ADVANCES IN HYDROPHILIC INTERACTION LIQUID
3. SEPARATION MECHANISM AND EFFECTS OF THE ADSORBED WATER AND MOBILE PHASEAND MOBILE PHASE
3.1 ADSORPTION OF WATER ON POLAR COLUMNS
The mechanism of separation on polar columns in aqueouseorganic mobile phases is complicated.
HILIC is traditionally understood as the partition process between the aqueous layer accumulated close to the solid surface and a bulk mobile phase containing high concentrations (usually more than 60%) of a polar organic solvent in water (Alpert, 1990). However, the actual separation mechanism is obviously more complex, and differences in the chromatographic selectivities of various polar com- pounds indicate that adsorption on polar functionalities on the solid phase surface may also play an important role in the retention (Jandera, 2011). Both polar and hydrophobic groups on the structure of adsorbents may cause excess adsorption of either acetonitrile or water, depending on the mobile
phase composition (Noga et al., 2013). In organic solvent normal-phase adsorption chromatography, the retention is primarily based on competition between the solute and the polar solvent for the localized adsorption sites on the surface of a polar adsorbent, usually bare silica gel. Polar solvents, especially water, are strongly adsorbed on polar adsorbents from mixed aqueouseorganic mobile phases. In the original HILIC model,Alpert (1990)considered a partition-driven retention mechanism in which the analytes distribute between the stagnant water-rich layer adsorbed onto the polar sorbent on one hand and the water-poor bulk mobile phase on the other, without contribution of the sorbent backbone.
However, the solid adsorbent cannot be considered as an inert support for the adsorbed “pseudo- stationary” aqueous phase in mobile phases containing 2%e40% water in organic solvent (usually acetonitrile). Although the water uptake is often characterized in terms of “adsorbed water layers,”
there is an essential difference between the water retained in HILIC and that in classical adsorption organic solvent normal-phase chromatography. A compact water layer can be accumulated on the adsorbent surface from traces of water present in nonpolar or weakly polar organic solvents. On the other hand, water is miscible at any proportion with acetonitrile, acetone, methanol, or other polar organic solvents. Hence, an adsorbed “water layer” is rather diffuse, without sharp boundaries, and the concentration of water progressively decreases from the solid surface toward the bulk organic-rich mobile phase outside (and, possibly even partly inside) the pores of the stationary phase (Fig. 2.1).
Specific solid phase effects such as hydrogen-bonding and ionic interactions with polar functional groups such as ionized silanol groups on bare silica, or residual silanols on silica-bonded phases, may contribute to the sample retention and also affect the amount of water adsorbed on the solid support;
even nonpolar siloxane groups may adsorb some water at very low organic solvent concentrations (McCalley, 2010).
The plethora of polar stationary phases used in the contemporary practice of HILIC separations show large variability in the amount of adsorbed water in aqueouseorganic mobile phases, depending on the column type. Computer molecular dynamics simulation studies indicate that the relative proportion of the amount of water contained in the pores of silica-based phases to the water concentration in the bulk mobile phase increases at low total water concentrations in the column (Melnikov et al., 2011). The water molecules close to the silica surface are almost immobilized by the hydrogen bonds to the silanol groups.
NMR studies of bare silica and silica with bonded zwitterionic sulfobetaine groups suggested that three types of water can be distinguished inside the 6e10 nm pores: (1) free water molecules, (2) “freezable”
bound water, and (3) water bound within the polymeric stationary phase network that does not freeze at the regular water freezing temperature (Wikberg et al., 2011). This may be the reason why separations are often irreproducible or fail in mobile phases containing less than 2% water in acetonitrile. Water adsorption can often be described by the Langmuir model,Eq. (2.1), which enables measuring the adsorbed amount of water over a varying composition of the bulk mobile phase (Langmuir, 1916):
qi¼ acm 1þcma
qs
(2.1)
whereqiis the adsorbed water concentration,cmis the water concentration in the bulk mobile phase, andais the distribution constant of water in the pores of the stationary phase, forming a layer between the column and acetonitrile at very low cm. The plateaus of the isotherms characterized by qs in Eq. (2.1) give the saturation capacities for water adsorption, which can be calculated from the
experimental parameters of the Langmuir equation determined by the frontal analysis technique, where a sample solution is pumped through the column and the breakthrough volume of the sample from the column is usually monitored online in the column effluent using a nonspecific (refractive index or low-wavelength UV) detector (Vajda et al., 2013). As the boundaries of the adsorbed water diffusion layer are difficult to determine and depend on the composition of the bulk mobile phase, the adsorbed water amount may be characterized in terms of the percentage of the water-occupied pore volume.
The water breakthrough volumes,VB, at varying concentrations of water in the column feed (from 0% to column saturation) could be measured using off-line coulometric Karl Fischer titration of small collected column effluent fractions. This method does not rely on the online nonspecific detection, which may be affected by interfering compounds. The amount of the water retained in the stationary phase can be conveniently expressed in terms of the volume fraction of the excess water contained in the inner pore volume,qex, (over the bulk mobile phase water concentration,cm) and was calculated fromEq. (2.2)(Soukup and Jandera, 2013):
qex¼ðVBVMÞ$cm Vi
(2.2) VMis the total column hold-up volume, including the inner pore volume,Vi¼εiVM, and the outer (interparticle) volume, V0¼ε0VM; εi and ε0 are the inner pore and the inter-particle porosities, respectively. Uracil or thiourea are often used as column hold-up volume,VM, markers in RPLC, which are quite strongly retained on polar columns under HILIC (NP) conditions (Urban et al., 2009).
