Preparation, Characterization, and Application ofan Enzyme-Immobilized Magnetic Microreactor forFlow Injection Analysis

Akira Nomura,† Shigemitsu Shin,† Othman Oulad Mehdi,‡ and Jean-Michel Kauffmann*,‡

National Metrology Institute of Japan, National Institute of Advanced Industrial Science and Technology, Tsukuba,Ibaraki, 305-8565 Japan, and Institut de Pharmacie, Universite Libre de Bruxelles, Campus Plaine 205/6,1050 Brussels, Belgium

Enzyme-immobilized magnetic microparticles (EMMP)have been prepared for use as a microreactor in flowinjection analysis (FI). The microparticles were directlyinjected into the FI system. Their retention occurredwithin the flow line by small permanent magnets locatednear the detector. The analytical utility of this concept wasillustrated by the assay of glucose using glucose oxidase(GOx), immobilized microparticles, and amperometricdetection of liberated hydrogen peroxide. The micropar-ticles were derived from silica gel (nominal pore diameter,15-80 nm) by impregnation with a citric acid/ethanolsolution and a ferric nitrate/ethanol solution and then bycalcination in a nitrogen atmosphere to produce ferrimag-netic fine particles of spinel-type iron oxide (γ-Fe2O3)inside the pore. They were characterized by X-ray diffrac-tion. The calibration curve of the glucose sample (2 µLinjected) was linear between 2.5 × 10-6 and 5 × 10-4

mol/L (R ) 0.9995), and the detection limit was 1.0 ×10-6 mol/L or 0.36 ng of injected glucose (S/N ) 3). Therepeatability for a 5 × 10-4 mol/L glucose solution wasRSD ) 1.5% (n ) 6). Application to the assay of glucosein a fermentation broth is illustrated. The GOx MMP werestable and active for more than eight months when keptat 10 °C.

With the increasing search for developing microfluidic systems,magnetic beads have attracted considerable interest since theymay serve as a solid support in a variety of applications, includingimmunoassays,1-4 support for enzymes,2,5 oligonucleotides,6 and

bioaffinity sorbents.2 Different magnetic microbeads are com-mercially available; they generally consist of iron oxide micro-particles silanized or coated with thin layers of polystyrene andfunctionallized for subsequent biocomponent immobilization.1-6

The magnetic beads may be retained at any location within amicroflow injection manifold1 or within a capillary electrophoresissetup2, i.e., flow systems where narrow fluidic paths and low flowrates are encountered. The use of magnetic beads does not requirefrits and allows an easy on-line renewal of the solid phase. Enzyme-immobilized microreactors offer several useful properties in termsof conversion efficiency and biocomponent stability for numerousanalytical applications.7 Recently, enzyme and antibody biomol-ecules entrapped in nanometer-sized silica-coated magnetite wereprepared by sol-gel technology and applied to batch fluorometricassays.8 Relatively few examples using magnetic microparticles(MMP) with immobilized enzymes in flow injection systems havebeen described.9 In this study, we inject, in the same manifold,the glucose oxidase (GOx) MMP and then the analyte. Of equalinterest in the field of enzymatic micoreactors is the synthesis ofhighly porous MMP allowing for a high surface activity and highbiocomponent loading. Illustration of the potentiality of thesuggested approach is provided by studying GOx as a model forimmobilized biocomponent and for glucose assays in complexsamples.

â-D-Glucose is converted to hydrogen peroxide and glucono-lactone by GOx. Many flow analysis methods using this reactionhave been implemented for the determination of glucose. Glucosewas determined by flow injection analysis (FI) using GOx-immobilized controlled-pore glass. The produced hydrogen per-oxide reacted with luminol and was detected by a chemilumines-cence detector.10 Electrochemical detection of hydrogen peroxidemay also be advantageously considered.11-13 Applications havebeen described using liquid chromatography with a postcolumnimmobilized GOx reactor followed by amperometric detection.14

* To whom correspondence should be addressed. Tel.: 32 2 6505215. Fax:32 2 6505225. E-mail: jmkauf@ulb.ac.be.

† National Institute of Advanced Industrial Science and Technology.‡ Universite Libre de Bruxelles.

(1) Pollema, C. H.; Ruzicka, J.; Christian, G. D.; Lernmark A. Anal. Chem., 1992,64, 1356-1361.

