Reactive molecular dynamic simulations of hydrocarbon dissociations on Ni(111) surfaces
Highlights
► Force field development for Ni/H/C systems based on first-principles data. ► MD simulations on CH4, C2H6 and n-C4H10 dissociative chemisorptions on Ni(111). ► Use of microkinetic models for CH4 dissociation and coke formation analyses.
Introduction
The quest for efficient use of fossil-based energy resources includes intense interest in the direct use of hydrocarbon feedstock by solid-oxide fuel cells (SOFCs) [1], [2]. This process can effectively avoid the limitation of storing and transporting hydrogen and is potentially of much higher efficiency than that can be achieved in a conventional energy use scenario. However, SOFCs are plagued by unwanted reactions on anodes such as soot formation [3] and sulfur poisoning [4], [5], [6], which must be resolved before larger scale application of this technology. Towards this end, it is of great value to extend the reach of computational inquiries so that they can provide both qualitatively and quantitatively accurate descriptions of the complex reactions associated with the decomposition of hydrocarbons and initiation of carbon coke formation. The current work takes a step in this direction.
Molecular Dynamics (MD) provides an explicit approach for studying chemical reactions. Molecular trajectories can be easily monitored in order to quantify the evolution of surface species. In addition, precise reaction conditions can be modeled and explored.
MD has the potential to generate macroscopic kinetic information for surface reactions. A semi-empirical potential energy surface constructed by Lee and DePristo[7] with many-body expansions (e.g. 3-body, 4-body and so on) approximated by the Morse potential to study hydrogen dissociation on Ni(111) and Ni(100) surfaces. However, the LEPS (London–Eyring–Polanyi–Sato potential energy surface) used does not provide a very accurate reactive potential energy surface. A theoretical model using reduced dimensions, i.e., Z (methane-surface distance) and D (methyl-H distance) presented by Harris and Luntz et al. [8], was used to explain the thermally assisted tunneling of methane dissociations on platinum surfaces. Nevertheless, possible steric hindering was eliminated and might affect the methane dissociation behavior. Based on Harris and Luntz's work, another reduced-dimension model was proposed by Weaver et al. [9] to extend its applications to other alkanes such as ethane, propane, and neo-pentane to investigate the steric factors in the C-H bond dissociation. The steric effect deficiency noted in the Harris’ model has been addressed in Weaver's work, but still in an empirical fashion. Some quantitative analyses were conducted using relatively computationally simple models. For example, Gislason and Sellers [10] used a modified bond order based Morse-type potential for C2 hydrocarbons so that Arrhenius rate constants, pre-exponential factors and activation barriers could be obtained. It would be ideal that qualitative and quantitative analyses on hydrocarbon dissociation on solid surface can be studied using a more sophisticated and relatively assumption free theoretical model so that more detailed and accurate chemistry can be properly captured especially when bond breaking and formation are involved.
In this article, we report the development of a reactive force field (ReaxFF) for the nickel/hydrocarbon system since nickel is a common and effective reforming catalyst, an active ingredient in the fabrication of SOFC anodes, but nickel is also susceptible to deactivation due to carbon formation. The Ni-H-C reactive force field parameters will be obtained by fitting to the ab initio data and applied within a classical molecular dynamics (MD) setting to investigate the decompositions of alkanes and elucidate the origins of carbon soot formation. The simulations were performed to provide some insights in the dissociation mechanisms and estimate the kinetic data needed for macroscopic analyses. We obtained an estimate for the sticking coefficient of methane dissociative adsorption on Ni(111) at the zero coverage limit as well as the activation energy. The MD analyses were then coupled with a microkinetic modeling technique to quantify the step dissociations of methane on Ni(111). Similar methodology was then applied to provide a preliminary understanding of the complex reaction sequences associated with the decomposition of ethane and n-butane on Ni(111).
Section snippets
Force field
The reactive molecular dynamics potential, known as ReaxFF [11], has been found to accurately capture the reaction dynamics of a variety of molecular systems in both catalytic [12] and non-catalytic [13] environments. The total energy is expressed as a functional of the bond-order, which depends on inter-atomic distance. Bonded inter-atomic interactions are implemented in terms of bond stretching, angle bending, and torsional rotation within a molecule. Special molecular structures, such as,
Reactive force field development
The development of the ReaxFF for hydrocarbon, and hydrocarbon/transition metal (e.g., Ni, Cu) in the gas phase have been reported by van Duin et al. for a variety of hydrocarbon molecules [11] and catalyzed carbon fullerenes formations [12]. Mueller and coworkers have reported a recent development for the nickel/hydrocarbon system [25]. In this work, we make a separate attempt at extending ReaxFF to Ni/C/H systems. The parameters for C–H interactions developed in [11] are used. Therefore, the
Methane dissociations on Ni(111)
Experimental and theoretical studies report the energy barrier associated with methane dissociative adsorption on Ni(111) in the range of 12–20 kcal/mol [7], [27]. One of the major challenges to perform MD to simulate this reaction is that it should be treated as a rare event on the classical MD simulation time scale, which is supported by the extremely low reaction probability (< 10− 9 on Ni(111) between 523 and 618 K) observed experimentally at low pressure [28].
