*3.1. Flat Substrates*

The 3C-SiC grown on silicon has still high defectivity even though strategies for the elimination of a large plethora of defects are achieved. Common three-dimensional defects of 3C-SiC grown epitaxially on silicon, such as protrusions, twinned regions, antiphase domains, and polytypes inclusion, are eliminated or strongly reduced. On the contrary, the elimination of dislocation, stacking faults, and stress in the 3C-SiC layer is far from being solved. The hetero-interface is the principal "source" of such defects. Indeed, lattice mismatch among Si and SiC, as well as the disparity in thermal expansion coefficients, induced the formation of stress in the film and the formation of both SFs and dislocations.

For example, stacking faults in 3C-SiC can come out from Lomer dislocations' dissociations at the hetero-interface: Lomer dislocation (that forms naturally due to the lattice mismatch at the hetero-interface) with the Burgers vector of a/2 - 110 can dissociate in two partial dislocations, with the Burgers vectors a/6h 211] and a/6- 121 where a is the lattice constant. The two stacking faults propagate through the epitaxial layer and can approach the surface. A cross-view image of the Si/SiC hetero-interface is shown in Figure 2. In this figure, SFs are the oblique bright lines. A region near the hetero-interface with a high number of SFs is highlighted and the density of SFs decreases, moving away from the hetero-interface. In the "on-axis" substrate (Figure 2a), the SF can intersect and form another kind of defect such as the "Lomer lock". The formation of this linear extended defect modifies the mechanical properties and more interestingly can avoid the propagation of the SFs in the epilayer. Unlike the case of the "on-axis" substrate on which the SFs can interact with each other in the "off-axis" substrate (Figure 2b), SFs can arrive at the surface because they have the same orientation and are not able to cross each other. Even in this case, a region of high density of SFs is apparent near the hetero-interface.

Another defect that is present only in the "on-axis" image is the anti-phase boundary (APB) or also called "inverted domain boundary" (IDB)) (Figure 2a). This defect is a 2D defect and it is the boundary between two 3C crystals rotated by each other by 180◦ around the [110] axes. It is observed in Figure 2a as a curved bright line. This defect preferentially lies on the (110) or (111) plane and has a particular atomic structure: it is coherent and can couple with SF in a complex way. In the (110) plane, the structure is made up of a bent Si-C bond that generates a square and a semi-octahedral configuration with unaltered bonds, while in the (111) plane it resembles a twin with a Si–Si bond. In the last case, it can be associated with an SF. For more detail, the reader can refer to [39,40]. The propagation of IDBs within the crystal appears to be extremely complex, resulting in "complex IDBs" interacting with SFs. Moreover, we noticed that IDBs can also end and generate SFs. The presence of "disconnection" (which are steps with a Burgers vector associated in the IDB) might cause such behavior.

The different orientations of the substrates ((100), (111), or (110)) produce a different structure of the material on the surface [41] and a difference in the stress after the deposition.
