Author Topic: water , triplet excited state geometry optimization  (Read 15058 times)

methane

  • Newbie
  • *
  • Posts: 7
  • Karma: +0/-0
water , triplet excited state geometry optimization
« on: February 22, 2007, 02:44:10 AM »
First time using TURBOMOLE for excited state caculations.

I am performing a RIDFT first triplet excited state geometry optimization for water. I want to compare the 3B1 excited state with other method.

basis :TZVPP

The work made some optimization steps but it stops with error:

egrad ended abnormally.


the relative part of my control file :

$scfconv   7
$denconv 1d-7
$scfinstab rpat
$soes
b1 1
$exopt 1

------------------------------------------
there are 4 real representations found :   a1   a2   b1   b2

What else should I look ? 

dimi

  • Newbie
  • *
  • Posts: 5
  • Karma: +0/-0
Re: water , triplet excited state geometry optimization
« Reply #1 on: February 22, 2007, 09:52:19 AM »
Hello,

the specifications in the control file look good; there should be no problem with that. What are the last messages in the job.[n] output
file (n: some number)? It should contain output from egrad, the TURBOMOLE module for calculation of excited state gradients.

Best,
Dmitrij

methane

  • Newbie
  • *
  • Posts: 7
  • Karma: +0/-0
Re: water , triplet excited state geometry optimization
« Reply #2 on: February 22, 2007, 08:16:54 PM »
job.7

Here are the last part of job.7, which shows a negative excitation energy.


 ==============================================================================

                              I R R E P   b1 

 ==============================================================================


 Excitation is electric dipole allowed

 Excitation is magnetic dipole allowed

 Excitation is electric quadrupole allowed




                         1 triplet b1 excitation


 Eigenvalue:                            -0.4979858778651368E-03


                    WARNING! NEGATIVE EXCITATION ENERGY!

                         REFERENCE STATE IS UNSTABLE!



                    WARNING! TRIPLET INSTABILITY DETECTED!


 
 MODTRACE: no modules on stack

 reference state instability
 egrad ended abnormally
error in gradient step (1)

dimi

  • Newbie
  • *
  • Posts: 5
  • Karma: +0/-0
Re: water , triplet excited state geometry optimization
« Reply #3 on: February 23, 2007, 10:40:59 AM »
Morning,

the negative excitation energy means that the triplet state is below the singlet reference at the given geometry. That is you don't need  TDDFT to calculate the triplet and, in fact, can't even use it since it always needs the energetically lowest state to calculate excitations from. Instead you can just use unrestricted ground-state DFT to optimize the triplet. It works as follows:
1. The state you're calculating is not totally symmetric in C2v (b1), so you need to go to a Cs subgroup of C2v to have a totally symmetric ground state. You can do it by calling define, going to the geometry menu and calling "susy" to reduce the symmetry. There are two inequivalent Cs subgroups of C2v, so you have to try out which one to take.
2. Change the irrep of excitation to a', either in the "ex" submenu of the general menu (last one in define) or directly in the control file (data group $soes).
3. Change $scfinstab to "$scfinstab triplet" and add data group "$start vector generation" to the control file.
The program will do a triplet instability calculation and generate start MOs for the triplet calculation.
4. Call the escf module to do the instability calculation. You should find the files "alpha" and "beta" in the working directory after it's done.
5. Start the normal optimization with jobex (without -ex option).

I hope it helps.
Dmitrij

methane

  • Newbie
  • *
  • Posts: 7
  • Karma: +0/-0
Re: water , triplet excited state geometry optimization
« Reply #4 on: March 05, 2007, 10:48:29 PM »
Thanks Dmitrij~

I followed your instruction for 1&2 and get some results after 15 cycles.

I have noticed that when calculate b1 excited state, either singlet or triplet, I have to go back to C2 subgroup of C2V and choose the irrep of "b". What should I do if I want to caculate b2 excieted state?

The results I got for 3B1 using aug-cc-pVTZ basis with TD-b3lyp:

R(OH)=1.081877A  Theta=109.8306457 deg Excited Energy=5.585853551eV

Comparing the results in the literature: J. Chem. Phys. 113,7084-7096(2000)

R(OH)=1.082A  Theta=110.1 deg Excited Energy=6.19eV

It seems I got a close result for the sturcture but not the excitation energy. Why?
----------------------------------------------------------------------------------------

I here list  my other caculation results on 2nd 1A1 excited state and 2nd 1B1 excited state using aug-cc-pVTZ basis with TD-b3lyp for future reference.

2nd 1A1 : R(OH)=0.9765A  Theta=114.8deg Excited Energy=10.32 eV
Ref:         R(OH)=0.976A    Theta=114.6deg Excited Energy=10.40 eV

2nd 1B1:   R(OH)=1.001A  Theta=105.15 deg Excited Energy=9.62 eV
Ref:          R(OH)=1.001A  Theta=105.2 deg Excited Energy=9.69 eV




dimi

  • Newbie
  • *
  • Posts: 5
  • Karma: +0/-0
Re: water , triplet excited state geometry optimization
« Reply #5 on: March 07, 2007, 07:01:52 PM »
Hello,

the results for th triplet state of water look quite good too me: the structural parameter are as good as you can expect from DFT. The energies are usually less accurate. Moreover, B3LYP often favours states of higher spin (due to the HF exchange contribution) which fits well with your results. So I would not worry too much about that -- 0.5 eV is not small, but the singlet-triplet energy differences are often a problem for DFT.

When going to a subgroup, you have to make sure that you are calculating the same state again, which is not always trivial. b1 and b2 irreps of C2v reduce both to the b representation of C2 point group, thus you will get both b1 and b2 excitations if "b" is specified. Usually, you can assign the excitations to "b1" or "b2" by comparing the energies which the C2v vertical excitations calculation. Also, one of the irreps, say b1, will have a nonvanishing transition dipole moment in the x direction, the other in the y direction.

Best,
Dmitrij