EXAFS can detect bond length differences of ca. 0.01-0.02 A (and even lower; in specific cases, where differential procedures can be applied you can detect pm differences). Of course, the bond length resolution depends on you system; the above mentioned resolution can not be obtained when two shells differ by less of ca. 0.3-0.4 A. The inherent fitting error is, in such case, typically larger tan 0.01-0.02 A.

Resolution in the r space in EXAFS analysis also depends on the k range available. Remind that the Fourier Transform of EXAFS is related to the radial distribution function around the photoabsorber.

To my knowledge, the best resolution obtained in an EXAFS experiment is that described in this paper:

J. Synchrotron Rad. (2007). 14, 421-425 [ doi:10.1107/S0909049507028452 ]

Probing atomic displacements with thermal differential EXAFS

M. P. Ruffoni, R. F. Pettifer, S. Pascarelli, A. Trapananti and O. Mathon

by applying differential techniques as mentioned by Marcos above

Hope this helps

@ Marcos Fernández-García thahnks for informative comments. @Paolo Ghigna thanks a lot for the above important paper and informative comment.

I agree with the previous answers: the resolution will strongly depend on the method of analysis you use. Using the ab-initio simulations of EXAFS (i.e., programs like FEFF and GNXAS) for the interpretation of your data will give you accuracy of about 0.01 angstroms. Differential techniques can provide much higher resolution, but usually they will work for the analysis of the first coordination shell only.

I can recommend the following papers:

J. Purans et al, Isotopic effect in extended x-ray-absorption fine structure of germanium, Phys. Rev. Lett. 100 (2008) 055901:1-4.

J. Purans et al, Femtometer accuracy EXAFS measurements: isotopic effect in the first, second and third coordination shells of germanium, J. Phys.: Conf. Ser. 190 (2009) 012063)

@Janis Timoshenko: thanks a lot for the above important papers and informative comment.

One thing I would like to add, if you look at the above mentioned paper by Ruffoni et al, which is measured on a dispersive EXAFS beamline, I doubt that the effects are real. Since you only have a limited number of pixels on the ccd detector used to measure the spectrum as a function of energy (each pixel representing a discrete energy, more precisely an energy range) you may easily estimate how large the shift of the EXAFS fine structure is in energy if you vary the bond distance by some pm or even fm as suggested in this paper. You will see that the EXAFS spectrum only shifts by small fractions of the size of an individual detector pixel, and thus I doubt that this resolution is real. Sorry for my criticism, but I think this is an important issue.

Nevertheless, any mechanical monochromator has certain limits in resolution, reproducibility etc, and if two shells that need to be separated are close to each other, then as a consequence, the measured K-range (equivalent the energy range) must be increased. Of course this K-range is quite limited for a dispersive EXAFS system, while it can be extended over several keV in a scanning double-crystal monochromator.

Thus, in conclusion, you have to consider all these points if you want to judge the precision and accuracy of your experiments!

Regards to all, Dirk

I'd add that resolution, relative bond length, and absolute bond length are three different questions with different answers.

"Resolution" would be the possibility of distinguishing between having a pair of bonds at a single length and two bonds with slightly different lengths (e.g., a distorted local environment). As Paolo said, this kind of resolution is largely determined by the k-space you measure: the Rayleigh criterion gives delta R = 1/2 pi/(delta k), where delta R is the resolution and delta k is the k-range measured.

"Relative bond length" is the ability to detect a change in a bond length when the structure is otherwise similar; for example, an expansion with temperature or the effect of doping. Relative bond length can often be determined with high precision by EXAFS, although there's no single formula. Under 0.01 angstroms is not uncommon if data quality is reasonably good.

"Absolute bond length" is the ability to measure the length of a bond accurately. Because EXAFS models often have a variety of systematic errors, this is often not nearly as good as relative bond length, although for reasonably good models with reasonably good data, accuracies better than 0.03 angstroms are not uncommon.

You asked for the best, and I gave you typical values, but the distinction between the three kinds of length accuracy still applies.

Hallo Syed when you say 1/100 Angstrom I do not think this is too small with a large k range should be quite possible . Processing e.g. back transforming, using different k weightings can improve precision but not necessarily the accuracy.

@Dirk Luetzenkirchen-Hecht, Scott Calvin and David Batchelor: Thanks for you knowledgeable and informative comment.

@ Aria Tsam: thanks a lot for guide me toward important sites related to EXAFS.

A resolution lower than 0.001 Å can be obtained for the “relative” distances between nearest-neighbors atoms. As an example, a resolution better than 10 fm was reached in germanium (see J. Purans et al, Phys. Rev. Lett. 100, 055901 (2008)). Differently, for “absolute” bond distances, the resolution is typically about 0.01 Å or a little less.