The Raman spectra were excited by
a 785 nm diode laser (Toptica, Munich, Germany) using an intensity
of 3.5 × 105 W cm–2 yielded when
focusing the excitation light with a 60× water immersion objective
into a droplet of liposome suspension. Raman light was directed to
a single-stage spectrograph equipped with a liquid-nitrogen-cooled
CCD detector (Horiba, Munich, Germany). The spectral resolution is
∼2 cm–1 considering the full spectral range
of 300–1900 cm–1. From each sample, a series
of 300 spectra was collected using an acquisition time of 1 s per
spectrum. The spectra were frequency-calibrated using a spectrum of
a toluene–acetonitrile mixture (1:1).
Preprocessing of
the spectra comprised removal of spikes, background correction using
asymmetric least-squares (AsLS) algorithm,44 (link) and vector-normalization in MatLab R2020b (The MathWorks, Inc.).
After removal of spectra with no signal, there were 98, 136, and 174
spectra remaining for the PA/PC/CER, PA/PC, and CER samples, respectively.
The remaining 98 spectra of PA/PC/CER in a range of 500–1700
cm–1 were subjected to hierarchical cluster analysis
(HCA) using SPSS 20 software (IBM Corporation, Armonk, NY). Band occurrences
in all individual spectra were analyzed using a script in Mathematica
12.1 (Wolfram) as detailed in ref (45 (link)). Average spectra of each HCA class were calculated
for further qualitative analysis of SERS signals.
a 785 nm diode laser (Toptica, Munich, Germany) using an intensity
of 3.5 × 105 W cm–2 yielded when
focusing the excitation light with a 60× water immersion objective
into a droplet of liposome suspension. Raman light was directed to
a single-stage spectrograph equipped with a liquid-nitrogen-cooled
CCD detector (Horiba, Munich, Germany). The spectral resolution is
∼2 cm–1 considering the full spectral range
of 300–1900 cm–1. From each sample, a series
of 300 spectra was collected using an acquisition time of 1 s per
spectrum. The spectra were frequency-calibrated using a spectrum of
a toluene–acetonitrile mixture (1:1).
Preprocessing of
the spectra comprised removal of spikes, background correction using
asymmetric least-squares (AsLS) algorithm,44 (link) and vector-normalization in MatLab R2020b (The MathWorks, Inc.).
After removal of spectra with no signal, there were 98, 136, and 174
spectra remaining for the PA/PC/CER, PA/PC, and CER samples, respectively.
The remaining 98 spectra of PA/PC/CER in a range of 500–1700
cm–1 were subjected to hierarchical cluster analysis
(HCA) using SPSS 20 software (IBM Corporation, Armonk, NY). Band occurrences
in all individual spectra were analyzed using a script in Mathematica
12.1 (Wolfram) as detailed in ref (45 (link)). Average spectra of each HCA class were calculated
for further qualitative analysis of SERS signals.