Anal. Chem. 2006, 78, 5729-5735
High-Performance Polymer-Based Monolithic Capillary Column Ken Hosoya,* Natsuki Hira, Katsuya Yamamoto, Masaru Nishimura, and Nobuo Tanaka
Department of Polymer Science, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan
In this report, we introduce a new entry of high-performance polymer-based monolithic capillary column for mainly small molecules. This capillary column was prepared using a newly introduced epoxy monomer with diamines. Simply heat-induced polycondensation in an appropriate porogenic solvent afforded a really homogeneous co-continuous monolithic structure having submicrometer-size skeletons with micrometer-size throughpores. We were also able to prepare chiral monolithic columns using a chiral epoxy monomer as well as a chiral diamine. A 21.5-cm-long, 100-µm-i.d. column afforded up to 40 000 theoretical plate numbers (N) for alkylbenzenes in 60% aqueous acetonitrile as a reversed-phasemode stationary phase. Due to a quite low column pressure drop, a 150-cm-long column was prepared. This long column afforded up to 200 000 plates for alkylbenzenes with only a 4-MPa column pressure drop. In contrast, in 100% acetonitrile, this column has “HILIC” property to show up to 60 000 plates for methanol with a 17.5-cm-long column. In this mode, we were able to separate nucleic acids. In addition, we have prepared a chiral column with both of the chiral epoxy monomers and an amine. This column was able to chirally discriminate a racemic alcohol in a reversed-phase mode. Monolithic columns packed with well-controlled co-continuous separation media have lots of advantages as high-speed separation media without loss of high performance. Silica-based monolithic columns as well as some polymer-based monolithic media are now commercially available, but polymer-based monolithic media have shown excellent properties for relatively large molecule separations,1 while silica-based ones also showed excellent efficiency for small molecules. In comparison to silica-based monolithic media, polymer-based media have difficulties in control of their macroporous structure, if conventional vinyl monomers are employed through typical radical polymerization. Conventionally, polymer monoliths have been prepared by in situ polymerization of functional monomer(s), cross-linking agent, and initiator in porogenic solvent (pore-forming diluent). As porogenic solvent, binary or ternary mixtures of poor or good solvents, or both, are occasionally used for controlling a variety of pore size and its structure.2 The porogenic solvents dominantly * To whom correspondence should be addressed. E-mail: kenpc@ kit.ac.jp. (1) Oefner, P. J.; Huber, C. G. Anal. Chem. 2002, 74, 4688-4693. (2) Peters, E. C.; Petro M.; Svec F.; Frechet J. M. J. Anal. Chem. 1997, 69, 3646-3649. 10.1021/ac0605391 CCC: $33.50 Published on Web 07/19/2006
© 2006 American Chemical Society
selected for the preparation of monolith-type media are usually poor solvents of the monomers utilized to form macro-throughpores required for liquid flow. In the case of the porogenic poor solvents, the growing polymer chains tend to aggregate with each other because van der Waals attraction surmounts the steric hindrance mutually expelling the polymer chains.3 Thus, in the case of the ordinary polymer monolith preparations, the phase separation between growing polymer chains and porogenic solvent proceeds so fast and the coarsening of the monolithic structure inherently leads to heterogeneous macroporous structures composed of tiny micrometer-size globular particles. The inherent disadvantages of these heterogeneous macroporous monoliths are cited by the adverse effects such as larger eddy diffusion through irregular interstitial channels (increase of the A term of the van Deemter equation), low permeability, monolith compressibility at high-pressure drops,4 and limited pore surface area for molecular recognition site, especially for small molecules. We have proposed a new method to realize nicely controlled bimodal co-continuous polymer media, the so-called viscoelastic phase separation concept.5 This concept has been introduced theoretically by Tanaka et al.,6,7 where good solvent for the monomer employed can be utilized with an ultrahigh molecular weight polystyrene standard. We have also reported a wellcontrolled 3D skeletal polymer monolith based on an epoxy resintype polymer.8 In this paper, we report the epoxy resin-based polymer monoliths, which is greatly improved in terms of chromatographic performance in high-performance liquid chromatograph. We mainly utilized a trifunctional epoxy monomer with diamino compounds to produce a nicely controlled monolithic capillary column. Some chiral discrimination will be included. EXPERIMENTAL SECTION Monomers. The structure of monomers utilized was illustrated in Figure 1. 4-[(4-Aminocyclohexyl)methyl]cyclohexylamine (BACM) and a chiral trans-1,2-cyclohexanediamine (CHD) were (3) de Gennes, P.-G. Scaling Concepts in Polymer Physics; Cornell University Press: Ithaca, NY, 1979; p 115. (4) Jiang, T.; Jiskra, J.; Claessens, H. A.; Cramers, C. A. J. Chromatogr., A 2001, 923, 215-227. (5) Aoki, H.; Kubo; Ikegami, T.; Tanaka, N.; Hosoya, K.; Tokuda, D.; Ishizuka, N. J. Chromatogr., A. In press. (6) Tanaka, H. Phys. Rev. Lett. 1996, 76, 787-790. (7) Tanaka, H.; Araki, T. Phys. Rev. Lett. 1997, 78, 4966-4969. (8) Tsujioka, N.; Hira, N.; Aoki, A.; Tanaka, N.; Hosoya, K. Macromolecules 2005, 38, 9901-9903.
