Happy to help Miguel. -Salvador
________________________________ From: [email protected] <[email protected]> on behalf of Miguel Ruphuy <[email protected]> Sent: Sunday, January 17, 2016 2:55 PM To: [email protected] Subject: Re: [SIESTA-L] Problem with pseudopotential I think Yes, I already did it with RPBE... I'll try the rest and see how it goes... Thanks so much!!! I really appreciate it Miguel 2016-01-17 15:51 GMT-05:00 Salvador Barraza-Lopez <[email protected]<mailto:[email protected]>>: Hola Miguel, Please accept my apologies. On the basis part: Yes, I meant trying a TZP one. That may require to add zeros, on an "explicit" PAO.Basis block... I don't remember the details at the moment. But indeed, you may want to increase basis extent towards the vacuum: that is essentially the main idea. It would be nice if your quantity improves as a result of this. I also was wondering if you add the flag to the INP files so that the pseudo is RPBE, as you indicate on your input file: Did you do that part already? It appears so, but I was curious. Cheers! -Salvador ________________________________ From: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> on behalf of Miguel Ruphuy <[email protected]<mailto:[email protected]>> Sent: Sunday, January 17, 2016 2:38 PM To: [email protected]<mailto:[email protected]> Subject: Re: [SIESTA-L] Problem with pseudopotential I think Hola Salvador, I understood that I didn't need to check for those things in this case. I was asking to learn for next time in case I need to do it in the future. But that's ok... Now, going back to the problem. When you say to free the PAO blocks you mean to go from using DZP to use TZP instead, right? Thanks, Miguel 2016-01-17 15:15 GMT-05:00 Salvador Barraza-Lopez <[email protected]<mailto:[email protected]>>: Hola Miguel, The analysis I suggested is provided in the paper already, I meant to say that when you work with a new pseudo later on, you carry these checks before embarking on your calculations. The main difference between SIESTA and other codes concerns the expansion of wavefunctions: In order to reach the precision you need, it may be necessary to enlarge your basis set, or to work with it otherwise. There is a paper by Ordejon and coworkers where an analysis of the basis set at a metal-vacuum interface is provided... it is cited in the paper, it is a PRB from about 2008 if I am not mistaken. In a few words, the discrepancy seems to be a matter of precision now 1) Work with the basis set: Free the PAO blocks to begin with. Explore a TZP basis size. 2) Increase the k-points until your total energy converges 3) Increase mesh cutoff Finally, take the INP files, and add the flag so that they are recompiled with the RPBE flavor. Best regards, -Salvador ________________________________ From: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> on behalf of Miguel Ruphuy <[email protected]<mailto:[email protected]>> Sent: Sunday, January 17, 2016 1:11 PM To: [email protected]<mailto:[email protected]> Subject: Re: [SIESTA-L] Problem with pseudopotential I think Dear Salvador, Thanks for the reply. I tried the Pseudopotentials you sent me and I got -0.47 which is still far from the -0.33. With respect to the part where you say "When you contrast pseudos, do make sure lattice constants and band structures match all-elecron (pseudopotential-free) values." I am wondering how? I don't how to make sure the lattice constants match? Do you mean that the lattice constant I use in Siesta match to a lattice constant from a .psf file? if so, where do you see the lattice constant in the psf file? For the band structures, I am guessing that you do a simulation with one electron in Siesta, right?. how do you get the all electron band structure plot? Please find below the fdf code in case there is another issue. Thanks so much, Miguel SystemName PtH SystemLabel PtH NumberOfAtoms 13 NumberOfSpecies 2 %block ChemicalSpeciesLabel 1 78 Pt 2 1 H %endblock ChemicalSpeciesLabel PAO.BasisSize DZP PAO.BasisType split PAO.EnergyShift 0.0001 Ry XC.functional GGA XC.authors RPBE LatticeConstant 1.0000 Ang %block LatticeVectors 5.685138521 0 0 2.84256926 4.923474383 0 0 0 41.1488 %endblock LatticeVectors #41.1488 MeshCutoff 200 Ry # # MaxSCFIterations 300 DM.MixingWeight 0.1 DM.NumberPulay 10 DM.Tolerance 1.d-4 SolutionMethod diagon OccupationFunction FD #Fermi ElectronicTemperature 300 K pp # Temp. for Fermi smearing DivideAndConquer true # WriteDenchar true # The HOMO is state 844 # States from 640 to 920 # %block WaveFuncKPoints # 0.000 0.000 0.000 from 640 to 920 # %endblock WaveFuncKPoints %block kgrid_Monkhorst_Pack 4 0 0 0.5 0 4 0 0.5 0 0 1 0.5 %endblock Kgrid_Monkhorst_Pack MD.TypeOfRun CG MD.MaxForceTol 0.01 eV/Ang MD.NumCGsteps 500 #SaveElectrostaticPotential true SaveHS true DM.UseSaveDM true ZM.UnitsLength Ang %block Zmatrix cartesian 1 0.00000 0.00000 4.64190 1 1 1 1 2.84257 0.00000 4.64190 1 1 1 1 1.42128 2.46174 4.64190 1 1 1 1 4.26385 2.46174 4.64190 1 1 1 1 1.42128 0.82058 2.32095 0 0 0 1 4.26385 0.82058 2.32095 0 0 0 1 2.84257 3.28232 2.32095 0 0 0 1 5.68514 3.28232 2.32095 0 0 0 1 2.84257 1.64116 0.00000 0 0 0 1 5.68514 1.64116 0.00000 0 0 0 1 4.26385 4.10290 0.00000 0 0 0 1 7.10642 4.10290 0.00000 0 0 0 2 2.84971775 1.36734714 6.77649 1 1 1 %endblock Zmatrix 2016-01-16 12:44 GMT-05:00 Salvador Barraza-Lopez <[email protected]<mailto:[email protected]>>: Dear Miguel, Please have a look at Figures 1 and 6 --both dealing with bulk Pt-- on the following paper: http://www.sciencedirect.com/science/article/pii/S0927025614007940<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.sciencedirect.com_science_article_pii_S0927025614007940&d=BQMFaQ&c=JL-fUnQvtjNLb7dA39cQUcqmjBVITE8MbOdX7Lx6ge8&r=n_Y76F1vumEs9EYNHN2gzA5FD9jzyPhrzl3eOzxCHIQ&m=lpSzS1apfAYaHPmBQ7Vpu7cy-bY4tMFfBYXQ4yiCfLc&s=Jpv1Jz_vixhVcKrGDzU9BBmBLij7u_yHjmKhsEDQtFo&e=> All pseudos discussed on that paper are available as a zip file free of charge at: http://www.sciencedirect.com/science/article/pii/S2352340914000353<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.sciencedirect.com_science_article_pii_S2352340914000353&d=BQMFaQ&c=JL-fUnQvtjNLb7dA39cQUcqmjBVITE8MbOdX7Lx6ge8&r=n_Y76F1vumEs9EYNHN2gzA5FD9jzyPhrzl3eOzxCHIQ&m=lpSzS1apfAYaHPmBQ7Vpu7cy-bY4tMFfBYXQ4yiCfLc&s=spRgEonQZ8979MbpXReoVBr7YCOYLmsplA5-lGLkDiU&e=> When you contrast pseudos, do make sure lattice constants and band structures match all-elecron (pseudopotential-free) values. Best regards, -Salvador ________________________________ From: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> on behalf of Miguel Ruphuy <[email protected]<mailto:[email protected]>> Sent: Saturday, January 16, 2016 9:47 AM To: [email protected]<mailto:[email protected]> Subject: [SIESTA-L] Problem with pseudopotential I think Hi, So, I am new at this. So, I am trying to reproduce the results of the following paper: doi: 10.1149/1.1856988 (J. K. Nørskova et al. ,Trends in the Exchange Current for Hydrogen Evolution) So, I got a decent result for Au. For the adsorption energy of hydrogen (at 0.25ML coverage) I got 0.20 whereas they got 0.21. Now, I tried again for Pt (at 0.25ML coverage) and I got -21.37. (It should be -0.33 according to the paper). So, I though it was the pseudo potential. I found two other pseudopotentials in internet and I got -0.47 and -0.51 respectively. So, I was wondering how can I get a pseudopotential that works? In the paper they used ultra-soft pseudopotentials, but I think it doesn't work on Siesta right? Thanks, Miguel
