Comparison of Positively and Negatively Charged Achiral Co-Monomers Added to Cyclodextrin Monolith: Improved Chiral Separations in Capillary Electrochromatography

Abstract

Cyclodextrins (CDs) and their derivatives have been one of the most popular and successful chiral additives used in electrokinetic chromatography because of the presence of multiple chiral centers, which leads to multiple chiral interactions. However, there has been relatively less published work on the use of CDs as monolithic media for capillary electrochromatography (CEC). The goal of this study was to show how the addition of achiral co-monomer to a polymerizable CD such as glycidyl methacrylate β-cyclodextrin (GMA/β-CD) can affect the enantioselective separations in monolithic CEC. To achieve this goal, polymeric monoliths columns were prepared by co-polymerizing GMA/β-CD with cationic or anionic achiral co-monomers [(2-acrylamido-2-methyl-1-propanesulfonic acid (AMPS) and vinyl benzyltrimethyl-ammonium (VBTA)] in the presence of conventional crosslinker (ethylene dimethacrylate) and ternary porogen system including butanediol, propanol and water. A total of 34 negatively charged compounds, 30 positively charged compounds and 33 neutral compounds were screened to compare the enantioresolution capability on the GMA/β-CD, GMA/β-CD-VBTA and GMA/β-CD-AMPS monolithic columns.

Introduction

Chiral capillary electrochromatography (CCEC) like generic capillary electrochromatography (CEC) combines the excellent selectivity of high-performance liquid chromatography (HPLC) with the superior efficiency of capillary electrophoresis (CE). The use of chiral stationary phase (CSP) in various modes of CCEC have been an area of extensive study and can be broken into packed column CEC (p-CCEC), open-tubular CEC (o-CCEC) and monolithic rod-CCEC (1). Traditionally, p-CCEC requires frit burning (2, 3) or tapering of the frit at the outlet end of the capillary (4) to retain the packed particulate CSPs inside the capillary. In case of o-CCEC, the chiral selector is bonded or coated to the inner wall of the open tubular capillary column. Monolithic (5, 6) rod CCEC column is typically formed by co-polymerizing the chiral selector with a mixture of crosslinker and porogens. Other approaches include grafting by polymerization, physical adsorption and encapsulation of the chiral selector. The advantages and disadvantages of the three aforementioned modes of CCEC are described in a recently published review article (1).

Cyclodextrins (CDs) and its derivatives have been one of the most popular and successful types of CSPs used bonded phase HPLC (7). In CE, CDs are added to the CE buffer or immobilized as CSP (2, 3, 8–10). Because of the presence of multiple chiral centers, CDs are very adaptable to function as a successful CSP in the monolithic mode of CEC (11, 12). One of the approaches recently developed in our laboratory in advancing the development of rod-CCEC monolithic column is to synthesize polymerizable monomers of CDs (13). For example, in our previous work, prior to on-capillary polymerization, the primary hydroxyl group of β-CD was reacted with glycidyl methacrylate (GMA) groups providing a polymerizable chiral monomer (i.e. GMA-β-CD) (13). Because GMA-β-CD is essentially neutral, any negatively or positively charged co-monomers could be simply added to an appropriate crosslinker and porogens. The porogens act to dissolve the monomers and work to control the formation of pores throughout the monolithic material.

A major challenge in the development of chiral monolithic CEC column is the limited enantioselectivity. The purpose of this study was to investigate how the addition of charge achiral co-monomers influences the porosity and permeability of GMA/β-CD as well as improve its chiral separation ability when used as a monolithic CSP. The GMA/β-CD (without any charge co-monomer) was compared with GMA/β-CD-2-acrylamido-2-methylpropane sulfonic acid (AMPS) and GMA/β-CD-vinyl benzyltrimethyl-ammonium (VBTA) columns containing negatively and positively charged achiral co-monomers, respectively. In preparing the later two monolithic columns, the AMPS and VBTA comonomers were simply added to the polymerization mixture (Figure 1) providing cathodic and anodic electroosmotic flow (EOF), respectively. Using the aforementioned approach to prepare chiral monolithic columns provides improved resolution with high success rate for enantioseparation of large number of negatively charged analytes. Screening of 34 acidic, 30 basic and 33 neutral compounds was performed to verify the complementary enantioselectivity of GMA/β-CD-AMPS- and GMA/β-CD-VBTA-based monolithic CSPs.

Figure 1.

Polymerization scheme for the formation of the GMA/β-CD-VBTA (A) and the GMA/β-CD-AMPS (B) monoliths.

Experimental

Reagents and materials

The reagents ethylene dimethacrylate (EDMA), 1-propanol and 2,2′-azobisiso-butyronitrile (AIBN) were purchased from Aldrich (Milwaukee, WI, USA). Dimethyl sulfoxide (DMSO), β-cyclodextrin (β-CD), γ-methacryloxypropyltrimethoxysilane, AMPS, VBTA, 1,8-diazabicyclo-umdec-7-ene (DBU), HPLC grade acetonitrile (ACN), anhydrous N, N-dimethylformamide (DMF) and all the chiral analytes were obtained from Sigma (St. Louis, MO, USA). The porogen, 1,4-butanediol and monomer, and GMA were purchased from Fluka (Buchs, Switzerland). All the reagents were used as received, except for the EDMA, which was purified by distillation under vacuum prior to use. Fused silica capillary (OD 375 μm, ID 100 μm) was obtained from Polymicro Technologies, Inc. (Phoenix, AZ, USA).

