How to find the active orbitals for an ORCA CASSCF calculation on a porphyrin

[@sgig9608]

Really great to see such content in KZbin..Thank you..

[@shartzstein]

Quite useful! Thank you very much.

[@niconeuman]

You're welcome!

[@priyambharadwaz5956]

Hey, thank you so much for the upload. I want to do CASSCF calculation for my copper (I) system. I first dis DFT calculation in Gaussian. So, is it possible to CASSCF single point calculation in ORCA from Gaussian optimized structure?

[@niconeuman]

@@priyambharadwaz5956 yes it is possible. You just need the xyz coordinates. I don't think you can use the DFT orbitals from Gaussian as guess for the CASSCF, but you don't really need that nowadays.

[@priyambharadwaz5956]

@@niconeuman you mean I dont to provide orbital guess for CASSCF in Gaussian, if not then how to do that. Does it automatically read orbital guess?

[@niconeuman]

@@priyambharadwaz5956 I don't think Orca will be able to read gaussian files at all. However, you don't usually need to start from other orbitals in Orca, because the CASSCF procedure is really good and can make a good guess by itself and find good orbitals. In case it doesn't, you can plot the orbitals from the first CASSCF calculation, look at them, and make a second CASSCF calculation where you read the .gbw file from the first one, and use the "rotate" keyword to bring orbitals from the inactive and virtual spaces into the active space. I explain how to do that in this video, but also in the manual it is explained clearly. Nowadays doing CASSCF with Orca is quite easy and fast, unless you have a very large active space.

[@kildexftw]

Heey, thank you so much for the upload, I was extremely happy when I suddenly saw this randomly in my feed. Im currently attempting to learn CASSCF and this video is super well explained and will help me a lot on my current attempt to learn CASSCF. I have one question about regarding your initials guess orbitals for the CASSCF. I see that you calculate UKS orbitals by DFT and use those for finding your active space. What Im currently doing (unsuccesfully), is the calculate ROKS orbitals from QRO's, as is written on the orca input library. What do you think of this method?

[@niconeuman]

Thank you for the enthusiasm! In my (limited) experience it is not very important which orbitals you use to start the casscf. Although alpha and beta orbitals from UKS may be different, when you do the casscf orbitals are re optimized. Then if you still don't have all the correct orbitals you do a second casscf calculation reading the first and finding and rotating the correct orbitals. And you can keep repeating this procedure if your system is complicated. Once you are working with casscf orbitals you are on a better track and it doesn't matter anymore where the orbitals came from originally. Of course there may be very tricky problems if your active space is large and you have very delocalized orbitals. I would recommend the CASSCF tutorial in the forum. It has a lot of good examples. Hope this helps

[@kildexftw]

@@niconeuman Thank you so much for your fast reply! I have almost given up on it, but you just motivated me to keep trying!

[@kavyavinod6274]

Hey! Very useful video!! Can we do SOC calculation with CASSCF in ORCA?

[@niconeuman]

Yes. SOC within CASSCF/NEVPT2 is called using the %rel module. You can search that in the manual. I always include it when I do these calculations. In the other CASSCF video, kzbin.info/www/bejne/boqphHp5rdCBl6s I discuss magnetic properties like g- and D-. Hope this helps!

[@m.r.crawley762]

Hi Nicolas, I've thoroughly enjoyed your videos and find them incredibly informative. I do have a quick question: at 11:53 you show the LUMO and LUMO+1 Gouterman orbitals after the first CASSCF calculation; however they don't look like the "classic" eg set, (i.e. the node is on two of meso carbons, not two of the nitrogen as the classic diagram predict), later (17:49) after you rotate the dz2 orbital into the active space to replace the erroneous pi-type orbital and re-run the CASSCF calculation you obtain the expected LUMO and LUMO+1 that agree with the classic orbital picture for the Gouterman eg set. Was there anything else you changed to obtain those orbitals? I ask because am also interested in Co-porphyrins and have obtained LUMO and LUMO+1 orbitals that match yours quite nicely for the first set (e.g. the orbitals you show at 11:53); however, even after a couple CASSCF runs using the .gbw from the previous run I do not obtain LUMO and LUMO+1 orbitals that match yours. Am I overlooking something obvious in terms of the 11:53 and 17:49 being symmetry equivalents? Thank you in advanced for your input, and thank you for your videos! Best, Matt

[@niconeuman]

Hi Matt, I'm not sure why you are getting the orbitals slightly changed. I didn't change anything, it probably just optimized the orbitals better when I completed the 3d set. Probably both sets of orbitals are very similar but the first ones (at 11:53) were slightly distorted due to lack of good convergence. I would say that having less roots in your casscf calculation is more likely to give you more "pure" orbitals which will look nicer. But the chemical significance of the shapes of the orbitals is very limited.