Analytical Chemistry, Vol. 78, No. 16, August 15, 2006 5729
Figure 1. Structures of monomers utilized.
Figure 4. Calibration curve of TEPIC-BACM by size exclusion chromatography. Conditions: mobile phase, THF; pressure, 73 kg/ cm2; detection, 210 nm; temperature, 23 °C; flow rate, 234 nL/min; column, 20 cm × 100 µm i.d. (TEPIC-BACM); off column, 9 cm × 50 µm i.d.; solute, benzene, polystyrenes. Table 1. BET Mesopore Data of TEPIC-BACM BET specific surface area (m2/g)
total pore volume (cm3/g)
average pore diameter (nm)
2.7
0.004
5.65
TEPCI-BACM Figure 2. Scanning electron micrographs of TEPIC-BACM packed capillary.
Figure 5. Retention properties of TEPIC-BACM capillary column in aqueous acetonitrile. Conditions: mobile phase, AN/water 95/540/60 (v/v); pressure, 55 × 116 kg/cm2; detection, 210 nm; temperature, 30 °C; column, 36.4 cm × 100 µm i.d. (TEPIC-BACM); off column, 9 cm × 50 µm i.d.; solute, nucleic acid (adenine, uracil, thymine, cytosine), benzene, and toluene. Figure 3. Pore size and pore size distribution of TEPIC-BACM by the mercury intrusion method.
purchased from Tokyo Kasei Co. (Tokyo, Japan) and utilized as received. Tris(2,3-epoxypropyl) isocyanurate (TEPIC) (racemic and chiral) was kindly donated by Nissan Chemical Co. (Tokyo, Japan) and utilized without further purification. Other Materials. Poly(ethylene glycol) 200 (PEG 200) and 300 (PEG 300) were purchased from Nacalai Tesque (Kyoto, Japan) and utilized as porogenic solvents. (3-Aminopropyl)triethoxysilane was purchased from Nacalai Tesque and used as a surface-modifying agent of capillary inner wall, while 100 µm i.d. × 375 µm o.d. fused-silica capillary was purchased from Polymicro Technologies). Solvents. Ultrapure water was obtained through Milli-Q GPA system, while methanol, acetonitrile (AN), and tetrahydrofuran (THF) were purified by suitable distillation techniques. Chromatographic Solutes. All the alkylbenzenes as well as alkyl phenyl ketones were commercially available and used as 5730 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
Table 2. K′ Values in Figure 6 left K′ benzene toluene
0.90 1.24 right K′
thymine uracil uridine adenosine adenine guanosine
0.37 0.42 0.49 1.01 1.18 1.76
received. Polystyrene standard samples were purchased from Showa Denko, Co. (Tokyo, Japan) as size exclusion makers. Equipment. A constant-temperature oven DNE 400 (Yamato Co.) was utilized as a polymerization reactor. The HPLC consisted of a LC-20AT chromatographic pump (Shimadzu), DGU-20A as an on-line degasifier, and CE-2075 UV detector (Jasco) equipped with a Rheodyne 7725 injector. The scanning electron micrograph
Figure 6. Typical chromatograms in two different mobile phases on TEPIC-BACM. Conditions: column, 36.4 cm × 100 µm i.d. (TEPICBACM); detection, 210 nm; off column, 9 cm × 50 µm i.d. (Left) Mobile phase, AN/water 40/60 (v/v); pressure, 115 kg/cm2; flow rate, 0.5 mL/min (split method); temperature, 30 °C; solute, uracil, benzene, and toluene. (Right) Mobile phase, AN/water 90/10 (v/v); pressure, 55 kg/cm2; flow rate, 0.5 mL/min (split method); temperature, 30 °C; solute, nucleic acid (adenine, thymine, uracil, cytosine) and nucleoside (adenosine, guanosine, uridine, cytidine).
