I want to find transition state (of maximum energy point along the reaction path) for certain chemical reaction.

I fully relaxed the reactant and product structures and make an initial guess reaction path using reaction preview.

I tried the three methods to find that

1. TS search: When I tried this method, it often fails to find the transition state (convergence issue or imaginary frequency issue) when I increased the max number of QST steps to 20. Even if it find the TS, it does not exhibit one imaginary frequency.

    

2. NEB -> TS optimization: successful converging of NEB calculation can be done. So, I picked up the maximum energy snapshots (TS candidates) along the optimized reaction path. Then, by calculating hessian matrix, I tried TS optimization task to refine the structure to TS. 

   1. Regarding this method, I tried with the maximum energy structures. From my knowledge, the TS optimization should be constrained optimization (only along the NEB-optimized path). But, it just uses frequency information of TS candidate. I want to know this NEB->TS opt method is theoretically correct or not.

       

3. NEB-> TS run(Minimum Energy Path): This method is just the MEP calculation using "full path" mode. It can calculate TS, but sometimes, this method might not use the highest energy snapshot to find the TS. So, it just gives the wrong TS and also does not provide frequency analysis information. So, I can't check the structure is TS or not. 

Can you give an advise for this problem?