Dear all,

I have questions on the EXAFS fitting of a spectrum of solid AuCl.
I find this quite important to resolve because AuCl is so widely used in the literature as a standard reference compound.
I obtained the spectrum from the 2013 XAS workshop in Diamond Light Source. It is one of the reference spectrum in the cyanobacteria tutorial.
The XANES in the spectrum looks similar to previously published XANES of AuCl.

Lengke, M. F. et al., Can. J. Chem. 85: 651-659 (2007)

Each Au in the compound has two Cl atoms attached to it, as seen in the crystal structure below. So I fitted the spectrum using two scattering paths:

(i) the Au-Cl single scattering, and

(ii) the forward through absorber (Au-Cl- -Cl-Au) multiple scattering path.

In the fitting model, I fixed So2 (or amp) at 1. The variables that I used for the paths are:

(i) N, drcl, sscl, dele;

(ii) N = 2, drcl*2, sscl*2, dele.

One would expect N to be 2, or perhaps even a little lower because amp = 1 is quite high,  but instead, I get N = 2.35, significantly higher than expected.

Reduced chi-square : 1706.6480592
R-factor : 0.0222459
guess parameters:
N = 2.34968550 # +/- 0.18385678 [1.00000]
dele = 0.82469594 # +/- 0.86953191 [0]
drcl = -0.02909926 # +/- 0.00608062 [0]
sscl = 0.00349927 # +/- 0.00072012 [0.00300]

set parameters:
amp = 1.00000000

Fixing the coordination number N at the expected value of 2 leads to unrealistically high amp values, as one would expect because of the strong correlation between the two parameters.

Next, I tried adding the third and fourth cumulant terms to perhaps account for anharmonic vibrations.

The variables for the two paths are now:

(i) N, drcl, sscl, dele, third, fourth;

(ii) N = 2, drcl*2, sscl*2, dele, third*2, fourth*2.

amp was still fixed at 1.

But that doesn’t seem to improve matters much, N is similar to the value found without cumulants.

Reduced chi-square : 1114.1247832
R-factor : 0.0106796
guess parameters:
N = 2.32392798 # +/- 0.21444748 [1.00000]
dele = -2.54795042 # +/- 1.27479319 [0.82470]
drcl = -0.07137580 # +/- 0.01429137 [-0.02910]
sscl = 0.00338118 # +/- 0.00154424 [0.00300]
third = -0.00056160 # +/- 0.00017873 [0]
fourth = -0.00000155 # +/- 0.00003392 [0]

set parameters:
amp = 1.00000000

So I have two questions here:
1.  Why is the EXAFS amplitude overall so high? The only possibility I can come up with is that the material is actually a mixture of AuCl3 and AuCl. The bond lengths are very similar and it is well known that AuCl is quite unstable in aqueous solution. I have seen the conversion of [AuCl2]-  in solution to Au(III) in my own XAFS experiments.

2.  I have tried to include a Au-Au first shell scattering path. But there doesn’t seem to be evidence for Au-Au scattering in the data. If the sample was pure crystalline AuCl, the first Au shell would be at 3.2 Å from the absorber, but I cannot fit such a path to the data.

So I am wondering whether the AuCl spectrum is unreliable...

Given how many times AuCl XANES has been published before, has anyone else perhaps run into this problem? Has anyone perhaps measured solid AuCl under exclusion of moisture and air, or several times as a function of time to assess beam damage?

Does anyone perhaps have suggestions what else may explain the high EXAFS amplitudes and the absence of Au-Au scattering…?

Attached are the Athena, Artemis and .cif files. I used Demeter 0.9.18 with perl 5.012003 and using Ifeffit 1.2.11d on Windows 7.

Thank you for your attention.

Best wishes,
Sin-Yuen, Chang
PhD Student
School of Chemical Engineering and Analytical Science
The University of Manchester