The biggest overlooked problem with peak fitting UV/Vis absorption spectra of nanoparticles (which is what the original request was about) is not so much the fitting of gaussian peaks to the spectra, but rather, the background scattering contributions. Depending on the size and refractive index of the particles, there will be either a Rayleigh (negative fourth power of wavelength) or a Mie (Riccatti-Bessel function) contribution to the absorption. The former is a monotonic decay, the latter is a multiperiodic oscillation, in wavelength space. You will need to include these functions as appropriate, in your fitting functions, in order to get accurate results. The good news is these contributions are usually smaller than the absorptions, so if your sample concentrations are low enough, you may be able to ignore them, but you'll have to look at the spectra carefully to figure it out. When you say there is a "hump" in the spectrum, that automatically triggers a "Mie Function" flag in my head--particularly when you are analyzing a nanoparticle suspension.
Re. the discussion on the use of GRAMS or other software, the way to think about it is, the software is just a tool. Like a hammer, or a saw. If you dont know how to use it, you can hurt yourself. if you know how to use it, you can make a beautiful, functional product. If you use it for its intended purpose, the results are great. if you use it for an unintended purpose, the results wont be nearly as elegant. Try opening a can of tomatoes with a circular saw one of these days, and you'll see what I mean. No, dont actually try that. Just do a thought-experiment...
Do you want to deconvolve or peak fit? They are not the same.
What do you want to deconvolve? The peak broadening effects from small particle sizes? The instrument resolution function? What do you want to peak fit? The components for different transitions?
Any software that can do deconvolution or peak fitting can be used for your task.
As Jeffrey Weimer mentioned, it would be better if you write more about what exactly you are trying to do.
If I can assume that you are trying to just deconvulate the broad absorption peak (I dont know the reason why you want to do it though!) , you can use origin software; and if you are trying to fit the peak I would recommend you to use fityk (and its free, you can download from: http://sourceforge.net/projects/fityk/).
Thank you for your replies, it seems i have misunderstood what deconvolution involves.
Could you please explain the difference?
I have a graph that has a broad bump of sorts where one should expect the peak for ZnS - i was looking for a way to resolve the peak. If there is a flaw in my approach please recommend what should i do.
You want to do peak fitting to pull out (resolve) component peaks.
Deconvolution as a mathematical operation is the inverse of convolution. Peak fitting as a mathematical operation is a simple optimization of a summation of signals. Unfortuntely the former term is often misused (by spectroscopists) when they really mean they want to do the latter operation.
So your question should say something similar to "Can anyone help me with peak fitting ..."
Could you please "de-couple" physics and mathematics. From the mathematical stand point of view de-convolution irequires sokving an inverse problem. For doing that there are many different approaches,mdepending on the type of a kernel function. The most universal is probably Tichonov regularization. For exponential kernels we have developed a simple method which works very well,mand we could helpmyou with that.
Now, coming back to physics. From your writing I am not sure if I understood completely and correctly what is the priblem you want to solve. Hence, what is an input information and what is the sought for information, and what is the process in which the two are interplaying. Are you shoure that the two are interelated throug an integral or differential relation?
Thank you all for your replies... so what would i need to do to fit peaks?
Could anyone walk me through the steps as i have never done it before; thank you all for taking the trouble to reply and your patience.
In answer to Igor's question and please bear with me, maths has never been my strong point- i have an absorption spectra for zinc sulphide and my "peak" is more of a "bump" i am essentially looking for ways to better resolve this. I hope this answers your question.
With regard to the request to walk you through a method to peak fit ...
* choose the software
* load the data file
* decide the initial parameters for your peak
* set those parameters in the peak fit portion of the software
* run the peak fit
* decide whether the results make sense
You have a lot to do if this is your first time. The experience is not something that can be transferred immediately in an on-line forum. You might be best to find someone at your university to show you by example. Here are your preparation tasks
* read the manual and do the tutorials for the software that you choose, especially with regard to peak fitting procedures
* practice loading and graphing your data files in the software
* determine from the literature what parameters have been used in the past for similar samples to yours
* run the peak fit over many spectra to convince yourself the results are reproducible
You could also connect with someone who has done this type of work before and build a research collaboration to learn how to do the work.
