Advances in the understanding of stationary phases for HILIC

A recent survey that was conducted by LCGC in late 2011 showed that since 2008, hydrophilic interaction liquid chromatography (HILIC) has more than doubled its application. HILIC is finding more and more applications for polar compounds that are either unretained or weakly retained on reversed-phase chromatographic columns. The technique has also been used for the separation of biomolecules. This article reviews the newer stationary phases intended for HILIC and studies performed to categorize columns for optimized separations based on their selectivities.

Hydrophilic interaction liquid chromatography (HILIC) has become an established chromatographic technique in widespread use in most application areas. This great interest has led to an accompanying increase in the number of new stationary phases explored, some of which have become commercially available. For the user, this multiplication of phases is beneficial, but it also presents some challenges. Where to start? What column to choose? How can one get optimal selectivity for a specific application? Therefore, approaches to classify different HILIC stationary phases into categories have emerged. This article will summarize the progress in this direction since a previous article was published on this topic in the 2008 North American supplement (1).

Majors recently reported in an LCGC reader survey (2) that the use of HILIC mode had doubled since the year 2007. This means that roughly 25% of chromatographers are using HILIC today. I had the pleasure to be responsible for the market introduction of the column brand ZIC-HILIC in January 2002, which was immediately well received by researchers at AstraZeneca. At that time, there were about 70 papers on HILIC in the scientific literature published since the first paper by Alpert in 1990 (3). Back then few chromatographers had a complete picture of the retention mechanism of HILIC or which column to use, but there was a multitude of unsolved application problems that were delaying research in several branches of science. Most important, the orthogonal selectivity of HILIC to reversed-phase chromatography was driving the users interest.

Later on, in April 2008 when the previous review in this series was written by McCalley (1), he had to consider in total 350 HILIC papers. Today, this number has risen to almost 1300 papers, with more than 300 new contributions only last year (see Figure 1). Now, scientific literature has become a valuable source of separation applications using HILIC. The question of whether separation is possible can often be answered based on the published work already done. More focus can be given to how it is done and to other aspects of analysis such as sample preparation, throughput and detection. Readers are referred to special journal issues on HILIC (4–6), to a textbook with various chapters addressing successful applications (7), and to reviews on different application areas such as pharmaceutical analysis (8), proteomics (9), metabonomic and metabolomic analysis (10), glycomics (11) and bioanalysis (12).

Figure 1: The number of HILIC papers per year (43).

It is clear that the improved understanding of HILIC and growing interest for selectivity, in general, also has led to better sample handling by complementary pretreatment protocols and, in many methods, to an overall faster throughput.

What Defines a HILIC Column?

Even by the Hemström and Irgum review in 2006 (13) it became clear that a wide range of polar stationary phases had been explored and used in HILIC mode applications. However, recently new stationary phases have been introduced specifically for HILIC separations. The rapidly increasing interest and demand for HILIC applications pushed column manufacturers to provide HILIC branded products. It was my experience that among users this created the impression that so-called "HILIC columns" were interchangeable just as reversed-phase columns (a somewhat faulty impression) are supposed to be. This belief persisted despite the often very obvious difference in chemical structure of the functional group on the stationary phase of different HILIC columns.

A few years ago, during the so-called "Chinese tainted milk scandal" and the urgent need for reliable assays of melamine and cyanuric acid (14), my personal experience was rather shattering. It was generally accepted that a HILIC column was required for the analysis, but it turned out that just a few were suitable and those few required quite different experimental conditions. Thus, column providers contributed to the confusion experienced in food laboratories, of whom from just one day to the next had to be introduced to a totally new separation mode — HILIC.

For this reason, I believe that HILIC columns require further classification. From a user perspective, during method development, it makes sense to try several different categories of columns and to avoid repeating experiments on similar types. In an early attempt to classify HILIC columns, a range of probes targeting various interactions were selected and run on different polar stationary phases. Principal component analysis (PCA) then clearly revealed that the columns could be divided into different classes and that selection of columns from the different categories actually made sense in method development (15). Another aspect linked to the column classification is the possibility to predict the analyte retention or at least to get a rough idea if HILIC is a suitable separation mode and to find starting conditions. Based on retention data for 40 different probes evaluated by multivariate analysis, we developed a small "predictor program" (16) that made it possible to get an initial idea on the suitability of the ZIC-HILIC column for a certain analyte. At least it confirmed that it was possible to predict retention based on the chemical nature of a solute.

