Retention, separation selectivity and system efficiency of selected basic psychotropic drugs on different RPLC columns

Abstract Retention parameters of psychotropic drug standards were determined on different columns, i.e., Octadecyl silica, Phenyl, Phenyl-Hexyl, Polar Reverse Phase, Pentafluorophenyl, and Cyanopropyl using aqueous eluent systems containing methanol or acetonitrile as organic modifiers, acetate buffer at pH 3.5 and addition of silanol blocker − diethylamine (DEA). The retention, separation selectivity, and sequence of elution were different when using eluents containing various organic modifiers. The significant differences were observed in retention parameters with a change of the used stationary phase. The various properties of stationary phases resulted in differences in analyte retention, peaks shape, systems efficiency and separation selectivity. The best shape of peaks were on Cyanopropyl (CN) column and the highest efficiency for most investigated psychotropic drugs were obtained on Phenyl-Hexyl and Polar RP columns. Graphical Abstract


Introduction
Column selection in reversed-phase liquid chromatography (RPLC) can become a challenge as the analytes interact with the silica-based stationary phases [1]. One of such interactions is the attraction of cationic form of solutes to the free silanols in silica-based columns, which causes slow sorption-desorption process that gives rise to tailed and broad peaks.
Very important in optimization of chromatographic systems is selection of the type of stationary phase. In case of analysis of basic compounds, optimization of the stationary phase is achieved by minimizing the interaction between analytes and residual silanols [2]. Octadecylsilica stationary phase (C18) is the most popular column packing material. Theoretically, in RPLC, the retention should depend only on a solute partition between hydro-organic mobile phase and the alkyl-bonded stationary phase, which is related to compound hydrophobicity. However, the presence of free silanols on silica-based surfaces is responsible for low sorption-desorption interaction process with cationic analytes.
Several new stationary phases containing improved chemical stability have been developed for more difficult chromatographic separations involving basic compounds in reversed phase conditions, but still for very polar and highly ionizable basic analytes asymmetric peaks, low system efficiency, and poor reproducibility is observed.
Alternatively, the introduction of embeded polar groups or hydrophobic π-π active aromatic moieties to the common n-alkyl chain RP sites generates a concerted π-π RP retention mechanism that, as a consequence of the new functionality, diversifies the common RP interaction properties without altering them significantly. The π-π interactions between solute and stationary phase are significant for retention on Phenyl, Phenyl-Hexyl, Polar RP, Pentafluorophenyl (PFP) and CN columns. In contrast to alkyl-bonded silica stationary phases, there have been only few studies on the retention characteristics of phenyl type stationary phases, but the popularity of these phases for the separation of aromatic compounds has increased significantly. Their ligands are comprised of two sections: the spacer chain that separates and bonds the aromatic ring to the surface and the ring, which imparts the selectivity to the separation. Phenyl type stationary phases exhibit better aromatic selectivity than C18 stationary phases because π-π interactions take place between the aromatic solute and the phenyl group of these stationary phases [3,4]. The length of the n-alkyl chain is very important to define the retention behavior and selectivity of stationary phase, producing phases with different ring orientations [4,5]. The π-π interaction in a chromatographic system can occur between π-electrons of the attached ligands on surface stationary phase and the analytes. An interaction between π-electron containing compounds such as the stationary phase with π-π ligand e.g., phenyl, is favored when one compound is electron-rich (Lewis base) and one is electron-poor (Lewis acid) such as the analyte [6].
The aim of our work was comparison of various columns which can be applied in RPLC systems for their use for basic psychotropic drugs' analysis. The use of the double protection against interaction between basic solutes and free silanol groups, when stationary phase with π-π ligands and mobile phases containing addition of silanol blocker are simultaneously applied, allows to obtain symmetrical peaks and good system efficiency. Obtaining good separation selectivity, symmetrical peaks and high systems efficiency is especially important in the analysis of drugs in different biological fluids when multidrug therapy is used, in pharmacokinetic studies or in the case of poisoning.

