Messages

1

General Questions and Answers / potential averaged over the B direction for a fixed A coordinate

« on: September 8, 2013, 17:38 »

Dear ATK developers and all,

May I know how to extract potential averaged over the B direction for a fixed A coordinate (A = 18 Angstrom) as you ploted in FIG. 1. and FIG. 5. (and also the Mulliken charge) in PHYSICAL REVIEW B 85, 165442 (2012). I would appreciate your help.

Best,
leslie

2

General Questions and Answers / Re: About charged electrodes

« on: January 19, 2013, 20:11 »

For devices it's per electrode unit cell. Also note that -1 means one electron extra, so it's not "in units of electrons" but "in units of electron charge", and thus in light of my previous statement about units we should probably have made it "charge=-1*e".

In your case it should be -0.09 (provided Al donates electrons and not holes).

Hi Anders,

Actually one formula unit of Co2FeSi0.97Al0.03 has 0.03 less electron than that of Co2FeSi, so in order to simulate Co2FeSi0.97Al0.03, we should subtract 0.03 electron/unit cell from Co2FeSi. And if the parameter "charge" means electron charge, it should be set as positive (-(-0.03)/unit cell * 3 unit cell= 0.09) in my case containing three formula unit of Co2FeSi0.97Al0.03 and the corresponding Fermi level would shift downwards. Am I right?

Another question is about the surface layer (in principle we must also include at least one unit cell of the electrode material in the scattering region as surface layer). The problem is that we cannot add electron to the scattering region which is an open system. So the device structure we really have is Co2FeSi0.97Al0.03/Co2FeSi/Spacer/Co2FeSi/Co2FeSi0.97Al0.03 instead of Co2FeSi0.97Al0.03/Spacer/Co2FeSi0.97Al0.03.... Is it correct?

Best,
leslie

3

General Questions and Answers / Re: About charged electrodes

« on: January 19, 2013, 08:41 »

Hi Anders,

Thanks. Is it number of electron per unit-cell of the electrode material? Or per any cell, say supercell? For example, if I use three formula unit of the heusler compound Co2FeSi as the electrode image, how should I set the parameter charge if I would like to simulate the case of Co2FeSi0.97Al0.03? Should it be -0.03? Or -0.03 multiplied by 3 as that in VASP ?

Best,
leslie

4

General Questions and Answers / Re: About charged electrodes

« on: January 18, 2013, 17:25 »

What is the unit of the parameter "charge"?  Does 0.03 mean 0.03 electron or 0.03 eV...

5

General Questions and Answers / charged electrode

« on: January 17, 2013, 09:35 »

Dear ATK developers and all,

I have noticed in the release letter of ATK ver. 12.2.2 that it is possible to use charged electrode. Is there any more detailed instruction demonstrating how to use such a function?  I would appreciate your help.


Best,
leslie

6

General Questions and Answers / Re: k-resolved transmission dos

« on: September 1, 2011, 10:26 »

Thanks, Blom! 

Yes, in my opinion it is quite useful. Wish it will appear in the new version ATK.

leslie

7

General Questions and Answers / k-resolved transmission dos

« on: September 1, 2011, 01:59 »

Dear Colleagues,

Anybody knows whether ATK can calculate the k-resolved transmission DOS? and how? I know that the software can map transmission coefficient at wave-vector space at Fermi level, but can we do the similar mapping for device transmission dos?

Thanks,
leslie

8

General Questions and Answers / Re: MTJ with ferrimagnets as electrodes

« on: June 18, 2011, 16:34 »

Dear zh,

Thanks very much.

By compensated ferromagnets, do you mean compensated ferrimagnets.....? 

leslie

9

General Questions and Answers / MTJ with ferrimagnets as electrodes

« on: June 18, 2011, 14:35 »

Dear Colleagues,

Is it possible to carry on transport calculation with ATK 2011.2 for the stacking trilayer device like Fe-MgO-Fe, but using compensated ferrimagnets as electrodes instead of ferromagnets? Will this induce any unexpected mistake?

Thanks,

leslie

10

General Questions and Answers / how to define confinement orbital as hard_well in ATK basis set ?

« on: June 7, 2011, 08:50 »

Hi friends,

In Siesta, the basis set (atomic orbitals) is generated using a confinement potential. The default for this confiment potential is Sankey Hard Well (V=0 for r<a and V=infinity for r>a) in Siesta. An alternative (and better) confinement potential is soft-confinement potential V(r)=Voexp[-(rc-ri)/(r-ri)]/(rc-r).

In ATK, however, the only choice is soft-confinement potential, which is defined in the class ConfinedOrbital
 ConfinedOrbital(
principal_quantum_number,
angular_momentum,
radial_cutoff_radius,
confinement_start_radius,
additional_charge,
confinement_strength,
radial_step_size
)

Anyone knows what to do if I want to use the Sankey Hard Well potential instead of soft-confinement potential with ATK?

11

General Questions and Answers / Re: complex band structure

« on: June 2, 2011, 15:43 »

Yes, I am also wondering how to calculate complex bandstructure with ATK...

12

General Questions and Answers / Re: how to set the initial spin for Fe/MgO/Fe MTJ

« on: June 2, 2011, 02:47 »

Thanks, the tutorial is really beneficial to me. However, it only illustrates single iron as electrode: first conduct a bulk calculation to get a good guess for the initial spin. How about the case for alloys, say Co2MnSi as electrode. Should we carry on bulk calculation for Co, Mn, Si seperately, or Co2MnSi as a whole?

13

General Questions and Answers / Re: Spintronics and MRAM in ATK

« on: June 1, 2011, 19:33 »

I am wondering how to analyze the barrier tunneling and the resonant tunneling with ATK....

14

General Questions and Answers / how to set the initial spin for Fe/MgO/Fe MTJ

« on: June 1, 2011, 08:50 »

Dear all,

Recently I am trying to follow the tutorial Fe/MgO/Fe MTJ to learn how to calculate transport properties for such structures while encontering one question:

Why the initial spin of Fe atoms at the same electrode is set to the same direction? Shouldn't there exist both spin-up and spin-down ions within iron electrode?

leslie

15

General Questions and Answers / Re: Restore scf result

« on: May 24, 2011, 02:27 »

Thanks, Anders. So we do not need to set the initial spin if it is already set before scf run, right? By nlread the .nc file after scf, the initial spin will be loaded ?