Classification of HILIC Columns

Using the criteria that any polar column used under HILIC conditions can be denoted as a "HILIC column," a recent literature search revealed that a wide range of columns were still being used (17,18). In his HILIC tutorial at HPLC 2011 in Budapest, McCalley (19) suggested five main classes of HILIC columns as follows:

1. Silica gel (Si-OH or charged form Si-O^- H⁺)

2. Neutral bonded ligands (for example, amide and diol)

3. Charged ligands (for example, positively charge amino phase; negative charge poly-2-sulphoethylaspartamide)

4. Zwitterionic phases (for example, sulphobetaine and phosphorylcholine)

5. Mixed reversed phase/HILIC phase (for example, alkyl chain diol).

In application work, three of them were identified as the most used: bare silica, neutral bonded ligand (amide), and zwitterionic bonded ligand (sulphobetaine), according to the literature search (17,18). Notably, the three most popular phases also represented three different categories of stationary-phase functional group chemistries. The popularity of these was also reported in a recent review on food analysis (20). Now, I will explore these three phase types in greater detail.

Silica Gel as a HILIC Phase

Silica gel has been used as a normal-phase HPLC packing and in reversed-phase chromatography as the base for bonded phases for a long time. Bare (nonbonded, plain) silica possesses silanol groups that can ionize at mid-values of pH give rise to a negatively charged surface and exhibit cation-exchanger behaviour. Apparently, the surface silanols on silica are sufficient to create a surface water layer and the material can provide "HILIC retention." It is well known that the silica sorbent quality has a significant impact on its suitability as a HILIC stationary phase (21), so indeed nonbonded phases like bare silica also can be divided into subcategories.

The often high surface area of bare silica greatly contributes to its ability to establish HILIC retention. McCalley (22) showed that bare silica columns better preserved high efficiency at high flow rates compared to bonded sulphobetaine and amide phases, which were elsewhere found to be even more suitable for high speed analysis than reversed-phase columns (23). It should be noted that the performance also differs between various types of bare silica columns. It is well known that the metal content of the silica influences its acidity, but more recently the selectivity differences between unbonded ethylene bridged hybrid silica (BEH) and the Atlantis HILIC silica were demonstrated. Moreover, bonding amide and diol functionalities to the BEH silica also created distinct selectivity differences (24). Actually, a method with these bonded BEH HILIC columns just recently, in time for the doping controls at the Olympic Games in London, proved to be superior to reversed-phase chromatography (BEH-based columns) for the analysis of ephedrines (25).

Among the bare silica materials recently introduced, the fused-core superficially porous shell particles have generated the most interest, marketed under the brand names Halo HILIC (AMT), Poroshell (Agilent Technologies), Ascentis Express Fused Core (Supelco) and Kinetex HILIC (Phenomenex). They seem to have a somewhat limited capacity, but bonded phases such as Nucleoshell HILIC (Macherey-Nagel) and Halo Penta HILIC (AMT) have also become available. The Halo Penta HILIC is a neutral hydroxy phase, which according to the report by the manufacturer show less silanolic interactions. However, at least the Nucleoshell particle has a significantly lower carbon load (1.3%) than the corresponding Nucleodur HILIC particle (7%) and it seems that the bonding density has been compromised on the core–shell particles. Thus, it is quite likely that these bonded phases also show some bare silica–related interactions.

The hypothetical benefit of a nonbonded bare silica column is that in absence of a functional group the risk for bleeding and reduced mass spectrometry (MS) detection sensitivity is eliminated. However, to my knowledge, there is no paper confirming this assumption. According to my experience, stationary phase bleeding generally is a smaller problem with HILIC columns, at least compared to mixed-mode columns used in reversed-phase chromatography. The lack of stationary-phase bleed may result from the relatively mild mobile-phase conditions typically used in HILIC separations — that is, dilute buffers and a low ratio of water, often at neutral pH. Some HILIC amino columns have been reported to bleed, but no paper was found where bleeding from different columns was compared. Sometimes a suspected bleeding "problem" can be explained by the release of buffer salts, matrix components or other solutes that in an earlier stage have been adsorbed or precipitated on the stationary phase (by the user). A proper washing of the column often solves this type of "bleeding problem."

Neutral Bonded HILIC Phases

A bonded phase implies that there is an opportunity for specific solute functional group interactions, as well as solute spacer interactions while achieving a desirable degree of solute to sorbent interactions. Moreover, the polar bonded phases used in HILIC may be divided into two groups depending on the surface grafting chemistry rendering: a monomeric or a polymeric surface layer. On the other hand a nonbonded phase, such as bare silica, actually relies on the functionality and interactions that the bonded category might be missing (or at least express less). So, this classification is indicative of the capacity and chemical interactions that can be expected to contribute to the selectivity.

The properties of the polar bonded stationary phases are strongly related to the hydrophilicity of the functional group and consequently its ability to retain and establish a water layer on the stationary phase, which is assumed to promote a partitioning retention mechanism.

A popular type of neutral bonded HILIC column has an amide functionality bonded onto silica gel as a support. In an interesting experiment, the effects of bonding a neutral hydrophilic amide to silica under comparable conditions was explored. The Daisogel silica (5 μm, 120 Å, 300 m²/g) was modified to a different extent (0–3.67 μmol/m²) with a urea-type ligand, thereby enabling a direct comparison of bare silica and bonded columns (26), see Figure 2. To link the results to commercially available amide-type columns, the TSKgel Amide-80 (Tosoh Bioscience), Unisol Amide (Agela Technologies) and XBridge Amide (Waters) columns were evaluated in parallel. The bonded experimental urea phase with the highest loading performed the most similar to the Unisol Amide. It was seen that all amide phases showed significant differences in performance. It was noted that the commercial amide phases studied varied considerably in specific surface area (185–450 m²/g), and the authors pointed out the importance of column selection during HILIC method development. The cationic behaviour of free silanols on bare silica leads to strong retention of bases and weak retention for acids, in particular at low mobile phase buffer concentrations. As expected, because of the influence on silanolic interactions, the retention of bases decreased and acids increased when the bonding ligand density increased. Also, the retention of neutral solutes increased and the bonded phase was more selective for this group, and the difference in solute charge state gave higher selectivity on bare silica because of a higher contribution by ionic interactions on that phase. For neutral solutes, it was seen that retention was not only promoted by increasing eluent ion strength, but also by higher ligand density.

Figure 2: Stationary phase structural formula of an amide phase used for comparison of chromatographic performance of bare silica and the corresponding bonded phase at different surface coverage (0–3.67 μmol/m2). Adapted from reference 26.

The retention and selectivity of various stationary phases commonly used was recently reviewed by Guo and Gaiki (27). They chose the charge of the polar functional group as a criterion to divide columns into three categories: charged, neutral or zwitterionic phases. This distinction between phases might not be as clear as it sounds because silica is most often the base material that is functionalized. As mentioned earlier, at mobile-phase pH values > 4–5 free silanolic groups that are still present on many silica bonded phase materials may give even neutral materials a net negative charge and cation-exchanger behaviour. Apparently this is less evident if the functional groups are attached by a process creating a shielding stationary-phase layer (26). However, spacers or polymeric grafting may, on the other hand, introduce more hydrophobic properties to the phase that may affect the retentivity and selectivity of the column. McCalley (22) demonstrated that for strongly basic compounds all column categories showed ion-exchange contribution to the retention even at a 10 mM buffer concentration. It also was seen that for the more hydrophobic bases a larger extent of retention was attributed to ion exchange and less to other interaction processes. The neutral bonded and cross-linked diol column (Luna HILIC, Phenomenex) was an exception as it showed less electrostatic interactions, and presumably free silanols were more shielded. However, it was also the most hydrophobic bonded stationary phase providing the least retention (Figure 3). Because different column brands are likely to be based on different silica base materials these, therefore, will have different acidity and would provide different proportions of free silanolic groups available for ion-exchange interactions. Hypothetically, at its extreme, a less acidic bare silica could contain a similar number of ionized silanols as a bonded phase exposing free silanols from a more acidic silica. Actually, McCalley speculated that column manufacturers deliberately chose to use more acidic silica base material to also obtain this potential electrostatic interaction selectivity with the column. Still, metal impurities would need to be kept low to avoid tailing peaks (21).

