ChemRxiv | 2021
Prediction of Co and Ru Nanocluster Morphology on 2D MoS2 from Interaction Energies
Abstract
Layered materials, such as MoS2, have a wide range of potential applications due to the properties of a single layer, which often differ from the bulk material. They are of particular interest as ultrathin diffusion barriers in semiconductor device interconnects and as supports for low-dimensional metal catalysts. Understanding the interaction between metals and the MoS2 monolayer is of great importance when selecting systems for specific applications. In previous studies the focus has been largely on the strength of the interaction between a single atom or a nanoparticle of a range of metals, which has created a significant knowledge gap in understanding thin film nucleation on 2D materials. In this paper, we present a density functional theory (DFT) study of the adsorption of small Co and Ru structures, with up to four atoms, on a monolayer of MoS2. We explore how the metal–substrate and metal–metal interactions contribute to the stability of metal clusters on MoS2, and how these interactions change in the presence of a sulfur vacancy, to develop insight to allow for a prediction of thin film morphology. The strength of interaction between the metals and MoS2 is in the order Co > Ru. The competition between metal–substrate and metal–metal interaction allows us to conclude that 2D structures should be preferred for Co on MoS2, while Ru prefers 3D structures on MoS2. However, the presence of a sulfur vacancy decreases the metal–metal interaction, indicating that with controlled surface modification 2D Ru structures could be achieved. Based on this understanding, we propose Co on MoS2 as a suitable candidate for advanced interconnects, while Ru on MoS2 is more suited to catalysis applications. Introduction Layered materials that can be exfoliated into 2D sheets continue to generate significant interest across various disciplines, including batteries [1,2], catalysis [3,4], electronics [5-10], photonics [11,12], and sensors [13-16]. This is due in part to the interesting properties of these 2D materials, which often differ from their bulk equivalent, as well as the flexibility in fabricaBeilstein J. Nanotechnol. 2021, 12, 704–724. 705 tion afforded by an ultrathin material [17]. The majority of applications are built on an interaction between a metal and the 2D material. There are multiple studies in this regard that involve the adsorption of or doping with transition metals [4,5,18-22], alkali and alkali earth metals [23-25], and nonmetals [25] on MoS2 and other 2D materials. While experimental studies can be used to probe the performance of the 2D material in a device or some of the interfacial interactions between metal and 2D materials [9,10,21,26], first principles modelling is a powerful tool that permits the investigation of the detailed interactions of metals and 2D materials at the atomic scale. In particular, understanding the nucleation of metals on 2D materials will be valuable for the design of catalysts or for preventing islanding of conductive metals. Typically, theoretical studies focus on the adsorption of either single atoms of a series of metals [21,23-25,27] or large nanoparticle-like structures [19,20]. In our previous study we identified that while these studies do deliver useful insights, there is a knowledge gap in the understanding of metal thin film nucleation on 2D materials [28]. We showed that we can investigate the first stages of thin film nucleation on 2D materials with first principles simulations, using the example of small Cun structures on an MoS2 monolayer (ML). MoS2 is a naturally occurring transition metal dichalcogenide (TMD) and one of the most frequently studied 2D materials. Unlike graphene, MoS2 is a semiconductor, which gives it an increased number of possible applications [11,29]. Our previous first principles study [28] of the interaction of Cu species on MoS2 showed how Cu can take different structures depending on the number of Cu atoms and whether the TMD is stoichiometric or defective. In the present study, we will expand the knowledge gained from our previous work on Cu on MoS2 and apply it to the adsorption of small Con and Run clusters on an MoS2 ML, where n = 1–4. Co and Ru are of great interest in conjunction with MoS2 for application in advanced interconnects as alternatives to Cu [30-35] and TaN. Applications in catalysis include Pt-free hydrogen evolution catalysts [36-41]. Interconnects require high-quality, conformal thin films with low resistivity, to avoid many of the typical failure mechanisms such as electromigration [42,43]. This means that 3D migration of atoms (agglomeration) should be inhibited, while 2D growth (wetting) should be promoted. In contrast, in catalysis applications the ratio of surface to bulk is of great importance in promoting catalytic activity. Therefore, 3D growth (agglomeration) is essential when creating a supported metal catalyst [4447]. In this work we aim to determine the atomic-scale interactions that control the stability of small Con and Run clusters (n = 1–4) on a single ML of MoS2. Based on this understanding, alongside the magnitude of metal–substrate and metal–metal interactions we will be able to predict the morphology of Co and Ru thin films on 2D MoS2. We have previously studied 2D and 3D Cu clusters on TaN, where we determined that there are two useful descriptors for 2D-vs-3D growth [48]: (1) If the metal–substrate interaction is more favourable than the metal–metal interaction, then 2D growth is preferred; and (2) if the total binding energy is more favourable than the cohesive energy of the bulk metal, then 2D growth is preferred. Predictions made using these descriptors can be used when deciding which metal–substrate combination will be suitable for a particular application where the shape of the metal is vital. Methods All calculations for this study were carried out with density functional theory (DFT) using the Vienna Ab initio Simulation Package (VASP) version 5.4 [49]. Three-dimensional boundary conditions were applied and the spin-polarized general gradient approximation (GGA) along with the Perdew–Burke–Ernzerhof (PBE) approximation to the exchange–correlation functional were used to describe the system [50]. Valence electrons were described explicitly using a plane-wave basis set with an energy cutoff of 450 eV. The valence electron configurations are as follows: Co = 4s2 3d7, Ru = 5s1 4d7, Mo = 5s1 4d5, and S = 3s2 3p4. The core electrons were treated with the projectoraugmented wave potential (PAW) [51]. A Monkhorst–Pack k-point grid of 2 2 1 was used. All forces acting on the atoms were converged to within 0.02 eV/Å. A Methfessel–Paxton smearing of order 1 was used and no symmetry was applied. The description of pristine and defective MoS2 monolayers (ML) was published in our previous work [28]. Bulk MoS2 is made up of two layers. To create the pristine ML one of these was removed, which also creates the vacuum necessary to avoid interaction along the z-axis; the vacuum region is 8 Å. A (5 × 5) super cell was used. No van der Waals (vdW) corrections were applied, as both the literature and our own tests (Supporting Information File 1, section S4) show that vdW forces do not dominate in these types of structures. The defective ML has the same structure as the pristine ML, except that a single S atom has been removed to create a vacancy and the ions are relaxed with no symmetry constraints. Using H2S as a reference, we have computed an exothermic vacancy formation energy of −6.16 eV. The bond lengths in bulk structures that are used for comparison are based on the crystal structures in [52-57]. Only a theoretical crystal structure was available for RuMo, all other structures used have been determined experimentally. Beilstein J. Nanotechnol. 2021, 12, 704–724. 706 To understand the binding of Co and Ru to the MoS2 monolayer, four different energies are computed: 1. Binding energy per metal atom: