So... things gets more weird in my PhD:

The graph shown here are IRC runs of a triplet TS (blue) and a singlet TS (red) of a molecule torsional surface

From the look of this graph, one might be tempted to think there is a conical intersection near the middle of the graph

However, the singlet and triplet geometries at the energetically degenerate points are actually 90 deg rotated from each other

Currently trying to find if gaussian or other software can compute an "SCF isoenergetic ridge" to see if these two isoenergetic structures are actually connected by some barrierless pathway

Having said that, since spin flipping is not a continuous parameter, it seems doubtful there is a smooth path that connects between the two isoenergetic conformers of different spin states

physicstoday.scitation.org/doi/abs/10.1063/…

What's even more bizarre is that the SCF energies of the calculated conical intersection using easyMECP has the same structure as the one at the centre of the graph, but their SCF energies differ by 0.04 Hatress, which is around 20 kJmol-1

So does the two TSs actually shared a conical intersection or not, I have no idea yet

Meanwhile the conical intersection calculated using the penalty function method is 60 degrees rotated from the easyMECP structure, and it would be located offscreen to the right of this graph, thus it is unlikely that is a conical intersection unless these surfaces happen to intersect outside the 2D plane of this graph

In short: A lot of confusing things where either the energy or the geometry does not match up with the calculated conical intersection structures and the calculated structures using two different but well established methods disagree with each other

Btw, the "singlet" has a $<S^2> = 1.610$ suggesting heavy triplet contamination

i would suggest to start looking at the contaminated part to find the problem

easier said than done, there are almost no codes on multi reference DFT

that's why they call it "research" :P

Trouble is, since this is my 3rd year, I do not have the luxury to write a code from scratch (and more importantly) test it with a benchmark to show it works

I have a long history in my life that whether good or bad near misses, showed up a lot more often than most of my peers, meaning the tiniest and most obscure errors can show up in a place overlooked and make the resulting output completely junk without knowing it has gone junk

Almost all my work requires basically a complete carpet search style check of everything that is ran in order to be 100% confident all these obscure errors are avoided

which means, no shortcuts unless it is well established product or program

has your advisor made any suggestions yet?

Well, the above problem on conical intersections turns out to lie outside of both of their fields of expertise so for the moment being, I am on my own. They have suggested some leads I can follow, and I have also done some further literature searches, but nothing very helpful yet

Current hope is a program that allows me to verify via surface hopping, but so far I have yet to get it to work without it complaining I am not putting an equilibrium geometry while in fact it is equilibrium already

don't they know anybody in this field who could help you out?

Tried 3 of these, nothing useful yet, the last guy I am still waiting for reply

Our school and my old uni do not have anyone that works in the domain of spin forbidden organometallic reactions

and none of their contacts are familiar with this domain as far I am aware

My literature searches have found 2 methods that deals with this domain, but the codes they use are not found anyway except mentioned in the article as an internal thing

I still have the remaining 15 articles to read which may or may not touch upon this conical intersection specifically in the domain of organometellics reactions

sounds like you've got a lot of searching to do, pal

and wrestling so many program things that are NOT even chemistry

I realise how 90% of my PhD time is not chemistry, but trying to get codes to work

@andselisk, May I know why this edit suggestion is rejected - chemistry.stackexchange.com/review/suggested-edits/105909. Since I think superposable is the correct term instead of superimposable in the context.

And I have made some minor fixes like spelling correction, removal of space, and removal of article 'a'.

@Martin-マーチン, Your guidance is also required. Could you please tell why my edit has been rejected. Sorry for disturbing you for this.

@Intellex Everything was OK, or at least OK enough to make your suggestion too superfluous.