Synthesis of GMA/β-CD

The GMA/β-CD was synthesized according to a previously published procedure (13). Briefly, 1.5 g β-CD (1.30 mmol), which had been dried previously in a vacuum oven for at least 48 h at 40°C, was added in a clean dry 50 mL round-bottom flask (RBF). About 28 mL hydrous DMF was added to dissolve β-CD. A moisture trap was placed on the top of flask and the solution was sonicated for 15 min to ensure that the entire β-CD is dissolved. Once the β-CD was dissolved, moisture trap was removed to quickly add 14.5 µL DBU, 1% (w/w) β-CD and 535 µL GMA in the RBF. Next, the moisture trap was reattached and the flask was placed in the oil bath to stir and heat overnight at 98°C. After overnight stirring, the golden brown solution was allowed to equilibrate at room temperature before filtering the solid precipitates. The filtrate was transferred into a 1000-mL beaker, containing 600 mL toluene, which formed white/brown precipitate. The precipitated solution was stirred gently for 10 min and was allowed to sit for an additional 10 min to maximize precipitation formation. The precipitate was washed with 400 mL of HPLC grade acetone after filtration. The solid product was washed and transfer to a weighing boat before drying in the vacuum oven overnight at 40°C. The GMA/β-CD material was then characterized by ¹H NMR spectroscopy performed with a 300 MHz Varian Unity spectrometer using deuterium oxide as a solvent (13).

Preparation of monolithic columns

First, the inner walls of the bare fused silica capillaries were vinylized with 3-(methacryloyloxy) propyltrimethoxysilane (γ-MAPS) using the procedure described elsewhere (13). The following pre-polymerization solution was used: (%w/w): 14.6% (75 mg) GMA/β-CD; 0.4% (2 mg) VBTA or AMPS; 5% (23.8 µL) EDMA; 30% (148 µL) 1,4-butanediol; 30% (136 µL) anhydrous DMSO; 20% (125 µL) AIBN. Subsequently, each pre-polymerization solution was mixed ultrasonically into a homogenous solution and purged with nitrogen for 10 min. Several bare-fused silica capillaries with a total length of 40 cm were filled with the polymerization mixture to a length of 28 cm, sealed with rubber septum and then placed in a GC oven to polymerize for 20 h at 60°C. After the polymerization was completed, all three versions of monolithic columns were washed with ACN using an HPLC pump to remove any unreacted monomers and porogens. On-column detection window was made next to the polymer bed using a thermal wire stripper. Finally, the column was cut to obtain a total length of 33.5 cm with an effective length of 25 cm.

Morphology, porosity and surface area measurements

A Hitachi X-650 (Hitachi, Japan) scanning electron microscope (SEM) was used to characterize the morphology of the monolithic columns. The SEM was operated at 7.5 kV; the filament current was set at 40 mA. Each monolithic column was cut to 2 mm in length and stuck on an aluminum stub by double-sided carbon tape. The column was then sputter-coated with gold/palladium alloy with an SPI sputter (SPI supplies Division of Structure Probe, West Chester, PA, USA) for 1 min at 30 mA to avoid charging using a gold/palladium alloy.

The porosity experiment was carried out on a micro-HPLC system and calculated by the flow method with 100% ACN as mobile phase. The linear velocity, u (m/s), of the mobile phase was measured by taking a ratio of column length and dead time (t₀) using DMSO as a marker. The porosity was calculated using the following equation:

$${ϵ}_{\mathrm{T}}=\frac{V}{\pi r^{2}u}\times 100\mathrm{\%}$$ (1)

where ε_T is the total porosity of the column; r (m) is the inner radius of the column; V (m³/s) is the volumetric flow rate. The flow rate was measured by collecting and weighing the mobile phase eluted out of the column in a certain amount of time.

The specific permeability, (K⁰), was determined by Darcy's law (13), which is directly proportional to the solvent viscosity and column porosity, and inversely proportional to back pressure (Pa). The K⁰ was calculated by the following equation:

$$K^0=\frac{u\eta L{ϵ}_{\mathrm{T}}}{\mathrm{\Delta }p}$$ (2)

In Equation (2), η (Pa s) is the dynamic viscosity of the mobile phase; L (m) is the effective length of the monolithic bed; Δp (Pa) is the pressure drop across the column, which is the pressure measured at various volumetric flow rates.

The surface area of the bulk monolithic was obtained by nitrogen adsorption experiments on a Micromeritics Tristar 3020 (Micromeritics Instrument, Norcross, GA, USA). The surface area was calculated via a multi-point BET method applied to nitrogen physisorption data. All samples were first heated under vacuum at 70°C for 24 h to remove physisorbed water before being analyzed.

Resolution (R_s) and efficiency (N) were calculated by the ChemStation software using peak width at half height method. Selectivity was measured as a ratio of capacity factor (k′) of the individual enantiomer. The k′ of each enantiomer was determined by the equation: k′ = (t_R− t₀)/t₀, where t_R and t₀ are the retention time and the dead time, respectively. DMSO was used to measure the t₀ value.

CEC instrumentation

The separation on all three monolithic columns were carried on an Agilent CE system (Agilent Technologies, Palo Alto, CA, USA) equipped with an auto-sampler, 0–30 kV power supply and a diode-array UV detector was used to carry out all the CEC experiments. Agilent 3D-CE ChemStation software (Rev. A. 08.04) was used for data acquisition and analysis. A series III isocratic HPLC pump (Lab Alliance, State College, PA, USA) was used to flush and condition the CEC columns.

