[Ifeffit] Modelling texture in Artemis

[Kirstine Dalgaard]

Dear all


How can texture in a powder sample be modeled in ARTEMIS?


My samples are fine powders of a layered material, mixed with cellulose and pressed to pellets. The data were collected in fluorescence mode with a 45 degree tilt.

PXRD data reveal that the samples are highly textured, including standards.


I am new to XAFS, and I need some advise to deal with this issue.


Best regards

Kirstine Dalgaard

[Joselaine Cáceres gonzalez]

Dear all,

I am new in this group so I don't know if this is the right place to ask
about it, but IÂŽd like to know about LCF performed with Athena in XANES
data. I had some difficulties in founding information about errors reported
by the program.
I am working with different techniques and in some cases I have to
calculate uncertainty in results reported. In LCF results reported by
Athena I don't know how to interpret the errors reported between
parenthesis along with each model spectra.
In my reports I carefully define if I am reporting the standard deviation
of the mean, a combined standard uncertainty,etc. depending each case, the
coverage probability, etc. So I need to understand the information Athena
is reporting as errors in LCF.

So grateful if anyone can help me with this issue.
Best regards.
Joselaine Cáceres

2018-08-14 6:05 GMT-03:00 Kirstine Dalgaard <***@inano.au.dk>:

> Dear all
>
>
> How can texture in a powder sample be modeled in ARTEMIS?
>
>
> My samples are fine powders of a layered material, mixed with cellulose
> and pressed to pellets. The data were collected in fluorescence mode with a
> 45 degree tilt.
>
> PXRD data reveal that the samples are highly textured, including
> standards.
>
>
> I am new to XAFS, and I need some advise to deal with this issue.
>
>
> Best regards
>
> Kirstine Dalgaard
>
> _______________________________________________
> Ifeffit mailing list
> ***@millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
> Unsubscribe: http://millenia.cars.aps.anl.gov/mailman/options/ifeffit
>
>

[Matt Newville]

Hi Joselaine,

On Tue, Aug 14, 2018 at 10:21 AM Joselaine Cáceres gonzalez <
***@gmail.com> wrote:

> Dear all,
>
> I am new in this group so I don't know if this is the right place to ask
> about it, but IÂŽd like to know about LCF performed with Athena in XANES
> data. I had some difficulties in founding information about errors reported
> by the program.
> I am working with different techniques and in some cases I have to
> calculate uncertainty in results reported. In LCF results reported by
> Athena I don't know how to interpret the errors reported between
> parenthesis along with each model spectra.
> In my reports I carefully define if I am reporting the standard deviation
> of the mean, a combined standard uncertainty,etc. depending each case, the
> coverage probability, etc. So I need to understand the information Athena
> is reporting as errors in LCF.
>

The reported uncertainties are the 1-sigma standard errors. However, the
uncertainties in the data are necessary to get the scale of the standard
errors correctly, and these uncertainties in the data are (almost always)
not supplied for the analysis, so the program has to guess what those
uncertainties should be.

To do this it follows a common if not-quite-rigorous approach of first
asserting that the fit is "good" (otherwise you probably wouldn't care
about the detailed scale of the uncertainties) so that reduced chi-square
should be 1. That is, it effectively rescales the uncertainty in the
data, epsilon, so that reduced chi-square is 1, and uses that to scale the
reported uncertainties. Another way to look at this is that the reported
uncertainties increase the reported chi-square by reported reduced
chi-square, without regard to the value of epsilon used.

Hope that helps. Please try to use a separate topic thread in the future,
as this question has little or nothing to do with the subject line...

--Matt

[Joselaine Cáceres gonzalez]

Dear Matt,

Thank you very much, your answer has been very helpful. Please, apologize
me for not opening a separate topic, I was not sure how this work, IÂŽll be
more careful in the future.
Best regards
Joselaine

El mié., 15 ago. 2018 a las 12:49, Matt Newville (<
***@cars.uchicago.edu>) escribió:

> Hi Joselaine,
>
> On Tue, Aug 14, 2018 at 10:21 AM Joselaine Cáceres gonzalez <
> ***@gmail.com> wrote:
>
>> Dear all,
>>
>> I am new in this group so I don't know if this is the right place to ask
>> about it, but IÂŽd like to know about LCF performed with Athena in XANES
>> data. I had some difficulties in founding information about errors reported
>> by the program.
>> I am working with different techniques and in some cases I have to
>> calculate uncertainty in results reported. In LCF results reported by
>> Athena I don't know how to interpret the errors reported between
>> parenthesis along with each model spectra.
>> In my reports I carefully define if I am reporting the standard deviation
>> of the mean, a combined standard uncertainty,etc. depending each case, the
>> coverage probability, etc. So I need to understand the information Athena
>> is reporting as errors in LCF.
>>
>
> The reported uncertainties are the 1-sigma standard errors. However, the
> uncertainties in the data are necessary to get the scale of the standard
> errors correctly, and these uncertainties in the data are (almost always)
> not supplied for the analysis, so the program has to guess what those
> uncertainties should be.
>
> To do this it follows a common if not-quite-rigorous approach of first
> asserting that the fit is "good" (otherwise you probably wouldn't care
> about the detailed scale of the uncertainties) so that reduced chi-square
> should be 1. That is, it effectively rescales the uncertainty in the
> data, epsilon, so that reduced chi-square is 1, and uses that to scale the
> reported uncertainties. Another way to look at this is that the reported
> uncertainties increase the reported chi-square by reported reduced
> chi-square, without regard to the value of epsilon used.
>
> Hope that helps. Please try to use a separate topic thread in the future,
> as this question has little or nothing to do with the subject line...
>
> --Matt
>
> _______________________________________________
> Ifeffit mailing list
> ***@millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
> Unsubscribe: http://millenia.cars.aps.anl.gov/mailman/options/ifeffit
>

[Matt Newville]

Hi Kristine,

On Tue, Aug 14, 2018 at 7:09 AM Kirstine Dalgaard <***@inano.au.dk> wrote:

> Dear all
>
>
> How can texture in a powder sample be modeled in ARTEMIS?
>
>
>
I am not sure that it can be. EXAFS is sensitive to local structure around
the absorbing atom and is not sensitive to crystallinity or long range
order. Well, EXAFS *is* sensitive to the relative orientation of the local
structure to the polarization of the incident X-ray beam (usually
horizontal unless you're purposely changing the polarization, but since you
didn't mention this, I'll assume you are not doing this).


My samples are fine powders of a layered material, mixed with cellulose and
> pressed to pellets. The data were collected in fluorescence mode with a 45
> degree tilt.
>
> PXRD data reveal that the samples are highly textured, including
> standards.
>

My understanding is that texture in powder diffraction reflects a limited
range of orientations in a sample, insofar as there may be preferential
ordering and an incomplete distribution of orientations. But crystal
orientation and local structure are not necessarily the same thing. If the
texturing means some interatomic bonds are preferentially oriented with
respect to the horizontal plane then maybe there would be an effect. But
that more or less requires the local structure (not the crystal structure)
to be anisotropic around the absorbing atom. That is, even if the
crystal is a layered material, if the absorbing atom has an isotropic
environment, there will likely be very little or no effect.

On the other hand, if the structure is layered and those layers are
preferentially in the horizontal plane, then EXAFS will be preferentially
probing bonds along the layer planes.

Sorry if that sounds too vague, but I cannot give a more specific answer
without knowing more about the system. Maybe someone else here can.

--Matt

[Kirstine Dalgaard]

(I am not sure this is the correct way to reply to this thread, hopefully it works)


Dear Matt


Thank you for your reply! The standard, which is highly textured, is NbSe2. The Nb K-edge was scanned.

The structure consists of layers with Nb coordinated to 6 selenium atoms in a trigonal prismatic manner. Thus, the environment is not isotropic. The nearest Nb-neighbors are e.g. only in the plane of the layer.

The preferred orientation is with the stacking-axis perpendicular to the pellet plane.


I then assume that I do have an effect of the non-random orientation of bonds/paths relative to the X-ray polarization?


On slide 28 in this presentation by Bruce Ravel http://xafs.org/Workshops/NSLS2002?action=AttachFile&do=get&target=Ravel.pdf it is stated that:

"FEFF is capable of computing the effect of polarization directly. The polarized theory is easily used in analysis with polarized data".


However, how do I implement this in my analysis? Is there maybe a written guide?


Best regards

Kirstine


________________________________
From: Kirstine Dalgaard
Sent: Tuesday, August 14, 2018 11:05 AM
To: ***@millenia.cars.aps.anl.gov
Subject: Modelling texture in Artemis


Dear all


How can texture in a powder sample be modeled in ARTEMIS?


My samples are fine powders of a layered material, mixed with cellulose and pressed to pellets. The data were collected in fluorescence mode with a 45 degree tilt.

PXRD data reveal that the samples are highly textured, including standards.


I am new to XAFS, and I need some advise to deal with this issue.


