Any computational chemists have any advice?

[gravenewworld]

I am trying to do a QST2 calculation on two geometry optimized structures that I obtained by doing DFT calcs. The QST2 calc runs fine and finds 1 imaginary frequency resulting in a transition state that appears reasonable. However, when I pick points to the "left" and "right" of the transition state structure and optimize them back down, I find that the 1 optimized strucutre matches exactly to the input strucuture I originally put in, even in total energy. The structure to the "left" of the transition state optimizes back down but results in a structure that is similar to the original input structure but is off by almost 6kcal/mol in energy to my original input structure. Are there any alternatives to the QST2 calc that can be done in order to find transition states between 2 optimized geometries? Are there any tricks I can do in order to get the geometry to the "left" of the transition state to optimize back down to my original input structure (energy wise)? BTW I am currently using gaussian to do the calcs.

-Thanks, GNW.

 

[gravenewworld]

I also just found a journal article by a guy from Yale who uses the B3LYP/6-31++G** basis set for calculations involving lithium containing organic compounds. Is anyone familiar with using this basis set and what kind of results did you get? Will this take much longer than a DFT calc (or any idea of how much cpu power this will chew up)? The reaction I am trying to optimize is the is a Biphenylene ring, which when introduced to Li2, opens up to give biphenyl. One structure I have tried to optimize is the Biphenylene system with 2 Lithiums exactly over the center the biphenylene, with 1 lithium on top and the other on the bottom.

 

[movies]

I think that basis set is pretty standard for computational chemistry. It's going to be a lot more intensive than a standard DFT calculation. The results will be better, but it's going to take longer and require more CPU power.

 

[gravenewworld]

Great, my worst suspicions came true. The computer I am using is a 10 year old Mac. The DFT calcs took several days, the 6-31g++** calcs will probably take a week :grumpy: .

 

[movies]

Yeah, sorry to break the news, but that's probably about right.