Non-platinum group metal (non-PGM) oxygen reduction reaction (ORR) catalysts, based on pyrolyzed C, N, and transition metal systems, may replace precious metal catalysts in Pt-based fuel cell cathodes in the future. Major barriers to the adoption of these catalyst materials include much needed improvements in both ORR activity as well as durability under operating conditions. Detailed knowledge of the atomic-scale structure of the active site(s) would aid in strategically synthesizing materials with a higher density of desired active sites (those with simultaneously high ORR activity and practical stability). Density functional theory (DFT) modeling has previously been utilized to consider both relative stability of atomic scale structures,1 as well as to evaluate ORR activity2-5. Through the use of easily calculated descriptors, high-throughput DFT-based materials searches are possible. Relative formation energies of structures can be calculated via DFT with uniform atomic reference states. These formation energies may serve as a descriptor of relative active-site structure stability. This approach also allows for consideration of shifts in chemical potentials that can affect relative stability. Once the relaxed structure has been simulated, calculation of an active site with a bound ORR intermediate allows for an estimate of the binding energy of that intermediate using a computational hydrogen electrode approach2. Previously, O binding has been utilized as an activity descriptor for ORR catalysts2 but for non-PGM catalysts, OH binding is often the potential determining step in the ORR reaction pathway. As such, for screening purposes, we utilize the OH binding energy as a descriptor for ORR activity. Ideally, this value should be close to 1.23 eV with values less than this indicating OH overbinding and values greater indicating underbinding. Scaling law relations6suggest that both O and OH binding energies should be equally valid activity descriptors. We have performed an initial screening of 22 proposed active-site structures containing C, Fe, N, and H, as suggested through previously established structural motifs7. Three graphene-like C host materials, a “bulk” periodic graphene sheet, a zig-zag edge graphene nanoribbon, and an arm-chair edge graphene nanoribbon were considered. Active-site structures with one and two N-coordinated Fe atoms were considered as embedded in each of the three host materials, acting as C substitutional defects. In some cases, iron replaced two carbons to give different Fe coordination symmetry. Direct comparisons between similar structures in different C host materials were made where possible. For the nanoribbon hosted active site structures, the N-coordinated Fe moiety was only considered at the nanoribbon edge. The outcomes of this study will be the focus of this talk. In addition to the previously reported Fe2N5(*OH) zig-zag edge active site structure,8 preliminary studies suggest that bulk and zig-zag edge Fe2N6 structures (Figure 1) also have favorable OH binding characteristics, suggesting further investigation of their full ORR reaction pathways is merited and will be discussed. In all three cases, Fe2N5(*OH) zig-zag edge, Fe2N6 bulk, and Fe2N6 zig-zag edge, Fe is five-fold coordinated with N and a neighboring Fe, suggesting such coordination is an important aspect for tuning OH binding energy. Fe-Fe bond distances of 2.20 Å, 2.25 Å, and 2.24 Å were calculated for the Fe2N5(*OH) zig-zag, Fe2N6 bulk, and Fe2N6 zig-zag structures, respectively. In the case of the Fe2N5(*OH) zig-zag structure, the Fe atoms are initially out of the graphene plane whereas in the Fe2N6structures, they are initially in the graphene plane. Both multi-metal structures show enhanced stability, compared to their single-Fe counterpart structures for a wide range of Fe chemical potentials. By combining these simulation data, all calculated using the same methodology, direct comparison of structure-function relations for activity and structural stability for a wide range of structures, including similar structures at bulk and multiple edges, will be available for comparison. References 1 Holby, E. F. & Taylor, C. D. Applied Physics Letters 101, 064102, (2012). 2 Norskov, J. K. et al. Journal of Physical Chemistry B 108, 17886-17892, (2004). 3 Studt, F. Catalysis Letters 143, 58-60, (2013). 4 Kattel, S., Atanassov, P. & Kiefer, B. The Journal of Physical Chemistry C 116, 17378-17383, (2012). 5 Kattel, S., Atanassov, P. & Kiefer, B. Physical Chemistry Chemical Physics 15, 148-153, (2013). 6 Abild-Pedersen, F. et al. Physical Review Letters 99, 016105, (2007). 7 Holby, E. F., Wu, G., Zelenay, P. & Taylor, C. D. The Journal of Physical Chemistry C 118, 14388-14393, (2014). 8 Holby, E. F. & Taylor, C. D. Scientific Reports 5, (2015). Figure 1. Model Fe2N6 structure at a zig-zag edge nanoribbon with bound OH ORR intermediate. Figure 1