Growth Strategies for Oxide Heteroepitaxy on Silicon by ALD
Prof. Brian G. Willis
Chemical Engineering
University of Delaware
12:00 PM
Room 260, Wright-Rieman Chemistry Laboratory
Abstract: For the past several decades, the semiconductor industry has grown by doubling the transistor density on a silicon chip roughly every two years (Moore’s Law). We are now in the era of one billion transistors on a single chip, and there is growing uncertainty about the feasibility and practicality of continuing the miniaturization much beyond the 22 nm node, which is only a few years into the future. As the exponential scaling declines, there is a growing interest to integrate new materials with new functionalities in order to extend innovation in semiconductor technology. One promising avenue is to integrate complex oxides with silicon devices through the heteroepitaxy of crystalline oxides on semiconductors. The integration of crystalline oxides with semiconductors may enable new device structures that harness the useful functional properties of oxide materials. These useful properties include ferroelectricity, pyroelectricity, piezoelectricity, and many others. In addition, the crystalline oxides are of interest for applications as high-k dielectrics for transistor or memory devices. The integration of oxides with semiconductors is a challenging materials engineering problem due to the differences in crystal structure and properties between the covalent bonded semiconductor and the ionic bonded oxide layer. Presently, the successful heteroepitaxy of crystalline oxides on semiconductors has only been achieved using molecular beam epitaxy (MBE). While MBE methods have advantages in terms of the precise control of the growth process, their disadvantages include high capital and operating costs, and growth rates are considerably lower than standard manufacturing practice. A more cost effective method to grow epitaxial oxides would be a significant advance for the practical implementation of these useful materials.
This talk will present research strategies for the use of metal-organic compounds to grow crystalline oxides using chemical vapor deposition or atomic layer deposition. The critical objectives are to control the composition and structure over an interface layer that is less than 1 nanometer thick. Using surface chemistry methods including ultra-high vacuum scanning tunneling microscopy along with Hafnium and Strontium compounds as model systems, we present several strategies for a chemical engineering approach to oxide heteroepitaxy. It will be shown that direct reaction of the common β-diketonate precursors with the semiconductor surface is unlikely to be a successful strategy due to adverse surface reactions of the organic components associated with these compounds. A second approach is to use a water-templated Si(100)-2x1 surface to form an atomically abrupt semiconductor/metal-oxide interface. It is shown that atomically abrupt interfaces are achieved, but forming an ordered two dimensional surface requires a more detailed understanding of the adsorption kinetics of H2O(g). Lastly, it is shown that the most promising approach is to use alkaline earth metal-oxide layers grown by ALD with a catalytic oxide desorption step. It will be demonstrated that the results are comparable to MBE data and that the metal-organic surface reaction approach is a promising alternative to MBE for the growth of crystalline oxides.
|