Anal. Chem. 2010, 82, 4925–4935
Supercritical Fluid Chromatography Larry T. Taylor Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061-0212 Review Contents Perspective General Reviews Instrumentation SFC/MS Preparative SFC Stationary Phases Method Development Chiral SFC Literature Cited
4925 4926 4927 4928 4929 4930 4931 4933 4934
PERSPECTIVE Supercritical fluid chromatography (SFC) has been practiced sporadically for approximately 50 years. Early studies used packed columns with varying degrees of success, and it would be fair to say that SFC did not generate much interest in the separation science community in this period. In the early 1980s, interest in SFC exploded when it was reported that SFC could be easily accomplished with wall coated open tubular columns similar to the columns currently used for gas chromatography. The columns, however, differed in two respects: (a) SFC columns had smaller inner diameters and (b) the polymeric stationary phase that coated the walls was more extensively cross-linked. The emergence of SFC with a mobile phase (i.e., carbon dioxide) that exhibited appreciable, enhanced solvating power relative to helium and hydrogen was primarily promoted by users of gas chromatography (GC) who saw SFC as a way to increase the limited sample base afforded by GC. This technology flourished for some time, but it had several disadvantages for widescale acceptance by the analytical community. Specifically, the technique which relied mostly on nonpolar stationary phases was traditionally limited to relatively nonpolar analytes because (a) pure CO2 as the mobile phase exhibited poor solvating power similar to hexane, (b) robust sample injection was not feasible, (c) a wide polarity range of stationary phases was not available, (d) high quality separations required approximately 60 min, (e) gradient CO2 pressure and flow-rate were coupled such that flow-rate changed as mobile phase pressure changed, (f) passive, fragile fused silica restrictors at the column outlet often plugged and the separation had to be aborted, and (g) multigram scale-up of a successful separation was not feasible. One’s view of SFC today is entirely different from that of 25 years ago. Today, SFC is a separation technique similar to high performance liquid chromatography (HPLC) using mostly the same hardware and software as HPLC. The mobile phase is a binary or ternary mixture with CO2 as the main component. The separation is usually performed as a gradient elution where composition of the mobile phase is changed versus time. Polar stationary phases such as bare silica, cyanopropylsilica, 3-aminopropylsilica, and 2-ethylpyridylsilica are routinely employed. 10.1021/ac101194x 2010 American Chemical Society Published on Web 05/13/2010
SFC has numerous practical advantages relative to reversed phase HPLC such as higher speed/throughput, more samples/ day, more rapid equilibration, and shorter cycle time. SFC yields lower operating cost and lower column pressure drop, and it is orthogonal to reversed phase HPLC. Solvent consumption is low; therefore, waste generation is also minimal. Finally, the compounds of interest can be isolated in a relatively small amount of solvent because CO2 vaporizes away. This feature is particularly important for preparative applications in which elution volumes can be large. SFC, nevertheless, will always be compared with HPLC. As a general observation, if you can perform the normal phase HPLC separation, you can probably carry-out the same separation or something analogous to it by SFC. On the other hand, replacement of reversed phase HPLC with SFC will not be totally favorable except in limited situations. Two differences distinguish SFC from HPLC systems. The SFC pump system must have a chilled pump head to maintain the CO2 in the liquid state, and the whole system for UV detection or just the packed column for MS detection has to be under pressure. SFC instruments come in two forms. Analytical scale instruments have flow rates less than or equal to 20 mL/min. Preparative instruments are for industrial-scale runs and can be further subdivided. Semipreparative instruments have flow rates from 20 to 200 mL/min, whereas larger preparative systems handle flow rates at liters per minute with columns measured in inches rather than millimeters. Only a handful of vendors sell SFC systems. Jasco, Inc. and Selerity Technologies (now known as Sandra Selerity Technologies) continue to offer analytical SFC instruments. TharSFC was purchased by Waters in 2009 with a complete name change although the product line continues to be analytical and preparative SFC instrumentation. Aurora SFC Systems, Inc. appeared during the report period with its Fusion module that converts a standard HPLC into an analytical SFC. One function of the module is the precompression of CO2, and the other is the control of the system outlet pressure. In late 2009, Aurora and Agilent in a formal arrangement introduced this new approach to SFC where one side of the HPLC pump is stated to deliver compressed CO2 with the accuracy and precision of a normal liquid. Academic acceptance of SFC is currently not high worldwide but even less so in the United States. Early instrumentation involved wall coated open tubular columns which were riddled with problems (1). Challenges to their use led to disappointment and disillusionment. Furthermore, the early packed column instruments were not as hardy as HPLC ones. Nowadays, SFC hardware and software are exceedingly robust, less pure “softdrink” grade CO2 is readily useable, and the range of samples Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
4925
Table 1. Application Notes from Internet and Trade Publications “Profiling the Isomerization of Biologically Relevant (E)-(Z) Isomers by SFC”, J. Cole, R. Chen, June 2009, LCGC North America (Supplement), pp 24-25. “Fast Separation of FFA, FAME, and Glycerol for Biodiesel Analysis by SFC”, J. Cole, J. Lefler, R. Chen, February 2008, LCGC North America (Supplement). “A Feasibility Study of Using SFC with UV-ELSD for Food and Beverage Analyses” J. L. Lefler, R. Chen, June 2008, LCGC North America (Supplement). “An Acetonitrile-Free Chromatographic Methodology for Melamine Detection and Quantification using SFC”, J. L. Lefler, R. Chen, February 2009, LCGC North America (Supplement). “Enhancement of UV Detection Sensitivity in SFC Using Reference Wavelength Compensation”, L. Subbarao, J. Cole, R. Chen, September 2009, LCGC North America (Supplement). “Separation of Ionic Analytes Using Supercritical Fluid Chromatography”, L. T. Taylor, September 2009, LCGC North America (Supplement), p 62. “Effect of Varying Co-Solvents in SFC Method Development on a Whelk-O 1 Chiral Stationary Phase”, T. Szczerba, P. Wrezel, June 2008, LCGC North