Introduction
Supercritical fluid chromatography (SFC) has been a powerful technique for many years and has experienced several surges in popularity over that time frame [1-3]. One of the most widely embraced applications of SFC in the first several years of the new millennium was preparative chiral resolution [4-5]. The appearance of commercially available SFC instrumentation capable of processing several grams of material per hour led to numerous labs transitioning away from established normal phase HPLC operations to embrace SFC. The benefits of SFC were quickly realized and early adopters sought to transition more than just chiral resolutions to this “green technology” [6-8]. The transition of chiral separations from normal phase HPLC to SFC is relatively straightforward due to the similarity of both chromatographic techniques employing the same stationary phases based on similar retention mechanisms.
Converting achiral separations to SFC proved to be somewhat more challenging. While normal phase techniques such as flash chromatography on silica gel are relatively easily translated to SFC, the more widely used reversed-phase separation techniques are not. The alkyl-chain based stationary phases such as C18 used in reversed-phase chromatography retain achiral molecules via a hydrophobic exclusion retention mechanism, while achiral SFC relies on polar interactions. The retention of achiral analytes by SFC has been studied by various groups to show that a combination of retention mechanisms is often responsible for the selectivity observed in SFC separations [9-12]. Consequently, several different stationary phase chemistries are often used for achiral SFC separations, with the lack of a truly “universal” achiral SFC stationary phase being one of the factors limiting further adoption.
Practical Utilization of SFC
High-throughput purification groups frequently rely on a rapid analytical screen to enable preparative method selection. The commercial availability of mass-directed SFC systems for high-throughput use has drawn additional attention to the fact that no single achiral SFC phase can match the widespread applicability of C18 [13-16]. High-throughput groups that are making the transition to SFC are finding it necessary to utilize several columns to obtain widespread coverage of their compound collection. M.L. de la Puente reports on using a crosslinked diol in addition to a 2-ethylpyridine (2-EP) column to deliver a 98% success rate [17]. This two-column approach is an improvement over the group’s past approach of utilizing a five-column solution consisting of diol, 2-EP, benzenesulfonamide, diethylaminopropyl (DEAP), and dinitrophenyl [18]. Aurigemma describes screening a pyridine/diol mix column, a hydroxyamino pyridinyl column, a hydroxyamino dipyridinyl column, and a diol/monol mix column in SFC mode on a walk-up system to enable moving compounds downstream to a preparative scale [19]. Ventura presented the tandem use of achiral and chiral columns in SFC for compound purification, and expressed a preference for the use of 2-EP, 4-ethylpyridine (4-EP), silica, pyridylamide, crosslinked diol, and imidazole based columns for the achiral portion of the platform [20].
Examination of these examples highlights the variety of different stationary phases that are required to cover the diverse molecules encountered by end users in the pharmaceutical industry. Nevertheless, the quest for a more universal achiral SFC stationary phase has been an important goal for some time, with ongoing evolution and continuous improvement in recent years. It is important to note the prevalence of basic heterocycles as well as hydrogen bonding groups on the ability of a stationary phase to deliver widespread selectivity. Farrell prepared several unique phases combining these functionalities and compared them to commercially available phases [21]. No single phase seemed to deliver universal selectivity to challenge the champion role held by C18 and RPLC.
Evaluation of Both Existing and New SFC Stationary Phases
Merck has used SFC for chiral applications for the past 15 years and desires to continually increase the number of achiral samples processed via SFC. In 2011, we collaborated with a leading column manufacturer to investigate interactions between monofunctional test probes and various achiral stationary phases in hopes of identifying a universal phase [22]. In addition to demonstrating that no single column was the best at chromatographing the analytes of all functional classes examined, another interesting aspect of the study was the gain in efficiency experienced when using a non-endcapped stationary phase versus its endcapped counterpart. The study selected the use of a nonendcapped 2-EP column for carboxylic acid analytes, a non-endcapped DEAP for alcohol analytes, a non-endcapped nitro column for amides, and a non-endcapped imidazole based column for amine containing compounds. Silanol activity was shown to be of high importance in achiral SFC and enhances stationary phase performance. The fact that two of the four best performing columns in this study contained basic heterocycles, 2-EP, and imidazole also confirmed the importance of these groups in new phases.
In a recent collaboration between Merck and Professor Myung Ho Hyun of Pusan National University (S. Korea) we prepared ten novel stationary phases, each containing a basic heterocycle attached to the silica support through a hydrocarbon based tether, and evaluated their general utility for the SFC separation of a number of achiral analytes [23]. The ten novel phases were based on the established arylalkylamide structure used in chiral stationary design to accentuate simultaneous multi-point interactions between analyte and stationary phase [23]. Building upon the considerable experience of the Hyun group in the design of chiral stationary phases for chiral separations, a group of racemic stationary phases containing a hydrogen bonding group rotated out of the plane of the basic heterocycle were prepared (Figure 1). The racemic nature of these stationary phases makes them incapable of resolving enantiomers but could render them very effective at delivering added selectivity for mixtures of achiral molecules. The 3D spatial arrangement of this design motif is believed to be important for enabling simultaneous multi-point interactions between analyte and stationary phase based on π-π and hydrogen bonding interactions [24]. Amide and urea groups were selected as the hydrogen bonding groups in the tether due to their past success in chiral stationary phase development [24-25].
Figure 1. Structures of novel stationary phases developed in collaboration between Merck and Professor Myung Hyun from Pusan National University.A high degree of emphasis was placed on designing the novel phases to combine the correct chemical functionality in the most ideal spatial conformation to give wider selectivity coverage than is currently available. Users today can often select the appropriate achiral column to utilize for an achiral SFC separation, but no single phase can be relied on to resolve all mixtures. For example, in these labs chromatographers frequently resolve diastereomers on the nitro column through utilizing a combination of ionic and π-π interactions to provide acceptable selectivity for these extremely similar compounds (Figure 2). One limitation of the nitro column is that it can display excessive retention, as highlighted by Ventura, with 15 of 70 compounds analyzed on the nitro column tested not eluting from the column [20]. We have also experienced increased retention on the nitro column in comparison to the 2-EP phase and consequently preferred its use when implementing mass-directed purification in our laboratory, as it provided faster clearance of sample from the injection system while doing the SFC version of at-column-dilution [13].
Figure 2. Chromatogram showing resolution of a pair of proprietary diastereomers on the nitro column in achiral SFC. The chromatographic conditions are 10% isocratic methanol in CO2 @ 3 mL/min flow rate. The column temperature is 25 °C and the back pressure setting is 100 bar. The peaks at 6.14 and 7.19 minutes are the diastereomers.When performing preparative separations requiring stacked injections, additional columns are often screened in hopes of identifying alternative stationary phases requiring less co-solvent than the nitro column. Figure 3 provides an example of a crude reaction analysis of a proprietary sample encountered in our laboratory containing a desired basic compound, a regioisomer, and excess reactant. The top trace is a gradient analysis performed on a nitro column with the desired peak eluting at 8 minutes while the bottom trace is the same gradient analysis performed on a hydroxyamino pyridine column with the desired compound eluting at 6.5 minutes. On the preparative scale, the hydroxyamino pyridine column would require 30% less solvent than the nitro column, leading to considerable solvent savings.
Figure 3. Chromatographic comparison of a crude reaction mixture analyzed on the nitro column (top trace) and the hydroxyamino pyridine column (bottom trace). Both are gradient analysis from 5 to 50% methanol in CO2 over 7 minutes @ 3 mL/min. The column temperature was 40°C and the back pressure was 100 bar. The peak at 8.017 minutes on the nitro column and 6.457 minutes on the hydroxyamino pyridine column represent the desired molecule.For the Merck/Hyun collaboration, we selected 27 diverse test probes to compare the widespread applicability of the novel phases against four commercially available columns that are widely embraced in the SFC field (Figure 4). The commercially available columns selected to serve as comparators in our study included a 2-EP, an imidazole based column, an amide linked propylpyridyl (PPA), and a urea linked propylpyridyl (PPU) column. The test probes included acidic, basic, and neutral compounds with the diversity present being extremely important in evaluating widespread applicability of the phases. Acidic compounds have been known to display excessive retention on basic type phases through ionic interactions, much like the NO2 retention trend noted by Ventura. Basic compounds are well known to tail significantly on silica based stationary phases [26]. Much of the success of the 2-EP column is derived from its ability to deliver efficient peaks for basic compounds without the use of a basic additive leading to our focus of incorporating basic heterocycles in our novel phases [27].
