Reversed-phase liquid chromatography (RPLC) is the most widely-used analytical technique to track organic impurities generated from the synthetic process or degradation of active pharmaceutical ingredients (APIs). Though RPLC is generally considered a mature analytical technique, certain challenges remain. One such challenge is peak tailing. Tailing peaks are a major cause of inaccurate peak quantitation and loss of separation specificity due to ambiguous integration and potential overlap with impurity peaks [1,2]. To provide a quantitative measure of this undesirable peak distortion, tailing factor (Tf) is often specified as a system suitability criteria for RPLC methods in pharmaceutical analysis. Current USP Chromatography Chapter 〈621〉 guidelines recommend peaks of interest tailing factor to be 0.8–1.8 [3]. Therefore, understanding peak tailing in RPLC is important for development of robust RPLC methods to ensure product quality and patient safety.

There have been many studies on the physical origins of peak tailing within RPLC. Peak tailing can be attributed to either saturation of strong sites (thermodynamic origin) or slow desorption kinetics (kinetic origin) due to structural heterogeneith of the stationary phase, which can cause analytes to have a large variation of surface adsorption energy distributions [4]. Such structural heterogeneities usually include different types of silanol groups [5], [6], [7], alkyl bonded layer non-uniformity[8], or metal impurities on the surface [9], [10], [11]. Mechanistic studies on peak tailing have typically employed basic compounds or Lewis acids as probe molecules [12,13]. When acidic compounds have been used, previous peak tailing mechanism studies mostly focused on ionic interactions. Mobile phase (MP) pH has been reported to have a significant effect on peak tailing of compounds with carboxylic acid functional groups [14], [15], [16], [17]. In particular, a higher pH of the mobile phase generally results in a large tailing factor since acidic compounds will remain negatively charged [18]. One common explanation is that increased ionic interaction between the deprotonated carboxylic acid functional groups and surface cationic charges contribute to increased tailing at high pH [14,16,19]. These cationic surface charges are usually associated with metal impurities in silica substrate [20,21] or basic catalyst used in silica-modification reactions [22,23]. On the other hand, end-capping silica-based RPLC columns [14,24,25], using sterically protected silane bonded phase [26], and embedding surface charged groups [27] have been shown to reduce peak tailing, which provides supportive evidence for surface ionic interaction as the root cause.

While peak tailing due to ionic interactions has been extensively studied, the cause of peak tailing for the neutral protonated form of acidic pharmaceutical compounds has received less attention. Schwartz recently used a fluorescent fatty acid as a single-molecule kinetic probe to reveal that probe adsorption on a C18 stationary phase was heterogeneous and involved a broad distribution of site binding energies [28]. The evidence suggests that the acidic compound-stationary phase interaction is more complicated. Gritti and Guiochon used 2-phenylbutyric acid to demonstrate that both ionized and neutral forms of acids can have peak tailing in RPLC but the tailing origins are different [18,29]. In the hydrophobic-subtraction model (HSM) studies, Marchand, Snyder, Dolan and Carr found that peak tailing of non-ionized acidic solutes could occur at pH = 2.8 on both type-A and type-B silica columns, and their tailing factors are weakly correlated to excess column basicity [24,30]. In subsequent studies, the authors attributed excess column basicity to either two adjacent vicinal silanols or metal contamination in the stationary phase [31]. Similarly, Englehardt discovered that in normal phase separation, hydroxyl‑containing compounds have larger retention on vicinal silanol pairs than isolated silanols, likely due to stronger hydrogen-bonding interactions [32]. Meanwhile, Zhang et al. reported that both metal–phosphate interactions and silanophilic interactions contributed to peak tailing of phosphate prodrugs when acidic mobile phases were used for RPLC, and use of non-endcapped columns should be avoided due to the presence of more silanol groups in the stationary phases [33]. So far, there are few systematic studies on how neutral acidic pharmaceutical compounds interact with strong sites in reversed-phase stationary phases and characterization or theorization of the strong sites are scant.

Cimlanod (Fig. 1) was an experimental drug in development for the treatment of acute decompensated heart failure. During RPLC impurity method development for Cimlanod, the peak of a sulfinic acid impurity (Compound A, Fig. 1) was observed to have a significantly larger tailing factor than the peak of an analogous sulfonic acid impurity (Compound B, Fig. 1). This observation offered an opportunity to examine the interaction between neutral acidic pharmaceutical compounds and strong sites on reversed phase stationary phases.

The goals of this work were to better understand the peak tailing mechanism of the sulfinic acid (Compound A) and other similar small organic acids in RPLC and to develop a general approach to minimize peak tailing for acidic pharmaceutical compounds by the rational selection of column and other conditions. To this end, we first delved into understanding strong sites on the acidic compounds, followed by careful study of the strong sites of interaction within the stationary phase. We have shown that tailing of compound A occurs for the protonated state and can be reduced by addition of a competing substrate, such as acetic acid. In general, sulfinic acids have much greater tailing than carboxylic acids. HSM study was further applied systematically to probe the interaction on several RPLC columns. Heated acid wash column treatment proved to be effective in reduction of peak tailing, which provided potential evidence to support the hypothesis that peak tailing was due to interactions between protonated acids and surface vicinal silanol pairs. We also applied density functional theory (DFT) calculations to explore the experimental differences observed between sulfinic acids and carboxylic acids.