Dear all, I have a question to ask all of you as follows. We have been investigating the EXAFS of some partially reduced Co materials (by Co EXFAS) and would wish to obtain some qualitative information on the Co metal particles. If, in fact, the Co had some average coordination environment because we have a distribution of Co particle sizes with oxygen adsorption, analyzing the windowed Co-O and Co-Co peaks together as a linear combination and assessing the average oxygen and Co coordination from the Co edge absorption (what we do with Athena) would appear to give a useful answer. However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size. Is there a way to do this with Athena or other method? Or do you know of a reference where this has been discussed? Of course, we can (and will) make experimental mixtures as references but wonder if there is a different way of doing the analysis or if there is some literature on this problem (which we have not been able to locate). Thank you. Steven
It seems like one of the problems we have been solving, the analysis of thiol-stabilized Pd nanoparticles where the nanoparticles coexist with low molecular weight Pd-thiol complexes, or polymers. It is another example of a heterogenous mixture you described. We found that it to be very speculative to attempt to determine the nanoparticle size by the method you described, since the Pd-S bonds in our cases were present in the surface of the nanoparticles, and in the polymers, both types contributing the same peak in r-space of EXAFS. In your case, I assume, the problem is how to discriminate the contribution of Co-O to the low r-peak: what portion of it is due to the nanoparticles, and what - due to the Co oxide. It is, perhaps, possible to construct a model that will account for the: 1) mixing fraction of the two phases, 2) size of the particles, and 3) the surface coverage of the oxides. However, there are many options in the S binding on the Pd particle, or O binding on Co particle, and I find this procedure a sort of circular reasoning. What we did with Pd particles: we obtained their size by TEM and used that information to calculate the average coordination number of Pd-Pd first nearest neighbors. After that, we obtained the relative mixture of nanopartlces and polymers, so it is sort of a backward approach, may be conservative but not too ambiguous. Anatoly ****************** Anatoly Frenkel, Ph.D. Associate Professor Physics Department Yeshiva University 245 Lexington Avenue New York, NY 10016 (YU) 212-340-7827 (BNL) 631-344-3013 (Fax) 212-340-7788 anatoly.frenkel@yu.edu http://www.yu.edu/faculty/afrenkel -----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov]On Behalf Of Steven S. Lim Sent: Thursday, May 05, 2005 3:13 PM To: ifeffit@millenia.cars.aps.anl.gov Cc: Gary L. Haller; Yanhui Yang; Yuan Chen; Lisa D. Pfefferle; Dragos Ciuparu Subject: [Ifeffit] Need an advice Dear all, I have a question to ask all of you as follows. We have been investigating the EXAFS of some partially reduced Co materials (by Co EXFAS) and would wish to obtain some qualitative information on the Co metal particles. If, in fact, the Co had some average coordination environment because we have a distribution of Co particle sizes with oxygen adsorption, analyzing the windowed Co-O and Co-Co peaks together as a linear combination and assessing the average oxygen and Co coordination from the Co edge absorption (what we do with Athena) would appear to give a useful answer. However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size. Is there a way to do this with Athena or other method? Or do you know of a reference where this has been discussed? Of course, we can (and will) make experimental mixtures as references but wonder if there is a different way of doing the analysis or if there is some literature on this problem (which we have not been able to locate). Thank you. Steven
Hi Steven, On Thu, 5 May 2005, Steven S. Lim wrote:
Dear all,
I have a question to ask all of you as follows.
We have been investigating the EXAFS of some partially reduced Co materials (by Co EXFAS) and would wish to obtain some qualitative information on the Co metal particles. If, in fact, the Co had some average coordination environment because we have a distribution of Co particle sizes with oxygen adsorption, analyzing the windowed Co-O and Co-Co peaks together as a linear combination and assessing the average oxygen and Co coordination from the Co edge absorption (what we do with Athena) would appear to give a useful answer. However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size. Is there a way to do this with Athena or other method? Or do you know of a reference where this has been discussed? Of course, we can (and will) make experimental mixtures as references but wonder if there is a different way of doing the analysis or if there is some literature on this problem (which we have not been able to locate).