Figure 4. The 27 components utilized in the Merck/Hyun collaboration to identify retention trends on the columns being examined.The test probes selected were each chromatographed on the ten novel phases as well as the four commercially available columns in a combination of three different standard mixtures. The chromatographic conditions consisted of a linear gradient of 4-rt40% methanol in CO2 over 15 minutes at 3.0 mL/min with a column temperature of 25°C and a back pressure setting of 150 bar. Peak shape was monitored as was each stationary phase’s ability to selectively differentiate the standard components from one another. Despite some encouraging results, it quickly became evident that the pyrazine columns and quinolone based columns were not able to outperform the commercially available columns, with clustering of peaks, poor resolution, inefficient peak shape, and tailing observed. The 4-Py-A and 2-Py-U columns were clearly the two superior columns out of the ten novel phases prepared. Unfortunately, it was also clear that our hypothesis regarding the potential importance of simultaneous multipoint interactions for improving the generality of achiral SFC was not borne out by the data, with the novel 2-Py-U column showing inferior resolution of the critical pair components O&P and T&S compared with the commercially available PPU column (Figure 5). Not surprisingly, six of the novel pyridyl columns containing the amide and urea linkages provided excellent peak shape and minimal tailing for the basic analytes. This follows the general trend noted for the 2-EP for the past ten years and is in line with the use of an imidazole phase in our lab for mass-directed achiral SFC separations.
Figure 5. Chromatograms of 20 of the test probes in the Merck/Hyun study analyzed on four of the novel phase as well as four of the commercial comparators.As previously reported by Perrenoud, ionizable analytes are extremely useful in evaluating stationary phases in achiral SFC [26]. We previously observed excessive retention of niflumic acid on an imidazole based stationary phase [22]. A similar trend with niflumic acid and several other carboxylic acid test probes was noted on the commercial PPA column but the trend was absent on the novel 2-Py-A column. One possible explanation for this phenomenon, based on analogy with chiral stationary phase studies, invokes the presence of underlying residual amino groups that are left over when using a grafting approach to prepare stationary phases by acylation of amino silica [25]. The presence of unreacted amino groups in the PPA column would be expected to lead to ionic interactions with carboxylic acid test probes, resulting in increased retention. In contrast, the presence of residual amino groups would be expected to lead to decreased retention of basic analytes by a charge repulsion mechanism, which we experienced as well. Interestingly, analysis of the retention of both the carboxylic test mix and amine test mix on an aminopropyl column confirmed this predicted retention behavior, with the extreme retention for niflumic acid very closely matching that observed on the PPA column.
A decrease in retention for basic analytes on the pyridyl amide column series was observed in the Merck/Hyun collaboration as the pyridine substitution position moved from the 2 to the 3 to the 4 position. The opposite trend was observed when examining the retention behavior of the carboxylic test probes on these amide-linked pyridine phases. This behavior could perhaps involve the increased steric availability of the pyridine nitrogen in the 4-Py-A stationary phase, leading to increased retention for the acid analytes, and increased ionic repulsion for the basic analytes. It should also be noted that the position of the pyridine nitrogen can also impact inter- and intra-molecular hydrogen bonding, which can be important for “tuning” retention characteristics. For example, the 2-Py-A stationary phase can readily engage in an intramolecular hydrogen bond with the tether amide hydrogen, whereas the 4-Py-A stationary phase cannot.
One investigator who has studied stationary phases for SFC use for many years has used the critical pair of fenoprofen and flurbiprofen in evaluating the selectivity of different phases. We examined the separation of these two carboxylic acid test probes on several columns. One of the most interesting separations of these two peaks was seen on a crosslinked diol column. Ironically, the two peaks co-eluted when unmodified methanol was used as co-solvent, yet were baseline resolved with the use of 0.1% NH4OH as an additive, as described by Hamman [28]. This can be seen in Figure 6. This example continues to highlight the difficulty in identifying a stationary phase for widespread use in achiral SFC without the addition of an additive.
