# Do changing opt=modredundant to opt in Gaussian makes geometry optimization not to take into account frozen angles?

I am a newbie to Gaussian and just generated an input for the geometry optimization for some molecules with multi ring system. However, in the article that was a reference for those calculations, some of the dihedral angles were freezed, so I did the same with adding the needed dihedrals to constrain with lines D X X X X F.

However, someone superior told me that I shouldn't use the opt=modredundant, but opt instead. Does it mean that I shouldn't constrain angles then? Or should I just manually change it to opt and the program will still understand that the angles are constrained?

• Have you looked at the Gaussian manual (which includes a description of the keyword)? Did this person say why you should not use modredundant? Dec 7 '20 at 10:37
• "Does it mean that I shouldn't constrain angles then?" Of course not, if you wan to recreate the result in the original article. However you may then want to try again without freezing coordinates. Why were they frozen in the original article in the first place? Dec 7 '20 at 10:42
• @BuckThorn, thank you so much for the comments. Angles were frozen due to the fact that this molecule is part of a protein (i. e. nonplanar distortions that are being caused by it), so to ensure the overall structure of these molecules and their alignment, wouldn't change so much, I believe. However, we are trying to make this work with very similar system, so it's not exactly about recreation and more about getting new results. Dec 7 '20 at 10:49
• Well, as I mentioned before, if you want to recreate the literature results impose the constraints. If you want to explore more of the energy surface, don't. This goes beyond technical aspects and is a question of opinion which might be impossible to answer without knowing more about your project. Dec 7 '20 at 10:54
• @BuckThorn, Yes, of course, I have looked into the manual. It is written like this: "Except for any case when it is combined with the GIC option (see below), the ModRedundant option will add, delete, or modify redundant internal coordinate definitions (including scan and constraint information) before performing the calculation. This option requires a separate input section following the geometry specification". My guess would be that deleting the modredundant keyword would mean that the angles wouldn't be constrained, but I am not sure, so that's why I am asking this question here. Dec 7 '20 at 10:54

The following illustrates the use of modredundant in a simple optimization (water angle at fixed bond lengths).

Input:

# B3LYP/6-31G(d) Opt=ModRedundant

test to see how to freeze coordinates

0 1
H     0.0 0.0 0.0
O     1 1.0
H     2 1.2 1 105.0

1 2 F
2 3 F


Optimized internal coordinates:

                           ----------------------------
!   Optimized Parameters   !
! (Angstroms and Degrees)  !
--------------------------                            --------------------------
! Name  Definition              Value          Derivative Info.                !
--------------------------------------------------------------------------------
! R1    R(1,2)                  1.0            -DE/DX =   -0.0225              !
! R2    R(2,3)                  1.2            -DE/DX =   -0.1043              !
! A1    A(1,2,3)               99.0431         -DE/DX =    0.0                 !
-------------------------------------------------------------------------------


Note the angle was not constrained so the final value differs from the input.

Now change the route to

# B3LYP/6-31G(d) Opt


and the output changes to

                       ----------------------------
!   Optimized Parameters   !
! (Angstroms and Degrees)  !
--------------------------                            --------------------------
! Name  Definition              Value          Derivative Info.                !
--------------------------------------------------------------------------------
! R1    R(1,2)                  0.9687         -DE/DX =    0.0                 !
! R2    R(2,3)                  0.9688         -DE/DX =    0.0                 !
! A1    A(1,2,3)              103.6489         -DE/DX =    0.0                 !
-------------------------------------------------------------------------------


So sure enough, removing modredundant loosens restraints under the optimization.

• Thank you very much for the help! Just tried it myself with $H_2O_2$, so ensured that keyword is a must for constraining, despite the additional input of constrains (just was thinking there is somehow more complicated methodology to this). Sorry for bothering with relatively trivial matters. Dec 7 '20 at 11:20
• @aerospace No trouble, and it's not entirely trivial. I just happened to have done this before. But there might be other solutions. I don't claim modredundant is the only way to constrain, someone else might have another answer. Dec 7 '20 at 11:55