Question about using a proper model

newer
Saving data as chi(E) from Athena?

pushkar shejwalkar

27 Oct 2014 27 Oct '14

1:14 a.m.

Dear All, I am new to XAFS data analysis and have very limited knowledge about using modelling structures for XAS analysis. I am not sure as to if this question is directly related to Demter or XAFS analysis as such but its a part of it for sure. * Question number one* I am trying to analyze the sample I recently collected at SPring-8 facility. The structure I am hoping to be formed during the reaction is more of a organometallic complex at silica interface. In order to establish the structure I was told that there are two methods I could use. 1) Use a counter part compound which is exact/looks like the structure I am hoping (so basically known compound) and then do traditional comparison method to establish the structure. UNFORTUNATELY I cannot synthesize such compound in lab nor such material is available in literature ( including crystal structure). 2) is to use the model. right now for modelling such structure I am using Spartan program (again I have not used it earlier in my graduate work) So I want to know as to how to generate the models in order to get the correct crystal parameters? (atoms positions?) Is spartan good enough to do job? is there any other method for doing such fitting? * Question number two* In order to get lower E0 values after the fitting, I read that there is a method to use in which delE value can be set to 0 intentionally and then recalibrating the spectra in athena again in order to accurately calibrate the E0. I read paper by Shelly kelly and Bruce Ravel. Also I have seen the steps in one of shelly kellys PDF formatted lecture, although I am still not sure about the exacts steps to take in order to do such calibration. Could someone explain me the steps in details (especially taking artemis files to athena and then using it as STD spectra for calibration purposes, I am confused about that) Thank you very much everyone in advance for your patience to go through such a long email and helping to solve my silly doubts. Your New Bee in this fantastic group Pushkar

Attachments:

attachment.html (text/html — 2.5 KB)

Show replies by date

Matt Newville

27 Oct 27 Oct

9:12 a.m.

On Mon, Oct 27, 2014 at 1:14 AM, pushkar shejwalkar wrote:

...

Dear All, I am new to XAFS data analysis and have very limited knowledge about using modelling structures for XAS analysis. I am not sure as to if this question is directly related to Demter or XAFS analysis as such but its a part of it for sure. Question number one I am trying to analyze the sample I recently collected at SPring-8 facility. The structure I am hoping to be formed during the reaction is more of a organometallic complex at silica interface. In order to establish the structure I was told that there are two methods I could use. 1) Use a counter part compound which is exact/looks like the structure I am hoping (so basically known compound) and then do traditional comparison method to establish the structure. UNFORTUNATELY I cannot synthesize such compound in lab nor such material is available in literature ( including crystal structure). 2) is to use the model. right now for modelling such structure I am using Spartan program (again I have not used it earlier in my graduate work) So I want to know as to how to generate the models in order to get the correct crystal parameters? (atoms positions?) Is spartan good enough to do job? is there any other method for doing such fitting?

Sorry, I'm not familiar with Spartan -- it can probably generate a list of atomic X, Y, Z, atomic_number coordinates.... is that correct? If so, converting that into a feff.inp input file is not too hard, but if you haven't done that before, I'd suggest starting by running Atoms on a crystal structure with roughly comparable composition and the same central atom, and then it can be modified from there.

...

Question number two In order to get lower E0 values after the fitting, I read that there is a method to use in which delE value can be set to 0 intentionally and then recalibrating the spectra in athena again in order to accurately calibrate the E0. I read paper by Shelly kelly and Bruce Ravel. Also I have seen the steps in one of shelly kellys PDF formatted lecture, although I am still not sure about the exacts steps to take in order to do such calibration. Could someone explain me the steps in details (especially taking artemis files to athena and then using it as STD spectra for calibration purposes, I am confused about that)

Why is it important to you to get a lower value for E0? --Matt

pushkar shejwalkar

7:57 p.m.