Figure 6. Chromatograms of fenoprofen and flurbiprofen analyzed on a crosslinked diol column. The top trace was performed with unadulterated methanol in CO2 while the bottom trace was performed with the addition of 0.1% NH4OH to the methanol.The stationary phases containing urea tethers exhibit more efficient peaks and fewer instances of co-elution than the corresponding amide-linked stationary phases. This trend is also apparent with the commercial PPA and PPU columns, with the PPU resolving three critical pairs that were not resolved by the PPU column, as can be seen in Figure 5. This finding is in keeping with previous results from Hyun showing the effectiveness of urea-containing tethers for chiral separations [25]. A typical separation performed in our lab can be viewed in Figure 7 to highlight the performance of the 2-Py-U. The desired proprietary compound is observed at 4.51 minutes in contrast to eluting in the void when analyzed under reversed-phase conditions on a C18 column. This example displays the high degree of selectivity, efficient peak shape, and minimal tailing encountered when using stationary phases containing a basic heterocycle and urea based tether.
Figure 7. Chromatogram showing resolution of a proprietary compound in our lab on the 2-Py-U column. The chromatographic conditions are a linear 7-minute, 5–50% methanol gradient in CO2 @ 3 mL/min fl ow rate. The column temperature is 25°C and the back pressure setting is 100 bar. The peak of interest elutes at 4.51 minutes.Conclusions and Path Forward
The development of improved chromatographic phases for more general achiral SFC separations is an important goal for enabling the increased uptake and impact of the SFC technique. While the quest for a truly “universal” achiral SFC stationary phase is still ongoing, experience to date suggests that nitrogen heterocycle-containing stationary phases are important, as are hydrogen bonding groups. A universal achiral SFC phase could perhaps utilize multiple retention mechanisms such as ionic, π-π, dipolar, and hydrogen bonding interaction and could simultaneously deliver the ability to resolve various isomeric forms like the nitro column, while providing a high degree of selectivity with efficient peak shape like the basic heterocycle columns such as the 2-EP and the imidazole columns. Ideally, the universal achiral SFC phase would deliver these capabilities while having its ionic and electrostatic properties tuned to avoid excess retention. The phase would also deliver the ability to selectively differentiate carboxylic acids like the crosslinked diol without the use of additive which is occasionally necessary for separation of both acidic and basic molecules. The accumulated knowledge derived from decades of hypothesis-driven design and development of chiral stationary phases can be applied to the problem of developing new SFC stationary phases.
To date, our hypotheses regarding the optimal positioning of hydrogen bonding elements adjacent to the heterocycle have not led to substantial improvements in stationary phase generality; however, the knowledge derived from additional hypothesis-driven investigations of this type can be expected to lead to cumulative improvements in our understanding of the requirements for achiral SFC design over time. At this time, the most general achiral stationary phases remain the commercially available imidazole, crosslinked diol, pyridylpropyl urea, and pyridypropyl amide columns. However, it is our hope that additional investigations into the next generation of stationary phases may afford improved generality as it is still unclear what the appropriate mix of functionality is to deliver ”universal selectivity” in achiral SFC, but each attack at the problem brings us closer to a solution.
Acknowledgement
The authors would like to thank Zhenyu Wang, Ph.D., Principal Scientist, Pharmaceutical Sciences & Clinical Supply, Merck & Co., Summit, NJ, for his valuable input.
Author Biographies
Ray McClain is an Associate Principal Scientist at Merck’s West Point, PA, facility. Ray has been at Merck for 17 years and currently manages the chromatographic support of West Point’s Discovery Chemistry Department. His responsibility includes both chiral and achiral resolutions on scales ranging from analytical to preparative. Ray enjoys the use of SFC and serves on the Board of the Green Chemistry Group.
