Spectra from Orca Calculation with many hydrogens

[SebastianP]

Hi all,

I have done an EPR computation with Orca to compare to an experiment, but I cant get a reproduction of the experimental spectra.
The molecule is an azobenze with two TEMPO groups added, so it has quite a lot of hydrogen atoms. While I initially planned on using the garlic method to generate the spectrum it apparently doesnt work with the nitrogen atoms of the azo-bridge, so I switched to the chili method. Since the resulting matrix was way too big I tried changing the gridsize, and I also tried reducing the space by using mt2mhz(0.14) as a cutoff to reduce the number of atoms to 8 hydrogens, 2 nitrogens and 2 oxygenatoms. I also changed the hydrogen-, oxygen- and nitrogen atoms to their most common isotope. However this still needs an enormous amount of memory, so i couldnt produce any spectra. However, if I reduce the number of atoms to consider even further, only the nitrogens and oxygen atoms remain and the resulting spectrum does not match the experimental spectrum (a singlett compared to a triplett).
Is there any other way to reduce the space further or get a result for a calculation with this many hydrogen atoms (it should be an isotropic epr)?

Thanks in advance,
Sebastian

[DanielKlose]

Hi Sebastian,

garlic supports equivalent nuclei, which can be many without much computation time when the correlation time is omitted. See example for 2 nitrogens and 12 protons:
Sys.g = [2.0084 2.0063 2.0024];
Sys.Nucs = '14N,1H';
Sys.n = [2,12];
Sys.A = [110, 5];
Sys.lwpp = 0.1;

Exp.mwFreq = 9.5;
Exp.Range = [315 365];

garlic(Sys,Exp);

Would this solve your problem since you asked for an isotropic spectrum?
And just as a hint, if you need anisotropies like the g-anisotropy in the example above, it will likely be necessary to obtain those from low-temperature EPR, or also reduce the g-tensor to an isotropic value.
For more specific help, it would be useful to share more information on the spectrum that you want to model.
Best wishes,
Daniel

[SebastianP]

Hi Daniel,

thanks for your advice. I had a look at my System again and realised that removing the Sys.S variable allows me to use garlic directly on the system gained from orca2easyspin with a low cutoff point (including 2 nitrogen, 2 oxygen and 12 hydrogen atoms). The spectrum is still sadly a single peak, but that might be a problem with my calculation. I'll go back to the computation and try again, but at least I can now generate a spectrum without breaking my computer

Thanks again and best wishes,
Sebastian

[SebastianP]

HI all,

I tried a lot in the last few weeks to get my spectra to match the experimental spectra. However it's still not working
First, I realised that since my molecule has two TEMPO groups I need to describe it as two spin systems with Spin=1/2 each. Garlic cant compute that and just putting S=1/2 there gives me a quintet which is not what I expected.
So then I computed a coupling constant J and tried to do it with chili, copying the g value over and splitting the HFC matrix into two where half of the the cores only interact with one spin system (basically following the advice in this thread: https://easyspin.org/forum/viewtopic.ph ... 0d61#p1592)
Chili is obviously much more expensive than garlic so I had to also reduce the amount of nuclei included. I got it down to eight nuclei or 10 when I treated the ones with low HFC constants with PostConvNucs. But even when I changed the tcorr value and the J value around, the resulting spectra was always the same quintet I got when I misused garlic. What got me to the result I wanted, however, was deleting all nuclei belonging to one of the spin systems and just keeping one, with S=1/2 (which is the same as just a single TEMPO radical without anything else included, the experiment yielded the same for my doubly substituted molecule).

My Sys struct looks like this:

Code: Select all

xyz: [84×3 double]
S: [0.5000 0.5000]
g: [2×3 double]
gFrame: [2×3 double]
Nucs: '1H,1H,1H,14N,16O,1H,1H,1H,14N,16O'
NucsIdx: [12 32 38 41 47 56 64 65 73 79]
A: [10×6 double]
AFrame: [10×6 double]
data: [1×1 struct]
lwpp: [0 0.1000]
J: 1.1992e+03

with Sys.A looking like

Code: Select all

 0.9541    1.5357    4.0206         0         0         0
    0.9452    1.5242    4.0095         0         0         0
    1.8701   -2.5631   -4.0170         0         0         0
    4.9182    5.1867   40.8939         0         0         0
   20.9661   23.0396  -99.2039         0         0         0
         0         0         0    1.0028    1.5917    4.0709
         0         0         0    1.8712   -2.5581   -4.0105
         0         0         0    0.9621    1.5426    4.0239
         0         0         0    4.9351    5.2034   40.8908
         0         0         0   20.9802   23.0543  -99.2749

Is there anything else I can try or did I forget something important? Thanks in advance again

[DanielKlose]

Hi Sebastian,

Of course you have quite a complex system, particularly in terms of hyperfine couplings. However, the description of the electron-electron interaction with only a scalar J coupling might be the problem yet to be solved.

This has been treated in simulations by EasySpin in detail in the paper here: https://doi.org/10.1039/C8CP05236K
for the case of DNP biradicals using multifrequncy EPR simulations. And I think this should help you - otherwise we would need to know more details of the system you are trying to describe.
Best wishes!