Chocolate melts between 30–36  °C. 100% chocolate was used to establish the laser-induced melting threshold due to its darker color. The darker sample absorbs more laser light and, as a result, melts at lower powers and shorter exposure times.

Optimizing laser power is critical to prevent thermal damage or structural changes during measurement. Raman spectra collected at 5%, 10%, and 15% laser power revealed notable shifts in cocoa-related peaks (Figure 4a), with visible melting at 10%. Lighter-colored chocolates tolerated higher powers, generally up to 15%.

However, melting is not the sole indicator of heat-induced structural changes. Even white chocolate exhibited subtle crystallinity shifts in the 1060–1100 cm⁻¹ range when laser power exceeded 10% (Figure 4b). These results highlight that chocolate can undergo thermal alteration at relatively low laser powers, emphasizing the need for careful power selection during quality assurance and quality control (QA/QC). Fluorescence rejection methods combined with lower-power 785 nm excitation offer potential solutions.

The major sugar peaks are clearly observed in white, milk, and 70% chocolates (Figure 5). However, in 85% chocolate, the only noticeable sugar-related spectral feature appears at 1460 cm⁻¹. This suggests that Raman effectively determines sugar content for QC measurements. Cocoa-related ingredients exhibit characteristic Raman bands around 1300 cm⁻¹ and 1420–1480 cm⁻¹. Confining a Partial Least Squares (PLS) model to these spectral regions resulted in the most accurate predictive model for cocoa content analysis.

PLS models built from key peaks between 1200–1600 cm⁻¹ in the Raman spectra of various chocolate types show strong agreement between predicted and measured cocoa content, with low standard error. This confirms Raman spectroscopy’s effectiveness for routine cocoa content analysis (Figure 6a). Adding data points would enhance confidence in predictions at higher cocoa levels.

The sugar content model demonstrates even greater predictive accuracy, attributed to distinct sugar peaks and the lack of temperature-related variation in sugar measurements. Sugar data may also refine predictions of cocoa-related content, as sugar’s Raman intensity varies proportionally with cocoa content (Figure 6b). Both sugar and cocoa content are important QC parameters measurable via Raman spectroscopy.