(2) Rashkovetsky, L. G.; Lyubarskaya, Y. V.; Foret F.; Hughes D. E.; Karger,B. L. J. Chromatogr., A 1997, 781, 197-204.

(3) Thomas, J. H.; Kim, S. K.; Hesketh, P. J.; Halsall, H. B.; Heineman, W. R.Anal. Chem. 2004, 76, 2700-2707.

(4) Hayes, M. A.; Polson, N. A.; Phayre A. N.; Garcia, A. A. Anal. Chem. 2001,73, 5896-5902.

(5) Choi, J. W.; Oh, K. W.; Thomas, J. H.; Heineman, W. R.; Halsall, H. B.;Nevin, J. H.; Helmicki, A. J.; Henderson, H. T.; Ahn, C. H. Lab Chip 2002,2, 27-30.

(6) Wang, J. Anal. Chim. Acta 2003, 500, 247-257.

(7) Crouch, S. R.; Shi, Y. Anal. Chim. Acta 1999, 381, 165-173.(8) Yang, H.-H.; Zhang, S.-Q.; Chen, X.-L.; Zhuang, Z.-X.; Xu, J.-G.; Wang, X.-R.

Anal. Chem. 2004, 76, 1316-1321.(9) Banu, S.; Greenway, G. M.; McCreedy, T.; Shaddick, R. Anal. Chim. Acta

2003, 486, 149-157.(10) Swindlehurst, C. A.; Nieman, T. A. Anal. Chim. Acta 1988, 205, 195-205.(11) Thevenot, D. R.; Sternberg, R.; Coulet, P. R.; Laurent, J.; Gautheron, D. C.

Anal. Chem. 1979, 51, 96-100.(12) Masoom, M.; Townshend, A. Anal. Chim. Acta 1984, 166, 111-118.(13) Petersson, B. A. Anal. Chim. Acta 1988, 209, 231-237.

Anal. Chem. 2004, 76, 5498-5502

5498 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004 10.1021/ac049489v CCC: $27.50 © 2004 American Chemical SocietyPublished on Web 07/29/2004

In the present work, glassy carbon or platinum electrodes wereused for the detection of liberated hydrogen peroxide. Bothoperated at a relatively high positive potential yet more sensitiveand selective electrochemical detectors may readily be imple-mented depending on the purpose of the assay.15

EXPERIMENTAL SECTIONPreparation of Magnetic Porous Particles. Seven kinds of

Super Micro Bead Silica Gel, each having average particlediameter of 10 µm, were purchased from Fuji Silysia ChemicalLtd. (Kasugai, Japan) to determine the preparation condition ofthe magnetic particles. Their pore structures are summarized inTable 1. Super Micro Bead Silica Gel having an average particlediameter of 5 µm and nominal pore diameter of 30 nm (specificsurface area, 103 m2/g; pore volume, 1.15 mL/g; mean porediameter, 30.1 nm) was also purchased from Fuji Silysia ChemicalLtd. for the preparation of GOx MMP. The mean pore diameterof the silica gel was calculated from the specific surface area andthe pore volume determined by the BET adsorption method ormercury intrusion method.

The magnetic silica gels supporting defect spinel-type ironoxide particles inside the pore were prepared as follows. Asuspension of the silica gel microbeads (0.5 g) shown in Table 1in a 0.75 mol/L ethanol solution of citric acid (3 mL) was sonicatedfor 30 min. After centrifuging, the silica gel was dried at 60 °C for12 h and then sonicated for 30 min in a 1 mol/L Fe(NO3)3‚9H2Oethanol solution (3 mL). After being centrifuged and dried underthe same conditions again, the silica gel was placed in an aluminaboat and calcinated in a quartz tube at 400 °C for 1 h in a streamof nitrogen. An immediate color change of the silica gel from black-brown to orange red was observed when the boat was taken outfrom the quartz tube for cooling.