Results and discussion
Psychotropic drug standards ( Table 2) were chromatographed on alkylbonded, Cyanopropyl, Pentafluorophenyl, Phenyl, Phenyl-Hexyl and Polar RP columns in eluent systems containing methanol or acetonitrile as organic modifiers, acetate buffer at pH 3.5 and diethylamine (DEA) as silanol blocker. These chromatographic systems were compared in terms of retention of psychotropic drugs on different stationary phases that were selected according to their potential differences in retention mechanism and hence possible changes in selectivity, peak shape and performance. Significant differences in retention, peak shape and system efficiency were obtained on different stationary phases. Fig. 1 presents a comparison of log k values obtained on the different columns with mobile phases containing methanol as organic modifier. Most investigated compounds were more strongly retained on phases with phenyl ligands in comparison with the retention of these compounds on C18 column. As can be seen in Fig. 1 the retention of all compounds was weakest on XBridge Phenyl column compared to results obtained on Xselect Phenyl-Hexyl column. Increased retention of investigated polar drugs was observed on Polar RP column -a monomeric ether-linked phenyl phase with a propyl chain as spacer. A further increasing in retention of most drugs was observed on the PFP stationary phase containing phenyl ligand substituted with five fluorine atoms. The electronegative fluorine atoms produce an electron deficient phenyl ring, so that the PFP phase acts as a Lewis acid or electron acceptor. This is the opposite to phenyl phases, which contain an electron rich aromatic ring. π-π interaction can occur with solutes that are rich in electrons (Lewis bases). The carbon-fluorine bonds of the PFP ring are very polar, thus enabling analytes to also be retained by dipole-dipole and H bonding interactions, resulting in increased analyte retention and selectivity differences. CN phases with short alkyl spacer group show minimal hydrophobic retention.
The presented diagram enables also observation of the selectivity and sequence of elution for investigated drugs by the use of different stationary phases. The selectivity diagrams can be used in practice for rapid choice of the best system for separation of individual pairs or groups of compounds. For example, imipramine and alprazolam, which are poorly separated on C18, CN, Phenyl and Phenyl-Hexyl columns, are well separated on PFP and Polar RP columns. Medazepam and sertindol are not separated on phenyl stationary phase, poorly separated on Phenyl-Hexyl column but are well separated on the other columns.
Similar diagram shows comparison of log k values of psychotropic drugs obtained on different columns in analogous eluent system containing acetonitrile as organic modifier (Fig. 2). With a change of stationary phase, separation selectivity and sequence of elution of the drugs was also different. Selectivity differences obtained on π-π type stationary phases depend on the used modifier. This is due to the fact that organic solvents such as acetonitrile that contain π electrons interact with the aromatic solute, the aromatic stationary phase, or both thus decreasing the selectivity of separation. Solvents (such as methanol), that do not contain π electrons are unable to interfere with the π-π interactions taking place between the stationary phase and the solute molecules.
The great differences in peaks' shape were observed on different stationary phases. Comparison of asymmetry factor (A S ) values obtained on six columns in the same eluent system containing methanol as organic modifier is presented in Fig. 3. The range of correct As values (0.8−1.5) is marked in Figs. 3 and 4 with lines orthogonal to the y-axis. The worse shape of peaks for most investigated psychotropic drugs was obtained on C18 and PFP column -for 10 psychotropic drugs As values were in acceptable range (0.8 < As < 1.5); on Phenyl-Hexyl column for 11 out of 17 compounds values of asymmetry factors were in acceptable range, on Polar RP and Phenyl columns for 12 drugs asymmetry factors were acceptable and on CN stationary phase for 13 psychotropic drugs peaks were symmetrical. For example asymmetry factor obtained for tramadol on C18 column was 2.0, but on Polar RP and CN columns peak shapes were excellent As = 1.18 and 1.03, respectively, but for escitalopram the worst peak shape  was obtained on phenyl column (As = 1.64), the most symmetrical peak was obtained on Phenyl-Hexyl column (As = 1.05).
Less symmetrical peaks were obtained in the mobile phase system containing acetonitrile as organic modifier on most used columns (Fig. 4). On C18 column good symmetry of peaks were obtained only for tramadol and ropinirol, on PFP column for 6 psychotropic drugs peaks were symmetrical, for 8 investigated drugs good symmetry of peaks were obtained on Phenyl-Hexyl, Polar RP and Phenyl columns. However on CN column for 14 drugs asymmetry factor values were in acceptable range. In terms of symmetry of the peaks worst results were obtained in both eluent systems on C18 and PFP columns, good peak shapes were on Phenyl-Hexyl and Polar RP columns, but the most symmetrical peaks were obtained on CN column.
Separation efficiency of investigated systems was compared by theoretical plate number per meter (N/m). Theoretical plate number values were low for most investigated drugs in system with methanol on C18 and Phenyl columns -only for 2 psychotropic drugs N/m was higher than 20 000, on PFP column for 4 drugs, but on Phenyl-Hexyl and Polar RP stationary phases values of N/m > 20 000 were in 8 and 11 cases respectively ( Table 2). Great differences were observed in N/m values obtained on different columns e.g. for escitalopram on C18 column N/m = 7600 but on Polar RP 54160 N/m was obtained; for alprazolam N/m = 7370 and 8310 on phenyl and CN stationary phases respectively but on Phenyl-Hexyl 24260 and on Polar RP 41270 N/m were obtained.
