Charles,

When you asked me this in private email last week, I did not completely understand the question.

One (simple) solution to your question is to run Atoms on crystalline Pt and have it write out a cluster of the approximate size. For instance, a 7 angstrom cluster contains 87 atoms. I am guessing that the exafs from that will be pretty similar to the cluster you are looking for.

Of course, there are truncation effects. The cluster you are interested in will have some geometry which may not be well approximated by a sphere. Also, one must worry about the attenuation of coordination numbers due to absorption by atoms at the perifery of the cluster.

Also, remember that Atoms is a crystallography code. If you are interested in a cluster of atoms that is not a crystal, then you will need some other mechanism for generating the list of atomic coordinates needed by Feff.

Many people, including some some of the frequent contributors to this mailing list, have done considerable work on exafs of metallic nanoparticles. You should look up papers by Anatoly Frenkel and by Scott Calvin (and probably others as well) to see how some of the exafs experts who have already done work in this field have addressed your problem.

HTH, B

I tried this out, but some difficulties when attempting to plot with Artemis. I generated the Pt cluster with TkAtoms ver. 3.0beta9 running perl S.006001 and Tk800.024 and produced the "feff.inp" file. (My computer operates under Windows XP.) I tried to plot with Artemis 0.7.015(c) after importing the *.inp. file. (To my understanding, these are the most up-to-date program versions.) I get a message, "Artemis trapped one or more errors! Error message dumped onto screen." I'd like to see what the errors are in order to figure out how to fix for proper plotting, but when I go to the "view message" window (under "Edit" menu), it is blank. Not obvious where the error message(s) are displayed. Is there another location in the Artemis program that the error message should be if not in this window? Charles

Re: [Ifeffit] can't see error message(s) Artemis after *.inp import from TkAtoms At 11:30 PM 5/23/2005, you wrote:

Hi Charles,

Sorry for the delay. There won't be anything to plot after importing the feff.inp file. First, you'll have to run feff, then import some of the feff paths into Artemis. At that point you can either simply add the paths together or do a fit to actual data.

For the "Artemis trapped one of more errors! Error message dumped to screen" message, these actually go to the Windows command session window (aka the DOS shell).

Hope that helps!

--Matt

Thanks Matt! I wouldn't have guessed that was where the message went. It immediately popped up before I could try plotting so I thought I had to do something before running feff. In any case, I found that the error had to do with multiple ways TkAtoms can save the *.inp file. In the program [ver 3.0beta9], the *.inp is generated in the "TkAtoms display Frame 1", displaying the xyz coordinates. In this "Frame 1" window, there is a button that gives you the option of saving the *.inp file. When I use it to do so, Artemis doesn't seem to recognize the file when I try to import it (hence the error message). I found that I had to ignore this window and use the "File" submenu (at the top in blue in TkAtoms) to save the *.inp file. I could then import with no error message & finally do the Artemis plot. -Charles

Anatoly, I did not consider magic numbers & don't know for certain what the exact # is. The 95 atom cluster is only an estimate of the size of the composition of particles that was observed in TEM images of Pt tethered onto a a surface with -COH, -C=O groups; the exact nature of the bonding is what I'm trying to determine. The Pt seems to be stabilized by its surface interactions; the exact nature of the binding is what I'm trying to figure out. The EXAFS data that I have "sees" both the Pt-Pt interactions as well as the Pt-surface interactions. I realize that it would be best to simple scan a sample containing Pt clusters of the same size as observed in the TEM; but, unfortunately, Pt has a tendency to agglomerate and it is difficult to get a monodispersed size in colloidal suspension in solution. Comparing with Pt foil doesn't work because of the quenching problems in the fluorescence since there are so many atoms. I wanted to get a simulated EXAFS spectrum of something that just represents the bulk Pt-Pt only to compare with the data that I have. Differences in the spectra (in principle) would be attributed to Pt-surface interactions present in the TEM. I would like to try to get an idea of the # of Pt interacting with the surface as opposed to the Pt-Pt interactions. Charles At 09:23 PM 5/16/2005, you wrote:

Charles,

The 95 atoms Pt cluster you describe sounds strange. It does not fit any sequence of magic numbers typical for regular polyhedral clusters (13, 55, 147... for cuboctahedral fcc and icosahedral clusters, or 13, 38, 79, 140... for truncated octahedral or : 10, 37, 92, 185...for supported cuboctahedral clusters).

If you are modeling an fcc cluster (Pt does pack into fcc structure, even for sizes smaller than your clusters, contrary, e.g., to Au which is possible to observer in icosahedral structure for 13-atom clusters) then your "spherical" cluster of 95 atoms is not a closed shell cluster and thus it is not a stable isomer. What are you trying to accomplish?

Anatoly

Anatoly Frenkel

-----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov]On Behalf Of Charles Chusuei Sent: Monday, May 16, 2005 4:48 PM To: ifeffit@millenia.cars.aps.anl.gov Subject: [Ifeffit] idealized 95-atom Pt cluster

Hi,,,

I'm a new user to Artemis and Atoms. I would like to generate a simulated EXAFS spectrum of an idealized bare 95-atom cluster of Pt (with no ligands attached to it). Looking at Cambridge data files, the largest cluster documented is 18, so I would have to make one up. I would like to arrange the 95 atoms into an idealized sphere with the standard geometry for bulk Pt-Pt bonding that can be predicted from the crystal structure (e.g. metallic radius = 139nm, metallic separation = 277.5nm separation, etc.)

I understand this can be done using Artemis and Atoms. I'm relatively new to using the software (prodominantly know how to use Athena for plotting the experimental EXAFS data , not much more).

Could anyone provide suggestions on how to get started on this task? E.g. Generating the crystal data and displaying the EXAFS of the resulting bare cluster?

Charles

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

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

Hi Scott, Thanks for your suggestion. I find that I need to generate FEFF files for larger clusters than I had initially started off with. In your recent Physica Scripta paper [T115 (2005) 744-748], you had looked at some pretty large Pd clusters, particles 0.1 to 200nm in diam. (page 745, col 2, para 2, lines 5-6). Some of these clusters are well over 500 atoms. How did you manage to create the FEFF? When I try to do so for a cluster this large in Artemis, the program flags an error saying that I have too many atoms (in my *.inp file generated from TkAtoms). Charles At 10:36 AM 5/17/2005, you wrote:

Hi Charles,

Here's what I would try for your situation:

1) Use Artemis to create FEFF files for bulk crystalline Pt. Also generate FEFF files for some plausible Pt-surface interactions.

2) Constrain the amplitude of the Pt-Pt paths to fall off with reff in the way expected for a homogenous sphere (I have a couple of publications on this, including a recent one in Physica Scripta). If you are confident you know the crystallite size, then use that for the size of the sphere; if not, you can leave it as a parameter for Artemis to fit. If you were really confident you knew the size of the crystallites, then it might be worth it to do something more precise (e.g. use a magic number cluster), but that doesn't sound like your situation. In clusters of this size that are "roughly" spherical, I've found that the best-fit results aren't all that different if you just use a sphere, although the uncertainties in the results may be considerably higher.

3) Use Artemis to guess the fraction of Pt's having a surface interaction and start trying the different candidate interactions.

In advance, and knowing very little about your system, I'd warn you that the problem will be much more tractable by the method I describe if the bonding to the surface is somewhat random and floppy; i.e. if there is not a fixed orientation and distance of the Pt particle from whatever the substrate is. Then you can be more confident that the high-R part of the FT is entirely due to the Pt-Pt interactions, and can thus gauge the relative contributions of Pt-Pt vs. surface interactions more easily.

--Scott Calvin Sarah Lawrence College

...

Anatoly,

I did not consider magic numbers & don't know for certain what the exact # is. The 95 atom cluster is only an estimate of the size of the composition of particles that was observed in TEM images of Pt tethered onto a a surface with -COH, -C=O groups; the exact nature of the bonding is what I'm trying to determine. The Pt seems to be stabilized by its surface interactions; the exact nature of the binding is what I'm trying to figure out. The EXAFS data that I have "sees" both the Pt-Pt interactions as well as the Pt-surface interactions. I realize that it would be best to simple scan a sample containing Pt clusters of the same size as observed in the TEM; but, unfortunately, Pt has a tendency to agglomerate and it is difficult to get a monodispersed size in colloidal suspension in solution. Comparing with Pt foil doesn't work because of the quenching problems in the fluorescence since there are so many atoms.

I wanted to get a simulated EXAFS spectrum of something that just represents the bulk Pt-Pt only to compare with the data that I have. Differences in the spectra (in principle) would be attributed to Pt-surface interactions present in the TEM. I would like to try to get an idea of the # of Pt interacting with the surface as opposed to the Pt-Pt interactions. Charles

Hi Scott, Thanks for your suggestion. I find that I need to generate FEFF files for larger clusters than I had initially started off with. In your recent Physica Scripta paper [T115 (2005) 744-748], you had looked at some

large Pd clusters, particles 0.1 to 200nm in diam. (page 745, col 2,

Re: [Ifeffit] Artemis FEFF for clusters >500 atomsIn the articles by Montejano-Carrizale et al, Nanosctructured Materials, v.1, 397 (1992) and v.8, 269 (1997), analytical expressions are published for the 1NN coordination numbers in different geometries, including cuboctahedral, as a function of the cluster order L. However, to fit the entire r-range, one has to know how the 2NN, 3NN and multiple-scattering paths depend on the cluster order. Very similar to what Scott described, we also calculated these truncation parameters for SS and MS paths, for any given cluster order L, but they were not parameterized during the fit since it requires to know analytical expression N(L) where N is the truncation parameter and L is the cluster order. Instead, we just wrote several programs - coordination number generators - for different geometries: cuboctahedral, icosahedral, octahedral, truncated octahedral, and for each geometry and for each size we obtained the truncation parameters for parsing the x,y,z coordinates through another program that we wrote, that calculates the radial distribution function relative to the central atom. By EXAFS data analysis we obtained the coordination numbers for SS and MS paths model-independently and then compared them against several models of different sizes (L) and morphologies. Here are some references: J. Am. Chem. Soc., 119 , 7760 (1997) , J. Synchrotron Rad., 6 , 293 (1999). J. Phys. Chem. B, 105, 12689-12703 (2001). Anatoly -----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov]On Behalf Of scalvin@slc.edu Sent: Monday, May 30, 2005 9:08 AM To: XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] Artemis FEFF for clusters >500 atoms Hi Charles, We didn't use FEFF to create the really large clusters. Instead, we just generated a FEFF cluster larger than the largest paths we fit, and then used constraints to deal with all the termination effects. For the detailed modeling of cuboctahedra (it's been a while--that was the expected morphology, right?), we reduced the amplitude of each path by the weighted average of the reduction in coordination number relative to the bulk for the given morphology and size. A very clever high school student I was working with (C. J. Riedel) automated the process of generating these constraints. This approach is marginally less accurate than letting FEFF figure out an entire cluster, but doesn't require nearly as much computer power, and should be pretty darn close for large clusters. --Scott Calvin Sarah Lawrence College pretty para

2, lines 5-6). Some of these clusters are well over 500 atoms. How did you manage to create the FEFF? When I try to do so for a cluster this large in Artemis, the program flags an error saying that I have too many atoms (in my *.inp file generated from TkAtoms). Charles

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