G09

go back to Main Page, Computational Resources, Chemistry & More, Computational Codes, GAUSSIAN

---

You MUST use the "p" keyword for extra-print in all your calculations.

This is an essential requirement to use afterwards the Chemical Repository, SCIPIO, where you will need to save your data.

---

Revision Issues

• Revision A1 and A2;
  • do NOT perform the Dispersion calculation properly. You need the revision B1 or onwards.
  • implementation of the QZVP basis set is incorrect, it has not the polarization functions. You need the revision B1 or onwards.

• Revision C1;
  • problems with the modredundant section. To freeze a distance you can not use "B atom1 atom2 value F", as this does not do anything. Instead, you will have to move the geometry to the value and use "B atom1 atom2 F".
  • for some geometries the calculation stops after reading the initial structure.

How to

• to view frequencies in gaussview3

• how to create .fchk files

• how to compute BSSE

Basis sets

• About 5d/6d basis sets. The keywords 5D, 6D, 7F, and 10F are used to specify use of Cartesian or pure d and f (and higher) functions. The defaults are different depending if writing the basis sets in the gen section (at the end of the input file) or in the command line (#p opt belyp/6-31+g*). When using the Gen keyword the defaults are 5D and 7F (pure functions). When specifying the basis in the command line the defaults are 6D and 10F functions (Cartesian functions). The defaults can be changed typing 5d or 6d in the command line accordingly. It's very important to keep the use of one or another number of functions in all calculations, as if not there will be different basis sets and that will induce an error.
• About the QZVP basis set. The implementation is different between the A.02 (and A.01) and the B.01 versions of g09. The basis set in A.01 and A.02 miss the polarization functions, which means that these versions stored a QZV basis set. If you want to use the true QZVP basis, you should use the B.01 version. DO NOT RUN GAUSSIAN QZVP with REVISIONS A1 and A2 = KIMIK2

Dispersion

One of the new features in Gaussian G09, is the possibility to include dispersion. This can be done in two different ways.

• By using the B97D functional, which includes dispersion. With this you can optimize or do single point calculations.

• With the iop: iop(3/124=3),which allows to compute the Dispersion on a given structure optimized at your favourite level of theory (to add Dispersion as single point). The good point of this is that can be done using any set of basis and method (using SDDall you can save lot's of computer time).

The issue is that the first and second revision of Gaussian A1 and A2 do NOT perform the Dispersion calculation properly. You need the revision B1. This revision is not available in kimik2 and you will need to use CESCA to work on it. DO NOT RUN GAUSSIAN B97D or iop(3/124=3) with REVISIONS A1 and A2 = KIMIK2 IF YOU RUN THEM ON CESCA CHANGE TO REVISION B1

Large systems frequency calculations

For some reason, it seems a numerical problem, frequency calculations with large systems take too long. To reduce the computer time you can try to use the keyword Int=Acc2E=11 According to the manual it increases the accuracy from ^10 to ^11, but probably due to some bug, decreases the computational cost.

ONIOM in Gaussian09

• New force field implantation for ONIOM(QM:MM)

• ONIOM tips

• Convergence problems with ONIOM in G09

• Transition state searches with ONIOM

Issues concerning the coordinate system

Some extra checks of the completeness of the coordinate system were added in Gaussian09. Thus in some cases you might get the error message "Error in internal coordinate system" where it did not appear in Gaussian03.

Typical cases where this error message appear, are for linear system. E.g. when three covalently bond atoms are arranged on a line. For organometallic systems a typical case is for carbonyl ligands coordinated to a metal centre, M-CO. But you can also find it in many other cases.

• If your system contains any linearly arranged atoms and you get this error message, a solution might be to add linear bend coordinates to these atoms. In your modredundant section you write: L atom1 atom2 atom3 -1 A

• In the error message you will get the dihedral/angle affected, you can try to kill the problematic angle/dihedral atom1 atom2 atom3 K

• Modify a bit the geometry, by changing a bit the angles/dihedrals or even just deleting from the third decimal of the xyz coordinates.

• To optimize in cartesian coordinates could be also a solution to this problem although your calculations are going to be slower and you are not going to use modredun (not possible to freeze and scan coordinated).

• Another solution is to use a Z-matrix and add ghost atoms, that will allow you to define the system without angles and dihedrals close to 0 and 180.

• Revision C1 of Gaussian09 might have improved the way in which the internal coordinates are defined, as for some systems exactly the same input stops for RevA1 and runs without problems with RevC1. Be aware that RevC1 has issues with the modredundant section (see above)

• The Error does of course depend on the geometry, but also on the method/basis, so the same geometry might give the error with, let's say, M06 and not with B3LYP. So you can optimize the geometry "a bit" with another method and then try again with the method you whish.

Solvent Effects

As with Gaussian03, the SCRF keyword requests that a calculation be performed in the presence of a solvent by placing the solute in a cavity within the solvent reaction field.

• The integral equation formalism variant IEFPCM, is the default SCRF method. It has not changed from Gaussian03, BUT the formalism used and its implementation has changed. That is: you will get different results in G03 and G09, using the same method.

• The default RADII used in G03 was UAO , whilst now is UFF with spheres placed by default on all Hydrogen atoms. No need to use the SPHEREONH= keyword, except if you use UAO or another radii model that does not have them explicitly.

• It seems to give no "convergence failure" problems, but maybe its too early to say so.

• It is able to perform frequency calculations in solvent, giving enthalpies, free energies, ZPE corrections...

• The G09 output for PCM gives only one energy which corresponds, by default, to the electrostic energy. The non-electroestatic components of the energy of solvation are not included. If you wish to include them you will need to include "scrf=(read,PCM,...)" in the command line, and Cav Dis Rep (in separate lines) at the end of the input.

• The SMD method includes automatically the non-electrostatic terms.

The RI-DFT method

From the Gaussian manual:

Gaussian 09 provides the density fitting approximation for pure DFT calculations. This approach expands the density in a set of atom-centered functions when computing the Coulomb interaction instead of computing all of the two-electron integrals. It provides significant performance gains for pure DFT calculations on medium sized systems too small to take advantage of the linear scaling algorithms without a significant degradation in the accuracy of predicted structures, relative energies and molecular properties. Gaussian 09 can generate an appropriate fitting basis automatically from the AO basis, or you may select one of the built-in fitting sets.

What all this means is that for all calculations that have more than about 1000 basis functions, it will speed up the calculation greatly if you use the RI-DFT method. Generally, the greater the number of basis functions, the better it compares to normal DFT. The accuracy of this approximation is very good, and generally you can afford to use a much higher basis set than you would with a normal DFT calcualtion (so you get rid of BSSE errors etc.).

How to set up an RI-DFT calculation.

You have to use a pure DFT functional... no hybrids allowed...

You also have to specify TWO basis sets... The first one is the normal basis set specification (GENECP is supported for complicated setups with transition metals and ECPs), then the second one is the "density fitting basis", which you can specify yourself or Gaussian can generate automatically.

For example:

# PBEPBE/TZVP/TZVPFit

would specify a calculation with the PBEPBE functional, TZVP "normal" basis set, and TZVPFit "density fitting" basis set. The / sign between each section is essential.

For an example of an input file, and more information on how to set up the calculation, see the following page... RI-DFT input

Gaussian09 vs Gaussian03

• Scaling and New Functional Testing

• Solvation models, Polarizable Continuum Model (PCM)

• Tests on solvation energies, PCM in G03 and G09 and SMD