|
Giovanna Agrosì, Università di Bari (Italy)
Giancarlo Capitani, Università di Bari (Italy)
Eugenio Scandale, Università di Bari (Italy)
Gioacchino Tempesta, Università di Bari (Italy)
|
|
A Multi-analytical study of a complex syntactic coalescence of polytypes in a PVT-grown 6H-SiC wafer was first performed by means of X-Ray Diffraction Topography - using both White-Beam-Synchrotron-Radiation-Source and MoKα1 conventional source - and μ-Raman Spectroscopy. Three lamellae of different polytypes were localized by XRDT: one lamella L1 and two micro-lamellae L2 and L3 . L3 was enclosed in L1. The diffraction contrast analyses and the μ-Raman spectra indicated that the L1 and L2 lamellae were iso-oriented and both consisted of the15R polytype. Unfortunately, the polytype of the L3 lamella could not be univocally determined. In order to establish the actual polytype of such lamella, a small fragment embedding L3 lamella was extracted from the wafer and prepared for Transmission Electron Microscopy (TEM) observation. Selected area diffraction patterns (SADPs) carried out on the lamella were consistent with the 15R polytype. Nevertheless, SADPs taken exactly above the interface showed a splitting typical of twinning. Therefore, it could be concluded that the L3 lamella is twin-related to the host 15R L1-lamella. In order to determine the twin law, comparisons among SAPDs and simulations of the diffraction pattern introducing all the possible twin operations were carried out. Moreover, taking into account also the XRDT and μ-Raman data previously obtained, the twin law was determined as -rotation around [001] with (001) as compositional plane.
High Resolution images were acquired in order to establish the structure of the twin boundaries. The interpretation of the contrast was not trivial because of the presence of numerous stacking faults parallel to (001). HRTEM images taken across one of such boundaries showed the following stacking sequence: (32) zig-zag pattern which switches to (23) zig-zag pattern trough a (33) lamella. HRTEM images taken across the opposite twin boundary showed a stacking sequences that consisted of (23) pattern which passed to (32) by means of a number of interleaving (33) plus (22) lamellae and one (33) isolated lamella. The Fourier Transform calculated across the boundary (33) plus (22) did not show any splitting, allowing for the conclusion that this sequence introduced no twinning within the 15R, but only 4H and 6H stacking faults. Consequently it could be concluded that the real twin interface consists of the (33), 6H like, lamella. Finally, it is worth mention the importance of the multi-analytical approach we used. The TEM is a very powerful technique to investigate very fine microstructural details, but also an extreme sampling tool. In this respect, XRDT is very useful technique to assess representativity of the results and for localizing portions of the sample to be more deeply investigated.
|
