Interfaces play an essential role in the performance of ever-shrinking semiconductor devices, making comprehensive determination of their three-dimensional (3D) structural properties increasingly important. This becomes even more relevant in compositional interfaces, as is the case for Ge/GeSi heterostructures, where chemical intermixing is pronounced in addition to their morphology. We use the electron tomography method to reconstruct buried interfaces and layers of asymmetric coupled Ge/Ge0.8Si0.2 multiquantum wells, which are considered a potential building block in THz quantum cascade lasers. The three-dimensional reconstruction is based on a series of high-angle annular dark-field scanning transmission electron microscopy images. It allows chemically sensitive investigation of a relatively large interfacial area of about (80 × 80) nm2 with subnanometer resolution as well as the analysis of several interfaces within the multiquantum well stack. Representing the interfaces as iso-concentration surfaces in the tomogram and converting them to topographic height maps allows the determination of their morphological roughness as well as layer thicknesses, reflecting low variations in either case. Simulation of the reconstructed tomogram intensities using a sigmoidal function provides in-plane-resolved maps of the chemical interface widths showing a relatively large spatial variation. The more detailed analysis of the intermixed region using thin slices from the reconstruction and additional iso-concentration surfaces provides an accurate picture of the chemical disorder of the alloy at the interface. Our comprehensive three-dimensional image of Ge/Ge0.8Si0.2 interfaces reveals that in the case of morphologically very smooth interfaces─depending on the scale considered─the interface alloy disorder itself determines the overall characteristics, a result that is fundamental for highly miscible material systems.
Keywords: Ge/GeSi heterostructures; electron tomography; interface alloy disorder; interface roughness; layer thickness fluctuations.