Ab Initio Studies of Solid Oxide Fuel Cell Cathode Materials

Yueh-Lin Lee^a, Milind Gadre^a, Shih-kang Lin^b, Brian Puchala^b, Benjamin Swoboda^c, Leland Barnard^a, Dane Morgan^a,b,c, Jesper Kleis^d, Jan Rossmeisl^d, Riza Dervisoglu^e, Derek Middlemiss^e, and Clare Grey^e

^a. Materials Science Program, University of Wisconsin-Madison

^b. Department of Materials Science and Engineering, University of Wisconsin-Madison

^c. Department of Nuclear Engineering - Engineering Physics, University of Wisconsin-Madison

^d. Center for Atomic-scale Materials Design, Technical University of Denmark

^e. Department of Chemistry, Stony Brook University

      Solid oxide fuel cell cathodes must catalyze the oxygen reduction reaction (ORR), which consists of the reaction O₂(gas) -> 2O^2-(bulk). The oxygen is initially in the gas phase, and after reduction must be incorporated into bulk for transport through the electrolyte to the anode. The ORR contribution to the overpotential is expected to become increasingly important as other sources of voltage loss are reduced (e.g., by the making thinner and less resistive electrolyte films) and as researchers push to lower the SOFC operating temperatures to reduce degradation rates and material costs. Perovskite oxides (with formula unit ABO₃, where A and B are cations or sets of cations) are active for the ORR under SOFC conditions (generally 800 K – 1300 K and air atmosphere) and have a number of advantages over competing materials, including their stability at high temperature, reasonable cost, and acceptable thermal expansion properties. The ORR processes on SOFC cathodes surfaces are inherently complex with many possible pathways, as illustrated in Figure 1. Due to the complexity of the ORR mechanisms and the difficulty of resolving surface ORR steps in experiments, factors governing the cathode performances are still poorly understood, including the rate-limiting steps and how different transition metal cations alter the catalytic properties.

Figure 1. Possible reaction mechanisms of oxygen reduction reactions (ORR) on SOFC cathodes.

      Ab initio methods offer a powerful tool to probe catalytic properties by investigating electronic structures, reaction energetics, and activation barriers at molecular scales, and have been successfully applied in understanding trends in reactivity for metals. In modeling the ORR on complex oxides, careful treatments are needed to take into account the correlated electron effects in transition metal perovskites, Jahn-Teller distortions, and magnetic ordering. Also the oxygen gas reference state must be corrected to properly describe the oxygen exchange reactions underlying the ORR energetics. Finally, the defect physics predicted by zero-temperature (T = 0 K) DFT approaches must be supplemented with proper thermodynamics to yield values accurate at SOFC operating temperatures. Detail discussions on the ab initio modeling approach for SOFC cathode materials are described in Ref. 1-3.

      Three trusts of research works conducted by the CMG SOFC team and collaborators are currently ongoing to tackle with different levels of material chemistry that play important roles in ORR on transition metal perovskites, which include: 1. Bulk defect chemistry and diffusivity, 2. Surface properties, and 3. Catalytic properties.

1. Bulk defect chemistry and transport properties:

• Ab-initio based CALPHAD modeling of (La,Sr)MnO₃ (LSM) and (La,Sr)CoO₃ (LSC) defect chemistry: In transition metal perovskites, defect energetics are coupled to oxidation/reduction of transition metal cations. In this work, we are developing ab initio doping approaches to allow calculation of defect energy dependence with respect to electron Fermi level.  These techniques will then allow practical and robust prediction of ab initio defect energetics in these systems, which will then be applied to develop complete defect models. Specifically, due to complexity of the LSM and LSC systems, current defect models mostly neglect non-ideal behavior of defects. As workers are pushing to lower SOFC operating temperature, the non-ideal contribution will become increasingly important. In this work, we incorporate ab initio defect energetics and interactions into a CALPHAD modeling to study LSM and LSC defect chemistry.

Participants: Milind Gadre, Shih-kang Lin, Yueh-Lin Lee, and Dane Morgan

• Ab-initio based thermokinetic modeling of cation and anion transport in LSM: We are combining ab initio calculations to parameterize energetics on the perovskite lattice and KMC simulations to study the LSM cation and anion transport.  Such transport processes are essential for the performance of the materials as a catalyst and its long term stability against kinetic demixing.

Figure 2. Ab-Initio based thermokinetic modeling approach for cation and anion transport in LSM

Participants: Brian Puchala, Yueh-Lin Lee, Benjamin Swoboda, Leland Barnard, and Dane Morgan

2. Surface properties:

•  LSM surface defect model and surface diffusivity [3]: We combine bulk defect models fit to experimental data with ab initio surface energies (Figure 3a) to predict surface vacancy concentrations in (La_1-xSr_x)MnO₃ (Figure 3b). It is found that the surface vacancy concentration is ~10⁶ times that of bulk under SOFC operating conditions [3].

