2:36 PM
4

I'm attempting to optimize n-alkane dimers using MP2/DFT in Gaussian09. However the first one or two frequencies are always negative. Can these modes be ignored or do I need to do a tighter optimization?

6 hours later…
8:46 PM
3

I searched a lot, they always said you should do scan between two atoms and then fit with Lennard-Jones Equation, but they don't explain which atom should I select or depend on what should I select the atoms. I am a bit confused. Normally, pairs of atoms connected by chemical bonds are excluded...

You said "do a single point". You mean a single point without optimization each distance or what? Can't I do scan between the two atoms from d=1 and increase the distance by 0.1 or 0.3 and opt at the same time ? Two questions, 1- depend on what should I put the initial distance, because for example if I put the distance between Ru and H equal 1, it will be a bond. Will this effect on the energy ? 2- If I have another atom like 4 but in a different position, Should I do only one and do the fitting or I need to do for every possible atom and take the average ? — Abd-Elazeem Mohamed 8 hours ago
Single point: no geometry optimization. The new system should have only two type of atoms. These two atoms correspond to the two atoms in your real system that are no bonded. Starting the scanning from 1 is just to got a more complete data (don't worry if they are bonded). — Camps ♦ 7 hours ago
Sorry for that, I am still confused. I am using Gaussian, How can I do scan as a single point ? What I understood, I need to do single point for d = 1, d=1.3, d=1.6 etc. and use this data to plot the curve, and my energy should be like E(at d=1) - E(at d = max distance), E(at d=1.3) - E(at d = max distance) etc. Is it correct now ? — Abd-Elazeem Mohamed 7 hours ago
@Abd-ElazeemMohamed "single point energy" is just the energy calculated at a given geometry, without any attempt to find the minimum energy geometry. The scan keyword in Gaussian does a rigid scan, where the energy is evaluated at each step along some coordinate without additional optimization. It's also possible to do a relaxed scan with the `opt` keyword, but this wouldn't be needed in your case. — Tyberius ♦ 7 hours ago
@Tyberius I got it. I misunderstood. I thought he wants me to use modredundant. The last part of my question, Is the Delta energy will be the energy of each distance minus the energy of the last distance ? — Abd-Elazeem Mohamed 7 hours ago
@Abd-ElazeemMohamed that's right, you can subtract off the last energy to have the 0 line correspond to the energy of the totally separated case — Tyberius ♦ 5 hours ago
@Tyberius. Great. My last question, What about the charge and multiplicity of the calculation, For example if I will do the scan between Ru and H ? The charge will be the same as the whole complex, or depend on each atom ? — Abd-Elazeem Mohamed 4 hours ago
@Abd-ElazeemMohamed the collective charge of the whole system should remain the same throughout the scan. The multiplicity may vary, with it mattering most at intermediate distances; the spin should be known at short distances and at long distances the spin of the two atoms will be far enough where they won't interact (assuming you use an unrestricted method). — Tyberius ♦ 2 hours ago
@Tyberius You mentioned a really good point. In this case, I need to have two scans, one scan will be for short distance and another one will be for long distance. I can use the data from both scans without any problem ?. But if I put multiplicity 1 for short distance between Ru and H, it didn't work, and the program complain about the multiplicity. — Abd-Elazeem Mohamed 53 mins ago
@Abd-ElazeemMohamed Ru has 44 electrons and H has 1, so for a neutral complex there is a total of 45. This means the lowest spin state you could have is a doublet, since you can't have a singlet spin with an odd number of electrons.

1 hour later…
10:01 PM
So in this case, I will use only the doublet for the whole scan