The lattice mismatch of Ξπ = 6.7%, with π being the lattice constant, between the zinc blende InAs and GaAs crystal structures [101] prevents the epitaxial growth of InAs-based compound systems on the commonly employed GaAs substrate. Thus, except for the ternary compound In0.53Ga0.47As, which is lattice matched to available - yet expensive -
InP substrates, we lack a lattice matched substrate for In-containing ternary alloy systems. To overcome this limitation in the choice of material system, sophisticated buffer layer
5 Design and characterisation of InGaAs/InAlAs systems
concepts have to be applied to accommodate for the lattice mismatch between the active layer and the employed substrate material whilst ensuring strain relaxation during the buffer layer growth in order to achieve defect-free active layer structures [35, 36, 38, 102β108]. In this work, we use semi-insulating GaAs (001) as substrate material for the
MBE growth process. In order to achieve high zero-field spin-splitting, we choose an (InAs/)In0.75Ga0.25As QW as host material for the 2D electron system. As the high band
gap material, we employ In0.75Al0.25As since it is lattice matched to In0.75Ga0.25As at this
particular In concentration.
Figure 5.1: (a)Schematic layer sketch of the applied step-graded InxAl1-xAs
buffer layer to accommodate for the lattice mismatch between the used GaAs (001) substrate and the active In0.75Ga0.25As/In0.75Al0.25As layers, which exhibit
a significantly larger lattice constant than GaAs. (b) The aligned cross-sectional TEM image along the [110] crystallographic direction illustrates the formation of in-plane misfit dislocations due to plastic strain relaxation inside the buffer layer. The constant composition layer, i.e. the In0.75Al0.25As VS, is almost free
of threading dislocations, implying a well-functioning buffer layer system. (c) The HRTEM image of the VS shows the lattice planes of the In0.75Al0.25As zinc
blende lattice. (d) Evaluating the grey-scale oscillations of a line-cut along the [110] direction yields the lattice constant of the corresponding In0.75Al0.25As
layer.
A key to defect-free strain relaxation is the minimization of the amount of threading dislocations which reach the sample surface, whereby they penetrate the active layers and hence the 2DEG. To this end, the interaction of misfit dislocations parallel to the sample surface has to be reduced, which can be tailored by the applied buffer layer. Following the epitaxial growth study of Capotondi et al. [103], we employ an InxAl1-xAs step-graded
buffer layer concept in the epitaxial growth process of our heterostructures. Therein, the In concentration of the ternary compound material InxAl1-xAs is stepwise increased
during the growth process. A compositional overshoot to an In concentration of 85% is commonly employed as it has been proven to be beneficial for the strain-free adaption
5.1 Buffer layer growth
of the lattice constant for the 75% In-containing In0.75Ga0.25As/ In0.75Al0.25As active
layers. Figure 5.1(a) displays a sketch of the employed buffer system. The epitaxial growth process is initiated with a deoxidation of the GaAs substrate wafer, followed by the growth of a 100ππ GaAs seed layer. The subsequent incorporation of short-period alternating Al0.50Ga0.50As/GaAs layers, i.e. the superlattice (SL), with each layer having
a thickness of 5ππ, has been proven to be crucial for the screening of the following InxAl1-xAs buffer layers from impurities, which reside at the GaAs substrate interface.
Our InxAl1-xAs step-graded buffer system starts with an In concentration of π₯ β 0.07 and
is increased in 5%-steps to π₯ = 0.75 by keeping the Al flux constant and ramping the In cell temperature. Each InxAl1-xAs buffer step has a thickness of 50ππ. The step-graded
buffer is terminated by two broader, auxiliary InxAl1-xAs steps with π₯ = 0.80 and π₯ = 0.85,
presenting the compositional overshoot, whereby residual compressive strain is reduced. Subsequently, the In concentration is reduced to 75% and several hundred nanometers of In0.75Al0.25As, the so-called virtual substrate (VS), are grown. The VS provides the
strain-relaxed substrate for the following active layer system.
During the growth process, the BEP of As4 is commonly kept at 8 Β· 10β6π ππ π. For the growth of the GaAs seed layer and the SL, the temperature of the substrate holder of the MBE system is set to π = 620β¦
πΆ. For the InxAl1-xAs buffer layer the substrate temperature is reduced to πππ’ π β360
β¦
πΆto prevent the formation of 3-dimensional crystal islands during growth [37, 109]. Based on the work of Loher [97], who determined that an increase of the substrate temperature during the growth of the VS and the active layer system to ππ ππ‘ β 460
β¦
πΆ proves to be beneficial to achieve a highly mobile 2D charge carrier system, we employ a similar substrate temperature for the growth of our active In0.75Ga0.25As/In0.75Al0.25As layers.
