Article pubs.acs.org/ac
Introducing Enantioselective Ultrahigh-Pressure Liquid Chromatography (eUHPLC): Theoretical Inspections and Ultrafast Separations on a New Sub-2-μm Whelk-O1 Stationary Phase Dorina Kotoni,† Alessia Ciogli,† Carmela Molinaro,† Ilaria D’Acquarica,† Jelena Kocergin,‡ Ted Szczerba,‡ Harald Ritchie,§ Claudio Villani,† and Francesco Gasparrini*,† †
Dipartimento di Chimica e Tecnologie del Farmaco, Sapienza Università di Roma P. le Aldo Moro 5, 00185 Roma, Italy Regis Technologies, Inc., 8210 Austin Avenue, Morton Grove, Illinois 60053, United States § Thermo Fisher Scientific, Tudor Road, Runcorn, WA7 1TA, United Kingdom ‡
S Supporting Information *
ABSTRACT: A new chiral stationary phase for ultrahigh-pressure liquid chromatography (UHPLC) applications was prepared by covalent attachment of the Whelk-O1 selector to spherical, high-surface-area 1.7-μm porous silica particles. Columns of varying dimensions (lengths of 50, 75, 100, and 150 mm and internal diameters of 3.0 or 4.6 mm) were packed and characterized in terms of permeability, efficiency, retention, and enantioselectivity, using both organic and water-rich mobile phases. A conventional HPLC Whelk-O1 column based on 5.0-μm porous silica particles and packed in a 250 mm × 4.6 mm column was used as a reference. Van Deemter curves, generated with low-molecular-weight solutes on a 100 mm × 4.6 mm column packed with the 1.7-μm particles, showed Hmin (μm) and μopt (mm/s) values of 4.10 and 5.22 under normal-phase and 3.74 and 4.34 under reversed-phase elution conditions. The flat C term of the van Deemter curves observed with the 1.7-μm particles allowed the use of higher-than-optimal flow rates without significant efficiency loss. Kinetic plots constructed from van Deemter data confirmed the ability of the column packed with the 1.7-μm particles to afford subminute separations with good efficiency and its superior performances in the high-speed regime, compared to the column packed with 5.0-μm particles. Resolutions in the time scale of seconds were obtained using a 50-mm-long column in the normal phase or polar organic mode. The intrinsic kinetic performances of 1.7-μm silica particles are retained in the Whelk-O1 chiral stationary phase, clearly demonstrating the potentials of enantioselective UHPLC in terms of high speed, throughput, and resolution.
I
have been reported in the literature,12−15 little has been done to consistently exploit sub-2-μm particles to enhance both speed analysis and column efficiency in the separation of chiral analytes, probably because small porous particles of high superficial area (necessary to obtain enantioselectivity similar to that of one of the HPLC materials) have only recently been introduced in the market. In 2010, our research group illustrated the transition of a brush-type chiral stationary phase (CSP), namely the DACH-DNB phase, from HPLC to UHPLC, demonstrating that small porous particles can be functionalized and successfully used for ultrafast separations.16 Since then, sub-1-μm mesoporous silica particles functionalized with a cyclodextrin derivative have been prepared by Ai et al.,17 while, very recently, Chankvetadze and co-workers presented some preliminary results on a polysaccharide-based CSP obtained by coating core−shell particles with a nominal diameter of 2.6 μm.18 Brush-type CSPs are prepared through totally synthetic procedures, which are easily reproducible, and
n the past decade, the combination of columns packed with small porous particles (sub-2-μm) and the use of ultrahighpressure liquid chromatography (UHPLC), commercially available since 2004, have become preponderant in both biomolecular and pharmaceutical analysis.1−10 Advantages of UHPLC are well-known and extensively described in numerous articles: essentially either ultrafast separations without loss in efficiency or higher efficiency without considerable speed gain (high-resolution UHPLC) can be achieved. UHPLC has opened completely new horizons, both in terms of column efficiency (with plates per meter values as high as 250 000−300 000) and analysis time (subminute separations). Furthermore, UHPLC has proven both cost saving (with regard to solvent consumption), as well as more environmentally friendly. Although important goals have been achieved in the ultrafast separation of achiral compounds, enantioselective LC remains solidly attached to 3- and 5-μm totally porous particles and back-pressure values in the HPLC range. Why should analysis of chiral molecules, a major field both in pharmaceutical analysis and in enantioselective synthesis,11 not benefit from this important instrumental and materials evolution in liquid chromatography? While some subminute chiral separations © 2012 American Chemical Society
Received: May 16, 2012 Accepted: June 24, 2012 Published: June 24, 2012 6805
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
diffusion (the A term), longitudinal diffusion (the B term), and mass transfer (the C term).
can become ideal candidates for the transition from enantioselective HPLC (e-HPLC) to enantioselective UHPLC (e-UHPLC), as they offer fast kinetics, essential for ultrafast separations. Among brush-type selectors, the WhelkO1 (see Figure 1) is probably the best known and one of the
H=A+
B + Cμ μ
(1)
For the construction of van Deemter plots, the interstitial linear velocity μinter was used: it can be obtained using eq 2, where μsf is the superficial linear velocity, calculated by dividing the flow rate (Φ) by the area of the section of the column across which μsf is calculated (πr2), and εe is the external porosity of the packed bed. μ Φ μinter = sf = 2 εe πr εe (2) The external porosity (εe) of the packed columns usually ranges between 0.37 and 0.4242 and can be determined through methods such as inverse size-exclusion chromatography (ISEC), total pore blocking measurements, and Hg-intrusion porosimetry. Besides van Deemter plots, the kinetic performance of different columns can be best evaluated using the kinetic plot method, which shows the highest plate number a column can achieve in the shortest time possible, while working at the maximum pressure of the system (ΔPmax).43,44 Dead time (t0) versus plate number (N) plots can be used to quickly estimate which column offers the fastest separation for a given efficiency or the highest N value that can be obtained in a determined analysis time. Other more complex kinetic plots can be used to correlate impedance (E) and N, or the column length and N. The following equations are useful for the conversion of experimentally determined μ0 (linear velocity of an unretained analyte) and H values into kinetic plots, where ΔPmax is the maximum backpressure of the system, η the viscosity of the mobile phase, and Ks the specific permeability of the column.
