Physics of Fermi-Level “Unpinning” at Metal/Ge Interfaces; First-Principles View

Abstract

Because of high carrier mobility and small band gap, germanium (Ge) is expected as a next-generation material for electronic and optical devices. One of serious issues for application is the “Fermi-level pinning” at metal/Ge contact [1]; the Fermi energy of metal is located around 0.1eV above the valence-band top of Ge not depending on the kinds of metals, which interrupts desirable metal/n-Ge contacts. Recently, the breakdown of such pinning, i.e., Fermi-level “unpinning”, was observed at several interfaces such as Fe3Si/Ge, Sn/Ge, Al/GeO2/Ge, germanide/Ge, and TiN/Ge. For the former three interfaces, we have already clarified the origins of unpinning [2]. However, the TiN/Ge interface is the most interesting among these interfaces because this interface has the largest unpinning and the pinning/unpinning features depend on the growth conditions of interface [3]. In this work, by the first-principles calculations, we first clarify why the Fermi-level unpinning occurs at TiN/Ge interfaces and depends on the growth conditions, and then discuss what are the common key factors to promote the unpinning at various metal/Ge interfaces. The experiments showed that the unpinning occurs at TiN/Ge(001) interface when TiN is grown on Ge and some interface layers are produced in N-rich and low-temperature conditions [3]. Thus, we consider various N-rich interface layers such as TiN2 and Ge3N4. We adopt the (1x1)-(3x3) superlattice geometries to simulate these interfaces. Atomic/electronic structures and Schottky barriers of various TiN/Ge interfaces are calculated by the standard first-principles method using VASP code in the density-functional theory. The main results are summarized as follows; (1) In the case of plain flat interface, the pinning occurs, being consistent with experiments. This is because interface Ge atoms have many evanescent states (metal-induced gap states; MIGS) that are connected with Ti metallic states in the band gap of Ge. (2) In the case of N-rich TiN/Ge interfaces with transition interface layers, the penetration of MIGS into Ge layers is highly suppressed. This occurs because interface Ge atoms are terminated by N atoms and has no gap states (MIGS). As a result, reflecting interface N-Ge bonds, a definite electron transfer occurs from Ge to TiN sides, which promotes the large Fermi-level unpinning. (3) We found that the unpinning is also realized even when 20% of interface N-Ge bonds are broken at N-rich TiN/Ge interface or when interface Ge atoms are reacted with supplied N atoms to produce stable Ge3N4 interface layers. (4) By comparing various “unpinning” interfaces such as Al/GeO2/Ge and βSn/αSn/Ge, we found that the termination of Ge dangling bonds at the interface by the N-Ge, O-Ge, Sn-Ge, or Si-Ge bonding is the key factor to promote the Fermi-level unpinning. This is because such termination eliminates the appearance of evanescent states (MIGS) in interface Ge layers. These results are discussed, together with reviewing recent experiments and focusing on the fundamental physics at metal/Ge interfaces. <References> [1] T. Nishimura et al., Appl. Phys. Lett. 91 (2007) 123123. [2] T.Nakayama et al., ECS trans. 75 (2016) 643, Jpn. J. Appl. Phys. 55 (2016) 111302. [3] K.Yamamoto et al, Appl. Phys. Lett. 104 (2014) 132109. Figure 1