Figure 3. 3a. Calculated oxygen vacancy formation energy as function of position in an 8-layer slab of LaMnO₃. 3b. Predicted oxygen vacancy concentrations at T = 1173 K as a function of oxygen partial pressure for (La_0.9Sr_0.1)MnO₃.

• LaBO₃ surface [1,3]: LaBO₃ (B=Mn, Fe, Co, and Ni) perovskites form a family of materials of significant interest for cathodes of solid oxide fuel cells (SOFCs). In this work ab initio methods are used to study both bulk and surface properties of relevance for SOFCs, including vacancy formation and oxygen binding energies. The LaBO₃ oxygen vacancy formation energies are predicted to be in the order Fe > Mn > Co > Ni (where the largest implies most difficult to form a vacancy). It is shown that (001) BO₂ terminated surfaces have 1-2 eV lower vacancy formation energies, and therefore far higher vacancy concentrations, than the bulk. The stable surface species at low temperature are predicted to be the superoxide O₂^- for B = Mn, Fe, Co and a peroxide O₂^2- with a surface oxygen for B = Ni. These results will aid in understanding the oxygen reduction reaction on perovskite SOFC cathodes.

Figure 4. 4a. Calculated oxygen vacancy formation energies for stoichiometric bulk LaBO₃ (B = Mn, Fe, Co, Ni) and (001) BO₂ terminated surfaces. Green diamonds are collected data from experiments. 4b. LaBO₃ BO₂ surface oxygen adsorption energies vs. transition metal types.

Participants: Yueh-Lin Lee and Dane Morgan

Collaborators: Jesper Kleis and Jan Rossmeisl (CAMD, DTU)

3. Catalytic properties:

      The ORR process at oxide surfaces is inherently complex with many possible pathways. To develop new and more optimal materials, it is not only important to gain fundamental insight into the details of a single system, but to find fundamental intrinsic descriptors for the reaction process that can efficiently aid the search for new materials. By extending approaches originally explored for simple metal catalysts we are developing descriptors for the ORR on perovskites to design better SOFC cathodes.

Participants: Yueh-Lin Lee and Dane Morgan

Collaborators: Jesper Kleis and Jan Rossmeisl (CAMD, DTU)

Other SOFC relevant work:

Ab initio modeling of oxygen-vacancy ordering in Ga doped Ba₂In₂O₅ brownmillerite

      Ga-doped Ba₂In₂O₅ (BIO) brownmillerite is a potential fast ionic conductor as the SOFC electrolyte. As our ab initio based thermodynamic model has accurately reproduces the phases and order of the phase transitions for oxygen-vacancy ordering in pure BIO [4], we are currently extending the work to study Ga-doped BIO to improve understanding of fast oxygen transport, local structural changes in BIO, and the impact from the Ga doping. This will combine variable temperature PDF analysis from experiments with results from ab initio modeling to have detailed information of oxygen-vacancy ordering in the system.

Participants: Riza Dervisoglu, Yueh-Lin Lee, and Dane Morgan

Collaborators: Riza Dervisoglu, Derek Middlemiss, and Clare Grey (Stony Brook University)

We gratefully acknowledge financial support from the NSF MRSEC program (0079983), DOE Office of Basic Energy Sciences (DE-SC0001284), and computing support from NSF National Center for Supercomputing Applications (NCSA - DMR060007).

References:

1.       Y.-L. Lee, J. Kleis, J. Rossmeisl, and D. Morgan, Ab initio Oxygen Reduction Reaction Energetics of LaBO₃ (B=Mn, Fe, Co, and Ni) (001) Surfaces for Solid Oxide Fuel Cell Cathodes, Phys. Rev. B, 80, 224101, (2009)

2.       Y.-L. Lee, J. Kleis, J. Rossmeisl, and D. Morgan, Ab initio Defect Energetics in LaBO₃ Perovskite Solid Oxide Fuel Cell Materials, ECS Transactions, 25 (2), 2761-2767 (2009)

3.       Y.-L. Lee and D. Morgan, Prediction of Surface Oxygen Vacancy Concentrations of (La_1-xSr_x)MnO₃, ECS Transactions, 25 (2) 2769-2774 (2009)

4.       Y.-L. Lee and D. Morgan, Ab Initio Study of Oxygen-Vacancy Ordering in Oxygen Conducting Ba₂In₂O₅, Mater. Res. Soc. Symp. Proc. 972, 0972-AA04-06 (2007)