<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">Colleagues, <div><br></div><div>I am working through a problem that I have assigned to my Computational Materials Science class - computation of the transition pressure for the diamond-cubic to beta-Sn structures in diamond using pw.x. The problem I have is that pw.x produces answers that are inconsistent with those produced by VASP, and those published in the literature. I have used similar ultrasoft pseudopotentials in both calculations (both based on PW91 correlation). There may be differences, between the two cases, but I have tried a number of pseudopotentials in pw.x and find essentially the same results). I have installed the QE software version 4.2.1 using both gfortran and ifort, using entirely internal libraries and with MKL (32 bit) libraries, both openMPI and single processor mode, and the tests of pw.x all conclude successfully). The software is installed on a Mac OSX 10.6 laptop. I have compiled in 32 and 64 bit mode for gfortran. If memory serves me correctly, all produce essentially the same results.</div><div><br></div><div>Though I have generated a number of discrepancies, I think the most simple manifestation is the difference in total energy per atom for the equilibrium structures as predicted by VASP. More specifically, I find:</div><div style="font-size: 13px; "><br></div><div><div>(1) Diamond cubic structure in VASP. Relaxed structure (pressure ~ 0). K=Point mesh is 12 x 12 x 12, symmetrized. The lattice parameter comes out to be 5.456 Ang. Ultrasoft pseudopotential, PW91 exchange-correlation, cutoff chosen to be 400 eV. Total energy of relaxed cell = -10.866660 eV. </div><div><br></div><div>(2) Diamond cubic structure run using beta-Sn POSCAR with c/a = sqrt(2). Produces the same lattice parameter and the same total energy per relaxed cell = -10.866649 eV.</div><div><br></div><div>(3) Beta-Sn structure also relaxed using the same potential yields a lattice parameter a = 4.832 Ang, and the ratio c/a = 0.547. Using the same k-point mesh (perhaps not completely converged, but close enough for our purposes) one finds a total energy of this phase to be -10.191998 eV.</div><div><br></div><div>(4) Using these numbers, and noting that there are two atoms per unit cell, one finds that the energy difference per atom between the structures is given by</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>E(BetaSn Si)- E(DC Si) <span class="Apple-tab-span" style="white-space:pre"> </span>= (-10.191998+10.866660)/2 eV/atom</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>= 0.34 eV/atom</div><div><br></div><div>This value is in very good agreement with published results. We can now take these structural parameters and run pw.x from QE to compute the energy difference. I will use the beta-Sn unit cell for both calculations (the natural diamond cubic cell produces essentially the same results). I also used a MP grid 12x12x12 and an energy cutoff of 400 eV (12*400 eV for the charge density).</div><div><br></div><div>(5) The total energy of the beta-Sn structure, per atom, predicted by QE is -130.192 eV. The total energy of the dc structure, per atom, predicted by QE is -130.982 eV. The difference in energy is, therefore, 0.79 eV/atom, more than a factor of two larger than predicted by VASP. I've attached the input files and the output files for your perusal.</div></div><div><br></div><div>Sorry about the size of this message. Any insight you can lend will be more than appreciated.</div><div><br></div><div>Thank you,</div><div><br></div><div><div>
<span class="Apple-style-span" style="font-size: 12px; "><div style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><span class="Apple-style-span" style="border-collapse: separate; -webkit-border-horizontal-spacing: 0px; -webkit-border-vertical-spacing: 0px; color: rgb(0, 0, 0); font-family: Helvetica; font-size: 12px; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; -webkit-text-decorations-in-effect: none; text-indent: 0px; -webkit-text-size-adjust: auto; text-transform: none; orphans: 2; white-space: normal; widows: 2; word-spacing: 0px; "><div>Daryl C. Chrzan</div><div>Professor, Materials Science and Engineering</div><div>University of California</div><div>Berkeley, CA 94720</div><div><div><br></div><div>and</div><div><br></div><div>Materials Sciences Division</div><div>Lawrence Berkeley National Laboratory</div><div>Berkeley, CA 94720</div><div><br></div></div><div>ph./fax: 510 643 1624</div><div><br class="khtml-block-placeholder"></div></span></div><br class="Apple-interchange-newline"></span></div></div></body></html>