HILIC: The Pros and Cons

Hydrophilic interaction liquid chromatography (HILIC) combines a polar stationary phase (bare silica or polar bonding) and a highly organic mobile phase (>70% acetonitrile). It is a powerful approach to retain polar compounds and is also a good alternative to reversed-phase LC for the analysis of ionizable compounds of diverse polarity, which are often encountered in the pharmaceutical field. In addition, HILIC presents various advantages such as a higher sensitivity when it is used with electrospray ionization mass spectrometry (ESI-MS) and a lower back pressure than in reversed-phase LC because of the higher volatility and weaker viscosity of the highly organic mobile phase. In a recent web seminar (chromatographyonline.com/HILIC_webcast), Davy Guillarme, a senior lecturer at the School of Pharmaceutical Sciences at the University of Geneva, University of Lausanne, in Geneva, Switzerland — and the winner of the 2013 LCGC Emerging Leader in Chromatography Award — discussed the possibilities and limitations of HILIC. Here, he answers common user questions about HILIC.

Seeded questions:

Do you have an idea how many people use HILIC in industry?

Guillarme: To be honest, HILIC is not yet a widespread strategy in industry but has proved to be useful for a few specific applications such as the determination of polar substances or inorganic ions. Today, most validated methods have been developed in reversed-phase LC and it is often time consuming to change a method from reversed-phase LC to HILIC. However, when dealing with bioanalysis in clinical or toxicological laboratories, I think there is a significant interest in using HILIC because of the large gain in sensitivity that can be expected. So, I think that more and more people involved in applications such as drug metabolism and pharmacokinetics (DMPK) will use HILIC in the future.

Based on your experience, do you think that developing a HILIC method is harder than developing a reversed-phase LC method?

Guillarme: People are used to working in reversed-phase LC, and it is always more difficult to develop methods with a new strategy that has never been employed previously in a laboratory. However, if the interaction mechanism of HILIC is understood and if the sample diluent is adequately considered, some nice separations can also be achieved under HILIC conditions. The problem with HILIC is that it can be a kind of on–off chromatography for some compounds, whereas reversed-phase LC appears to be much more generic. For this reason, it can be sometimes more difficult to develop HILIC methods than reversed-phase LC methods, but the difficulty is highly dependent on the nature of the compounds.

Why is there only a limited number of chemistries for HILIC core–shell technology?

Guillarme: Core–shell technology is a recent development and the providers of core–shell columns have probably focused their attention on the largest market, namely reversed-phase LC. Today, core–shell reversed-phase LC technology seems to be mature and there are many companies that commercialize core–shell particles. All of them are now beginning to develop more exotic chemistries, including phases dedicated to HILIC operation. For example, core–shell bare silica and amide are now available from various providers.

HILIC–MS offers an impressive increase in sensitivity compared to reversed-phase LC–MS. However, are there some limitations with HILIC–MS, when dealing with biological fluids?

Guillarme: It is true that the gain in sensitivity achieved with HILIC is impressive, on average around 10-fold. This is particularly useful when dealing with biological fluids because the limits of detection and quantitation have to be as low as possible. However, in the case of quantitative bioanalysis, matrix effects are also a major concern that must be considered. For example, endogenous phospholipids are found in plasma at significant concentrations, and are typically not completely removed during extractions. Because phospholipids are much more strongly retained in HILIC than in reversed-phase LC, it is expected that matrix effects related to phospholipids will be more pronounced in HILIC than in reversed-phase LC. On the other hand, because the retention mechanism of HILIC is very different from that of reversed-phase LC, the reverse situation could also occur with other types of biological matrixes and matrix effects could possibly be reduced in HILIC.

Mobile Phases, Buffers, and pH

HILIC requires more organic solvent than aqueous solvent in the mobile phase. Can we use a gradient mobile phase from 60% to 95% (aqueous to organic)?

Guillarme: Contrary to reversed-phase LC, water is considered as the strongest solvent in HILIC. This is the reason why a HILIC gradient experiment should always be performed from high (for example 95% acetonitrile) to low proportion of organic modifier (around 60% acetonitrile). Below a proportion of 60% acetonitrile, the water layer at the surface of the stationary phase cannot be formed and hydrophilic partitioning cannot occur.

What is the proper buffer concentration for HILIC–MS?

Guillarme: It is generally advised to work with a buffer concentration of 10–20 mM on the maximum when working with MS devices. A higher proportion of salts could provide source contamination, ion suppression, and a loss of sensitivity. Then, a compromise should be made in HILIC–MS because peak shape is generally enhanced with high buffer concentration but MS sensitivity could be strongly reduced.

