Epitaxial growth of high-κ oxides on silicon ☆ C. Merckling a,b , G. Saint-Girons a , G. Delhaye a , G. Patriarche c , L. Largeau c , V. Favre-Nicollin d , M. El-Kazzi a , P. Regreny a , B. Vilquin a , O. Marty a , C. Botella a , M. Gendry a , G. Grenet a , Y. Robach a , G. Hollinger a, ⁎ a Université de Lyon, INL, UMR/CNRS 5270, Site Ecole Centrale de Lyon, 36 avenue Guy de Collongue, 69134 Ecully cedex, France b ST Microelectronics, 850 rue Jean Monnet, 38926 Crolles, France c LPN/UPR20 CNRS, Route de Nozay, 91460 Marcoussis, France d CEA/DRFMC/SP2M/NS, 17 rue des Martyrs, 38054 Grenoble Cedex 9, France abstract article info Available online 27 August 2008 Keywords: Molecular beam epitaxy High-κ oxides Integration X-ray diffraction Transmission electron microscopy The growth of epitaxial high-κ oxides on silicon has recently become a field of intense researches, due to the potential applications of these heterostructures in Complementary-Metal-Oxide-Semiconductor (CMOS) systems, but also for the integration of functional oxides or III–V semiconductors on Si. Some of the key advances in epitaxy of SrTiO 3 , γ-Al 2 O 3 and Gd 2 O 3 epitaxial films on Si(001) and Si(111) substrates are reviewed. It is shown that MBE affords unique advantages in interface engineering by controlling growth parameters. The conformation of the O sublattice of the oxide is shown to play a predominant role on the crystallographic orientations of the oxide layers with respect to the substrate. © 2008 Elsevier B.V. All rights reserved. 1. Introduction The demonstration a few years ago that high quality epitaxial oxides, such as SrTiO 3 (STO) , can be grown on silicon by Molecular Beam Epitaxy (MBE) was considered as a materials breakthrough and opened the door to a wide variety of new potential applications. The first and most demanding application was replacing SiO 2 by alternative oxides in Metal-Oxide-Semiconductor Field-Effect Transistors (MOS- FETs). In future scaling trends predicted by the International Tech- nology Roadmap for Semiconductors (ITRS), SiO 2 is continually being made thinner and thinner approaching its fundamental limit and direct current tunnelling can only be limited by using oxides with higher dielectric constant (κ). Up to now, most industrial research has focused on hafnium-based amorphous oxides, however epitaxial oxi- des could lead to more abrupt oxide-Si interfaces and consequently could offer solutions for the end of the roadmap. The fundamental material requirements for high-κ gate dielectrics are very challenging since, in addition to high dielectric constant and band offsets, low interfacial state density and low leakage, the oxides have to be stable on Si at temperatures exceeding 800 °C. The availability of epitaxial oxide-on-silicon templates is now opening the door to the integration of functional oxides (ferroelectric, ferromagnetic, piezoelectric…) which are generally epitaxially grown on oxide substrates. This could help the integration of new functionalities on silicon platforms (e.g. non-volatile ferroelectric memories) and the development of new classes of micro and nanosystems. Finally, epitaxial oxides are also a route to monolithic integration of heterostructures and devices com- bining epitaxial oxides and semiconductors. The integration of new materials (Ge, III–V, oxides,…) with silicon could offer opportunities to produce high speed microprocessor systems (Ge or III–V channels), III–V optical communication systems and high frequency systems, at low cost. In this context, we give here an overview of the molecular beam epitaxy and of the structural properties of SrTiO 3 /Si(001), γ-Al 2 O 3 /Si (111), γ-Al 2 O 3 /Si(001) and Gd 2 O 3 /Si(111) and Gd 2 O 3 /Si(001), as well as a discussion on the growth mechanisms of these oxides on Si. The structural quality of the oxide layers is evaluated on the basis of various characterization techniques. We discuss the influence of the substrate orientation and of the conformation of the O sublattice of the oxide layers on their crystallographic orientation. 2. SrTiO 3 /Si(001) The growth procedure of SrTiO 3 /Si(001) was first demonstrated by McKee et al. [1]. It was also rapidly shown theoretically that perfect coherent SrTiO 3 -Si(001) interfaces could exist [2]. Later, several groups at Motorola [3], IBM Zürich [4], and in various other parts in the world [5–8] have also studied the growth of STO on Si. It was shown that this system presents some intrinsic limitations (e.g. poor thermal stability and low conduction band offsets) for CMOS appli- cation. It is however the prototypical system for integration of oxide electronics on silicon since most functional oxide heterostructures are grown on SrTiO 3 (001) substrates. The easiest way to grow SrTiO 3 is to use a re-crystallization ap- proach. An initial SiO 2 free silicon surface is obtained through a Thin Solid Films 517 (2008) 197–200 ☆ This research was supported by the European Commission's Information Society Technology Program, under PULLNANO project (contract No. IST-026828). ⁎ Corresponding author. E-mail address: Guy.Hollinger@ec-lyon.fr (G. Hollinger). 0040-6090/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2008.08.165 Contents lists available at ScienceDirect Thin Solid Films journal homepage: www.elsevier.com/locate/tsf