Fermi-Dirac smearing and magnetism

[mpol]

Dear CP2k users,

We try to model a semiconducting material (the gap is close to zero) with CP2k.

The system has an even number of electrons and DFT-mGGA (TPSS functional with SR-MOLOPT-GTH BS

and GTH pseudopotentials for PBE functional) is used to describe the electronic structure.

It is practically impossible to converge the wavefunction of the system without smearing and thus

Fermi-Dirac smearing is used. And here an interesting issue arises: when
the smearing with electronic temperature of 1 K or 50 K is used, the
system is paramagnetic. However when the smearing with electronic
temperature of 300 K is used, the system becomes diamagnetic.

May it be an artifact caused by smearing (so the system is truly
paramagnetic at the room temperature) or it may be an evidence that the
quenching of the magnetic moment occurs and the system is diamagnetic
at the room temperature?

With best regards,

MP

P.S. An example input file:

&FORCE_EVAL
  METHOD QS
  &DFT
    BASIS_SET_FILE_NAME [path]
    POTENTIAL_FILE_NAME [path]
    &MGRID
      NGRIDS 5
      CUTOFF 600
      REL_CUTOFF 60
    &END MGRID
    &PRINT
      &MO
	ADD_LAST SYMBOLIC
	COMMON_ITERATION_LEVELS 2000
	EIGENVALUES
	NDIGITS 8
	OCCUPATION_NUMBERS
	&EACH
	  GEO_OPT 1
	  QS_SCF 0
	&END EACH
      &END MO
    &END PRINT
     UKS .TRUE.
     MULTIP 1
    &QS
      EPS_DEFAULT 1.0E-14
      METHOD GPW
    &END QS
    &POISSON
	PERIODIC XY
	PSOLVER ANALYTIC
    &END POISSON
    &SCF
      EPS_EIGVAL 1.0E-10
      SCF_GUESS RESTART
      EPS_SCF 5.0E-06
      MAX_SCF 300
      &DIAGONALIZATION
        ALGORITHM STANDARD
        MAX_ITER 5
      &END DIAGONALIZATION
      &MIXING
	METHOD BROYDEN_MIXING
	NBUFFER 8
	ALPHA 0.2
	BETA 1.0
      &END MIXING
      &SMEAR
	METHOD FERMI_DIRAC
	ELECTRONIC_TEMPERATURE 300.0
      &END SMEAR
      ADDED_MOS 32
    &END SCF
    &XC
      &XC_FUNCTIONAL TPSS
      &END XC_FUNCTIONAL
      &vdW_POTENTIAL
         DISPERSION_FUNCTIONAL PAIR_POTENTIAL
         &PAIR_POTENTIAL
            TYPE DFTD3
            PARAMETER_FILE_NAME [path]
            REFERENCE_FUNCTIONAL TPSS
            CALCULATE_C9_TERM .TRUE.
            &PRINT_DFTD SILENT
	      COMMON_ITERATION_LEVELS 100
            &END
         &END PAIR_POTENTIAL
      &END vdW_POTENTIAL
    &END XC
  &END DFT
  &SUBSYS
    &CELL
    PERIODIC XY
      ABC 19.64800 51.04800 16.00000
       &END CELL
    &TOPOLOGY
	&CENTER_COORDINATES .FALSE.
	&END
    &END TOPOLOGY
    &COORD
[coords]
    &END COORD
[KIND sections with BSs and PPs specified in the message]
  &END SUBSYS
&END FORCE_EVAL
&GLOBAL
  RUN_TYPE GEO_OPT
  PRINT_LEVEL MEDIUM
  &TIMINGS
     THRESHOLD 0.000001
  &END
  &DBCSR
    MM_DRIVER BLAS
  &END
&END GLOBAL
&MOTION
    &GEO_OPT
	MAX_ITER 2000
    &END
&END

[Marcella Iannuzzi]

Dear MP,

By increasing the temperature of the Fermi-Dirac smearing, more states get occupied and a different redistribution of electrons between the two spin channels is possible. I wouldn't call it an artefact, assuming that the distribution of states provided at this level of theory can be considered correct.

Just plot the PDOS at the different smearing temperatures to identify which are the responsible states for the behaviour that you are observing.

Kind regards,

Marcella

[mpol]

Dear Marcella,

Thank  you for your suggestion.

With best regards,

MP