Figure 1. Structure of (R,R)-Whelk-O1 chiral stationary phase (CSP).
most widely used in e-HPLC. It was originally designed in the early 1990s for the separation of the enantiomers of naproxen,19,20 a 2-aryl propionic nonsteroidal anti-inflammatory drug (NSAID), which was commercialized as a single enantiomer.21 Later, the Whelk-O1 phase proved a broadspectrum CSP for the separation of compounds bearing an aromatic system with a hydrogen-bond acceptor group located near the stereogenic center.22 It has been, thereafter, used in the last 20 years for the separation of many chiral compounds whose structure fits the general mechanism of separation, including alcohols, amides, esters, ethers, epoxides, carbamates, aldehydes, ketones, carboxylic acids, aziridines, phosphonates, and ureas.23−33 The phase is commonly used under normalphase (NP) conditions but interesting examples have been reported with water-rich mobile phases,22,34 as well as in polar organic mode (POM).35 Further applications include the use of supercritical/subcritical CO2 solvent,36−38 while its high loading capacity has been exploited in preparative LC.39,40 The WhelkO1 selector has also been used in Inverted Chirality Columns Approach to determine the enantiomeric excess of (S)namitecan (a water-soluble camptothecin derivative) in the absence of the minor enantiomer or the racemate as reference material.41 Our goal was to prepare a highly stable and efficient brushtype CSP of wide applicability using sub-2-μm totally porous particles, which hopefully will introduce enantioselective UHPLC (e-UHPLC) in the contemporary analytical laboratory. In our opinion, brush-type CSPs are ideal candidates for e-UHPLC, because of the fast kinetics of the separation mechanism involved. To the purpose, the broad spectrum Whelk-O1 selector, shown in Figure 1, was chosen and a new, totally porous, high-surface-area 1.7-μm CSP was prepared, which was extensively characterized for its thermodynamic and kinetic performances.
N=
t0 =
ΔPmax ⎡ K s ⎤ ⎢ ⎥ η ⎢⎣ μ0 H ⎥⎦ ΔPmax ⎡ K s ⎤ ⎢ ⎥ η ⎢⎣ μ0 2 ⎥⎦
experimental
experimental
(3)
(4)
The ΔPmax values were set at 400 bar for the HPLC system and 1000 bar for the UHPLC system, while μ0 was calculated from the column dead time t0,c, corrected for the extra-column dwell time, according to eq 5:
μ0 =
L t0, c
(5)
where L is the column length.
■
EXPERIMENTAL SECTION Chemicals and Columns. Unless otherwise noted, all reagents and analytical-grade solvents were purchased from Sigma−Aldrich (Milano, Italy) and used without further purification. HPLC gradient-grade solvents were further filtered on 0.2-μm Omnipore filters (Merck Millipore, Darmstadt, Germany), prior to use in the UHPLC system. Polystyrene standards of HPLC grade were purchased from Fluka (Buchs, Switzerland). Chiral samples were available from previous studies or were provided by Regis Technologies, Inc. (Morton Grove, IL, USA). Syncronis silica 1.7 μm (pore size = 120 Å,
■
THEORY Van Deemter plots correlating the lowest achievable plate height Hmin with the corresponding optimal linear velocity (μopt) are typically used to assess the efficiency of columns. They are based on the van Deemter equation (eq 1), where the plate height H is the sum of three different factors: eddy 6806
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
particle size = 1.7 μm, and specific surface area = 320 m2 g−1) was a gift from Thermo Scientific (Waltham, MA, USA), while the (R,R)-Whelk-O1 selector, together with an HPLC (R,R)Whelk-O1 analytical column (250 mm × 4.6 mm ID), were donated by Regis Technologies, Inc. Different column geometries were used in this study, with the column length being either 50, 75, 100, or 150 mm, while the internal diameter was either 3.0 or 4.6 mm. Stainless steel columns (IsoBar Systems purchased by Idex, Wertheim-Mondfeld, Germany) were packed using an acetone slurry composition (10% w/v), a 900-bar packing pressure, and flushing with hexane. End frits of 0.5 μm were used. Hold-up volumes (Vpyc 0 ) were determined by static pycnometry: wCHCl3 − wTHF V0pyc = ρCHCl − ρTHF (6)
commercial (R,R) denomination for clarity reasons. FT-IR (KBr): 2940, 2850, 1675, 1629, 1548, 1513, 1348, 1085 cm−1. The surface coverage of the silica gel with the selector, calculated on the nitrogen content, was 286 μmol g−1 silica (260 μmol g−1 of matrix). The derivatization of the silica was carried out in a laboratory-modified rotavapor (Büchi, Flawil, Switzerland), in which the reaction flask was fitted with a solvent condenser, a solvent collector, and an argon inlet. Mixing was obtained by spinning the flask around its axis, providing an even dispersion, limiting such phenomena as perturbation of particle aggregation and/or particle breaking, which become fairly frequent with sub-2-μm particles. Methodology. The van Deemter equation (eq 1) was used to fit the experimental data, allowing us to compare the efficiency of the new UHPLC-Whelk-O1 columns with the commercially available 5-μm HPLC Whelk-O1 column. Data fitting of the van Deemter curves was performed using Origin 6.0 software. Van Deemter plots were produced via inspection of the column efficiencies for a mixture of achiral test compounds at different flow rates both under normal-phase (NP) and reversed-phase (RP) conditions. The data obtained were not corrected for the extra-column peak broadening. In NP