I have been running excited state (Mostly TD) geometry optimization in Gaussian or Q-chem. Most of my compounds have exclusive HOMO -> LUMO (amplitude^2 ~ 1) at excitation state 1 (B3LYP). But when it comes to CAM-B3LYP, many transitions comprised 1 excitation state, usually I can still identify the state I want for optimization.
The thing is, during optimization, let's say state 3, the excited states crossed. So in the half-way of optimization, previous state 3 could become state 2, and I ended up optimizing a different state 3, which was previously state 2. Then the new state 3 was not the state I am interested in (comprised different linearly combined transitions)
Is there any commands that I can try to keep the running "locked to" to specific transitions?
Dear Bryan Po-Wen Chen ,
When you perform a TD-DFT optimization with Gaussian (I haven't used QChem, but it might be the same situation), you do not optimize the excited states according to their energetic ordering, but to their electronic nature (this is, to the character of the excitation).
For instance, assume that you have a molecule for which the S1 state is of np* character (where p means pi) in the Franck-Condon (FC) region, and the S2 is a pp* state. If you want to relax the pp* state, you say to Gaussian "optimize the S2 state of this structure", but it really means "optimize the pp* state". Therefore, the program starts following the steepest descent pathway of the pp* state, and moves out of the FC region. If this pathway leads to a point where the np* and pp* states invert their energetic order, the program does not care at all, because it has been instructed to follow the pp* state.
So, don't worry about state reorderings because they are normal and expected for TD-DFT in these packages (conical intersections are the essence of photochemistry). At least for Gaussian, you can check in the output file that you are still optimizing the state you selected at the very beginning. Just look for the string This state for optimization and/or second-order correction, it should be assigned to the excited state that originally was your S3, despite it no longer is the third in energy.
Hope you find it helpful
Thanks, Enrique Manuel Arpa, but does that mean I can manually grab geometry coordinations right after state crossing and start a new geometry optimization with different state of interest input, until reaching geometry optimization criterium (such as energy change).
I know there's something like multireference approach for excitation simulations, but so far Im incapable of doing that. So I might do whatever Im capable of, and after concluded from preliminary data, I shall proceed to console simulation ppl.
Bryan Po-Wen Chen ,
In principle yes, you can locate the crossing point on the intermediate steps of the optimization and try to optimize both steps from that point, so you can locate the minimum of the original state and the minimum upon internal conversion. Just to remind, most likely that crossing will not be the minimum energy crossing point, that one should be located by other means.
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Powders