What about the additives to organic phase? For example, when dealing with nicotine and its derivatives, I have observed peak-shape improvement when adding 0.01% of formic acid to the organic phase, using a bare silica column.

Guillarme: Additives and buffer solutions are extremely important in HILIC because ion exchange is a strong contributor to the HILIC mechanism. Because the amount of organic modifier is much larger in HILIC than in reversed-phase LC, it is much more useful to add additives to the organic phase in HILIC. Based on our experience, the combination that provides the best peak shape in HILIC consists of using a buffer solution as aqueous media (such as 10 mM ammonium formate or ammonium acetate at a controlled pH value) and organic modifier containing a low concentration of additive (such as 0.01–0.1% formic acid or acetic acid).

You suggested using a limit of 3% of water/buffer. Sometimes I encounter 2% or 1%. How did you estimate this 3%?

Guillarme: This value of 3% is just an estimate. Indeed, this cut-off value also depends on the type of bonding at the surface of the stationary phase and its ability to retain water, to have a hydrophilic partitioning mechanism. In an interesting study published in the Journal of Chromatography A in 2008, McCalley and Neue (1) suggested that about 4–13% of the pore volume of a silica phase was occupied by a water-rich layer when using acetonitrile–water containing 95–70% (v/v) acetonitrile.

Have you tried using high pH levels (pH 8–10) with hybrid bare silica?

Guillarme: According to the provider of the hybrid bare silica that I have used, it can withstand pH levels up to 9. We have never tried to work under these conditions in our laboratory but I suppose that column lifetime would be reduced under such extreme conditions. Indeed, mobile phase with an elevated pH has the ability to dissolve the silica matrix. In the case of bare hybrid silica phase, the accessibility to silica is superior compared to that of a C18 phase and column resistance under alkaline conditions is reduced. As an alternative, I would advise HILIC users to select a hybrid amide phase, which can be used up to pH 11, according to the same provider.

When you are talking about pH, are you referring to the pH value in aqueous solution only or in a mixture of aqueous and organic solvents?

Guillarme: Measuring pH in hydro-organic mixtures is quite difficult using a conventional pH meter and electrode. In any paper related to HILIC, the reported pH values correspond to aqueous pH, which can be considered an apparent pH value. It is evident that the pH could be strongly affected by the amount of organic modifier and that it would be better to measure the real pH value under real HILIC conditions.

You mentioned the importance of pH for the charged state–polarity. How can one manage this property if the analyte of interest has more than two pK_a values? For example, tetracycline hydrochloride has three pK_a values: 3.32, 7.78, and 9.58. What are your thoughts on analysing this type of ionizable compound using HILIC?

Guillarme: Because of the presence of large amounts of organic modifier in HILIC, it is always difficult to know precisely the mobile-phase pH, the pK_a of silanols at the surface of the stationary phase, and the pK_a values of the analysed compounds. For this reason, even if the aqueous pK_a values of tetracycline are reported in the literature, I would work empirically and would try different mobile-phase pH values (for example, pH 3 and 6), to determine the retention and peak shape under these conditions.

You mentioned that at least 3% water is required for HILIC. Why isn't it possible to use 100% acetonitrile? Can you have phase collapse when going to 100% acetonitrile?

Guillarme: If 100% acetonitrile were used in HILIC, hydrophilic partitioning could not occur because of the absence of a water layer at the surface of the stationary phase. There is no risk of phase collapse in HILIC because there are no hydrophobic alkyl chains such as the C18 chains in reversed-phase LC. Under such conditions (100% acetonitrile), it would probably be possible to retain the ionizable compounds thanks to an ion-exchange mechanism but in this case, we could not call this approach HILIC.

Is dimethylformamide (DMF) or DMF–water as a sample solvent compatible with HILIC?

Guillarme: Personally, I have never worked with DMF as sample solvent but I expect that its behaviour may be close to that of dimethyl sulphoxide (DMSO) because the polarity is similar. We have recently published a systematic study (2) showing that the peak shape remains acceptable with a maximum of 20% DMSO within the sample diluent. The same rule probably could be applied for DMF. However, because DMF is polar and UV active, it will be retained under HILIC conditions and could interfere with compounds of interest during the chromatogram.

Mass Spectrometry

Can we use an atmospheric pressure chemical ionization (APCI) probe with HILIC–MS?

Guillarme:There is no restriction on the type of ionization source that can be employed with HILIC because the mobile phase is very close to that of reversed-phase LC, except for the amount of organic modifier. However, because APCI is normally devoted to lipophilic compounds and HILIC is dedicated to the analysis of hydrophilic and ionizable compounds, the number of applications for HILIC–APCI-MS is rather limited.

