Hi Travis,

Thanks for the answer. I have implemented 'noeigentest' and it has started running. I'll post here what kind of a result I get. Thanks again.

So, having made the change to:

# opt=(qst3,NoEigenTest,calcfc) b3lyp/6-31G* Symmetry=None

The final lines are:

Error termination request processed by link 9999.

Error termination via Lnk1e in /usr/local/g09/l9999.exe at Thu May 22 17:38:59 2014.

Job cpu time: 1 days 6 hours 34 minutes 46.3 seconds.

And the final optimisation step reports:

Item Value Threshold Converged?

Maximum Force 0.001913 0.000450 NO

RMS Force 0.000291 0.000300 YES

Maximum Displacement 0.113158 0.001800 NO

RMS Displacement 0.012981 0.001200 NO

Predicted change in Energy=-8.917354D-06

Optimization stopped.

-- Number of steps exceeded, NStep= 143

-- Flag reset to prevent archiving.

I am going to restart this run and specify more optimisation steps. So far, this is the furthest this particular TS has come.

Hi Travis,

Thanks for a further reply. I am picking up more from asking questions than from trawling through the Gaussian documentation. I have successfully found 2 TS out of a pathway with potentially 15 transition states. However, I am guessing that it was by pure luck alone.

I have restarted the QST3 just as a TS with noeigentest. Would you suggest using a lower basis set to initially find a TS then revert to a more accurate basis from that structure?

Hi Travis,

A question regarding your comment:

"However if you use local searches [ # opt=(ts,noeigentest,calcall) or # opt=(ts,noeigentest,calcfc) ] initially to refine the structure and get rid of the extra imaginary modes, any subsequent QST3 calculation may have a better shot at finding your TS."

I believe I can't see the number, and subsequent reduction of the imaginary frequencies unless I am at a stationary point. So if I was to run a opt=ts which runs out of steps (there is a limit on the cluster I am using), I can literally copy one of the best fitting structures from the Gaussian output and continue refining in a separate run before trying QST3?

Thanks to you both. I took on board all suggestions and I've managed to find several TS. All in one day, a bit of a record. I have used a B3LYP with a low basis set and with noeigentest. Unfortunately, the one imaginary number did not yield the correct vibration. Interestingly, one of the unsuccessful TS optimisations did yield the correct vibration, but a long with several others (so not a saddle point). So I am using that TS calculation and increasing the basis set slowly.

I am very new to ab initio, and Gaussian. If all I want to do is protonate a linear glucose molecule (carboxyl) using the "environment" (e.g., in a water solution), would it be correct to translate hydrogen from H3O to generate a transition state, or just a single H or a H from H2O?

Many thanks

HI Oleg,

I mean to study a complete mechanism involving the gradual breakdown of glucose to eventually cross link with a lipid. The protonation of glucose is only the very first intermediate (there are somewhere between 8 to 15 intermediates). There are a few proposed mechanisms, hence I need free energy values along the mechanism reaction coordinate.

Hi Oleg,

Is it possible to translate a single hydrogen in gaussian? I only ask, as I might have followed some bad advice from a colleague who suggested that I don't perform protonation from single lone H+.

Hi Oleg,

Thanks for the response - I think I will need to get back to the real chemists about this (I am more of a statistical mechanic guy!). However, in the meanwhile I really appreciate the responses I have had here. I have had a lot of success with finding TS (most of them not the right ones) recently, but as of this morning I had one TS which I performed IRC on and it returned the correct product and reactant.

Hello everyone! I am in similar situation as that of Anthony who pointed out his difficulties in optimization of TSs. I have just stared learning computation chemistry and only month has spent. So i m a very beginner in this field.

I have got TSs for oxygen atom transfer from dioxiranes and Ni-peroxo system to CH4 molecule where oxygen inserts in between C-h bond leading the methanol formation. But these TSs are with low  basis sets of B3LYP such as STO-3G. When I tried higher basis stes of B3LYP such as 6-31G(d,p) or 6-31++G(d,p), unfortunately both showed the similar error as that of Anthony's(see avove the first post by Anthony).

Theese problems were two years back I can see dates. Please provide me updated and latest solutions for this problem.

Thanking you,

Best regards,

Rahul

Explanation of Error

This means that the Gaussian job terminated abnormally in some fashion. Typically, it means that a geometry optimization has not converged.

Fixing the Error

Geometry optimizations usually fail to converge for one of a few reasons.

If your initial starting structure is not good, then you should ask if you can provide a better starting structure, for example, one optimized at a lower level of theory. However, if it looks as if the structure is converging to what you want, as seen in your visualizer of choice, then one should restart the optimization from the last step, for example by using geom=allcheck in the route line, and it generally is a good idea to also use opt=CalcFC in these situations if it is not too expensive (say HF or DFT).

If your starting force constants matrix (Hessian) is poor, use a better one. This typically manifests itself when they vary a lot between levels, or if there is a large geometry change during the optimization. One can carry out a series of linked jobs (--Link1--). If you have a previous job, then usually Opt=ReadFC works well, but occasionally Opt=CalcFC, or rarely Opt=CalcAll are needed. In these cases, the forces are often converged, but the steps are not, and the final output will look like

-----------------------------

                       Item          Value           Threshold          Converged?

Maximum   Force              0.000401   0.000450             YES

RMS            Force              0.000178   0.000300              YES

Maximum Displacement 0.010503    0.001800             NO

RMS Displacement           0.003163   0.001200             NO

----------------------

Rarely, the coordinate system itself may be at fault. If z-matrix coordinates are being used, then in some cases, poor choices can be made, which result in angles, or three consecutive atoms of the four atoms used to define a torsion angle (dihedral angle) may be collinear (the angle is close to 0 or 180 degrees), which can give problems. In this case one can either formulate a better z-matrix or use the default redundant internal coordinates.

If these methods fail, another option would be to change the optimization method from the default to another type, such as opt=ef (if the number of variables is less than 50) or opt=gdiis (for floppy molecules).

best of luck

Hamid Mosaddeghi

Just add "noeigentest"   key word in the opt section. It will solve your problem. Thanks

I also have the same kind of problem. Actually I want to do CP correction to B2H6 2- , while doing I am getting the Error 9999. Please help me out and suggest me the best possible way to rectify it.

Hey Hi, you can vary the maxcycle step during optimization,

# opt=(maxcycle=N)

N: Whole Number :{50-500(or Max if this doesn't work)}

Regards

Ajay Khanna

following

i have some problem for calculated in semiempirique