Preparation Strategy for Uniformly Sized,Polymer-Based HPLC Packing Materials HavingPractically Acceptable Column Efficiency. 1.Copolymerization Technique

Ken Hosoya,* Masashi Teramachi, and Nobuo Tanaka

Department of Polymer Science, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan

Atsushi Kobayashi, Taketoshi Kanda, and Yutaka Ohtsu

Basic Research Center, Research & Development Division, Shiseido Research Center (Shin-Yokohama),2-2-1 Hayabuchi, Tsuzuki-ku, Yokohama 224-8558, Japan

Uniformly sized, polymer-based packing materials afford-ing excellent chromatographic performance in semimicroHPLC were prepared from mixtures of alkyl methacrylateand glycerol dimethacrylate by a copolymerization tech-nique using a multistep swelling and polymerizationmethod. A column efficiency of up to 13 000 plates wasobtained for a 2-mm-i.d. × 150-mm column with beadsprepared from a 1:1 (v:v) mixture of an alkyl methacrylateand glycerol dimethacrylate. The packing materials exhibita considerable improvement in column efficiency, asdemonstrated on the separation of polyaromatic hydro-carbons (PAHs) having a relatively long retention time,which thus far has been a serious problem of polymer-based packing materials. Detailed studies suggested arather limited optimum pore size, and its distribution wasdepicted to get the excellent column efficiency on polymer-based packing materials.

Polymer-based packing materials for high performance liquidchromatography (HPLC) are commercially available.1 However,the application field of polymer beads is rather limited whencompared to that of widely utilized silica-based stationary phases,despite their good chemical stability and a broad variety of surfacechemistries. The reasons for this are lower column efficiencies,as compared to that of inorganic oxide-based stationary phases,a complicated retention mechanism, and difficult preparationmethods.2-6

Uniformly sized polymer packing materials that have beenrecently developed7-12 have solved some of these problems. For

example, a much lower column pressure drop results in bettercolumn efficiency, as compared to that of more polydispersepolymer particles prepared by the traditional suspension poly-merization process.13 Although the size uniformity was a clearadvantage,14-15 the column efficiency remained unacceptably lowtoward planar solutes having longer retention times, such aspolyaromatic hydrocarbons (PAHs).16-17

This problem can be clearly observed even using a typicalcommercial polymethacrylate-based 150-mm-long column forwhich the manufacturer claims column efficiency over 10 000plates (Figure 1). The packing material used for this commercialcolumn has good size uniformity. According to Figure 1, excellentcolumn efficiency was observed only for solutes having shorterretention times, and severe peak broadening occurred for PAHs,such as naphthalene or anthracene, in an aqueous acetonitrilemobile phase and resulted in a rapid decrease in the columnefficiency, much below the value of 10 000 plates claimed by theproducer.

These findings suggest that particle size uniformity and theirmorphology do not lead automatically to high column efficiency.13

It is likely that the retentivity of the solutes depends on nanoscalesurface properties of the polymer-based packing materials. Theseproperties can be controlled through the polymerization condi-tions, such as the initiation method, polymerization technique,

(1) Ohto, M.; Yamamoto, A.; Matsunaga, A.; Mizukami, E. Bunseki Kagaku1994, 43, 71-74.

(2) Tanaka, N.; Ebata, T.; Hashizume, K.; Hosoya, K.; M. Araki J. Chromatogr.1989, 475, 195-208.

(3) Hosoya, K.; Maruya, S.; Kimata, K.; Kinoshita, H.; Araki, T.; Tanaka, N. J.Chromatogr. 1992, 625, 121-129.

(4) Kimata, K.; Hosoya, K.; Tanaka, N. Bunseki 1992, 450-457.(5) Tanaka, N.; Hashizume, K.; Araki, M. J. Chromatogr. 1987, 400, 33.(6) Nevejans, F.; Verzele, M. J. Chromatogr. 1987, 406, 325.(7) Ugelstad, J.; Kaggerud, K. H.; Hansen, F. H.; Perge, A. Makromol. Chem.

1979, 180, 737-744.

(8) Ugelstad, J.; Mork, P. C. Adv. Colloid Interface Sci. 1980, 13, 101-140.(9) Ugelstad, J.; Mfutakamba, H. R.; Mork, P. C.; Ellingsen, T.; Berge, A.;

Schmid, R.; Holm, L.; Jorgedal, A.; Hansen, F. K.; Nustad, K. J. Polym. Sci.,Polym. Symp. 1985, 72, 225-240.