Dear Matt, Thank you very much for your reply. Yes it generates list of XYZ cordinates. I am wondering if its ok to use such information. The reason is that while converting the XYZ file into .inp file, the format is different and hence I am simply typing the coordinates into it, however I am not sure as to should I change nleg values, Rmax values and other mphase mfeff values and all. Is it same for all the program? just changing the coordinates is enough? About lower E0 value, I read in Shelly's paper that mostly ppl do not report E0 value and if they and if it is >10 then there may be problem with modelling or so. one of the method to reduce such chances of high E0 is by properly aligning spectra in athena. She mention in this a method to use first artemis for getting first shell fitting using E0=0 and then coming back to the athena and using this file as reference. But I am not sure how to open artemis file in athena because if I open .fpj file in athena it doesnt recognise and says that the format is not readable if you continue then athena shutdown. In artemis I cannot save file as .prj. is there a way around? OR IT IS JUST THAT I AM DOING IT ALL WRONG? Thank you very much for your help. Pushkar On Mon, Oct 27, 2014 at 11:12 PM, Matt Newville wrote:

...

...
Dear All, I am new to XAFS data analysis and have very limited knowledge about using modelling structures for XAS analysis. I am not sure as to if this question is directly related to Demter or XAFS analysis as such but its a part of it for sure. Question number one I am trying to analyze the sample I recently collected at SPring-8 facility. The structure I am hoping to be formed during the reaction is more of a organometallic complex at silica interface. In order to establish

On Mon, Oct 27, 2014 at 1:14 AM, pushkar shejwalkar wrote: the

...
structure I was told that there are two methods I could use. 1) Use a counter part compound which is exact/looks like the structure I am hoping (so basically known compound) and then do traditional comparison method to establish the structure. UNFORTUNATELY I cannot synthesize such compound in lab nor such material is available in literature ( including crystal structure). 2) is to use the model. right now for modelling such structure I am using Spartan program (again I have not used it earlier in my graduate work) So I want to know as to how to generate the models in order to get the correct crystal parameters? (atoms positions?) Is spartan good enough to do job? is there any other method for doing such fitting?

Sorry, I'm not familiar with Spartan -- it can probably generate a list of atomic X, Y, Z, atomic_number coordinates.... is that correct? If so, converting that into a feff.inp input file is not too hard, but if you haven't done that before, I'd suggest starting by running Atoms on a crystal structure with roughly comparable composition and the same central atom, and then it can be modified from there.

...
Question number two In order to get lower E0 values after the fitting, I read that there is a method to use in which delE value can be set to 0 intentionally and then recalibrating the spectra in athena again in order to accurately

...
the E0. I read paper by Shelly kelly and Bruce Ravel. Also I have seen

calibrate the

...
steps in one of shelly kellys PDF formatted lecture, although I am still not sure about the exacts steps to take in order to do such calibration. Could someone explain me the steps in details (especially taking artemis files to athena and then using it as STD spectra for calibration purposes, I am confused about that)

Why is it important to you to get a lower value for E0?

--Matt _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

-- Best Regards, Pushkar Shejwalkar. Post-doctoral -Researcher,JSPS Fellow Hokkaido University, Sapporo, Japan

Matt Newville

8:29 p.m.

Pushkar, On Mon, Oct 27, 2014 at 7:57 PM, pushkar shejwalkar wrote:

...

Dear Matt, Thank you very much for your reply. Yes it generates list of XYZ cordinates. I am wondering if its ok to use such information. The reason is that while converting the XYZ file into .inp file, the format is different and hence I am simply typing the coordinates into it, however I am not sure as to should I change nleg values, Rmax values and other mphase mfeff values and all. Is it same for all the program? just changing the coordinates is enough?

Please understand that we only know as much about what you are trying to as you actually tell us. While we can give you advice, we cannot actually solve your problems for you. I think that what you are trying to do is create a feff.inp file from atomic coordinates generated by some other program called Spartan (coincidentally, Greek-themed!). A feff.inp has several parts to it, one being a list of X, Y, Z, IPOT, where X, Y, Z are atomic coordinated (to be clear -- in Angstroms) and IPOT is an integer index of one of the Potentials listed elsewhere in the file. There must be between 2 and 7 IPOTs, number sequentially starting with 0. There is exactly 1 IPOT '0' -- this is the absorbing atom. Each scattering element must have its own IPOT (including any other elements that are the same Z as the absorber). Starting with an existing feff.inp file and modifying the list of atomic coordinates and, if necessary, the list of potentials, is highly recommended. It's not that hard.

...

About lower E0 value, I read in Shelly's paper that mostly ppl do not report E0 value

People do report E0.

...

and if they and if it is >10 then there may be problem with modelling or so.

An E0 shift > 10 (or < -10) is a warning that something may not be right. It's not necessarily wrong, but should cause you to pause and try to understand such a large discrepancy between what Feff thought E0 should be and what the maximum of the first derivative is.

