Hello,
A yearly allocation for the LCRC cluster has been requested with the
following updated information:
Submitter/PI: John J. Low
Project Name: DFT_NiOxHy_on_Pt3Ni
Division: APS
Project title: First principles studies of stability and reactivity of Ni (hydr)oxide films on Pt3Ni(111) surface in alkaline environments.
Associated funding: DOE, Office of Basic Energy Science
Other Systems: Carbon (CNM-ANL), Carter (Purdue)
Science: This project will employ Density Functional Theory (DFT) plus Hubbard U and van der Waals corrections, and ab inito molecular dynamics, combined with atomistic thermodynamic models and computational hydrogen electrode, to study the structures of the interfaces between monolayer Ni (hydr)oxide films, close-packed Pt3Ni alloy substrates, and vacuum or aqueous bulk phases; additional work will focus on probing the catalytic properties of these interfaces for the hydrogen evolution reaction (HER) in alkaline solution. The relevant films, i.e. Ni (hydr)oxide films on Pt(111) surface, which can be synthesized using standard electrochemical techniques, have been investigated experimentally in the group of Nenad Markovic (ANL-MSD) and have been shown to exhibit bifunctional catalytic properties for hydrogen evolution and other electrocatalytic reactions. To understand the bi-functional mechanisms, we have performed an extensive series of studies on the electrochemical phase
diagrams/pourbaix of monolayer Ni-(hydr)oxide films on Pt(111) and the three-phase boundaries (TPB) at the film edges at different potential. The kinetic studies have shown that water dissociation, a key step in alkaline HER, is significantly accelerated at the TPB edges. Based on these understanding, we have extended the effort to probe the alloy effect of the substrate in the last proposal. The preliminary results have shown that the TPB edge of Ni-(hydr)oxide films on Pt3Ni may have similar promotion effects toward water dissociation to that on Pt(111) substrate. Due to limitation of core hours, however, the kinetic analyses so far are only performed on a small set of film-electrode-vacuum TPB edges and without extra hydrogen adsorption on Pt3Ni electrode part. In the present proposal, besides including more TPB models to generalize the understanding on the water dissociation kinetics, we will consider two realistic effects catalytic conditions, hydrogen coverage effect
and the solvation effect. We anticipate that these two effects may have a substantial impact on both the thermodynamics and kinetics, and the present study may reveal true chemistry at the film-electrode-liquid-water TPB edges.
Project description: Electrochemical surface studies on well-defined, single crystal metal and oxide surfaces have provided tremendous understanding of the fundamentals of chemistry and catalysis at interfaces in the past 20 years2, 3. The applications of such systems are wide-ranging, from fuel cells to solar fuels production. However, very few studies have rigorously probed the fundamentals of electrocatalytic processes at the interface between oxide/hydr(oxy)oxide films, metal or alloy substrates, and bulk solution phases. Metals and oxides have dramatically different structures, properties, and affinities to reactive species, which bring potentially unique properties to electrocatalytic process at interfaces between these disparate classes of materials. Experimental results by Nenad Markovic’s group (ANL/MSD) have demonstrated that certain electrochemically deposited oxides/hydr(oxy)oxides (nickel, cobalt, iron, and manganese-based, deposited as islands with diameters
of 5-10 nanometers) on Pt(111) can efficiently catalyze the hydrogen evolution reaction (HER) with activities never before seen on either metal or oxide catalysts in alkaline electrolytes1. The bi-functional mechanisms behind have begun to be understood through our recent work. To study the generality of the understanding, we have extended out study from Pt substrate to Pt3Ni substrate to evaluate the alloy effect. Indeed, some Pt alloys have shown higher electrocatalytic activity than Pt itself. Investigation of alloy substrates is thus a natural extension of the work on pure Pt substrates and may provide an extra variable to tune the activity of water dissociation and hydrogen evolution reactions at three phase interfaces. The preliminary results also show the monolayer Ni-(hydr)oxide/Pt3Ni is a promising bi-functional electrocatalyst. Though basic thermodynamic and kinetic analyses make this systems very interesting, a few realistic effects (e.g. coverage effect and elect
rochemical double layer) have not been included in the studied yet, which in some instances, yield completely unexpected physical and catalytic properties. To evaluate the influence of true electrocatalytic environment, we will conduct more careful study by considering hydrogen coverage and explicit solvent effect. As there is still no experiment on these combined systems, the true chemistry predicted from present study could provide a blueprint for the future search for and preparation of catalysts with higher activity.