Figure 3: Average retention of neutral solutes on different HILIC columns. Mobile phase: 80:20 (v/v) acetonitrile–25 mM aqueous ammonium acetate (pH 6.8); flow rate: 0.5 mL/min; temperature: 22 °C. Toluene was used as unretained marker to determine t0. On the LiChrospher CN, 5-μm, 100-Å column several analytes had less retention than toluene (data not shown). Data adapted from reference 33.

Zwitterionic Bonded HILIC Phases

Zwitterionic bonded phases are a relatively new type of stationary phases in chromatography and have been popular since the early days of HILIC. Ideally, the zwitterionic functionality should have a zero net charge and therefore be neutral. Because of the short but distinct charge separation within the functional group the phase expresses weak electrostatic interactions that contribute to the selectivity, but without being the dominating retention mechanism. The most popular sulphobetaine functionality bear a positively charged quaternary ammonium and a negatively charged sulphonic acid group and because both of these are strong ion-exchange groups, the stationary phase is "charged but neutral" regardless of the mobile-phase pH.

The popularity of zwitterionic columns is also reflected among the new bonded phases. In a recent paper on the development of a weak zwitterionic stationary phase, a trimethylchlorosilane "end-capped" variant was investigated (28). The 3-P,P-diphenylphosphoniumpropylsulphonate functional group is presented in Figure 4. The most significant effect of "end-capping," using relatively low overall buffer concentration at pH 4.1, seemed to be less retention in general for various categories of analytes (strong bases not tested). This indicates that free silanols may play a role on the sterically hindered nonendcapped phase and contribute to the overall hydrophilicity and retentivity of that material. Thus, functional groups and functionalization chemistries introducing hydrophobic elements at the silica surface tend to give less retentive stationary phases. To some extent, this is because it may influence the electrostatic interactions, but probably more so because it counteracts a "water layer" on the stationary phase and retention by a hydrophilic partitioning "HILIC mechanism.

Figure 4: A zwitterionic 3-P,P-diphenylphosphoniumpropylsulfonate stationary phase endcapped by trimethylchlorosilane functionalization. Adapted from reference 28.

Another weak zwitterionic phase was developed by the functionalization of a silica monolith with a lysine moiety (29), and it proved to be more suitable for HILIC than a less hydrophilic diol type monolith. A range of zwitterionic stationary phases have also been studied in the development of polymeric monolithic columns (30), and recently a zwitterionic sulphobetaine silica monolith capillary column was introduced by Merck (CapRod ZIC-HILIC) (31).

While selecting a column stationary phase based on "charge state" classification it is necessary to keep in mind that it depends on the mobile-phase pH. Functional groups like silica or amino columns that show weak ion exchanger behaviour are defined as charged (27). Also, depending on the surface coverage and type of silica base material, neutral phases on silica may show weak ion exchanger properties because of acidic silanols. Even zwitterionic functional groups may be affected by pH if they are "weak" rather than "strong" zwitterions.

Classification and Phase Selection in HILIC

To select the optimal stationary phase for a separation problem, one has to know the retention, selectivity and efficiency of HILIC columns. Recently, two comprehensive HILIC column characterization studies were reported in which 14 columns and 13 test compounds (32), and 22 columns and 21 test compounds (33) were used to categorize columns and study the HILIC retention mechanisms. The aim was also to find probes that may be in common use for HILIC column testing. Both studies were designed to avoid probing pH effects either by the choice of test solutes or by using constant chromatographic conditions with respect to pH. Changing pH in the mobile phase is very important for controlling selectivity, but it is not a desired effect to probe in column comparisons as the ionization state may change for both solutes and for some stationary phases. The test compounds in these studies were selected to probe all kinds of interactions that are relevant for retention and selectivity and the results were evaluated by principal component analysis (PCA).