Table 3. K′ Values in Figure 7 K′ benzene toluene ethylbenzene proplybenzene
0.17 0.21 0.24 0.29
K′ butylbenzene amylbenzene hexylbenzene
0.36 0.43 0.51
Figure 8. Chromatogram of methanol in 100% acetonitrile. Conditions: mobile phase, AN; pressure, 24 kg/cm2; detection, 210 nm; temperature, ambient; column, 17.5 cm × 100 µm i.d.(TEPICBACM); solute, methanol.
Figure 7. Separation of alkylbenzenes on TEPIC-BACM capillary column. Conditions: mobile phase, AN/water 60/40 (v/v); pressure, 194 kg/cm2; detection, 210 nm; temperature, ambient; flow rate, 1 mL/min (split method); column, 21.5 cm × 100 µm i.d. (TEPICBACM); off column, 9 cm × 50 µm i.d.; solute, uracil, and alkylbenzenes.
was obtained using a Hitachi S-510. Pore size distribution was performed on a Micromeritics Poresizer 9320. Surface Modification of Inner Wall of the Capillary. The 100-µm-i.d. capillary was washed with 1 N NaOH aqueous solution
Figure 9. On column detection of TEPIC-BACM capillary column. Conditions: mobile phase, AN/water 60/40 (v/v); pressure, 194 kg/ cm2; detection, 210 nm; temperature, ambient; flow rate, 1.85 mL/ min (split method); column, 17.8 cm × 100 µm i.d. (TEPIC-BACM); solute, thiourea, alkylphenone.
and kept at 70 °C for 30 min, followed by washing with pure water. The capillary was washed with 1 N HCl and kept at 70 °C for 30 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
5731
Figure 10. Scanning electron micrographs of TEPIC-BACM-S and TEPIC-BACM-L. TEPIC-BACM-S column length, 18.2 cm; TEPICBACM-L column length, 150.5 cm. Table 4. K′ Values in Figure 9 K′ acetophenone propiophenone butyrophenone valerophenone hexanophenone
0.08 0.14 0.19 0.26 0.34
K′ heptanophenone octanophenone nonanophenone decanophenone
0.42 0.52 0.63 0.76
Table 5. Components of TEPIC-BACM-S and TEPIC-BACM-L TEPIC(g)
BACM(g)
PEG 200 (g)
temperature (°C)
1.60
0.36
7.00
80
min, followed by washing with pure water and acetone. After complete removal of acetone by air, THF and (3-aminopropyl)triethoxysilane (1:1 v/v) were flowed through the capillary and kept at 80 °C for 24 h. After the reaction, the modified capillary was washed repeatedlywith ethanol. Preparation of Capillary Column. TEPIC (1.60 g) and BACM (0.37 g) were completely dissolved in PEG 200 (7.00 g). This solution was injected into the modified capillary by a syringe. The polymerization reaction was carried out at 80 °C for 12 h. The resulting capillary was washed with water and methanol and dried in vacuo for 5 h. In the case of chiral amine, 0.63 g of CHD was admixed with TEPIC in PEG 300 (12.64 g) and polymerization was carried out at 120 °C for 24 h followed by the same washing process. The ratio of monomers affected the formation of cocontinuous structure and will be discussed later in Results and Discussion. Measurement. Monolithic bulk material was prepared in a test tube as well as in the preparation of the capillary column. 5732 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