America. “Screening Process for Development of Enantioselective SFC Separation Methods”, G. B. Cox, February 2008, LCGC North America (Supplement), p 15. “Feasibility of Using ELSD to Trigger Fraction Collection in Small Scale Purification by SFC”, J. Cole, R. Chen, July/August 2009, LCGC North America (Supplement), pp 1-2. “Keith Bartle on SFC and his Career in Chromatography”, Chromatography Today, Feb/Mar, 2009, pp 20-22. “A Brief Review of Interfacing SFC with MS”, R. Chen, Chromatography Today, Feb/Mar, 2009, pp 11-13. “Recent Developments and Future Challenges in Supercritical Fluid Chromatography”, S. J. Rumbelow, Chromatography Today, February/March 2009, pp 15-19. “Modern Chiral Separations Using HPLC and SFC for Method Development and Prep Purification”, G. W. Yanik, Chromatography Today, Sept. 2009, pp 44-46. “Meeting Review: SFC 2009-3rd International Conference on Packed Column SFC”, L. T. Taylor, The Column, September 2009, p 12. “Multicolumn Preparative SFC: An Advanced Solution to Scale-up Difficulties”, Z. Ali, J. Kocergin, V. Edwin, The Peak, March 2009, pp 16-21. “Meeting Review: SFC for Pharmaceutical Analysis-Opportunity or Challenge?”, M. Hanna-Brown, T. Lynch, The Column. “Separation of Ionic Analytes via SFC: Achieving the Impossible”, L. T. Taylor, Chromatography Online, June 2009. “Packed Column SFC-Based Analysis of Lipid Modification Reactions: Overview and Applications”, D. G. Hayes, Amer. Pharm Rev., April/March 2009, pp 46-52. “The Use of SFC for Chiral Pharmaceutical Analytical and Preparative Separations”, J. Van Anda, Amer. Pharm. Rev., April 2009, pp 48-53. “SFC of Peptides-State of the Art”, L. Taylor, Amer. Pharm. Rev., May 2009, pp 48-53. “Pushing the Limits of SFC to Resolve Chiral Molecules”, C. Kraml, Amer. Pharm. Rev., July 2008, pp 80-83. “Preparative SFC in Drug Discovery”, L. Miller, M. Potter, Amer. Pharm. Rev., September 2008, pp 112-117. “SFC as a Green Chromatographic Technique to Support Rapid Development of Pharmaceutical Candidates”, J. O. DaSilva, H. S. Yip, V. Hegde, Amer. Pharm. Rev., Jan/Feb, pp 98-104.
continues to expand. In other words, the range of applicability is getting closer and closer to the range of applicability of reversed phased HPLC. Yet, the inertia holds and many academicians who are accustomed to HPLC are loath to invest time and energy to switch. In this regard, one would think that SFC would be the first to get on the “environmentally friendly” bandwagon. Being green is apparently fine and not a bad thing, but most people seemingly go for SFC because of its speed and fast method development. Experts in the field agree that SFC has established itself as the preferred way of doing chiral analysis on both the analytical and preparative scales. They also say that SFC will become the norm for small-scale purifications. Increased interest in (a) the petrochemical and food industries, (b) the determination of environmental air quality, (c) biodiesel quality control, (d) protein separations, etc. can be expected in the future. The field of SFC is solely examined in this biannual review. The published literature in the period from January 1, 2008 to December 31, 2009 has been considered. The document builds upon the previous biannual review of the author published during June 2008 in which (a) analytical scale chiral and achiral separations, (b) supercritical fluid simulated moving bed, (c) preparative SFC, (d) SFC coupled with mass spectrometry, (e) natural product applications, and (f) open tubular column studies were reviewed (2). The searched database was compiled by SciFinder Scholar which contained English and non-English journal articles, patents, books, and abstracts of regional and National American Chemical Society meeting presentations. The search keywords were “SFC” 4926
Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
and “supercritical fluid chromatography”. Although the number of citations fails to exceed 110 for a second two year period, the individual contributions represent workers from all over the world. The bulk of the peer reviewed, published investigations concerns packed columns and pharmaceutical applications. A somewhat disturbing trend, however, appears to be developing in that references to articles from Internet chromatographic sites (
[email protected], www.chromedia.org, chromtoday@ chromtoday-mailers.com, and www.chromatographyonline.com) and trade publications are on the rise (see Table 1). Yet, the customary search engines list these documents as bona fide references. A few of these reports are highly significant and will, thus, be covered in this biannual review. The author, however, has serious questions regarding their peer reviewed status and long-term availability to the chromatographic community. GENERAL REVIEWS The field has been well reviewed during the two year period by a number of publications. A critical review of activity in industry versus academia with numerous expert commentaries was published in 2008 (3). The vendor situation in the U.S. was considered along with the current available instrumentation for conducting analytical and preparative separations. The separation of ionic analytes via SFC was reviewed wherein the use of additives was shown to dramatically extend the range of solute polarity amenable to separation (4). Ion suppression and ion pairing were suggested to be important mechanisms for SFC of ionic analytes. Packed column SFC as the new modern normal phase chromatography
has been reviewed (5). Fundamental differences between packed column and open tubular column were discussed. Divisions in chromatography were noted to not be useful. SFC for the 21st century appeared as a review during the biennium (6). A field guide to new SFC instrumentation which lists information taken from correspondences with manufacturers and their Web sites and printed brochures appeared (7). Companies that were polled included Chiral Technologies Inc., Jasco Inc., Modular SFC, Novasep Process, Princeton Chromatography, Inc., Regis Technologies, Inc., Selerity Technologies, Inc., and Thar Instruments, Inc. The enantioseparation of 123 clinically used racemic drugs by SFC on commercial chiral stationary phases that were available in 2006 was reviewed (8). The mobile phase compositions with organic modifier and additives were listed. The data were extracted and compiled from the ChirexBase database (Marseille, France). All the drugs included are listed according to the 13 therapeutic classes of the Anatomical Therapeutic Chemical classification. The nature of the mobile phase precluded the use of several classes of chiral stationary phases (CSPs) such as crown ethers, ligand exchange, and protein-based CSPs. The polysaccharide-based CSPs were responsible for two-thirds of the enantiomer separations listed. Another review highlighted the separationsofchiralpharmaceuticalsanddrugsbyliquidchromatographic modalities (i.e., HPLC, capillary electrochromatography, SFC, SubSFC, and thin layer chromatography) utilizing polysaccharide CSPs (9). Coated versus immobilized CSPs were discussed. A comparison of CSPs commercialized by different companies was considered. Another review summarized a variety of novel supercritical fluid chromatography/mass spectrometry (SFC/MS) methods for chemical analysis that have been reported in peer-reviewed publications (10). Efforts directed toward the establishment of a hardware framework for SFC/MS was predicted to elevate the technique to become the method of choice for high-throughput chemical analysis in both academic and industrial areas. Representative SFC/MS methods for chemical analysis were summarized. Several additional reviews should also be mentioned here. The utility of SFC in metabolomic studies was also reviewed (11). A review of applications of SFC in cancer research, toxicology, peptides, lipids, etc. appeared (12). Main developments and applications of multidimensional chromatographic techniques in food analysis have also been reviewed (13). Approaches for efficient method development with immobilized polysaccharidederived chiral SP were reviewed (14). Multidimensional SFC methodologies such as SFC × SFC, SFC × HPLC, and silver ion SFC × RPLC were reviewed (15). INSTRUMENTATION A novel method of interfacing an acoustic flame detector (AFD) with modified SFC was presented (16). By applying resistive heating directly to the burner region between the restrictor outlet and the oscillating acoustic flame, infrequent severe noise, baseline drift, and peak deformations that can occasionally be observed with the AFD are eliminated. The approach worked equally well for modifier levels from 0 to 100% in the HPLC mode. The performance facilitated reliable application of the AFD as a universal detector in modified SFC and other separation modes that employ organic solvents in the mobile phase. The interface
was observed to reduce detector noise to near 10-25 Hz when an appropriate temperature was achieved. The method was easy to assemble, inexpensive to construct, and robust in its daily operation. Remote control of the vent/detector split flow ratio in packed column SFC (pcSFC) with flame ionization detection (FID) was demonstrated during the two year period by the same workers using a dual heated restrictor method (17). Restrictors stemming from a Tee at the separation column outlet were respectively fixed into an FID and a vent port, and their individual temperatures were controlled using resistively heated wires. Both system pressure and split flow could be manipulated. Isobaric altering of the split flow ratio was possible when opposing positive and negative temperature gradients were applied at the two restrictors. The method was used to establish stable detector operation in the analysis of n-alkanes under pcSFC-FID conditions that normally exhibit flame instability. Results indicated that this technique could be a relatively simple and inexpensive means of controlling system pressure and detector split flow ratios in pcSFCFID. Methodology was predicted to be compatible with methods where a conventional backpressure regulator is desired for pressure control. An alternative means of independently controlling column pressure in SFC by resistively heating the postcolumn restrictor was demonstrated by Li and Thurbide (18). Compared to conventional block heating methods, resistive heating provided at least 4 times greater pressure programming rates and allowed for much faster cooling times in between runs, thereby increasing sample throughput. The methodology was found to be equally effective for either capillary or packed column operating modes. The technique was stated to be a simple, inexpensive, and convenient alternative to limited passive restrictors or more costly and complex backpressure regulators. Four papers were published by Takahashi, et al. during the period. A corona-charged aerosol detector (CAD) was developed to improve the sensitivity, reproducibility, and quantitativeness of detection as compared with evaporative light scattering detection (ELSD) using a certified reference material of poly(ethylene glycol) (PEG) (19). The corona CAD was able to detect a 10 times more dilute solution of uniform oligomers compared to ELSD. Repeatability of the corona CAD was greater than ELSD. The CAD had better signal-to-noise at very low concentrations and exhibited a lower minimum detectable quantity. In another study, the quantitativeness of ELSD for SFC was evaluated using an equimass mixture of uniform PEG oligomers (20). All molecules have an identical molecular mass in uniform oligomers. They are, thus, useful for the accurate calibration of detectors. Using chromatograms of the equimass mixture of uniform oligomers to calibrate SFC-ELSD, it was possible to determine exact values of not only the average mass but also the molecular-mass distribution for a PEG 1540 sample. The average molecular mass was shifted to a higher value by several percentage points after calibration of the ELSD. The ELSD 2000 was from Alltech Associates. PEGs were separated by preparative SFC using an SFCpak SIL-5 silica gel column (Jasco Co.) with pore size ) 6 nm and particle size ) 5 µm. Column dimensions were 250 × 4.6 mm. Equimass mixtures of uniform PEG oligomers with various degrees of polymerization were subjected to a precise calibration Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
4927
of SFC-ELSD by the same workers to examine the quantitativeness of the detector in terms of the concentration and degree of polymerization (21). The concentration dependence of the ELSD was small, and the response of the ELSD was linear against the injected concentration of analyte. Calibration curves for SFC-ELSD were able to be determined for various degrees of polymerization using data concerning the equimass mixture of uniform oligomers. The method has been developed to permit the certification of fractions of all the components in PEG samples that are needed for certified reference materials issued by the National Metrology Institute of Japan. The material is suitable for calibration against both masses and intensities regarding matrix-assisted laser desorption/ionization time-of-flight MS. A condensation nucleation light scattering detector (CNLSD) was adapted for use as a detector for SFC (22). The performance of the CNLSD was evaluated and compared to ELSD using a welldefined equimass mixture of poly(ethylene glycol) 1000. The CNLSD was able to detect a 10 times less concentrated solution of uniform oligomers compared to ELSD. The quantitativeness of CNLSD in terms of concentration and degree of polymerization was evaluated using an SFC system and PEG standards, and it was high enough to obtain the molecular mass distribution of poly(ethylene glycol) 1000 without any calibration. On column density measurement of CO2 in packed capillary columns using Raman microspectroscopy of the position of the Fermi doublet has been reported (23). Correlation of the spectrum with density was calibrated over a pressure range of 15-290 atm at 125° and 150 °C which allowed for determination of the density gradient of fluid flowing through a packed capillary column. Analysis was said to follow in a later publication. Open tubular capillary column SFC with 1-n-butyl-3-methylimidazolium methyl sulfate [bmin][MeSO4] as the stationary phase and supercritical CO2 as the carrier fluid was employed to measure retention factors of organic solutes within 313-353 K and 8.5-23.2 MPa (24). Solute selection included 18 compounds of diverse volatilities and chemical functionalities. At constant temperature, an increase in CO2 density produced distinct shifts in relative retention, thus providing some pressure-tunable selectivity. Analysis of the relative retention data by regular solution theory resulted in approximate values of the solubility parameter of CO2-expanded [bmin][MeSO4]. A new approach to SFC instrumentation, which allows a standard HPLC system to reversibly function as an SFC system, was reported (25). The system removes the compression requirements from an HPLC pump, allowing it to perform pulseless metering. By isolating compression and metering, the detection limits and dynamic range were improved well over an order of magnitude. Detection limits (S/N ) 3) of 0.004% of the parent peak were demonstrated, while both were on the same linear calibration curve. Linearity was 0.99999 from 0.01 to 100% (0.001-10 mg/mL racemate or 2.5 ng to 25 µg of each enantiomer injected). Average S/N at 0.1% was 58. Retention time stability was ±0.15% to ±0.70% over sets of five injections. Area reproducibility was ±0.27% to ±0.46% RSD when S/N was >100. It was noted that SFC is almost never used for trace analysis. A significant improvement in detection limits and dynamic range could allow much more extensive use of SFC including use in regulated 4928
Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
environments. The improvement in background noise, sensitivity, and dynamic range appeared to be related to the new approach to pumping CO2. SFC/MS The summarization of a variety of novel SFC/MS methods for chemical analysis that have been reported in peer-reviewed publications has been published (26). Efforts toward the establishment of a hardware framework for SFC/MS were predicted to elevate the technique to become the method of choice for highthroughput chemical analysis in both academic and industrial areas. Another review appeared which extensively discussed applications of SFC/tandem MS in pharmaceuticals (27). An interesting paper appeared that describes the ionization of samples in the absence of an applied electrospray voltage when SFC/MS was used with some compounds showing increased sensitivity (28). A series of test standards were analyzed with a range of pressures and modifier percentage conditions. The methodology was compared with three established liquid-to-gas ionization mechanisms: thermospray, charge residue model of electrospray ionization, and sonic spray ionization. The most important point to note from this report was that enhanced ionization was observed under novo-spray conditions in a SFC/ MS configuration, which in certain cases provided a lowering of the overall limit of detection. SFC/MS was used for the separation and analysis of integral membrane proteins and hydrophobic peptides (29). Detergents were rapidly and effectively separated from the proteins and peptides, yielding them in a state suitable for direct mass spectrometric analysis. Detergents are generally used to solubilize membrane proteins, but they interfere with the crystallization process which is essential to X-ray studies. They also cause severe ion suppression effects that hinder MS analysis. Gramicidin and bacteriorhodopsin were purified from detergents and lipids. Photosystem II was also investigated by SFC/MS, and a total of 16 out of 26 core proteins were eluted in 15 min. The hyphenation of SFC to atmospheric pressure chemical ionization ion trap MS has been detailed (30). Temperature, flow rates of gas, and mobile phase introduced in the source, position of the restrictor, ionization additives, and conditions of autotune were studied. The latter had to be performed with parameters as close as possible to analytical conditions (i.e., subcritical conditions). Unambiguous identification and structure elucidation of several additives in formulated car lubricants were presented. In another study, the separation of all six dimethylaniline isomers by SFC without derivatization was reported (31). The combination of SFC with ESI-MS provided selective detection in crude extracts of spiked (40 ppb of 3,5-dimethylaniline) raw materials that are commonly used in the manufacture of consumer hair-dye products. An analytical system that enables the simultaneous rapid analysis of lipids with varied structures and polarities through the use of SFC/MS has been reported (32). Lipid mixtures included phospholipids, glycolipids, neutral lipids, and sphingolipids. When cyanopropylated silica gel was used for the separation, all lipids were successfully detected and the analysis time was less than 15 min. The use of an octadecylsilylated column resulted in separations which were dependent on differences in the unsaturation of the fatty acid side chains. This system is a powerful tool for studies on lipid metabolomics because it is useful not only
as a fingerprinting method for the screening of diverse lipids but also for the detailed profiling of individual components. Carotenoid analysis was carried out by SFC/MS (33). The use of an ODS packed column separated seven analytes within 15 min. The use of a monolithic ODS column resulted in additional improvement in both resolution and throughput. The analysis time was reduced to 4 min by increasing the flow rate. Carotenoids in biological samples containing complex matrixes were separated effectively using three monolithic columns in series whose backpressure was very low. PREPARATIVE SFC Exergetic life cycle analysis for the selection of chromatographic separation processes in the pharmaceutical industry has compared preparative HPLC versus preparative SFC (34). It was concluded that, for this case, the most sustainable process as far as integral resource consumption is Prep HPLC, unlike the general perception that Prep-SFC outperforms Prep-HPLC. The poor score for Prep SFC was related to the production of liquid CO2 and the use of electricity for heating and cooling. Prep HPLC required 26.3% more resources due to its higher use of organic solvents and 29.1% more resources quantified in exergy. In this regard, a paper that discusses the use of preparative HPLC and SFC to generate individual enantiomers for discovery activities appeared (35). The usefulness of preparative chromatographic resolution of racemates was demonstrated through the presentation of numerous nonroutine, less straightforward case studies from the laboratories of Amgen. There were some racemates that do not scale-up as expected due to low solubility or poor sample loading or for other unknown reasons. The introduction of immobilized chiral stationary phase enabled separation of racemates that would be very time-consuming and/ or impossible due to poor compound solubility. The evaluation and implementation of a commercially available off the shelf analytical SFC/MS and a mass directed semipreparative SFC/MS system for use in a high throughput purification laboratory was described (36). The utilization of standard FractionLynx/AutoPurify functionalities to enable rapid incorporation into high throughput environments was emphasized. The analytical SFC allowed for rapid screening of crude reaction products and purity confirmation analysis of isolated fractions on a 2-ethylyridine column using a 2 min gradient. The mass directed system provided rapid separation of libraries of compounds on 10 mm (i.d.) × 10 cm semiprep columns with no modifier for the vast majority of the time. The expansion and controlled release of CO2 upon fractionation leaves the desired isolated product in several milliliters of methanol, thus decreasing evaporation time from greater than 8 h previously encountered with mass directed reversed phase HPLC to 1 hour. A paper that traces the economics and pressures involved in scale-up was published (37). As a new drug moves through the early stages of development and increasing amounts of material are needed, the initial chromatographic method can grow in scale with the needs of the project, often reaching the isolation of kilogram quantities for Phase 1 clinical trials. The focus at this point would be to find the most economical procedure for the production of the material in time for Phase II. The cost of chromatography for the first few grams of material will be very different from that of the first 10 t. The cost decreases quickly
with scale and with further optimization of the separation process. Chromatography was suggested to always be considered as one of the options for the manufacture of pure enantiomers. A critique of SFC, HPLC, and simulated moving bed (SMB) at the prep scale was provided. Several approaches for purifying difficult samples more efficiently for discovery research support were mentioned in another paper (38). Various specialty columns were employed such as hydrophilic interaction liquid chromatography (HILIC), hydrophilic end-capped columns in a reversed phase purification mode, and SFC on various chiral or bonded silica gel phases in a normal phase purification mode. Analytical method development strategies, sample preparation, scale-up from analytical to preparative, and purification results for various discovery samples were discussed. Difficult samples were defined as follows: (a) minimal retention on regular reversed phase columns due to high polarity, (b) minimal separation on reversed phase columns due to structurally similar impurities such as diastereomers and regioisomers, (c) requirement for large sample amounts for numerous injections, and (d) degradation in aqueous solutions. Preparative batch HPLC, SFC, SMB, and steady state recycling (SSR) were used to resolve a total of 12.2 kg of a racemic pharmaceutical intermediate (39). The separation conditions and results of these techniques were discussed. The productivities and solvent costs of SFC versus HPLC and then SMB and SSR versus HPLC were compared. Higher productivities and lower solvent usage were observed with SMB relative to SFC, SSR, and HPLC. SFC improved the separation efficiency when compared with batch HPLC. At 10% cosolvent, the solvent usage of SFC was reduced to half the amount compared with that of 20% cosolvent. However, the productivity was also reduced due to the longer cycle time. A review of SMB (i.e., continuous multicolumn chromatographic process) for the separation of enantiomers appeared (40). The review stressed design methods for robust operation, scaleup using data obtained from analytical experiments, process schemes, areas of design, and practical issues concerning the operation of SMB units. Another review presented the major developments and applications to those embarking on using SFC for high throughput applications in other fields (41). The semipreparative chiral separation of lansoprazole and two related compounds (pantoprazole and rabeprazole) using SFC was communicated (42). The volumes injected were 1, 2, and 4 mL. The concentrations of the racemic mixtures were 3 and 6 g/L for lansoprazole and 1.5 g/L for pantoprazole and rabeprazole. In all cases, the recoveries, for purity higher than 99.9%, were better for the second eluted enantiomer than for the first one. In the case of lansoprazole, it was possible to obtain 0.025 and 0.090 mg/ min of the first and second enantiomer, respectively, with an enantiomeric purity of 99.9%. A comprehensive approach was described to develop a chiral purification method for an analyte that was found to be unusually difficult to scale-up in SFC (43). Major factors were studied such as solubility of the analyte in the SFC mobile phase, impurity profiles, and cycle time. Solubility in the mobile phase was measured by a packed column technique coupled with a novel trapping mechanism to enhance measurement precision under SFC conditions. The SFC methods were developed to purify a sample containing 15% of an impurity. An equal volume mixture Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
4929
of acetonitrile and ethanol was chosen for the final purification method since this mixture demonstrated high SFC solubility among all solvent combinations with enhanced resolution between the analyte and the impurity and the shortest run time. The separation of the enantiomers of flurbiprofen on an amylase-derived chiral stationary phase, Chiralpak AD-H, by SFC under both linear and nonlinear conditions was studied (44). The primary objective of the work was to elucidate adsorption isotherms in SFC and study the effect of modifier concentration and pressure on the isotherms in order to apply this knowledge to process optimization. At linear conditions, the isotherm was determined directly from the chromatogram. Under overload conditions, the elution profiles were described by competing Langmuir and bi-Langmuir isotherms. The value of selectivity was from 1.9 to 2.1; while the value of resolution was from 5.3 to 11.8. The number of theoretical plates was always greater than 5000 indicating high efficiency of SFC. STATIONARY PHASES An overview on fluorocarbon stationary phases as alternative reversed phases was reported (45). Solvophobicity and fluorophilicity of the fluorinated phases provided enhanced selectivity for organofluorine compounds. The dual normal and reversed phase characteristics make fluorinated phases suitable for analysis of polar pharmaceutical and biological samples such as proteins, peptides, nucleotides, steroids, and alkaloids. Little work was stated to have been done in SFC. Another review traced the development and application of polysaccharide-based CSPs for the efficient separation of enantiomers (46). The chiral recognition mechanism of the polysaccharide-based CSPs was discussed. Current applications such as chiral separation via SFC, immobilization of polysaccharide derivatives, use of monolithic silica columns, and preparative separations were summarized. A series of papers appeared during the period authored by West and Lesellier. In their first paper, they evaluated specific retention properties of 28 different packed columns based on isocratic retention data using quantitative structure-retention relationships (QSRR) (47). The approach gives great promise of allowing the rapid selection of appropriate column types to produce the desired selectivity for a particular separation. In addition, it shows which columns would provide similar separations. The selectivity differences observed between stationary phases (SP) were essentially due to the stationary phase itself and not to the silica support or to differential adsorption of mobile phase components. The oversimplified classification scheme (i.e., reversed phase and normal phase) exists in LC because of the different polarities of SP and the necessary change in mobile phase (MP) required when working with liquids. This frontier does not exist in SFC. SFC provides analytical capability not readily accessible with HPLC using the same MP. Thus, it is possible to couple columns of widely differing polarity in SFC because the MP can be identical regardless of the polarity of the SP. Naming techniques are misleading and considering both modes as one with the SP as variable is a more correct approach. This becomes the basis of “unified chromatography from the SP perspective” as a complement to Chester’s “unified chromatography” from the MP perspective” In other words, normal phase and reversed phase 4930
Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
modes are limiting behaviors, and they are bridged in SFC as they may be bridged by emerging techniques in HPLC, such as HILIC. Later, it was investigated whether the number of SPs can be reduced to an optimized set with only highly orthogonal systems (48). The initial set of columns was defined with emphasis on efficient SPs covering a broad selectivity range. The following columns were picked: (a) a polar phase, 2-ethylpyridine bonded silica, (b) an aromatic phase of intermediate polarity, phenyloxypropyl bonded silica, (c) a nonpolar phase, octadecyl- and phenylpropyl-bonded silica, and (d) two unique phases butylsiloxane bonded silica and pentafluorophenyl- propyl-bonded silica. Further study (49) investigated the different chromatographic behaviors of a variety of octadecylsiloxane-bonded SPs: (a) classical, (b) protected against silanophilic interactions, and (c) containing polar groups (i.e., end-capping groups or embedded groups). Two approaches were chosen: carotenoid test and solvation parameter model. Both tests proved to be complementary and provided precise information on the chromatographic behavior of ODS phases. In a final paper, retention mechanisms in super/subcritical fluid chromatography on packed columns were reviewed (50). Whereas the retention rules of achiral compounds are well-defined in HPLC on the basis of the nature of the SP, some difficulties appear in SFC on packed columns. This is mainly due to the supposed effect of volatility on retention behaviors in SFC and to the nature of CO2 which is not polar; thus, SFC is classified as a normal phase separation technique. Moreover, additional effects related to density changes of the MP or to adsorption of fluid on the SP causing a modification of its surface are not well-known. In this paper, all these behaviors are discussed mainly using log k-log k plots which allow a simple comparison of SP properties. The synthesis of mono-6-(3-methylimidazolium)-6-deoxyperphenylcarbamoyl-β-cyclodextrin chloride (MPCCD) and its application in chiral SP for SFC was reported (51). The chiral selector was coated onto the silica gel in different weight percentages (15, 20, and 35% w/w) to obtain CSPs having different loading content. The best enantioseparation results were obtained using a column with 20% of MPCCD. The resolution obtained for p-halophenylethanol analytes ranged from 3.83 to 5.65. The first comparison (52) of a full screen-based approach using two series of commercially available Pirkle SPs with a polysaccharide-based SP screen (i.e., AD, OD, AS, OJ) has been reported. When running corresponding chromatography conditions, oftentimes Pirkle SPs can provide resolution opportunities similar to the polysaccharide phases prevalent in the pharmaceutical industry. Literature suggested an 80% success rate for polysaccharide phases, while in this study a 70% success rate was observed for Pirkle phases. An advantage of the additional capabilities and solvent choices (i.e., 1,2-dichloroethane, methylene chloride, chloroform, and THF) of Pirkle SPs was noted. Two commercially available Pirkle chiral phase platforms were chosen for comparison: Chirex Pirkle columns from Phenomenex and Pirkle columns from Regis. Poly(trans-1,2-cyclohexanediyl-bis-acrylamide) (P-CAP) with SFC was investigated via the analysis of 40 commercial and 100 proprietary compounds using a 12 min gradient with methanol as a modifier (53). P-CAP demonstrated separation of 25% of the
