Analytica Chimica Acta, accepted, 07/07/2015. This is the accepted version without proofing
corrections. DOI: 10.1016/j.aca.2015.06.011
.
Page 6 of 26
3.2 Chemometric analysis: To identify which fluorophores generated the aniso-TSFS
changes, MCR was undertaken on the raw (HH/VV/HV/VH) and aniso-TSFS data. NPFPCA
first estimated the number of principal components in the augmented datasets. This showed
that there were always four significant components (i.e. fluorophores) present in all four
differently polarized datasets, which always explained >99% of the data variance (SI, Table
S-1). MCR modelling using a confidence level of 0.95, non-negativity constraints, and
typically 100 iterations, was repeated three times on each individual replicate set of
measurements (no benefit was seen from undertaking more iterations). This yielded emission
and excitation profiles (Figure 2), and the scores (confidence intervals calculated from the
three replicate measurements) for each component (Figure 3) which were used to generate
reconstructed TSFS/EEM spectra of the individual components (SI, Fig. S-6) as well as the
relative contribution of each fluorophore to each measured polarization state.
Figure 2: Normalized excitation (left) and emission, Δλ (right) profiles of the MCR model components
extracted from the HH (a/b), VV (c/d), VH (e/f), and HV (g/h) polarized fluorescence datasets recorded during
corrections. DOI: 10.1016/j.aca.2015.06.011
.
Page 6 of 26
3.2 Chemometric analysis: To identify which fluorophores generated the aniso-TSFS
changes, MCR was undertaken on the raw (HH/VV/HV/VH) and aniso-TSFS data. NPFPCA
first estimated the number of principal components in the augmented datasets. This showed
that there were always four significant components (i.e. fluorophores) present in all four
differently polarized datasets, which always explained >99% of the data variance (SI, Table
S-1). MCR modelling using a confidence level of 0.95, non-negativity constraints, and
typically 100 iterations, was repeated three times on each individual replicate set of
measurements (no benefit was seen from undertaking more iterations). This yielded emission
and excitation profiles (Figure 2), and the scores (confidence intervals calculated from the
three replicate measurements) for each component (Figure 3) which were used to generate
reconstructed TSFS/EEM spectra of the individual components (SI, Fig. S-6) as well as the
relative contribution of each fluorophore to each measured polarization state.
Figure 2: Normalized excitation (left) and emission, Δλ (right) profiles of the MCR model components
extracted from the HH (a/b), VV (c/d), VH (e/f), and HV (g/h) polarized fluorescence datasets recorded during