spin-orbit matrix elements

If you have a suggestion and a plan for implementing it, please file an issue on the Dalton or LSDalton GitLab. Otherwise, feel free to discuss it here first.
Post Reply
quartarolo
Posts: 22
Joined: 15 Dec 2013, 22:25
First name(s): Angelo
Middle name(s): Domenico
Last name(s): Quartarolo
Affiliation: High School
Country: Italy

spin-orbit matrix elements

Post by quartarolo » 05 Jul 2015, 13:02

Hi,
I use Dalton for spin-orbit matrix elements calculation between exited states (singlet-triplet states) as:
....
**INTEGRALS
.MNF-SO (or nothing for ECP calculations)
....
*QUADRATIC
.DOUBLE RESIDUE
.ISPABC
1 0 1
.PROPRT
X1MNF-SO (or X1SPNSCA for calculation including ECP)
.PROPRT
...
.ROOTS

Can it be possible to do it with LSDalton future releases? Do you think this kind of calculation could go faster, taking advantage for the LSDalton implementation?
Thanks
Domenico

taylor
Posts: 545
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: spin-orbit matrix elements

Post by taylor » 05 Jul 2015, 14:39

(Disclosure: I am not currently an LSDalton developer. But I certainly see the advantages of extending the program, especially if I don't have to do any of the work...)

Your proposal is interesting but you do not suggest what level of calculation you want to see implemented for such calculations? SCF, DFT, CC...? How will you determine that this is the appropriate level (just agreeing with experiment is not the answer)? And what level of accuracy you think is needed for your purposes? How many excited states do you need? Are they valence-like, Rydberg-like, or mixed? What type of basis sets are needed to describe them at your desired wave function level?

Best regards
Pete

quartarolo
Posts: 22
Joined: 15 Dec 2013, 22:25
First name(s): Angelo
Middle name(s): Domenico
Last name(s): Quartarolo
Affiliation: High School
Country: Italy

Re: spin-orbit matrix elements

Post by quartarolo » 05 Jul 2015, 18:51

For the purpose I adopted the atomic mean-field approximation (AMFI) at B3LYP/cc-pVDZ to get an estimation
just for electronic intersystem crossing between the lowest excited states (from S1-S3 to T1).
The inclusion of the vibrational part for the rate constant calculation is a more difficult computational task.
For small selected systems, I compared the results with an higher-level theoretical approach which makes use of correlated DFT/MRCI
wavefunctions and AMFI(see Marian et al. papers).The implementation of DFT-MRCI could be a further suggestion for the developers.
best regards
Domenico

taylor
Posts: 545
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: spin-orbit matrix elements

Post by taylor » 05 Jul 2015, 19:05

"For the purpose"... What purpose? What are you doing, and what criteria are you applying for the reliability of any computed results? How do you calibrate these?

This may sound harsh, but simply saying (by implication) "it would be good to have [fill in the blanks]" says nothing about
(a) what a user may want to do
(b) whether it makes any sense at all
(c) whether the proposed methodology is then useful for that purpose.

Anyone who thinks that DFT-based methods parametrized for ground-state calculations can tell us anything reliable about excited states, well, I got this bridge in Sydney, with an Opera House nearby, and we could talk about a real good deal if they want to put down a deposit: bridge, Opera House, or a fantastic deal on both...

To my knowledge, there are no MRCI-type methods programmed in LSDalton. In fact, the Duesselldorf/Muenster/Bonn work (and I have worked with and published with these groups, so I can claim I am expert on the methodology) is specifically a selected/perturbationally-extrapolated multireference method: we have never had anything like this in Dalton or LSDalton. So while I don't like the idea of just saying "might be a good idea but probably not going to happen", I think this is a long way outside any Dalton/LSDalton development! At least that I'm aware of.

Best regards
Pete

quartarolo
Posts: 22
Joined: 15 Dec 2013, 22:25
First name(s): Angelo
Middle name(s): Domenico
Last name(s): Quartarolo
Affiliation: High School
Country: Italy

Re: spin-orbit matrix elements

Post by quartarolo » 05 Jul 2015, 21:02

Strating from matrix elements evaluation,the aim is to calculate an approximate value for nonradiative and radiative transition rate constants and
try to connect them to the real photochemical behaviour. I'm aware DFT approach is not the best but I'm not expert about multiconfigurational methods, a little about cc2 calculations. I will improve myself about this.
thanks
domenico

taylor
Posts: 545
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: spin-orbit matrix elements

Post by taylor » 09 Jul 2015, 16:28

There has been correspondence about aspects of this in the past on this forum, I think. While I appreciate what you are talking about doing, there are several dimensions that are completely outside what not just Dalton, but any electronic structure code is designed to do.

At the risk of seeming tedious, having excited, by photon-based electronic excitation from (for reasons of simplicity) the zero-point vibrational level of the electronic ground state, to a particular excited vibronic state (I ignore rotation here), several things may happen. The state may immediately decay radiatively to the ground state. It may radiatively or nonradiatively decay down its vibrational ladder to lower vibrational levels of the electronically excited state and fluoresce from there to the ground state. It may undergo intersystem crossing via singlet/triplet coupling and eventually phosphoresce to the ground state. Etc...

The important point here is that although there are some processes (intersystem crossing, say) that are almost entirely dependent on intramolecular properties like spin-orbit coupling, there are processes that are not. In real-world experiments molecules are studied in the gas-phase or (even worse) condensed-phase: in this situation vibrational relaxation, say, will largely happen nonradiatively via collisions with the bath-gas or solvent. There is no straightforward way to treat these things in a molecular electronic structure code like Dalton! They are dynamical phenomena that involve degrees of freedom that are effectively eliminated from Dalton from the beginning. This is not a limitation of Dalton: it is what the code (like its various competitors) is designed to do!

If you are indeed looking to tackle the general problem of relaxation of excited states to the ground state via all the possible mechanisms listed in the traditional Jablonski diagram, you need much more than an electronic structure code! You need a way of treating the nuclear motion, and you need a way of incorporating a bath gas or explicit solvent that can accommodate relaxation of nuclear motion. If you do not have all this, comparisons with experiment will be pointless, because you will never know whether good agreement is a reflection of genuine success in your calculation, or simply a cancellation of errors, in a particular case, between the errors in a particular DFT functional (or, so as not to victimize DFT, truncated CC or perturbation theory, or basis set issues), and the inability to treat all of these other dimensions of the problem.

By all means seek to calculate the various contributions. But Dalton alone (nor any other electronic structure program) cannot treat all the degrees of freedom you need to describe, and your goal, while praiseworthy, is a very complicated and demanding one.

Best regards
Pete

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest