I want to deconvolute the PL spectra for ZnO nanoparticles. How can I check which deconvoluted spectra is correct?
As Sergio already pointed out, the recognition of a meaningful fit is not easy whereas a good fit always means that the deviation between theoretical and experimental curves are as small as possible. However, this is only a mathematical description. Physically it does not mean anything. You have to understand the profile functions you are using, and how many of them you are applying. It should be clear that with increasing number of functions the mathematical deviation becomes smaller and smaller, but physically the consideration of many peaks is useless. In contrary, the smaller the number the simpler the model used and the easier any application of the entire procedure. Important is that you never forget what you are doing. It is a fit, nothing more. You have to define limits and constraints in order to get a meaningful interpretation. And: everything could be wrong since there are possibly also other opportunities to find good matches selecting different models or constraints.
The deconvolution is a process which we use a mathematical function (depending of the experiment it is usually a peak function) to fit determined experimental data, usually it helps us observe features that are hidden by stronger signals, or to see shifts that the raw experimental data do no show clearly, among other occasions. The question regarding to check which deconvolution is correct is hard, usually some conditions can be applied, like usually the smaller quantity of functions to reach a reasonable fit, and pay attention to the resolution of the experiment you are fitting. Commonly you know a deconvolution is good if all the fitted function have a known physical meaning.
Some very good papers exist containing the size, defect and structure relations of ZnO nanoparticles with the PL. I general you will observe green PL as in our own publication. I highly recommend the works of Djurisic.
https://www.google.fr/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&ved=0ahUKEwjNzfXwpP3VAhVCVBQKHSTqBmIQFgg8MAQ&url=http%3A%2F%2Fxa.yimg.com%2Fkq%2Fgroups%2F13354653%2F7837681%2Fname%2Fik.pdf&usg=AFQjCNEGLD59x8m4Fntw-tOrWuEcRuacTA
Article ZnO nanoparticles as polymerisation Photo-Initiator: levulin...
As Sergio already pointed out, the recognition of a meaningful fit is not easy whereas a good fit always means that the deviation between theoretical and experimental curves are as small as possible. However, this is only a mathematical description. Physically it does not mean anything. You have to understand the profile functions you are using, and how many of them you are applying. It should be clear that with increasing number of functions the mathematical deviation becomes smaller and smaller, but physically the consideration of many peaks is useless. In contrary, the smaller the number the simpler the model used and the easier any application of the entire procedure. Important is that you never forget what you are doing. It is a fit, nothing more. You have to define limits and constraints in order to get a meaningful interpretation. And: everything could be wrong since there are possibly also other opportunities to find good matches selecting different models or constraints.
Thanks everyone for your answers.You guys helped me to understand the topic.
So the peaks in my PL spectra (in experimentally deconvoluted fit) should have the same profile as that of theoretical fit plus it should also have minimum number of peaks nearer to the peaks in theoretical fit. But about what constraints you are talking?
You should also concider that your optical receiver most likely have a wavelength dependent quantum yield.
Article About the possibility of calibrating optical detectors by so...
- Badges
- Science topic
- Similar topics
- Chemistry
- Nanotechnology
- Nanostructured Materials
- Nanoparticles