EasySpin forum

Hi !
I have a problem with defining the proper line broadenings.
I am trying to fit 3 Systems to my EPR spectra (W-band EPR) and this is the best what I can obtain (spectrum added as an attachment).
I have used the following parameters:
mwFreq=94490000000.0000
Sys1:
weight=0.502
g=[2.0045 2.00509 2.00511]
lwpp=1.306
Sys2:
weight=0.35
g=[2.00532 2.00322 2.00352]
lwpp=4.18
Sys3:
weight=1.269
g=[2.00483 2.00469 2.00311]
lwpp=2.73966

Could you provide me with any suggestions of how I can improve my fit?

Best regards
Krystian

You could add some Lorentzian character into your linewidths, that may let you capture the trailing to low field feature better. Alternatively you could think about g-strain. But that adds a whole additional dimension of parameters to what in my opinion is something that is already a little over parameterized, given the observed structure.

Hi Matt!

What do you mean by "over parametrized" ? Do you think that the number of systems should be reduced ?

I would go with the g-strain as Lorentzian did not provide me with satisfactory improvement. Thank you very much for your advice !

By over parameterized, I mean (in my opinion) that you are trying to fit too much in your spectrum, three different spins each with 3 g-values, a linewidth and a weight. 15 parameters and they are all likely to be highly correlated. So while technically you can pull all of that from your spectrum, in practice there are likely to be several equally good looking minimal with vastly different parameter sets.

Unless you have good reason to include three spins you might want to reconsider the number, I would say there is maybe 2 in there or a single one with some large hyperfine. If you have any idea of what might actually be present, perhaps try some electronic structure calculations and use Orca to estimate the g-tensors. That way you can restrict the parameter space and seed it with good initial guesses.

I agree with Matt, the spectrum is too featureless to allow you to extract 15 parameters from 3 assumed components if you don't have any other additional information.