Tomoaki Shimpo, Tohru Ikegami
The robustness of a test for characterizing hydrophilic interaction liquid chromatography (HILIC) columns against variations in aqueous mobile-phase composition was evaluated using mobile phases of acetonitrile/aqueous mixture (9:1, v/v) in the pH range of 3.6-6.8 of the aqueous part. Ten columns (commercially available and home-made columns) were examined under five mobile phases differing in anion species (formate or acetate), NH4 + concentration and salt concentration. The specific hydrophilicity of the columns was assessed via the retention factor of uridine (k(U)), and the relative retention characteristics (i.e., retention selectivity: α) were evaluated using structure-related probe pairs. The k(U)-values exhibited low variability under low-salt conditions (20 mM in the aqueous part; CVs = 0.01-0.08), but higher variability under high-salt conditions (100 mM in the aqueous part; CVs = 0.03-0.18). Under comparable pH conditions (ammonium formate at pH 4.5 vs. ammonium acetate at pH 4.7), k(U)-values were consistently higher in the former, indicating that differences in NH4 + concentration affect hydration at the stationary-phase interface and thereby influence retention. This suggests that k(U), commonly used as an indicator of hydrophilic retention, can be affected significantly by partition interactions that are controlled by ionic concentration rather than the pH value of the mobile phase. In contrast, the characteristic α-values primarily associated with solute retention via partitioning exhibited high robustness (CVs ≤ 0.03). Notably, α(k(U)/k(2'-deoxyuridine)) values correlated with the estimated phase ratio, suggesting that it reflects the effective volume of the hydrated layer on the surface of the packing materials. Furthermore, the relationships between α(k(1-β-D-arabinofuranosyl uracil)/k(U)) and α(k(vidarabine)/k(adenosine)) indicated that nucleobase‑dependent interactions contribute significantly to stationary‑phase retention. Overall, these results demonstrate that each k(U) and α, particularly those associated with partition-dominated retention, exhibit generally high robustness within this test framework and provide a basis for more robust and transferable HILIC column characterization.