Abstract
Vibrational spectroscopy combined with chemometric analysis is widely applied across chemistry, pharmaceutical sciences, and the agro-food sector. In this thesis, this approach was used to investigate the effects of pulsed electric field (PEF)-assisted extraction and related mild processing techniques on the structural, compositional, and functional properties of plant-based systems, with particular emphasis on semi-refined flaxseed extracts (SRFE). Multimodal vibrational spectroscopy—near-infrared (NIR), mid-infrared (MIR), and Fourier-transform Raman (FT-Raman)—integrated with chemometric analysis, was applied in Chapters 3 to 5 to characterize processing-induced changes in SRFE, produced from defatted flaxseed meal under four PEF energy input levels (0, 139, 347, and 534 kJ/kg) and two extraction media (water and alkali). In Chapter 6, the spectroscopic–chemometric framework was extended using FTIR-based analysis to additional plant-based systems of increasing matrix complexity, including PEF-extracted chia seed mucilage (CSM), germinated oat flour (GOF), and an oat–pea milk alternative (OPMA) subjected to combined mild heating and PEF treatment.
In Chapter 3, NIR spectroscopy coupled with partial least squares regression (PLSR) was evaluated for rapid, non-destructive prediction of protein content in SRFE. Alkali extraction consistently yielded higher protein content (35.6–45.5%) than water extraction (33.2–37.9%). A plateau effect was observed in both media at medium PEF energy input (347 kJ/kg), beyond which no significant improvement in protein recovery was recorded. Notably, medium PEF-assisted water extraction yielded protein concentrations of alkali-extracted controls, suggesting its potential as a more sustainable extraction alternative. The optimal PLSR model achieved a ratio of performance deviation (RPD) of 8.26, a coefficient of determination for prediction (R2P) of 0.99, and a root mean square error of prediction (RMSEP) of 0.09%, demonstrating suitability for process control applications. Principal component analysis (PCA) of the preprocessed NIR spectra confirmed that the extraction medium and PEF energy input jointly produced distinct spectral fingerprints that reflected differences in protein conformation and matrix composition.
Chapter 4 characterized the extraction-dependent molecular changes in SRFE by integrating MIR, FT-Raman, and NIR spectroscopy with PCA, amide I peak fitting, and two-dimensional correlation spectroscopy (2D-COS). Alkali and water extraction produced fundamentally distinct structural responses to increasing PEF energy input: alkali extraction, progressively disrupted β-sheet structures and enriched α-helical content, while water extraction promoted β-sheet formation and aggregation from an initially α-helix dominant structure. Homo-spectral 2D-COS revealed that alkali extraction proceeds through early polysaccharide–phenolic disruption followed by protein reorganization and lipid removal, whereas water extraction follows a phenolic- and protein-accessibility-first pathway. NIR–MIR hetero-spectral 2D-COS further distinguished extraction-selective from extraction-universal molecular components, with β-sheet changes showing network-wide reorganization and α-helix transitions showing localized, independent behavior.
Chapter 5 established a quantitative relationship between vibrational spectral features and the functional properties of SRFE through compositional, microstructural, thermal, and functional characterization, combined with PLSR modeling using Raman and MIR spectral data individually and as low-level fused data. Alkali-extracted SRFE was globulin-enriched with lower baseline solubility (~58%), high surface hydrophobicity (H0 ~157 × 106 a.u.), and moderate emulsifying activity index (EAI ~80 m2/g). Moderate PEF treatment improved these properties, with solubility and EAI both peaking at 347 kJ/kg (~82% and ~131 m2/g, respectively). Water-extracted SRFE, enriched in albumin fractions and protein–polysaccharide complexes, exhibited higher baseline solubility (~72%), superior emulsifying stability that increased monotonically at 534 kJ/kg, and progressive improvements in water-holding capacity (WHC). Surface hydrophobicity declined monotonically with increasing PEF energy in both systems. PLSR modeling demonstrated that Raman spectroscopy was the most informative single-technique input for predicting solubility and H0, with low-level fusion with MIR further improving prediction performance for both properties. Spectral interpretation showed solubility was influenced by both protein compaction and interactions with polysaccharides and phenolics, whereas H0 mainly reflected lipid‑rich, aliphatic regions on the protein surface and their degree of compaction. PEF treatment reduced these hydrophobic domains, increased their hydration, and therefore lowered H0.
Chapter 6 validated the generalizability of Fourier-transform infrared (FTIR)-chemometric analysis across three plant-based systems of increasing compositional complexity. For PEF-extracted CSM, FTIR-PCA resolved dose-dependent enrichment in protein and oil content concurrent with selective polysaccharide restructuring, including a ~95% increase in galactose at the highest PEF energy input. These compositional shifts were spectrally corroborated by a progressive increase in the lipid carbonyl band at 1722 cm-1, a non-monotonic response of uronic acid bands at 1597 and 1416 cm-1 consistent with PEF-induced cell wall disruption followed by structural degradation at higher energy, and a progressive increase in the amide III band at 1252 cm-1 reflecting enhanced protein release from the cellular matrix. FTIR-PCA also discriminated between ungerminated and germinated samples with ungerminated oat flour (UGOF) samples characterized by higher loadings of storage lipids (1747 cm-1) and complex carbohydrates (988 and 1167 cm-1), while GOF samples were defined by proteins, phenolic compounds, free fatty acids, and cell wall polysaccharides — consistent with the biochemical changes induced by germination. Amide I peak fitting further revealed a temperature-driven secondary structure transition: the β-sheet-dominant native protein (α/β = 0.37) progressively shifted to α-helix- and random coil-enriched conformations at 15–20°C (α/β = 1.57–1.65), with partial reversal to β-sheet and β-turn-stabilized structures at 25°C (α/β = 0.65), concurrent with a 2.4-fold increase in total phenolic content. In OPMA, FTIR-PCA resolved both inter-ingredient compositional differences and processing effects, with preheating temperature dominating the structural response and inducing reorganization into intermolecular β-sheet-rich crystalline aggregates. Spectroscopic evidence showed that the blended OPMA product exhibited enhanced carbohydrate–protein interactions not detected in the constituent ingredients.
The findings of this thesis demonstrate that PEF is an effective and versatile processing technology capable of restructuring plant-based systems at the molecular level, with outcomes strongly dependent on extraction conditions and matrix composition. PEF energy input and extraction medium together control the structural, compositional, and functional behavior of flaxseed‑based systems and these changes can be effectively quantified and interpreted through the integration of NIR, FTIR, and Raman spectroscopy with chemometric tools such as PCA, PLSR, peak fitting, 2D-COS, and data fusion. This approach enabled the development of spectroscopic–chemometric models capable of discriminating treatment-induced variations in protein content, secondary structure, lipids, phenolic compounds, and possible interactions across matrices of increasing compositional complexity. Collectively, these outcomes provide a rapid, non-destructive basis for translating specific PEF-assisted and mild processing conditions into measurable molecular signatures and functional performance in plant-based ingredients and foods.