I am fairly new at using this software and at computational chemistry in general. I achieved what is believed to be a good TS before running the IRC but we are repeatedly receiving the same "bad initial gradient" error. The verbose output says I need to loosen my tolerances but try as we might it still gives the same error. Any help would be greatly appreciated.
The error is a result of the convergence criteria for the transition state geometry optimization being much looser than the criteria for an IRC calculation. This means that what the TS optimization routine considers a stationary point doesn't meet the criteria for the IRC calculation to consider it a stationary point, hence the IRC routine telling you the initial gradient is bad, despite the fact you already converged it. The solution is actually to TIGHTEN the convergence criteria for your TS optimization, not loosen them as you were apparently trying. I recently solved this problem by reoptimizing my TS geometry with the keyword GRADIENTTOLERANCE=0.0001 (the default is 0.0007); hopefully that helps.
- Badges
- Science topic