Hi Alan,
     Looking at the real part of FT, I am convinced that O atom will justify for peaks at 1.8 and 2.3 A in sample 1. One of the possibilities for sample 2 to be different is, it has higher ss2 for Cd-O (as can be seen from lower amplitude of first peak). If Cd loading in sample 1 is lower than sample 2, then it makes sense to me that Cd is bound to higher affinity (and less disordered) sites on Sodium Titanate nanotube in the sample1. As the Cd loading increases, Cd starts going to lower affinity albeit more disordered sites, making Cd-O bonding more disordered than sample 1.
 
    Other possibilities include coexistence of O and Na in the first shell of sample 2, which might be interfering destructively and dampening the peak at 2.3 A). Have you tried that scenario? It might not be trivial to distinguish O bonding with high ss2 from O and Na in the first shell. But you can try that out by splitting the first shell at two different distances with smaller ss2 values.  
  It is not obvious to me that your data supports 2nd shell Cd-Ti interaction. In the fit you described, your E_not2 and ss_2 are very high. The amp_2 is highly correlated with ss_2, and amp_2 value is close to its error bar. Put together, these two parameters makes me suspicious of your 2nd shell fit. Your Chi data does not necessarily show clear Cd-Ti interaction either. In my opinion Cd-Ti interaction would result in high amplitude of oscillation in chi (and correspondingly strong peak in FT). However, things can behave differently in case of nanomaterials and you might have some contribution of Ti in your spectra. If this is a significant part of your manuscript, you will have to convince the reviewers.
 
  
Good Luck,
Bhoopesh

On Sun, Jan 9, 2011 at 11:43 PM, Alan Du <duji0003@ntu.edu.sg> wrote:
Hi Bhoopesh and Scott,

I should have given a description of my project. Yes Scott, the work is to investigate the binding mechanisms of aqueous cadmium onto sodium titanate nanotubes. Spectrum of Sample 1 and 2 obtained from merging 9 scans and 4 scans, respectively.

A quick check in Athena and, indeed, the white line of Samples are higher than CdO. I'm not sure the reason behind it though. It is likely that cadmium binds to the surface of substrate rather than inside the bulk. The lack of distinct peaks after 1.8 Å means that there are not many scatters around the absorber?

Bhoopesh, as requested, I have attached the real part of FT (http://img585.imageshack.us/i/ftreal.jpg/). I haven't got a chance to interpret them.

From preliminary fitting of Sample 1, the major and minor peaks at 1.8 and 2.3 Å could be described by a Cd-O path (CdO). This interests me because Sample 2 does not have a peak at 2.3 Å, meaning there is another single scattering path for Sample 2?.

The peaks at 3 Å were fitted with Cd-Ti path (CdTiO3). No multiple scattering paths used. The best fit goes something like this:

****************************************************************************
Independent points          =      13.166992187
Number of variables         =       8.000000000
Chi-square                  =    1534.709946959
Reduced Chi-square          =     297.021921317
R-factor                    =       0.000128095
Measurement uncertainty (k) =       0.000060423
Measurement uncertainty (R) =       0.004455442
Number of data sets         =       1.000000000

Guess parameters +/- uncertainties  (initial guess):
  amp             =     0.9242390   +/-      0.0509920    (1.0000)
  enot            =     1.3950420   +/-      0.5425380    (0.0000)
  delr            =    -0.0872060   +/-      0.0051060    (0.0000)
  ss              =     0.0113250   +/-      0.0008480    (0.0030)
  amp_2           =     0.2441320   +/-      0.1905250    (1.0000)
  enot_2          =    22.5261260   +/-      4.5990590    (0.0000)
  delr_2          =     0.2510860   +/-      0.0719080    (0.0000)
  ss_2            =     0.0274690   +/-      0.0128040    (0.0030)

Correlations between variables:
       amp_2 and ss_2       -->  0.9342
      enot_2 and delr_2     -->  0.9133
         amp and ss         -->  0.8865
        enot and delr       -->  0.8632
       amp_2 and delr_2     -->  0.3040
      delr_2 and ss_2       -->  0.2888
All other correlations are below 0.25

  k-range             = 2.000 - 9.000
  dk                  = 1.000
  k-window            = hanning
  k-weight            = 3
  R-range             = 1.000 - 4.000
  dR                  = 0.000
  R-window            = hanning
  fitting space       = R
  background function = none
  phase correction    = none
 

  R-factor for this data set   = 0.00270

***********************************************************************************

The above enot_2 is on the high side. I am not entirely familiar with the parameters yet. Are there other parameters I should worry about?


Cheers,

Alan J. DU
Nanyang Technological University, Singapore