I've noticed during HILIC method development that sometimes even when my run-to-run retention time has stabilized, the ESI-MS sensitivity fluctuates wildly between runs. Could this be due to the retention and release of buffer on the HILIC phase?

Guillarme: We have not observed a higher variation in peak intensity in HILIC–MS compared to reversed-phase LC–MS in our laboratory, but we have not yet performed any systematic study of this topic. Of course, this observation could strongly depend on the electrospray conditions, the type of MS analyser and brand, the type of mobile phase, and the nature of the analysed compounds. However, because the type of buffer used in HILIC is very close to that employed in reversed-phase LC (such as ammonium acetate, ammonium formate, or formic acid), I don't think that the problem you observed is related to the buffer itself.

Injection and Sample Volume

What is the typical injection volume for a HILIC column? How large an injection is too big?

Guillarme: On average, a good starting value for HILIC is an injected volume equal to 1% of the column volume. This corresponds to about 1–2 µL and 15–20 µL for 50 mm × 2.1 mm and 150 mm × 4.6 mm columns, respectively. Nevertheless, it is hard to answer this question precisely as the injected volume depends on the nature of the sample diluent and the retention time of the compound. If the compound is dissolved in a significant proportion of aprotic solvent (such as acetonitrile) and it is strongly retained, the injection volume may be high (up to 2–10% of the column volume). On the other hand, if the compound is dissolved in a critical solvent (which contains a significant proportion of water in the sample diluent) and it is poorly retained, the injected volume should be limited to around 0.1–0.5% of the column volume.

Specific Separations

Can a molecule with a phosphate group and an amine group be separated using HILIC?

Guillarme: Because both phosphate and amine groups are polar, a molecule containing these two functional groups may be hydrophilic and should be suitable for a HILIC experiment. I don't know precisely the structure of the molecule you are discussing, but the risk is that the molecule may be too polar and thus strongly retained at the surface of the stationary phase. So, if the molecule is too strongly retained under HILIC conditions using a bare silica or zwitterionic phase, I would recommend evaluating amide or diol phases to reduce retention, as these phases possess significantly less charges at the surface (lower ionic interaction).

Does HILIC work well with nano-LC?

Guillarme: Unfortunately, I don't have any expertise with nano-HILIC. Similarly to what can be done with reversed-phase LC, HILIC can also theoretically be used at the nano scale. The only constraint is the limited number of stationary-phase dimensions and chemistries, which are available from a very limited number of providers.

Glyphosate is an example of an analyte that is retained poorly in reversed-phase LC but is retained too well using a zwitterionic HILIC column at any higher concentration of acetonitrile. It finally starts to be eluted at <25% acetonitrile; the less acetonitrile, the better. In many published methods for this analyte, a HILIC column is used with a highly aqueous phase. Do you have any thoughts about using highly aqueous mobile phases with a HILIC column?

Guillarme: This is a problem that can be encountered with very polar compounds such as glyphosate but also aminoglycosides, polar pteridines, and others. Because glyphosate possess an amino, a phosphate, and a carboxylic acid group, the ionic interactions between the compound and the zwitterionic stationary phase are very strong. I would recommend evaluating an alternative stationary phase such as the amide or diol phases to reduce the strength of ionic interaction.

Pat Sandra's group (3) has demonstrated the possibility of working with a bare silica phase and a purely aqueous buffered mobile phase. This mode has been called pure aqueous liquid chromatography (PALC). It cannot be considered HILIC because there is no hydrophilic partitioning between a water-enriched layer at the surface of the stationary phase and the mobile phase. However, under such conditions, MS sensitivity is drastically reduced.

I am quantifying amino acids using a C18 column with a 10 mM borate buffer at pH 7.8 and acetonitrile–methanol–water. Can I use the HILIC column with these mobile phases to improve the separation?

Guillarme: Because the interaction mechanisms in reversed-phase LC and HILIC are completely different from those of reversed-phase LC, I cannot guarantee that the separation will be improved under HILIC conditions. In theory, amino acids are very polar compounds and are thus well adapted for HILIC. In practice, we have made numerous attempts in our laboratory and we have not yet found a suitable mobile phase and stationary phase condition allowing the retention and selectivity of a majority of amino acids. In your question, you mentioned that amino acids were analysed on a C18 phase. To the best of my knowledge, I have never seen any successful separation of intact amino acids on a C18 phase without ion-pairing reagents. The amino acids you are analysing are probably derivatized and cannot be easily analysed in HILIC under this form.