mode, a test mixture consisting of naphthalene and nitrobenzene was eluted through the columns using a mobile phase of n-hexane/CHCl3 (ethanol stabilized) 9:1, v/v (viscosity of the mobile phase given as η = 0.43 cP at 25 °C)50 and ultraviolet (UV) detection at 254 nm. In RP mode, ethylbenzene was eluted with a mobile phase of MeCN/H2O (6:4), v/v (viscosity of the mobile phase given as η = 0.75 cP at 25 °C)51 and UV detection at 214 nm. The temperature of the column was set at 25 °C in all experiments. Van Deemter curves were based on naphthalene (k = 0.82 on the UHPLC columns and k = 1.09 on the HPLC commercial column) and nitrobenzene (k = 1.87 on the UHPLC columns and k = 2.81 on the HPLC commercial column) in NP mode and on ethylbenzene (k = 3.12 on the UHPLC columns and k = 2.38 on the HPLC commercial column) under RP conditions. The number of theoretical plates (N) was calculated for every sample, according to the European Pharmacopeia, using the peak width at half height as implemented in the Chromeleon 6.8 software (Dionex, Sunnyvale, CA). An average of three measurements was used for each determination. The interstitial linear velocity (μinter) was calculated using eq 2, as previously described (after correction for the extra-column dwell-time). Recorded pressure drops (ΔPc, obtained by subtracting the pressure drop in the connecting tubing, ΔPext, from the total pressure drop) and the corresponding superficial linear velocity (μsf) were used to determine specific permeability (Ks) values for each column. The specific permeability Ks [m2] was calculated according to Darcy’s equation (eq 8):
3
where w and ρ are the mass of the column and solvent density, respectively.45,46 [Caution: CHCl3 and THF are toxic and care should be exercised to avoid exposure and inhalation.] An ISEC analysis was performed to determine the external porosity (εe) of both the 1.7-μm and 5-μm columns. A wide range of polystyrene standards (MW = 500 ÷ 3.6 × 106 Dalton) was injected into the columns, using neat THF as a mobile phase, according to an already reported procedure.47 The analysis yielded an εe value of 0.38 in all cases. The total porosity (εt) of the columns was also calculated according to the correlation εt = Vpyc 0 /Vg, where Vg is the geometrical volume of the column, considered as a cylindrical tube of radius r and long L, given by eq 7: Vg = Lπr 2
(7)
Instrumentation. The chromatographic system used was an UltiMate 3000 UHPLC system from Dionex (Sunnyvale, CA, USA), consisting of a dual-gradient RS pump (pressure up to 1034 bar under reversed-phase conditions, up to 800 bar under normal-phase conditions; flow rates up to 8.0 mL/min), an in-line split loop Well Plate Sampler, a thermostatted RS Column Compartment (temperature range = 5−110 °C), and DAD detector with a 2.5-μL flow cell. The DAD was set at a filter time constant of 0.002 s, a data collection rate of 100 Hz, and a response time of 0.025 s. Viper capillaries and fittings were used, with the two capillary Viper tubes (350 mm × 0.13 mm ID), producing an extra-column volume of 9.3 μL. The total volume of the two connectors and of the detection cell amounted to 11.8 μL.48 Data acquisition and processing were performed with Chromeleon 6.8 software from Dionex. Detection of all tested analytes was carried out at two different wavelengths (214 and 254 nm). The injection volume ranged between 1 and 2 μL on the UHPLC 1.7-μm columns and between 5 and 10 μL on the HPLC column. The UHPLC system was carefully characterized before the analysis, yielding a total extra-column volume of 19 μL, obtained by injecting toluene and using a zero dead volume connector. The peak variance (second statistical moment calculated at the base of the peak, as implemented in the Chromeleon software) was 9.8 μL2 at 1.0 mL/min. Preparation of UHPLC-Whelk-O1 Stationary Phase. The UHPLC (R,R)-Whelk-O1 chiral stationary phase (CSP) was synthesized according to the procedure described by Pirkle in 1992,19,20 starting from the new 1.7-μm Syncronis matrix. The correct configuration of the final CSP (see Figure 1) is (3S,4R),49 but in this work, we decided to maintain the
⎛ ηL ⎞ ΔPc = ⎜ ⎟μsf ⎝ Ks ⎠
(8)
where L is the column length [mm], ΔPc is the pressure drop across the column [bar], and μsf is the superficial linear velocity [mm/s]. Plots of μsf vs ΔPc were also prepared and Ks was obtained from the slope value of such plots. The column permeability K0 values [m2] were calculated using eq 9: εt = 6807
Ks K0
(9) dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
Figure 2a and Table 1. The number of theoretical plates per meter was as high as 244 000 with the UHPLC column. Figure
K0 can also be obtained from Darcy’s equation by substituting μsf with μ0, considering the linear mobile phase velocity for an unretained compound (μ0 = μsf/εt). Plots of μ0 vs ΔPc were also designed. The data obtained from the permeability studies were then used to create kinetic plots. For the construction of kinetic plots, obtained in NP mode, viscosity values were considered constant in the range of pressure used, as few data are available in the literature on the variation of the viscosity of binary mixtures of apolar solvents. Kinetic plots were obtained using the data of the van Deemter analysis into the Kinetic Plot Analyzer software (version 6.7), provided by the Vrije Universiteit Brussel.44,52
■
RESULTS AND DISCUSSION Van Deemter Analysis. Many chiral HPLC columns present in the market give an excellent enantioselectivity at the expense of solvent choice. Sample solubility and its link to preparative separations can mean that a compromise must be reached between selectivity and solvent choice. Brush-type stationary phases are able to perform under NP, RP, and polar organic mode (POM) conditions, as well as in SFC, because of the fact that they have no solvent restrictions and no memory effect. When working in NP, which is characterized by lower mobile-phase viscosity, the columns can be operated at high speed rates (as much as 2−3 mL/min for the 4.6-mm-ID columns and 0.8−1.6 mL/min for the 3.0-mm ID columns) without exceeding ΔPc values of ∼700 bar. This pressure value is typically considered to be the limit dividing Rapid Speed LC (RSLC) from UHPLC (between 600 and 1000 bar). It should be considered, moreover, that most commercially available UHPLC instruments are not fully compatible with all apolar solvents used in NP-LC; in particular, very few allow the use of chlorinated solvents. Furthermore, the pressure limits declared by the manufacturer are often lower in NP-LC than in RP-LC. Finally, few UHPLC systems allow flow rates of >2 mL/min, which was necessary to perform a complete van Deemter analysis under NP conditions. Therefore, we were driven by multiple factors in the choice of the apparatus. In our experiments, we were able to use a maximum backpressure of 800 bar in NP mode, which proved more than acceptable for the kinetic evaluation of our columns, given the high permeability of the phase and the low mobile phase viscosity (if compared to RP). Considering the application field of the Whelk-O1 columns, their kinetic efficiency was tested both in NP and RP mode. The data reported are referred to a 100 mm × 4.6 mm ID UHPLC column and to a 250 mm × 4.6 mm ID commercial HPLC column. To perform the van Deemter analysis, an internal diameter of 4.6 mm was chosen, as the effect of extra-column volume of the system on peak broadening is less pronounced, if compared to the more typical formats of 3.0 mm or 2.1 mm. For the present work, the HETP (H) values were not corrected for the extra-column volume because, although theoretically correct, it would, nevertheless represent a different situation from the one experimentally observed in the laboratory. To the every-day analyst, the number of theoretical plates and the H value obtained experimentally are usually the parameters that matter the most. Initially, the performance of the UHPLC column was tested under NP conditions (using a mobile phase composed of n-hexane/CHCl3, 9:1, v/v): a Hmin of 4.10 μm was obtained for naphthalene (k = 0.82), at the relative μinter,opt of 5.22 mm/s (corresponding to a flow rate of 2 mL/min and a μsf of 2.01 mm/s for the 100 mm × 4.6 mm ID column), as shown in
Figure 2. (a) Experimental van Deemter plot of naphthalene under normal-phase (NP) conditions on the HPLC (R,R)-Whelk-O1 column packed with 5-μm particles (blue curve, ▲) and on the 100 mm × 4.6 mm ID UHPLC-(R,R)-Whelk-O1 column packed with 1.7-μm particles (green curve, ●). Mobile phase: n-hexane/CHCl3, 9:1 (v/ v); η = 0.43 × 10−3 Pa s; Tcol = 25 °C; UV detection at 254 nm. (b) Experimental van Deemter of ethylbenzene under reversed-phase (RP) conditions on HPLC (R,R)-Whelk-O1 column packed with 5μm particles (blue curve, ▲) and on the 100 mm × 4.6 mm ID UHPLC-(R,R)-Whelk-O1 column packed with 1.7-μm particles (green curve, ●). Mobile phase: MeCN/H2O, 6:4 (v/v); η = 0.75 × 10−3 Pa s; Tcol = 25 °C; UV detection at 214 nm.
Table 1. Experimental van Deemter Analysis Data in Normal-Phase (NP) and Reversed-Phase (RP) Mode for the UHPLC-(R,R)-Whelk-O1 (100 mm × 4.6 mm ID) and the HPLC (R,R)-Whelk-O1 (250 mm × 4.6 mm ID) Normal-Phase (NP) Mode column 1.7-μm UHPLC 5-μm HPLC
Reversed-Phase (RP) Mode
Hmin [μm]
N/m
μinter,opt [mm/s]
Hmin [μm]
N/m
μinter,opt [mm/s]
4.10
244 000
5.22
3.74
278 000
4.34
13.71
73 000
2.77
12.64
78 000
2.48
S-1 in the Supporting Information reports the van Deemter plots of nitrobenzene (k = 1.87) in NP mode. Van Deemter plots of the UHPLC columns having different geometries (see Figure S-2 in the Supporting Information) show a similar trend, with Hmin ranging between 4.01 μm (for column length, L = 15 cm) and 4.46 μm (L = 5 cm). The flat C-term observed on all UHPLC columns allowed the use of interstitial linear velocities higher than the optimum values (registered between 5.22 and 6.83 on all four columns) without significant loss in efficiency. It therefore becomes possible to perform subminute separations 6808
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
of enantiomeric pairs, consistently reducing analysis time in enantioselective LC, as it will be shown later in this work. The 5-μm column (250 mm × 4.6 mm ID, see Figure 2a) showed a Hmin value of 13.71 μm and an optimal linear velocity of 2.77 mm/s (corresponding to a flow rate of 1.05 mL/min), again in accordance with the theoretically expected values (Hmin ≅ 2dp) for fully porous particles. The optimum column impedance value (opt ) was also calculated according to eq 10, yielding a value of 3646 for the HPLC (R,R)-Whelk-O1 (250 mm × 4.6 mm ID) and 3179 for the UHPLC (R,R)-Whelk-O1 (100 mm × 3.0 mm ID). opt =
H2εt H2 = Ks K0
(10)
