# Is it possible to do a Gaussian redundant scan with some fixed bonds?

I'm trying to do a redundant PES scan of a hydrogen abstraction in a Gaussian job while maintaining the rest of the system fixed. I need to scan the coordinate where H5 approaches to H6, maintaining the rest of the methane in place (that is, increasing the C1 H5 bond length). When I do a scan, the H6 just approaches H5. Is there a way so I can freeze everything except the coordinates of H5? So the distance between C1 and H5 increases and the distance between H5 and H6 decreases.

• +1. Welcome to our new community and thank you for contributing your question here! We hope to see more of you in the future!! I removed the footnote because the asterisk was a bit confusing and I think almost everyone here knows what PES means. If you want to define it you can write "PES (potential energy surface) scan". Welcome again! – Nike Dattani Jun 18 at 17:31

I think @romaichenko's answer covers everything. But for a full answer I believe a sample input file is necessary.

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

methane with H

0 2
C         -1.41531        0.37839        0.00000
H         -0.34531        0.37839        0.00000
H         -1.77197        0.64640       -0.97255
H         -1.77198        1.08664        0.71838
H         -1.77198       -0.59787        0.25418
H         -2.67803       -3.10703        0.79353

B 1 6 F
B 5 6 S 20 -0.1

! there has to be a newline at the end of input file


Here, replace the coordinates and the charge/multiplicity with your own coordinates. If you want atom 5 to move while keeping atom 1 and atom 6 fixed, then you would have to fix the distance between 1 and 6. B indicates bond and F requests Gaussian to freeze that coordinate.

The last line indicates that the bond (B) between atoms 5 and 6 are scanned (S) with 20 steps of size -0.1 Angstrom. In the image you posted in the question, the bond between 5 and 6 is long, so during the scan, the bond length would have to decrease which is why the step size is negative. Make sure that the number of the scan steps and the step size are consistent with the bond lengths of your system, because I cannot say what is the distance between 5 and 6 from an image. 20 steps of step size -0.1 Angstrom will reduce the bond length by 2 Angstrom.

Note that the coordinates of the other three H atoms on methyl group (2,3,4) are not frozen, so they can move during the optimization, i.e. this is a relaxed scan to some extent.

• Thanks for your answer but unfortunately its not working. I get an error during the optimization process that says: "New curvilinear step failed, DQL= 1.19D-02 SP=-7.15D-02. RedCar failed. Error termination via Lnk1e in /Applications/g16/l103.exe at Sun Jun 20 12:18:30 2021" Even using your input throws an error that says: "New curvilinear step not converged. Error imposing constraints Error termination via Lnk1e in /Applications/g16/l103.exe at Sun Jun 20 12:20:52 2021." – christian Jun 20 at 10:18
• @christian You shouldn't use my coordinates!! They are just dummy values I put to make a full input file. Are you using ROHF or UHF? What is your spin multiplicity? Also can you try the following? Remove X 1 F and X 6 F, and then put B 1 6 F before the B 5 6 S 20 -0.1 line. ( I have edited my answer with this) – S R Maiti Jun 20 at 12:13
• That worked! Using B 1 6 F correctly freezes the two atoms. The 5H can move towards H6 as I wanted. I use an unrestricted shell (UHF). Thanks! – christian Jun 20 at 15:50
• Don't use modredundant when you can do it with GIC! (You can do everything with GIC.) – Martin - マーチン Jun 21 at 19:51

It can be done using ModRedundant option in Gaussian (more on that here). You just freeze everything in space except for the 5th and 6th hydrogen.

However, in my opinion it would be better to perform a so called relaxed scan (a series of optimizations with the scanned coordinate fixed to certain values) as it yields better geometries for system states during the reaction.