Thank you so much Jeffery, i think il follow your advise, as you said this may be a bit daunting since this is the first time i am trying something like this. I shall probably have to reach out to my colleagues to help me with this.
Thank you so much for your patience and taking the time to help me with this.
If you can, buy the GRAMS32 program which is dedicated to handling all kind of spectra. Can handle most major instrument outcomes and Excel files. Very easy to use.
A simple quick fix if you can transfer your graph into Excel: the book "Excel for Chemists" gives an algorithm for spectra deconvolution and also explains in one or two pages what you're actually doing so you can modify it as you wish. Your library might have this book or you should be able to get it from Amazon or similar site for a few dollars, or I also saw free to download e-book version of this book.
While it may be a good start, you may want to consider a more robust analysis program than GRAMS (or any others solely for peak fitting), especially if you know that you will want to do more than just peak fitting later. In this case, you will want to start with a software application that you can grow to use comfortably and know that you will never feel limited by. You will not want to start with one that is defined for a limited, very specific task. Eventually it will have you going in circles just to get it to do something else, perhaps something even as simple as to prepare a high quality graph of your final results for a publication (or do non-linear regression fits with a complex function that models other data you collect).
I recommend Excel only to accountants or otherwise only to folks who need to do robust spreadsheet-type calculations. It certainly is highly adept for this. I avoid using Excel beyond this. The fact that "Excel for Chemists" apparently does not recognize the important fundamental difference between algorithms for peak fitting and those for deconvolution says quite a bit in this regard.
As to finding further references for peak fitting, any good software tool that is designed properly for peak fitting will have plenty of resources for you to read about how to do peak fitting, such as manual pages, tutorials, and on-line forum discussions. You can even do a Web search with the query "tutorials on peak fitting".
You might be absolutely right with all these notions, especially with the further growth option. Yet for me as a mortal being with many different tasks, GRAMS is complicated enough. Rarely have spectroscopy problems beyond its capabilities.
Abinaya does not sound like a deconvolution or peak fitting expert hence the more simple is his beginning the easier he can start. This is why I recommended that I did. The simplicity of these things I know from experience. Maybe I just don't know the selection of programs out there, but with handling ten different softwares one does not feel the urge to learn a new complicated one unless absolutely necessary.
Anyway Abinaya is free to accept or decline any suggestion from this web site.
The biggest overlooked problem with peak fitting UV/Vis absorption spectra of nanoparticles (which is what the original request was about) is not so much the fitting of gaussian peaks to the spectra, but rather, the background scattering contributions. Depending on the size and refractive index of the particles, there will be either a Rayleigh (negative fourth power of wavelength) or a Mie (Riccatti-Bessel function) contribution to the absorption. The former is a monotonic decay, the latter is a multiperiodic oscillation, in wavelength space. You will need to include these functions as appropriate, in your fitting functions, in order to get accurate results. The good news is these contributions are usually smaller than the absorptions, so if your sample concentrations are low enough, you may be able to ignore them, but you'll have to look at the spectra carefully to figure it out. When you say there is a "hump" in the spectrum, that automatically triggers a "Mie Function" flag in my head--particularly when you are analyzing a nanoparticle suspension.
Re. the discussion on the use of GRAMS or other software, the way to think about it is, the software is just a tool. Like a hammer, or a saw. If you dont know how to use it, you can hurt yourself. if you know how to use it, you can make a beautiful, functional product. If you use it for its intended purpose, the results are great. if you use it for an unintended purpose, the results wont be nearly as elegant. Try opening a can of tomatoes with a circular saw one of these days, and you'll see what I mean. No, dont actually try that. Just do a thought-experiment...
I prefer the analogy that one might try to open a can of tomatoes with a plastic spoon when it comes to recommendations to use Excel for data analysis. Eventually you may get what you want, but it will be a long and painful process (and you will be even more hungry at the end).
Thank you everyone, for clarifying a number of things; i think i will first have to read up to get a better understanding of what i hope to accomplish. Any suggestions on that front would be greatly appreciated.
@ Ananth most of the particles in my sample seem to be spherical or quasi spherical so how would i include mie contribution to my calculation.