Figure 11. H-u plot of TEPIC-BACM columns as well as C18 particle packed column. Conditions: mobile phase, AN/20 mM sodium phosphate buffer pH 7.0 60/40 (v/v); detection, 210 nm; temperature, 23 °C; column, TEPIC-BACM-S 18.2 cm × 100 µm i.d., TEPICBACM-L 150.5 cm × 100 µm i.d.; Shim-pack VP-ODS 15.0 cm × 4.6 mm i.d.; off column, 9 cm × 50 µm i.d.; solute, benzene. Table 6. A and C Terms of Van Deemter Equation of TEPIC-BACM
TEPIC-BACM-S TEPIC-BACM-L
A term (µm)
C term (ms)
1.3 5.0
2.5 2.0
Pore size distribution was measured by the mercury intrusion method. The surface area was measured by BET. RESULTS AND DISCUSSION Formation of Co-Continuous Structure. In our previous paper, we discussed the ratio of epoxide monomer and amine monomer to realize a well-controlled co-continuous structure.8 The stoichiometric ratio did not afford good structure; this was easily
Figure 12. Separation of alkylbenzenes on TEPIC-BACM-S and TEPIC-BACM-L. Conditions: mobile phase, AN/20 mM sodium phosphate buffer pH 7.0 60/40 (v/v); pressure, (left) 8.7, (right) 4.0 MPa; detection, 210 nm; temperature, 23 °C; column, (left) TEPIC-BACM-S 18.2 cm × 100 µm i.d., (right) TEPIC-BACM-L 150.5 cm × 100 µm i.d.; off column, 9 cm × 50 µm i.d.; solute, uracil and alkylbenzenes.
Figure 13. P-u plots and permeability of the columns. Conditions: mobile phase, AN/20 mM sodium phosphate buffer pH 7.0 60/ 40 (v/v); detection, 210 nm; temperature, 23 °C; column, TEPICBACM-S 18.2 cm × 100 µm i.d., TEPIC-BACM-L 150.5 cm × 100 µm i.d.; Shim-pack VP-ODS 15.0 cm × 4.6 mm i.d.; off column, 9 cm × 50 µm i.d.; solute, benzene.
understood based on the difficulty in stoichiometric polymerization in the growing viscous polymerization medium. One of the optimized ratios was 2.77 (epoxide):1 (amine). First, we employed this ratio for TEPIC and BACM, but well-controlled co-continuous structure has never been obtained. We studied the ratio of monomers as well as the amount of PEG and reaction temperature; finally the feed recipe mentioned in the Experimental Section was chosen. Observation by scanning electron microscope revealed the typical morphology of the prepared epoxy polymer as shown in Figure 2. A narrow skeleton of ∼0.5 µm in diameter and relatively large through-pores of ∼3-4 µm were observed. The boundary between monolithic polymer and the inner wall looked very smooth without any disconnection. Pore size measurement also supported unimodal pore size distribution as shown in Figure 3, where mean pore size was calculated at ∼3.3 µm in diameter. The total porosity was calculated as 82.3%, while porosity calculated from the composition of feed solution was 80%. Although the pore size distribution measured by the mercury intrusion method as described above did not afford significant surface area, size exclusion chromatography in THF afforded total porosity at 90.1%, while the volume of through-pores was calculated at 67.8% as illustrated in Figure 4. This means significant pore volume contributing to retention of small molecules was
Figure 14. Relationship between separation impedance E and linear velocity. Conditions: mobile phase, AN/20 mM sodium phosphate buffer pH 7.0 60/40 (v/v); detection, 210 nm; temperature, 23 °C; column, TEPIC-BACM-S 18.2 cm × 100 µm i.d., TEPICBACM-L 150.5 cm × 100 µm i.d.; Shim-pack VP-ODS 15.0 cm × 4.6 mm i.d.; off column, 9 cm × 50 µm i.d.; solute, benzene.