(10) Kasai, K.; Hattori, M. Gosei Gomu 1986, 94, 11-17.(11) Cheng, C. M.; Micale, F. J.; Vanderhoff, J. W.; El-Aasser, M. S. J. Polym.

Sci., Part A: Polym. Chem. 1992, 30, 235-244.(12) Cheng, C. M.; Vanderhoff, J. W.; El-Aasser, M. S. J. Polym. Sci., Part A:

Polym. Chem. 1992, 30, 245-256.(13) Hosoya, K.; Frechet, J. M. J. J. Liq. Chromatogr. 1993, 16, 353-365.(14) Haginaka, J.; Hosoya, K.; Kimata, K. Bunseki 1997, 474-480.(15) Hosoya, K.; Tanaka, N. In Advanced Technology of Microspheres and Powders;

Kawaguchi, H., Ed.; CMC: Tokyo, 2000, pp 229-237.(16) Hosoya, K.; Kimata, K.; Tanaka, N.; Araki, T.; Terashima, M.; Frechet, J.

M. J. J. Liq. Chromatogr. 1993, 16, 3059-3071.(17) Hosoya, K.; Frechet, J. M. J. J. Polym. Sci., Part A: Polym. Chem. 1993,

31, 2129-2141.

Anal. Chem. 2001, 73, 5852-5857

5852 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001 10.1021/ac0105991 CCC: $20.00 © 2001 American Chemical SocietyPublished on Web 11/10/2001

reaction temperature, and properties of the cross-linking monomerand by copolymerization with other functional monomers.

As a result of our extensive studies in the preparation ofuniformly sized polymer particles, we now wish to demonstratethat a copolymerization technique using a hydrophilic cross-linkingagent together with hydrophobic functional monomers is aneffective method leading to significant improvement of the columnefficiency, which makes these packings very useful for semimicroHPLC. Our objective is also the creation of a “database” ofpolymer-based HPLC packing materials that have practicallyacceptable column efficiency.

EXPERIMENTAL SECTIONSolvents. Acetonitrile and tetrahydrofuran (THF) were HPLC

grade and were used as received. Water utilized for the prepara-tion of the mobile phase was ultrapure water produced in ourlaboratory using a Yamato Auto Still model WG-22 followed by atreatment in Branstead E-Pure equipment. Cyclohexanol (poro-gen) was purified using the standard distillation.18

Materials. Styrene for the preparation of the seed particlessuitable for a multistep swelling and polymerization method waswashed using 5% aqueous sodium hydroxide solution and satu-rated sodium chloride solution, dried over calcium chloride, and

distilled in a vacuum. Glycerol dimethacrylate (GDMA) was a giftfrom Kyoeisya Chemicals Inc., (Osaka, Japan), and alkyl meth-acrylate monomers [methyl methacrylate (MMA), butyl meth-acrylate (BMA), and 2-ethylhexyl methacrylate, (2-EHMA)] werepurchased from Wako Pure Chemical Ltd., (Osaka, Japan). All ofthese monomers that were utilized for preparation of stationaryphases were distilled under reduced pressure to remove impuritiesand inhibitors.18 The radical initiator benzoyl peroxide (BPO) wascrystallized from mixtures of chloroform and methanol to removethat was water serving as a stabilizer. Potassium peroxodisulfateas a water-soluble initiator was also recrystallized prior to its use.18

Preparation of the Seed Particle.19 Oxygen was removedfrom the boiling deionized water (300 mL) by purging with heliumfor 30 min, and the water was allowed to cool to the roomtemperature. Then 0.39 g of sodium chloride and 6 mL of thepurified styrene were added. The temperature of this dispersionwas increased to 75 °C under argon atmosphere. Then a solutionof 0.27 g of potassium peroxodisulfate in 50 mL water was addedto initiate the emulsifier-free emulsion polymerization. After theinitiation, 7 mL of styrene was added each hour, and finally, 2mL of styrene was added after 7 h after the first initiation. Thetotal amount of styrene was 50 mL. The mixture was stirred foranother 17 h. After the completion of the polymerization reaction,the emulsion was purified using repeated centrifugation (5000 rpmfor 30 min) and redispersion in water. The yield of the polystyreneseeds was 61%, and their concentration in the aqueous dispersionwas 0.0552 g/mL.

Multistep Swelling and Polymerization.7,17 A microemulsionconsisting of the activating solvent (dibutyl phthalate, 0.405 mL),sodium dodecyl sulfate (0.028 g), and 10 mL of water preparedby sonication was admixed to a 1.44-mL suspension of the seedparticles. The swelling was completed in 4 h at room temperature.

Another microemulsion was prepared by sonication of theporogen (cyclohexanol 5 mL) and BPO (0.10 g) in 25 mL of anaqueous solution of poly(vinyl alcohol) (degree of polymerization500, 0.48 g) and added to the activated seeds, and the swellingcontinued for 5 h. Then a microemulsion consisting of 5 mL ofmonomers (GDMA + alkyl methacrylate) and another 25 mL ofaqueous poly(vinyl alcohol) was added, and the swelling continuedfor 5 h at room temperature.

(18) Perrin, D. D.; Armarego, W. L. F.; Perrin, D. R. Purification of LaboratoryChemicals, 2nd ed.; Pergamon Press: Oxford, 1980.

Figure 1. Chromatogram on a commercial polymer-based column.Column size, 4.6 mm i.d. × 150 mm; mobile phase, 80% (v/v)aqueous acetonitrile; flow rate, 1.0 mL/min; temp, 30 °C; UV detection,254 nm. Solutes are depicted on the figure.

Table 1. Prepared Stationary Phases

symbol volume ratio (v/v/v)

E2 2-EHMA/GDMA/CHO (2/8/10)M5 MMA/GDMA/CHO (5/5/10)B5 BMA/GDMA/CHO (5/5/10)E5 2-EHMA/GDMA/CHO (5/5/10)

Table 2. Chemical Yields, Particle Sizes, and TheirDistributions

packing yield (%) particle size (µm) CV value (%)a

E2 99.1 4.17 3.72M5 80.8 4.06 3.86B5 87.8 3.81 2.95E5 87.2 3.99 3.43

a CV (%) ) {∑(d - dav)2/N}1/2dav × 100 (d, particle size; dav, meanparticle size; N, no. of particles).

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001 5853

The polymerization was carried out at 70 °C for 24 h. Theparticles were washed using water, methanol, THF, and acetoneand were dried.

Chromatography. The porous polymer beads (stationaryphases) were packed in semimicro-sized stainless steel columns(2 mm i.d. × 150 mm) from a slurry in water, acetonitrile, and

2-propanol. The actual composition of the slurry solvent wasvaried, depending on the properties of the packing materials. Asemimicro HPLC system, Nanospace S-1 (Shiseido Co., Tokyo,Japan), equipped with a UV detector was used for chromato-graphic evaluation of the stationary phases.

RESULTS AND DISCUSSIONSEthylene dimethacrylate (EDMA) is the most widely used

cross-linker for the preparation of methacrylate-based separationmedia for reversed-phase chromatography, because it is deemedto contribute to the overall hydrophobicity of the surface. Incontrast to this general notion, we used an apparently hydrophiliccross-linking monomer, GDMA instead. We assumed that themore hydrophobic EDMA facilitates a strong retention forhydrophobic solutes in the reversed-phase HPLC mode thatpresumably leads to the peak broadening and tailing of soluteshaving long retention times, as demonstrated in Figure 1. Incontrast, the use of more hydrophilic cross-linking GDMA,together with a variety of hydrophobic monovinyl monomers,allows for better control of the retention of hydrophobic solutes.

Stationary Phases. The volume ratios of monomers used forthe preparation of the stationary phases are listed in Table 1.Methyl methacrylate and butyl methacrylate were copolymerizedwith GDMA in a ratio of 50:50 (v:v), and 2-ethylhexyl methacrylatewas mixed with GDMA at two different levels of 20:80 or 50:50 toevaluate the effect of the concentration of the comonomer on thechromatographic properties and column efficiency. A porogenicsolvent, cyclohexanol, was used in a volume equal to that of themonomer mixture.

We attempted the preparation of alkyl brush-type stationaryphases similar to an ODS phase by copolymerization of GDMAand alkyl methacrylates. Obviously, the most hydrophobic mono-mer, 2-EHMA, best suits the purpose. Although more hydrophobicmonomers, such as long-alkyl-chain dodecyl (C12) or octadecyl(C18) methacrylates are commercially available, they cannot beeasily used in the multistep swelling and polymerization method,

Figure 2. Chromatogram of alkylbenzenes on E5 column. Columnsize, 2.0 mm i.d. × 150 mm; mobile phase, 60% (v/v) aqueousacetonitrile; flow rate, 160 µL/min; temp, 30 °C; UV detection, 254nm. Solutes: benzene, toluene, ethylbenzene, propylbenzene, butyl-benzene, and amyl(pentyl)benzene in the order of elution. (a)Determined with amylbenzene.