[@m.r.crawley762]

@@niconeuman Thank you very much for the response. I agree that the chemical significance is limited, just wanted to make sure I wasn't making a silly oversight. Thanks again, looking forward to future videos!

[@Chem-iu5jx]

If I want to use orbitals from a broken symmetry calculation, do I use the uco, uno or the non-localized orbitals from the gbw file? Technically it shouldn't matter since they are the same but just obtained tthrough unitary transformation. But the weird thing is: The unrestricted corresponding HOMO is of dz² caracter while the HOMO from the gbw file is dx²y² character...

[@niconeuman]

You would have to see for your particular problem. The programs I know don't show the different types of orbitals, only the canonical orbitals, so you would have to visualize them using orca_plot. Other than that I don't think it matters that much, as you would only be using those orbitals as a guess, and then the casscf procedure would optimize the orbitals, which if I'm correct are natural orbitals.

[@kritikagupta3117]

Hii, i am investigating spin-statr reactivity towards C-H activation. Here i am looking for the alternate triplet state with more unpaired electrons, and i specifically need to select those orbitals. In gaussian, i use SCF(Vshift=1000) command to shift orbital energies to achieve higher energy state and Guess=alter keyword for to SWAP the orbitals, like one of the occupied orbital is replaced by other virtual orbitals. Please let me know if such an option is available in ORCA.

[@kritikagupta3117]

Also, i want to know if we can generate natural orbitals for the spin density using ORCA

[@niconeuman]

thank you for your comment. I don't really understand what you want to do. if you want to study the reactivity of a triplet state just choose multiplicity equal 3 in the input file. do you want to do DFT or CASSCF? if you want a specific triplet state that is not the first one, you could use TDDFT or do a broken symmetry calculation where you start from a quintet, and flip two electrons. that may give you the right configuration or not. in the 2nd case you want to rotate the orbitals. look for the rotate keyword, using MOread to read the orbital from a previous calculation. how to do that with CASSCF, I don't know. it's probably possible to manually define a specific configuration and run a state specific calculation. I would say that is rather advanced.

[@kritikagupta3117]

@@niconeuman Thank you so much for your response. Actually, ia m looking for the transition state that maximizes the number of exchange interactions and have max number of unpaired electron (Exchange Enhanced Reactivity Concept)....i am working on 3[Mn(IV)=O(PC)•]-1 complex , where i am getting 2 unpaired electrons of complex , but i want 4 unpaired electrons after analysis of natural orbitals for spin density...and i have no idea how to do it using orca..in gaussian we swap the orbitals.

[@niconeuman]

@@kritikagupta3117 I think you just need to calculate the TS with a multiplicity of 5, and also do a BS 2,2 calculation. That may work. Using the rotate option for each orbital you want may also work, but it may also optimize the orbitals to a different state.

[@kritikagupta3117]

@@niconeuman Thank you for replying and solving my query so quickly! I really appreciate your help. I'll try to implement all your suggestions.

[@sgig9608]

Can you suggest some papers and books to get basic of CASSCF?

[@niconeuman]

hi, I don't know any particular books. I have read different papers over the years. but to start it's really good to follow the casscf tutorial that can be downloaded from the orca forum. beyond that, casscf is related to configuration interaction, so a book like Szabo and Ostlund can be a good start on that.

[@aarzoosaharan1655]

how can I find active electrons and orbitals in the triphenylamine molecule?

[@niconeuman]

The triphenylamine molecule is diamagnetic and neutral. So your calculation will be done with charge 0 and multiplicity 1 (there could be excited triplet states if you want to include them), and therefore you have to include an even number of electrons in your active space. If you only want to include pi-electrons and the N lone pair, then each phenyl group has 6 pi electrons and 6 orbitals (the first 3 are doubly occupied). This is because phenyl is similar to benzene. So you could choose an active space of 18 electrons in 18 orbitals, plus 2 electrons in 1 orbital for the N lone pair. This would give you a pretty large active space of 20 electrons in 19 orbitals. Although this may be possible to do with special methods, I would not recommend it for your first calculations. A smaller and easier choice would be to choose 2 orbitals with 2 electrons for each phenyl, plus the N lone pair. This would give you CAS(8,7). You would have to see that the orbitals that end up in the active space are the correct ones. I have never studied this molecule, but I would look for bonding and antibonding combinations of orbitals that look like the HOMO and LUMO of benzene. If you want to see any spectroscopic property such as UV-Vis then you would have to use around 10-20 roots at least. If that gives you something reasonable you can keep adding occupied and empty orbitals (CAS(10,9), CAS(12,11), etc). Hope this helps!