Faceting and stress of missing-row reconstructed transition-metal (110) surfaces

We present ab initio total energy and stress calculations for the unreconstructed and (2x1)-missing-row reconstructed Ir (110) and Rh (110) surfaces. We then use those results to set up a model rationalizing the (2x1) reconstruction as a faceting transition to a long-wavelength-corrugated (111)-like surface. Next, we discuss the qualitative extension of such model to a general nx1 reconstruction, using ab initio results, elasticity theory, and classical dynamics simulations for Al (110). Remarkably, despite the severe inherent limitations of the model, the 3x1 structure is found to be the most stable for Ir. Finally, we use the stress density to analyze the stress increase upon reconstruction, and find it to be due to a changed balance of tensile and compressive contributions in the near-surface region, which closely matches previous interpretations of the reconstruction mechanism. We conclude that, as for the (100) surface, the reconstruction basically originates from the strong relativistic contraction effects on the electronic structure of end-of-series 5d metals. [S0163-1829(99)04643-3].