Voigt parameters

Share icon

Hi all,

I have been trying to fit 3 overlapping peaks in my Raman spectra.....I have done the fitting using Gaussian profiles and now i am trying to repeat the process using Voigt profiles. However, i am trying to understand what the parameters stand for. I am aware that voigt is the convolution of a gaussian and a Lorentzian. When i choose to fit voigt, i use some contraints....As far as i can see, the parameters that i can constrain are the following: position, width, height and shape. Nevertheless, i am not sure what this height and shape represent exactly..... 1). The height is different from the amplitude given in the fitting result , 2). Changing the shape i guess i modify the contribution of either the Gaussian or Lorentzian. Is there somewhere that i can see what these parameters represent upon modification? Can i somehow pre-determine (when setting the constraints before the fitting) how much Gaussian or Lorentzian contribution i apply in my voigt profile? Thanks a lot
Thanks for this Kurt....However, this is basically showing how you can use multipeak fitting to fit overlapping peaks and this is not my concern. No matter if its a single peak or multiple peaks, i just want to understand what the parameters of the voigt profile mean. As i previously mentioned, how is the relative contribution of Gaussian and Lorentzian change with respect to how you modify the shape values? You can set initial guesses for the height, but the fitting results give you amplitude as well....What is the difference between height (before fitting) and amplitude (after fitting)? Is there any information somewhere that is specifically explaining what changes someone introduces in the voigt profile by giving various values for shape for example (0.1, 0.5, 1)? Does it become predominantly Gaussian and less Lorentzian that way or vice versa? I hope what im asking seems reasonable. Thanks

October 29, 2015 at 06:59 am - Permalink

Hi Konstantinos,

Konstantinos Chatzipanagis wrote:
However, this is basically showing how you can use multipeak fitting to fit overlapping peaks and this is not my concern. No matter if its a single peak or multiple peaks, i just want to understand what the parameters of the voigt profile mean. As i previously mentioned, how is the relative contribution of Gaussian and Lorentzian change with respect to how you modify the shape values?

In the discussion in http://www.igorexchange.com/node/5612 JohnWeeks points out that you can use the Multi-peak Fit Package to visualise a single peak. - It provides a convenient interface to adjust the peak parameters and see how that affects the Voigt peak-shape.

A quick search using the word 'voigt' in the Igor help files gives the following link:
DisplayHelpTopic("MPFXVoigtPeak")

The shape parameter is described thus: Zero results in pure Gaussian, infinity results in pure Lorenzian, one is 50% Gaussian and 50% Lorenzian.

Hope this helps.
Kurt

October 29, 2015 at 07:25 am - Permalink

Hi Kurt,

Thanks a lot this is more clear now....I did a bit of . searching there before but i missed that bit....Thanks for pointing this out for me. However, i still struggle to see the difference between height and amplitude for a voigt profile but maybe i am missing something again. Thanks a lot again

October 29, 2015 at 07:37 am - Permalink

Konstantinos Chatzipanagis wrote:
However, i still struggle to see the difference between height and amplitude for a voigt profile but maybe i am missing something again.


Technical note TN026 (which is referenced in the MPFXVoigtPeak section of the help) gives the amplitude of the peak as (after translating between different coefficient waves, so that this posting is self-consistent):

amp= cw[2]*exp(cw[3]^2)*erfc(cw[3])

where:
yw = cw[2]*Voigt(cw[1]*(xw-cw[0]), cw[3])

cw[2] is described as the 'amplitude factor', but also notes that the actual amplitude depends upon other parameters as well.

I can't help any more - there are a bunch of references in the technical note that may be of use.
Kurt

October 29, 2015 at 08:59 am - Permalink