...

one of the method to reduce such chances of high E0 is by properly aligning spectra in athena. She mention in this a method to use first artemis for getting first shell fitting using E0=0 and then coming back to the athena and using this file as reference. But I am not sure how to open artemis file in athena because if I open .fpj file in athena it doesnt recognise and says that the format is not readable if you continue then athena shutdown. In artemis I cannot save file as .prj. is there a way around? OR IT IS JUST THAT I AM DOING IT ALL WRONG?

Just change E0 in Athena, perhaps copying a Group first, and use that chi(k) with the updated E0 as the data to fit. Usually Feff wants E0 further up the edge than the maximum of the first derivative, but not always. --Matt

pushkar shejwalkar

8:57 p.m.

Dear Matt thank you very much for the comments. I will try to change the E0 value in athena accordingly. It appeared to me from the lecture notes that we need to do some saving file in artemis and then opening it in athena or something like that. but may be I read it all wrong. Thanks again Pushkar On Tue, Oct 28, 2014 at 10:29 AM, Matt Newville wrote:

...

Pushkar,

On Mon, Oct 27, 2014 at 7:57 PM, pushkar shejwalkar wrote:

...
Dear Matt, Thank you very much for your reply. Yes it generates list of XYZ cordinates. I am wondering if its ok to use such information. The reason is that while converting the XYZ file into .inp file, the format is different and hence I am simply typing the coordinates into it, however I am not sure as to should I change nleg values, Rmax values and other mphase mfeff values and all. Is it same for all the program? just changing the coordinates is enough?

Please understand that we only know as much about what you are trying to as you actually tell us. While we can give you advice, we cannot actually solve your problems for you.

I think that what you are trying to do is create a feff.inp file from atomic coordinates generated by some other program called Spartan (coincidentally, Greek-themed!). A feff.inp has several parts to it, one being a list of X, Y, Z, IPOT, where X, Y, Z are atomic coordinated (to be clear -- in Angstroms) and IPOT is an integer index of one of the Potentials listed elsewhere in the file. There must be between 2 and 7 IPOTs, number sequentially starting with 0. There is exactly 1 IPOT '0' -- this is the absorbing atom. Each scattering element must have its own IPOT (including any other elements that are the same Z as the absorber).

Starting with an existing feff.inp file and modifying the list of atomic coordinates and, if necessary, the list of potentials, is highly recommended. It's not that hard.

...
About lower E0 value, I read in Shelly's paper that mostly ppl do not report E0 value

People do report E0.

...
and if they and if it is >10 then there may be problem with modelling or so.

An E0 shift > 10 (or < -10) is a warning that something may not be right. It's not necessarily wrong, but should cause you to pause and try to understand such a large discrepancy between what Feff thought E0 should be and what the maximum of the first derivative is.

...
one of the method to reduce such chances of high E0 is by properly aligning spectra in athena. She mention in this a method to use first artemis for getting first shell fitting using E0=0 and then coming back to the athena and using this file as reference. But I am not sure how to open artemis file in athena because if I open .fpj file in athena it doesnt recognise and says that the format is not readable if you continue then athena shutdown. In artemis I cannot save file as .prj. is there a way around? OR IT IS JUST THAT I AM DOING IT ALL WRONG?

Just change E0 in Athena, perhaps copying a Group first, and use that chi(k) with the updated E0 as the data to fit. Usually Feff wants E0 further up the edge than the maximum of the first derivative, but not always.

--Matt _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

-- Best Regards, Pushkar Shejwalkar. Post-doctoral -Researcher,JSPS Fellow Hokkaido University, Sapporo, Japan

Georges Siddiqi

28 Oct 28 Oct

7:45 a.m.

