<HTML><BODY style="word-wrap: break-word; -khtml-nbsp-mode: space; -khtml-line-break: after-white-space; ">Dear Wien-Users,<DIV><BR class="khtml-block-placeholder"></DIV><DIV>I'm using Wien2k and the virtual crystal approximation (VCA) to simulate the metallic state in highly Boron doped diamond. K.W. Lee and W.E. Pickett [PRL93 (2004) 237003] used VCA in the Wien2k code and got a metallic hole Fermi surface with a Fermi energy of about 0.8 eV at 2.5 at% doping. When I put 2.5 at% doping, i.e. I replace Carbon with Z=6 by a virtual atom with Z=5.975, the calculation converges into a insulating solution. At 2.9% and above I get a metallic band structure with E_F of about 1 eV, i.e. the Fermi level is about 1 eV below the (now unoccupied) valence band maximum. I don't do any phonons (yet) like Lee and Pickett did, I just want to get the ground state of the undistorted lattice. I tried volume optimization and found a 4% volume expansion, but at 2.5% doping I get an insulating state for all lattice parameters. </DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV>How can I reproduce the result of Lee and Pickett? Is there a trick for VCA? Where do my 0.025 electrons per atom go? Or is this some kind of Mott transition in the GGA? </DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV>I use GGA with the 1s as core electron (energy separation -6 Ry). To sample the small Fermi surface well I use a dense k-mesh of 1240 k-points in the reduced wedge (36 x 36 x 36 in total). I use WIEN2k_05.3 (Release 3/4/2005) and I got the same result also with Wien97. I attach my .struct and .inst file to this e-mail.</DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV>Many thanks for any elucidating remarks</DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV>Moritz</DIV><DIV><BR class="khtml-block-placeholder"></DIV><DIV></DIV></BODY></HTML>