Results and discussion

Characterization of the monolithic columns

The SEM images of the three monolithic columns are shown in Figure 2. From the three sets of SEM micrographs (Figure 2A1, B1 and C1), one can easily conclude that the monolithic material was successfully formed inside the capillaries. Looking at the SEM images, it appears that the most porous monolithic material is the one without any co-monomer (i.e. GMA/β-CD (Figure 2C2) monolithic material), which has a structure composed a lot of granular particles. On the other hand, GMA/β-CD-VBTA (Figure 2A2) and GMA/β-CD-AMPS (Figure 2B2) had the skeletal structures formed with larger size globular clusters, with more through pores on the monolithic matrix. As listed in Table I, there is very little difference in the porosity values between the three monolithic materials, which is similar to what we expect as all three monolithic column were prepared with 80% porogens and 20% monomers. However, the GMA/β-CD monolith provided a much higher permeability than the GMA/β-CD-AMPS and GMA/β-CD-VBTA materials. This trend is supported by the SEM of GMA/β-CD monolith (shown in Figure 2C2) exhibiting flow more through pores in the monolithic material. On the other hand, the GMA/β-CD-AMPS and GMA/β-CD-VBTA monolithic materials could only provide flow paths around but not through the particle clusters.

Figure 2.

Scanning electron micrographs of the three monolithic materials A1, A2: GMA/β-CD-VBTA; B1, B2: GMA/β-CD-AMPS; C1, C2: GMA/β-CD. A1, B1 and C1 are the edge sections of the monolith, whereas A2, B2 and C2 are the magnified center sections of the monolith. For conditions, see Experimental section. Inset green bar represents scale of 2 µm.

Table I.

Porosity and Permeability as Determined by the Micro HPLC Flow Method and BET Surface Area Measurements as Determined by Nitrogen Adsorption Experiments

	% Porosity	Permeability (m2)	BET surface area (m2/g)
GMA/β-CD-VBTA	83 ± 2	3.9 × 10−16 ± 2.1 × 10−17	15.7 ± 0.9
GMA/β-CD-AMPS	88 ± 2	3.7 × 10−16 ± 2.0 × 10−17	33.4 ± 0.9
GMA/β-CD	86 ± 3	1.6 × 10−15 ± 3.1 × 10−16	33.4 ± 1.1

The overlaid plots of the back pressure versus the volumetric flow rate observed on the three monolithic columns are shown in Figure 3. Clearly, at any given pressure, GMA/β-CD column provided the highest flow rate. On the other hand, the GMA/β-CD-AMPS and GMA/β-CD-VBTA columns provided very similar flow rate, which is consistent with the permeability data listed in Table I. In addition, the linearity of all three plots is very good, indicating that the mechanical stability of all three monolithic columns is excellent. The specific permeability of the three monolithic columns follows the order: GMA/β-CD > GMA/β-CD-AMPS–GMA/β-CD-VBTA.

Figure 3.

Pressure versus flow rate for the permeability study of the three monolithic materials GMA/β-CD (filled triangle), GMA/β-CD-VBTA (filled diamond) and GMA/β-CD-AMPS (filled squary). The inset R² values represent the linearity of the best-fit line.

Selective enantioseparations on GMA/β-CD, GMA/β-CD-AMPS and GMA/β-CD-VBTA monolithic column

Similar to CD bonded phases (7), one of the features of CD monolithic column, which could impart to its unique selectivity and influences retention, is the inclusion complexation. This complexation involves interaction of non-polar CD cavity with the non-polar portion of the analytes. In addition, the size and shape of analytes are also essential factors that may influence the retention. For example, large size chiral molecule will not fit in the CD cavity and the complexation will be very poor. Hydrogen bonding of polar molecules can also help with inclusion formation. Because aromatic groups can share electrons with the glycosidic oxygens of CD cavity, analytes with those groups can gain the stability of the inclusion complexation (14). By varying the charges of the co-monomer, selective enantioseparations could be achieved. In the following sections, selective enantioseparation capability of GMA/β-CD, GMA/β-CD-AMPS and GMA/β-CD-VBTA monolithic columns are discussed for various acidic basic and neutral chiral compounds.

Acidic compounds screening

The structures of 34 structurally diverse acidic compounds (A1–A34) screened are shown in Supplementary data, Figure S1, whereas the respective electrochromatographic data are listed in Table II. The enantioseparations of all 34 acidic compounds were compared on two monolithic CSP with positively charged (VBTA) and negatively charged (AMPS) achiral monomers. In addition, a monolithic column (i.e. GMA/β-CD) without any ionic co-monomer was also fabricated and compared. Note that in this study compound with a chiral resolution value (i.e. R_s > 0) is considered separated for which enantioselectivity was observed. Every hint of chiral resolution is therefore considered important because subsequent optimization could most likely provide improve resolution. Looking at the bar graph in Figure 4A, it can be seen that the GMA/β-CD-VBTA monolithic column was able to separate 23 of the 34 anionic compounds (with a success rate of ∼68%) under the mobile phase composition: 75% ACN/25% water, 5 mm NH₄COOH, at pH 3.0. The aforementioned mobile phase conditions provided chiral R_s values of ≥1.5 for 10 acidic compounds, another five provided R_s between 1.0 and 1.5, whereas the remaining eight compounds showed resolutions between 0.1 and 1.0. In contrast, under the same mobile phase conditions, the GMA/β-CD column was unable to resolve the same set of compounds, whereas the GMA/β-CD-AMPS column provided slight hint for only one acidic compound (A12, 4-bromo-mandelic acid, R_s ∼0.1).

Table II.