To investigate the functionality of the applied buffer layer system, detailed TEM-based studies of the step-graded InxAl1-xAs buffer are conducted. Thereby, we are able to
analyse the crystalline quality of the semiconducting layers, the strain relaxation, as well as the incorporation of structural defects in the system. Figure 5.1(b) shows an exemplary cross-sectional TEM image of the InxAl1-xAs buffer system along the [110]
crystallographic direction, aligned to the schematic sketch of the buffer layer sequence in figure 5.1(a). We find that misfit dislocations efficiently annihilate each other inside the buffer layers since the VS is almost free of threading dislocations. This demonstrates effective strain relaxation by means of the employed buffer concept. A single piercing threading dislocation can be seen as dark line in the centre of the TEM image in figure 5.1(b). Figure 5.1(c) depicts a high-resolution TEM (HRTEM) image of the In0.75Al0.25As
VS layer from figure 5.1(b). The zoom into the image reveals the individual crystal lattice planes as small grains. Since we know the crystal orientation of the HRTEM image, we are able to determine the effective lattice constant π(πΌππ₯π΄π1βπ₯π΄π )of the VS layer from the
oscillation of the grey-value in a line-cut through this image. From the thereby obtained value of π(πΌππ₯π΄π1βπ₯π΄π ), we are able to deduce the material composition of the layer.
This presents an experimental opportunity to cross-check the applied In calibration in the epitaxy of our heterostructure. The applied calculation method is illustrated in the sketch in figure 5.1(d). With π(πΌππ΄π ) = 6.0583 Λπ΄ and π(πΊππ΄π ) = 5.65325 Λπ΄ at π = 300β¦
5 Design and characterisation of InGaAs/InAlAs systems
[101], we can write in first approximation π(πΌππ₯π΄π1βπ₯π΄π ) =5.65325 Λπ΄ + 0.40505 Λπ΄ Β· π₯.
Evaluating the grey-value oscillation period yields an In composition of π₯ = 0.745. This value is in excellent agreement with the targeted In concentration of 75% for the VS and thus confirms our employed In-cell calibration. Further in-depth investigations of the applied InxAl1-xAs step-graded buffer concept via TEM can be found in [93, 97].
Figure 5.2(a) displays an exemplary plan view AFM image of the wafer surface of an In0.75Ga0.25As/In0.75Al0.25As heterostructure, for which the above described step-graded
buffer layer concept has been utilised.
Figure 5.2:Wafer C160428A: (a) 20 π₯ 20ππ plan view AFM image of the wafer
surface, revealing 5 β 15ππ deep trenches along the h110i crystallographic directions. (b) Profile sections along the π¦ k [Β―110] and π₯ k [110] direction. (c) Three dimensional plot of the profile of the wafer surface shown in (a).
We see clear undulations in the surface structure of the wafer in form of a cross hatching pattern, which is perfectly aligned along the [Β―110] and [110] crystallographic directions with a pronounced anisotropy in the corresponding oscillation periods. Such a cross hatched pattern is generally considered as characteristic for MBE-grown buffer layer systems [103, 110β114]. The trenches in our samples generally exhibit a depth of 5β15ππ in both directions. As shown in figure 5.2(b), displaying two profile sections along the two marked crystallographic directions, we find for the undulations along the [Β―110] direction a periodicity of 1.0 β 1.6ππ, for the [110] direction the undulation period takes values of 0.4 β 1.0ππ. The surface texturing is clearly visualized in the three dimensional plot of the wafer surface in 5.2(c). It has been experimentally shown that locally varying strain fields inside the heterostructure lead to anisotropic growth rates along the [Β―110] and [110] directions, as well as to In segregation, which causes a modulation of the underlying band structure. Such processes are likely to contribute to the generation of the surface structuring in buffer systems. Yet, there is no unified explanation for the cross hatching phenomenon, however, it is also widely suggested that the surface modulations can be associated with the formation of an orthogonal network of the misfit dislocations along the [Β―110] and [110] crystallographic directions during the epitaxial growth process. Thus, the development of a cross-hatched surface texture is generally regarded as an indicator of a well-functioning buffer system [103, 106, 111, 113, 115].
In a next step, building on the obtained strain-relaxed and defect-free In0.75Al0.25As
VS, we growth-engineer the active In0.75Ga0.25As/In0.75Al0.25As layer design under the