involved, because benzene was eluted last. The surface area of the polymeric materials is shown in Table 1. Chromatographic Properties. A prepared 36.4-cm-long capillary column was chromatographically studied in aqueous AN mobile phases. As shown in Figure 5, the content of AN (φ) was varied from 40 to 95% in volume, we plotted k′ values of benzene, toluene, uracil, cytosine, adenosine, and thymine calculated based on retention time of water as t0. With the increment of φ, k′ values of hydrophobic solutes, benzene, and toluene were decreased. This means a reversedphase-mode separation dominantly worked on this stationary phase. On the other hand, hydrophilic solutes, uracil, cytosine, adenosine, and thymine afforded much larger k′ values at higher acetonitrile content. This is the so-called hydrophilic interaction chromatography (HILIC) property and is probably due to a relatively hydrophilic polymer backbone containing OH as well as amine functional groups. Therefore, by control of the AN content, we are able to choose an appropriate separation mode. Using this column, separations between benzene and toluene, as well as some nucleic acids were performed as shown in Figure 6, where theoretical plate numbers were noted in the figure. K′ values in Figure 6 are shown in Table 2. A 21.5-cm-long column afforded separation of alkylbenzenes (from benzene to hexylbenzene) in 60% AN as shown in Figure Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
5733
Figure 15. Column performance of half-cut columns. Conditions: mobile phase, AN/20 mM sodium phosphate buffer pH 7.0 60/40 (v/v); detection, 210 nm; temperature, ambient; off column, 9 cm × 50 µm i.d.; column, TEPIC-BACM, (A) 72 cm × 100 µm i.d., (B) 36 cm × 100 µm i.d., (C) 36 cm × 100 µm i.d.; pressure, (A) 18 kg/cm2, (B) 9 kg/cm2, (C) 9 kg/cm2; solute, uracil and alkylbenzenes.
Table 7. Permeability K K (× 10-14 m2) TEPIC-BACM-L TEPIC-BACM-S C18-silica particle
26.8 2.1 4.2
7, where up to 40 t000 theoretical plate numbers were obtained for all the solutes. K′ values in Figure 7 are shown in Table 3. On the other hand, in 100% AN, methanol was eluted with up to 60 000N on the 17.5-cm-long column as shown in Figure 8. These are quite interesting chromatographic properties and quite a high chromatographic performance comparable to those of silica monolith-based capillary columns. According to our knowledge, this is the first example of a high-performance, polymer-based monolithic column for small molecules. This might be because no aromatic functional groups are involved in the polymer monolith. Due to this property of polymer utilized, on-column detection was successfully done to show the separation of alkyl phenyl ketones as shown in Figure 9. The 17.8-cm-long column afforded around N ) 30 000 in 60% AN mobile phase. K′ values in Figure 9 are shown in Table 4. Detailed Chromatographic Performance. We have prepared two different types of capillary columns using the same method described before. Components of TEPIC-BACM-S and TEPICBACM-L are shown in Table 5. As shown in Figure 10, the 18.2cm-long column (TEPIC-BAMC-S) and 150.5-cm-long column (TEPIC-BAMC-L) were prepared, but by change of prepolymerization time, the morphology of the two monolithic columns looked rather different. The TEPIC-BACM-S column had small domain size (size of skeleton + size of through-pore), while the TEPICBACM-L column had a rather large domain size. At this moment, the reason the different domain sizes were obtained is not clear. 5734
Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
Usually epoxy resin is easily polymerized once s curing agent is admixedl therefore, the relatively long time for filling of monomer solution into the long capillary somehow affected the morphology. Using a C18 particle packed column (Shim-Pack VP-ODS 15.0 cm × 4.6 mm i.d., particle size 4.6 ( 0.3 µm) as a reference, the H-u plots of the prepared capillary columns were examined in AN/20 mM sodium phosphate buffer pH 7.0 60/40 (v/v) as mobile phase using benzene as the solute. As shown in Figure 11, much smaller H values were obtained compared with those on a C18 column, and on a TEPIC-BACM-S column realized smaller than H ) 5 for the solute utilized, benzene. A and C terms of the Van Deemter equation of TEPIC-BACM