Skip to content

Week n+13

Monday 27/9

What I did today:

  • I decided that I want charged of each of the derivatives, so I’m optimising them in ORCA and and will use Multiwfn to get their charges
  • In this porcess I realised that the geometry for the \(\ce{NO2-NH2}\) geometry was actually just \(\ce{NO2-H}\), so I guess I have to re-do that :/

Tuesday 28/9

What I did today:

  • After chatting with Michael, I’ve started to re-structure my thesis a bit, so that I can write more of a combined methods and results section.
  • Since all the TS geometry have been completed, I’ve rendered out those as a pretty figure
  • Woo! I’m making progress on my writing!!! I’mma make a writing matrix to keep track of how I’m going
  • I’ve gotten my results from the ba21t job, and I have a sneaky negative frequency in there… normally I wouldn’t care too much, but my transmission coefficient is at 13.4, so the reaction is pretty much instantaneous. I’m not particularly happy with that assessment, so I’m doing a full hessian guided optimisation to see if it comes down a bit
  • I’m getting a bit desperate, so I’m running a crest job to see if I can find a more stable configuration of ba22r as well 😑

Wednesday 29/9

What I did today:

  • Second dose AZ 😄
  • Finished the first draft of \(F_Z\)
  • Put a lot of work into figuring out the Wigner transmission coefficient approximation, only to find that it didn’t fix my problem anyway

Thursday 30/9

What I did today:

  • Since the benchmarks are all done, I’m removing that matrix from my notes
  • Working on the CPCM section was interesting, as all of the figures that I had used ended up being replaced by new ones that served the narrative a bit better
  • I’ve completed the field scans sections, but I’m not sure that I’m saying enough there. I just don’t really know what I have to say to say about it. I think that In general I should consider this primarily a methods writeup and I can come back to flesh out the discussion components a bit later.
  • I’ve been trying to understand the mechanisms behind the derivatives, and I’ve reached a point where I think I know what’s going on, but I’m creating some EDDs to try and double check the logic.

Friday 1/10

What I did today:

  • EDDs finished last night at like 10pm, so I’ve left them till this morning to interpret. Preliminary results seems to support my theories about how they’re working though!
  • I got pretty distracted getting PyMOL running with openVR…
  • Trying to figure out these derivatives…
    • I think I got them sorted in the notes below, and have made some generalisations about how they work.

Screen Shot 2021-10-01 at 4.42.55 pm

Notes table

System Effect on S relative to H-H Effect on R relative to H-H S R
\(\ce{R1 = H, R2 = H}\) - - - helps hybridisation (\(\ce{N->\beta}\)) - prevents hybridisation
\(\ce{R1 = H, R2 = NH2}\) ⬇︎ ⬇︎ - makes C more electrophilic (\(\ce{\beta->\pi}\)) - makes C more electrophilic (\(\ce{\pi->\beta}\))
- makes N more nucleophilic (\(\ce{\pi->N}\))
\(\ce{R1 = H, R2 = NO2}\) ⬆︎ ⬆︎ - Helps hybridisation (\(\ce{N->\beta}\))
- makes C less electrophilic (systematic) (\(\ce{\beta->NO2}\))
- makes C less electrophilic (systematic) (\(\ce{\beta->NO2}\))
- prevents hybridisation
\(\ce{R1 = NH2, R2 = H}\) (reversed) ⬆︎ ⬆︎ - flipped
- Pushes electron density back onto the cyclising amine
- flipped
- makes C less electrophilic (\(\ce{O->\pi->\beta}\))
\(\ce{R1 = NH2, R2 = NH2}\) ⬆︎ ⬆︎ - flipped
- Pushes electron density back onto the cyclising amine (more available than in NH-H)
- makes C less electrophilic (\(\ce{O->\pi->\beta}\))
\(\ce{R1 = NH2, R2 = NO2}\) ⬆︎ ⬆︎ - Helps hybridisation (\(\ce{N->\beta}\)) - prevents hybridisation
\(\ce{R1 = NO2, R2 = H}\) ⬇︎ ⬆︎ - Helps hybridisation (\(\ce{N->\beta}\))
- prevents hybridisation
- Negative charge accentuates OEEF
\(\ce{R1 = NO2, R2 = NH2}\) ⬇︎ ⬆︎ - Helps hybridisation (\(\ce{N->\beta}\))
- prevents hybridisation
- Negative charge accentuates OEEF
\(\ce{R1 = NO2, R2 = NO2}\) ⬇︎ ⬆︎ - Helps hybridisation (\(\ce{N->\beta}\))
makes C less electrophilic (systematic) (\(\ce{\beta->NO2}\))
- prevents hybridisation
- Negative charge accentuates OEEF
- makes C less electrophilic (systematic) (\(\ce{\beta->NO2}\))makes C less electrophilic (systematic) (\(\ce{\beta->NO2}\))
  • General trends:
    • \(\ce{R2}\) has more effect on the R isomer than the S
      • Effects in the order of \(\ce{NH2<H<NO2}\)
    • \(\ce{R1}\) has more effect on the S isomer than the R
      • Effects also in the order of \(\ce{NH2<H<NO2}\)
    • Both \(\ce{R1}\) and \(\ce{R2}\) on their own, with normal \(F=Z/-Z\) (where the other \(\ce{=H}\)) causes splitting to occur with S decreasing in energy and R increasing
  • General mechanisms:
    • S
      • Helps hybridisation (\(\ce{N->\beta}\))
      • Makes C more electrophilic (\(\ce{\beta->\pi}\))
      • When OEEF flipped, Pushes electron density back onto the cyclising amine
    • R
      • Prevents hybridisation
      • Makes N more nucleophilic (\(\ce{\pi->N}\))
      • When \(\ce{R2=NO2}\), makes C less electrophilic (systematic) (\(\ce{\beta->NO2}\))
      • When \(\ce{R1=NO2}\), engative charge accentuates OEEF
      • When OEEF flipped, makes C less electrophilic (\(\ce{O->\pi->\beta}\))

Writing Matrix

Section Status Notes In Progress
Abbreviations Later
Abstract Later
Intro Second draft done
Computational details First draft done
Reaction benchmarking To do
Static \(F_Y\) First draft done
Static \(F_Z\) First draft done
Static \(\varepsilon_r\) First draft done
Efield Scans First draft done I’m not sure what else to add here, but it feels weak.
EDD maps First draft done perhaps needs a discussion of the implication of these mechanisms?
Relaxed \(\vec F\) To do ✅
Derivatives First draft done
Conclusion Later
Acknowledgements Later
Appendices Later/WIP