Nitride materials for use in downscaled interconnect technology

Abstract

Copper interconnects will struggle to keep up with the advances in transistor miniaturisation. This creates a significant limiting factor in trying to keep Moore’s Law on track into the 2020’s. At smaller dimensions, Cu prefers to form non-conducting Cu islands. To prevent this, a liner material that promotes 2D conducting Cu needs to be deposited in the interconnect via in addition to the diffusion barrier that prevents the migration of Cu into the surrounding dielectric. However, the lowest, and smallest, levels of interconnects have extremely high aspect ratios and the bilayer of diffusion barrier and liner material takes up too much volume to allow for the deposition of a sufficient amount of Cu. A change to an alternative metal could solve some of these issues, but this brings its own challenges. Additionally, Cu is expected to remain in use at global interconnect level. For these reasons it is essential to optimise the performance of Cu interconnects and extend the use of Cu as the interconnect metal for as long as possible. One approach to this is to replace the separate diffusion barrier and liner bilayer for a single material that exhibits both properties and can be deposited in an ultra-thin layer, e.g. using atomic layer deposition. In this thesis I explore, using first principles simulations, two different approaches to developing such a combined barrier+liner material. One is to study how known liner and barrier materials could be combined through doping and passivation to create a new material that delivers both properties. Our model system is TaN (diffusion barrier) modified with Ru, Co and W. Secondly, an intrinsically ultra-thin 2D material that has both barrier and liner properties could be used to replace the barrier/liner bilayer. MoS2 has diffusion barrier properties but its liner properties have not been studied and is thus used as a model system for this approach. The literature on studying barrier and liner materials using theoretical methods is quite limited. Therefore, this thesis also presents an approach for predicting thin film morphology using density functional theory (DFT). Applying this method to Cu adsorption on TaN modified with Ru, Co and W as well as monolayers of MoS2 showed that the most important factor in controlling thin film morphology is the competition between metal-metal and metal-substrate interaction. Results show that 50% doping of TaN with Ru and 25% doping with Co or W in the surface layer can promote liner properties in TaN and thus promote the growth of 2D, conducting Cu while also preventing Cu diffusion through the material. The study of Cu as well as Ru and Co, which are of interest as alternative interconnect metals, on MoS2 showed that the metal-substrate interaction of Cu and Ru on MoS2 must be improved in order to promote a 2D film, while Co shows very strong metal-substrate interactions and should thus form a 2D film on MoS2.