Multiwfn forum

may01dz

Member

Registered: 2018-07-17

Posts: 54

aromaticity index for Porphyrin-like systems

Hello,

1/Which aromaticity index is appropriate for Porphyrin-like systems?

2/
I used ORCA and the corresponding input below to calculate the magnetic shielding tensor at a ghost atom H: 1 A° above the ring plan (-0.0676346525,0.0191628382,-0.8365329255). the order of the ghost atom is 95.

the structure of the input is:

%pal nprocs 10 end                                         
%maxcore 3675                                               
! B3LYP def2-TZVP def2/J def2/JK TightSCF NMR RijCosX GridX4
* xyz 0 1                                                   
C   -2.82353463539941     -1.13724734028106      0.44846883140457                                   
C   -4.18974744461116     -0.69584407152261      0.53629948317338
.
.
.
H: -0.0676346525          0.0191628382         -0.8365329255 # This H atom is defined as a ghost atom
*                                                 
%EPRNMR                                           
     NUCLEI = 95 {SHIFT}                         
END

Somehow, ORCA outputted only the shielding tensor of the last real atom (94H) and has not shown up that of 95H (the ghost atom).

On the other hand, although I asked the program to calculate the magnetic shielding tensor of one atom, the time spend to do that was not less than that it spend to calculate the magnetic shielding tensor of all the atoms of the molecule, Does this make sense?
thank you very much

Last edited by may01dz (2021-05-25 17:58:11)

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,638

Website

Re: aromaticity index for Porphyrin-like systems

I am not familiar with this feature of ORCA. What I can say is that multicenter bond order (also known as multi-center bond index) and AV1245 in Multiwfn are suitable for investigating aromaticity of porphyrin-like systems. The ICSS supported by Multiwfn is very useful in visually studying aromaticity of this system, though the cost is high.

may01dz

Member

Registered: 2018-07-17

Posts: 54

Re: aromaticity index for Porphyrin-like systems

well, thank you very much

I'll post my inquiry in the ORCA forum, then I'll share the reply here (If you don't mind).

I am not familiar with this feature of ORCA. What I can say is that multicenter bond order (also known as multi-center bond index) and AV1245 in Multiwfn are suitable for investigating aromaticity of porphyrin-like systems. The ICSS supported by Multiwfn is very useful in visually studying aromaticity of this system, though the cost is high.

may01dz

Member

Registered: 2018-07-17

Posts: 54

Re: aromaticity index for Porphyrin-like systems

Hello,
The following are the responses I received from the Orca Forum's leaders:

For the second question (ORCA outputted only the shielding tensor of the last real atom (94H) and has not shown up that of 95H (the ghost atom).) the answer is as follows:

One of the many idiosyncrasies of the EPRNMR module is that counting for the NUCLEI keyword starts at 1, unlike most of the program. This is noted in the manual. So 94H is the 95th atom in the molecule. You can ask for atom 96, but it's probably easiest to just let the program calculate all shieldings - as discussed, the calculation time is basically the same.
As for the basis set on the ghost atom: it is not wrong per se to use a regular basis like def2-TZVP - indeed, it gets the globally assigned basis for hydrogen (or whatever element the ghost atom is). However, it does introduce a basis set error if you compare multiple calculations which only differ in the position of the ghost atom (or lack thereof). Since def2-TZVP is probably not at the CBS limit for shieldings, this error may be significant.
In other words, the basis set of the ghost atom is added to the basis set of the system, but what you probably want is to probe different spatial positions of the system with its basis set unchanged. You can do that with a dummy atom (X) but then the problem is that the latter has no grid points, so for DFT at points just a couple of Angstrom away from real atoms, you get totally wrong results. Hence, the workaround of using ghost atoms (which do get grid points) but giving them a practically non-interacting basis of a single tight s-function (and corresponding aux functions, if necessary).

the answer on - although I asked the program to calculate the magnetic shielding tensor of one atom, the time spend to do that was not less than that it spend to calculate the magnetic shielding tensor of all the atoms of the molecule, Does this make sense?- the respond was:

Most of the time is spent in the CPSCF equations that are done for the x,y and z component of the magnetic field.
The shielding tensor is a second derivative, so one could formulate it to first take the derivative with respect to the nuclear magnetic moment and then wrt to the magnetic field. Then you would benefit greatly in performance if you have only one nucleus (see work by Ochsenfeld et al who report this "sublinear scaling") but as soon as you would have more than just a few nuclei (as you would typically for 1H NMR) you would need much (!) more time. This is why most codes first do the field derivative, and also why you hardly save any time if you only need few nuclei as the second derivative is comparably cheap after the CPSCF equations have been done.