Clearly, classification of columns into different categories is possible, as this was also found in early studies (15). Kawachi and colleagues (26) found three clusters in the PCA analysis and summarized the column properties in radar plots. As previously seen, even columns with (a supposed) equal functionality show different chromatographic behaviour. They therefore suggest that HILIC columns can be divided into two classes:

1. Strongly retentive showing high selectivity (amides, sulphobetaine, sulphonates)

2. Weakly retentive with limited selectivity (silica, diol, amino, hydroxy, triazol).

Although peak shape was not evaluated in either of these studies, both papers underline the importance to avoid or control strong electrostatic interactions leading to a dominating ion-exchange mechanism, which otherwise ruin both peak shape and column efficiency. It was also concluded that increased ion strength of the mobile phase not only reduces ion-exchange contribution, but also reduces the selectivity differences between columns. Moreover, the differences between columns and the individual column selectivity are promoted by a high content of acetonitrile in the eluent. Thus, the conditions providing the most between-column and in-column selectivity are also generally ideal for the most used (17) MS detection sensitivity.

The findings of these two studies were also coherent in comments on details in selectivity differences found and they manifest that there is no significant correlation between retention and selectivity. The average retention for 22 commercially available columns is presented in Figure 3. When these columns were investigated by injection of the test solutes at identical conditions the PCA evaluation revealed clustering into four major groups (see Figure 5) providing different main selectivity characteristics;

I: Bare silica phases — cation exchange and adsorption

II: Bonded neutral phases — low specificity

III: Bonded amino phases — anion exchange

IV: Bonded zwitterionic, sulphoethyl amino phases — partitioning and weak ion exchange

Figure 5: Classification of column types based on retention for solute probes selected for representing various stationary phase interactions. Data adapted from reference 33.

The zwitterionic HILIC columns in group IV stand out also in a multivariate classification of columns for peptide analysis (34). It was suggested that the behaviour may be caused by their polymeric stationary-phase layer (33).

Today's data confirms that the HILIC mechanism may contain several types of interactions and much depends on the category of the column used. Chirita and colleagues (35) proposed a generic guideline for HILIC method development and optimization based on a multivariate study of neurotransmitters (see Figure 6). Although not claiming perfection, they suggest that different HILIC columns should be chosen depending on the nature of the analyte (neutral, zwitterionic, cationic or anionic) and that only a few columns are suitable for all type of analytes. Thereafter, the separation is fine-tuned by the optimization of organic solvent content followed by buffer concentration, type and pH. In fact, nowadays most application papers report on the effect of these experimental parameters and it is clear that the optimization of buffer pH value is generally the most important for selectivity.

Figure 6: Decision tree for HILIC separation optimization. Adapted from reference 35.

Trends in HILIC Column Development

On average, a new HILIC paper is published almost every day. Jandera (36) recently commented in a comprehensive review that there is a new HILIC stationary phase presented every month. Obviously, very few of these will become commercially available. It merely illustrates the significance of HILIC and the renewed interest for chromatographic selectivity.

Besides the exploration of new novel functionalities for bonded stationary phases, more focus can be expected on the grafting chemistries used and the support materials. In particular, development to control or to avoid silanolic interactions because these show a weak ion-exchanger behaviour dependent on mobile-phase pH overlaid the selectivity from the functional group. Obviously, a polymer particle support is completely free from silanolic interactions, however, only one such column is commercially available (ZIC-pHILIC, Merck SeQuant). Another incentive for using a polymeric support may be complete elimination of silica related interferences while using the charged aerosol detector, exemplified by the simultaneous detection of positive and negatively charged counterions to pharmaceutical salts (37). A more common reason for using a special support material is the use of high-pH eluents, where conventional silica stationary phases are not stable. For example, high pH was reported to increase MS sensitivity for analytes ionized with negative electrospray ionization mass spectrometry (ESI-MS) using the pH-stable BEH-silica (38). Considering these benefits, it is likely that additional pH stable stationary phases will be developed.

