Chromatographic Evidence of Silyl Ether Formation (SEF) in

Dec 16, 2014 - We have found that storage conditions of a column are paramount, which can either prevent or accelerate the process. Because SEF exists...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/ac

Chromatographic Evidence of Silyl Ether Formation (SEF) in Supercritical Fluid Chromatography Jacob N. Fairchild,* Darryl W. Brousmiche, Jason F. Hill, Michael F. Morris, Cheryl A. Boissel, and Kevin D. Wyndham Waters Corporation, Milford, Massachusetts 01757, United States ABSTRACT: In this article, we propose that silyl ether formation (SEF) is a major contribution to retention and selectivity variation over time for supercritical fluid chromatography (SFC). In the past, the variations were attributed to instrumentation, but high performance SFC systems have shed new light on the source of variation. As silyl ethers form on the particle surface, the hydrophilicity is decreased, significantly altering the retention and selectivity observed. SEF is expected to occur with any chromatographic particle containing silanols but is slowed on hybrid inorganic/organic particles. The SEF reaction is between alcohols on the particle surface and in the mobile phase solvent. We have found that storage conditions of a column are paramount, which can either prevent or accelerate the process. Because SEF exists as an equilibrium between the liquid phase and the particle surface, the process is also reversible. The silanols can be hydroxylated (regenerated) to their original state upon exposure to water. The next generation of stationary phases will either advantageously utilize SEF or effectively mitigate its effects. Mitigation of SEF would be a significant improvement in SFC that has the potential to vault their performance to levels of similar reproducibility and reliability observed for high performance liquid chromatography (HPLC). Further research in SEF may lead to a better understanding of the mechanism of interaction between the solutes and chromatographic surface.

I

achiral separations.11−14 Much work has been done to expand the separation space by using mobile phase additives with traditional stationary phases.15−18 As the technique grew, the need for SFC specific and high performance stationary phases led to the development of the first generation of SFC columns.19 These SFC specific, bonded materials were led by a few, well-received phases: 2-ethylpyridine, diethylaminopropyl, dinitrophenyl, and similar phases.20−22 More recently, in 2012, Waters introduced a series of hybrid particle based sub-2 μm phases in conjunction with the release of the ACQUITY UPC2 (Waters Corporation, Milford, MA, USA) instrument.23,24 This brought together some of the most important improvements to the technique: technique specific bondings, scalable hybrid particle technology, and high performance instrumentation.25,26 Prediction of the potential SFC instrumentation in the future is relatively easy, following the progress of HPLC as a guide. One could expect that lower dispersion, higher pressures and miniaturization are all viable improvements. Stationary phase technology for SFC, on the other hand, is much more difficult not only to predict but ideate. From an application space perspective, SFC can continue to be a niche technique used only when LC and

mprovements in instrumentation and chromatographic media and hardware have allowed for significant improvements in separation science. The evolution of separation technology from short, inefficient columns to long, high efficiency ones, has brought gas chromatography (GC) to high performance capillary GC. High performance liquid chromatography (HPLC) underwent a similar transition from HPLC to ultra high performance LC (UHPLC) with low dispersion, very high pressure instruments and sub-2 μm particle packed columns. Carbon dioxide based separations, specifically supercritical fluid chromatography (SFC), has been evolving through instrument and stationary phase improvements. SFC initially presented great promise as a chromatographic technique, but was unable to deliver on those promises, due to performance or practical limitations, including solubility.1−7 The initial SFC columns used for polar compound retention were repurposed from normal phase liquid chromatography (NPLC), after the convention of using high purity silica. These phases typically were unbonded silica, cyanopropyl ligands or other HPLC column chemistries. Quite literally, any column that could be connected to a SFC system was a candidate to be utilized.8 This practice is still used often, as practitioners find use for materials designed for HILIC use (amide and aminopropyl bondings, etc.) and other polar phases, such as diols and poly(vinyl alcohol)s.9 Chiral separations are ideally performed using SFC with specially designed phases,10 which have even been successfully used for © 2014 American Chemical Society

Received: September 23, 2014 Accepted: December 16, 2014 Published: December 16, 2014 1735

DOI: 10.1021/ac5035709 Anal. Chem. 2015, 87, 1735−1742

Article

Analytical Chemistry

Figure 1. Twenty-five consecutive injections on a 2.1 × 150 mm column packed with 5 μm Viridis Silia 2-EP. Chromatographic conditions: 0.65 mL/min, 90/10 CO2/MeOH, 40 °C, 2175 psi backpressure.

