Mod:Hunt Research Group/preferred method

preferred methodology

• the sequence of calculations to undertake for each molecule, ion-pair, cluster etc
• if you are not sure or just want to check, talk with Tricia

basic

• start by optimising at the B3LYP-D3(BJ)/6-31G(d) level

b3lyp/6-31g(d,p) empiricaldispersion=gd3bj int=ultrafine scf=conver=9

opt

this is for starting from scratch

if you already have a pre-optimised structure or a structure from a very similar molecule use that and go straight to 2

• then improve basis-set to 6-311+G(d,p) and optimise

b3lyp/6-311+g(d,p) empiricaldispersion=gd3bj int=ultrafine scf=conver=10

opt guess=read geom=checkpoint

note we increase the scf convergence

• carry out frequency analysis using 6-311+G(d,p) including property analysis

in this case transfer over the xyz coords so we have a record of the geom in the input *.com file

b3lyp/6-311+g(d,p) empiricaldispersion=gd3bj int=ultrafine scf=conver=10

freq=raman pop=(full,NBO7,dipole,chelpg) prop

• save frequency data to excel

then open the spectrum in GView and save it make sure

start point is 0 (not -500!)

end point is 3500,

record for 0.5 cm^-1 ie 7000 values

new key words "freq pop=(full,NBO7,dipole,chelpg) prop"

watch that chelpg has a radius for all your atoms!

• use goodvibes to evaluate Gibbs free energy

set limit to 100cm^-1

use Cramer and Truhlar approach

• generate a *.wfx file using the frequency job geometry/checkpoint file

b3lyp/6-311+g(d,p) empiricaldispersion=gd3bj int=ultrafine scf=conver=10

output=wfx density=current

make sure you have density=current otherwise you get the HF density!

key words "output=wfx density=current"

and at end of the com file give the wfx file name eg myfile.wfx

don't forget to add a command to save the *.wfx file in your BATCH runscript

• generate the density and ESP cube files

run the following interactively on the hpc

see wiki for instructions! and more details

but briefly key commmands are:

formchk *.chk *.fchk

cubegen 6 density=scf *.fchk *_density.cube 0 h

cubegen 6 potential=scf *.fchk *_esp.cube 0 h

note if you are comparing cube files ensure they have same sampling

confirm structures with other dispersion functionals

optimisation then frequency

M06-2X(main group) or M06-L(TM)

ɷB97-XD

m06-2x/6-311+g(d,p) int=ultrafine scf=conver=10

opt guess=read geom=checkpoint

more accurate vibrations

see the papers in "key papers" under Scale factors for vibrational frequencies DFT/MP2 on the wiki

large molecules B97-1/6-31G*

B97-1 is not available in orca so use PBE/6-31G*

mid range is B97-1/6-31+G*

best is B2PLYP-D3(BJ)/def2-TZVPD but is expensive!

NMR or UV-vis calculations

see Gaussian advanced on the wiki

both cases check the solvent being used

• for NMR typically B3LYP/6-31+G(d,p) gas phase followed by mPW1PW91/6-311+G(2d,p)
• for UV-vis typically cam-b3lyp/def2tzvp

you will need to test for the number of states to include ball-park: td=(nstates=60)

sp cam-b3lyp/def2tzvp empiricaldispersion=gd3bj int=ultrafine scf=conver=10

td=(nstates=60) geom=cartesian scrf(pcm,solvent=methanol)

more accurate energies

carry out single point energy analysis

note that properties do not improve as much so only compute energies

• G16 include correlation

B2PLYP-D3(BJ) note that empiricaldispersion=GD3BJ is NOT required for this functional, call with B2PLYPD3

MP2

basis set use aug-ccpVTZ is preferred

but can be too large, use ccpVTZ or maybe aug-ccpVDZ, its better to increase zeta than to add aug functions

• use ORCA for DLNPO-CCSD(T)

key words "DLPNO-CCSD(T) NoAutoStart VeryTightSCF RIJCOSX defGrid3 TightPNO"

higher level basis-sets "aug-cc-pVTZ aug-cc-pVTZ/C aug-cc-pVTZ/J)"

level and faster "def2-TZVPD def2-TZVPD/C def2/J"