Evaluation of Transition Metal Dichalcogenide Encapsulation to Improve Copper Interconnects: An ab Initio Study

Benjamin A Helfrecht, Purdue University

Abstract

Two-dimensional materials, such as graphene, hexagonal boron nitride, and transition metal dichalcogenides (TMDs) have gained much interest for their applications in nanoscale devices, especially nanoelectronics. In this regard, TMDs are particularly interesting, as this class of materials contains several stable compounds with a wide variety of properties, existing as semiconductors with tunable band gaps, metals, superconductors, and/or ferromagnets. However, in order to incorporate TMDs into devices, a fundamental understanding of the interfaces TMDs form with other materials, including metals, is required. This is especially true for Group IV and V dichalcogenides, which have received less attention than the Group VI dichalcogenides to date. The goal of this work is to identify trends in properties most relevant to TMDencapsulated copper nanowires, which are envisioned as a candidate for ultra-scaled interconnect technology. In the spirit of this goal, the stability, adhesion, and electronic properties of interfaces between Cu(111) surfaces and TMDs of the form MX2, where M is a transition metal and X a chalcogen, are examined with ab initio computational methods. Select Group IV and V TMDs that are metals or small bandgap semiconductors (M = Ti, V, Nb) are studied alongside the Group VI semiconductors (M = Mo, W); all three chalcogenides (X = S, Se, Te) are studied for each transition metal. In addition, the ability of these TMDs to serve as diffusion barriers to copper and oxygen is also evaluated. Ground state TMDs forming an interface with a Cu(111) surface are predicted to be stable, with binding energies per unit area three to five times higher than hexagonal boron nitride or graphene. The interactions between the Cu and TMD are dominated by van der Waals forces, but also have some covalent or ionic character. As a TMD becomes adsorbed on the Cu surface, charge rearrangement occurs at the interface resulting in a net dipole and a modulation of the work function at the Cu surface that depends strongly on the chemistry of the TMD: increasing by up to ≈ 1 eV by encapsulating with a Group IV or V TMD and decreasing by up to ≈ 1 eV by encapsulating with a Group VI TMD. However, the Schottky barrier at the interface does not depend strongly on chemistry. All semiconductor TMDs studied form a Schottky barrier of roughly 0.5 eV with the Cu surface. No Schottky barrier is formed for the metallic TMDs. Many Group IV, V, and VI disulfides are also predicted to serve as adequate Cu and O diffusion barriers, with energy barriers of 2.5–5.5 eV. However, all TMDs studied are predicted to perform more poorly as diffusion barriers than h-BN and graphene, which exhibit energy barriers to Cu and O diffusion of 4–11 eV.

Degree

M.S.M.S.E.

Advisors

Strachan, Purdue University.

Subject Area

Engineering|Materials science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS