Development of a ReaxFF potential for Ag/Zn/O and application to Ag deposition on ZnO
Evidence and attribution¶
Authority of statements
Prose sections below (Summary, Methods, Findings, etc.) are curated summaries of the publication identified by doi, title, and pdf_path in the front matter above. They are not new primary claims by this wiki.
For definitive numerical values, reaction schemes, and interpretations, use the peer-reviewed article (and optional records under normalized/papers/ when present)—not this page alone.
Summary¶
Lloyd et al. derive Ag–Zn–O interactions for ReaxFF by extending an established ZnO parameter set, fitting new terms to density functional theory (DFT) data from SIESTA for bulk references (elemental silver, Ag–Zn alloy, silver oxides) and for Ag on ZnO surface configurations, including equations of state, binding energies, and works of separation. The reported fits reproduce the DFT bulk benchmarks and track Ag–ZnO surface energetics with useful accuracy for the training scope. The work is motivated by low-emissivity (Low-E) glazing, where Ag films are sputtered onto ZnO seed layers but the Ag/ZnO junction is mechanically weak (large lattice mismatch, order 10%). The parametrized field is exercised in reactive MD using single-atom Ag deposition onto wurtzite ZnO, comparing O-terminated polar (0001) and nonpolar (1010) orientations for impact energies from 0.1 eV to 30 eV, bracketing magnetron sputtering-like conditions. Over that campaign, adsorption is strongest when deposition energies stay at or below ~10 eV, whereas higher energies favor reflection or subsurface behavior (per the article’s summarized trajectories).
Methods¶
Force-field training¶
Parent field / elements: Starting from the established ZnO ReaxFF set, new terms describe Ag–Zn–O interactions while retaining the reduced energy expression used for ZnO (bond, van der Waals, shielded Coulomb with EEM charges, valence, lone-pair, over/undercoordination terms as in the article’s Eq. (4) narrative). QM reference: SIESTA DFT on bulk crystals (Ag, Ag–Zn alloy, silver oxides) and Ag-on-ZnO surface configurations—equations of state, binding energies, and works of separation among the targets described in the abstract and Sec. 2. Training set: expanded and distorted lattice configurations plus surface binding motifs referenced to supplementary material in the paper. Optimization: ReaxFF parameters adjusted to reproduce those DFT benchmarks (see article and SI for weights/objectives).
MD application (atomistic dynamics)¶
Reactive MD with the fitted ReaxFF field uses the velocity Verlet integrator with Δt = 1 fs on wurtzite ZnO slabs, comparing O-terminated polar (0001) and nonpolar (1010) surfaces for single-atom Ag deposition at 0.1–30 eV incident energy (motivated by magnetron sputtering in the introduction). Boundaries / cells: periodic replication in x and y with a fixed bottom ZnO layer; supercells are about 22.80 × 26.33 × 30 Å (512 atoms) for (0001) and 26.51 × 26.33 × 30 Å (640 atoms) for (1010) (Surface Science Sec. 2.4). A Berendsen thermostat acts on the second and third double ZnO layers beneath the impact region. Barostat / NPT: N/A — constant-volume slab impacts. Electric field / enhanced sampling: N/A. The article’s deposition subsection does not name a specific MD package; coupling to standard ReaxFF workflows is implicit. Thermal conditions: the Berendsen coupling is applied to specified ZnO layers, but a single numeric target temperature (K) for the thermostated region is not stated in the Sec. 2.4 excerpt checked here (N/A beyond “temperature control” wording in Surface Science). Equilibration / production durations (ps–ns): N/A in that same excerpt beyond the 1 fs timestep—the paper moves immediately to Results without quoting total trajectory lengths in the subsection summarized.
Findings¶
The fitted ReaxFF reproduces DFT equations of state for Ag, Ag–Zn, and silver oxide training crystals and tracks Ag–ZnO binding and work of separation trends versus DFT in the authors’ tables. Deposition MD indicates that maximum Ag adsorption occurs for deposition energies at or below about 10 eV, whereas higher energies favor reflection or subsurface behavior—consistent with the sputtering-relevant energy window emphasized in the abstract. Sensitivity: outcomes depend on facet (polar (0001) vs nonpolar (1010)) and on incident energy along the 0.1–30 eV grid. Comparisons: the introduction connects the interface science problem to Low-E glazing, where Ag on ZnO seed layers suffers weak adhesion and roughly 11% lattice mismatch relative to experiment-oriented motivation. Limitations: high-energy impacts access non-equilibrium chemistry; translating trajectories to industrial sputtering requires validation against experiment for real terminations, defects, and plasma composition.
Limitations¶
High-energy impacts access non-equilibrium chemistry; validate sputtering-like outcomes against experiment for each facet termination and defect population.
Relevance to group¶
A.C.T. van Duin and RxFF Consulting contributors appear on the author list, reflecting industry-facing ReaxFF parameterization for oxide/metal interfaces.