Enantioselective Screening of 34 Acidic Compounds Using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS Columns

	Acidic analytes	GMA/β-CD-VBTA			GMA/β-CD			GMA/β-CD-AMPS
		Rs	k′	α	Rs	k′	α	Rs	k′	α
	Phenoxy propionic acids and analogs
A1	2-Chlorophenoxypropionic acid	1.5	6.31	1.06	0	5.92	1	0	8.88	1
A2	2,2(Chlorophenoxy)propionic acid	0	2.17	0	0	1.79	1	0	2.51	1
A3	2,3(Chlorophenoxy)propionic acid	1.5	8.9	1.04	0	8.49	1	0	12.86	1
A4	2,4(Chlorophenoxy)propionic acid	1.3	11.6	1.07	0	11.15	1	0	16.96	1
A5	2,2,4(Dichlorophenoxy)propionic acid	0.4	5.67	1.02	0	5.28	1	0	7.89	1
A6	2,2,4,5(Trichlorophenoxy)propionic acid	0.3	15	1.01	0	14.6	1	0	12	1
A7	2-Phenylbutyric acid	0	0.79	0	0	0.42	1	0	0.39	1
	α-Hydroxy acids and analogs
A8	Mandelic acid	0.1	4.35	1.01	0	3.97	1	0	5.87	1
A9	Acetylmandelic acid	0.5	6.13	1.03	0	5.73	1	0	8.59	1
A10	Atrolactic acid	2.1	12.6	1.07	0	12.15	1	0	18.5	1
A11	4-Chloro-mandelic acid	1	16.3	1.03	0	15.88	1	0	13.1	1
A12	4-Bromo-mandelic acid	0	0.58	1	0	0.21	1	0.1	0.07	1.02
A13	α-Bromophenyl acetic acid	0.5	6.65	1.03	0	6.25	1	0	7.18	1
A14	β-Phenyl lactic acid	0.4	11.3	1.02	0	10.88	1	0	12.81	1
A15	α-Methoxyphenyl acetic acid	1.3	11.4	1.04	0	10.94	1	0	12.89	1
A16	3,4(Methylenedioxy) mandelic acid	1.1	12.4	1.05	0	11.96	1	0	14.12	1
	Dinitro amino acids
A-17	2,4-DNP-norleucine	0.9	8.52	1.06	0	8.12	1	0	12.28	1
A-18	2,4 -DNP-norvaline	2.4	7	1.17	0	6.6	1	0	9.94	1
A-19	2,4-DNP-threonine	2.9	6.63	1.11	0	6.23	1	0	9.36	1
A-20	3,5-(Dinitrobenzoyl)phenylglycine	0.5	5.04	1.03	0	4.65	1	0	6.93	1
A-21	3,5-(Dinitrobenzoyl)leucine	1.5	10.7	1.09	0	10.32	1	0	12.13	1
A-22	2,4-DNP-methionine sulfone	1.1	4.08	1.04	0	3.7	1	0	5.45	1
	Profens
A-23	Ibuprofen	0.5	5.85	1.02	0	5.46	1	0	8.18	1
A-24	Indoprofen	0	1.94	1	0	1.56	1	0	2.15	1
A-25	Flurbiprofen	0	3.56	1	0	3.18	1	0	4.65	1
A-26	Suprofen	0	4.19	1	0	3.8	1	0	5.62	1
A-27	Fenoprofen	0	2.56	1	0	2.18	1	0	3.12	1
A-28	Ketoprofen	0	2.79	1	0	2.41	1	0	3.47	1
A-29	Carprofen	0	3.9	1	0	3.51	1	0	5.17	1
	Dansylated amino acid
A-30	DNS-phenylalanine	0	0.29	1	0	0.05	1	0	0.11	1
A-31	DNS-serine	2.2	10.6	1.14	0	10.15	1	0	15.42	1
A-32	DNS-threonine	2.9	8.75	1.23	0	8.35	1	0	12.63	1
	Phenylthiohydantoin amino acids
A-33	PTH-glutamic acid	0	2.45	1	0	1.8	1	0	3.66	1
A-34	PTH-aspartic acid	0.5	2	1.02	0	1.62	1	0	2.25	1

Conditions: 75/25 ACN/5 mM NH₄COOH pH 3.0; Capillary 35 cm × 100 µm i.d.; 25°C at –30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS).

Figure 4.

Bar plots representing the number of acidic (A and B), basic (C and D) and neutral (E and F) compounds separated on GMA/β-CD, GMA/β-CD-AMPS and GMA/β-CD-VBTA monolithic columns. Conditions for A, C and E: 75% (v/v) ACN/25% (v/v) H₂O, 5 mM NH₄COOH, pH 3.0; capillary 35 cm × 100 µm i.d.; 25°C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS). Conditions for B, D and F: 50% (v/v) ACN/50% (v/v) H₂O, 5 mM NH4OAc, 0.3% (v/v) TEA (pH 5.0); capillary 35 cm × 100 µm i.d.; 25°C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS).

Next, for the same set of negatively charged compounds, the second set of mobile phase composition (50/50 ACN/5 mM NH₄OAc 0.3% TEA, pH 5.0) was screened on three chiral monolithic columns. Again, irrespective of pH both GMA/β-CD and GMA/β-CD-AMPS columns partially separated only one acidic analyte. For example, A12 and A13 were separated on GMA/β-CD and GMA/β-CD-AMPS columns, with R_s of 0.1 and 0.4, respectively. In contrast, the GMA/β-CD-VBTA column was able to baseline enantioseparate one compound and partially separate four additional acidic compounds at the same pH 5.0 mobile phase (Figure 4B). Comparing the bar plots for separation on GMA/β-CD-VBTA column (Figure 4A versus B) clearly shows the success rate of acidic compound is 68% at pH 3.0 and only 14% at pH 5.0. Thus, acidic compounds seem to resolve better in partially anionic form.

The differences in enantioseparation capability of acidic compounds between the three monolithic CSP under two mobile phase pH conditions suggest that not only the presence of charged co-monomer is important but also the analyte charge state has a significant effect on chiral recognition. For example, the GMA/β-CD-VBTA column monolithic column with pH-independent positive charge will attract the negatively charged chiral analytes to facilitate complexation with the β-CD cavity. The GMA/β-CD-AMPS column is just the opposite of the GMA/β-CD-VBTA-based column, in that its negatively charged sites likely repelled the negatively charged analytes away from the chiral centers of the β-CD cavity hampering chiral separations. Although the GMA/β-CD monolithic material lacks any ionization site, it can still form inclusion complexation with some aromatic analytes due to π–π interactions. Interestingly, the inclusion complexation alone can barely cause the anionic analytes to allow enantioselective interactions.

