I have a transition metal-anion system with slightly distorted octahedra. I want to be able to set the z-axis and x-axis of the local rotation matrix so that it points (approximately) along the bonds of the octahedra. Essentially this is because even with the slight distortion I still expect to see a fairly clear t2g to eg splitting. Is it possible to do this in Elk? I could do it in Wien2k with the QTL program.
you could rotate your crystal structure (unit cell) such that it aligns the way you like with the cartesian axes. Usually, this will break many symmetries, so this is not really efficient.
Sorry, I think there was an option (not in manual)
one can tune the direction of z-axis just by
flag. it takes 3 real numbers
Thanks for the responses.
I'll need to set both the z-axis and x-axis to align with the octahedra. Can I also set the x-axis? What are the units of the three number? Cartesian? Units of lattice vectors?
Oh, sorry, I did not get, the first question properly and maybe still dont understand it.. happens))
you probably want to see that splitting using the DOS plots for spin-polaraised calculation?
with 'sqados' you choose the z-axis for the spin-up and spin-down channel in dos plots.
it takes lattice coordinetes as input.
The thing that it is cheap and if it will help to see that splitting - good, if not- whatever
Ohh, it is the crystall field splitting…
nothing to do with spin, 'sqados' will not help, sorry
Thanks again for the reply. But I'm not sure what spin polarisation has to do with this. The t2g to eg splitting is a result of the crystal field effects. The nuts and bolts is that I want to align my z-axis and x-axis with the bonds of the octahedra. Then the t2g will simply be the dxy, dxz and dyz. The eg will be the dx2-y2 and dz2.
But if I can't align my axes it is very difficult to identify these.
Yeah, I already got the point..
the best way to do it - what Markus said, and it indeed will increase the cost of calculation.
but as soon as your calculations are without spin-orbit couplings, and your system is not 100 atoms, you will not suffer too much)
I think elk choose that quantisation axis with respective to Cartesian coordinates, if it is like that (I also want to know) you can just try to rotate lattice vectors ('avec'), till the cartesian basis, xyz, more or less coincide with octahedral axis, It should be enough to see that splittings.
but I never heard about introducing local octahedral axis in the ELK, most probably it is not implemented there yet
I think you don't have to change the axis manually due to the way things are implemented in Elk, in principle. For example, I have tried the following code to print the dx2-y2 character
bc(dx2-y2) = 0.5 * dble( dmat(idxlm(2,-2), idxlm(2,-2))
+ dmat(idxlm(2,-2), idxlm(2,2))
+ dmat(idxlm(2,2), idxlm(2,-2))
+ dmat(idxlm(2,2), idxlm(2,2)) )
And this seems to work when I change the crystal axes or tilting angles. (See dos.f90 file for more info. You also need to set lmirep to .false.)
However, I don't quite understand how this works. Presumably, the Y_lm's are symmetrized according to the symmetry of the site (like stars for planewaves), and this picks out the correct z and x axes. But I could not find where this is done in the code.
Perhaps Lars or Kay would elaborate on this. I would be very grateful!
Thanks for the useful information. For me I think the chief problem is that the site is not "perfectly" octahedrally coordinated. The only site symmetry operation for the transition metal ion is the identity. Therefore the local rotation axes for the Y_lm basis on the transition metal site will not align with a sensible geometric angle with respect to the bonds.
Any input from developers would be welcome.
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.