Axial ligand-related issues in XAFS calculations

[Eason Jane]

Dear Community，

  I would like to ask a question regarding a recent calculation. After adding an O2 molecule as an axial ligand to an Fe-N4 structure, the calculated XAFS spectrum using CP2K shows an intense peak at ~6999 eV. Could anyone kindly provide some insights into why this happens? I have found that this specific feature appears whenever the Fe center is bonded to O or S atoms. 

  （ GLOBAL ： RUN_TYPE ENERGY    theoretical method：PBEh_ADMM with HFX of 45.000 % without RI-HFX    basis set and pseudopotential：DZVP-MOLOPT-SR-GTH     OT   GAPW     TD-DFT

   Fex ： POTENTIAL ALL  BASIS_SET：def2-TZVP）

  Thank you in advance for your time and help.

Eason

[Marcella Iannuzzi]

Hi Eason

What is exactly the problem. 

The output seems OK

Regards

Marcella

[Eason Jane]

Dear  Marcella

  After performing the calculation using FDMNES, I observed a small peak before the main peak (the black line). However, the spectrum obtained from the CP2K calculation (the blue line) shows a huge peak in the same region. This makes me suspect that the data for this peak might be incorrect, and I am also concerned that my parameter settings may be improperly configured. Therefore, I am seeking some assistance.  

  Thank you in advance for your time and help.

Eason

[Marcella Iannuzzi]

Hi

Have you tried to modify some settings in XAS_TDP like the GRID, the RI_REGION, the RI basis set?

Regards

Marcella

[Eason Jane]

Dear Marcella

  I have already tried (1) changing the GRID to CUTOFF 500 and REL_CUTOFF 60; (2) changing the GRID Fex parameters to 200, 300; (3) increasing the RI_REGION to 5.0. However, none of these adjustments resulted in any significant changes in the values.  

  Thank you in advance for your time and help.

Eason

[Nan Dong]

Hi Eason,

the GRID of Fe seems still small, you could check this https://doi.org/10.1021/acs.jpclett.5c02836 

It simulated Fe L-edge, and tested functionals (HF portions), basis-sets, RI REGION, and also GRID (LEBEDEV, RADICAL) in SI. 

Some factors influenced the shape of spectrum a lot. 

Best,

Nanchen 

[Eason Jane]

Dear  Nanchen

   After reading this  literature, I have conducted the following tests: (1) increasing the GRID Fex to 500 and 1000; (2) modifying the RI_REGION to 6.0; and (3) performing both modifications simultaneously. However, the resulting spectra do not show significant differences from previous ones. Subsequently, I attempted to change the functional (as shown in the figure), but the issue of a significant peak appearing before the main peak remains unresolved. Could you please advise on what steps I should take next?

  Thank you in advance for your time and help.

Eason

[Nan Dong]

Hi Eason,

have you tried to make the x-axis (energy interval) as the same with your FDMNES png. I tried with one of your CP2K results attached here. you can see you should have this pre-peak. If you plot them together, it could be more clear. The pre-peak is not a "problem" or "something wrong". 

Another option, with the same spectrum data, but use different broadening methods, and adjust the sigma. 

BTW, for the exact energy, TDDFT cannot give a correct excitation energy, so the rigid shift is needed. but the relative peaks positions should be fine. 

Best,

Nanchen 

[Eason Jane]

Dear  Nanchen

  Thank you for your guidance. I will look into adjusting it accordingly.

Best regards

Eason