Figure 2. Retention losses for three different 2-ethylpyridine bonded phases, Viridis BEH 2-EP, Viridis Silica 2-EP, and PrincetonSFC 2ethylpyridine. Approximately 470 mL of methanol were pumped through each column over a 24 h period. Chromatographic conditions: 2.0 mL/ min, 90/10 CO2/MeOH, 40 °C, 2175 psi backpressure.



GC fail or are significantly limited or SFC can become more universal and define new application areas. The latter requires research and a better understanding of the interactions present in SFC, to produce columns that exhibit wider applicability and universality similar to the perceived role of the C18 phases in RPLC. The universality of C18 is a mis-nomer, as there are at least a few hundred variations of C18 bonded RPLC materials, yet there is still a growing need for alternate selectivity.27,28 Recently, an article was published comparing SFC columns from many different manufacturers with regards to retention variation.29 Their observations detail that the significant retention changes could be attributed to the relative acidity of the particle surfaces. The work presented here further identifies the source of retention shifts with supporting chromatographic evidence.

EXPERIMENTAL SECTION

All chromatography experiments were performed using an ACQUITY UPC2 system (Waters Corporation, Milford, MA), except for the testing depicted in Figure 2, which was performed on a TharSFC Method Station (Waters). The UPC2 system was equipped with a Binary Solvent Manager, Sample Manager equipped with a fixed loop injector and a 2 μL sample loop, Convergence Manager containing the dynamic backpressure regulator, Column Manager equipped with active eluent preheating, and photodiode array detector containing a high pressure UPC2 flow cell. The CPMAS (cross-polarization magic angle spinning) NMR (nuclear magnetic resonance) spectra were obtained for solid-state samples using a Bruker Corporation (Billerica, MA) 300 MHz NMR. Chromatographic Materials. The 4.6 × 150 mm Viridis Silica 2-EP and BEH 2-EP (5 μm) and 3.0 × 100 mm ACQUITY UPC2 BEH (1.7 μm) columns used in this study were obtained from Waters. The two BEH stationary phases 1736

DOI: 10.1021/ac5035709 Anal. Chem. 2015, 87, 1735−1742

Article

Analytical Chemistry utilize ethylene bridged hybrid inorganic/organic particles.23 The 4.6 × 150 mm PrincetonSFC 2-ethylpyridine (5 μm) column was obtained from Princeton Chromatography Inc. (Cranbury, NJ). Test analytes (flavone, thymine, papaverine, ketoprofen, prednisolone, sulfanilamide, 3-benzoylpyridine, coumarin, caffeine, prednisone, and N,N-dimethylbutylamine) were obtained in high purity from Sigma-Aldrich (St. Louis, MO). Optima and HPLC grade methanol and n-propanol, respectively, were obtained from Thermo Fisher Scientific (Waltham, MA). A solution of 2 M ammonia (NH3) (Thermo Fisher Scientific) in methanol was diluted to prepare ammonia containing cosolvents. All water used in this study was from a Milli-Q (EMD Millipore, Billerica, MA) filtration system. Food grade, gaseous carbon-dioxide was obtained from Airgas (Radnor, PA).

