Firefly and PC GAMESS-related discussion club

 
Learn how to ask questions correctly  
 
 

 

Amir Nasser Shamkhali
amir_n_shamkhali@yahoo.com

On Thu Apr 3 '14 7:34am, Siddheshwar Chopra wrote
-------------------------------------------------
>On Wed Apr 2 '14 11:02am, Amir Nasser Shamkhali wrote
>Dear Sir,
>Thank you again Sir. As suggested by I performed the TDDFT calcs. for the singlet excited state. The output file shows me various transitions with the probabilities and oscillator strengths.. How do I discuss MOST probable transitions on this data basis? I find 90% and above probability for 17 transitions including the one between HOMO---> LUMO (90%)... But there are several others with more than 90%... I just wish to discuss the prominent transitions (Is there any classification of these bands like Soret bands etc).. Also for sure the optical band gap is the "eV" corresponding to the HOMO---> LUMO transition??
>
>
>Regards,

>-----------------------------------------------------
>>Dear Siddheshwar
>>Yes, you need to optimize ground state only. The MULT key in $CONTROL is for your ground state, but in $TDDFT group is the multiplicity of excited state.
>>In fluorescence and phosphorescence both of vertical and adiabatic excitation may occur. First a vertical excitation by a photon is taken place, then molecule geometry changes with some energy loss without emission of a photon. Then a new electronic state is achieved and relaxation into ground state emits a photon with higher wave length than initial one. Now, in fluorescence the multiplicity of ground state and the mentioned new electronic state is the same. However, in phosphorescence this is not the case. If the non-photonic energy loss is due to the geometrical changes, thus the second excited state is adiabatic, otherwise, all of your excitations are vertical and changes from first to second excited states (non-photonic energy loss) can be due to the tunneling of electron between two states.
>>If all of excitations are vertical, your work is too easy. You should perform TDDFT caculation to obtain some of excitation energies. Then the wave length of fluorescence and phosphorescence emission should be one of vertical excitation energies. Also, you may find first excitation (initial photon) by the same way. So, you should increase NSTATE keyword in order to find the wave length initial photon. This may be higher numbers, please check this. Higher NSTATE maybe restricted by memory of your computer.
>>If you can find the wave length of initial photon, not the emitted photon, then the second excited state is an adiabatic one and you need CI or CASSCF calculations to find it. But I think that this has less possibility. Because fluorescence and phosphorescence are reversible processes and molecules are in chemical equilibrium.

>>Best Wishes      

>>On Wed Apr 2 '14 7:27am, Siddheshwar Chopra wrote
>>-------------------------------------------------
>>>Dear Amir Sir,
>>>First of all let me thank you for explaining in depth. I really appreciate the way you have explained. It at times becomes so difficult to read from a book. I being a physicist find it really interesting to discuss like this. Thank you again.
>>>Sir, from your explanation about UV-Vis. spectra, I understood two things::
>>>1) A hessian run is sufficient. I don't need to re-optimize..
>>>2) MULT will be added twice in the i/p file for TDDFT..

>>>Could you confirm the above observations saying YES?? Now how do I decide upon studying phosphorescence or fluorescence? Say, I don't have any prior information about the sample... practically in which applications are both processes used?
>>>Also is there any way I can comment on radiative or non-radiative nature of transitions??
>>>Sir, I am quite ambiguous while choosing NSTATE value.. Can I keep it very high? What does it actually infer?

