Pepper simulation depends on field range

[howardjou]

Hi everyone,
I try to simulate an 1 electron and 2 nuclei system by pepper.
However, I notice that the simulation depends on the exp.range I use.
I have set exp.npoint = 100000, so I don't think number of points is the reason.

Here is my code. The result of Exp.range = [0 18000] and Exp.range = [10000 18000] are different.
I don't know how to post pictures in this topic. You can run this code in different Exp.range and notice that the simulation depends on field range.

Code: Select all

Sys.g = [1.99 1.999];
Sys.S = 1/2;
Sys.lwpp = 20;

Sys.A = [500 10; 100 10]; 
Sys.Nucs = 'Mn, Mn';
Exp.nPoints = 100000;
Exp.mwFreq =397.3; %in GHz
Exp.Temperature = 10; %in K
Exp.Range = [0 18000]; %in mT
Exp.Harmonic = 1;

Exp.Mode = 'perpendicular';
[B, spec] = pepper(Sys,Exp);
spec = (spec - min(spec)) ./ (max(spec) - min(spec));

plot( B/1000, (spec - mean(spec)), 'r')
xlim([14 14.5]);

[Stefan Stoll]

What's happening here is that the automatic transition selection is not identifying all EPR transitions. This happens for two reasons: (a) the field range is much wider than the spectral range, (b) there are several nuclei with hyperfine couplings that give (almost) degenerate energy levels.

There are several ways around this:

• Turn automatic transition selection off with Opt.Threshold = 0.
• Use perturbation theory instead of matrix diagonalization, with Opt.Method = 'perturb2'.

Here is a script were the simulated spectrum is now independent of the field range:

Code: Select all

clear

Sys.S = 1/2;
Sys.g = [1.99 1.999];
Sys.A = [500 10; 100 10]; 
Sys.Nucs = 'Mn, Mn';
Sys.lwpp = 20;

Exp.nPoints = 100000;
Exp.mwFreq = 397.3; %in GHz
Exp.Temperature = 10; %in K

Exp.Range = [0 18000]; %in mT
%Exp.Range = [10000 18000]; %in mT
%Exp.Range = [14000 14500]; %in mT

Opt.Method = 'perturb2';
[B,spec1] = pepper(Sys,Exp,Opt);

Opt.Method = 'matrix';
Opt.Threshold = 0;
[B,spec2] = pepper(Sys,Exp,Opt);

plot(B/1e3,spec1,B/1e3,spec2)
xlim([14 14.5]);

Another comment: It looks like you don't have any resolved hyperfine splittings. In this case, it might be more efficient to exclude the nuclei and instead use an anisotropic line broadening with Sys.HStrain.

[howardjou]

Thank you so much