An even higher efficiency (N/m = 278 000, see Table 1) was obtained under reversed-phase conditions (mobile phase consisting of MeCN/H2O, 6:4, v/v), where ethylbenzene (k = 3.12 on the UHPLC columns and k = 2.38 on the HPLC commercial column) was used for the construction of van Deemter plots (see Figure 2b). The Hmin was as low as 3.74 μm (at a μinter,opt value of 4.34 mm/s, corresponding to a flow rate of ∼1.65 mL/min) on the UHPLC-(R,R)-Whelk-O1 column (100 mm × 4.6 mm ID, 1.7 μm), while the HPLC (R,R)Whelk-O1 column (250 mm × 4.6 mm ID, 5 μm) reached a minimum HETP of 12.64 μm at μinter,opt = 2.48 mm/s (flow rate = 0.94 mL/min). The expected increase of column efficiency with low-viscosity eluents (NP) was not observed in our case. However, it should be noted that HETP figures that were not corrected for extra-column effects were used, and that this can become a severe source of peak broadening for early eluting peaks. Thus, the different retentions of the solute probes used for the kinetic evaluation in the NP and RP modes could be responsible for the lower efficiency recorded under NP elution. Nonetheless, the higher viscosity of the water-rich eluent does account for the lower optimal interstitial linear velocity in RP mode. Finally, a 100 mm × 3.0 mm I.D. column was included in the van Deemter analysis (see Figure S-3 in the Supporting Information). However, because of the fact that no correction for the extra-column volume was used, the Hmin observed was slightly higher than that reported for the 4.6-mm-ID columns. In fact, the instrument peak variance becomes quite important for 3.0-mm-ID columns (and even more for 2.1-mm-ID columns). In particular, a Hmin value of 5.31 μm was obtained at μinter,opt = 5.11 mm/s (corresponding to a μsf value of 1.94 mm/s and a flow rate of 0.8 mL/min). Physical Characterization of Whelk-O1 Columns. In order to provide a full characterization of the newly prepared UHPLC-Whelk-O1 columns, the specific permeability (Ks) and the column permeability (K0) were determined using μ vs ΔPc linear plots, as shown in Figure 3. As is well-known from Darcy’s law, linear velocity (μsf) and ΔPc are correlated by the constant (ηL)/Ks; the specific permeability can thus be easily calculated from the slope of the linear plots (see Figure 3a).53 The same approach was used to determine the column permeability (K0), plotting μ0 against ΔPc values. In detail, for all 1.7-μm packed columns and for the commercially available analytical 5-μm packed column, the linear plots shown in Figure S-4 in the Supporting Information were determined at 25 °C in a wide range of flow rates. The calculated column permeability values of the UHPLC columns were between 0.502 × 10−14 m2 and 0.585 × 10−14 m2, while the specific
Figure 3. (a) Backpressure ΔPc vs μsf plot (blue diamond, ◆) and ΔPc vs μ0 plot (red square, ■) for the UHPLC-(R,R)-Whelk-O1 column (1.7 μm; 100 mm × 4.6 mm ID). (b) Backpressure ΔPc vs μsf plot (red square, ■) and ΔPc vs μ0 plot (blue diamond, ◆) for the HPLC (R,R)-Whelk-O1 column (5 μm; 250 mm × 4.6 mm ID). The permeability values were calculated from the slope of the trend lines.
permeability (Ks) range was between 0.292 × 10−14 m2 and 0.346 × 10−14 m2 (for complete data, see Figure S-4 and Table S-1 in the Supporting Information). A 10-fold reduction in permeability was, in fact, observed when comparing these values to those obtained for the analytical 5-μm column (see Figure 3b), coherently with the reduction in particle size. Finally, the total porosity of the columns was determined using eq 9. As listed in Table S-1, the same εt values were obtained for the 1.7-μm and 5-μm columns. These data were in excellent agreement with the porosity data obtained through ISEC analysis, where values of εt = 0.59 and 0.60 were obtained for the 1.7-μm and 5-μm columns, respectively (see Figure S-5 in the Supporting Information for the ISEC plot). Kinetic Plots of Whelk-O1 Columns. Besides van Deemter plots, the kinetic performance of different columns can be best evaluated using the kinetic plot method, which shows the highest plate number a column can achieve in the shortest time possible, while working at the maximum pressure of the system.44 The first kinetic plots present in the literature are reported as early as 1965 by Giddings,1 while the method was further developed by Guiochon54 and Knox.2 Later, Poppe introduced t0/N vs N instead of t0 vs N to obtain a clearer view on the C-term of the van Deemter equation and direct speedefficiency relationships.3 The basic kinetic plot is the one that correlates t0 vs N: starting from this basic plot, more complex plots can be devised. Kinetic plots are ideal to compare different columns and especially different analytical conditions (such as UHPLC vs HPLC or SFC or CEC, silica-based versus monolithic columns), which is nowadays becoming necessary, given the wide range of instrumentation, methodologies, and 6809
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
Figure 4. Kinetic plots showing a comparison of the two different particle sizes under NP conditions (mobile phase: n-hexane/CHCl3, 9:1 (v/v); η = 0.43 × 10−3 Pa s; Tcol = 25 °C; UV detection at 254 nm): (a) t0 vs N plot, (b) t0/N vs N plot, (c) L vs N plot, and (d) t0/N2 vs N plot at 400 bar for the e-HPLC column (5 μm; blue curve, ▲) and at 400 bar (red curve, ◆) and 1000 bar (green curve, ●) for the e-UHPLC 1.7-μm column.
