frequency analysis internal coordinates

[acapobia]

Hi all

I'm using the frequency analysis tool of aoforce. It's really a very powerful tool, a little jewel in the TM suite, especially if one needs to manipulate normal modes, for getting e.g. Franck Condon factors and so on.

However I wonder if it is possible to use internal coordinates referred to a ghost atom. It would be very useful to me describing the normal modes of ammonia using the <HNQ bending angle, where Q is a point located on the C3v axis, say, opposite to N and to the plane spanned by the hydrogen atoms.

I tried the 'pyrm' and 'bipy' types of internals (since I guessed they could help) but 'curiosity killed the cat' and I discovered they are not  implemented yet (and so they are  undocumented).

Is there a way to define the internal coordinate I am interested in for the frequency analysis with the current version of TM ?

Any help will be appreciated.

Thanks in advance.

Amedeo

[uwe]

Hi,

did you try to define the dummy atom as e.g. hydrogen without basis and charge? The Turbomole modules treat such dummy atoms like usual nuclei without electrons and basis sets. Add the hydrogen to the coordinates and modify the $atoms section:

h  4                                                                           \
   basis =none                                                                 \
   charge=  0.00000000


here the 4th atom has no basis and no nuclear charge (if you are using RI, you have to give also the jbas line here - simply copy the lines of the other atoms and replace the settings without deleting the \ at the end of the lines...). If you run the usual procedure, define internal coordinates, run frequency calculation, etc., aoforce should print out what you would like to see - hopefully, I did not check that myself.

Uwe

[acapobia]

Unfortunately the strategy proposed by Uwe (thank you so much for your quick and kind reply anyway) doesn't seem to work, or, at least, I was not able to make it work.
Indeed aoforce finds five atoms for the NH3Q system instead of four (NH3), so it needs all 9 internal coordinates and -since it found only six internals- exits with the following warning:

[...]
 WARNING: The internal coordinate analysis
 needs all 3N-6/3N-5 internal coordinates!
 I ignore the keyword!

 MODTRACE: no modules on stack

 Huch
 force ended abnormally
 force ended abnormally
program stopped.

I tried to define three more internal coords assigning them to 'f' or 'i' types, but I did not success in getting normal modes and obtained the above warning again (aoforce is not 'stupid', of course! :-)

I also defined three more  'k',  internal coordinates and run a dscf/grad step, followed by a full 'aoforce' computation and an 'aoforce analysis only' run, but I obtained 9 normal modes, one of which corresponding to an imaginary frequency; moreover all the frequencies previously computed were dramatically modified by the new computation.

I  add, for the sake of completeness, that the dscf/grad steps with the dummy hydrogen atom included and  the key 'internal' switched 'on' in $optimize went fine. In fact the cartesian gradient for the coordinates of dummy H were zero as expected, while for the 'real' coordinates of NH3 I obtained the same values of the earlier cycles of optimization, i.e. less than 10^-5.

Suggestions or tips?

amedeo

[acapobia]

I wrote my last post in a very confused manner, I'm sorry.

"moreover all the frequencies previously computed were dramatically modified by the new computation."

I meant that the frequencies computed (before the step concerning analysis with internals) with the dummy atom included are dramatically different from the frequencies previously obtained by the computation on 'regular' NH3 without the ghost atom.

Of course after the analysis, frequencies remain the same !

[uwe]

Hi,

I guess that aoforce reads in old data which does not match to the new number of atoms.

Just start from scratch, i.e. take just the coord file and take only the coordinates from there - without the definition of internal coordinates. Start dscf or even jobex and then do the aoforce step with this 'fresh' control file again.

The only problem I see is to get a good set of internal coordinates for your NH3-Q which is a) mathematically stable (iaut will do that) and b) simple enough to be understandable.

Note that you will get 3 more modes than before, because you have one atom in addition now. The frequencies of the run without dummy atom should almost be the same as before, most likely you will have negative or smaller wave numbers in addition (move the dummy atom closer or further from the N to get different frequencies for the three new modes - do not forget to run dscf first and then aoforce after changing the coordinates in the coord file).

I would recommend to do everything without the analysis option first. If you are sure which modes you have and which internal coordinates you are interested in, add the analysis only intcoord option to $drvopt.

Hope this helps,

Uwe

[acapobia]

I started a new calculation (B3LYP/def2-TZVPP by the way) from scratch, dummy atom included without defining any type of internal (except redundant internals as generated by 'ired') then I ran jobex and optimized structure (dummy atom included); variations in geometry with respect to the computation already done  without dummy atom were as small as 10^-6 Angstrom. Then, as suggested by Uwe, i ran an aoforce step, but aoforce ended with the error reported below.


CONSTRUCTING second deriv. of 2e energy    -> Hessian
   treating Coulomb (and exchange) contribution
     integrals will be neglected if total contribution <  6.172839506172839E-009
      ...terminated. cpu:       0.96       wall:       0.96
   treating exchange-correlation contribution
 Dummy atoms without basis are not allowed in force !
Remove dummy atoms !
 
So I never reached the step of analysis. Anyway, despite of my results I wish to thank Uwe again for his kindness.

amedeo

[uwe]

Hi,

aoforce does not like dummy atoms for DFT calculations. The reason is that dummy atoms without basis functions might cause numerical problems with the quadrature - if grid points for the quadrature are being assigned to the dummy atoms. In that case the density will be very small at those points, as well as the gradient of the density. This can (but does not have to) cause large numerical errors (since this gives 0 over 0).

So aoforce avoids possible inaccuracies by not allowing dummy atoms at DFT level - unfortunately in your case...

You can do Hartree-Fock instead, although this might not really be what you want.

I remember that there have been some patches concerning DFT quadrature and dummy atoms in the last release, and the new version should be much more (numerically) stable also in such critical cases. Please send an email to us (the Turbomole support), and we will send a modified aoforce version which will allow the usage of dummy atoms.

Regards,

Uwe