Theoretical molecular chemist/physicist. Interested in finding stable geometries of molecules (mostly research chemicals) as a base for further research by other scientists. You can contact me if you need any specific molecule's geometry or its infrared spectra.
- What do I do here?
- I calculate the stable geometry of the molecules with AM1 semiempirical method, and B3LYP density functional approximation and upload my results here. These geometries can be used for docking tests in ligand binding to receptors or finding molecular characteristics such as IR absorption frequencies, etc.
- How accurate are these results?
- It is hard to say. Although AM1 method considers only the valence electrons, but its results are fairly good. B3LYP results are much more accurate but they need comparatively huge amount of computation time. A simple geometry search which takes ~1 min using AM1, will take ~50-100 hours using B3LYP method on the same machine.
- Why the molecular bonds/shapes look different than the skeletal formula representation?
- Skeletal formula predictions for bonds/shapes are not accurate. My calculations find the geometry which have the least energy (therefore is the most stable one). These geometries may or may not be the same shape as their skeletal formula as most the complex molecules are not planar.
- The proper method for finding the bond order is by NBO analysis. I'm using the bond order from my B3LYP calculations for all my molecules. Other than highly symmetric structures, the mesomerism (resonance) is not stable in molecules. Note that NBO is not possible in NDDO methods such as AM1.
- Are these geometries valid for molecules in water/SBF/blood/urine, etc?
- No! All my calculations are done on isolated molecules (basically molecules in low pressure gas phase). The geometries depend on the environment and do change in presence of solvents. But based on the property you are interested in, they are generally a good approximation or a first guess.
- Are these geometries the most stable molecular configuration?
- There's no guarantee that these geometries are fully relaxed or are the most stable one. These molecules have a high degree of freedom and I have not enough time to check for all stable geometries and find the most stable one. I'm just searching for local energy minima on potential energy surface with Berny algorithm using GEDIIS in redundant internal coordinate. If a molecule have several stereoisomers, my calculations only finds one of them which is similar to my initial guess.
- I start my B3LYP calculations from the geometry obtained by AM1 method.
- Also, I do not check for existence of any imaginary vibrational frequencies.
- Can people reuse these geometries for their own research?
- If you are going to do extra computations, just keep in mind that B3LYP implementation of Gaussian (VWN functional part) may be different from the one you are going to use in other computational packages.
- What is the Z-matrix?
- A method for representing molecular geometries. You can use that for you research or to make a better image of molecules (if you want to!). You can find more information on this format here.
- Which software do I use?
- I use Gaussian 09 for calculations and GaussView 5 for visualization. You can also use other free software which supports Z-matrix format like MOLDEN.
- Which basis set I am using?
- Unless stated otherwise:
- Polarized split-valence Pople-style Gaussian basis set (6-31G**).
- What are my convergence criteria?
- For AM1:
- Maximum Force: 0.000002 Ha = 0.00005 eV
- RMS Force: 0.000001 Ha = 0.00003
- Maximum Displacement: 0.000006 a0 = 0.000003 Å
- RMS Displacement: 0.000004 a0 = 0.000002 Å
- For B3LYP:
- Maximum Force: 0.000450 Ha = 0.012 eV
- RMS Force: 0.000300 Ha = 0.008 eV
- Maximum Displacement: 0.001800 a0 = 0.0010 Å
- RMS Displacement: 0.001200 a0 = 0.0006 Å
- Why I am not using more accurate method (larger basis set, other XC-functionals, solvents effect, multireference methods, etc.) for my calculations?
- More accurate calculations need lots of extra computation power which currently I don't have.
- Development and use of quantum mechanical molecular models. 76. AM1: a new general purpose quantum mechanical molecular model, J. Am. Chem. Soc., 1985, 107 (13), pp 3902–3909 doi:10.1021/ja00299a024
- Comparison of SCC-DFTB and NDDO-Based Semiempirical Molecular Orbital Methods for Organic Molecules, J. Phys. Chem. A, 2006, 110 (50), pp 13551–13559 doi:10.1021/jp064544k
ToDo, Bugs, Junk-yard
- Missing MediaWiki tags in templates imported from Wikimedia projects. e.g. <wm-license-self-one-license>, <wm-license-information-description>, etc.
- New version of MediaWiki must be installed or the current version of localization must be updated.
- Uploading image file with size bigger than 1MB leads to error "413 Request Entity Too Large".
- This is related to client_max_body_size directive in Nginx server configuration file.
- Austin Model 1
- Becke, 3-parameter, Lee-Yang-Parr
- Natural Bond Orbital
- For example, if you replace only one of the benzene ring hydrogens with a different group, most probably you will loose the stability of delocalized electronic structure.
- Neglect of Diatomic Differential Overlap
- Simulated Body Fluid
- GEometry optimization using Direct Inversion of the Iterative Subspace