000: however, the use of the UHPLC column is justified up to 100 000 plates/column (for ΔPmax = 1000 bar). For higher N values, the B term of the van Deemter equation becomes preponderant, as the flow rate diminishes. The 1.7-μm column is, in fact, at its best when working with high flow rates, because of the flat C-term (see Figures 2a and 2b). For lower flow rates, the 5-μm HPLC column becomes more convenient (higher N value reached in less time). In the same plot, the curve of the eUHPLC column operated at a ΔPmax value of 400 bar is also shown (red curve, ◆): the curve shifts to the left, as a lower number of theoretical plates (Nmax = 57 000) can be reached when operating a 1.7-μm column on a standard HPLC system. When ΔPmax is 400 bar, the use of the HPLC column becomes convenient for N > 35 000. Poppe plots (Figure 4b) for the two Whelk-O1 columns show the separation speed that can be obtained when operating at 400 bar and 1000 bar (only for the UHPLC column). The relevant part of the Poppe plot is the bottom left corner when considering fast separations: it is in this area of the plot, in fact, where the separation speed is at its maximum. Given the flat Cterm of the van Deemter plot (see Figure 2), the e-UHPLC columns provide a good efficiency even in subminute separations. L vs N plots only confirm the already-shown results. If an efficiency of 55 000 plates per column is desired, the analyst should either use a 20.2-cm-long UHPLC column, operated at 1000 bar, or, when working with an HPLC system, a column length of 49 cm for the 1.7-μm column and 82 cm for
support formats available in the market. Because efficiency data (theoretical plates number) are becoming ever more important, kinetic plots can be used as a powerful geometry-independent comparative tool, showing the kinetic potential of the different available solutions, becoming an important aid in the final choice. In enantioselective HPLC, given the limited range of analytical devices, kinetic plots are rarely used: considering the novelty of the silica support, kinetic plots would be important to better estimate the gain in terms of performance and analysis time obtained in the transition from the enantioselective HPLC (e-HPLC) to the e-UHPLC. Kinetic plots can be used to quickly estimate which column offers the fastest separation for a given efficiency or the highest N value that can be obtained in a determined analysis time. Retention time (t0) versus plate number (N) plots for the e-UHPLC (1.7 μm) and the e-HPLC column (5 μm) are shown in Figure 4, together with the Poppe plots, the impedance time tE vs N (tE = t0/N2) plots, and the Lcol vs N plots. Kinetic plots were prepared using linear velocities μ0 and H values of an unretained solute. The three kinetic curves in Figure 4a show the kinetic performance limit: the value needed will be experimentally reached in the shortest time possible using a data point (H vs μ0) on the kinetic plot curve, when considering a column operated at the maximum backpressure (ΔPmax, can be chosen at free will). In our case, maximum backpressure values of 400 and 1000 bar were selected for the HPLC and UHPLC systems, respectively. Note that the maximum N value on the e-UHPLC column is ∼145 6810
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
the 5-μm column. Finally, Figure 4d shows impedance time (tE) vs N plots, where the values on the x-axis (N-values) are in reversed order, thus resembling a van Deemter curve. For the eUHPLC column, the minimum corresponds to N = 51 200 plates (per column) for L = 21 cm at ΔPmax = 1000 bar. If an HPLC system is used (ΔPmax = 400 bar), the minimum for the e-UHPLC column corresponds to N = 20 500 plates (obtained with an 8-cm-long column). The HPLC column shows a minimum for a column length of 1.63 m, with N = 118 000. It is easily deducible that the kinetic plots of the e-UHPLC column are similar to those observed with other achiral UHPLC columns. The 5-μm column performs best only for N > 100 000, but it would require a column length of 3.5 m to achieve N ≈ 200 000, which is not practicable for most HPLC applications (see Figure 4c). Note that the same 1.7-μm column (100 mm × 4.6 mm ID), when used on an HPLC (see Figure 4, red curve) or on an UHPLC (Figure 4, green curve) instrument performs very differently, with considerably lower N values, which, in practical applications, results in a lower resolution of enantiomers. In fact, the low viscosity of the mobile phase in NP mode and the excellent permeability of the UHPLC column make the column compatible with a traditional HPLC instrument, but at the price of lower efficiency. Applications to Chiral Compounds. After studying the kinetic performance and the physical proprieties of the new CSP for e-UHPLC, this study focused on the evaluation of its enantioselective abilities. Chiral compounds, typically separated on the HPLC Whelk-O1 CSP, were separated under a variety of conditions, on the new UHPLC 1.7-μm column (see Table S-2 in the Supporting Information for complete analytical results obtained on the 75 mm × 3.0 mm ID column). A first screening was performed at the optimal linear velocity obtained through van Deemter analysis, that, for the 75 mm × 3.0 mm ID column, corresponds to a flow rate of 0.8 mL/min. As reported in Table S-2 in the Supporting Information, a wide range of compounds, including sulfoxides, agricultural compounds, alcohols, and nonsteroidal anti-inflammatory drugs, was resolved on the e-UHPLC column, with excellent resolution values ranging between 1.86 for fenoxaprop-ethyl and 16.83 for trans-stilbene oxide. Good retention and α values (with remarkable enantioselectivity shown toward benzoin, naproxen, and trans-stilbene oxide, in all cases, >2) were obtained, maintaining analysis time compatible with the UHPLC time scale (as typically observed with achiral compounds). Only a phosphine oxide compound was too strongly retained to allow analysis in less than 5 min. In most cases, run times were as low as 1.5−2.0 min: using a 5-cm-long column or higher flow rates yielded chiral separations in the time scale of seconds. As shown in Figure 5, acenaphthenol and benzoin were separated in NP mode (mobile phase: n-hexane/ iso-propanol; 7:3; v/v; flow rate = 2 mL/min; hold-up time: 15.4 s) in ∼50 s, with a resolution of 3.19 and 11.02, respectively (with α values of 1.25 for acenaphthenol and 2.06 for benzoin). Whelk-O1 phases are very often used to separate highly polar chiral compounds such as 2-aryl propionic acids used as anti-inflammatory drugs and normally commercialized as a single pure enantiomer. Racemic flurbiprofen (Figure 5c, α = 1.27) and ketoprofen (α = 1.25) were separated in less than 40 s in polar organic mode on the same 50 mm × 3.0 mm ID eUHPLC Whelk-O1 column at a flow rate of 2.5 mL/min (holdup time: 12.3 s), with resolution values of 2.32 and 2.18, respectively. The more-retained naproxen (Figure 5d) was
Figure 5. Examples of separation of chiral compounds on the UHPLC 1.7 μm (R,R)-Whelk-O1 column (50 mm × 4.6 mm ID): (a) acenaphthenol and (b) benzoin separated in NP mode (mobile phase: n-hexane/iso-propanol = 7:3 (v/v); flow rate = 2.0 mL/min; UV detection at 214 nm; Tcol = 25 °C); (c) Flurbiprofen and (d) Naproxen separated in POM (mobile phase: MeCN + 0.2% AcOH + 0.07% DEA (v/v/v); flow rate = 2.5 mL/min; UV detection at 254 nm; Tcol = 25 °C).