So far, new polymeric support materials have frequently been studied in the development of monolithic HILIC columns, details on this are given in a recent review (39). Various synthetic protocols have been explored, such as direct polymerization of monoliths including the hydrophilic functionality, typically zwitterionic or amide, or introduction of the functionality in a stepwise procedure. This field is still developing, but the benefits and problems seem to be similar to those seen for polymeric monoliths in general. Currently, bonded silica monoliths look more promising. In particular, highly efficient bonded polyacrylamide silica monoliths were developed by Ikegami and colleagues (40) for the analysis of underivatized carbohydrates; they demonstrated faster analysis with the monolith compared to the particulate TSKgel Amide-80 column (3 μm).

As more phases are explored, a better understanding of the underlying mechanism behind "HILIC retention" emerges. Most progress and new commercially available phases can be expected to be among zwitterionic and neutral phases, but the latter should be more hydrophilic than conventional diol-type. For example, a cyclofructan 6 phase showed promising results for sugar separations (41) and a tris(hydroxymethylmethyl amine) phase provided some unique selectivity (42). In fact, quite recently several amide type (XBridge Amide, Unisol Amide, Inertsil Amide) and zwitterionic type (PC HILIC, Nucleodur HILIC and ZIC-cHILIC) columns have become commercially available, presumably based on the success of the forerunners TSKgel Amide-80 and Sequant ZIC-HILIC.

Conclusions

Significant steps have been achieved toward the understanding of interactions contributing to retention and selectivity in HILIC columns. Clearly, because of the multimodal nature of the retention mechanism and the wide range of different stationary phases the selection of HILIC columns is worth some careful consideration before method development.

There are great opportunities for optimization of selectivity and several different categories of HILIC columns to test during this work. However, a cost efficient approach may be to first investigate the selectivity by testing different eluent pH values, buffer salts and ion strengths using a stationary phase that is likely to show satisfactory results.

Considering the differences between columns and even between those that are supposed to be equal since they are claimed to have the same functionality, it seems difficult to meet the need for a second source supplier, which is a common requirement in method validation.

Still, because of the increased understanding and wide applicability, the use of HILIC will continue to grow considerably. In the long run, it will probably become just as popular as reversed-phase chromatography.

Einar Pontén, PhD, is an independent consultant who previously was Managing Director of Merck SeQuant AB and closely involved in chromatography R&D and the product development for HILIC. He has a background as research engineer at the pharmaceutical company AstraZeneca and associate professor at Umeå University. Please direct correspondence to: einarponten@qast.se.

References

(1) D.V. McCalley, LCGC N. Amer. 26(S4), 53–58 (2008).

(2) R.E. Majors, LCGC N. Amer. 30(1), 20–34 (2012).

(3) A.J. Alpert, J. Chromatogr. 499, 177–196 (1990).

(4) M. Lämmerhofer, J. Sep. Sci. 31, 1419 (2008).

(5) M. Lämmerhofer, J. Sep. Sci. 33, 679–680 (2010).

(6) A. Alpert, J. Chromatogr. A. 1218, 5879 (2011).

(7) Hydrophilic Interaction Liquid Chromatography (HILIC) and Advanced Applications, Weixuan He, Ed. (CRC Press, 2011).

(8) B. Dejaegher and Y.V. Heyden, J. Sep. Sci. 33, 698–715 (2010).

(9) P.J. Boersema, S. Mohammed and A.J.R. Heck, Anal. Bioanal. Chem. 391, 151–159 (2008).

(10) K. Spagou, H. Tsoukali, N. Raikos, H. Gika, I.D. Wilson and G. Theodoridis, J. Sep. Sci. 33, 716–727 (2010).

(11) G. Zauner, A.M. Deelder and M. Wuhrer, Electrophor. 32, 3456–3466 (2011).

(12) W. Jian, R.W. Edom, Y. Xu and N. Weng, J. Sep. Sci. 33, 681–697 (2010).

(13) P. Hemström and K. Irgum, J. Sep. Sci. 29, 1784–1821 (2006).

(14) D.N. Heller and C.B. Nochetto, Rapid. Commun. Mass Spectrom. 22, 3624–3632 (2008).

(15) M. Sims, "Towards the Characterisation of HILIC Stationary Phases," Presentation at HILIC Day, Reading, UK (2007).

(16) Merck SeQuant website, http://www.sequant.com/prediction — accessed March 9, 2012.

