alan-jarvis
Greetings,
I've got some synchrotron data for an alumina coating on aluminium: 3 ceramic phases and some aluminium from the substrate.
I'm running a Pseudo-Voigt refinement where I'm trying to refine each peak separately.
I have the following for each peak:
xo_Is
xo @ VALUE_0`
peak_type pv
pv_lor @ VALUE_1`
pv_fwhm @ VALUE_2`
I @ VALUE_3`
Where VALUE_i is a number.
Couple questions please:
(1) How should I be building up the refinement for a new spectra?
Should I add one of these xo_Is refinements for each big peak and THEN run the refinement?
Or can I do it piecemeal?
And perhaps limit the range (by start_X and finish_X)?
But at some point I have to add xo_Is terms.
Does it matter how I do all this?
(2) Using the above example, what starting values should I give for VALUE_1, VALUE_2 and VALUE_3?
I suppose I can give an estimated value from the spectra intensity for VALUE_3.
But what about VALUE_1 and 2?
I know that they can range as follows:
pv_lor: 0 to 1
pv_fwhm: 1e-6 to "2Val+20" (which I'm not sure what it means)
Should I make an estimate for pv_fwhm, based on the peak itself?
And for pv_lor maybe start with 0.5?
Will it affect the refinement if I don't make good estimates?
(3) What happens when I add a new xo_Is term to my refinement?
Let's say I have a funny shaped peak that needs to have two phases (two xo_Is).
Or if I want to account for a small peak, or increase the range I'm refining.
Do I need to re-set all values, or just the nearby refinements of pv_lor, pv_fwhm and I?
I find sometimes I get a better fit doing this, just by trial and error.
But I don't really know what I'm doing.
Is there a more rigorous way of doing this?
Thanks,
Alan Jarvis
Sheffield
alancoelho
Use the following order:
xo_Is
peak_type pv
pv_lor @ VALUE_1`
pv_fwhm @ VALUE_2`
xo @ VALUE_0` I @ VALUE_3`
xo @ ... I @ ...
Note, if you use indentation (optional) then you will see the dependencies more clearly.
Start VALUE_1 at 0.5
Start VALUE_2 to an approximate FWHM from your data
Start VALUE_2 to the approximate peak position
Start VALUE_3 at 100 say
>Will it affect the refinement if I don't make good estimates?
Try a few refinements changing parameters values and you will get feel for what's possible.
alan-jarvis
Thanks: I shall do that.
And what should I do when I'm adding an additional xo_Is peak refinement?
Say that I've done a refinement for a pattern looking at the major peaks and now want to try to account for a smaller peak.
Or else I increase the range of my refinement.
Do I need to "re-set" the values in the nearby xo_Is peaks that have ALREADY been refined?
I HAVE done some experimentation here by trial and error and it DOES seem to make a difference.
But it seems a bit unpredictable; I get different results depending on what I reset.
I don't know how many of the peaks I need to "re-set".
And if some refinements are at LIMIT_MAX or LIMIT_MIN I am ASSUMING that they don't stay there.
I don't know how your software works.
Do I need to reset these manually?
I have read Section 3.8 but it's not clear to me what happens.
alancoelho
>Do I need to "re-set" the values in the nearby
>xo_Is peaks that have ALREADY been refined?
Only if the parameter refined to something way off. This is really a general refinement type question then you need to look into the literature for explanation.
>I have read Section 3.8 but it's not clear to me what happens.
LIMIT_MIN / LIMIT_MAX are warnings that lets the user know that the parameters are close to thier limits. You can turn them off if you like using:
no_LIMIT_warnings