[Wien] A question about the value of energy loss in TELNES.2

Laurence Marks L-marks at northwestern.edu
Mon Apr 9 21:26:00 CEST 2007


If I wear my original hat as an electron microscopist, the fact that
ab-initio calculations have (for many years) done a rather bad job at
matching experiment in EELS is a BIG issue which really needs to be
addressed. Part is of course the fact that dynamical diffraction
effects are not included, but this should not affect the offsets (just
the relative peak heights).

On 4/9/07, Jorissen Kevin <kevin.jorissen at ua.ac.be> wrote:
> Thanks for your remarks, Laurence.
> I seem to remember that Slater (1/2 core hole) did give the best onset energy for C K edge, but it's a while since I did that test, and I'm not sure where to look for the numbers now :-).  There are actually theoretical arguments as for why it should give correct onset energies - outlined in the Paxton paper I mentioned, and probably in other places as well.
>
> I think it's important that we realize that the whole core hole approach is intrinsically an approximation, and we shouldn't expect it to work perfectly ; nor should we overanalyze.  I'm not saying we shouldn't strive for the best possible results - of course we should.  At the same time, one should be realistic about what's expected to come out right within a given theoretical framework.  It would be great if we could get correct onset energies from the Slater transition method.  On the other hand, there's no guarantee that this will reproduce the shape of the spectrum better than a full core hole (or, indeed, for some edges, a ground state calculation!), and core hole calculations are fairly expensive to run and converge.  So, unless one really cares to get the threshold energy ab initio, I think it makes sense to just shift and align to experiment.
>
>
> More discussion about this is certainly welcome!
>
> Happy computing, everyone
>
>
> Kevin.
>
>
>
>
> Kevin Jorissen
> University of Washington
> Dept. of Physics
> Box 351560
> Seattle, WA-98195-1560
> U.S.A.
>
> phone +1 206 543 3904
> fax +1 206 685 0635
> e-mail kevin.jorissen at ua.ac.be
> web page http://fraangelico.phys.washington.edu/~jorissen
>
>
>
> -----Original Message-----
> From: wien-bounces at zeus.theochem.tuwien.ac.at on behalf of Laurence Marks
> Sent: Mon 4/9/2007 11:31
> To: A Mailing list for WIEN2k users
> Subject: Re: [Wien] A question about the value of energy loss in TELNES.2
>
> One addition; I came across this when trying to figure out what Wien
> does (or should do) with charged cells. A core hole calculation is
> probably better than just taking the energies, BUT has a problem -- it
> is a charged cell calculation so all the energies need to be corrected
> in a way that is difficult to uniquely determine. From tests it
> appears that the Slater transition method approximately cancels out
> the charged cell corrections so is better. I don't think this has been
> fully analyzed -- if anyone wants to take a more serious look at this
> I would be interested in collaborating.
>
> On 4/9/07, Jorissen Kevin <kevin.jorissen at ua.ac.be> wrote:
> > Hi Ruijuan,
> >
> > Wien2k (or any other comparable code) cannot accurately calculate threshold energies.  This is not a problem of numerical accuracy, but simply of formalism (we calculate one electron states, while EELS is a two particle problem).
> >
> > Say you're doing a K edge, you can find the energy of the 1s orbital w.r.t. the Fermi level in the output of WIEN2k, and add a small amount if the edge in case.elnes doesn't start at 0 but a little higher ; but this will only give you a very rough idea of where the edge is.
> >
> > If you do a core hole calculation, you will obtain a different (lower) threshold energy.  But it will also be off.
> > Supposedly, a calculation with a half core hole ("Slater transition state") should give you the correct threshold energy (paper by Paxton a few years ago - can't remember the exact reference now).  But I'm not sure that it actually does in our calculations. (Anyone?)
> >
> > The number in case.innes is there so that we can calculate the q-vectors for the cross section accurately.  (It doesn't matter if it's off by a small fraction of a few eV.)
> >
> >
> > I stress that these are not limitations of the TELNES2 and WIEN2k implementation, but simply of the formalism that WIEN2k uses.
> >
> >
> > On the bright side, energy differences tend to be fairly accurate.  Eg., the L23-splitting came out well in the calculations that I've done ; and I also expect reasonably good chemical shifts.
> >
> >
> >
> > To summarize : believe energy differences, but not absolute threshold energies calculated by WIEN2k+TELNES2.  Just shift to align with experiment.
> >
> >
> >
> > Kevin.
> >
> >
> >
> > Kevin Jorissen
> > University of Washington
> > Dept. of Physics
> > Box 351560
> > Seattle, WA-98195-1560
> > U.S.A.
> >
> > phone +1 206 543 3904
> > fax +1 206 685 0635
> > e-mail kevin.jorissen at ua.ac.be
> > web page http://fraangelico.phys.washington.edu/~jorissen
> >
> > ________________________________
> >
> > From: wien-bounces at zeus.theochem.tuwien.ac.at on behalf of Ruijuan Xiao
> > Sent: Sun 4/8/2007 19:58
> > To: wien
> > Subject: [Wien] A question about the value of energy loss in TELNES.2
> >
> >
> >
> > Dear all,
> >
> > When I use the TELNES.2 to calculate energy loss near edge spectrum, I found that the value of the energy loss(x axis) in the simulated spectrum begins from "0". In the past, I always shift this x axis and make the onset of the theoretical spectra to well align with the experimental one, and make some comparison between them. I do not think there is any problem.
> >
> > But recently, someone asked me about the meaning of the x axis. He emphasized that the value of energy loss is the difference between the energies of final and initial states, so there should be an absolute value calculated from the program, and any shift of the theoretical energy loss to compare with the experiments does not make sense. I do not know how to explain it to him. So I wonder why the program gives a theoretical spectrum starting from zero instead of calculating the absolute value of energy loss. Is it because some initial states are core states, their energies can not be calculated as accurately as valence states?
> >
> > I also searched in this mailing list, and found some statements related to this question. It says "the '0' of the elnes spectrum corresponds to the threshold energy that you specify in case.innes.  So the 'real' energy loss is given by that on the x-axis, plus the threshold energy of case.innes.". But it seems that the threshold energy specified in case.innes is given by the user, so it can not reflect the real energy loss of the spectrum.
> >
> > I am looking forward to your reply. Thank you very much in advance.
> >
> >
> > Best regards,
> >
> > Sincerely,
> > Ruijuan
> >
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> >
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> >
> >
>
>
> --
> Laurence Marks
> Department of Materials Science and Engineering
> MSE Rm 2036 Cook Hall
> 2220 N Campus Drive
> Northwestern University
> Evanston, IL 60208, USA
> Tel: (847) 491-3996 Fax: (847) 491-7820
> email: L-marks at northwestern dot edu
> Web: www.numis.northwestern.edu
> EMM2007 http://ns.crys.ras.ru/EMMM07/
> Commission on Electron Diffraction of IUCR
> www.numis.northwestern.edu/IUCR_CED
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>
>


-- 
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
EMM2007 http://ns.crys.ras.ru/EMMM07/
Commission on Electron Diffraction of IUCR
www.numis.northwestern.edu/IUCR_CED


More information about the Wien mailing list