An Atomistic Investigation of Vicinal Calcite Surfaces: The Effect of Terrace Width on Surface Structure and...

The {10.4} calcite surface is commonly chosen for computational studies because unless some growth modifier is present, it dominates crystal form in nature and industry. Its rhombohedral symmetry produces growth and fracture terraces separated by acute and obtuse steps that have been studied experimentally in vacuum, air, and solution, during growth and dissolution. The two favored model systems for computational studies are vicinal surfaces with {10.4} terraces and etch pits or islands, manually created on {10.4} surfaces. Although experimental studies demonstrate that even for high step frequencies, terraces are hundreds to thousands of nanometers wide between steps, computational costs have driven density functional theory, DFT, and early molecular dynamics, MD, studies to work with relatively narrow terraces. An obvious question is how wide must a terrace be to correctly predict stepped surface structure and properties? We made dispersion-corrected DFT (DFT-D2) calculations for 16 vicinal surfaces and analyzed the effect of terrace width on surface structure and adsorption energy. For accurate theoretical treatment, the distance between the steps must be around ∼22 Å for acute and ∼26 Å for obtuse steps. On terraces, structure and adsorption energies resemble those of {10.4} surfaces at ∼13 Å away, above and below acute steps; on obtuse steps, the distances are ∼16 Å away, above and ∼10 Å below. The terrace width required for convergence is therefore significantly wider than assumed in most previous studies. This means that the standard approach for modeling atomic steps on calcite introduces systematic errors, namely up to 0.2 Å for Ca-Ca distances at the step and up to 0.2 eV for the difference between terrace and step sites, for the adsorption energy of water. In this study, water served as a probe for adsorption energy, and though not explicitly studied, adsorption energies for other molecules are likely affected similarly.