Elemental analyses were conducted using a CHN analyzer typeEA 1110 from CE Instrument (Milan, Italy). Magnetization of thesamples was measured at room temperature by using a samplevibrating magnetometer (TM-VSM1550HGC) from TamagawaSeisakusho (Sendai, Japan) in the magnetic field range of (15kOe. The powder X-ray diffraction patterns of silica gel supportingiron oxide inside the pore were recorded on a X-ray diffractometerfrom Mac Science (Yokohama, Japan) using Cu KR radiation. Theelectron micrographs were obtained by a Hitachi S-800 scanningelectron microscope or a Hitachi 2000 transmission electron

microscope (TEM) both from Hitachi, Ltd. (Tokyo, Japan). The57Fe Mossbauer spectroscopic measurements were made on aconstant-acceleration-type instrument, which has been describedpreviously.16 The isomer shifts are reported relative to metalliciron foil.

Preparation of GOx MMP. Aminopropyltriethoxysilane waspurchased from Chisso, Ltd. (Minamata, Japan), glutaraldehydesolution (∼25%) was from Wako Pure Chemicals Industries, Ltd.(Osaka, Japan), and GOx (from Aspergillus niger) was from SigmaChemical Co. (St. Louis, MO). The immobilization of GOx wasadapted from the literature.12,14,17 Briefly, the magnetic silica gelobtained according to this method (0.5 g) was taken into a flat-bottom flask equipped with a water condenser, added to a 10 wt% aminopropyltriethoxysilane solution in toluene (20 mL), andheated under reflux for 2 h. Then, after vacuum filtration, thesuspension was washed with benzene (30 mL) and dichlo-romethane (30 mL), each of which was repeated for three times,and dried in the vacuum drier at room temperature overnight.Glutaraldehyde (2 wt %) was reacted with the particle suspensionfor 2 h, rinsed thoroughly, and reacted with a 2 mg/mL GOx,0.03 mol/L phosphate buffer solution for 8 h at 4 °C. Thesuspension was placed in a glass filter, separated from the residualGOx solution by vacuum filtration, and washed with 30 mL ofdistilled water five times. Besides this magnetic silica gel, sevenkinds of untreated bare silica gel were freshly immobilized withGOx according to the same procedure, for the evaluation of theGOx MMP thus obtained.

Procedure for FI of Glucose on GOx MMP. The schematicdiagram of the FI system is illustrated in Figure 1. The GOx MMPwere obtained from the silica gel having an average particlediameter of 5 µm and nominal pore diameter of 30 nm (specificsurface area, 103 m3/g; pore volume, 1.15 mL/g; mean porediameter, 30.1 nm). â-D-Glucose was purchased from Carlo Erba(Milan, Italy), and hydrogen peroxide was from VWR Int. (Leuven,Belgium) whose content was determined by titration using apotassium permanganate standard solution. Pump 1 (P1) for largerflow rates was a solvent delivery system model PM-80 from BAS(West Lafayette, IN). Pump 2 (P2) for lower flow rates was a single-syringe type BEE syringe pump (BAS). The potentiostat was aLC-4B amperometric detector (BAS). The glassy carbon orplatinum electrodes were poised at +1.000 or + 0.650 V,respectively. Injector 1 (I1) for slurry injection was a Rheodynetype 7125 injector with a 20-µL sample loop. Injector 2 (I2) forsample injection was Rheodyne type 7725(i) with an injection loop(∼2 µL) made of Teflon tubing (0.17 mm i.d. × 90 mm length).The mobile phase for the FI was a 0.1 mol/L phosphate buffersolution adjusted to pH 4.5.

(14) Murakami, K.; Kakemoto, M.; Harada, T.; Yamada, J. Bunseki Kagaku 1991,40, 125-129.

(15) Kauffmann, J.-M.; Guilbault, G. G. In Methods of Biochemical Analysis: Bio-Analytical Applications of Enzymes; Suelter C., Kricka, L.-J, Eds.; J. Wiley:New York, 1992; Vol. 36, pp 63-113.

(16) Iijima, S.; Nomura, A.; Mizukami, F.; Shin, S.; Mizutani, F. J. Radioanal.Nucl. Chem. 1999, 239, 297-302.

(17) Weibel, M. K.; Dritschilo, W.; Bright, H. J.; Humphrey, A. E. Anal. Biochem.1973, 52, 402-414.

Table 1. Pore Structure of Raw Silica Gel

nominal porediameter (nm)

specific surfacearea (m2/g)

pore volume(mL/g)

mean porediameter (nm)

3 669 0.60 3.67 535 0.83 6.2

10 344 1.14 13.315 173 1.21 16.230 101 1.14 34.250 65 1.00 48.380 36 0.83 82.1

Figure 1. Schematic diagram of magnetic microflow system: P1,conventional pump; P2, syringe pump; I1, slurry injector; I2, sampleinjector; D, electrochemical detector.

Analytical Chemistry, Vol. 76, No. 18, September 15, 2004 5499

Permanent magnet (i.d. 6 × 5 mm) used was a Neody Magnet(Nd-Fe-B magnet) purchased from As One, Ltd. (Osaka, Japan).One pair of these permanent magnets was assembled using a yokeiron supporter so that the gap between the pair was just the samesize as the outer diameter of the Teflon tube (∼1.6 mm) for fixingthe magnetic particles inside the tube (Figure 2). The magneticflux density measured at the center between the pair of themagnets using the Gauss/Tesla meter model 5080 made by Sypris(Orlando, FL) was 8.65 kG. The Teflon tube used for the fixationof the magnetic particles was 1.6-mm o.d., 0.25-mm i.d., and 150-mm length. The concentration of the slurry applied to the magnetvia I1 with a 20-µL sample loop was 10 µg/µL GOx MMP; thus,0.2 mg of GOx MMP was loaded into the manifold.

After stabilization of the system, by flowing the mobile phaseat the flow rate of 100 µL/min using P1, the slurry of the magneticparticles was loaded from the I1 using P2 at a flow rate of 50 µL/min and retained at the magnet part. When the EMMP wereretained between the pair of magnets inside the Teflon tube, theflow rate was adjusted for the analysis of glucose samples. Todetermine the conversion efficiency of glucose by GOx MMP,glucose and hydrogen peroxide solutions were prepared usingthe same solvent as the mobile phase. After baseline stabilizationof the electrochemical detector, the I1 was fixed at the loadposition, and glucose was injected through the I2 injection valve(∼2 µL).

Glucose Assay in a Fermentation Broth. A fermentationbroth YEPG based on 10 g/L yeast extract (Merck, Darmstadt,Germany), peptone from soymeal 10 g/L (Merck), and D-glucose75 g/L (Merck) kept at -20 °C was defrosted and diluted 104

times with the mobile phase before injection into the FI manifold.The glucose assay was performed by the method of standardadditions.

RESULTS AND DISCUSSIONPreparation of Magnetic Porous Particles. The specific

surface area of original silica gels (untreated) and after calcinationtreatments are presented in Table 2. In the case of the silica gelhaving smaller pore diameters such as 3 and 7 nm, the specificsurface area after treatment decreased compared with the originalstructure, presumably because of the iron oxide formed duringthe calcination step. Yet for silica gel having larger pore diameters,

namely, above 10 nm, the specific surface area before and aftertreatment was similar.

The powder X-ray diffraction (XRD) patterns of the treatedsilica gels are shown in Figure 3. The broad peaks around 2θ )25° are of silica gel base. The cell parameter for the cubic latticeof the nominal pore diameter 50 nm was 8.374 Å. This valuewas intermediate between those of Fe3O4 (8.396 Å) and γ-Fe2O3

(8.352 Å),16,18 which implies that the iron oxide obtained in thepresent work was a solid solution of the system Fe3O4-γ-Fe2O3

with the defect spinel structure, which has cation vacancies inthe octahedral sites. We call the iron oxide thus obtained γ-Fe2O3

or simply maghemite, though the plausible composition is(Fe3+)[Fe3+

5/3Fe2+1/6[ ]1/6]O4 according to the Vegard’s law. The

size of the γ-Fe2O3 produced was estimated by these XRD patterns.From the analysis of full width at half-maximum (fwhm) of thepeak around 2θ ) 38° corresponding to the (311) plane diffraction,the particle sizes of the γ-Fe2O3 were estimated using the equationshown as follows and summarized in Table 3.

where Dhkl is the crystallite size (nm) perpendicular to the (hkl)plane, λ is the wavelength (nm) of the X-ray radiation (0.154 nmfor Cu KR), â is the breadth (rad) of the diffraction peak, θ is theBragg angle (deg), K is the shape factor of the average crystallite(expected shape factor is 0.9), â ) B - b (when the size ofcrystallite is smaller than 50 nm), B is the fwhm of the relevant

(18) Powder Diffraction File, Inorganic and organic, Set 47, International Centerfor Diffraction Data, 1997.