(17) E. Pontén, P. Appelblad and T. Jonsson, The Column, April 6, 15–20 (2010).

(18) P. Hemström, T. Jonsson, P. Appelblad and W. Jiang, Chromatog. Today, 4, 4–8 (2011).

(19) D. McCalley, "Hydrophilic Interaction Chromatography: Is it a Viable Complimentary Method to Reversed-phase for the Separation of Polar or Ionisable Compounds?" Presentation T01 at HPLC 2011, June 2011.

(20) J. Bernal, A.M. Ares, J. Pól and S.K. Wiedmer, J. Chromatogr. A 1218, 7438–7452 (2011).

(21) N. Nagae, N. Fujita and T. Enami, G.I.T. Lab. J. 5, 32–33 (2005).

(22) D.V. McCalley, J. Chromatogr. A 1217, 3408–3417 (2010).

(23) P. Appelblad, T. Jonsson, W. Jiang and K. Irgum, J. Sep. Sci. 31, 1529–2536 (2008).

(24) K.J. Fountain, J. Xu, D.M. Diehl and D. Morrison, J. Sep. Sci. 33, 740–751 (2010).

(25) J. Heaton, N. Gray, D.A. Cowan, R.S. Plumb, C. Legido-Quigley and N.W. Smith, J. Chromatogr. A 1228, 329–337 (2012).

(26) W. Bicker, J.Y. Wu, H. Yeman, K. Albert and W. Lindner, J. Chromatogr. A 1218, 882–895 (2011).

(27) Y. Guo and S. Gaiki, J. Chromatogr. A 1218, 5920–5938 (2011).

(28) H. Qiu, E. Wanigasekara, Y. Zhang, T. Tran and D.W. Armstrong, J. Chromatogr. A 1218, 8075–8082 (2011).

(29) G. Huang, Q. Lian, W. Zeng and Z. Xie, Electrophor. 29, 3896–3904 (2008).

(30) Z. Jiang, N.W. Smith and Z. Liu, J. Chromatogr. A 1218, 2350–2361 (2011).

(31) J. Wohlgemuth, M. Karas, W. Jiang, R. Hendriks and S. Andrecht, J. Sep. Sci. 33, 880–890 (2010).

(32) Y. Kawachi, T. Ikegami, H. Takubo, Y. Ikegami, M. Miyamoto and N. Tanaka, J. Chromatogr. A 1218, 5903–5919 (2011).

(33) N.P. Dinh, T. Jonsson and K. Irgum, J. Chromatogr. A 1218, 5880–5891 (2011).

(34) S. van D., V. Vergote, A. Pezeshki, C. Burvenich, K. Peremans and B. De Spiegeleer, J. Sep. Sci. 33, 728–739 (2010).

(35) R-I. Chirita, C. West, A-L. Finaru and C. Elfakir, J. Chromatogr. A 1217, 3091–3104 (2010).

(36) P. Jandera, Anal. Chim. Acta. 692, 1–25 (2011).

(37) Z. Huang, M.A. Richards, Y. Zha, R. Francis, R. Lozano and J. Ruan, J. Pharm. Biomed. Anal. 50, 809–814 (2009).

(38) K.J. Fountain, J. Xu, D.M. Diehl and D. Morrison, J. Sep. Sci. 33, 740–751 (2010).

(39) Z. Jiang, N.W. Smith and Z. Liu, J. Chromatogr. A 1218, 2350–2361 (2011).

(40) T. Ikegami, K. Horie, N. Saad, K. Hosoya, O. Fiehn and N. Tanaka, Anal Bioanal. Chem. 391, 2533–2542 (2008).

(41) H. Qiu, L. Loukotková, P. Sun, E. Tesarová, Z. Bosáková and D.W. Armstrong, J. Chromatogr. A 1218, 270–279 (2011).

(42) N.T.H. Bui, J.J. Verhage and K. Irgum, J. Sep. Sci. 33, 2965–2976 (2011).

(43) Search on Web of Science using HILIC or "hydrophilic interaction" or "aqueous normal" and chromatogr* as search criterion at http://apps.webofknowledge.com, accessed February 14, 2012.