140 compounds, while each of the polysaccharide phases (AD-H, AS-H, OD-H, OJ-H) separated 46%. A key advantage of P-CAP is the fact that it is available in both enantiomeric forms, allowing manipulation of the elution order of enantiomers which is helpful for preparative applications. P-CAP, however, showed less chiral discrimination power compared to the derivatized polysaccharidebase CSPs. Four cationic β-cyclodextrin derivatives have been synthesized and physically coated onto porous spherical silica gel to obtain a novel chiral SP (54). The performance of these columns was studied using 18 racemic aryl alcohols as test analytes. Mono-6(3-octylimidazolium)-6-deoxyperphenylcarbamoyl-B-cyclodextrin chloride (OPCCD) showed the best separation results for all the analytes via both HPLC and SFC. Chromatographic studies revealed that CSPs consisting of an n-octyl group on the imidazolium moiety and phenylcarbamoyl groups on the CD ring provided enhancement of analyte-chiral substrate interactions over CSPs bearing a methyl group on the imidazolium moiety and 3,5-dimethylphenylcarbamoyl groups on the CD ring. The applicability of monolithic silica columns versus conventional columns packed with spherical particles for the analysis of complex hydrophobic metabolites was reported (55). The structural isomers of carotenoids were separated (monolithic ODS column (100 × 4.6 mm) for 10 min, seven carotenoids, methanol 1-5% in 8 min, 0.1% ammonium formate, inlet pressure as low as 15 mPa, outlet pressure 10 mPa, and flow rate of 3 mL/min). A method for profiling biological samples containing complex matrixes was also developed. Two new synthetic polymeric chiral stationary phases based on trans-(1S,2S)-cyclohexanedicarboxylic acid bis-4-vinylphenylamide and trans-N,N-(1R,2R)-cyclohexanediyl-bis-4-ethenylbenzamide monomers were prepared and evaluated by HPLC and SFC (56). A variety of chiral compounds were separated on these two CSPs. The different orientation of the amide groups in the two CSPs resulted in a striking difference in the enantioselective properties. Resolutions generally were higher with HPLC than with SFC; while SFC provided faster separations. Another study concerned the synthesis of monosubstituted positively charged cyclodextrins and their application to SFC in chiral separations (57). An ionic liquid (IL)-functionalized SP based on 1-octyl-3propylimidazolium chloride covalently bonded to silica gel was communicated (58). A 3.2 mm × 250 mm column for the simultaneous separation of acidic, basic, and neutral compounds using CO2 was considered. The effects of pressure, temperature, cosolvent, and additive on the retention behavior of the analytes were studied. Simultaneous separation of acidic, basic, and neutral compounds via SFC was successful at a cosolvent content of 20% methanol, a pressure of 110 bar, and a column temperature of 35 °C. RSDs of the retention times and peak areas at 50 ppm were all less than 4% and 8% (n ) 6), respectively. METHOD DEVELOPMENT The separation of phenacyl esters of the fatty acids originating from a fish oil extract by means of a comprehensive analysis using silver ion (SI) SFC and reversed phase liquid chromatography in the first and second dimensions, respectively, is described (59). The construction of the SI-SFC × RP-LC interface consisted of
two 2-position/ten port switching valves, of which one was equipped with two loops packed with ODS particles. In the SFC dimension, high efficiency and loadability were obtained by coupling two wide-bore columns (4.6 mm id) in series. Evaporative light scattering detection (ELSD) and UV detection with standard and high pressure flow cells were evaluated in terms of data acquisition speed and suppression of signal interferences originating from supercritical CO2 expansion. A quantitative structure retention relationship model to predict SFC retention of some organic compounds in various percents of organic modifiers in the MP using linear and nonlinear feature mapping techniques was published (60). The data set contained retention information for 35 various organic compounds in a MP which contained 0, 2, 4, and 6% methanol. Prediction of SFC retention based upon the solvation parameter model was attempted, given only the knowledge of the solute structures without any prior experiment (61). The Derringer desirability function proved to be useful for ranking chromatographic systems according to their separation performance. Even if retention factors were not exactly matched, elution order and separation factors were correctly predicted within 10%. The capacity of the method to determine the column with the largest probability to succeed in a given separation was suggested to have been proven. Chlorotriazine pesticides were separated. The effects of particle size and thermal insulation on retention and efficiency in packed column SFC with large pressure drops was described for the separation of a series of model n-alkanes (62). The three principle causes of band broadening were stated to be (a) the normal dispersion processes described by the van Deemter equation, (b) changes in the retention factor due to axial density gradient, and (c) radial temperature gradients associated with expansion of the MP. Positive effects were attributed to the elimination of radial temperature gradients and the concurrent enhancement of the axial temperature gradient. Thermal insulation had no significant effect on chromatographic performance at the higher density. The possibility for achieving high resolution separations concurrently with HPLC equivalent analysis times through serial coupling of five columns was presented (100 000 plates) (63). A highly effective screening method for a set of 17 structurally and physico-chemically diverse pharmaceutical standards was obtained in less than 20 min with optimum results obtained on five serially connectedcyanopropylsilicacolumnswith(a)acetonitrile-methanol (1:3) modifier, (b) diisopropylamine and trifluoroacetic acid additives (0.5% each), and (c) 100 bar outlet pressure. The effect of temperature on both efficiency and selectivity was significant and even amplified when a long column setup was used. The effect of increased concentrations of ammonium acetate (AA) in SFC was studied on silica, 2-ethylpyridine, and end-capped 2-EP phases (64). The study involved the addition of increasing concentrations of AA either in the MP modifier or in the sample solvent. Compounds that exhibited satisfactory chromatographic behavior in the absence of the additive were virtually unaffected by its presence in the MP or sample solvent. Compounds that exhibited late elution and strong peak tailing when pure methanol was used showed dramatically improved chromatographic behavior in the presence of the additive. Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
4931
