Recently, a spin-orbit (SO) interaction observed in narrow-gap semiconductor heterojunctions has attracted much attention,since it can provide a basic operation principle for various novel spintronics devices, such as spin-filters and spin field-effect transistors (spin FETs). The concept of the spin-FET is based on the SO coupling in the two dimensional electron gas (2DEG)causing a Rashba spin precession which can be controlled by a gate voltage . Spin-injection and detection are carried out by ferromagnetic (FM) electrodes. As candidate structures for Rashba effect based spin-FET devices, various high electron mobility transistor (HEMT) structures have been studied [2,3]. The Rashba mechanism has been found to be especially enhanced in narrow gap InAs-based heterostructures .
Our recent investigations into the SO interaction in higher Incontent(~75%) InGaAs 2DEGs have shown a very large Rashba SO interaction coupling constant, α~ 30 × 10-12 eVm, in the normal-HEMT (n-HEMT) structure . As used here, α describes the sensitivity of the spin precession when the electron travels across the electric field (
In this work, molecular beam epitaxy growth of the In0.5Ga0.5As/In0.5Al0.5As thin channel (≤60 nm) narrow-gap i-HEMT interface
and the transport analysis results of the 2DEG confined at the heterointerface are herein reported. The analysis includes the characterization of the in-plane transport as well as the Rashba SO interaction investigated by measuring the Hall effect and magnetoresistance (MR) oscillations in the van der Pauw and Hall bar samples using three kindred in-plane current directions(, [1-10] and ). In the analysis, we focused on the dependency of the low temperature 2DEG mobility and the SO coupling constant (α) upon the different HEMT types (inverted or normal) and the In0.5Ga0.5As channel layer thickness. We also paid attention to the effects of the very thin In0.5Al0.5As cap layer.
The samples used in this study were grown on semi-insulating GaAs (100) substrates using a conventional molecular beam epitaxy(MBE) apparatus. Prior to the growth, an oxide desorption was carried out under As4 flux at a beam equivalent pressure of 1.3× 10-5Torr. The surface oxide desorption process involved slowly ramping up the substrate temperature at a rate of 20℃/min until the reflection high-energy electron diffraction pattern showed a clear 2 × 4 surface reconstruction. This surface reconstruction transformation was adopted as a means to calibrate the substrate temperature, which was set at ~590℃, as measured by an infrared pyrometer. Four different i-HEMT structures were designed and grown in order to investigate the In0.5Ga0.5As channel thickness dependency on the spin-related transport properties and the effects of the very thin In0.5Al0.5As cap layer. An n-HEMT structure was also grown as a reference. The layer structures of the grown samples are shown in Fig. 1 and the corresponding conduction band profiles and charge distributions calculated by taking into account the structure and doping conditions adopted in the growths are shown in Fig. 2. Before depositing the modulated doping heterostructure, the substrate temperature of 350℃,selected to reduce the dislocation that occurs during the growth of the step graded InxAl1-xAs buffer (which has good isolation properties; its final inverse step was adopted in order to reduce the residual strain), was increased up to 450℃ to avoid te point defects that can result from the excess arsenic atoms . The silicon δ-doping density was fixed at ~7 × 1012 cm-2. The channel In0.5Ga0.5As layer was separated by a 20 nm thickness undoped In0.5Al0.5As spacer in order to reduce the remote ionized donor
[Fig. 2.] The calculated conduction band potential profiles (full lineleft axis) and electron charge distributions (dashed line right axis) in the (a) dc = 30 nm and (b) dc = 60 nm inverted high electron mobility transistor (i-HEMT) samples and (c) the normal HEMT (n-HEMT)sample.
scattering. Figure 1(a) shows the inverted In0.5Ga0.5As/ In0.5Al0.5As HEMT structure with the two different channel thicknesses (
The van der Pauw samples were initially characterized by Hall measurements. Figure 3 shows the temperature dependencies of the sheet electron density (
The summary of ns μ and α estimated from the hall bar samples via the Hall effect and MR measurements.
Figure 4 shows the second derivatives (
[Fig. 4.] The magnetoresistance (MR) results for the (a) dc = 60 nm without a cap-layer (WOC) samples and for the (b) dc = 30 nm WOC samples with different current directions. The inset shows the fast Fourier transform result for the [1-10] MR oscillation.
from the traces is that there is almost no in-plane anisotropy for the beating oscillation period, suggesting that there is no inplane anisotropy for α as well. The WAL effect, an alternative and complementary method, was applied to the
Here, Ψ is the digamma function,
[Fig. 5.] The conductivity correction fitting plotted as a function of the magnetic field for the dc = 30 nm with cap-layer (WC) and without cap-layer (WOC) samples in the [1-10] direction. The experimental data are given by the circles and the calculated fitting curve is represented by the solid curve.
Figure 5 presents the experimental data of conductivity correction for the dc = 30 nm WC and WOC samples compared to the calculated curve obtained from the Iordanskii equation. The results of the curve fitting are τφ = 8 ps and Ω1 = 8 × 10-23 J, respectively.From the Ω1 = α
We investigated the basic electronic properties as well as the magnitude and in-plane anisotropy of the Rashba SO interaction in thin channel In0.5Ga0.5As/In0.5Al0.5As i-HEMTs. The low temperature mobility of the 2DEG decreased with a decrease in the thickness of the In0.5Ga0.5As channel layer. In addition, the low temperature mobility of the 2DEG in the sample with thin In0.5Ga0.5As channels is found to be improved by adding a very thin undoped In0.5Al0.5As cap layer onto the top of the layered structure. The former is considered to be due to the increase of the surface charge scattering in the thinner channel sample and the latter probably originates from the suppression of the scattering by the Fermi level pinning with the cap layer. The structural dependent characteristic of the Rashba spin transport was observed through the low temperature magneto resistance and weak anti localization measurements. We determined that a WAL analysis is difficult to apply to a high mobility sample that has a strong Rashba SO interaction giving the relation τSO < τtr. Consequently,a larger α is confirmed in the i-HEMTs than that found in the reference n-HEMT; the α value likely increases with a decreasing channel layer thickness