[Pw_forum] Problem with PAW vs. USP

[Giovanni Cantele]

A quick run using your input (that demonstrates that providing inputs results in quicker answers!!!) as is (so non lattice constant optimisation),
gives:

bulk Zr : -614.4200082344 Ry
isolated atom: 

where the isolated atom was reproduced using the same input with just one atom, a=12, c=15 (just more or less safe values for an isolated atom), nspin=2,
starting_magnetization(1) = 0.5, KPOINTS { gamma }. The final magnetic moment on the isolated Zr atom is 2.00.

Cohesive energy: (-614.4200082344 - 2 * -306.74916142) / 2 * 13.605692 ~ -6.27 eV per Zr atom

So, the value matches what you reported from the USPP calculations. Are you sure that in calculating with PAW the cohesive energy you are using, for the
isolated Zr atom, its PAW energy (so, did you remake the calculation for the atom and use that energy for calculating the cohesive energy)? Energy coming from
different pseudo potentials cannot be used within the same formula, because they are referred to a different zero.

Hope this helps,

    Giovanni


PS because you fix a quite strict criterion for the convergence of the scf cycle (conv_thr=1.0D-9) I stopped the calculation for the isolated atom when
estimated scf accuracy    <          5.5E-09 Ry


> On 13 Oct 2015, at 14:54, Vic Bermudez <c.h.bermudez at cox.net> wrote:
> 
> Hello,
> 
> 	I'm encountering some strange behavior in calculations for the lattice
> constants and cohesive energy of Zr metal. All calculations are being done
> using version 5.0.
> 
> (1) I first find the energy of the free Zr atom, taking care to obtain the
> proper triplet ground state of the free atom.
> 
> (2) I then optimize the (hexagonal) Zr metal lattice constants using a
> script that varies a0 over a range of values while doing vc_relax to
> optimize c0 at each value of a0. From this I find the values of a0 and c0
> that give an absolute-minimum total energy
> 
> 	When I do this using PBE with a USPP for Zr (Zr.pbe-nsp-van.UPF), which I
> obtained from the QE website, the results are very good. I find a0=3.235
> Angstroms, c0=5.148 Angstroms and a cohesive energy of 6.23 eV/Zr, all of
> which are very close to the experimental values. So far, so good.
> 
> 	When I repeat this procedure using PBE with a PAW pseudopotential
> (Zr.pbe-spn-kjpaw_psl.0.2.3.UPF), which I obtained from the THEOS website, I
> get essentially the same lattice constants, but the cohesive energy is
> wildly wrong (about 41 eV/Zr !!). I get the same results for
> ecutwfc=40/ecutrho=320 and ecutwfc=70/ecutrho=560. Except for the choice of
> PP, I'm using all the same input parameters for PAW as I did for USPP. Also,
> I get the same ridiculous cohesive energy whether I use the PAW total
> energies or the all-electron energies. Since the lattice constants are good,
> and, according to THEOS, the pseudopotential is well-tested, I must be doing
> something spectacularly wrong in obtaining or interpreting the energy
> values.
> 
> 	Any advice would be much appreciated. In case it will help, I'm inserting a
> representative input file below (for relaxing with the optimized lattice
> constants).
> 
> ******************
> 
> &CONTROL
> calculation='relax',
> restart_mode='from_scratch',
> title='single-crystal alpha-Zr - testing Zr PBE/PAW pseudopotential',
> pseudo_dir='/lustre/cmf/scratch/b/bermudez/QE_PP/',
> ! outdir='/lustre/cmf/scratch/b/bermudez/aZr/',
> etot_conv_thr=1.0D-5,
> forc_conv_thr=1.0D-4,
> nstep=10000,
> wf_collect=.TRUE.,
> verbosity='default'
> /
> 
> &SYSTEM
> ibrav=4,
> a=3.24, c=5.147,
> nat=2,
> ntyp=1,
> ecutwfc=40.0,
> ecutrho=320.0,
> occupations='smearing',
> smearing='gaussian',
> degauss=0.002
> /
> 
> &ELECTRONS
> electron_maxstep=200,
> conv_thr=1.0D-9,
> mixing_mode='plain',
> mixing_beta=0.7D0
> /
> 
> &IONS
> ion_dynamics='bfgs',
> upscale=10000.D0
> /
> 
> ATOMIC_SPECIES
> Zr 91.22   Zr.pbe-spn-kjpaw_psl.0.2.3.UPF
> 
> ATOMIC_POSITIONS {crystal}
> Zr   0.000     0.000     0.000
> Zr   0.333     0.667     0.500
> 
> K_POINTS {automatic}
> 10 10 6 0 0 0
> 
> *******************
> 
> Best Wishes,
> Vic Bermudez
> 
> Victor M. Bermudez
> E-mail: bermudez at alum.mit.edu
> 
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
> 

-- 

Giovanni Cantele, PhD
CNR-SPIN
c/o Dipartimento di Fisica
Universita' di Napoli "Federico II"
Complesso Universitario M. S. Angelo - Ed. 6
Via Cintia, I-80126, Napoli, Italy
e-mail: giovanni.cantele at spin.cnr.it
Phone: +39 081 676910
Skype contact: giocan74

ResearcherID: http://www.researcherid.com/rid/A-1951-2009
Web page: http://people.na.infn.it/~cantele