Multiwfn official website: http://sobereva.com/multiwfn. Multiwfn forum in Chinese: http://bbs.keinsci.com/wfn. E-mail of admin: sobereva[at]sina.com

You are not logged in.

- Topics: Active | Unanswered

**I_was_a_baby****Member**- Registered: 2017-09-11
- Posts: 29

A Doubt about Dipole Moment of water molecule

By I_was_a_baby

first publish: 2017.10.16

modified: 2017.10.17

There is a simple test by Gaussian and Multiwfn to reaserch the difference of dipole moment of water molecule by two methods, (modified at 2017.10.17)relaxation and non-relaxation. All of computational details can be listed below:

Geo: B3LYP/6-311G* by Gaussian 09 D.01

Relaxation:

ground state: wB97XD/aug-cc-pVTZ td(nstate=10) by Gaussian 09 D.01

excited state: wB97XD/aug-cc-pVTZ td(nstate=10,root=M) [M=1-5] by Gaussian 09 D.01 with the key word of “density”

Non-relaxation:

1, wB97XD/aug-cc-pVTZ td(nstate=10,root=M) [M=1-5] by Gaussian 09 D.01 with the key word of “density=rhoci” (no way to obtain the result of ground state by non-relaxation)

2, wB97XD/aug-cc-pVTZ td(nstate=10) by Gaussian 09 D.01 with the key word of “iop(9/40=5)” then analyzed by Multiwfn. (the result of ground state by non-relaxation can be achieved)

The results of computations have been shown in the table:

Here are some issues about the differences of results between two methods

1, On the computation of ground state, the consequence is amazing that the difference between two methods is tiny. Why?

2, The difference of excited states of water molecule by two methods is large. In detail, in the first, second, third and fifth excited state, the result of non-relaxation is double of the one of relaxation nearly, except the fourth. why?

3, Why the results of non-relaxation are different by two software slightly?

4, The conclusion fails to be same as blog of "http://sobereva.com/227". The The reason, I suggest, can be that the water molecule is differ to the phenol, which the water molecule is too small or too relaxed. Is it right?

*Last edited by I_was_a_baby (2017-10-17 00:59:11)*

Offline

I feel difficult in understanding your description. I think what you want to refer is "difference between dipole moment of excited state and that of ground state" rather than "transition dipole moment", the relaxed and unrelaxed results are distinguished only for the former.

About point 3, the slight difference arises from trivial numercial reason, which can be simply ignored.

Offline

**I_was_a_baby****Member**- Registered: 2017-09-11
- Posts: 29

sobereva wrote:

I feel difficult in understanding your description. I think what you want to refer is "difference between dipole moment of excited state and that of ground state" rather than "transition dipole moment", the relaxed and unrelaxed results are distinguished only for the former.

About point 3, the slight difference arises from trivial numercial reason, which can be simply ignored.

I am sorry to make this mistake, and the errors have been modified. Please explain the problems once again. Thank you very much.

Offline

Which procedure did you use in Multiwfn to compute ground state dipole moment? In fact there are different ways of outputting dipole moment of the whole system. If grid based method is used, there must be very few difference with respect to Gaussian result since the the grid is finite.

The data in the table actually denote dipole moment of ground state (column 2) and different excited state (column 3~7), right? If yes, from the the excited state data, it can be seen that the non-relaxed density usually overestimated dipole moment, since further relaxation of electron density is not taken into account. However, the 4th excited state is an exception, to explain the underlying reason, you should compare corresponding relaxed and non-relaxed density difference map.

Offline

**I_was_a_baby****Member**- Registered: 2017-09-11
- Posts: 29

sobereva wrote:

Which procedure did you use in Multiwfn to compute ground state dipole moment? In fact there are different ways of outputting dipole moment of the whole system. If grid based method is used, there must be very few difference with respect to Gaussian result since the the grid is finite.

The data in the table actually denote dipole moment of ground state (column 2) and different excited state (column 3~7), right? If yes, from the the excited state data, it can be seen that the non-relaxed density usually overestimated dipole moment, since further relaxation of electron density is not taken into account. However, the 4th excited state is an exception, to explain the underlying reason, you should compare corresponding relaxed and non-relaxed density difference map.

Dear Big Doctor:

Actually, the data of ground and excited states by Multiwfn are analyzed by the way to compute the transition dipole moment. The procedure is that, after the computations of TD-DFT with iop(9/40=5) have been taken by 18th main module of Multiwfn, all of transition dipole moments between all excited states have been calculated. It can be found the dipole moments of all excited states.

It is correct that the data in the table actually denote dipole moment of ground state (column 2) and different excited state (column 3~7). So, in your opinion, should I take a series of ways to compute dipole moment?

Thanks again.

Offline

It seems that you used "5 Calculate transition dipole moments between all excited states" in main function 18 to simultaneously obtain dipole moment of ground state and excited state? In this case, the marginal difference between Multiwfn result and Gaussian result arises from the fact that even 9/40=5 is used, still not all configurational coefficients are outputted by Gaussian and then loaded by Multiwfn.

I don't understand what exactly is "take a series of ways to compute dipole moment".

Offline

**I_was_a_baby****Member**- Registered: 2017-09-11
- Posts: 29

sobereva wrote:

It seems that you used "5 Calculate transition dipole moments between all excited states" in main function 18 to simultaneously obtain dipole moment of ground state and excited state? In this case, the marginal difference between Multiwfn result and Gaussian result arises from the fact that even 9/40=5 is used, still not all configurational coefficients are outputted by Gaussian and then loaded by Multiwfn.

I don't understand what exactly is "take a series of ways to compute dipole moment".

Another question is why the results by Multiwfn is the same as that by Gaussian with "density=rhoci" nearly, but different from that by Gaussian with "density". and what is the discrepancy between that with "density=rhoci" and "density"?

Offline

I_was_a_baby wrote:

sobereva wrote:It seems that you used "5 Calculate transition dipole moments between all excited states" in main function 18 to simultaneously obtain dipole moment of ground state and excited state? In this case, the marginal difference between Multiwfn result and Gaussian result arises from the fact that even 9/40=5 is used, still not all configurational coefficients are outputted by Gaussian and then loaded by Multiwfn.

I don't understand what exactly is "take a series of ways to compute dipole moment".

Another question is why the results by Multiwfn is the same as that by Gaussian with "density=rhoci" nearly, but different from that by Gaussian with "density". and what is the discrepancy between that with "density=rhoci" and "density"?

Difference between relaxed density (default case of "density" keyword) and non-relaxed density (density=rhoci) can be found in this discussion: http://bbs.keinsci.com/forum.php?mod=vi … d&tid=5738

The non-relaxed density corresponds to the excited density directly evaluated based on the printed configurational cofficients and molecular orbitals. While the relaxed density requires additional calculation step (using Z-vector method), Gaussian support this but Multiwfn itself is unable to do that.

Offline