This paper discusses recent progress in thin film epitaxy across the misfit scale through the paradigm of domain matching epitaxy (DME). This epitaxy across the misfit scale is critical for integrating multifunctionality on a chip and creating smart structures for next-generation solid-state devices. There are three sources of strains that are cumulative at the growth temperature, and the relaxation process starts during the growth process. Upon cooling, unrelaxed lattice, thermal and defect strains give rise to net residual strains. In large misfit (ε⩾10%) systems, where lattice misfit strain is predominant, it can be relaxed completely, and then only thermal and defect strains remain upon cooling. In low misfit systems, all three sources contribute to the residual strain upon cooling, as result of incomplete lattice relaxation. The predominant strain relaxation mechanism in thin films is by nucleation of dislocations at the free surface, as the nucleation energy in the bulk is considerably higher. At the free surface, the activation barrier for dislocation nucleation is considerably lower at the steps. Since the step formation energy is lower under a compressive stress compared with tensile stress, it reduces nucleation energy under compressive stress and lowers the critical thickness compared with tensile stresses in thin films. Once the dislocation nucleates, it propagates or glides to the interface to relieve the strain. However, if lattice frictional stress in the film is high, most dislocations may not reach the interface, depending upon the growth temperature and rate. Thus, these two key steps, dislocation nucleation and propagation, play a critical role in the thin film relaxation process. Once the dislocations reach the interface, the atomic structure of the dislocation at the heterointerfaces determines its electronic properties, specifically trapping and recombination characteristics. It is found that the atomic structure of the dislocation is determined by the interplay between strain and chemical free energies. Thus, the dislocations (representing missing or extra planes) play a critical role in the relaxation of thin film heterostructures. This paper focuses on epitaxy across the misfit scale, based upon matching of integral multiples of lattice planes. If the misfit falls between the integral multiples, it is accommodated by the principle of domain variation, where domains alternate to accommodate the misfit. Details of epitaxy from low misfit (∼4%) in Ge/Si) to large misfit (∼22%) in TiN/Si are shown. In III-nitride/sapphire and II-oxide/sapphire systems, this paper deals with polar orientations, where misfit is uniform in the basal plane, and non-polar orientations, where misfit varies over an order of magnitude in the film plane. It is shown that the DME paradigm is key to the integration of thin film heterostructures across the misfit scale and other complex systems such as vanadium oxide and PZT systems on Si(100) substrates for the integration of functionalities on a computer chip. Finally, it is shown that the formation of epitaxial and self-assembled nanodots on Si(100) provides a critical advance, with tremendous implications for information and data storage and related nanomagnetics applications.
Read full abstract