Chemometric evaluation of temperature-dependent surface-enhanced Raman spectra of riboflavin: What is the best multivariate approach to describe the effect of temperature?

• We examined the influence of substrate temperature on SERS spectra of riboflavin.
• Changes of SERS features depend on adsorption mechanism of riboflavin molecules.
• Changes of band widths are not proportional to changes of band intensities.
• Spectra can be classified according to temperature using multivariate analysis.

Riboflavin is an essential nutrient involved in energetic metabolism. It is used as a pharmacologically active substance in treatment of several diseases. From analytical point of view, riboflavin can be used as an active part of sensors for substances with affinity to riboflavin molecules. In biological environment, metal substrates coated with riboflavin are exposed to temperatures that are different from room temperature. Hence, it is important to describe the influence of temperature on adsorbed molecules of riboflavin, especially on orientation of molecules towards the metal surface and on stability of adsorbed molecular layer. Surface-enhanced Raman scattering (SERS) spectroscopy is a useful tool for investigation of architecture of molecular layers adsorbed on metal surfaces because the spectral features in SERS spectra change with varying orientation of molecules towards the metal surface, as well as with changes in mutual interactions among adsorbed molecules.In this study, riboflavin was adsorbed on electrochemically prepared massive silver substrates that were exposed to temperature changes according to four different temperature programs. Raman spectra measured at different temperatures were compared considering positions of spectral bands, their intensities, bandwidths and variability of all these parameters. It was found out that increase of substrate temperature up to 50 °C does not lead to any observable decomposition of riboflavin molecules, but the changes of band intensity ratios within individual spectra are apparent. To distinguish sources of variability beside changes in band intensities and widths, Principal Component Analysis (PCA) was applied. Discriminant Analysis (DA) was used to explore if the SERS spectra can be separated according to temperature. The results of Partial Least Squares (PLS) regression demonstrate the possibility to predict the sample temperature using SERS spectral features. Results of all performed experiments and chemometric evaluation procedures show that the temperature dependence of SERS spectral features varies with the temperature program applied. The adsorption mechanism of riboflavin molecules and the molecular re-orientation in the layer adsorbed on silver surface are the most important effects.