Figure 3. Retention profiles of M5, B5, and E5 columns. Column size, 2.0 mm i.d. × 150 mm; mobile phase, 60% (v/v) aqueous acetonitrile;flow rate, 160 µm/min; temperature,: 30 °C; UV detection, 254 nm. All solutes are mentioned in the figure.

5854 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

because they form very stable emulsions7 that make the swellingof activated seeds difficult.

Table 2 shows the yields, the mean particle sizes, and thecoefficients of variation (CV value) of the prepared packings.Interestingly, the least hydrophobic monomer, MMA, tends tolower the chemical yield of the polymer particle without affectingthe excellent monodispersity. Because we aimed at polymerpackings for semimicro HPLC, the particle size was set to be about4 µm.

Column Performance of M5, B5, and E5. Figure 2 showsthe column performance of E5. The column efficiencies20 for bothM5 and B5 exceed 11 000 plates/15-cm column (73 000 pL/m),but the column efficiency observed for E5 was 13 000 plates/15-cm column (87 000 pL/m). These numbers are sufficiently highto make the phases well-suited for the application in semimicroHPLC.

As expected, Figure 3 demonstrates that the hydrophobicselectivity (RCH2) for the packings increases in the order M5 <B5 < E5. However, relative retentivities of several solutes, whencompared with that of amylbenzene, decrease in the same orderexcept for that of the standard solute, amylbenzene. This findinglooks weird, because E5 contains the most hydrophobic alkyl

(19) Smigol, V.; Svec, F.; Hosoya, K.; Wang, Q.; Frechet, J. M. J. Angew.Makromol. Chem. 1992, 195, 151-164.

(20) Snyder, L. R.; Kirkland, J. J. Introduction to Modern Liquid Chromatography,2nd ed.; John Wiley & Sons: New York, 1979.

Figure 4. Comparison of the relative N. Chromatographic conditionsare the same as those in Figure 3.

Figure 5. Calibration curves on M5, B5, and E5. Mobile phase,100% tetrahydrofuran (THF); flow rate, 100 µm/min; temp, 30 °C; UVdetection, 254 nm; solutes, polystyrene standards and alkylbenzenes.

Figure 6. Comparison of column efficiency toward PAHs: (a)chromatogram on E5 column in 80% (v/v) aqueous acetonitrile, (b)chromatogram on the commercial polymer-based column. Numberin parentheses is relative N value (Nsolute/Nbenzene).

Figure 7. Retention profiles of M5, B5, E5, and E2 columns.Chromatographic conditions are the same as those in Figure 3.

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001 5855

group when compared with B5 and M5. Because amylbenzenewas selected as the standard in Figure 3, its longer retention timeis due to stronger hydrophobic interaction with the bulky alkylchain at the internal surface, and as a result, relative retentivitiesfor another solutes decrease. Most of the solutes tested werearomatic compounds; therefore, E5 appears to have more aliphatic-like character similar to that of alkyl-type stationary phases.21

The results mentioned above are rather important for thecomparison of relative effciencies of the columns. Figure 4presents data calculated as a ratio of efficiency for specific soluteand the efficiency found for a standard, amylbenzene (Nrel )N(solutes)/N(amylbenzene)). For example, Nrel ) 1.0 means that the samecolumn efficiency is observed for both solute and amylbenzene.These data show that M5 exhibits a low relative column efficiencyfor aromatic compounds having more phenyl rings, such aspyrene, triphenylmethane, and triphenylene. B5 and E5 havesimilar chromatographic properties; however, E5 affords the bestrelative column efficiency.

Figure 5 shows the calibration curves for these columnsobtained from size exclusion chromatography (SEC) of poly-styrene standards in tetrahydrofuran. Similarly to the finding ofFigure 4, M5 has a different calibration curve, but those for B5and E5 are similar. Unfortunately, current data do not allowdrawing any serious conclusion concerning the relation betweencolumn performance and pore size distribution.

Figure 6 shows an example of the separation of aromaticcompounds on E5, as compared to the separation using thecommercial polymethacrylate-based column, as shown in Figure1. Although peaks of a chromatogram on E5 exhibit a slightfronting, the column performance is much better than that of thecommercial column, for which a severe peak tailing is observed.This is due to the advantageous characteristics of E5 discussedabove.

