Author Topic: Troubling results in Energy Descomposition Analysis (EDA)  (Read 5233 times)

romarine

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
Troubling results in Energy Descomposition Analysis (EDA)
« on: September 27, 2017, 01:26:38 PM »
Dear all,
I would like to perform an energy decomposition analysis (EDA) on a system with biological interest.

To know if I follow the procedure correctly I tried to reproduce previous calculations.
Turbomole manual (v 7.1, section 6.6) indicates the paper of Su and Li (J, Chem. Phys. 130, 014104 (2009)) for further details.

Taking an example of this reference I performed the EDA for the Cu+(SCH3)- complex at the BLYP/ACCQ//MP2/ACCT level: EDA at BLYP/aug-cc-pVQZ level using the CuSCH3 optimized structure at MP2/aug-cc-pVTZ level (the SCH3- fragment was not reoptimized), and compared the results with the Su&Li's work (Table V of the paper).

The comparison of the results is:
 
                   DEele   DEex  DErep   DEpol  DEdisp  DEtotal   (kcal/mol, where D means Delta)
My Calcs: -199.37    4.77  0.000    0.001  -12.40 -207.00
Su&Li:      -257.96 -46.95 190.97 -81.21  -12.27 -207.41
 
I am comparing polarization and dispersion (from paper) with orbital relaxation and correlation energy differences, respectively as it is commented in the paper (as I understand from points (2) and (4) of page 2 of the paper)

As you can see that my results reproduce the total and dispersion energy differences assuming that differences are due to the use of RI approx. in my calculations (what do you think?) but I am obtaining very different values in the rest of energy differences.

But this doesn't finish here. As I am interested in solvated systems, I performed EDA of the same system using COSMO model (EDA-COSMO) and comparing with the results in J. Chem. Phys. 137, 034111 (2012) where the EDA-CPCM analysis is performed on the same system at B3LYP-cc-pVTZ level (Table II of the paper, B3LYP/CCT results).

The comparison is:

                                       DGele   DGex  DGrep   DGpol  DGdisp  DGotal   (kcal/mol, where D means Delta)
EDA-COSMO,eps=78.4):  -84.96  25.80  0.00     -0.001  -9.16  -68,20
EDA-CPCM,eps=78.4):  -264.23 -60.99  203,92 -69,44  -9,99  -67,95

EDA-COSMO,eps=1.00) -214.54  21.92    0.00     -2.75   -9.11   -204.48
EDA-CPCM,eps=1.00):  -268.16 -60.36  205.23  -67.71 -10.22  -201.22 

I must confess that in this case I have to do a deeper bibliographical analysis to confirm that both methods are comparable (for example, are free energy data the turbomole's results in the case of EDA-COSMO?), but again, the comparison in dispersion and total energy difference values are the only reasonable, although the differences for epsilon=1.00 are the biggest.

Dear community,

Am I missing something important/evident?
Is it normal that repulsion energy difference is always zero in turbomole's results?
Is there other systems (computationally fast) I can compare results with?
Any other idea?

I am using TURBOMOLE V7.1, jobex -ri -level mp2 command for MP2/aug-cc-VTZ geometry optimization, and rift command for EDA method making the modifications on control files exposed in the manual.

Thanks in advance,

R.

romarine

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
Re: Troubling results in Energy Descomposition Analysis (EDA)
« Reply #1 on: October 04, 2017, 04:31:43 PM »
Problem solved.
I was missunderstanding the results and effect on control file of promowa module.
It was my fault.
The comparison with Su&Li is good.

Regards,
R.