If I understand correctly, you are asking where there is way to distinguish a two possible cases: 1) a physical mixture of metallic cobalt + cobalt oxide (what I would call "mixed phases") ; and 2) a chemically intermediate phase, that was neither fully metallic nor oxidized. I don't know enough about the system you're working on to know the exact details. In general the EXAFS analysis (as with Artemis) allows one to distinguish these case, though it can sometimes be complicated. For the first case, you could create models of Co-Co and Co-O and fit the spectrum as a linear combination of those two phases without allowing the bond distances to change from the known values for metal and oxide phases. That gives one fit, with a result for metal / oxide fraction. For the second case, you could allow both Co-Co and Co-O contributions but allow the distances and coordination numbers to vary. If the distances move significantly from those of the isolated phases, that's good evidence that a simple physical mixture is not sufficient. But also, you can compare the goodness-of-fit parameters for the two cases to help decide which of the two cases better explain the data. Ideally, you could also include further "shells" in this analysis, especially for second neighbor Co-Co in CoO. This is more work, generally requires fairly good EXAFS data, but can really improve the confidence that you have a component that is the isolated oxide. Hope that's enough to get you started in the right direction. If not, let us know! --Matt
Re: [Ifeffit] Need an adviceMatt, the complication here is that the Co-O bonds will be present in the surface of Co particles because they are not just "metallic cobalt" but can be nano-size particles, and thus their surface contribution may be significant. If it is this type of a mixed phase, not the other type, when the particles are larger than, say, 5nm in size, then these bonds will count and it is not possible to deconvolute them from the Co-O bonds in the oxide. Therefore, the results of the Co-O and Co-Co bond coordination numbers cannot be easily interpreted in terms of the oxide/metal ratio if, again, the particles are smaller than 5nm and if the binding geometry of O on the Co surface is not known. For example, in fcc metal nanoparticles, ligands may bind above the bridge, the face, or the vertex. In each case, the coordination number of M-O will be different, and the oxide/metal ratio will thus change accordingly to the model used. I think, the best way to deal with it is to rely on TEM or, perhaps, XRD, to actually measure the particle size and then infer the Co-Co coordination numbers. The Co-Co coordination numbers measured from EXAFS can then be compared to the TEM-derived, and if the answers are different (namely, the experimental numbers are smaller than the TEM-derived), this discrepancy can be interpreted in terms of the "mixed phase" situation, and the mixing fraction of metal is exactly the ratio of these two numbers. What I've just described, is actually going to published, and I will send the reference when it is out. Regards, Anatoly ****************** Anatoly Frenkel, Ph.D. Associate Professor Physics Department Yeshiva University 245 Lexington Avenue New York, NY 10016 (YU) 212-340-7827 (BNL) 631-344-3013 (Fax) 212-340-7788 anatoly.frenkel@yu.edu http://www.yu.edu/faculty/afrenkel -----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov]On Behalf Of Matt Newville Sent: Friday, May 06, 2005 4:06 PM To: XAFS Analysis using Ifeffit Cc: Gary L. Haller; Yanhui Yang; Yuan Chen; Lisa D. Pfefferle; Dragos Ciuparu Subject: Re: [Ifeffit] Need an advice Hi Steven, On Thu, 5 May 2005, Steven S. Lim wrote:
Dear all,
I have a question to ask all of you as follows.
We have been investigating the EXAFS of some partially reduced Co materials (by Co EXFAS) and would wish to obtain some qualitative information on the Co metal particles. If, in fact, the Co had some average coordination environment because we have a distribution of Co particle sizes with oxygen adsorption, analyzing the windowed Co-O and Co-Co peaks together as a linear combination and assessing the average oxygen and Co coordination from the Co edge absorption (what we do with Athena) would appear to give a useful answer. However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size. Is there a way to do this with Athena or other method? Or do you know of a reference where this has been discussed? Of course, we can (and will) make experimental mixtures as references but wonder if there is a different way of doing the analysis or if there is some literature on this problem (which we have not been able to locate).
If I understand correctly, you are asking where there is way to distinguish a two possible cases: 1) a physical mixture of metallic cobalt + cobalt oxide (what I would call "mixed phases") ; and 2) a chemically intermediate phase, that was neither fully metallic nor oxidized. I don't know enough about the system you're working on to know the exact details. In general the EXAFS analysis (as with Artemis) allows one to distinguish these case, though it can sometimes be complicated. For the first case, you could create models of Co-Co and Co-O and fit the spectrum as a linear combination of those two phases without allowing the bond distances to change from the known values for metal and oxide phases. That gives one fit, with a result for metal / oxide fraction. For the second case, you could allow both Co-Co and Co-O contributions but allow the distances and coordination numbers to vary. If the distances move significantly from those of the isolated phases, that's good evidence that a simple physical mixture is not sufficient. But also, you can compare the goodness-of-fit parameters for the two cases to help decide which of the two cases better explain the data. Ideally, you could also include further "shells" in this analysis, especially for second neighbor Co-Co in CoO. This is more work, generally requires fairly good EXAFS data, but can really improve the confidence that you have a component that is the isolated oxide. Hope that's enough to get you started in the right direction. If not, let us know! --Matt _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
At 04:20 PM 5/6/2005 -0400, Anatoly wrote:
<fontfamily><param>Arial</param><color><param>0000,0000,ffff</param><smaller>I think, the best way to deal with it is to rely on TEM or, perhaps, XRD, to actually measure the particle size and then infer the Co-Co coordination numbers. The Co-Co coordination numbers measured from EXAFS can then be compared to the TEM-derived, and if the answers are different (namely, the experimental numbers are smaller than the TEM-derived), this discrepancy can be interpreted in terms of the "mixed phase" situation, and the mixing fraction of metal is exactly the ratio of these two numbers. What I've just described, is actually going to published, and I will send the reference when it is out. </smaller></color></fontfamily> Using TEM in this way sounds like a good idea; I look forward to seeing this paper. But I would strongly caution people to carefully consider if anything is known about the size distribution before using XRD to estimate sizes. If the size distribution of crystallites is broad, XRD preferentially "notices" the largest crystallites (it's roughly weighted as the square of the volume). EXAFS has a different weighting (more like volume-weighting, or even a little more biased toward smaller particles). Therefore attempts to use an XRD-determined size to constrain EXAFS models can be highly distorted if the size distribution is moderately broad. --Scott Calvin Sarah Lawrence College
RE: [Ifeffit] Need an adviceVery good point! And, with TEM, it is a similar trend (I think, Scott, you mentioned it in your JAP paper): if the distribution is broad, TEM overemphasizes larger particles because of the resolution/contrast problems at the lower size tail of the distribution. Anatoly -----Original Message----- From: Scott Calvin [mailto:scalvin@slc.edu] Sent: Friday, May 06, 2005 4:46 PM To: Frenkel, Anatoly; XAFS Analysis using Ifeffit Subject: RE: [Ifeffit] Need an advice At 04:20 PM 5/6/2005 -0400, Anatoly wrote:
I think, the best way to deal with it is to rely on TEM or, perhaps, XRD, to actually measure the particle size and then infer the Co-Co coordination numbers. The Co-Co coordination numbers measured from EXAFS can then be compared to the TEM-derived, and if the answers are different (namely, the experimental numbers are smaller than the TEM-derived), this discrepancy can be interpreted in terms of the "mixed phase" situation, and the mixing fraction of metal is exactly the ratio of these two numbers. What I've just described, is actually going to published, and I will send the reference when it is out. Using TEM in this way sounds like a good idea; I look forward to seeing this paper. But I would strongly caution people to carefully consider if anything is known about the size distribution before using XRD to estimate sizes. If the size distribution of crystallites is broad, XRD preferentially "notices" the largest crystallites (it's roughly weighted as the square of the volume). EXAFS has a different weighting (more like volume-weighting, or even a little more biased toward smaller particles). Therefore attempts to use an XRD-determined size to constrain EXAFS models can be highly distorted if the size distribution is moderately broad. --Scott Calvin Sarah Lawrence College