particle.23 The selectivity change should not be related to ligand bleed, as the retention would be expected to increase as the 2ethypyridine ligand is removed and more silanols are exposed. These results can, however, be potentially explained by a change in the particle surface chemistry e.g. reaction of the silanols resulting in a change in stationary phase selectivity. The phenomenon causing this change appears to be relatively slow and was found to be related to the mobile phase, as using different analytes did not improve or reverse the retention loss. The reaction also should be considered universal for materials using silanols to retain polar compounds. The relative reactivity of surface silanols is believed to be a contributing factor, given the observation that the BEH particle showed half the retention loss compared to two separate silica particles, from different manufacturers, as suggested by Ebinger et al.29 Silanol activity has been shown to be significantly different between silica and hybrid silica particles, with the former having pKa’s of approximately 4.5 and 7 and the latter having a pKa around 9.30,31 As retention and selectivity changes have been observed for multiple analytes, the mobile phase is a likely source of the interaction or reaction with the surface silanols. Mixtures of carbon dioxide and methanol are the most common mobile phases and important to examine. Carbon dioxide is not expected to readily react with silanols and would not interact to a high degree. However, the polar nature of methanol causes direct interaction with silanols and could result in a reaction. We propose here that retention and selectivity changes are due to, what we have termed, silyl ether formation (SEF) caused by a condensation reaction between silanols and methanol (alcohol) cosolvent.32 This reaction has been widely used to create silica polymers from tetraethoxysilane (TEOS) monomers, which form silica particles used in chromatography. The equilibrium, depicted in Figure 3, converts silanols to methyl



RESULTS AND DISCUSSION Modern chromatographic instrumentation is expected to deliver reliability and reproducibility.29 By performing calibration and diagnostic tests of instrument modules, it becomes much easier to identify failing components and changes in dayto-day performance. We first identified significant retention changes when older chromatographic results of SFC columns were compared with more recent ones and changes over the course of a single test. A continuous retention loss is shown in Figure 1 for six different analytes over the course of 25 consecutive injections (about 7 h). In this test, standard SFC conditions were used (90/10 CO2/MeOH mobile phase, 40 °C and 2175 psi ABPR pressure) resulting in drifting retention as the column was used. Two important results were noted from this experiment: (1) all the probes tested had less retention, relative to their initial retention measured at the first injection, and (2) each of the probes tested resulted in a different rate of retention loss during the test. To determine if the results were related to the mobile phase, three different 2-ethylpyridine bonded (no end-caps) packing materials were tested by flowing hundreds of column volumes of CO2/MeOH through the columns. Over 24 h, about 470 mL of methanol was pumped through each of the following 4.6 × 150 mm columns: Viridis BEH 2-EP, Viridis Silica 2-EP (Waters Corp.) and PrincetonSFC 2-ethylpyridine (Princeton Chromatography Inc., Cranbury, NJ). In Figure 2, the initial and final chromatograms are shown for each of the columns tested. Each peak of the final chromatograms is denoted with the percent retention loss, which is then averaged and represented on the right-hand side of the figure, 3.1, 6.6, and 6.0% average retention loss for Viridis BEH 2-EP, Viridis Silica 2-EP, and PrincetonSFC 2-ethylpyridine, respectively. Trends of the same test probes used during the consecutive injections test are consistent with these results. The largest percent of retention loss came from papaverine for each column tested and the smallest from flavone. The two silica based materials, Viridis Silica 2-EP and PrincetonSFC 2-ethylpyridine have very similar values for retention loss, except for ketoprofen, expressing the selectivity difference between the particles. A surprising result is that the Viridis BEH 2-EP column had approximately half the retention loss of either of the silica based materials. There is significantly less retention for the BEH based material, but that is mostly due to the lower phase ratio compared to these silica based materials. It is obvious that there is a significant selectivity change in each of these materials over the course of the test, which is less dramatic on the hybrid inorganic/organic BEH

Figure 3. Silyl ether formation (SEF) equilibrium of a methyl silyl ether on a silica based particle surface.

silyl ethers using methanol, present in the mobile phase of most SFC separations.33,34 This equilibrium is most likely present in all forms of liquid and supercritical fluid chromatography where silanols are used, however, the reaction is driven toward SEF by the lack of water present in SFC separations. It is quite common for SFC separations to be performed at or near waterfree conditions with a consequential increase in chemical complexity compared to water-rich separation modes.35 Hence, a traditionally ignored alcohol condensation equilibrium becomes magnified under these conditions. Etherification of silanols is quite comparable to bonding a hydrophobic ligand to the silica surface. The silanol-based retention mechanism, often used in SFC, requires direct interaction between the silanols and polar analytes. In SFC, the roles of the bonded ligands are to affect selectivity and manage secondary interactions. Retention loss and selectivity changes are then expected as the silica surface is modified. The proposed equilibrium also indicates that SEF should be reversible or mitigated by the addition of water in the mobile phase. To test the validity of SEF, we chose columns packed with ACQUITY UPC2 BEH particles and exposed them to 1737