The simulations of methane
Conclusions
We have generated a new reactive force field potential tailored for the interactions of hydrocarbons with nickel systems. The force field was applied within a classical MD-NVT setting to study the catalytic decomposition of several small hydrocarbon molecules, i.e., methane, ethane and n-butane on Ni(111). Quantum mechanical calculations were performed on nickel complexes molecules, nickel crystals as well as the (111) surface so that hydrocarbon–nickel interactions, more importantly
Acknowledgement
We would like to thank Dr. William A Goddard III and Dr. Adri C. T. van Duin for their valuable comments and suggestions on ReaxFF force filed and its parameter training during the preparation of this work. This work was financially sponsored by the U.S. Department of Energy, Office of Science, Grant No. DE-ER9542165. Computations were performed on the supercomputing cluster located at Golden, CO, which is managed by the Golden Energy Computing Organization (GECO) that uses sources acquired
References (35)
- et al.
J. Mol. Catal. A: Chem.
(2008) - et al.
J. Power Sources
(2008) - et al.
Appl. Catal. A: Gen.
(2009) - et al.
Surf. Sci.
(1998) - et al.
Comput. Mater. Sci.
(2003) - et al.
J. Catal.
(1992) - et al.
J. Catal.
(2000) - et al.
J. Catal.
(2002) - et al.
J. Catal.
(2004) - et al.
Appl. Catal A: Gen.
(2006)
Nature
Chemical Review
J. Electrochem. Soc.
J. Chem. Phys.
Phys. Rev. Lett.
J. Chem. Phys.
J. Phys. Chem. A
Cited by (20)
Soot growth mechanism in C<inf>2</inf>H<inf>2</inf> combustion with H<inf>2</inf> addition: A reactive molecular dynamics study
2023, International Journal of Hydrogen EnergyCitation Excerpt :Therefore, the ReaxFF MD can smoothly describe the bonding and breaking processes between atoms in the simulation system, and the simulation system does not need to pre-set the chemical reaction path. It has been widely used to study pyrolysis [36], combustion [37], explosion [38], oxidation [39], catalysis [40], and other systems involving physical chemistry. Recently, several studies have been reported the application of ReaxFF MD to explore soot formation process successfully [31,32,41,42].
LAMMPS molecular dynamics simulation of methane decomposition on nickel thin films at high temperatures
2021, Surface ScienceCitation Excerpt :ReaxFF has been widely applied for the MD treatment of surfaces, defects, diffusion, gases, liquids, and solids [9,10,11]. Reactions between methane and nickel catalysts have also been intensively investigated using MD coupled with ReaxFF [12, 13, 14]. In this study, we applied the ReaxFF potential provided by Mueller et al. [12] to measure the activation energy in CDM by Ni thin-film catalysts using a large-scale atomic/molecular massively parallel simulator (LAMMPS) [15].
Ab initio molecular dynamics thermal decomposition of methane on Ni-Cu/γ-Al<inf>2</inf>O<inf>3</inf> catalysts for CNT production: Effect of Cu promoter
2017, Computational Condensed MatterCitation Excerpt :Some applications of the Catalytic Thermal Decomposition of Methane (CTDM), one of the most common CCVD technologies [7], utilize Ni-Cu/γ-Al2O3 catalysts due to the susceptibility of Ni and γ-Al2O3 to carbon deposition and the catalytic-promoting property of Cu [7,8]. Molecular Dynamics (MD) simulations have been performed on the dissociation and reaction of CH4 on various catalytic surfaces [9–13], but so far only few have conducted simultaneous theoretical and experimental investigation on them. In a previous work [13], the decomposition of CH4 on unsupported amorphous Ni and Ni-Cu purely through ab initio Molecular Dynamics (AIMD) simulation was investigated.
A review on methane transformation to hydrogen and nanocarbon: Relevance of catalyst characteristics and experimental parameters on yield
2017, Renewable and Sustainable Energy ReviewsCitation Excerpt :Considering two sets of experiments, one with a slab of 720-atoms and 50 methane molecules and another slab with 2880-atoms and 200 methane molecules, authors’ concluded that number of dissociated molecules are proportional to the surface areas of the two slabs, which indicates the insignificants of size of decomposition system. They found the activation energy and sticking coefficient as 22 kcal/mol and 1×10−8, respectively, which are consistent with experimental results [255,256]. Additionally, they observed a faster decomposition at higher temperature in accordance to the higher number of collisions.
Classical atomistic simulations of surfaces and heterogeneous interfaces with the charge-optimized many body (COMB) potentials
2013, Materials Science and Engineering R: ReportsCitation Excerpt :Examples in the literature can be found where these two types of EE schemes have been applied to all types of interatomic bonding depending on the type of short-range potential with which the scheme is coupled [12,13,28,40,41]. For example, the EEM method is employed in the ReaxFF potential family [12,13,42–47] for a wide range of materials, while the QEq scheme is coupled to a Finnis–Sinclair potential [9] to model Al/Al2O3 interfaces [39], to a Morse-stretch potential in MS-Q [48] to model TiO2, to a second-moment tight-binding potential in SMTB-Q [49] to model TiO2 and ZrO2, and to Tersoff type potentials in COMB potentials to model metals, metal oxides and covalent materials. As exemplified above, modern empirical potentials for multi-component and multi-functional systems normally include both the bond-order concept for strong short-range interactions and a variable charge scheme to describe electrostatic energies.
Mechanisms of CH<inf>4</inf> activation over oxygen-preadsorbed transition metals by ReaxFF and AIMD simulations
2024, Journal of Computational Chemistry