**5. Conclusions**

High-permittivity binary oxides for silicon carbide (SiC) and gallium nitride (GaN) electronic devices have attracted significant interest in the last decade because of the potential benefit they can bring in the device performances. In particular, special attention has been placed on the most suitable deposition techniques for their synthesis and on their implementation in real device fabrication, in which all the processes must be compatible with industrial environments and scalable to large areas. Surely the most widely investigated binary oxide is Al2O3, as well as its combination with HfO<sup>2</sup> and other materials. In fact, Al2O<sup>3</sup> provides a good compromise among all the basic physical properties to be fulfilled by the gate dielectric for wide band gap semiconductors, namely, a dielectric constant close to that of the semiconductor, a large band gap, an appropriate band offset, a high critical electrical field, and good thermal stability. On the other hand, HfO<sup>2</sup> and other oxides possess higher dielectric constants than Al2O3, but their band alignments and crystallization temperatures represent a concern in application. The most affirmed method for their synthesis has been demonstrated to be the ALD approach, which can be considered the deposition technique of choice for the fabrication of very thin films with high uniformity and conformal growth on large areas. All these capabilities render ALD as very appealing for industrial implementation. In this context, beyond the fundamental study on the impact of the deposition parameters on the films' properties, the pre- and post-deposition conditions are relevant features for the development of a reliable high-κ technology for SiC and GaN. Cleaning treatments before high-κ thin film deposition, e.g., based on wet chemical solutions are the most suitable approach for both SiC and GaN substrates in order to limit the creation of interface defects. In spite of the "gentle" nature of the wet cleaning, interface states, as well as fixed charges within the binary oxides, still represent a great concern in practical applications. Hence, post-deposition and post-metallization annealing treatments need to be optimized in order to achieve the desired device performance. A common problem in SiC technology is the formation of an uncontrolled SiOx layer at the interface as well as residual carbon. Hence, the intentional Al2O3/SiO<sup>2</sup> combination has been proposed as a possible solution, although the presence of the SiO<sup>2</sup> interfacial layer partially reduces the advantage offered by the high-κ Al2O3. For that reason, the search for other material combinations and/or post-deposition treatments limiting the interfacial interaction has become mandatory.

In regard to GaN-based devices, the implementation of Al2O<sup>3</sup> thin films is also the most investigated and promising solution. The interaction at the interface is limited to a partial oxidation of the substrate, which in turn might be source of electrically active defects when oxynitride bonds are present. In this case, the epitaxial growth of crystalline oxides has also been widely explored as a possible route to gate insulation in GaN-based devices, considering other oxides, such as lanthanide oxides (Gd2O3, Sc2O3, and La2O3) or NiO and CeO2. However, the main limitations of the epitaxial oxides' implementation are the number of structural defects occurring after the initial layers and the presence of preferential leakage current paths at the grain boundaries.

In terms of practical device application, high-κ binary oxides have already been implemented in both 4H-SiC MOSFETs and GaN-based MISHEMTs, with Al2O<sup>3</sup> being the most widely used system. In this case, while promising results in terms of channel mobility and RON have been reported, charge-trapping effects occurring in these oxides remain a limiting factor that has to be addressed by appropriate surface preparation techniques and post-annealing conditions. In particular, the integration of high-κ oxides as gate insulators in 4H-SiC MOSFETs will require optimization of the process flow, with particular attention to the thermal budget required for ohmic contact formation, which must be compatible with the crystallization temperature of the oxide.

**Author Contributions:** Conceptualization, R.L.N. and F.R.; methodology, P.F., G.G. and E.S.; formal analysis, P.F., G.G. and E.S.; investigation, E.S., P.F. and G.G.; resources, E.S.; data curation, E.S.; writing—original draft preparation, R.L.N., P.F. and G.G.; writing—review and editing, R.L.N. and F.R.; visualization, P.F. and G.G.; supervision, R.L.N. and F.R.; project administration, F.R.; funding acquisition, F.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was partially funded by the ECSEL-JU project WInSiC4AP (Wide Band Gap Innovative SiC for Advanced Power)—grant agreement no. 737483 and the national project EleGaNTe (Electronics on GaN-based Technologies)—PON ARS01\_01007. Moreover, the authors would like to acknowledge the European project GaN4AP (GaN for Advanced Power Applications)—grant agreement no. 101007310 for funding part of their current GaN activities.

**Data Availability Statement:** The data that support the findings of this study are available from the corresponding author upon reasonable request.

**Acknowledgments:** The authors would like to acknowledge their colleagues at CNR-IMM: F. Giannazzo and M. Vivona, for the fruitful discussions and contributions in SiC and GaN experiments, and S. Di Franco and C. Bongiorno, for the precious technical support during device fabrication and TEM analyses. M. Saggio and F. Iucolano from STMicroelectronics are greatly acknowledged for their fruitful collaboration on wide band gap semiconductor research activities. The authors also thank Graziella Malandrino of the Department of Chemistry, University of Catania, for the fruitful collaboration in the realization of NiO and CeO<sup>2</sup> gate dielectrics by MOCVD.

**Conflicts of Interest:** The authors declare no conflict of interest.
