## page was renamed from AJMGrpOnly/teaching/electronic-structure/practical-dft-S66 ## page was renamed from ajm/teaching/electronic-structure/practical-dft-S66 <> Navigation: * AJM Main * [[ajm/teaching/electronic-structure|ESM]] * [[ajm/camcasp/start|CamCASP]] = DFT : evaluation using the S66 database = This exercise is strongly related to [[ajm/teaching/electronic-structure/argon2-pes-dft|the previous one on Ar$_2$]], only here you get the chance to perform calculations on somewhat more interesting systems from the S66 data set. The S66 data set is a comprehensive set of intermolecular complexes that display a wide range of types of intermolecular interactions. It has been put together by [[https://pubs.acs.org/doi/10.1021/ct2002946|Rezac, Riley and Hobza in 2011]] and can be found online at [[http://www.begdb.com/|the Benchmark Energy and Geometry DataBase (BEGDB) site]] run by Pavel Hobza and co-workers. In this exercise you will go to that site, find the S66 set (search for //S66: A Well-balanced Database of Benchmark Interaction Energies Relevant to Biomolecular Structure//) and select dimers from the following molecule pairs: * 01 Water ... Water * 02 Water ... MeOH * 03 Water ... MeNH2 * 31 Uracil ... Ethene * 59 Ethyne ... Water (CH-O) * 66 MeNH2 ... Pyridine Choose one or two sets from here. If you'd like to try something else, ask me first as the computational cost of some of these could be high. You will see that for each set of molecules, Hobza et al. have provided reference interaction energies computed using a variety of methods. The first task you have is: {{{#!wiki note Note Read the Hobza S66 paper and the BEGDB page to figure out which of the reference energies are relevant to you. In short, which one/s should you use? Basically what you need to do is understand (to some extent) what all these methods mean and choose one or more from them. }}} == Single-point calculations == Having decided on your system/s and having got the reference energies, find the complex geometries from the database. They will be in the form: [[attachment:02 Water..MeOH]] {{{ 9 O -0.525329794 -0.050971084 -0.314516861 H -0.942006633 0.747901631 0.011252816 H 0.403696525 0.059785981 -0.073568368 O 2.316633291 0.045500849 0.071858389 H 2.684616115 -0.526576554 0.749386716 C 2.781638362 -0.426129067 -1.190300721 H 2.350821267 0.224964624 -1.943414753 H 3.867602049 -0.375336206 -1.264612649 H 2.453295744 -1.445998564 -1.389381355 }}} This is an XYZ file and if you save it in a file ending with '.xyz', say **02_water_MeOH.xyz** then you will be able to display the contents using Avogadro or JMol. The BEGDB uses a fixed format for this file: * 9 = number of atoms in the complex (3 for water and 6 for Methanol) * Blank line (usually some descriptors are present here, but BEGDB keeps it blank * ATOM Rx Ry Rz: atom label and coordinates in Cartesian form in Angstrom. The order of the atoms is not fixed in an XYZ file, but BEGDB use the convention: * Molecule 1 atoms first (in this case H2O). * Molecule 2 next (CH3OH) Using this convention it is easy to split the XYZ file into the coordinates for two molecules. Now, using the Ar$_2$ example as a guide, write out the NWChem (or Psi4 if you know how to use it) command file for the systems you have chosen. When doing this, pay attention to the following: - Basis set: which one should you use? - Functional: Which one/s? - Dispersion correction: this is essential, but again, you need to make a choice. If unsure, try a few. - LC correction: We will see in the lectures that the LC correction can be quite important. What do you find? {{{#!wiki note Note Construct the NWChem command files for your system/s and method/s. Run them. Compare with the reference energies from the BEGDB. Tabulate the data appropriately. Use the appropriate units. Make an analysis. }}} == S66 dissociation curves == Single-point energies are sometimes misleading as a method may appear to be very good for a particular geometry, but it may exhibit the wrong //functional form// when used to compute energies on a surface. This is the reason Hobza et al. have put together the S66x8 dataset of S66 dissociation curves. Basically this consists of 8 geometries for each of the S66 complexes. So you can plot a curve. Find the S66x8 dataset on the BEGDB site (search for //S66x8: Dissociation curves for the S66 dataset//). For each complex you will see 8 geometries that are indicated like this: [[attachment:Water..MeOH S66x8]] {{{ Complex Optimization CCSD(T) CP/CBS 02 Water ... MeOH (0.90) MP2/cc-pVTZ CP -5.19 02 Water ... MeOH (0.95) MP2/cc-pVTZ CP -5.55 02 Water ... MeOH (1.00) MP2/cc-pVTZ CP -5.57 02 Water ... MeOH (1.05) MP2/cc-pVTZ CP -5.38 02 Water ... MeOH (1.10) MP2/cc-pVTZ CP -5.08 02 Water ... MeOH (1.25) MP2/cc-pVTZ CP -3.94 02 Water ... MeOH (1.50) MP2/cc-pVTZ CP -2.39 02 Water ... MeOH (2.00) MP2/cc-pVTZ CP -0.95 }}} The number in parenthesis indicates the scaling used compared with the equilibrium (1.00) configuration. {{{#!wiki note Note Compute the interaction energy using your method of choice for each of these complex geometries. Tabulate neatly with appropriate units. Plot the results and compare with the reference energies. What do you conclude? }}}