Magnetism of the V(001) surface: Contradictory results from pseudopotential and linearized augmented plane-wave calculations

Recently using a pseudopotential and the generalized gradient approximation (GGA) we determined that V(001) with the experimentally determined surface relaxation has a strongly ferromagnetic surface. More recently, others also using the GGA found that with full-potential linearized augmented plane-wave (FLAPW) calculations, the unrelaxed surface is only weakly magnetic and the relaxed surface is nonmagnetic. We report here other pseudopotential and FLAPW calculations. The FLAPW results are consistent with the previous ones except that the surface relaxation is much smaller. A nonmagnetic pseudopotential calculation yields a -13.5% surface relaxation. slightly larger than the FLAPW calculation, while the ground state has a -12.5% surface relaxation with a surface magnetization of 0.75 mu (B). The unrelaxed surface has a 1.77 mu (B) magnetization in good agreement with our previous calculation. Thus we are forced to conclude that what are usually assumed to be the most accurate methods of electronic structure calculation are in complete disagreement insofar as the magnetic nature of the V(001) surface is concerned. We speculate on the source of this disagreement.