Dear Nicola, Yes, indium is metallic. The parameters I use to treat it are: occupations = 'smearing', smearing = 'cold', degauss = 0.02,
There are 13 valence electrons in indium and 34 Kohn-Sham states. I have checked the band occupations for all the k-points; at least the two uppermost bands are empty, in most case even more. Here are the results of my tests. Exp. : 3.2449 x 3.2449 x 4.9420 angstroms -------------------------------------------------------------------------- K-points | Ecut/Ry | Ecutrho/Ry | PSP | Fntal | cell parameters -------------------------------------------------------------------------- 8x8x8=40 120 1200 PAW nrel PBE 3.2168 x 5.3095 8x8x8=40 150 600 NC nrel PBE 3.2213 x 5.3194 8x8x8=40 50 500 PAW rel+SO PZ 3.0786 x 5.0095 8x8x8=40 120 1200 PAW rel+SO PZ 3.1032 x 5.0938 12x12x12=126 50 500 PAW rel+SO PZ 3.1692 x 4.7429 16x16x16=288 50 500 PAW rel+SO PZ 3.1340 x 4.8412 12x12x12=126 120 1200 PAW rel+SO PZ 3.2094 x 4.7811 16x16x16=288 120 1200 PAW rel+SO PZ 3.1647 x 4.9076 The best result is the last one. So the PAW of In is very hard (!) and need a huge amount of k-points… This is the first time I see this problem with a PAW PSP of the library… Thank you all for your hints. Pascal Pascal Boulet — Professor in computational chemistry - DEPARTEMENT OF CHEMISTRY Aix-Marseille University - ST JEROME - Avenue Escadrille Normandie Niemen - F-13013 Marseille - FRANCE Tél: +33(0)4 13 55 18 10 - Fax : +33(0)4 13 55 18 50 Site : http://madirel.univ-amu.fr/pages_web_BOULET_PASCAL/infos - Email : [email protected] Le 7 déc. 2017 à 17:35, Nicola Marzari <[email protected]> a écrit : > > Thanks Pascal - this is metallic, > and you are using smearing, correct? Something like 0.02 Ry, and > Marzari-Vanderbilt cold smearing? > > Also, how many electrons does your pseudo have? Do you have enough bands > (i.e. are the top ones empty)? > > nicola > > > > On 07/12/2017 14:23, Pascal Boulet wrote: >> Dear Nicola, >> You are right I also noticed that the ‘pseudo' state energies match very >> well the AE ones. >> Regarding the alloys Vegard’s law should hold according to experimental >> data. We are indeed using supercells. >> I am currently running tests on pure indium. As mentioned by Paolo, the mass >> is irrelevant in the results for static calculations (as expected!). I have >> preliminary results with relativistic+SO calculation that are a bit better: >> 3.0786 x 3.0786 x 5.0095 >> Recall: >> exp.: 3.2449 x 3.2449 x 4.9420 >> previous calc.: 3.2168 x 3.2168 x 5.3095 >> This is obtained with Ecut=50 Ry Ecutrho=500 Ry, 8x8x8 kpts=40 kpts >> With these parameters c improves substantially (1.4% off) but a and c worsen >> a bit (5% off). >> I am running a test with more kpts first. In a second step I will increase >> Ecut and Ecutrho and will report to the forum. >> Pascal >> Pascal Boulet >> — >> /Professor in computational chemistry - DEPARTEMENT OF CHEMISTRY/ >> Aix-Marseille University - ST JEROME - Avenue Escadrille Normandie Niemen - >> F-13013 Marseille - FRANCE >> Tél: +33(0)4 13 55 18 10 - Fax : +33(0)4 13 55 18 50 >> Site : http://madirel.univ-amu.fr/pages_web_BOULET_PASCAL/infos - Email : >> [email protected] <mailto:[email protected]> >> Le 6 déc. 2017 à 23:24, Nicola Marzari <[email protected] >> <mailto:[email protected]>> a écrit : >>> >>> Dear Pascal, >>> >>> >>> the comparison with all-electron calculations looks good to me - see >>> here for tests and for the other pseudos: >>> http://materialscloud.org/sssp/results/In_conv_patt.png >>> >>> Comparing with expts is more delicate - c/a with PBE is >>> quite a bit off e.g. in ferroelectric perovskites. >>> >>> Re the alloy, how are you constructing the intermediate concentrations? >>> Supercells? (Vegard law should hold) >>> >>> nicola >>> >>> On 06/12/2017 23:03, Paolo Giannozzi wrote: >>>> You mean: the value 49.0 here? >>>> ATOMIC_SPECIES >>>> In 49.0 In.pbe-dn-kjpaw_psl.1.0.0.UPF >>>> Iit is used only by molecular dynamics, not by structural optimization. >>>> >>>> I don't remember In to be an especially nasty element, at least for >>>> simple III-V compounds like InAs. I would examine the effect of a denser >>>> k-point grid, and of spin-orbit >>>> >>>> Paolo >>>> >>>> >>>> On Wed, Dec 6, 2017 at 10:46 PM, Saif Ullah <[email protected] >>>> <mailto:[email protected]>> wrote: >>>> >>>> Dear Boulet, >>>> >>>> I think I found where the error is. You are using the atomic number >>>> of In instead of its atomic mass. >>>> >>>> Regards >>>> Saif >>>> Department of physics, UFJF, Brazil >>>> >>>> >>>> >>>> On Wed, Dec 6, 2017 at 6:45 PM, Pascal Boulet >>>> <[email protected] <mailto:[email protected]>> wrote: >>>> >>>> Thanks Lorenzo for the suggestion. I have tried with Ecut=120 Ry >>>> and Ecutrho=1200 Ry. I get a slightly better result (but c is >>>> still very bad): >>>> 3.2168 x 3.2168 x 5.3095 >>>> >>>> as compared with: >>>> exp.: 3.2449 x 3.2449 x 4.9420 angström >>>> previous calc.: 3.2715 x 3.2715 x 5.4541 angström >>>> >>>> >>>> Best, >>>> >>>> Pascal Boulet >>>> — >>>> /Professor in computational chemistry - DEPARTEMENT OF CHEMISTRY/ >>>> Aix-Marseille University - ST JEROME - Avenue Escadrille >>>> Normandie Niemen - F-13013 Marseille - FRANCE >>>> Tél: +33(0)4 13 55 18 10 <tel:+33%204%2013%2055%2018%2010> - Fax >>>> : +33(0)4 13 55 18 50 <tel:+33%204%2013%2055%2018%2050> >>>> Site : http://madirel.univ-amu.fr/pages_web_BOULET_PASCAL/infos >>>> <http://madirel.univ-amu.fr/pages_web_BOULET_PASCAL/infos> - >>>> Email : [email protected] <mailto:[email protected]> >>>> >>>> >>>> >>>> >>>> Le 6 déc. 2017 à 18:54, Lorenzo Paulatto <[email protected] >>>> <mailto:[email protected]>> a écrit : >>>> >>>>> Hello, >>>>> can you try with a much higher cutoff (100 Ry or more) and see >>>>> if the lattice parameter improves? I've noticed that sometimes >>>>> pslibrary pseudos are much a harder than they look, especially >>>>> when converging stress >>>>> >>>>> Cheers >>>>> >>>>> -- >>>>> Lorenzo Paulatto >>>>> Written on a virtual keyboard with real fingers >>>>> >>>>> On Dec 6, 2017 18:32, "Pascal Boulet" >>>>> <[email protected] <mailto:[email protected]>> >>>>> wrote: >>>>> >>>>> Dear all, >>>>> >>>>> We are working on some In-containing semiconductors >>>>> M(1-x)In(x)L. M is supposed to be a smaller atom than In. >>>>> We are facing the problem that when we vary x from 0 to 1 >>>>> the cell parameters is decreasing instead of increasing. >>>>> Although this may happen, it should not be the case here >>>>> since ‘pure’ ML has a smaller cell parameter than ‘pure’ InL. >>>>> >>>>> We have run some calculations on crystal indium using the >>>>> PSL library pseudo potentials (both PAW and USPP). one of >>>>> the calculated cell parameters are miles away from the >>>>> experimental ones: >>>>> exp.: 3.2449 x 3.2449 x 4.9420 angström >>>>> calc.: 3.2715 x 3.2715 x 5.4541 angström >>>>> >>>>> Does anyone have tested the In pseudo potentials of the >>>>> PSL library or faced this problem with indium? >>>>> >>>>> Thank you for your help and time. >>>>> >>>>> Best regards >>>>> >>>>> PS1. here are some information from the input file: >>>>> >>>>> calculation = 'vc-relax’, >>>>> etot_conv_thr = 1d-7, >>>>> forc_conv_thr = 5d-5, >>>>> ibrav = 0, >>>>> nat = 2, >>>>> ntyp = 1, >>>>> ecutwfc = 45.d0, >>>>> ecutrho = 450.d0, >>>>> nbnd = 18 >>>>> &ELECTRONS >>>>> conv_thr = 1d-8, >>>>> &CELL >>>>> press_conv_thr = 0.01d0, >>>>> cell_factor = 1.5, >>>>> ATOMIC_SPECIES >>>>> In 49.0 In.pbe-dn-kjpaw_psl.1.0.0.UPF >>>>> K_POINTS automatic >>>>> 8 8 4 1 1 1 >>>>> CELL_PARAMETERS angstroms >>>>> 3.2449 0. 