Help with fitting parameters

[lkrishna]

Hi,
The ESR data (attached) of Na in Si is broad on the left side compared to right. This is a powder pattern.
Any suggestions on what fitting parameters should I use to capture that broadening.
I was thinking this could be axially symmetric g. I did use 2 separate g-values but that does not help to fit that feature in the left hand side of ESR data.
Thank you
Lakshmi

clear

% below is a standard fitting code
[B,spc] = eprload('test.par')
Sys0.S = 1/2;

Exp.mwFreq = 9.501715; % GHz
Exp.Range = [335.5 340.5]; % in mT Field range use for sims and esfit
Exp.CrystalOrientation = [ ];
Exp.CrystalSymmetry = 63;
Temperature = 78;
ModAmp = 2;
nPoints = 512; % number of points in spectral data vector
B = linspace(Exp.Range(1),Exp.Range(2),nPoints); % field axis


% Now we set up the least-squares fitting.
% First comes a starting set of parameters (which we
% obtain by copying the spin system from the simulation
% and changing a few values)
Sys0.g = [2.003163 2.00600];
%Sys0.lw = 0.5; % mT
Sys0.gStrain=[ 0.00294 0.0166];


% Next, we specify which parameter we want to be fitted
% and by how much the fitting algorithm can vary it approximately.
Vary.g = [0.05 0.05];
Vary.gStrain = [0.002 0.05];
SimOpt.Method = 'perturb';

% Calling the fitting function
FitOpt.Method = 'simplex int'; % simplex algorithm, integrals of spectra
esfit('pepper',spc,Sys0,Vary,Exp,SimOpt,FitOpt);

[Stefan Stoll]

Two things: (1) In the fitting, you could try to vary the g-strain as well. (2) Include some Lorentzian broadening (Sys.lwpp = [0 0.1]).

[lkrishna]

Thanks Stefan!
I tried varying the g-strain and added some Lorentzian broadening. It seems to help.
When I start the simulation I see that at some point there are 2 peaks and then eventually it becomes one broad peak. I am wondering which parameter causes that to happen?
There is a new error that appears when I run the code:
** Spectrum exceeds sweep range. Artifacts at lower and upper limits possible.
Here is my code:
clear

% below is a standard fitting code
[B,spc] = eprload('test.par')
Sys0.S = 1/2;

Exp.mwFreq = 9.501715; % GHz
Exp.Range = [335.5 340.5]; % in mT Field range use for sims and esfit
Exp.CrystalOrientation = [ ];
Exp.CrystalSymmetry = 63;
Temperature = 78;
ModAmp = 2;
nPoints = 512; % number of points in spectral data vector
B = linspace(Exp.Range(1),Exp.Range(2),nPoints); % field axis


% Now we set up the least-squares fitting.
% First comes a starting set of parameters (which we
% obtain by copying the spin system from the simulation
% and changing a few values)
Sys0.g = [2.003163 2.00600];
Sys0.lwpp = [0 0.2]
%Sys0.lw = 0.5; % mT
Sys0.gStrain=[ 0.00294 0.0166];


% Next, we specify which parameter we want to be fitted
% and by how much the fitting algorithm can vary it approximately.
Vary.g = [0.05 0.05];
Vary.gStrain = [0.001 0.005];
SimOpt.Method = 'perturb';

% Calling the fitting function
FitOpt.Method = 'simplex int'; % simplex algorithm, integrals of spectra
esfit('pepper',spc,Sys0,Vary,Exp,SimOpt,FitOpt);

[Stefan Stoll]

Try to search around a little more in the parameter space you have, but it might be that you need to include two lines. However, always keep in mind your sample and make sure your model (one or two spin systems, anisotropic g, presence of g strain etc) makes physical sense.