Super Beton Wisdom

This question can be raised anytime when we use computational chemistry. Why? it is because computational chemisty contains ‘black’ parts of chemistry which probably not yet proofable or clearly correct. Easy example, the transition structure (TS). It is still controvercy whether it is really exist, exactly follows the way that we imagine or not? More than that, still many abundant questions on capability of what the computational chemistry is able to do or not possible to calculate.

So, it is not easy task to use computational chemistry for calculating the real case. The important point is selecting the correct method and basis set. For example, hybrid correlation B3LYP DFT (density functional theory) is very good to represent the existance of transition structure. It is very common method used to calculate TS, but it has lack correction in term of calculation the energy of TS and minimum energy pathways. When you need to do thermochemistry calculation, probably you have to compare with more expensive method and basis set (ofcourse you will sacrifice your time). There are some more better method, such as MPW1K or others which presents in the paper of “The Design of Density Functionals that are Broadly Accurate for Thermochemistry, Thermochemical Kinetics, and Nonbonded Interactions,” by Y. Zhao and D. G.. Truhlar published in Journal of Physical Chemistry A 109, 5656-5667 (2005).

Another example is calculating the dispersion forces for 2 dimers, just make an example two benzene. When you use PW91 or B3LYP with the basis set 6-31G(d), you will not find a perfect face-to-face parallel configuration. After doing frequency analysis, it will go to ‘transition state’ mean not occured in the real situation. But, when you use MP2 method, you will get it (as many papers present the result using this method). Now, which one is correct? With many functions, you will have more numbers of probability. As MP2 is more higher level that previous one, MP2 could find the things that the previous methods could not find it. And it means MP2 is correct? no! MP2 is well-known to have over-estimated value of energy. Again don’t believe directly to the values resulted by this method.

So, there are quite common strategy to ’say’ that what we said is approaching the correct-real situation. The easies way is compared with the real case. For example, you want to calculate the bond distance. You can compare with IR analysis that many papers are available for many molecules. The second is read the basic concept of the method. Actually for this case, there are no clear explanation. People usually know this method match for specific case by analyzing the tendency and its values. But, some method already built try to represent the specific phenomena, for example B-97D basis set which made to be used for weak bond correlation.

All above issues are the keys when you want to make a paper or presentation in good journal or symposium, e.g. Journal of Physical Chem. or Journal of Amer. Chem. But if you just use it for ‘brief’ data, I think B3LYP is good enough already. I ever used HF (the lowest rank method), and it can describe already the ratio composition that I want. The example at that time is reaction between CF₃-CH₂OH with CF₃-CF=CF₂.