Myung Ho Hyun is a Professor at the Department of Chemistry, Pusan National University, Busan, South Korea. He has been involved in the development of HPLC chiral stationary phases based on smart chiral selectors such as chiral crown ethers, synthetic small chiral molecules, and chiral bidentate ligands for more than 25 years. He has also been involved in the development of chemical sensors for chiral recognition by bonding the smart chiral selectors to appropriate signaling units.
Christopher J. Welch is Science Lead for Analytical Chemistry within the Process and Analytical Chemistry area at Merck Research Laboratories in Rahway, NJ. Chris co-chairs the New Technologies Review and Licensing Committee (NT-RLC), the organization that oversees identification, acquisition, and evaluation of new technologies of potential value to Merck Research Laboratories. Chris also co-chairs the MRL Postdoctoral Research Fellows Program.
References
- W. Farrell; C. Aurigemma; and D. Masters-Moore. J. Liq. Chromatogr. Relat. Tech. 2009; 32: 1689-1710.
- L. Taylor. http://www.chromatographonline.com; Oct 2, 2013 (accessed on 24-Mar-2014).
- D. Speybrouck; D. Corens; and J. Argoullon. Current Topics in Medicinal Chemistry 2012; 12: 1250-1263.
- C. Welch; W. Leonard; and J. DaSilva, et al. LC GC N. Am. 2005; 23: 16-29.
- C. White. J. Chromatogr. A 2005; 1074: 163-173.
- T. Berger; K. Fogleman; and T. Staats, et al. J. Biochem Biophys. Meth. 2000; 43: 87-111.
- M. Ventura; W. Farrell; and C. Aurigemma, et al. J. Chromatogr. A 2004; 1036: 7-13.
- C. White and J. Burnett. J. Chromatogr. A 2005; 1074: 175-185.
- C. West and E. Lesellier. J. Chromatogr. A 2006; 1110: 181-190.
- C. West and E. Lesellier. J. Chromatogr. A 2006; 1110: 191-199.
- C. West and E. Lesellier. J. Chromatogr. A 2006; 1110: 200-213.
- C. West and E. Lesellier. J. Chromatogr. A 2006; 1115: 223-245.
- R. McClain; A. Dudkina; and J. Barrow, et al. J. Liq. Chromatogr. Relat. Tech. 2009; 32: 483-499.
- K. Ebinger; H. Weller; J. Kiplinger; and P. Lefebvre. JALA 2011; 16: 241-249.
- V. Lazarescu; M. Mulvihill; and L. Ma. http://www.chromatographyonline.com; May 1, 2011 (accessed on 24-Mar-2014).
- A. Mich; B. Matthes; R. Chen; and S. Buehler. http://www.chromatographyonline.com; Nov 1, 2010 (accessed on 24-Mar-2014).
- M.L. Puente; P. Soto-Yarritu; and C. Anta. J. Chromatogr. A 2012; 1250: 172-181.
- M.L. Puente; P. Soto-Yarritu; and J. Burnett. J. Chromatogr. A 2011; 1218: 8551-8560.
- C. Aurigemma; W. Farrell; and J. Simpkins, et al. J. Chromatogr. A 2012; 1229: 260-267.
- M. Ventura, Presentation: Utilizing Coupled Achiral and Chiral Columns to Improve SFC Purifications, Pittcon2013, Philadelphia, Pa., March 21, 2013.
- W. Farrell. Green Chemistry Conference 2011, http://greenchemistrygroup.org/Program2011.html.
- R. McClain and M. Przybyciel. LC GC N. Am. 2011; 29: 894-906.
- R. McClain; M. Hyun; Y. Li; and C. Welch. J. Chromatogr. A 2013; 1302: 163-173.
- W. Pirkle; M. Hyun; and B. Bank. J. Chromatogr. 1984; 316: 585-604.
- W. Pirkle and M. Hyun. J. Chromatogr. 1985; 322: 295-307.
- A. Perrenoud; J. Boccard; J. Veuthey; and D. Guillarme. J. Chromatogr. A 2012; 1262: 205-213.
- J. Zheng; J. Pinkston; P. Zoutendam; and L. Taylor. Anal. Chem. 2006; 78: 1535-1545.
- C. Hamman; D. Schmidt; M. Wong; and M. Hayes. J. Chromatogr. A 2011; 1218: 7886-7894.