In the previous proposal cycle, we have developed a new technique to highly accurately describe the stability of (hydr)oxides, and have found that using water reference, van der Waals correction and optimized Hubbard U together, the formation energy of (hydr)oxides can be predicted highly accurately, i.e. the standard deviation with respect the experiment values could be around 0.04 eV. This accuracy has never been reached by standard DFT calculations in previous study, and has strong implication for the application on the extended systems. For example, we have used this technique to study the structural evolution of Ni-(hydr)oxides films deposited on Pt(111) substrates, and the bi-functional mechanism of the NiOH-Pt-H2O TPB. The experimental confirmation of on the predicted film structures and the clarification of the bi-functional mechanism not only
only a big step forward towards accurate identification and prediction of a variety of oxide/electrode interfacial structure-properties relationship, but also provides the foundation for rational design and control of ‘targeted active phases’.
In the last proposal cycle, we have applied the same methodology to screen the electrode potential-dependent structures of monolayer Ni-(hydr)oxides films deposited on Pt3Ni(111) substrates. Based on it, we have generated a few simple TPB models at HER potential region, and obtained basic water dissociation kinetics therein. In order to investigate on the influence of liquid water on the thermodynamics and kinetics at the TPB, we have started the AIMD simulation of liquid water on well-defined Pt(111) surface to accumulate fundamental knowledge.
In the present cycle, we will focus on the influence of hydrogen coverage and/or liquid water on the thermodynamics and kinetic at TPB. The TPB models developed in the last proposal cycle is composed by 1 nm NiOH ribbons deposited on Pt3Ni(111)-(6x2) substrate. Though they have been larger than most of models used on single crystal surfaces and have been large enough to consider the influence of H coverage, they are still not large enough to accommodate extra water molecule at the TPB to roughly evaluate the influence of liquid water on the thermodynamics and kinetics. Thus, we will firstly built up similar but large TPB models, e.g. 1 nm NiOH ribbons deposited on Pt3Ni(111)-(8x4) substrate, and study dissociation kinetics of water monomer. Though we anticipate the trends on the large TPB models may be somehow similar to that we have obtained on small TPB models, these studies could provide richer and more general information. To study the influence of H coverage, we will fir
stly establish the equilibrium H coverage through Langmuir isotherm analysis. Then will use the corresponding H coverage and geometric information in the following up studies without or with extra water in the systems. To investigate the influence of H2O on the thermodynamics and kinetics, we will initially introduce one extra water in the systems without and with pre-covered H. We anticipate that it (this extra water) is likely having substantial influence on the thermodynamics and kinetics at TPB, and could establish basic but reliable information on the role of liquid water. To obtain more precise information, we will introduce liquid water in a few promising systems by filling the vacuum region of models. We will conduct geometry optimization to find local minimum, and then perform AIMD simulation to find more realistic water-electrode interfaces. Technically, we will firstly heat the systems up to 400 K to obtain liquid phase, then we will perform canonical ensemble (NV
T) simulation with time upto 10 ps. We will cool the systems down to 0 k again every 1-2 ps during the simulation to find the most stable geometry. Then, it will be used as the starting point in the thermodynamics and kinetic analysis to obtain final data.
Ultimately, we would like to compare the water dissociation properties for different oxide/hydr(oxy)oxide films on both pure Pt and alloy substrates. If the hypothesis that H produced from water dissociation moves to metal sites is correct, then the thermodynamics and kinetics of water splitting will be substantially different on the different metal substrates. In combining with the influence of substrates to the thermodynamics and kinetic of hydrogen evolution (2H→H2), this could provide significant additional information that would be of use in further searching for and optimizing bifunctional catalysts, by tuning the films and substrates.