The retention behaviors of pharmaceutical and druglike compounds in SFC with polar SPs were evaluated using linear solvation energy relationships (LSER) (65). More than 200 compounds were used. The dominant contribution to positive retention was the hydrogen bond donor acidity of the solutes particularly for pyridine and amino columns. Another significant contributor to retention behavior was the hydrogen bond acceptor basicity of the solutes made particularly for diol and amino columns. The LSER results showed that the SFC retention behavior of compounds using CO2-MeOH with polar SP was close to that reported for normal phase HPLC using hexanemethanol. A greater knowledge of the nature of solute/SP and solute/MP interaction allows a more rapid and efficient choice in high throughput screening of compound libraries. In a related study, the connection between the observable output in column chromatography (retention time, retention volume, retention factor, separation factor, etc.) and system properties (hold-up volume, pressure, temperature, isotherm behavior, etc.) was discussed from a practical and mechanistic perspective for SFC and other chromatographies (66). The unifying feature of these techniques was that retention can be described by a partition model, although not always exclusively. Applications and prospects of two phase tunable solvent systems composed of ionic liquids and supercritical fluids with an emphasis on supercritical CO2 were reviewed (67). SFC has emerged as a powerful technique to study the interphase distribution of highly dilute solutes. A major issue to be resolved in the future is the reliability of SFC data, namely the role of spurious retention mechanisms and the ensuing systematic errors in the resultant partition coefficients. An overview of methods for predictive thermodynamic modeling of binary (IL-CO2) and ternary (solute-IL-CO2) systems was included. Several papers discussed very unusual experimentation. Enantiomeric resolution by chiral SFC of a series of neutral iridium complexes with greater than 95% enantiomeric purity was communicated (68). Water has been used as a polar modifier, and µ-Porasil has been used as a saturator column (69). The µ-Porasil column was inserted between the pump outlet and the injection valve. During the passage of the supercritical fluid MP through the silica column, water could be dissolved in the pressurized fluid. Under these conditions, PEG was separated via SFC. More peaks were separated with the saturator column than with pure CO2. These results agreed well with those reported in the case of PDMS. A Nucleosil packed diol column was used (i.e., 100 × 2 mm id). The application of SFC to the analysis of naturally occurring polyprenols and lipid mixtures has been documented as an example for the separation of hydrophobic metabolites (70). Under optimized SFC conditions, individual homologues with 10-100 mers were separated. When a cyanopropylated silica gel packed column was used, the separation of 14 lipids was successfully detected, and the time required for analysis was less than 15 min. 2-Propanol and methanol were compared as alcohol/water modifiers of supercritical CO2 in packed column SFC (71). 2-Propanol was found to allow 5 times more MP water content in SFC than methanol. Of note, a 1:1 2-propanol/water ratio was attainable relative to a maximum 9:1 ratio realized for 4932
Analytical Chemistry, Vol. 82, No. 12, June 15, 2010
methanol/water. In test separations of polar analytes, the methanol system easily eluted compounds from a nonpolar PRP-1 column but failed to elute the most polar analytes from a polar silica gel or diol column. By comparison, 2-propanol systems could elute all of the analytes from each of these columns. Thus, 2-propanol can assist in increasing the water capacity of supercritical CO2 and potentially facilitate the analysis of various polar hydrophilic analytes. Isolation, fractionation, and identification of sucrose esters from aged oriental tobacco employing supercritical fluids have been completed (72). In addition to supercritical fluid extraction, semipreparative SFC provided for an additional purification of the sucrose ester enriched fraction after column optimization. Structural assignments of the SFC fractions were facilitated using GC/ MS accompanied by N,O-bis(trimethylsilyl)trifluoroacetamidedimethylformamide derivatization of the free hydroxyl groups and HPLC-MS. The path forward for validation in analytical SFC was discussed (73). Few SFC instruments can be found in drug development laboratories where current Good Manufacturing Practice (cGMP) is followed. Drug discovery group emphasis is on isolating the individual chiral compound in the enantiomeric mixture. Drug development group emphasis is on quantifying the enantiomeric impurity in the chiral active. Method validation, method transfer, and instrument qualification, etc. must be addressed. Certain improvements concerning instrument sensitivity, chromatographic data system and system qualification were touted to be necessary for SFC to fulfill the analytical needs in drug development laboratories. Separation of furocoumarins has become a great interest for the cosmetic industry and human health. Due to the variety of compounds and their subtle structural differences, their separation requires high performance methods. Isocratic conditions were established to obtain a satisfactory separation in 10 min (74). A two-dimensional analysis was also investigated by performing first a class fractionation of compounds on a 2-ethyl pyridine phase, then by separating each class on a pentafluorophenyl phase with the selected isocratic MP. These approaches allowed the structure of the compound to be related to retention behavior which was unusual due to the selected pentafluorophenyl SP. A study aimed to develop a generic gradient method and validate it, for a pharmaceutically relevant application was the goal of this publication (75). A secondary goal was the determination of thiourea in a pharmaceutical intermediate at low level (0.01% w/w). The applied method validation acceptance criteria were consistent with those used for Active Pharmaceutical Ingredient late stage development and regulatory submission. This suggested that with some further research and instrumental developments, SFC could potentially reach the same maturity as GC and LC. Generic conditions were CO2/MeOH containing 20 mM ammonium acetate (AA). The modifier was programmed from 5%, held for 1 min, to 40% at 2%/min. The pressure was 150 bar; the temperature was 40 °C, and the flow rate was 2.0 mL/min. The advantage of AA as an additive was that this is the preferred volatile salt for SFC/MS. System suitability and precision based on six injections of a 7.5 µg/mL thiourea (0.05% w/w) solution showed an RSD of 3.13% for peak area
and of 0.09% for retention time. Late stage development and regulatory submission performance criteria were met. An achiral 22-component mix, representing wide diversity and previously found to be extremely difficult to resolve, was chromatographed between 20° and 70 °C in 10 °C increments using a fixed, doubling gradient (76). Between 17 and 22 maxima were found depending upon the temperature although baseline resolution of 22 peaks was not achieved in the 12 min of a fixed gradient. Overlapped trace peaks, representing
