I have been trying to study defects in oxypnictides using QE. Whilst it is easy to relax the pristine oxypnictide structure, the defective structure does not converge, i.e. convergence accuracy never falls under 1e-5 I have been trying: - different smearing methods and degauss values - increasing number of bands - decreasing mixing beta - different mixing modes - different ferrum pseudopotentials - slightly different initial ionic positions - possibility of LDA+U
Would very much appreciate your suggestios Quoting below the input file I use: QUOTE &control calculation='relax', prefix='defective supercell', nstep=20, etot_conv_thr=1.d-4, forc_conv_thr=1.d-3 pseudo_dir='./pseudo' / &system ibrav = 0, nat=24, ntyp=5, nbnd=500, occupations='smearing', smearing='m-v', degauss=0.011, ecutwfc = 80, !lda_plus_u=.true., Hubbard_U(4)=4.3 / &electrons mixing_beta=0.3, conv_thr=1.d-6 mixing_mode='local-TF' mixing_ndim=12 electron_maxstep=1000 / &ions / ATOMIC_SPECIES La 138.90547 La.pbe-mt_fhi.UPF O 15.999 O.pbe-mt.UPF !F 18.998403 F.pbe-hgh.UPF H 1.00794 H.pbe-mt_fhi.UPF Fe 55.845 Fe.pbe-sp-mt_gipaw.UPF !Fe.pbe-mt_fhi.UPF !Fe.pbe-sp-mt_gipaw.UPF As 74.921596 As.pbe-mt_fhi.UPF CELL_PARAMETERS angstrom 11.97 0.00 0.00 0.00 3.991600575 0.00 0.00 0.00 8.636372966 ATOMIC_POSITIONS crystal La 0.084333 0.251 0.144227 La 0.417667 0.251 0.144227 La 0.751 0.251 0.144227 La 0.251 0.751 0.857773 La 0.584333 0.751 0.857773 La 0.917667 0.751 0.857773 O 0.251 0.251 0.001 H 0.584333 0.251 0.001 O 0.917667 0.251 0.001 O 0.084333 0.751 0.001 O 0.417667 0.751 0.001 O 0.751 0.751 0.001 Fe 0.251 0.251 0.501 Fe 0.584333 0.251 0.501 Fe 0.917667 0.251 0.501 Fe 0.084333 0.751 0.501 Fe 0.417667 0.751 0.501 Fe 0.751 0.751 0.501 As 0.084333 0.251 0.648885 As 0.417667 0.251 0.648885 As 0.751 0.251 0.648885 As 0.251 0.751 0.353115 As 0.584333 0.751 0.353115 As 0.917667 0.751 0.353115 K_POINTS automatic 1 1 1 0 0 0 UNQUOTE -- Eleftheria Gkogkosi Applied Mathematics & Physics Dept. National Technical University of Athens http://users.ntua.gr/elefthe _______________________________________________ Pw_forum mailing list [email protected] http://pwscf.org/mailman/listinfo/pw_forum