are shown in Table 6. This is excellent column performance with a polymer-based monolithic column. In Figure 12, separation of alkylbenzenes was performed on both columns. TEPIC-BACM-S afforded up to 40 t000 plates with 8.7-MPa column pressure drop, while the TEPIC-BACM-L column was driven only by 4.0 MPa to show up to 200 000 plates for the separation of alkylbenzenes. The observed low column pressure drop on a TEPIC-BACM-L column was explainable due to wider through-pores of this column. In fact, this column efficiency is extremely high. If P (column pressure drop) plotted against u (linear velocity) was examined, as shown in Figure 13, a TEPIC-BACM-L column showed much lower P than that on a C18 column. According to the rather small domain size of a TEPIC-BACM-S column, higher P was observed on this column compared with that on a C18 column. The permeability of the columns was calculated based on the following equation, as expected,
K ) uηL/∆P
where K is permeability, ∆P is pressure drop, η is viscosity of
Table 8. Plate Heights H (µm) of Half-Cut Columns
uracil benzene toluene ethylbenzene propylbenzene butylbenzene amylbenzene hexylbenzene heptylbenzene
A (36 cm)
B (36 cm)
13.0 10.9 12.0 13.3 13.1 14.2 14.1 15.6 18.8
13.6 11.8 14.0 14.3 15.0 16.0 17.0 20.1
Figure 16. Example of chiral separation. Conditions: mobile phase, AN/20 mM sodium phosphate buffer pH 7.0 50/50 (v/v); pressure, 112 kg/cm2; detection, 210 nm; temperature, 28 °C; column, TEPIC(S,S,S)-CHD(S,S) 17.5 cm × 100 µm i.d.; off column, 9 cm × 50 µm i.d.; solute, (R,S)-1,1′-bi-2-naphthol.
mobile phase, u is linear velocity, and L is column length. TEPICBACM-L showed high permeability as shown in Table 7. On the other hand, separation impedance E was calculated based on the following equation9
E)
∆Pt0 ηN2
)
( )( )( )
H2 ∆P t0 1 ) N N η K
and plotted against linear velocity, where ∆P is column pressure drop, t0 is elution time of not retained solute, η is viscosity of mobile phase, N is theoretical plate number, H is plate height, and K is permeability. We had much smaller E values for TEPICbased two capillary columns as shown in Figure 14. These values (9) Bristow, P. A.; Knox, J. H. Chromatogr. 1977, 10, 279-289.
Table 9. Efficiency of chiral column
(S)-1,1′-bi-2-naphthol (R)-1,1′-bi-2-naphthol
K′
Rs
R
N
3.65 3.85
0.82
1.06
6027 4562
obtained with both polymer-based monoliths were very low, which means high performance. For example, the lowest E on TEPICBACM-L column was only 200, which was 1/30 of C18 particle packed column. To study more about the long capillary column, we have prepared a 72-cm-long column and cut it into two 36-cm-long columns. Figure 15 revealed the column performance of the 72cm column and two 36-cm-long columns. Interestingly, two 36cm columns showed vanishingly small differences in plate height H as shown in Table 8. These facts strongly suggest that homogeneity of the polymer monolithic structure is acceptable. Chiral Monolith. If we utilized a chiral TEPIC (TEPIC(S,S,S)), a chiral diamine CHD(S,S) monolithic chiral column could be prepared. On this column, (R,S)-1,1′-bi-2-naphthol was chirally discriminated as shown in Figure 16. Table 9 displays the efficiency of a chiral column. This means a monolithic chiral column using chiral monomers has a chiral discrimination. At this moment, experimental conditions involving preparation method, monolithic structure, and separation condition are still being investigated, but this chiral monolith is obviously a good target of this epoxy-based polymer monolith. CONCLUSION We have prepared a new entry of a high-performance polymerbased monolithic capillary column using epoxy resin-based polymer. The TEPIC-BACM column afforded really high performance based on its homogeneous co-continuous structure having submicrometer-size skeleton and micrometer-size throughpores. If we utilized chiral monomers, chiral discrimination took place under the reversed-phase condition. ACKNOWLEDGMENT The financial support from Grant-in-Aid for Basic Scientific Research (14042232 and 16350082) from the Ministry of Education, Science, Sport, and Culture of Japan is gratefully acknowledged. Received for review March 24, 2006. Accepted June 6, 2006. AC0605391
Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
5735