Best regards

Kirstine Dalgaard

[Matt Newville]

Hi Kristine,


On Thu, Aug 16, 2018 at 10:28 AM Kirstine Dalgaard <***@inano.au.dk> wrote:

> (I am not sure this is the correct way to reply to this thread, hopefully
> it works)
>
>
> Dear Matt
>
>
> Thank you for your reply! The standard, which is highly textured, is
> NbSe2. The Nb K-edge was scanned.
>
> The structure consists of layers with Nb coordinated to 6 selenium atoms
> in a trigonal prismatic manner. Thus, the environment is not isotropic. The
> nearest Nb-neighbors are e.g. only in the plane of the layer.
>
> The preferred orientation is with the stacking-axis perpendicular to the
> pellet plane.
>
>
> I then assume that I do have an effect of the non-random orientation of
> bonds/paths relative to the X-ray polarization?
>
>
> On slide 28 in this presentation by Bruce Ravel
> http://xafs.org/Workshops/NSLS2002?action=AttachFile&do=get&target=Ravel.pdf it
> is stated that:
>
> "FEFF is capable of computing the effect of polarization directly. The
> polarized theory is easily used in analysis with polarized data".
>
>
> However, how do I implement this in my analysis? Is there maybe a written
> guide?
>
>
>
Like Bruce wrote, to account for the polarization of the beam relative to
your sample, you can specify the direction of polarization in the feff.inp
file. That is done by adding a line ike

POLARIZATION 1 0 0

which will put the polarization vector along the X axis for the cluster of
atoms in that feff.inp file. The feffNNNN.dat files calculated this way
will include the effect of polarization.

I'm not entirely sure I have you geometry correct, but I think you are
saying that the layers are perpendicular to the axis of your sample disk,
which sits at 45 degrees to the incident beam. I think that means the
polarization vector would be along (1 0 1) (that is, in the X-Z plane)
with Z beling the sample normal. But I'm not sure what the orientation of
your structure is relative to the sample normal...

Hope that helps,

--Matt

[Kirstine Dalgaard]

Dear Matt


Thank you for the advice. I could figure it out myself, if I know how the POLARIZATION parameter is interpreted or defined.


Where can I find that information? I found this on my own:

"This card specifies the direction of the electric field of the incident beam, or the main axis of the ellipse in the case of elliptical polarization. x, y, z are the coordinates of the nonzero polarization vector?"


But I lack a definition of the coordinate system - the polarization direction with respect to what, exactly?


Best

Kirstine

________________________________
From: Kirstine Dalgaard
Sent: Thursday, August 16, 2018 5:27 PM
To: ***@millenia.cars.aps.anl.gov
Subject: Re: Modelling texture in Artemis


(I am not sure this is the correct way to reply to this thread, hopefully it works)


Dear Matt


Thank you for your reply! The standard, which is highly textured, is NbSe2. The Nb K-edge was scanned.

The structure consists of layers with Nb coordinated to 6 selenium atoms in a trigonal prismatic manner. Thus, the environment is not isotropic. The nearest Nb-neighbors are e.g. only in the plane of the layer.

The preferred orientation is with the stacking-axis perpendicular to the pellet plane.


I then assume that I do have an effect of the non-random orientation of bonds/paths relative to the X-ray polarization?


On slide 28 in this presentation by Bruce Ravel http://xafs.org/Workshops/NSLS2002?action=AttachFile&do=get&target=Ravel.pdf it is stated that:

"FEFF is capable of computing the effect of polarization directly. The polarized theory is easily used in analysis with polarized data".


However, how do I implement this in my analysis? Is there maybe a written guide?


Best regards

Kirstine


________________________________
From: Kirstine Dalgaard
Sent: Tuesday, August 14, 2018 11:05 AM
To: ***@millenia.cars.aps.anl.gov
Subject: Modelling texture in Artemis


Dear all


How can texture in a powder sample be modeled in ARTEMIS?


My samples are fine powders of a layered material, mixed with cellulose and pressed to pellets. The data were collected in fluorescence mode with a 45 degree tilt.

PXRD data reveal that the samples are highly textured, including standards.


I am new to XAFS, and I need some advise to deal with this issue.


Best regards

Kirstine Dalgaard

[Matt Newville]

Hi Kristine,

On Mon, Aug 20, 2018 at 5:36 AM Kirstine Dalgaard <***@inano.au.dk> wrote:

> Dear Matt
>
>
> Thank you for the advice. I could figure it out myself, if I know how the
> POLARIZATION parameter is interpreted or defined.
>
>
> Where can I find that information? I found this on my own:
>
> "This card specifies the direction of the electric field of the incident
> beam, or the main axis of the ellipse in the case of elliptical
> polarization. x, y, z are the coordinates of the nonzero polarization vector
> "
>
>
> But I lack a definition of the coordinate system - the polarization
> direction with respect to what, exactly?
>

With respect to the coordinate system used by the cluster of atoms.

--Matt