The present proposal require a remarkable amount of time consuming kinetic analysis using nudged elastic band (NEB) method. Spin-polarized calculations with the VASP code have shown that full convergence of water dissociation on magnetic TPB model NiOH/Pt3Ni(111)-(6x2) require approximately 50 hours on 8 16-cores Intel Xeon-E5-2670 for an 8 image case (640 core hours). Consider more than two time larger super cells (NiOH/Pt3Ni(111)-(8x4)), the computational time would be quadruple (2,560 core hours per NEB calculation). For each case, i.e. with or without pre-covered H, and with or without one extra water, we need to run at least 10 NEB calculations (40 in total), which require total 102,400 core hours. The AIMD simulations on 3ML Pt(111)-c(3x4) filled with 48 water molecules have shown that it usually takes 12 hours on 4 16-cores Intel Xeon-E5-2670 to perform 1 ps NVT simulation (16,896 core hours). For the AIMD simulations, we at least need to run on two models (with and wi
thout pre-covered H) upto10 ps simulations. Considering heating up (2ps) and cooling down (2ps) every 2ps, we need actually run 22 ps per simulation. The systems here, i.e. NiOH/Pt3Ni(111)-(8x4) with around 100 water molecules, is around two time larger than that on Pt(111)-c(3x4). Consider the spin-polarization calculations here, it is actually equivalent 4 time larger, and may be 8 time more demanding in the simulation. Then it requires around 270,336 core hours to complete two AIMD simulations for screening the realistic TPB between Ni-hydroxide film, Pt3Ni electrode and the electrolyte (liquid water). The following up kinetic analyses require to run around 10 NEB calculation for each model (with and without pre-covered H). As the systems filled with water are likely four time more demanding, this require another 200,800 core hours. Consider we may spend extra 10% of above core hours on routine geometric optimization, e.g. to find most stable initial states, final states,
realistic H coverage, etc, and another 10% extra time on issues come out randomly, we need 685,843 core hours to finish the work proposed.
The models of fully two dimensional hydroxide films on Pt(111), which were developed in previous proposal cycles, have been published in Nature Materials1. The newly developed scheme towards first-principles based prediction of highly accurate electrochemical Pourbiax diagrams, proposed general strategy towards controlling the structural evolution of the ultra-thin film/electrode interfaces, and the understanding of the bi-functionial mechanism of electrocatalyst NiOH/Pt111, which are generated from previous proposal cycles, could be the interest of the broad community and three papers are in preparation. The present study, a new thrust based on previous studies, not only would provide new insight towards the complex electrochemical interface, but also may suggest new materials for electrocatalytic application.
References
1. R. Subbaraman, D. Tripkovic, K. C. Chang, et al., Nature Materials 11, 550 (2012).
2. N. M. Markovic and P. N. Ross, Surface Science Reports 45, 121 (2002).
3. D. M. Kolb, Surf. Sci. 500, 722 (2002).
4. J. Greeley, T. Jaramillo, J. Bonde, et al., Nature Materials 5, 909 (2006).
5. J. Greeley, I. E. L. Stephens, A. S. Bondarenko, et al., Nat. Chem. 1, 552 (2009).
6. Z.-H. Zeng, J. L. F. Da Silva, and W.-X. Li, Phys. Chem. Chem. Phys. 12, 2459 (2010).
7. Z.-H. Zeng, J. L. F. Da Silva, and W.-X. Li, Phys. Rev. B 81 (2010).
8. M. J. Piotrowski, P. Piquini, Z. Zeng, et al., J. Phys. Chem. C 116, 20540 (2012).
9. D. R. Butcher, M. E. Grass, Z. Zeng, et al., J. Am. Chem. Soc. 133, 20319 (2011).
10. Z. Zeng, M. E. Bjorketun, S. Ebbesen, et al., Phys. Chem. Chem. Phys. 15, 6769 (2013).
11. J. Greeley, L. Kibler, A. M. El-Aziz, et al., ChemPhysChem 7, 1032 (2006).
12. J. K. Nørskov, J. Rossmeisl, A. Logadottir, et al., Journal of Physical Chemistry B 108, 17886 (2004).
Industry partnership:
Project URL:
Current FY Hours Used: undetermined amount
New FY Requested allocation: 700000
Q1: 175000
Q2: 175000
Q3: 175000
Q4: 175000
Justification: The current version of VASP has made some major improvements in parallel scaling. Specifically data is distributed to minimize the amount of memory per node and parallelism over K points has been introduced. For the size of the models we are using in this project good parallel scaling is observed up to 640 cores on blues and fusion.
Storage requirements:
Thank You,
The LCRC Accounts System