analyzed in ∼70 s with an α value of 2.26 and a resolution of 9.60. In all three cases, the mobile phase in polar organic mode consisted of acetonitrile +0.2% v/v of acetic acid +0.07% v/v of diethylamine (DEA), because these conditions proved rather advantageous for the separation of 2-aryl propionic acids, as recently illustrated by Dossou et al.55 Note that, using a lower flow rate (1.5 mL/min), the analysis time is almost doubled, while using a flow rate of 2.5 mL/min, the loss in resolution is only slight (−17%, passing from 11.56 to 9.60, as shown in Figure 6). Using a 50 mm × 4.6 mm ID column at higher flow rates, an even lower analysis time could be reached, arriving at a minimum of 33 s at a flow rate of 5 mL/min (Rs = 6.14). The loss in resolution observed was 46.9% passing from a flow rate of 1.5 mL/min (corresponding to the optimal linear velocity, as determined by van Deemter analysis in RP mode; see Figure 2 and Table 1) to a flow rate of 5 mL/min (see Figure 6), with a speed gain of almost 4 times (3.8). 6811
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
Figure 6. Ultrafast resolution in polar organic mode (POM) of the enantiomers of Naproxen on the UHPLC 1.7 μm (R,R)-Whelk-O1, 50 mm × 4.6 mm column at different flow rates (mobile phase: MeCN + 0.2% AcOH + 0.07% DEA (v/v/v); UV detection at 254 nm; Tcol = 25 °C).
Given the excellent kinetic performances of the column and the low backpressure of the system (determined by the high permeability of the phase, the low viscosity of the mobile phase in POM and the column geometry), high flow rates could be reached and very-high-speed chiral separations could be obtained. This is, to the best of our knowledge, the first time that a chiral compound has been separated so rapidly, while maintaining a good resolution and efficiency (N/m at 5 mL/ min, corresponding to a μ0 of 8.57 mm/s, was registered at 50 000). The high permeability of the e-UHPLC column and the low viscosity of the mobile phase in NP allow the use of traditional HPLC systems to perform ultrafast chiral separations. However, the use of the HPLC system is inconvenient, both in terms of efficiency and of throughput of the column. Finally, the HPLCWhelk-O1 was included in the screening of chiral compounds for a direct comparison with the new e-UHPLC columns. An example is reported in Figure S-6 in the Supporting Information, where a sulfoxide (1-chloro-4-(methylsulfinyl)benzene) was separated on four e-UHPLC columns (5, 7.5, 10, or 15 cm long) having an internal diameter of 4.6 mm, and on the e-HPLC column. Passing from the commercial 5-μm column (250 mm × 4.6 mm ID, red chromatogram in the center of Figure S-6 in the Supporting Information) to the 1.7μm UHPLC (50 mm × 4.6 mm ID, blue chromatogram, top
left of Figure S-6 in the Supporting Information), a factor of ∼15 speed gain was registered, with a loss of 33% in resolution. However, using a longer UHPLC column (150 mm × 4.6 mm ID), not only did the resolution grow (+5%) but the analysis time was four times lower, even considering the slightly lower α values obtained on the e-UHPLC columns. A similar behavior was also observed with other chiral molecules, both in NP and POM, as illustrated in Table S-3 in the Supporting Information.
■
CONCLUSIONS This paper reports the synthesis and evaluation of a broad spectrum brush-type chiral stationary phase for UHPLC applications. The well-known Whelk-O1 selector was covalently immobilized onto 1.7-μm high-surface-area porous spherical silica particles and the resulting CSP was packed in columns that were characterized using flow-pressure plots, inverse size exclusion chromatography (ISEC), van Deemter analysis, and kinetic plots. With achiral test solutes, 100 mm × 4.6 mm UHPLC columns generated 244 000 and 278 000 N/m under normal-phase (NP) and reversed-phase (RP) elution modes, respectively. Kinetic plot analysis for the UHPLC columns and for a conventional HPLC column packed with 5μm particles, clearly demonstrated the superior performances of the UHPLC columns in the fast separation regime, where high chromatographic efficiency was obtained for subminute 6812
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
Analytical Chemistry
Article
