Key Words

  • I have had feedback that some of you would like more information on what the keywords mean. This section is for especially for you!
  • In a research environment, we don't use the graphical interface, we create and edit the filename.gjf files directly. What the code does is determined by key words given on a specific line of the input file
  • You can view the filename.gjf file by clicking on the file and allowing it to open in gaussview, then choose results and view file, a window will open with the input file
  • Here is one of my input files for a job: 
    %chk=nicl2ph32_sp_cis_631gd_eclipsed_opt.chk
# opt b3lyp/6-31g(d) geom=connectivity int=ultrafine scf=(conver=9)

NiCl2PH32 square planar cis PH3 eclipsed 6-31G(d)

0 1
 Ni                 0.00000400    0.06169800    0.00000400
 Cl                 1.58734700    1.48893600    0.63664400
 Cl                -1.58724100    1.48901200   -0.63665400
 P                 -1.65783500   -1.39475400    0.52293600
 H                 -1.36963800   -2.70844600    0.99652300
 H                 -2.38275400   -0.88873500    1.62817500
 H                 -2.70724500   -1.65687100   -0.39472300
 P                  1.65772300   -1.39486100   -0.52292300
 H                  2.70712900   -1.65713700    0.39470500
 H                  1.36953600   -2.70847900   -0.99674200
 H                  2.38273200   -0.88875900   -1.62807700

 1 2 1.0 3 1.0 4 1.0 8 1.0
 2
 3
 4 5 1.0 6 1.0 7 1.0
 5
 6
 7
 8 9 1.0 10 1.0 11 1.0
 9
 10
 11
  • So what does this information mean?
    • The first line, is telling the computer what to name the checkpoint file, this is a file where all key information used in running the job is stored, this file is in binary and not ascii so it cannot be read by a human unless it is reformatted
    • The next line starts with a "hash" this is the key words line and this is the line that controls the job, in the graphical interface you click on a button and a key word is added to this file. I will outline the basic keywords shortly after describing the rest of the file. This section is finished by a blank line.
    • Then comes a line with the title you entered and then another blank line. NOTE that blank lines mean something, they terminate a section of input; this is a "formated" input file
    • Then comes the charge and multiplicity, these are obvious, don't forget that multiplicity is (2S+1) where S is the total spin, so if S=0 for a spin paired system then the multiplicity is 1 and we have a singlet. This section is finished by a blank line.
    • The next part defines the molecule, it is given by the atomic name (or number) and then the x, y and z cooridinates of that atom. This section is finished by a blank line.
    • Last is the connectivity data, reading across my file we can see that atom 1 is connected to atom 2 by a single bond, it is also connected to atome 3, atom 4 and atom 8 by a single bond. Atom 2 on the next line is only connected to atom 1 which has already been defined and so there is no extra data on this line. We can see that the P must be atom 4 and atom 8 because they are connected to three other atoms which are the hydrogen atoms. This section is finished by a blank line.
  • More key words Actually there are many many keywords we could choose from, here is the gaussian manual of kew words: G09 key words
  • B3LYP this is a particular type of method, you will need to wait until you have learnt some more quantum mechanics before you can understand the different types of methods and the various flavours of DFT ... just trust me for now
  • UB3LYP. You might also notice that in the calculation summary the method is identified as UB3LYP:
    • The U stands for an "unrestricted" spatial orbtial calculation. A MO wavefunction is made from two components a spatial part (which can be further split into a radial and an angular component) and a spin part
    • In spin paired systems the correlation between all the electrons of alpha spin and then all those of beta spin is the same because there is the same number of alpha and beta spin electrons. The spatial part of the orbitals is thus the same for both electrons and we "pair" them up in a molecular orbital.
    • However, when there are unpaired spins as in a doublet or triplet configuration there will be more electrons of one spin type, lets say alpha spin. This means that the correlation between the alpha spin electrons will be different from that between the beta spin electrons, this can have a large impact on the spatial part of the orbitals
    • In this case we can no longer pair electrons in the same spatial orbital, each electron has to have its own spatial orbital, thus the spatial orbitals are unrestricted. We will go on to consider this in more detail shortly.
  • 6-31G(d) is the basis set, we have already talked about what this means.
  • geom=connectivity this means use the connectivity data at the end of the input section to help build the molecule
  • geom=cartesian this means just use the xyz coordinates, don't use any connectivity data
  • scf=(conver=9) This keyword relates to how accurately the energy and electronic density are evaluated. If you look inside one of your optimised jobs you will fine a section that looks like this: 
     Keep R1 ints in memory in canonical form, NReq=10515645.
 Requested convergence on RMS density matrix=1.00D-08 within 128 cycles.
 Requested convergence on MAX density matrix=1.00D-06.
 Requested convergence on             energy=1.00D-06.
 No special actions if energy rises.
 EnCoef did   100 forward-backward iterations
 SCF Done:  E(RB3LYP) =  -3099.90946319     A.U. after   14 cycles
            NFock= 14  Conv=0.21D-08     -V/T= 2.0033
    We are interested in the part that is related to the convergence on the density matrix and on the energy. This means the calculation keeps going until the difference between the previous step and this step is less than 0.000001 on the root mean square of the density matrix and less than 0.000001 in the energy. Here is part of another file with the key word turned on, ie I have specified that scf=(conver=9): 
     Keep R1 ints in memory in canonical form, NReq=32125793.
 Requested convergence on RMS density matrix=1.00D-09 within 128 cycles.
 Requested convergence on MAX density matrix=1.00D-07.
 Requested convergence on             energy=1.00D-07.
 No special actions if energy rises.
 SCF Done:  E(RB3LYP) =  -3114.92377526     A.U. after   17 cycles
    You can see that the convergence criteria is one order of magnitude tighter on the density and on the energy. You might guess that since we are using "density" functional theory (DFT) we want to have a good and accurate density.
  • int=ultrafine this keyword relates to the integration grid used to evaluate the density. You know that the electronic density is the integral of the wavefunction squared. When it is evaluated numerically we need to define a grid on which the integral is evaluated. The finer the integration grid the more accurate the density! The figure below shows a cartoon which highlights this point:
    integration
  • opt=z-matrix Normally a job is optimised in cartesian coordinates, however there are some cases where optimisation is better performed in internal coordinates. Internal coordinates are just bond distances, bond angles and dihedral angles. A molecule can be described in terms of where the atoms are in space, we can choose different "coordinate" systems to define the atomic positions, examples are, cartesian ie (x,y,z) another is spherical (r,θ,φ) and another is internal (bond,bond-angle,torsion), you have also meet mass weighted cooridnates.
Top