Dear fellow researchers,
I have performed a transition state search for a suggested reaction coordinate at B3LYP/6-311+G** and was able to identify and confirm both the reactant and products along the reaction coordinate using frequency computation at the same theoretical level. The system consists of nearly 50 atoms (C, H and O) with 100/100 alpha/beta electrons. I've computed single point energies at B3LYP/aug-cc-PVTZ method also to get more accurate estimates of barrier heights.
I would like to know whether it is required to go up to the use post-HF method such as CCSD(T) for better accuracy or wouldn't there be much difference in energies given my system has no metal atoms. Furthermore how much computational time would I need to spent on such calculation?
I can only allocate maximum of 12-cores per job and 64GB of RAM
Thank you
Generally, regular CCSD(T) with cc-pVTZ can only calculate a system consisted of up to 20 atoms (if the system does not have any point group symmetry). Clearly, the CCSD(T) cost for your system is prohibitively high. If you need better result, I suggest using double-hybrid functionals. The DLPNO-CCSD(T) supported by ORCA is an alternative choice.
For organic systems, M06-2X works much better than B3LYP, though the cost is slightly higher.
Normally CCSD is not useful in reaction mechanism due to its extreme computational cost and scaling for large systems.
More DFT methods such as M06-2X can be used to verify your results with B3LYP.
Normally CC series are used for extreme small system in gas phase.
Generally, regular CCSD(T) with cc-pVTZ can only calculate a system consisted of up to 20 atoms (if the system does not have any point group symmetry). Clearly, the CCSD(T) cost for your system is prohibitively high. If you need better result, I suggest using double-hybrid functionals. The DLPNO-CCSD(T) supported by ORCA is an alternative choice.
For organic systems, M06-2X works much better than B3LYP, though the cost is slightly higher.
You can also use MP2 for better accuracy. RI-MP2 in Orca is quite fast for small molecules.
Eduardo Maurina Morais it was noted that MP2 overestimates barrier heights. Thats why I didn't chose Moller-Plesset methods in the first place. Would you be able to clarify me on the said fact?
Senal Dinuka My recommendation is of course just a generalization, since MP2 is even used as a benchmark in many research papers. But it's accuracy has to be analyzed for each individual case. I just recommended it because RI-MP2 is usually considered to be a good middle ground between coupled cluster methods and DFT.
Thank you Eduardo Maurina Morais for your sincere reply. I will further look in to this. As suggested by Zhaoxi Sun and Tian Lu I am proceeding with MO6-2X theory level. Given that I have already found the geometries of four systems I only need a good estimate of the energies in minimal computational time.
A note I would like to add is that M06 series do not necessarily outperforms B3LYP. The reliabilities of them require extensive tests on your system and related structures.
Releted tests can be found in literatures and I think you can get more insights from them.
Normally consistency between them indicates correct result and if inconsistency occurs you should test more functionals.
Phys. Chem. Chem. Phys., 19, 32184 (2017) is a good guidance of selection DFT functional for thermochemistry of organic type of systems. As you can clearly see, M06-2X outperforms B3LYP (with D3 correction) significantly.
It is well recognized that the MP2 has been completely out-of-date and should no longer be employed without special reasons (only for evaluation of some molecular properties such as (hyper)polarizability and NMR, the MP2 is still useful), since its cost is comparable with double-hybrid functionals, while its accuracy is even lower than normal DFT functionals (as long as the DFT functional you used was properly chosen). This point can be clearly seen in the test data (especially Table X) of J. Chem. Phys. 131, 174105 (2009).
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Quantum Physics
- Quantum Mechanics