Benzene and toluene are more suitable nonretained markers of column hold-up volume in acetonitrile- rich mobile phases (Urban and Jandera, 2013). Their elution times on silica columns slightly increase as the water content in the mobile phase grows from 0% to 30%, which means that the water amount adsorbed close to the polar adsorbent surface depends on the water concentration in the bulk mobile phase (McCalley and Neue, 2008). Water adsorption isotherms on nonmodified and monomeric functionalized silica phases show monolayer formation followed by multilayer adsorption, whereas water uptake on polymeric functionalized silica stationary phases may lead to the formation of hydrogels (Vajda et al., 2013). The water uptake correlates with the retention factors of neutral ana- lytes, supporting the idea of the coexistence of adsorption and partitioning of neutral solutes in the water concentration regime normally encountered in HILIC (Dinh et al., 2013).
Frontal analysis with Karl Fischer titration was used for the determination of water uptake on 19 stationary phases, including fully porous and coreeshell silica gel and silica-bonded phases with different polarities: octadecyl and cholesterol, phenyl, nitrile, pentafluorophenylpropyl, diol, zwit- terionic sulfobetaine and phosphorylcholine ligands bonded on silica, hybrid organic-silica, and hydrosilated silica stationary phases, together with a few nonpolar alkyl-bonded phases (Soukup and Jandera, 2014). All columns adsorb water from acetonitrile, and even though there are large differ- ences between the individual stationary phases, the adsorption can be described by Langmuir iso- therms, Eq. (2.1). Six isotherm examples are shown in Fig. 2.6. Nonpolar long alkyl chain (YMC Carotenoid) and hydrosilated (Cogent Silica C, Cogent C18 Bidentate) columns show steep isotherms with low plateau water concentrations (low water uptake), whereas the water adsorption isotherms on polar columns (YMC Triart Diol HILIC, ZIC HILIC, and TSKgel Amide-80) have a much less steep slope and at least one order of magnitude higher water saturation concentrations, which is often not achieved even in mobile phases containing 20% water in acetonitrile (Soukup and Jandera, 2014).
Fig. 2.7compares the column saturation capacities,qs, calculated from the Langmuir isotherms.
The TSKgel amide, ZIC-HILIC and ZIC-cHILIC, and TSKgel NH2columns have high saturation capacities, corresponding tow45% (vol/vol) water in the bulk mobile phase, which is consistent with their high affinity to water reported elsewhere (Dinh et al., 2013). A relatively high saturation capacity, FIGURE 2.6
Langmuir isotherms of water adsorbed on two hydrosilated (Cogent Silica, Cogent Bidentate) and four silica gel (C-30 YMC carotenoid, YMC Triart Diol HILIC, zwitterionic sulfobetaine ZIC-HILIC, and TSkgel Amide-80) bonded stationary phases.cm, volume fraction of water in the mobile phase in equilibrium with the stationary phase;cs, volume fraction of excess water contained in the pores of the stationary phase.
Based on unpublished data from J. Soukup, P. Jandera, P. Jana´s.
FIGURE 2.7
Water adsorption on high-performance liquid chromatography columns: (A) excess water saturation capacities, qsatur, calculated fromEq. (2.1)and (B) the equivalent number of adsorbed monomolecular water layers,Nw, inside the pores at full saturation capacity of the columns.
Based on the data from Soukup, J., Jandera, P., 2014. J. Chromatogr. A 1374, 102e111.
35.4% (v/v) water, was observed for the Ascentis Express OH column, whereas the Xbridge HILIC, Atlantis HILIC, Ascentis Express ES-CN, and Ascentis Express F5 columns have lower saturation capacities, less than 9% v/v. At full column saturation, the excess adsorbed water,Vex, fills up to 45.3%
of the pore volume of normal silica-based columns but only 2.6%e5.5% of the pore volume of hydrosilated silica columns. Because of the low affinity of the hydrosilated silica material to water, saturation capacities are as low as 0.2%e0.4% water in the inner pore volume, which are achieved in mobile phases containing 3%e6% v/v water (for the hydrosilated bare silica or bonded C18 Bidentate columns), in agreement with the Pesek ANP model (Pesek and Matyska, 2009). This low water saturation capacity is similar to the long nonpolar alkyl chain (C30) columns. On the other hand, polar columns used frequently in HILIC show much less steep water isotherms and relatively high saturation capacities, which are not approached even in mobile phases containing 20% v/v water in acetonitrile (Soukup and Jandera, 2014).
Fig. 2.7(B)shows the water uptake in terms of the number of “hypothetical” monomolecular water layer equivalents,Nw, at full saturation capacity of the 19 columns tested (Soukup and Jan- dera, 2014). The number of adsorbed water layer equivalents generally agrees with the order of column sorption capacities inFig. 2.7(A)but with some exceptions from the rule. Less than one monomolecular water layer equivalent (full horizontal line) was adsorbed on the silica hydride- based stationary phases and on moderately polar coreeshell columns (Ascentis Express F5 and Ascentis Express CN) at the column saturation capacity. On strongly polar stationary phases, several water layer equivalents are captured from the mobile phase. The sample partition between the bulk mobile phase and a submonomolecular layer of adsorbed water lacks a physical meaning. Rather, competition between the adsorbed water and polar solutes based on a NP adsorption mechanism (Pesek and Matyska, 2009) is more realistic. Hence, a low number of the adsorbed monomolecular equivalents,Nw<1, may be used to distinguish between aqueous normal phase (ANP) and tradi- tional HILIC (Soukup and Jandera, 2014). The strongest affinity to water was observed on the ZIC- cHILIC stationary phase, where more than nine water layer equivalents were adsorbed at the saturation capacity. Columns with bonded hydroxyl and diol ligands show stronger water adsorption in comparison to bare silica.