I am trying to find a transition state for a nucleophile addition and my jobs never seem to converge. Calculating the force constants at the final geometry shows multiple negative frequencies, the largest of which is clearly the one I'm looking for. The other two imaginary frequencies are less than 100 cm-1. Is there a way to force Gaussian to converge to a TS in such scenarios? My current input line is: #p opt=(calcfc,ts,noeigentest, modredund) freq b3lyp/6-31g(d) scrf=(solvent=diethylether, PCM)
Helloo
Check this article for some inputs
http://onlinelibrary.wiley.com/doi/10.1002/jcc.540070402/full
Dear mate
The problem with Gaussian optimization algorithm is that once it finds a stationary point it will keep fixed on that position, sometimes the "noeigentest" keyword is not enough.
There is a possible solution. Open the output file that gave you several imaginary frequencies, and if you are using GaussView select one of the imaginary vibrations you do not want and distort the structure with the "Manual displacement" option. Move it from 0.00 to 0.75 or above, save that structure and relaunch the calculation with the same keywords, but changing "calcfc" for "calcall" (to calculate force constants in every step, making sure you are going through the vibration you want) and removing "freq" (vibrations are already computed when you ask for a TS, you don't need to specify again)
If you get to find the TS you want, with 1 imaginary frequency that's perfect. If you get several ones but less than in the previous calculation, repeat the procedure to "purify" the TS.
Hope you find it helpful
Hello,
I recommend trying the opt=QST3 method for finding the TS - in which case your input file should include the reactant, product and a guess for the TS.
If this does not work - consider increasing the accuracy by changing the basis set or functional.
Hi,
1. You can try to change the maxstep: # opt=(ts,calcfc,noeigentest,maxstep=N), where N defaults to 30.
2. Another solution is to optimize the structure with the key word "loose" # opt=(ts,calcfc,noeigentest,loose). if the geometry is ok, reoptimize it again using # opt=(ts,calcfc,noeigentest).
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry