ADVANCES IN HYDROPHILIC INTERACTION LIQUID
4. HYDROPHILIC INTERACTION LIQUID CHROMATOGRAPHY MODE IN TWO-DIMENSIONAL LIQUID CHROMATOGRAPHY SEPARATIONTWO-DIMENSIONAL LIQUID CHROMATOGRAPHY SEPARATION
SYSTEMS
The number of peaks that can be separated in a single HPLC run can hardly exceed 100. However, in clinical, pharmaceutical, food, natural and environmental analysis, molecular biology, and elsewhere, we encounter samples that may contain even millions of compounds, in concentrations possibly spanning over 10 orders of magnitude. The peak capacity, i.e., the maximum number of peaks that can be separated in a sample, can be significantly increased by combining two or more different separation mechanisms in multidimensional separation systems (Stoll et al., 2007; Guiochon et al., 2008; Fair- child et al., 2009; Grosskreutz et al., 2012; Jandera, 2012b).
To quantify the differences in selectivity of the individual HILIC columns, the correlation coef- ficient,r2, between the retention factors,k, of a suitably selected group of test compounds, or of their logarithms, on various polar columns and a reference column (such as bare silica) may be measured to calculate the coefficientss2, characterizing the relative separation selectivities of the individual polar columns (Neue et al., 2006):
s2¼1r2 (2.10)
s2¼1 for fully noncorrelated (orthogonal) systems (r2¼0), whereass2¼0 for equivalent separation systems. The 2D peak capacity can theoretically reach the product of the peak capacities of the individual systems in orthogonal 2D separation systems, where the retention in the first dimension is not correlated with the retention in the second dimension. In practice, most 2D systems show a lower number of resolved peaks because of more or less significant retention correlation in the two dimensions (Gilar et al., 2005). Coupled HILIC and RP separation systems offer two completely different retention mechanisms and a very high degree of orthogonality in comparison to other 2D LC systems (Jandera, 2008). Hence, 2D separation systems combining HILIC and RP modes allow the number of compounds resolved in complex samples to be significantly increased (Guiochon et al., 2008).
The expected increase in the number of really separated compounds (the practical peak capacity) that can be achieved in various off-line and online multidimensional setups and the price to be paid for it in terms of the analysis time and sample dilution (i.e., decreased sensitivity) have been reviewed in detail (Guiochon et al., 2008). Later, the performance of off-line, online stop-flow, and comprehensive HILIC chromatography combined with RPLC was compared in terms of peak capacities, analysis times, and peak production rates (Kalili and de Villiers, 2013). An online comprehensive LCLC system is best for samples requiring peak capacities up to 600. The off-line or stop-flow systems provide higher peak capacities, at the cost of long separation times. The contribution of the stop-flow to band broadening was found negligible.
Generally, the peak capacity is higher in gradient mode than that under isocratic conditions. Hence, the number of resolved peaks increases when simultaneous gradients are used in the first dimension and in the second dimension of a comprehensive online 2D setup (Jandera, 2012a). Optimization of the gradient range and gradient profile, especially in the second dimension, can significantly increase the practical 2D peak capacity (Cacciola et al., 2007; Jandera et al., 2011).
The off-line approach using two (or more) separate columns is very simple, does not necessitate any special instrumentation, and has been practiced for many years. It allows independent optimization
of the two separation systems; the fractions collected from the first column can be pretreated before injection onto the second column, such as using evaporation to dryness of manually collected HILIC fractions before introduction to a C18column in the second dimension (Liang et al., 2012). Off-line coupling of HILIC and RP-LC was shown to provide a powerful separation system for procyani- dins (Kalili and de Villiers, 2010) and flavonoids (Beelders et al., 2012). Another recent example reports an off-line RP-HILIC method coupled with MS for impurity profiling of infusion solutions (Schiesel et al., 2012). Unfortunately, off-line procedures are labor demanding and time demanding.
Online combinations of HILIC and RP-HPLC (Fig. 2.9) are subject to compatibility problems originating from the differences in the mobile phase elution strengths in the HILIC and RP modes.
High concentrations of the organic solvents used for HILIC separations usually provide weak retention in the RP systems, whereas the mobile phases rich in water used in RP-HPLC are usually too strong as HILIC eluents. If the mobile phase from the first (HILIC) dimension is used for the fraction transfer to the second dimension, a significant decrease in retention and unsymmetric or even split peaks may appear, with detrimental effects on the separation (Jandera et al., 2012).
The compatibility of the mobile phases used in 2D HILIC-RP systems can be improved in several ways (Kalili and de Villiers, 2013):
1. Transferring only small volume fractions online from the first dimension to the second dimension column (for example, 2mL onto a 503 mm I.D. C18 column) often minimizes the sample solvent effects in the second dimension, however, sometimes with possible negative impacts on the sensitivity of 2D separations (Guiochon et al., 2008).
2. The mobile phase strength in the acetonitrile-rich fractions transferred from the HILIC column can be modulated by diluting with water before introducing the fractions onto an RP column in the second dimension using a make-up flow mediated by an additional auxiliary pump (Wang et al., 2008c).
3. Fractions from the first column can be trapped on a small column and another solvent can be used for the transfer to the second dimension (Wilson et al., 2007).
Approaches 1 and 2 can be combined. For example, 2D HILICRP separation of phenolic compounds in green tea was reported using a diol HILIC column in the first dimension and a gradient of water and methanol in acetonitrile with acetic acid additive. 50mL fractions were collected with 1 min frequency and evaporated under nitrogen to 2mL before introduction onto a C18 column in the second dimension (Kalili and de Villiers, 2013).
Online 2D separations can employ either serial or parallel column setups. Serial column coupling does not require complex instrumentation, and allows only a moderate increase in peak capacity, because of the additivity of the contributions of the individual columns, in contrast to comprehensive online 2D LCLC providing multiplicative effects on peak capacity and considerably higher peak production rate (i.e., the number of resolved compounds in a preset separation time). A set of serially coupled columns containing different stationary phases behaves like a new column with modified selectivity, which may enable separations of samples with widely differing properties (Alvarez Segura et al., 2016). An octadecylsilica column in the first dimension serially coupled with a HILIC column in the second dimension, to which a gradient of acetonitrile in water up to the end concentration of 80%
or more was applied, allowed separation of a broad range of pharmaceuticals in a single run (Louw et al., 2008). A serial combination of a ZIC-HILIC sulfobetaine column and an amide column with a HILIC gradient run from 95% to 35% aqueous buffer in acetonitrile allowed simultaneous separations
of polar and nonpolar metabolites in a mouse serum sample (Chalcraft and McCarry, 2013). Tandem coupling of a C18 column with a zwitterionic column was employed for separation of polar and nonpolar phenolic compounds in wine, with a single gradient of simultaneously increasing concen- tration of acetonitrile and decreasing salt concentration (Greco et al., 2013).
A 2D capillary liquid chromatographyeFourier transform mass spectrometry method, employing a 100 mm150mm I.D. C18 RP column serially coupled with a 250 mm200mm I.D. (poly- hydroxyethyl aspartamide) HILIC column, was used for the analysis of 11 quaternary ammonium compounds in brain extracts, including acylcarnitines of low polarity. To overcome the mobile phase compatibility problem on the two columns, they were connected via a T-piece allowing addition of a make-up flow of 95% acetonitrile to the effluent from the RP column before transfer to the HILIC column (Falasca et al., 2012).
An interesting solution to the problem of different elution strengths required to elute polar and nonpolar compounds employed a column-switching setup, where the sample was first injected onto two serially coupled 100 mm2.1 mm BEH columns. A plug of early eluting weakly polar com- pounds was directed onto a trapping RP column, where the analytes were stored until the separation of polar compounds on the HILIC columns had been finished. Then the configuration of switching valves was changed to redirect the weakly polar compounds onto a Phenyl Hexyl RP column, where they were separated using a gradient of increasing acetonitrile concentration in 0.02% aqueous formic acid (Cabooter et al., 2014).
A crucial point affecting the separation time in comprehensive 2D liquid chromatography is the performance of the column used in the second dimension, which should allow highly efficient fast chromatographic separations. For this purpose, UHPLC with a short column packed with sub-2mm particles can be used at a very high operating pressure (Cacciola et al., 2011). A coreeshell column represents another possibility that can be used with conventional liquid chromatographic instrumen- tation (Jandera et al., 2015). Online connection of a capillary HILIC column in the first dimension and an ultra-high performance RP-LC in the second dimension, coupled with MS, was used for high- resolution separation and detailed characterization of anthocyanins and related pigments in berries and aged red wine (Willemse et al., 2014, 2015). Because of lower flow resistance, monolithic columns can be used in the second dimension at higher flow rates in comparison to particle-packed columns operated at the same operating pressure. Very fast second-dimension gradient separations on coree shell columns were achieved at ambient temperature without excessive backpressure and without compromising optimal first dimension sampling rates. Very good band symmetry and retention time repeatability in gradient separations of phenolic compounds and flavonoids could be achieved in optimized comprehensive HILICRP on a 0.5 mm I.D. monolithic sulfobetaine HILIC capillary column coupled with various 2.5e5 cm long, 3 mm I.D. monolithic and coreeshell C18columns. A flow rate of a few microliters per min was used on the capillary column in the first dimension; the short C18 columns were operated at flow rates of 3e5 mL/min in the second dimension (Jandera et al., 2012).
Polar columns showing a dual RP-HILIC mechanism allow a combined 2D RPRP and HILICRP setup to be used. In the first dimension, the RP mode in a highly aqueous mobile phase alternates with the HILIC mode in a mobile phase with a high acetonitrile concentration. In the second dimension, the RP mode can be used. A recently introduced zwitterionic polymethacrylate sulfobetaine BIGDMA-MEDSA capillary column shows a dual HILIC/RP mechanism at high con- centrations of acetonitrile and RP behavior in water-rich mobile phases (Stankova´ and Jandera, 2016).
The BIGDMA-MEDSA column in the first dimension coupled online with a short monolithic or coreeshell C18column was used for combined alternating HILICRP and RPRP comprehensive 2D separations of polyphenolic compounds. During the HILICRP period, a gradient of decreasing acetonitrile gradient was used for the separation in the first dimension. At the end of the gradient, the polymeric monolithic microcolumn was equilibrated with a highly aqueous mobile phase and was ready for the second sample injection in the RPRP period. This time a gradient of increasing concentration of acetonitrile was used in the first dimension.Fig. 2.10presents 2D chromatograms of flavones and related polyphenolic compounds, acquired with a single first-dimension BIGDMA- MEDSA capillary column in two experiments with consecutive injections of the sample, the first one into a decreasing and the second into an increasing acetonitrile gradient (Ha´jek et al., 2016). The automated dual LCLC approach allows obtaining three-dimensional data in a relatively short time.