0. >>>>> 0. 3.2449 0. >>>>> 0. 0. 4.9420 >>>>> ATOMIC_POSITIONS crystal >>>>> In 0.00000000 0.00000000 0.00000000 >>>>> In 0.50000000 0.50000000 0.50000000 >>>>> >>>>> PS2. And some from the output: >>>>> >>>>> bfgs converged in 18 scf cycles and 15 bfgs steps >>>>> (criteria: energy < 1.0E-07 Ry, force < >>>>> 5.0E-05Ry/Bohr, cell < 1.0E-02kbar) >>>>> CELL_PARAMETERS (angstrom) >>>>> 3.271466870 0.000000000 0.000000000 >>>>> 0.000000000 3.271466870 0.000000000 >>>>> 0.000000000 0.000000000 5.454067219 >>>>> ! total energy = -821.46750731 Ry >>>>> total stress (Ry/bohr**3) (kbar) >>>>> P= 0.30 >>>>> >>>>> Pascal Boulet >>>>> — >>>>> /Professor in computational chemistry - DEPARTEMENT OF >>>>> CHEMISTRY/ >>>>> Aix-Marseille University - ST JEROME - Avenue Escadrille >>>>> Normandie Niemen - F-13013 Marseille - FRANCE >>>>> Tél: +33(0)4 13 55 18 10 <tel:04%2013%2055%2018%2010> - >>>>> Fax : +33(0)4 13 55 18 50 <tel:04%2013%2055%2018%2050> >>>>> Site : >>>>> http://madirel.univ-amu.fr/pages_web_BOULET_PASCAL/infos >>>>> <http://madirel.univ-amu.fr/pages_web_BOULET_PASCAL/infos> - >>>>> Email : [email protected] >>>>> <mailto:[email protected]> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Pw_forum mailing list >>>>> [email protected] <mailto:[email protected]> >>>>> http://pwscf.org/mailman/listinfo/pw_forum >>>>> <http://pwscf.org/mailman/listinfo/pw_forum> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Pw_forum mailing list >>>>> [email protected] <mailto:[email protected]> >>>>> http://pwscf.org/mailman/listinfo/pw_forum >>>>> <http://pwscf.org/mailman/listinfo/pw_forum> >>>> >>>> >>>> _______________________________________________ >>>> Pw_forum mailing list >>>> [email protected] <mailto:[email protected]> >>>> http://pwscf.org/mailman/listinfo/pw_forum >>>> <http://pwscf.org/mailman/listinfo/pw_forum> >>>> >>>> >>>> >>>> _______________________________________________ >>>> Pw_forum mailing list >>>> [email protected] <mailto:[email protected]> >>>> http://pwscf.org/mailman/listinfo/pw_forum >>>> <http://pwscf.org/mailman/listinfo/pw_forum> >>>> >>>> >>>> >>>> >>>> -- >>>> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, >>>> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy >>>> Phone +39-0432-558216, fax +39-0432-558222 >>>> >>>> >>>> >>>> _______________________________________________ >>>> Pw_forum mailing list >>>> [email protected] >>>> http://pwscf.org/mailman/listinfo/pw_forum >>>> >>> >>> >>> -- >>> ---------------------------------------------------------------------- >>> Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL >>> Director, National Centre for Competence in Research NCCR MARVEL, EPFL >>> http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project >>> _______________________________________________ >>> Pw_forum mailing list >>> [email protected] >>> http://pwscf.org/mailman/listinfo/pw_forum > > > -- > ---------------------------------------------------------------------- > Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL > Director, National Centre for Competence in Research NCCR MARVEL, EPFL > http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project
_______________________________________________ Pw_forum mailing list [email protected] http://pwscf.org/mailman/listinfo/pw_forum