In eluent system containing acetonitrile, phosphate buffer and DEA in almost all cases system efficiency was higher in comparison to system with methanol as organic modifier. On C18 and Phenyl columns for 4 psychotropic drugs N/m was higher than 20000, on PFP for 8 investigated compounds, while on Phenyl-Hexyl and CN for 13 and 14 respectively. On PFP column for 20 psychotropic drugs N/m was higher than 10000. The highest values N/m were obtained for most investigated drugs (for 14 out of 17) on Polar RP column and for 15 compounds N/m > 20 000. The best system efficiency in both eluent systems was obtained on Polar RP column. Table 3 summarizes the results of correlations of log k 1 vs. log k 2 obtained on all applied columns and two mobile phases containing addition of DEA as silanol blocker and methanol or acetonitrile as organic modifier. Slope values of log k 1 vs log k 2 plots give information about the selectivity of separation of investigated drugs on different columns. When the slope value is near 1, the selectivity of separation for most investigated compounds on both columns is similar. Such situations were observed when log k values obtained on Phenyl/ Phenyl-Hexyl, Phenyl/PFP or Polar RP/PFP columns in both eluent systems and log k obtained on C18/PFP with eluent containing methanol or on Phenyl/Polar RP with mobile phase containing acetonitrile were correlated. Low values of slops were observed for correlation of log k values obtained on C18/CN, Phenyl-Hexyl/CN columns in both applied eluents. It indicates on the worse separation selectivity on CN column compared to C18 or Phenyl-Hexyl columns.    Better correlation (r ≈ 1) was obtained for all cases when the correlated log k values were obtained on different columns with eluent containing acetonitrile as organic modifier, which strongly interact with stationary phases containing π electrons. The best correlation was observed between log k values obtained on Phenyl and Phenyl-Hexyl columns in eluent system containing methanol as organic modifier and between log k values obtained on C18 and Phenyl-Hexyl stationary phases in mobile phase containing acetonitrile. Good correlation were also for log k C18 vs log k Phenyl-Hexyl and log k C18 vs log k Phenyl obtained in eluent with methanol as well as for k C18 vs log k Polar RP and k Phenyl vs log k Polar RP when acetonitrile was used as organic modifier. These results suggest that the solute-stationary phase interactions on these columns are more similar than on the other columns and separation selectivity of investigated drugs in these chromatographic systems is similar. The most different separation selectivity was obtained on Polar RP and CN columns in eluent containing methanol and on C18 and PFP columns when acetonitrile was applied as organic modifier. The low correlations were also between log k Phenyl-Hexyl vs. log k CN , k C18 vs. log k CN in system with methanol and between log k Polar RP vs log k PFP , between log k Phenyl vs. log k PFP . The application of these stationary phases enables to obtained greatest differences in separation selectivity, which can be applied in multidimensional separations.

Conclusions
The changes of stationary phases enable variations in retention of compounds, separation selectivity, peaks' symmetry and systems efficiency. The greater differences in obtained results were observed when methanol was applied as organic modifier in comparison to acetonitrile. The separation selectivity for most investigated psychotropic drugs was similar on almost all columns, only on CN column separation of most drugs were worse in both eluent systems.
The best shape of peaks for most investigated drugs was obtained on CN phase, the worse on C18 column in systems with both organic modifiers. Good peaks shape was also obtained for most compounds on Phenyl, Phenyl-Hexyl and Polar RP columns.
The highest efficiency for most psychotropic drugs was obtained on Polar RP column both in eluent system with methanol and as in system with acetonitrile.
The more symmetrical peaks on all columns were obtained in chromatographic system containing methanol in comparison to system with acetonitrile, but system efficiency was higher when acetonitrile was applied as organic modifier.
The differences in retention, separation selectivity and systems efficiency can be applied to select the most optimal chromatographic conditions for the analysis of selected drugs or their mixtures including biological samples.
Comparing, separation selectivity, peak shapes and systems efficiency for most investigated psychotropic drugs the best results were obtained on Polar RP and Phenyl-Hexyl columns in both eluent systems.
The obtained results indicate that the stationary phases containing ligands with π-π electrons compared to C18 stationary phase offer better peaks shape and system efficiency for most investigated basic psychotropic drugs.