Hi, I may as well chime in since I actually have some expertise with trying to get XAS of organometallic compounds on silica surfaces (mainly pertaining to quesion#1). These are the papers I've previously published with my group for inorganic and organometallic compounds grafted on silica surfaces: http://www.pnas.org/content/111/32/11624 http://onlinelibrary.wiley.com/doi/10.1002/adsc.201400125/abstract http://onlinelibrary.wiley.com/doi/10.1002/anie.201308983/abstract the methodology we've had the most success with consists of, synthesize inorganic/organometallic precursor, and get XRD crystal structure, then grafting the compound onto the silica surface. We then take XAS measurements of both the precursor and the grafted species. From other characterization methods (NMR, IR) we have a good idea that the compound hasnt changed too much upon grafting to the silica surface. Then we take XAS measurements of both of these samples, fit the structure of the molecular precursor to the XRD data, then use some of the parameters of the fit of the molecular structure to fit the surface species. So in the previously described method we went around the problem of not knowing the structure of the species on the surface by using complementary characterization methods in conjunction with XAS to get a good fit. now let's say the structure on the surface starts to deviate more from the molecular analog (ie we do a heat treatment on the surface species on silica), then it gets trickier and I have to add more variables to the structure, but I was usually successful in fitting the surface species by playing around with the molecular model. Now, in some cases (not yet published) we havent been able to crystallize the molecular compound, and it's structure is not published. In those cases we had to dig through crystallographic databases to find structures that are similar (at least first shell coordination, and distances) to the one we have on the surface, and try to fit it. I've usually managed to get a reasonable fit by doing this. but if you cant synthesize a molecular analog, and you cant find a crystallographic structure reasonably close to what you hope to find on the surface (based on other characterization methods), then I'd say you're out of luck. best of luck, georges On Tue, Oct 28, 2014 at 2:57 AM, pushkar shejwalkar < pshejwalkar2004@gmail.com> wrote:

...

Dear Matt thank you very much for the comments. I will try to change the E0 value in athena accordingly. It appeared to me from the lecture notes that we need to do some saving file in artemis and then opening it in athena or something like that. but may be I read it all wrong. Thanks again Pushkar

On Tue, Oct 28, 2014 at 10:29 AM, Matt Newville < newville@cars.uchicago.edu> wrote:

...
Pushkar,

On Mon, Oct 27, 2014 at 7:57 PM, pushkar shejwalkar wrote:

...
Dear Matt, Thank you very much for your reply. Yes it generates list of XYZ cordinates. I am wondering if its ok to use such information. The reason is that while converting the XYZ file into .inp file, the format is different and hence I am simply typing the coordinates into it, however I am not sure as to should I change nleg values, Rmax values and other mphase mfeff values and all. Is it same for all the program? just changing the coordinates is enough?

Please understand that we only know as much about what you are trying to as you actually tell us. While we can give you advice, we cannot actually solve your problems for you.

I think that what you are trying to do is create a feff.inp file from atomic coordinates generated by some other program called Spartan (coincidentally, Greek-themed!). A feff.inp has several parts to it, one being a list of X, Y, Z, IPOT, where X, Y, Z are atomic coordinated (to be clear -- in Angstroms) and IPOT is an integer index of one of the Potentials listed elsewhere in the file. There must be between 2 and 7 IPOTs, number sequentially starting with 0. There is exactly 1 IPOT '0' -- this is the absorbing atom. Each scattering element must have its own IPOT (including any other elements that are the same Z as the absorber).

Starting with an existing feff.inp file and modifying the list of atomic coordinates and, if necessary, the list of potentials, is highly recommended. It's not that hard.

...
About lower E0 value, I read in Shelly's paper that mostly ppl do not report E0 value

People do report E0.

...
and if they and if it is >10 then there may be problem with modelling or so.

An E0 shift > 10 (or < -10) is a warning that something may not be right. It's not necessarily wrong, but should cause you to pause and try to understand such a large discrepancy between what Feff thought E0 should be and what the maximum of the first derivative is.

...
one of the method to reduce such chances of high E0 is by properly aligning spectra in athena. She mention in this a method to use first artemis for getting first shell fitting using E0=0 and then coming back to the athena and using this file as reference. But I am not sure how to open artemis file in athena because if I open .fpj file in athena it doesnt recognise and says that the format is not readable if you continue then athena shutdown. In artemis I cannot save file as .prj. is there a way around? OR IT IS JUST THAT I AM DOING IT ALL WRONG?

Just change E0 in Athena, perhaps copying a Group first, and use that chi(k) with the updated E0 as the data to fit. Usually Feff wants E0 further up the edge than the maximum of the first derivative, but not always.

--Matt _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

-- Best Regards, Pushkar Shejwalkar. Post-doctoral -Researcher,JSPS Fellow Hokkaido University, Sapporo, Japan

_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

3708

Age (days ago)

3709

Last active (days ago)

List overview

Download

5 comments

3 participants

participants (3)

Georges Siddiqi
Matt Newville
pushkar shejwalkar