Although AV can be directly assessed from a material using Raman spectroscopy, the hardener was first dissolved in glacial acetic acid (AcOH) following the protocol described in Application Note AN-T-239. This approach ensures that both Raman spectroscopy and potentiometric titration (Figure 1) were performed on identical test solutions, enabling a valid comparison between the two methods.

Samples prepared from a commercially available epoxy resin kit were categorized into calibration, validation, and unknown (blind) sets. The calibration set was prepared by dissolving 0 (blank), 68, 116, 208, 315, and 554 mg of the hardener in 25 mL of AcOH. Validation samples were prepared with 308 and 514 mg of the hardener in the same solvent volume. Additionally, five blind samples (A–E) with unknown hardener amounts were prepared to evaluate the model's performance. All samples were prepared in 100 mL beakers.

According to titration, AVs of the calibration set were 30.8, 54.9, 95.2, 147.7, 196.0, and 258.7 mg KOH/g. The calibration model based on titration volume yielded a near-perfect linear correlation, with R² = 1.0000, and RMSEC (root mean square error of calibration) of 0.018 (Figure 2). The predicted AVs for the validation samples were 144.1 and 241.9 mg KOH/g for the 308 and 514 mg samples, respectively, deviating ±0.2% from measured values.

The hardener exhibited a strong Raman peak at 1002 cm⁻¹ which is consistent with aromatic amines such as aniline and phenylenediamine (Figure 3). Glacial acetic acid showed a significant peak attributed to C–C vibration at 893 cm⁻¹. The regions 650–850, 930–1270, and 1550–1630 cm⁻¹ demonstrated minimal spectral overlap between the hardener and AcOH, making them suitable vibrational bands for quantitative analysis of AV.

The intensity of the 1003 cm⁻¹ Raman peak increased proportionally with AV, exhibiting a strong linear correlation (Figure 4). Simple linear regression yielded R² = 0.9965 even without advanced chemometric techniques. This result highlights Raman spectroscopy’s intrinsic quantitative capabilities through direct peak intensity/concentration correlation.

A more comprehensive calibration model incorporating key vibrational bands further improved performance, achieving R² = 0.9999 and RMSEC = 0.79. This model accurately predicted the AV of validation samples with a deviation of ±0.5% from measured values (Figure 2). The Raman-based results were highly consistent with those obtained by titration. These findings support the use of Raman spectroscopy as a reliable secondary technique for the rapid, nondestructive estimation of AV in epoxy formulations.