Nanoscale superlattices of samarium-doped ceria layers with varying doping levels have been recently proposed as a novel fuel cell electrolyte. We calculate the equilibrium composition profile across the coherent f100g interfaces present in this system using lattice-gas Monte Carlo simulations with long-range interactions determined from electrostatics and short-range interactions obtained from ab initio calculations. These simulations reveal the formation of a diffuse, nonmonotonic, and surprisingly wide (11 nm at 400 K) interface composition profile, despite the absence of space charge regions. The performance of solid oxide fuel cells [1] is steadily increasing, thanks to the development of novel electrode and electrolyte materials. A recent trend is the use of nanostructured materials to devise electrolytes and electrodes with improved properties [2]. In particular, nanoscale superlattices consisting of layers of the samariumdoped ceria ion conductor with various levels of doping have been proposed as an avenue to improve ionic conductivities of the electrolyte in the low-temperature regime [3,4]. Optimizing this approach as well as understanding the mechanisms underlying the increased ionic conductivity requires a detailed understanding of the interfacial properties in this system.In this Letter, we present ab initio calculations of the equilibrium composition profile (of dopants and vacancies) across an array of coherent f100g interfaces mimicking the experimentally synthesized samarium-doped ceria superlattices. These calculations reveal two surprising findings. First, the coherent interface width is large and represents a significant fraction of the period of superlattices [3,4], thus suggesting a very plausible explanation for the observed significant ionic conductivity improvements. Interestingly, this phenomenon is not driven by space charge effects [5,6], the most commonly invoked explanation for longrange deviations from the bulk equilibrium vacancy composition. The composition profile also exhibits composition oscillations, a feature that, to our knowledge, has never been noted in ceramic coherent interfaces.The determination of the equilibrium f100g interface composition profile in this system demands simulation cell sizes that are beyond the reach of a brute-force ab initio approach: In order to obtain results that are converged with respect to system size, a system consisting of over 70 000 atoms is needed. Moreover, finding the equilibrium profile requires extensive sampling of configuration space that would be prohibitive via ab initio methods alone. To address these issues, we rely on the cluster expansion formalism [7][8][9][10], which represents the energy E of a crystalline alloy with a computationally efficient Hamiltonian taking the form of a polynomial in terms of occupation variables i indicating the type of atom residing on each lattice site i:The unknown coefficients J . . . of this polynomial are called effective cluster interactions (ECI) and are fit to a database of ab initio stru...