Getting rid of Rayleigh Åsmund Rinnan
Introduction Fluorescence Light source Sample Detector Excites sample Emitted from sample
Introduction Fluorescence
Introduction PARAFAC X is the EEM a are the scores b are the emissionspectra c are the excitationspectra E is the residuals Can be seen as an expansion of PCA from two-way data to multi-way data
= A B C Introduction PARAFAC & Fluorescence
Catechol Hydroquinone
Introduction ”Faking” fluorescence
Introduction Light scatter
Introduction Light scatter – The trouble maker Excitation Emission 2nd order Rayleigh 1st order Rayleigh Raman
Introduction Light scatter
Introduction Bi-linearity
Introduction Why is this a problem? X X
Example Fluorescence & PARAFAC
Getting rid of Rayleigh Subtraction of standard Cut off and insert missing/ zeros Weights Modeling of Rayleigh
Subtracting a standard
Missing values Missing values Zeros Signal/ Data area Thygesen, Rinnan, Barsberg & Møller
Example 18 wood samples 4 different levels of p-benzoquinone adsorbed in the fiber cell walls 30 emission wavelengths x 35 excitation wavelengths Thygesen, Rinnan, Barsberg & Møller
WOW! None Weighted Non-Negativity Zeros
So, now Rayleigh is finished, right? The data presented so far is a bit simple Sugar data Excitation Emission 1st order Rayleigh
Weighting - MILES Emission loadingsExcitation loadings
Band of missing values
Using a band of missing values Hard weights Emission loadingsExcitation loadings
Using a band of missing values MILES weights Emission loadingsExcitation loadings
Another method? Why, why, why? The Rayleigh scatter width has to be estimated quite accurately The band width of missing data should also be correct What about an automatic method of removing the Rayleigh scatter, that was not so prone to the estimation of the width of the Rayleigh scatter? Modeling the Rayleigh is the answer!
Modeling Rayleigh A Gauss-Lorentz curve fitting method
Modeling Rayleigh Rinnan, Booksh & Bro
Modeling Rayleigh
With constraints even better Emission loadingsExcitation loadings Rinnan, Booksh & Bro
Thanks to: Rasmus Bro, Karl Booksh, Lisbeth G Thygesen, Søren Barsberg, Jens K S Møller and Charlotte Andersen Thank you for your attention