QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Hasan Sahin on January 11, 2009, 16:03

Title: unexpected peaks in IV curves
Post by: Hasan Sahin on January 11, 2009, 16:03
I calculated an IV curve and plotted spin up-spin down currents.(see attachment) There are unexpected peaks...

Calculation was tested for various parameters

1- DZP and DZ
2- various buffer atoms (from 4 to 10)
3- LDA and GGA
4- tolerance 1e-5 and 1e-6
5-  InitialDensityType.EquivalentBulk and  InitialDensityType.NeutralAtom


but these peaks still occur. What is the reason for these strange peaks ?


my parameters are:


iteration_mixing_parameters = iterationMixingParameters(
    algorithm = IterationMixing.Pulay,
    diagonal_mixing_parameter = 0.05,
    quantity = IterationMixing.Hamiltonian,
    history_steps = 30)

electron_density_parameters = electronDensityParameters(
    mesh_cutoff = 200.0*Rydberg,
    initial_scaled_spin = 10*[0]6*[0.5]+35*[0])

basis_set_parameters = basisSetParameters(
    type = DoubleZeta,
    radial_sampling_dr = 0.001*Bohr,
    energy_shift = 0.01*Rydberg,
    delta_rinn = 0.8,
    v0 = 40.0*Rydberg,
    charge = 0.0,
    split_norm = 0.15)

iteration_control_parameters = iterationControlParameters(
    tolerance = 1e-06,
    criterion = IterationControl.TotalEnergy,
    max_steps = 1000)


two_probe_algorithm_parameters = twoProbeAlgorithmParameters(
    electrode_constraint = ElectrodeConstraints.Off,
    initial_density_type = InitialDensityType.EquivalentBulk)

energy_contour_integral_parameters = energyContourIntegralParameters(
    circle_points = 30,
    integral_lower_bound = 3*Rydberg,
    fermi_line_points = 10,
    fermi_function_poles = 4,
    real_axis_infinitesimal = 0.05*electronVolt,
    real_axis_point_density = 0.02*electronVolt)

two_center_integral_parameters = twoCenterIntegralParameters(
    cutoff = 2500.0*Rydberg,
    points = 1024)

Title: Re: unexpected peaks in IV curves
Post by: Nordland on January 11, 2009, 16:51
Strange!

Is the peak(s) located at the same bias? always?

Alternative if you have the newest version of ATK, you can try the new constraint option DensityMatrix instead -
it is better at higher voltage!
Quote
two_probe_algorithm_parameters = twoProbeAlgorithmParameters(
    electrode_constraint = ElectrodeConstraints.DensityMatrix,
    initial_density_type = InitialDensityType.EquivalentBulk)
Title: Re: unexpected peaks in IV curves
Post by: Hasan Sahin on January 11, 2009, 16:56
yes peaks occur at same voltages. I also calculated the transmission at the vicinity of this points but there no any significant contribution from states.

i will try Density matrix.
Title: Re: unexpected peaks in IV curves
Post by: Hasan Sahin on January 11, 2009, 22:13
densitymatrix does not work too. :(
Title: Re: unexpected peaks in IV curves
Post by: kstokbro on January 11, 2009, 22:49
It might be that there is some problem with the scf convergence at the bias where you see the peak.

Do you use the density matrix of the  previous voltage as initial guess?

Maybe you can try to increase the voltage in smaller steps between 0.8 and 1.2 Volt.

Title: Re: unexpected peaks in IV curves
Post by: Nordland on January 12, 2009, 10:48
densitymatrix does not work too. :(

What do you mean by does not work? Same behavior? - I would go for a closer voltage scan in the area around this peak, perhaps it has some unique features in that area....
Title: Re: unexpected peaks in IV curves
Post by: Hasan Sahin on January 12, 2009, 11:24
densitymatrix does not work too. :(

What do you mean by does not work? Same behavior? - I would go for a closer voltage scan in the area around this peak, perhaps it has some unique features in that area....

yes I see the same result with densitymatrix.

Before, I had calculated this region with smaller steps and seen the same peaks. So, now I am thinking its physical meaning.