Figure 8. Relationship between pore volume percent up topolystyrene Mr 2000 and relative N value. Relative N values weremeasured in 60% (v/v) aqueous acetonitrile.

Figure 9. Column performance on E5 and commercial C18 silica-based column. Column size, 2 mm i.d. × 150 mm; mobile phase, 50% (v/v)aqueous acetonitrile; flow rate, 190 µm/min; temp, 40 °C; UV detection, 254 nm. Solutes: 1, uracil; 2, caffeine; 3, 2-ethylpyridine; 4, phenol; 5,butyl benzoate; 6, benzene; 7, N,N-diethylaniline; 8, toluene; 9, phenylacetyl acetone; and 10, naphthalene. C18 column: Shiseido, CAPCELLPAK C18 UG120.

5856 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

Effect of 2-Ethylhexyl Methacrylate on the Column Per-formance. It can be seen from direct comparison of efficien-cies of E2 (8000 plates/15-cm column) and E5 (13 000) that theformer affords a lower plate number. E2 has a hydrophobicityhigher than M5 but lower than B5, as demonstrated in Figure 7.The relative retentivities are different when compared to thoseobserved for the other tested columns. For example, planarsolutes, such as pyrene and triphenylene exhibit longer retentiontimes, but sterically bulky solutes, such as triphenylmethane andtriptycene, are eluted earlier. These findings might be explainedby the effect of the cross-linking monomer.22 Similarly, the columnperformance of E2 in terms of Nrel is lower for aromaticcompounds having planar structure. This makes this columnsimilar to an M5 column. Interestingly, E2 and M5 also possessa similar pore size distribution, as determined by SEC.

Figure 8 plots a new parameter we propose to characterizethe stationary phase,23 that is, the percent volume of poresaccessible for molecules smaller than the polystyrene standards,with MW ) 2000 against Nrel values for triphenylene andtriptycene. These plots exhibit a good linearity. A SEC pore sizedistribution typical of E5, which involves a higher volume ofsmaller pores, affords columns with higher Nrel values, but a poresize distribution characteristic of M5 and E2 results in poor Nrel

values. The reason for this interesting observation remainsunclear.

The performance of the E5 column is demonstrated on theseparation of 10 solutes shown in Figure 9 and compared to atypical C18 silica-based stationary phase (Shiseido, CAPCELL PAKC18 UG120). E5 affords comparable retentivity for a number ofthe solutes under the same chromatographic conditions and a

good column efficiency. Surprisingly, phenol which is a relativelyhydrophilic solute, has a longer retention time on the E5 columnthan on the C18 stationary phase. This is probably due to thehydrophilic nature of the cross-linking monomer glyceroldimethacrylate used in E5. The column efficiency is good evenfor naphthalene, the solute that has the longest retention time.

CONCLUSIONUniformly sized polymer-based semimicro columns for reversed-

phase chromatography can be prepared using a hydrophilic cross-linking monomer together with hydrophobic alkyl methacrylates.The optimum ratio of these two types of monomers is 1:1, butmore hydrophobic alkyl methacrylates, such as butyl or 2-ethyl-hexyl methacrylat, are preferable to obtain good column efficien-cies.

Our columns E5 and B5 showed practically acceptable columnperformance, and in addition, E5 performed well even for moreretained planar aromatic compounds, such as anthracene, pyrene,and triphenylene. The column performance of E5 for a variety ofsolutes is comparable to that of a commercial silica-based C18

column and better than that of a commercial column packed withpolymethacrylate beads.

A detailed study using different polymerization systems andvariables, involving polymerization temperature, porogenic solvent,cross-linkers, and monomers is now under progress.

ACKNOWLEDGMENTFinancial support of this work by the Japanese Ministry of

Education (13640604) and the Mitsubishi Foundation is gratefullyacknowledged. We also thank Mr. Shimbo of Showa Denko Co.for his useful comments and discussions.

Received for review May 30, 2001. Accepted September17, 2001.

AC0105991

(21) Jinno, K. Chromatographic Separations Based on Molecular Recognition; Wiley-VCH: New York, 1997.

(22) Iwakoshi, Y., M.S., Kyoto Institute of Technology, Kyoto, 1999.(23) Teramachi, M., M.S., Kyoto Institute of Technology, Kyoto, 2000.

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001 5857