DOI: 10.1021/ac5035709 Anal. Chem. 2015, 87, 1735−1742

Article

Analytical Chemistry

Figure 4. Selectivity change (3-benzoylpyridine and flavone) as a column was stored in pure methanol between data points. The peaks are nearly resolved at the day 12 data point, indicating significant levels of silyl ethers. Chromatographic conditions: 3.0 × 100 mm ACQUITY UPC2 BEH 1.7 μm column, 1.2 mL/min, 95/5 CO2/MeOH, 50 °C, 2200 psi backpressure.

Figure 5. Separation of 3-benzoylpyridine and flavone before (day 0, top) and after storing a column in 0.1% DMBA in methanol (day 2, bottom) for 2 days. Chromatographic conditions are the same as in Figure 4.

flavone are found in Figure 4. The initial data point, day 0, shows a complete coelution of the two analytes, but by day 12, the analytes are approaching a resolution of unity. Both of the compounds lose retention as SEF occurs, but a change in selectivity is found and the components gradually resolve. It appears that the SEF process is quite slow when storing a column packed with BEH particles in pure methanol, so an attempt was made to catalyze the reaction. Given that chemical equilibria can be acid and/or base catalyzed, we chose to focus on base catalysis as the base particle should be stable and a very strong acid may damage the chromatograph or column hardware. A column was stored for 2 days in methanol with 0.1% N,N-dimethylbutylamine (DMBA) in Figure 5, tested before and after storage. DMBA was chosen as it would not act as a nucleophile and potentially convolute the experiment. Injections of 3-benzoylpyridine and flavone revealed that the material stored in 0.1% DMBA in methanol showed selectivity yielding complete resolution of the two analytes, indicating SEF. The peaks are baseline resolved, meaning higher levels of silyl ethers were achieved in the same storage time frame by using a basic additive than pure methanol. Supporting the

various storage conditions. The BEH particle was chosen because (1) there is no ligand bonded to the surface, providing access only to silanols and (2) the particle has been proven to exhibit high chemical and physical stability.23 An unbonded particle simplifies the interaction between analytes and provides a single source of retention on the stationary phase. Assuming no change in the chromatographic conditions, any retention change should be related to a modification of the silica surface. Chromatographic conditions were developed using a 3.0 × 100 mm 1.7 μm BEH column that produced a coelution of two compounds, 3-benzoylpyridine and flavone. To separate the compounds, a change in selectivity is needed, hence a change in surface chemistry as the mobile phase conditions were held constant. We observed that when relatively high levels of silyl ethers were formed on the particle surface, the pair of peaks could be resolved. To track the selectivity of the two compounds, daily tests were performed on a new column, never exposed to methanol. Six replicate injections of the compounds were made each day and then the column was subsequently filled with pure methanol to be stored until the next day. Example chromatograms of 3-benzoylpyridine and 1738

DOI: 10.1021/ac5035709 Anal. Chem. 2015, 87, 1735−1742

Article

Analytical Chemistry

Considering that SEF exists as an equilibrium, the observed chromatographic effects should be reversible by introducing water to the system. To demonstrate this, the same column that had been stored for 2 days in methanol with 0.1% DMBA was rinsed in short time segments at 1 mL/min with pure water, then filled with pure isopropanol before being retested under SFC conditions. The retention times of 3-benzoylpyridine and flavone were tracked after each rinsing segment and the results are shown in Figures 8 and 9. As the silyl ethers are removed,

chromatographic evidence, C13 CPMAS NMR was used to examine packing materials removed from columns that exhibited retention loss and selectivity changes due to SEF. In Figure 6, the NMR spectra of a batch of BEH 2-EP before

Figure 6. CPMAS NMR of BEH 2-EP particles (2-ethylpyridine ligands) before (lower) and after (upper) exposure to SFC conditions. Characteristic peaks for methanol and methoxy groups are present on the particles, indicating silyl ether formation.

(lower) and after (upper) SEF are compared. The common peaks between the two spectra are from the bonding (2ethylpyridine) and the ethylene bridges in the BEH particle (