noncollinear-spin functionality

[baizq]

Dear ATK developers，

May I know some background/theory stuff of the newly implemented noncollinear-spin funtionality? Say, the original papers you refer to about the noncollinear DFT. We are currently writing a paper and would like to include a few lines of simple description of the theory..but there is no tutorial for the functionality so far...

Thanks in advance!

baizq

[zh]

1. V M García-Suárez et al 2004 J. Phys.: Condens. Matter 16 5453 doi:10.1088/0953-8984/16/30/008
2.  V. M. García-Suárez, et al, The European Physical Journal B,  2004, Volume 40, Issue 4, pp 371-377,   http://link.springer.com/article/10.1140%2Fepjb%2Fe2004-00265-y
3. José M Soler et al 2002 J. Phys.: Condens. Matter 14 2745 doi:10.1088/0953-8984/14/11/302

[Anders Blom]

Try

G. Theurich and N. A. Hill, Phys. Rev. B 64, 073106 (2001)
T. Oda, A. Pasquarello, and R. Car, Phys. Rev. Lett. 80, 3622 (1998)
Ph. Kurz, F. Forster, L. Nordstrom, G, Bihlmayer, and S. Blugel, Phys. Rev. B 69, 024415 (2004).

and references therein.

[baizq]

Thanks a lot for both of your replies.

I have read the references you recommended. I have a question here. For a magnetic tunnel junction, the Fe/MgO/Fe sandwich for example, is the global spin-quantization axis always confined in the transverse plane perpendicular to the transport direction? If I align all the spin pointing along z-axis (transport direction), would the calculated conduction the same as that obtained from collinear calculation (parallel)?

By the way, can we extract the k-dependent transmission spectrum from a noncollinear transport calculation(parallel, antiparallel), just as what we do for the collinear case (parallel majority/minority, antiparallel)?

[zh]

It may depends on the magnetic anisotropy of electrode. If the electrode has large magnetic anisotropy, the setup of spin parallel or perpendicular to the transport would result in different transport properties, otherwise the effect may be negligible.

[Anders Blom]

Just a note on the coordinate system. For noncollinear spin, when you specify the spin direction through InitialSpin, the angles theta and phi are the polar and azimuthal angles, respectively, in a coordinate system corresponding to the "usual" orientation, i.e. where theta=0 is the positive Z axis. The collinear case is thus indeed obtained when theta=0 (or pi).

As for the k-point dependence, yes, it's still possible. Just note that all quantities are obtained as 2x2 matrices, instead of just up/down components.