During 4H-SiC homoepitaxy and post-growth annealing, the process of stress relaxation has previously been reported to lead to the formation of misfit dislocations or interfacial dislocations (IDs). [1] The driving force for such a relaxation process is believed to be the lattice misfit induced by the difference in nitrogen doping concentrations across the substrate/epilayer interface. According to van der Merwe [2], at initial stage of epitaxy growth, misfit is accommodated by elastic strain in epilayers; as the growth continues, misfit strain builds up until a critical amount when crystals start to deform plastically with misfit dislocations introduced at the substrate/epilayer interface, relaxing additional elastic strain and minimizing the free energy of the epi-substrate system. Matthews and Blakeslee, [3] from a mechanics point of view, predict the onset of stress relaxation process to be the moment when the mismatch stress applied on a single threading glide dislocation in the epilayer (basal plane dislocation in the case of 4H-SiC homoepitaxy) exceeds the sum of the line tension and the Peierls-Nabarro stress as the epilayer grows thicker.
Nevertheless, the direct observation of the stress relaxation process in silicon carbide crystals has not yet been reported. In this paper, we present an in-situ study in which a sample cut from a 150mm commercial n-/n+ 4H-SiC homoepitaxial wafer was subject to high-temperature heat treatment while sequential synchrotron white beam X-ray topographs were recorded simultaneously. The heat treatment was carried out in a halogen lamp furnace and the local temperature was elevated to as high as 1600°C. Figure 1 shows a series of topographs recorded in this manner and the time interval between consecutive topographs is 10 seconds. As time advances, a basal plane dislocation (BPD) in the epitaxial layer is observed to glide from left to right, depositing a straight segment of misfit dislocation right at
the substrate/epilayer interface. Figure 2 schematically shows the entire process. Such dynamic observations of relaxation process suggest that the lattice misfit once again exceeds the critical value during high temperature heat treatment. Since the epilayer thickness does not change, it can be inferred that the misfit strain simply increases as temperature is raised. The reason may be that the thermal expansion coefficient of 4H-SiC crystal is a function of doping concentration. Sasaki et. al [4] also confirmed this by measuring the lattice constants of both n+ substrate and n- epilayer at elevated temperatures.
[1] M. Dudley, H. Wang, J. Guo, Y. Yang, B. Raghothamachar, J. Zhang, B. Thomas, G.Y. Chung, E. Sanchez, D.M. Hansen, S.G. Mueller, MRS Advances, 1(2), 91-102 (2016).
[2] J.H. van der Merwe, J. Appl. Phys. 34, 117 (1963).
[3] J.W. Matthews, and A.E. Blakeslee, Journal of Crystal Growth, 27, 118–125 (1974).
[4] S. Sasaki, J. Suda, and T. Kimoto, Materials Science Forum, 717-720, 481-484 (2012)