Comparing the enantioseparation of acidic compounds on VBTA-based column, it seems likely that two essential factors, π–π interactions and analytes acidic strength, are important. As shown in Table II, most of phenoxy propionic acids (A1 to A7), α-hydroxy acids (A8 to A16), dinitrophenyl amino acids and their analogs (A17 to A22) as well as dansylated amino acids analogs (A30 to A32) provided enantioseparation on the GMA/β-CD-VBTA monolithic column. In contrast, profens (A23 to A29), which are relatively weaker acids than other classes of acidic compounds, provided essentially no enantioseparation when screened on GMA/β-CD-VBTA monolithic column. In addition, note that how a small change in structure of some acidic analytes affects the enantioresolution on the VBTA CSP. For example, the two dansylated amino acids [dansyl-serine (A31) and dansyl-threonine (A32)], with two aromatic rings, were baseline enantioseparated. In contrast, dansyl-phenylalanine (A30) with three aromatic ring was the weakest binder and was not enantioseparated, suggesting that the size and shape of analytes may influence the inclusion complexation, which in turn affects the enantioseparation ability on GMA/β-CD-VBTA column. The enantiomers of PTH-AAs with acidic chain such as PTH-glutamic acid (A33) had no chiral R_s, whereas PTH-aspartic acid (A34) had an R_s of 0.5.

The enantioseparations of 10 representative acidic compounds profiled on three monolithic columns, with the first five identified on the left, and the next five identified on the right are shown in Figure 5. The select compounds show contrasting abilities of the monolithic columns. For example, the enantioseparation of the compound A1, [2-chlorophenoxypropionic acid], was observed in Figure 5 using GMA/β-CD-VBTA column. Moreover, the GMA/β-CD and GMA/β-CD-AMPS monolithic columns have respective lower and higher k′ compared with GMA/β-CD-VBTA column, still no chiral separation was observed on the former two columns. Similar trends were observed for other PPA (e.g. A3 and A4). The exception was A2 (i.e., 2,2 (chlorophenoxy)-propionic acid), Table II) perhaps due to the shape and size of this PPA. Among the α-hydroxy acids analogs, atrolactic acid (A10), α-methoxyphenyl acetic acid (A15) and 3,4 (methylenedioxy) mandelic acid (A16) were the three best resolved α-hydroxy acids on VBTA column, while no enantioseparations were observed on GMA/β-CD and GMA/β-CD AMPS monolithic columns. Dintiro amino acids (e.g., dinitrobenzoyl leucine, A21) and dansylated amino acids [e.g. dansylated serine (A31) and dansylated threonine (A32)] separated with high enantioresolution on VBTA column. Again, no chiral resolutions were seen on the other two monolithic columns. In general, the run times and the k′ for most weak acids on the three columns follows the order: GMA/β-CD-VBTA > GMA/β-CD AMPS > GMA/β-CD. Clearly, the trend suggests that the increase in k′ does not necessarily provide better chiral R_s, but chiral selectivity plays a major role in enantioresolution.

Figure 5.

Representative electrochromatogram of acidic compounds using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS columns. Conditions: 75/25 ACN/5 mM NH₄COOH pH 3.0; capillary 35 cm × 100 µm i.d.; 25°C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS).

Basic compounds screening

The structure of a total of 30 structurally diverse basic compounds (B1 to B30) screened on three monolithic columns is shown in Supplementary data, Figure S2. When these set of compounds were screened on three monolithic columns in the optimized mobile phase at pH 5.0 (successfully used for separation of acidic compounds on VBTA column), no significant difference in chiral resolution was seen among the three columns (Figure 4C). On the other hand, the bar plots in Figure 4D shows that the AMPS-based monolithic column provided a very high success rate of 90% by separating 27 of the 30 anionic compounds under the optimized mobile phase, namely 50% (v/v) ACN/50% (v/v) H₂O, 5 mM NH₄OAc 0.3% (v/v) TEA, pH 5.0. A total of six basic compounds provided R_s values ≥1.5, another two with R_s between 1.0 and 1.5 and 19 compounds with R_s between 0.1 and 1.0. Under the same mobile phase composition, the VBTA-based column shows partial separation of four basic analytes (B1, pindolol; B10, norephedrine; B12, pseudoephedrine and B21, promethazine). The worst results were again seen on the GMA/β-CD column, which provided partial R_s of only 0.3 for one basic compound promethazine (B21). Nevertheless, a comparison of two mobile phase compositions suggests that unlike for acidic compounds, GMA/β-CD-AMPS column at pH 5.0 is most suitable combination for the enantioseparation of basic compounds (Table III).

Table III.

