Computational Inorganic Chemistry: Hunt Research Goup, Imperial College London

Trouble Shooting

You try to submit a job to the SCAN server and you get something like this:
Database error: SQLSTATE[22P02]: Invalid text representation: 7 ERROR: invalid input syntax for integer: "null", at 29 of /opt/portal-deploy/deploy/uportal/uportal-config/db-functions.inc doing INSERT INTO job_list ( job_id, app_id, user_id, description, staging_dir, is_published, project_id ) VALUES ( ?, ?, ?, ?, ?, ?, ? ) with values:Array ( [0] => 10029168 [1] => 2240 [2] => 10238 [3] => first td nicl4 631g [4] => run/10029168 [5] => 0 [6] => null )
The problem is that you have not setup a project! Go back to "running calculations" page and read see the text in bold, where it says "create a project for your jobs"
You try to "publish" and an error box comes up. In the first instance please wait 5-10min and try again. It maybe that the system is very busy (there are a lot of you trying to publish jobs at the same time). If it still doesn't work go to the portal profile page and tick "Publish to Dspace" then save. If this doesn't work do as the error message suggests and e-mail m.harvey with your job ID. Check back a little later and then try to publish them again. If the job is still not published within 24hrs then contact one of the demonstrators in the lab next lab session. Do not get hung up on waiting to publish but continue on with the lab. Do not delete the job and do not repeat the job as you will still be able to publish it later, the jobs will stay in your list until you delete them so there is no harm in leaving them a couple of days before publishing.
You go to "edit point group" and you cannot apply the right symmetry (T_d or D_4h) point group. This problem can arise if you are you trying to "correct" the symmetry of a higher level calculation. ONLY correct the symmetry if the calculation is at the 3-21G level. This problem can occur when the coordinates are not accurate enough, so we need to loosen the criteria a bit. In the tolerance click on "loose" and hopefully the symmetry point groups will become active. If not then try "very loose". If that doesn't work there may be something more serious going on and you should ask a demonstrator to have a look at your file.
Your B3LYP/6-31G(d) vibrational wavenumbers do not match mine by more than 1-2 cm^-1? For example my lowest frequency is 54cm^-1 and your lowest frequency is more like 65 cm^-1. When you check the "Item" part of your frequency file, you find that the molecule is not converged! What is wrong? You have symmetrised your optimised geometry before doing the frequency analysis, this was the wrong thing to do! By doing this you have destroyed the optimisation, and a frequency calculation should only ever be performed on a fully optimised structure. How do you fix it? Find your old B3LYP/6-31G(d) and this time when you set up the frequency job do not symmetrise the geometry. Alternatively use the geometry from your current frequency job to start a new optimisation. (Be careful to check the link0 information as discussed below).
```
        Item               Value     Threshold  Converged?
 Maximum Force            0.002612     0.000450     NO
 RMS     Force            0.001533     0.000300     NO
 Maximum Displacement     0.141613     0.001800     NO
 RMS     Displacement     0.077553     0.001200     NO
 Predicted change in Energy=-1.387744D-03
 GradGradGradGradGradGradGradGradGradGradGradGradGradGradGradGradGradGrad
```
Have you have added extra extra words in the additional keywords? Gaussian is very sensitive to the exact text. For example you should be adding "int=ultrafine scf=(conver=9)" your job will fail if you used "int=ultrafine and scf=(conver=9)"! Any other extra text not exactly as stated in the instructions will cause a job to fail.
Have you used the "opt=z-matrix" on the square planar NiCl₂(PH₃)₂ complex? This option was only to be used on the [NiCl₄]^2- complex because of the D_4h symmetry. NiCl₂(PH₃)₂ is C₁ and so it does not need this option, and your job will take too long and then fail if you use it. Do not use "opt=z-matrix" with NiCl₂(PH₃)₂.
Your job ends with the Maximum Force and RMS Force converged but not the distances! What is going on? This happens when you have a very flat PES and the slope is essentially zero over a very large range. As long as your forces are well converged, this is OK, go on and carry out your frequency analysis. The frequency analysis will tell you if you are converged properly or not.
Your NiCl2(dppe) job finishes (fails), but has not completed and there is no error message. Make sure there is no error message in the bottom of the text version of the log-file! In this case it may be your job has run for too long and has been "booted off" the server. Take the LAST step and resubmit using the Gaussian 4px queue.
Your frequency job has no negative frequencies BUT the "Item" options are not converged! What is going on? This happens because we are not really using a good enough basis set, however if we used a better basis set the jobs would take too long. How to fix it? First check that your symmetry is C1, ie that you have no symmetry. If you don't have C1 symmetry, manually break the symmetry by shifting atoms that are not on the symmetry element. If you do have C1 symmetry then you want to run the job with the optimisation criteria much tighter, this improves the convergence criteria on the structure (rather than the SCF), see below for the box to tick when you resubmit the job. You will need to re-optimise the structure and then repete the frequency calculation.
```
Low frequencies ---    0.0045    0.0060    0.0077   46.2233   47.9823   54.2204
Low frequencies ---   74.7746  131.9792  136.0728 
```
...
```
         Item               Value     Threshold  Converged?
 Maximum Force            0.009631     0.000450     NO
 RMS     Force            0.004838     0.000300     NO
 Maximum Displacement     0.122841     0.001800     NO
 RMS     Displacement     0.050627     0.001200     NO
```

Do you have one negative frequency for the cis square planar NiCl₂(PH₃)₂ complex? This can occur when the PH₃ groups are staggered. In this complex they prefer to be eclipsed. If you animate the negative frequency does it look like a PH₃ rotation forward and back? If so use the torsion tool to rotate one PH₃ group such that the H-atoms are eclipsed and start the optimisation from this point, you should be able to optimise to a lower energy structure with no negative mode. If you are unsure how to do this ask a demonstrator to help you.
Do you have one negative frequency for the trans square planar NiCl₂(PH₃)₂ complex? This can occur when the PH₃ groups are not aligned properly for the lowest energy structure. First check that your symmetry is C1, if it is not manually break the symmetry by moving atoms. If your symmetry is C1 then look at the orientation of the PH₃ groups. Use the torsion tool to rotate one H (in a PH₃ group) to have a zero (H-P-Ni-Cl) torsion angle with one Cl (which remains fixed), then rotate the other rotate PH₃ group so a H alignes with the opposite Cl. Starting from this point you should be able to optimise to a lower energy structure with no negative modes. If you are unsure how to do this ask a demonstrator to help you.
If you have used a file that has already been submitted to the SCAN queue it will have had some "extra" information added at the head of the file (this controls how many processors are used, the memory allocated and the checkpoint file). If you try to sumbit a job with this extra information a second time it will fail. You have two options (do one OR the other, not both!):
- 1. For example if you use "view file" you will see something like this at the top of the file:
```
%nprocshared=8
%mem=13000MB
%NoSave
%Chk=chk.chk
%rwf=/tmp/pbs.7098272.cx1b/rwf
(blank line in the file)
# opt b3lyp/6-311g(d,p) geom=connectivity int=ultrafine scf=conver=9
```
  edit the input file in word-pad to remove the extra material, that is everything before the line beginning with the hash symbol, be sure to remove any blank lines as well. You should be left with something that starts like this:
```
# opt b3lyp/6-311g(d,p) geom=connectivity int=ultrafine scf=conver=9
```
- 2. when you prepare the input file under "gaussian set-up" click on the link 0 tab. You will see something like the first image below, you need to change these options to reflect those in the second image below:
  
  incorrect file options
  
  correct file options

Top