Elsevier

Surface Science

Volume 645, March 2016, Pages 67-73
Surface Science

Development of a ReaxFF potential for Ag/Zn/O and application to Ag deposition on ZnO

https://doi.org/10.1016/j.susc.2015.11.009Get rights and content

Highlights

  • DFT calculation for the Ag–ZnO system carried out

  • New reactive force field parameters fitted to the DFT results

  • Energetic deposition of Ag on ZnO performed using the model

  • Ag absorbs preferentially to the polar ZnO surface.

Abstract

A new empirical potential has been derived to model an Ag–Zn–O system. Additional parameters have been included into the reactive force field (ReaxFF) parameter set established for ZnO to describe the interaction between Ag and ZnO for use in molecular dynamics (MD) simulations. The reactive force field parameters have been fitted to density functional theory (DFT) calculations performed on both bulk crystal and surface structures. ReaxFF accurately reproduces the equations of state determined for silver, silver zinc alloy and silver oxide crystals via DFT. It also compares well to DFT binding energies and works of separation for Ag on a ZnO surface. The potential was then used to model single point Ag deposition on polar (0001¯) and non-polar (101¯0) orientations of a ZnO wurtzite substrate, at different energies. Simulation results then predict that maximum Ag adsorption on a ZnO surface requires deposition energies of ≤ 10 eV.

Introduction

Silver has many scientific applications due to its high electrical and thermal conductivity [1] as well as its unique properties as an optical reflector [2], [3]. These properties make Ag perfect for coatings on solar cells and low-emissivity (Low-E) windows amongst other applications [4], [5]. Low-E coatings are used to reduce heat loss through windows via the reflection of infra red radiation. The reflective part of the coatings consist of a silver thin film applied to a seed layer. Zinc oxide is often chosen as the material for the seed layer due to its relatively low cost and semiconductor properties (having a wide band gap of 3.37 eV [6]) along with a reasonable propensity for Ag growth. There are many thin film layers applied to create a Low-E window coating—often including a tough, scratch resistant layer and an antireflective layer. However, the Ag/ZnO interface is known to be one of the weakest interfaces used within the reflective coating due to its low adhesive energy and large lattice mismatch (≈ 11%) [7], [8]. An investigation of the adhesive properties of this interface and Ag growth could directly benefit the Low-E window industry. Any improvement in the production of this coating could reduce the amount of Ag used (reducing costs) and increase durability.

To model surface interactions and deposition events at an atomic scale, molecular dynamics (MD) [9] is often a method of choice. In MD simulations, all interatomic interactions are modelled by a potential function. The ReaxFF potential for zinc oxide [10] has proved to be useful in identifying growth mechanisms for ZnO by Blackwell et al. [11]. However, an existing Morse type potential for the Ag–ZnO interaction [12], fitted to reproduce works of separation, does not accurately predict binding energies.

The purpose of this work is to provide a model for Ag–ZnO interactions that encapsulates important surface characteristics. This model should provide binding energies for Ag on a ZnO surface similar to those found via DFT calculations whilst being computationally inexpensive (compared to DFT). The ReaxFF potential for Ag–ZnO interfaces has been produced and used to model deposition of Ag on polar (0001¯) and non-polar (101¯0) ZnO over an energy range from 0.1 to 30 eV, such as might be the case in magnetron sputtering [13], [14].

Previous modelling of ZnO polar surfaces suggest that O vacancies [15] could be present as a means of stabilising the polar 0001¯ face. However, small scale structures used for fitting this model agree with experimental results indicating that the O-terminated polar face closely resembles the bulk termination [16] and that no defects other than 101¯0 steps on the surface appear [17]. Growth has previously been studied on both polar (0001¯)-O terminated and (0001)-Zn terminated surfaces experimentally [18] as polar surfaces have been shown to be very stable [19]. This model has been fitted to the O-terminated (0001¯) polar face, (101¯0) non-polar face and several simple crystal structures, providing a range of structures which can be simulated accurately, because these surfaces were of more interest to our experimental collaborators.

Section snippets

The ReaxFF Reactive Force Field

The Reactive Force Field (ReaxFF) potential, developed by van Duin et al. [20], depends on the bond order of atoms in a system. This potential has proven to be highly transferable, with an ability to describe both covalent, ceramic and metallic materials and their interfaces. The total energy of the system is expressed as a sum of bond order dependent and non-bonded energy terms (Eq. (1)).Esystem=Ebond+EvdWaals+ECoulomb+Eval+Elp+Etors+Epen+Econj+Eover+Eunder.

Ebond represents the bond energy

Results and discussion

The objective for this fitted parameter set is to reproduce equations of state (EoS) for a number of crystal structures and give binding energies for different surface structures with reasonable accuracy.

Application: single point depositions

As a precursor to considering how Ag grows on the ZnO surface, the interaction of Ag atoms and Ag2 dimers with a ZnO surface was investigated. A study into how Ag deposits onto a ZnO substrate at different depositions energies (ranging from 0.1 to 30 eV) has been carried out via MD simulations. Ag atoms were deposited normal to the surface over the regions shown in Fig. 5. Each Ag atom is initially placed over 10 Å above the surface outside of the ReaxFF cutoff. Here, a sample of 400 separate

Conclusions

A new potential to describe the Ag–ZnO system has been developed. The new parameters were fitted against DFT calculations with the aim of reproducing equations of state for simple crystal structures and Ag on ZnO binding energies. Overall, the fitted parameter set agrees well with the DFT data used within the fitting procedure as well as agreeing with experimental data and calculated works of separation. As a preliminary phase to undertaking growth simulations [37], single point depositions

Acknowledgments

The research at Loughborough was supported by EPSRC, AGC Glass Europe and the Loughborough HPC unit. The work in Mons was supported by the Région Wallonne, (Aide à la subvention—Project 7189 with AGC Glass Europe) and the Belgian National Fund for Scientific Research. J.C. and D.B. are FNRS research directors. The authors would like the thank Hugues Wiame and Benoit Lecomte for useful discussions.

References (37)

  • E. Barthel et al.

    Thin Solid Films

    (2005)
  • D. Raymand et al.

    Surf. Sci.

    (2008)
  • M. Galeotti et al.

    Chem. Phys. Lett.

    (1994)
  • T.M. Parker et al.

    Surf. Sci.

    (1998)
  • G. Kaye et al.

    Tables of Physical and Chemical Constants

    (1995)
  • A. Raki et al.

    Appl. Opt.

    (1998)
  • G. Hass

    J. Opt. Soc. Am.

    (1955)
  • H. Glaser

    Appl. Opt.

    (2008)
  • Z. Wang

    J. Phys. Condens. Matter

    (2004)
  • Z. Lin et al.

    Phys. Rev. B

    (2007)
  • S. Blackwell et al.

    J. Phys. Condens. Matter

    (2013)
  • I. Gheewala et al.

    Philos. Mag.

    (2009)
  • K. Chopra et al.

    Thin Film Device Applications

    (1983)
  • R. Wahl et al.

    Phys. Rev. B

    (2013)
  • L.V. Koplitz et al.

    J. Phys. Chem. B

    (2003)
  • Cited by (0)

    View full text