Screening Results of 30 Basic Compounds Using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS Columns

	Basic analytes	GMA/β-CD-VBTA			GMA/β-CD			GMA/β-CD-AMPS
		Rs	k′	α	Rs	k′	α	Rs	k′	α
	β-Blockers
B1	Pindolol	0.7	1.56	1.05	0	0.65	1	0.3	1.1	1
B2	Propranolol	0	1.6	1	0	0.67	1	1.57	1.29	1.04
B3	Oxprenolol	0	1.53	1	0	0.63	1	2.1	1.02	1.07
B4	Alprenolol	0	1.45	1	0	0.59	1	0.82	3.2	1.04
B5	Nadolol	0	1.49	1	0	0.61	1	0.42	1.05	1.02
B6	Metoprolol	0	1.95	1	0	0.86	1	0.6	3.16	1.02
B7	Atenolol	0	1.68	1	0	0.71	1	0.61	0.88	1.02
B8	Labetalol	0	1.17	1	0	0.45	1	0.21	0.85	1.01
	Ephedrine and derivations
B9	Methylephedrine	0	1.91	1	0	0.84	1	0.87	1.32	1.03
B10	Norephedrine	1	2.17	1.06	0	0.98	1	0.45	1.49	1.02
B11	Ephedrine	0	1.05	1	0	0.39	1	0.1	1.01	1.01
B12	Pseudoephedrine	0.3	4.7	1.01	0	2.38	1	2.71	0.87	1.14
	Miscellaneous
B13	Terbutaline	0	5.71	1	0	2.95	1	0.46	0.6	1.03
B14	Doxylamine	0	4.84	1	0	2.46	1	0.62	4.96	1.09
B15	Aminoglutethimide	0	7.63	1	0	4.04	1	1.81	3.5	1.11
B16	Bupivacaine	0	4.13	1	0	2.06	1	0.62	1.34	1.04
B17	Nefopam	0	1.76	1	0	0.75	1	0.31	1.22	1.02
B18	Troger's Base	0	5.85	1	0	3.03	1	2.81	0.88	1.02
B19	Octopamine	0	1.45	1	0	0.59	1	0.6	1.09	1.15
B20	PTH-arginine	0	1.21	1	0	0.47	1	0.41	1.02	1.03
B21	Promethazine HCL	0.9	0.8	1.03	0.3	0.27	1.02	0.2	3.09	1.02
B22	Isoproterenol	0	1.25	1	0	0.49	1	1.22	0.63	1.01
B23	Verapamil	0	1.25	1	0	0.49	1	0	0.9	1.06
B24	Ipratropium bromide monohydrate	0	3.27	1	0	1.58	1	0.43	0.9	1
B25	Laudanosine hydrobromide	0	4.06	1	0	2.02	1	0.4	2.83	1.03
B26	Sulconazole	0	1.56	1	0	0.65	1	0	0.68	1.02
B27	Prilocaine	0	5.4	1	0	2.77	1	2.47	1.1	1
B28	PTH-histidine	0	0.84	1	0	0.29	1	0.62	1.77	1.09
B29	Synephrine	0	0.27	1	0	0.02	1	0	0.31	1
B30	Miconazole	0	1.37	1	0	0.55	1	1.04	4.53	1.12

Conditions: 50/50% (v/v) ACN/50% (v/v) H₂O, 5 mM NH₄OAc, 0.3% (v/v) TEA (pH 5.0); capillary 35 cm × 100 µm i.d.; 25°C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or +20 kV (GMA/β-CD-AMPS).

In our previous section for acidic compounds, we discussed that the differences in enantioseparation capability between the AMPS and VBTA CSPs is mainly due to the presence of negatively charged co-monomers in the polymerization mixture. With negatively charged moieties covered on the monolithic material, the AMPS-based monolithic column can form stable complexes by attracting basic analytes. In contrast, because of electrostatic repulsion, a great majority of basic chiral compounds cannot be separated or only partially resolved on the VBTA-based monolith. Representative electrochromatographic separations for the basic enantiomers screened on three monolithic CSP are compared in Figure 6. Each electrochromatogram includes the analyte identification (see Table II), the k′ and R_s values. The run times for most weak bases are within 10–20 min, except for aminoglutethimide (B15), for which the separation time was substantially longer (c. 30 min on VBTA column with no enantioseparation). Notably, majority of the basic compounds provided significantly higher enantioresolution using GMA/β-CD-AMPS compared with GMA/β-CD-VBTA or GMA/β-CD columns.

Figure 6.

Representative electrochromatogram of basic compounds using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS columns. Conditions: 50% (v/v) ACN/50% (v/v) H₂O, 5 mM NH₄OAc, 0.3% (v/v) TEA (pH 5.0); capillary 35 cm × 100 µm i.d.; 25°C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS).

Neutral compounds screening

The structure of a total of 33 neutral chiral compounds (N1 to N33) screened are shown in Supplementary data, Figure S3. The two mobile phase pHs (3.0 and 5.0) were used to compare the resolution of neutral chiral compounds on three monolithic columns. First, using the mobile phase at pH 3.0, 75% (v/v) ACN/25% (v/v) water, 5 mM NH₄COOH, the GMA/β-CD-VBTA column separated a total of 10 compounds, whereas GMA/β-CD-AMPS column and GMA/β-CD column separated 4 and 2 neutral compounds, respectively (Figure 4E). For the second mobile phase, namely 50% (v/v) ACN/50% (v/v) H₂O, 5 mM NH₄OAc 0.3% (v/v) TEA, pH 5.0, the AMPS column showed a better enantioseparation with 15 compounds separated compared with VBTA column (six compounds separated) and GMA/β-CD column (three compounds separated) (Figure 4F). At pH 3.0 and 5.0 mobile phase, almost all compounds were uncharged. In general, the k′ of neutral chiral compounds follows the trend k′_{GMA/βCD-AMPS} > k′_{GMA/βCD >} k′_{GMA-β−CD-VBTA} (Table IV) and k′_GMA/β_CD-VBTA > k′_{GMA-β−CD-AMPS} > k′_GMA-βCD (Table V) under the two mobile-phase conditions. The exceptions were five benzoin derivatives, which were retained exceptionally longer on GMA/β-CD-AMPS column at pH 5.0. Overall, it appears that the success rate for enantioseparation of neutral compounds were higher at pH 5.0 using AMPS compared with pH 3.0 using VBTA column.

Table IV.