>>>Regards,

>>>On Wed Apr 2 '14 1:41am, Amir Nasser Shamkhali wrote
>>>----------------------------------------------------
>>>>Dear Siddheshwar
>>>>UV-Vis spectra in most cases is a collection of vertical excitations. Please note that when you measure UV-Vis spectra by a spectrophotometer, the most important interaction of light with molecules is its electric filed. Due to the movement of molecules, the direction of interaction varies by time, also. Therefore the Hamiltonians of the molecule in various times within a period between t=0 and t=T(1/freq)do not commute and expectation values of Hamiltonians (energies) vary with time. However, this time is very short ( in order of 10^-17 sec) in which geometry of the molecule have not enough opportunity to change.Thus your material is stable. This means that molecules are in chemical equilibrium. So, in order to calculate UV-Vis spectra, you should only optimize the ground state, by your DFT method. Then, taking the optimized structure, you should run a single point TDDFT calculation with some higher vertical excitations (NSTATE keyword). So, for example a simple input is required as follows:
>>>> $CONTRL scftyp=RHF CITYP=TDDFT icharg=0 mult=1                                
>>>>  NZVAR=??? Coord=ZMT MAXIT=800 DFTTYP=??? UNITS=ANGS $END
>>>> $TDDFT NSTATE=7 ISTATE=1 MULT=1 $END
>>>>Please note that when you are interested to phosphorescence, it means that the multiplicity of excited state is different that of the ground state. So, if the ground state has 2S+1=1, then the MULT keyword in $CONTROL should be 1 and in $TDDFT should be 3.
>>>>Electron-phonon interaction is a solid state effect which is responsible for superconductivity. Please note that Firefly is based on the Gaussian-type basis sets which are localized and are appropriate for molecules. But electron-phonon interaction is delocalized in nature and you should use a code based on plane wave basis sets (such as ABINIT, espresso, WIEN2k, exciting, and etc.).
>>>>  

>>>>On Tue Apr 1 '14 11:15pm, Siddheshwar Chopra wrote
>>>>--------------------------------------------------
>>>>>Dear Sir,
>>>>>That means if I have to study UV-Vis Spectra, then studying vertical excitation only, is correct?? If yes then running a CITYP=TDDFT is correct? Is that right? In vertical excitation if the geometry of the molecule (in first excited state???) doesnt change, then optimization still is needed? How time consuming is this job? Does the optimization and oscillator strengths get calculated in the same run RUNTYP=optimize?

>>>>>Also how do I decide upon studying the adiabatic excitation for my sample?? Say it is a new sample/molecule.. Do you mean phosphorescence transitions by this? Thanks again for explaining so well Sir.

>>>>>Also is there any way I can comment on radiative or non-radiative nature of transitions?? or can I find electron-phonon interactions using Firefly?

>>>>>Regards,

>>>>>On Tue Apr 1 '14 3:00pm, Amir Nasser Shamkhali wrote
>>>>>----------------------------------------------------
>>>>>>Dear Sidheshwar
>>>>>>There are two types of excited states: vertical and adiabatic excited states. Vertical excitation is more important for UV-Vis spectra in which lifetime of excitation is too short in such a way that Franck-Condon effect is important. Thus in vertical excitation, the geometry of the molecule is not changed considerably. TDDFT is appropriate method for vertical excitation. However, in adiabatic excitation, lifetime of excited state is longer and geometry of the molecule is changed. For this excitation you should use configuration interaction (CI) or CASSCF methods.  
>>>>>>
>>>>>>
>>>>>>On Tue Apr 1 '14 12:06pm, Siddheshwar Chopra wrote
>>>>>>--------------------------------------------------
>>>>>>>Dear Sir,
>>>>>>>I went through the forum regarding this but couldn't get my answer exactly. I am running TDDFT calculations on the ground state configuration. Steps were:
>>>>>>>1) Copied the lowest energy config. coordinates to TDDFT i/p file.
>>>>>>>2) Using RUNTYP=optimize and CITYP=TDDFT in $CONTRL, to optimize for the first excited state only.
>>>>>>>Am I performing it in the correct manner? Will I get "Equilibrium geometry located" and the coordinates of Lowest first excited state? Will the same run give me UV-Vis. spectra too?
>>>>>>>I wish to know how similar is this optimization to the Hessian run, say in terms of the memory and CPU time requirement? How to speed up these calcs?
>>>>>>>And secondly, I read on the forum about CIS.. But do I need that here?

>>>>>>>Regards,