Anatoly, Scott, Sorry to be so dense here, but did the original message even mention nanoparticles?? I see 'partially reduced Co materials'. Without knowing more, it seems like a pretty big stretch to read that as 'nanoparticles', no? As you say, if there are oxidized nanoparticles, that could show up as something different than a simple physical mixture of Co metal and Co oxide. Of course, one can also want to distinguish 'chemically reacted' from 'physical mixed' in other situations. --Matt
Hi Steven, If I understand correctly what you're describing, I would certainly recommend against trying to use the first Co-Co peak as any indication of the size of the metal particles. As Anatoly indicates in his post, there's a bunch of interacting factors that control the size of that peak, making it very dicey to use it as any indication of size. If the oxide really doesn't contribute much beyond the first peak, it might be possible to look at the relative sizes of the Co-Co peaks as a function of R (i.e. go well beyond the first shell; say, out to 5 Angstroms if the data supports it) and compare to a Co metal standard (or a FEFF calculation). At least this has the benefit of making use of redundant data (the amplitude of more than one Co-Co peak). --Scott Calvin Sarah Lawrence College At 03:12 PM 5/5/2005 -0400, you wrote:
<fontfamily><param>Arial</param>Dear all, </fontfamily> <fontfamily><param>Arial</param>I have a question to ask all of you as follows. </fontfamily> <fontfamily><param>Arial</param>We have been investigating the EXAFS of some partially reduced Co materials (by Co EXFAS) and would wish to obtain some qualitative information on the Co metal particles. If, in fact, the Co had some average coordination environment because we have a distribution of Co particle sizes with oxygen adsorption, analyzing the windowed Co-O and Co-Co peaks together as a linear combination and assessing the average oxygen and Co coordination from the Co edge absorption (what we do with Athena) would appear to give a useful answer. However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size. Is there a way to do this with Athena or other method? Or do you know of a reference where this has been discussed? Of course, we can (and will) make experimental mixtures as references but wonder if there is a different way of doing the analysis or if there is some literature on this problem (which we have not been able to locate). </fontfamily> <fontfamily><param>Arial</param>Thank you. </fontfamily> <fontfamily><param>Arial</param>Steven </fontfamily>_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit <<<<<<<<
Hi Matt, The quote below is where I got the idea that Steven was interested in size-determination of nanoparticles, but I may have misunderstood...maybe he meant to use the fraction of oxide as a marker for the surface to volume ratio? In any case, that still makes me think that he expects to have nanoscale particles... --Scott Calvin Sarah Lawrence College At 03:12 PM 5/5/2005 -0400, Steven wrote:
<fontfamily><param>Arial</param> However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size.</fontfamily>
Re: [Ifeffit] Need an adviceLet me break the tie: here is what I just received from one of the people cc'd to in Steven's email. This problem is indeed even more complex than we all think. Dear Anatoly, The problem is even more complex: - there is one Co2+ species in tetrahedral environment; - there are small (20 - 100 atoms) metallic clusters; - there is an intermediate Co+1 like species in tetrahedral coordination. All these species may be present in some of the samples. Plus, I assume the Co-O distance between Co in small metallic clusters and oxygen in the silica is different than that between Co2+ ions and O2- ions in the cobalt oxide - like compound, which adds to the complexety. We feel like we're chasing our tail with this thing...:) Dragos -----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov]On Behalf Of Scott Calvin Sent: Friday, May 06, 2005 5:09 PM To: XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] Need an advice Hi Matt, The quote below is where I got the idea that Steven was interested in size-determination of nanoparticles, but I may have misunderstood...maybe he meant to use the fraction of oxide as a marker for the surface to volume ratio? In any case, that still makes me think that he expects to have nanoscale particles... --Scott Calvin Sarah Lawrence College At 03:12 PM 5/5/2005 -0400, Steven wrote: >>>> However, if what we have is closer to a physical mixture of small Co metal particles and Co oxide particles, it would seem that it would be more appropriate to analyze separately the windowed Co-O and Co-Co peaks in R-space, to somehow normalize the Co-Co to the fraction of metal in the sample and use this as a qualitative measure of metal particle size.
participants (4)
-
Anatoly Frenkel -
Matt Newville -
Scott Calvin -
Steven S. Lim