Screening Results of 33 Neutral Compounds Using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS Columns

	Neutral analytes	GMA/β-CD-VBTA			GMA/β-CD			GMA/β-CD-AMPS
		Rs	k′	α	Rs	k′	Α	Rs	k′	α
	PTH-amino acids
N1	PTH-methionine	0	0.67	1	0	0.84	1	0	1.4	1
N2	PTH-valine	0	0.63	1	0	0.79	1	0	1.33	1
N3	PTH-leucine	0	0.58	1	0	0.74	1	0	1.25	1
N4	PTH-isoleucine	0	0.67	1	0	0.84	1	0	1.4	1
N5	PTH-norvaline	0.5	0.92	1.04	0	1.13	1	0	1.84	1
N6	PTH-norleucine	0	0.77	1	0	0.96	1	0	1.58	1
N7	PTH-tryptophan	1.2	1.08	1.07	0	1.32	1	0	2.13	1
N8	PTH-tyrosine	0	1.29	1	0	1.56	1	0	2.5	1
N9	PTH-4-hydroxyproline	0	1.33	1	0	1.61	1	0	2.57	1
N10	PTH-proline	1.0	1.35	1.06	0.4	1.63	1.03	1.9	2.61	1.11
N11	PTH-serine	0.5	0.73	1.04	0	0.91	1	0	1.51	1
N12	PTH-asparagine	0.7	0.65	1.05	0	0.82	1	0	1.36	1
N13	PTH-alanine	0	0.56	1	0	0.72	1	0	1.22	1
N14	PTH-aminobutyric acid	0	0.98	1	0	1.2	1	0	1.95	1
	Benzoins
N-15	Hydrobenzoin	0	0.65	1	0	0.82	1	0	1.36	1
N-16	Benzoin	0	0.69	1	0	0.86	1	0	1.44	1
N-17	Benzoin methyl ether	0	0.73	1	0	0.91	1	0	1.51	1
N-18	Benzoin ethyl ether	0	0.69	1	0	0.86	1	0	1.44	1
	Barbiturates
N-19	Secobarbital	0	0.94	1	0	1.15	1	0.5	1.88	1.04
N-20	Pentobarbital	0	0.85	1	0	1.06	1	0	1.73	1
	Miscellaneous
N-21	Chlorthalidone	0.5	0.46	1.04	0	0.6	1	0	1.03	1
N-22	Flavanone	0.8	0.81	1.05	0	1.01	1	0	1.66	1
N-23	3,4-Dihydroxylphenylalanine	0.5	0.37	1.02	0	0.46	1	0	0.99	1
N-24	Ethyl-3-hydroxy-4-methoxy mandelate	1.2	5.65	1.07	0	6.58	1	0	10.17	1
N-25	5-Methylphenyl-5-phenylhydantoin	0.5	0.38	1.05	0	0.5	1	0	0.89	1
N-26	2-Phenylcyclohexanone	0	0.48	1	0	0.62	1	0	1.07	1
N-27	Camphor	0	0.38	1	0	0.5	1	2.0	0.89	1.09
N-28	Mandelic acid ethyl ester	0	0.69	1	0	0.86	1	0	1.44	1
N-29	5-Methyl-5-phenylhydantoin	0	0.71	1	0	0.89	1	0	1.47	1
N-30	BOH	0	0.4	1	0	0.53	1	0	0.92	1
N-31	Tetrahydronaphthol	0	0.35	1	0	0.77	1	0	1.29	1
N-32	Thalidomide	0	1.37	1	0	2.16	1	0	3.42	1
N-33	trans-Stilbene oxide	0	1.58	1	0	2.45	1	0	3.86	1

Conditions: 75/25 ACN/5 mM NH₄COOH (pH 3.0); capillary 35 cm × 100 µm i.d.; 25 °C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS).

Table V.

Screening Results on 33 Neutral Compounds Using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS Columns

	Neutral analytes	GMA/β-CD-VBTA			GMA/β-CD			GMA/β-CD-AMPS
		Rs	k′	Α	Rs	k′	Α	Rs	k′	α
	PTH-amino acids
N1	PTH-methionine	0.7	1.33	1.03	0	0.53	1	0	0.9	1
N2	PTH-valine	0	1.48	1	0	0.66	1	0.6	0.73	1.01
N3	PTH-leucine	0	1.44	1	0	0.65	1	1.3	0.64	1.03
N4	PTH-isoleucine	0	1.56	1	0	0.70	1	0.6	0.81	1.03
N5	PTH-norvaline	0	1.79	1	0	0.78	1	0.1	1.19	1.01
N6	PTH-norleucine	0	1.53	1	0	0.63	1	0	1.02	1
N7	PTH-tryptophan	0	2.1	1	0	0.94	1	0	1.38	1
N8	PTH-tyrosine	0	2.47	1	0	1.14	1	0	1.62	1
N9	PTH-4-hydroxyproline	0	2.54	1	0	1.18	1	0	1.67	1
N10	PTH-proline	1.0	2.58	1.05	0	1.2	1	0	1.69	1
N11	PTH-serine	0.3	1.45	1.02	0	0.59	1	0.1	0.98	1.01
N12	PTH-asparagine	0	1.29	1	0	0.51	1	0	0.88	1
N13	PTH-alanine	0	1.13	1	0	0.43	1	0.2	0.79	1.01
N14	PTH-aminobutyric acid	0	1.91	1	0	0.84	1	0	1.26	1
	Benzoins
N15	Hydrobenzoin	0	1.29	1	0	0.51	1	1.7	4.71	1.08
N16	Benzoin	0	1.37	1	0	0.55	1	1.0	5.72	1.04
N17	Benzoin methyl ether	0	1.45	1	0	0.59	1	1.7	8.22	1.07
N18	Benzoin ethyl ether	0	1.37	1	0	0.55	1	1.4	8.89	1.06
	Barbiturates
N19	Secobarbital	0	1.83	1	0	0.8	1	0	1.21	1
N20	Pentobarbital	0	1.68	1	0	0.71	1	0	1.12	1
	Miscellaneous
N21	Chlorthalidone	0.6	0.93	1.02	0.2	0.33	1.01	2.0	0.49	1.08
N22	Flavanone	0.9	1.31	1.03	0.3	0.47	1.03	2.5	0.46	1.08
N23	3,4-Dihydroxylphenylalanine	0.6	1.25	1.03	0.2	0.49	1.02	1.1	0.9	1.04
N24	Ethyl-3-hydroxy-4-methoxy mandelate	0	9.81	1	0	5.29	1	0	6.6	1
N25	5-Methylphenyl-5-phenylhydantoin	0	0.76	1	0	0.24	1	0.3	3.08	1.03
N26	2-Phenylcyclohexanone	0	0.97	1	0	0.35	1	0	0.69	1
N27	Camphor	0	0.76	1	0	0.24	1	0	0.57	1
N28	Mandelic acid ethyl ester	0	1.37	1	0	0.55	1	0	0.93	1
N29	5-Methyl-5-phenylhydantoin	0	1.41	1	0	0.57	1	0.5	0.95	1.02
N30	BOH	0	0.8	1	0	0.27	1	0	0.6	1
N31	Tetrahydronaphthol	0	1.21	1	0	0.47	1	0	0.83	1
N32	Thalidomide	0	3.38	1	0	1.64	1	0	2.21	1
N33	trans-Stilbene oxide	0	3.81	1	0	1.88	1	0	2.5	1

