In regards to my research, I do something similar except with small peptides. My primary tools are Spartan '11 and Gaussian09. You have to use a "hierarchical" approach by funneling conformers down to higher and higher levels of theory.

Here is my protocol as follows:

Considering my peptides have quite a bit of flexibility, it is possible that the thermochemistry is influenced by higher energy conformers. For this case I derive the Boltzmann weighed thermochemical values and use that for my calculations. 

Let me know if you need any help with the details, I'd be glad to help you out.

Patrick

Hello Johan,

  There isn't a quick solution to your problem, especially for molecules that have many rotatable bonds. I agree with your conclusion that the DFT minimization is providing you with local DFT minima, which are directly depended upon the input conformation (and hence its initial location on the PES).

  In addition to this, different QM-based theories will provide different potential energy surfaces -- i.e. B3LYP, HF, MP2, PM3, etc. can give you different results depending on your molecule. This is especially true for molecules that have lots of weak intramolecular forces. Ultimately, you are going to have to validate why the structure you found is the global minimum. Furthermore, you should consider if you are looking for the gas-phase minimum or an aqueous minimum, which can be different.

 I would suggest that you proceed as you have been, and generate random conformations for input into QM/DFT calculations. You can be judicious about which ones to actually run, based on your previous calculations. Then validate the finding using experimental and/or higher theory levels.

  Off the top of my head, I could imagine one using MD simulations to explore the molecule conformational space. However, this can be difficult to do reliably, involving molecular mechanics model building and optimization.

Best regards,

Karl

Thanks for the answer. I have been using the option of conformer distribution Spartan 14 calculations MM, I know this is not the best way to get the minimum at the highest levels of theory but I can give you a shortcut to get these conformers may be minimal in DFT I mean calculating the energies and with these levels of theory and optimizing the most stable structure. Thanks a lot

Dear Johan,

In Gaussian, DFT keeps the electron density as the functional and calculates other parameters like energy, force constants, geometry, vibrations etc, by bring the molecule to the lowest energy state. We call this as stable state or optimised state. So the purpose of bringing the molecule is not just bring to the lowest state but for the reason to find out other parameters. 

Thanks for reply Xavier Mariasusai, I want to get the lowest state energy of my systems because I want to calculate some thermodynamic properties, and I think the variance in the results may be increased when you are in a local minimum instead a global one.

Hello Johan,

Sure, you know that DFT is more reliable than MM. Hence, you should use MM only for finding a proposal and it can be the initial guess in DFT theory. I hope than you are checking the frecuencies. The negative frecuencies mean local minimun and positive frecuencies mean global minimun. If you have all the geometries with positive frecuencies, then I would choose the structure with minimun energy.

Best Regards

Rody

Hallo, Rody!

 Your statement  that "The negative frecuencies mean local minimun and positive frecuencies mean global minimun" is incorrect.

 All local minima must have only positive frequencies (more exactly  the eigenvalues of the matrix of the second derivatives of the energy according to the coordinates - i.e. the Hessian matrix). The local minimum with the lowest energy is the global minimum. If a single vibration is imaginary (i.e. the frequency is negative) and all gradients are zero (or vanishing) you have obtained a saddle point of the energy hypersurface which corresponds to the top of the barrier between two stable configurations. Analogous structures with several imaginary frequencies are not interesting from the chemical point of view.

  Best wishes

                     mb

In regards to my research, I do something similar except with small peptides. My primary tools are Spartan '11 and Gaussian09. You have to use a "hierarchical" approach by funneling conformers down to higher and higher levels of theory.

Here is my protocol as follows:

Considering my peptides have quite a bit of flexibility, it is possible that the thermochemistry is influenced by higher energy conformers. For this case I derive the Boltzmann weighed thermochemical values and use that for my calculations. 

Let me know if you need any help with the details, I'd be glad to help you out.

Patrick

Thanks a lot Mr Batoon, It really helps

I am so glad I came across this post. I think I will implement Dr. Batoon's method

I like the Dr. Batoon's method. In fact, some time i was doing conformers in Materials software with MM but I tested some conformation with Semiempirical and the structure with minimun energy was different each. Maybe taking several geometries the problem is resolved.

Thanks

I think the above suggestions are very godd. To search the global minimun, I have one more question. Can I trust Spartan/Sybyl for the primary selection of conformations? Because Spartan/Sybyl doesn't perform optimization for each conformations, it merely rotate the bond without moving the atoms attached to it. Maybe one conformer has lower energy after an optimization, then the lowest 100 out of total 10000 conformers are actually not the lowest ones. I calculate something that looks like 3-methylpentane-2,4-dione, for which I think Spartan/Sybyl might be erroneous to provide the lowest energy conformer because they don't do optimization. Can I just trust the softwares ? How different could it be, to compare the results from the softwares to, those from a higher level calculation that (hypothetically) optimizes all the 10000 possible conformers ?

thanks ~

Hi Rina,

Sorry for the late response but, I can say that I trust my calculations, as we use them to determine Proton Affinity, Gas-Phase Basicity, Deprotonation Enthalpy, and Gas-Phase Acidity. The experimental values that we determine are reasonable and match quite well (especially when considering the possibility of other isomers and applying a Boltzmann weighted distribution). In the case of these studies however, I am only utilizing a 6-311+G(d) basis set at most on tetrapeptides ~50 atoms-70 atoms. 

You can also open the conformational search to include 1,000,000+ or however many you want as I do when I calculate longer peptides. Just keep in mind that this does not give you the global minimum, but increases your likelihood of finding it by narrowing it down to the possibly (not possible) lowest energy conformations.

how i can use gaussian

i have probleme i want otimise my structre but always the computur told me the memory 64 G =Ram can you help me