(23) Villani, C.; Pirkle, W. H. J. Chromatogr. 1995, 693, 63. (24) Villani, C.; Pirkle, W. H. Tetrahedron: Asymmetry 1995, 6, 27. (25) Casarini, D.; Lunazzi, L.; Alcaro, S.; Gasparrini, F.; Villani, C. J. Org. Chem. 1995, 60, 5515. (26) Pirkle, W. H.; Brice, L. J.; Caccamese, S.; Principato, G.; Failla, S. J. Chromatogr. A 1996, 721, 241. (27) Pirkle, W. H.; Brice, L. J.; Widlanksi, J.; Roestamadji, J. Tetrahedron: Asymmetry 1996, 7, 2173. (28) Pirkle, W. H.; Koscho, M. E.; Wu, Z. P. J. Chromatogr. A 1996, 726, 91. (29) Pirkle, W. H.; Spence, P. L. J. Chromatogr. A 1997, 775, 81. (30) Pirkle, W. H.; Gan, K. Z. J. Chromatogr. A 1997, 790, 65. (31) Pirkle, W. H.; Lee, C. J. Enantiomer 1997, 2, 423. (32) Welch, C. J.; Szczerba, T.; Perrin, S. R. J. Chromatogr. A 1997, 758, 93. (33) Magora, A.; Abu-Lafi, S.; Levin, S. J. Chromatogr. A 2000, 866, 183. (34) Kennedy, J. H. J. Chromatogr. A 1996, 725, 219. (35) Dungelova, J.; Lehotay, J.; Krupcik, J.; Cizmarik, J.; Armstrong, D. W. J. Sep. Sci. 2004, 27, 983. (36) Pirkle, W. H.; Brice, L. J.; Terfloth, G. J. J. Chromatogr. A 1996, 753, 109. (37) Oswald, P.; Desmet, K.; Sandra, J.; Krupcik, J.; Majek, P.; Armstrong, D. W. J. Chromatogr. B 2002, 776, 283. (38) Kraml, C. M.; Zhou, D. H.; Byrne, N.; McConnel, O. J. Chromatogr. A 2005, 1100, 108. (39) Blum, A. M.; Lynam, K. G.; Nicolas, E. C. Chirality 1994, 6, 302. (40) Hamper, B. C.; Dukesherer, D. R.; Moedritzer, K. E. C. J. Chromatogr. A 1994, 666, 479. (41) Badaloni, E.; Cabri, W.; Ciogli, A.; D’Acquarica, I.; Deias, R.; Gasparrini, F.; Giorgi, F.; Kotoni, D.; Villani, C. J. Chromatogr. A 2010, 1217, 1024. (42) Neue, U. D.; El Fallah, M. HPLC Columns: Theory, Technology, and Practice; Wiley−VCH: New York, 1997. (43) Neue, U. D. LC-GC Eur. 2009, 22, 570. (44) Desmet, G.; Clicq, D.; Gzil, P. Anal. Chem. 2005, 77, 4058. (45) Gritti, F.; Kazakevich, Y.; Guiochon, G. J. Chromatogr. A 2007, 1161, 157. (46) Gritti, F.; Kazakevich, Y.; Guiochon, G. J. Chromatogr. A 2007, 1169, 111. (47) Al-Bokari, M.; Cherrak, D.; Guiochon, G. J. Chromatogr. A 2002, 975, 275. (48) Gritti, F.; Guiochon, G. J. Chromatogr. A 2011, 1218, 907. (49) Caccamese, S.; Principato, G.; Chimirr, A.; Grasso, S. Tetrahedron: Asymmetry 1996, 7, 2577. (50) Wei, I. C.; Rowley, R. L. J. Chem. Eng. Data 1984, 29, 332. (51) Thompson, J. W.; Kaiser, T. J.; Jorgenson, J. W. J. Chromatogr. A 2006, 1134, 201. (52) Desmet, G.; Gzil, P.; Clicq, D. LC-GC Eur. 2005, 7, 403. (53) Cramers, C. A.; Rijks, J. A.; Schutjes, C. P. M. Chromatographia 1981, 14, 439. (54) Martin, M.; Guiochon, G. J. Chromatogr. 1974, 99, 357. (55) Dossou, K. S. S.; Farcas, E.; Servais, A. C.; Chiap, P.; Chankvetadze, B.; Crommen, J.; Fillet, M. J. Chromatogr. 2012, 1234, 56.
separations. Resolutions of the enantiomers of a broad set of compounds, including alcohols, polar sulfoxides and phosphine oxides, and acidic drugs, were accomplished on the new UHPLC chiral columns with analysis times as low as 30 s. The excellent kinetic performances of the UHPLC columns combined with the broad spectrum enantioselectivity and complete solvent tolerance of the Whelk-O1 CSP clearly showed the large potentials of UHPLC in the field of very fast, high-throughput enantioselective analysis. Hopefully, the present work will be a realistic starting point in the process aimed at finally introducing enantioselective UHPLC in the contemporary analytical laboratory.
■
ASSOCIATED CONTENT
* Supporting Information S
Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*Tel.: +39 06 4991 2776. Fax: +39 06 4991 2780. E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors are thankful for the financial aid supported by the MIUR, PRIN 2009 (Contract No. 2009ZSC5K2_001z) and Sapienza University (Contract No. C26A11RT5X).
■
REFERENCES
(1) Giddings, J. C. Anal. Chem. 1965, 37, 60. (2) Knox, J. H. J. Chromatogr. Sci. 1977, 15, 362. (3) Poppe, J. H. J. Chromatogr. A. 1997, 778, 3. (4) Wu, N.; Lippert, J. A.; Lee, J. J. Chromatogr. A. 2001, 911, 1. (5) Jerkovich, A. D.; Mellors, J. S.; Jorgenson, J. W. LC-GC Eur. 2003, 16, 20. (6) Majors, R. E. LC-GC N. Am. 2005, 23, 1248. (7) Majors, R. E. LC-GC N. Am. 2006, 24, 248. (8) Desmet, G.; Gzil, P.; Nguyen, D.T.-T.; Guillarme, D.; Rudaz, S.; Veuthey, J.-L.; Vervoort, N.; Torok, G.; Cabooter, D.; Clicq, D. Anal. Chem. 2006, 78, 2150. (9) Guillarme, D.; Nguyen, D.T.-T.; Rudaz, S.; Veuthey, J.-L. J. Chromatogr. A 2007, 1149, 20. (10) Carr, P. W.; Wang, X; Stoll, D. R. Anal. Chem. 2009, 81, 5342. (11) Ward, T. J.; Ward, K. D. Anal. Chem. 2012, 84, 626. (12) House, D. W.; Oroskar, A. A.; Oroskar, A. R. LC-GC N. Am. 2009, 9, 22. (13) House, D. W.; Oroskar, A. A. LC-GC N. Am. 2010, 9, 32. (14) Chankvetadze, B.; Yamamoto, C.; Okamoto, Y. Chem. Lett. 2003, 32 (9), 850. (15) Chankvetadze, B.; Yamamoto, C.; Kamigaito, M.; Tanaka, N.; Nakanishi, K.; Okamoto, Y. J. Chromatogr. A 2006, 1110, 46. (16) Cancelliere, G.; Ciogli, A.; D’Acquarica, I.; Gasparrini, F.; Kocergin, J.; Misiti, D.; Pierini, M.; Ritchie, H.; Simone, P.; Villani, C. J. Chromatogr. A 2010, 1217, 990. (17) Ai, F.; Li, L.; Ng, S.-C.; Tan, T. T. Y. J. Chromatogr. A 2010, 1217, 7502. (18) Lomsadze, K.; Jibuti, G.; Farkas, T.; Chankvetadze, B. J. Chromatogr. A 2012, 1234, 50. (19) Pirkle, W. H.; Welch, C. J.; Lamm, B. J. Org. Chem. 1992, 57, 3854. (20) Pirkle, W. H.; Welch, C. J. J. Liq. Chromatogr. 1992, 15, 1947. (21) Decamp, W. H. Chirality 1989, 1, 2. (22) Pirkle, W. H.; Welch, C. J. Tetrahedron: Asymmetry 1994, 5, 777. 6813
dx.doi.org/10.1021/ac301335b | Anal. Chem. 2012, 84, 6805−6813