Peptides and Proteins

Is HILIC better than reversed-phase LC for analysing hydrophobic peptides?

Guillarme: HILIC is a very powerful strategy for the analysis of peptides. Indeed, these compounds are sufficiently retained in HILIC, the selectivity is orthogonal to that of reversed-phase LC, and the MS sensitivity is strongly enhanced under HILIC conditions. It is also important to notice that the ion-exchange mechanism, always present with HILIC phases, plays a major role for the retention of peptides. Based on my own experience, I would indicate that 95% of peptides can be adequately analysed in HILIC thanks to the ion-exchange contribution. It is possible that for some extreme, very hydrophobic peptides such as cyclosporine A, HILIC might be unsuitable, but this has to be checked experimentally.

Do you think HILIC is applicable in dealing with large molecules, like proteins?

Guillarme: Until now, there is only one paper demonstrating the applicability of HILIC for the analysis of intact, soluble proteins (4). The achieved performance was not impressive, and there is always a risk to strongly denature the proteins in the presence of large amount of acetonitrile, which is a strategy that is commonly used to precipitate protein in sample preparation. In addition, because wide-pore HILIC phases are not yet available, the peak shape of large proteins (>20 kDa) may be unacceptable.

General Method Development

What are the major factors that affect the consistency of HILIC performance?

Guillarme: Because an ion-exchange mechanism plays a major role in HILIC mode, it is important to control precisely the mobile-phase pH and ionic strength to avoid unexpected shifts in retention times. In addition, the equilibrating time for a gradient operation should correspond to ~20 column volumes in HILIC, instead of five column volumes in reversed-phase LC.

Does temperature have a similar effect on separation in HILIC mode as it does in other types of separations, such as reversed-phase LC?

Guillarme: Mobile-phase temperature has a similar impact in HILIC and reversed-phase LC; namely, an increase in temperature causes a slight change in selectivity, lower retention, less tailing for ionizable compounds (because of faster secondary ionic interaction kinetics), less band broadening because of the reduction in viscosity and improvement of diffusion coefficients, and less back pressure in the column. However, because HILIC phases are not bonded with a long alkyl chain such as the C18 moiety in reversed-phase LC, HILIC phases are less resistant and the upper temperature limit of most HILIC columns is 45–50 °C. In addition, because back pressure is not a strong constraint in HILIC (because the mobile phase contains a high proportion of acetonitrile and thus has low viscosity), the effect of elevated mobile-phase temperature may be more limited in HILIC than in reversed-phase LC.

System Maintenance, Column Life

Given the greater use of buffers in HILIC compared to reversed-phase LC, what is your experience regarding the effect of HILIC on the amount of maintenance required for the mass spectrometer, particularly in terms of cleaning the front plate, electrode tips, and lens on the vacuum side of the MS–MS system?

Guillarme: In fact, the amount of maintenance required for the mass spectrometer is equivalent in both modes, and we have no special maintenance procedure for HILIC–MS compared to reversed-phase LC–MS. The reason is that we use similar buffers in both modes (10 mM ammonium formate, 10 mM ammonium acetate, 0.1% formic acid). However, some HILIC methods have been developed with huge amount of salts (up to 50–100 mM in some cases) and may be incompatible, or at least poorly compatible, with MS.

Is it true that HILIC columns generally do not last as long as with reversed-phase LC columns?

Guillarme: Because of the absence of long alkyl chains (C8 or C18) in HILIC columns, the silica matrix is less protected in HILIC than in reversed-phase LC. As a result, the silica matrix could be dissolved when using very aggressive conditions (elevated pH, high temperature). On the other hand, because of the absence of bonding on the bare silica, these phases are not prone to hydrolysis under acidic conditions. In our laboratory, we are able to perform 700–800 injections on a HILIC stationary phase compared to an average of 1000 injections on a reversed-phase LC phase. So column lifetime is about 20–30% lower in HILIC, but it also strongly depends on the mobile phase conditions, the type of column, and the sample matrix. This reduction in lifetime can be partially explained by the longer equilibrating time between successive HILIC analyses.

References

(1) D.V. McCalley and U.D. Neue, J. Chromatogr., A 1192, 225–229 (2008).

(2) J. Ruta et al., J. Chromatogr., A 1217, 8230 (2010).

(3) A. Dos Santos et al., J. Sep. Sci. 32, 2001–2007 (2009).

(4) T. Tetaz et al., J. Chromatogr., A 1218, 5892–5896 (2011).