacing of the peaks and the close alignment between SmN and Si peaks are evidence for the cube-on-cube epitaxial relationship.

Figure 6

(a) Field cooled (FC) and zero-field cooled (ZFC) total magnetic moment of a SmN sample which includes a large diamagnetic substrate contribution. Also shown is the total moment of a Si substrate, measured in a magnetic field of 25 mT. (b) FC and ZFC magnetization of the SmN film after subtracting the diamagnetic Si contribution shown in (a). (c) Temperature dependent resistivity measured in zero magnetic field with increasing temperatures.