50/50 ACN/5 mM NH₄OAc 0.3% TEA (pH 5.0); capillary 35 cm × 100 µm i.d.; 25°C at −30 kV (GMA/β-CD-VBTA and GMA/β-CD) or 20 kV (GMA/β-CD-AMPS).

Examination of the resolution data for structurally similar PTH-AA (N1 to N14, Table IV) indicated that five PTH-AA could be enantioseparated using VBTA column at pH 3.0 and only one with AMPS. On the other hand, at pH 5.0 reverse was observed with more PTH-AAs resolved on the latter column. The benzoin class of compounds (N15 to N18, Table V) was all enantioseparated with the latter AMPS column at pH 5.0 and none by the VBTA monolithic column at pH 3.0. In contrast, secobarbital was partially resolved at pH 3.0 but not at pH 5.0, whereas its structurally similar analog pentobarbital remain unresolved at both pH. In general, more miscellaneous compounds were enantioseparated at pH 5.0 with AMPS column compared with VBTA column, whereas reversed was true at pH 3.0, suggesting that the pH and composition of the mobile phase are important factors in each type of column optimization. Figure 7 shows some of the representative electrochromatographic separations on the three columns. No general trend was found as to which structural class of neutral chiral compounds was better resolved on AMPS and which ones on VBTA column. At least two compounds showed chiral resolution on both VBTA and AMPS columns. For example, N10 (PTH-proline) had similar resolution values but their optimum resolution were seen at pH 3.0 with VBTA column, and at pH 5.0 using AMPS column. Flavanone (N22) showed better resolution (R_s = 2.5) compared with AMPS column (R_s = 0.9) at pH 5.0 (electropherogram on the last right row of Figure 7), whereas a reverse trend was seen for the same enantiomeric pair at pH 3.0 (electropherogram on the fourth left row of Figure 7) where this enantiomeric pair was better resolved on VBTA monolithic column.

Figure 7.

Representative electrochromatogram of neutral compounds using GMA/β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS columns. Conditions are the same described in Figure 3

Conclusions

To demonstrate the role of achiral monomers in chiral separation in monolithic media, a chiral monomer GMA-β-CD was first synthesized. Next, three different type of monolithic columns GMA-β-CD-VBTA, GMA/β-CD and GMA/β-CD-AMPS were polymerized and characterized by SEM images, porosity, permeability and BET surface measurement. Improved chiral separations were achieved by the addition of achiral co-monomer to the polymerization mixture of GMA-β-CD and porogens in monolithic CEC. The improved chiral separation is accompanied by decrease in permeability but the k′ is not always lowest with the most permeable GMA-β-CD. Perhaps improved chiral separation is probably due to more rigid ternary complex of achiral co-monomer–chiral monomer–analyte compared with binary complex of chiral monomer–analyte. A compound library of 34 acidic, 30 basic compounds and 33 neutral compounds was screened using the GMA/β-CD-VBTA, GMA/β-CD-AMPS, and GMA/β-CD monolithic columns. The comparison suggests that the positively and negatively charged comonomers not only determines the direction of EOF but also provide the appropriate electrostatic interaction sites for chiral recognition. For example, GMA/β-CD-VBTA monolithic column separated negatively charged compounds with the highest success rate under negative polarity, whereas GMA/β-CD-AMPS monolithic separated positively charged compounds with highest success rate under positive polarity. Neutral compounds were investigated under both polarity conditions and found to have relatively higher success rate at pH 5.0 with GMA/β-CD-AMPS compared with pH 3.0 with GMA/β-CD-VBTA column. In general, GMA/β-CD chiral monomer is a good template that can be modified with different charged achiral co-monomers to develop a generic and harmonized approach for chiral separations. Our future studies will focus on predicting the enantioseparation for a combinatorial library of structurally similar positively and negatively charged achiral monomer.

Supplementary data

Supplementary data are available at Journal of Chromatographic Science online.

Funding

This work was supported by a grant from financial support by NIH (Grant # GM-062314) and Petroleum Research Foundation (PRF-Grant # 47774-AC7).