Figure 2. Photograph of the magnet assembly for keeping themagnetic particles inside the Teflon tube.

Table 2. Effect of Magnetic Treatment on PoreStructure

specific surface area (m2/g)nominal porediameter (nm) original treatment

3 669 5037 535 419

10 344 32415 173 16030 101 9750 65 7780 36 44

Figure 3. Powder XRD pattern of treated silica gel.

Dhkl ) Kλ/â cos θ

5500 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

diffraction peak, and b is the fwhm of the diffraction peak of thestandard material (silicon was used in this experiment, 2θ )28.486°, b ) 0.00256 rad).

These results indicated that the pore size of the silica gelcontrols the particle size of the iron oxide formed.

The presence of citric acid was essential to produce theγ-Fe2O3. When the calcination treatment was carried out withoutthe citric acid impregnation, formation of R-Fe2O3 was observedby XRD analysis.

The magnetization degree of these products determined at15kOe using a vibrating sample magnetometer is presented inTable 3, together with the particle sizes of the γ-Fe2O3 inside thepore of the silica gel. From these results, it appears that γ-Fe2O3

particles having a larger diameter were produced in the largerpores of silica gels, and the larger γ-Fe2O3 particles exhibited ahigher magnetization. The products from the nominal porediameters of 3, 7, and 10 nm did not show saturation in themagnetization curves, which implies that these γ-Fe2O3 particlesshow a superparamagnetic property. This result was also ascer-tained by the 57Fe Mossbauer spectroscopic measurement,16,19 inwhich no magnetic hyperfine structure was observed. Theproducts from the nominal pore diameter of 15, 30, 50 ,and 80nm, on the other hand, showed the degree of magnetizationproperty in the curves, which implies that γ-Fe2O3 particles above∼6 nm of particle size show a ferrimagnetic property from thisexperiment. These ferrimagnetic products are attracted by mag-nets and can be used advantageously in this study as the substratefor the immobilization of GOx.

The TEM images of the original and magnetic silica gels whosenominal pore diameters were both 50 nm are shown in Figure 4.The small particles are observed in the image (Figure 4b). Theparticles were estimated to be ∼8.8 nm in average diameter, andthe result was comparable to that from the XRD (9.0 nm). Thereare no such particles observed in the image (a) of the originalsilica gel.

Preparation of GOx MMP. The process for the immobiliza-tion of GOx was traced by elemental analysis. In the modificationby aminopropyltriethoxysilane and GOx, nitrogen content wasused for calculation of the amount of aminopropyl group (AP)and GOx introduced. Carbon content was used for calculation ofthe amount of glutaraldehyde (GA) introduced. The amount ofGOx introduced was estimated from the molecular weight of GOx,186 000. The immobilization of GOx was done for both originalsilica gel (bare) and magnetic silica gel having γ-Fe2O3 particleswhose nominal pore diameters were 15, 30, 50, and 80 nm,respectively. The results are summarized in Table 4. The efficiency

of the immobilization process was similar between the bare silicagel and the magnetic one. More aminopropyl groups wereintroduced in the smaller pore diameter, because it had a largersurface area and accordingly more silanol groups for reaction withthe aminopropyl group. Molar ratio of glutaraldehyde withaminopropyl group (GA/AP) was found to be ∼1:1 for all thesamples from these results. The reaction of GOx with glutaral-dehyde (GOx/GA) occurred more readily in the silica gels havinga larger pore diameter, presumably because GOx can more readilyreach GA molecules entrapped inside the silica gel. Glutaralde-hyde is a small molecule, and its diffusion was likely not hamperedby the pore size of silica gel. The hydrodynamic diameter of GOxis 7.6 nm;20 thus, GOx should also readily penetrate into the poresof the magnetic particles (nominal pore diameter, 30 nm), butGOx adsorption phenomena onto the silica surface20 may hinderits diffusion into particles with small pore diameter.

Procedure for FI of Glucose. The GOx MMP having anaverage particle diameter of 5 µm and a nominal pore diameterof 30 nm were used for the glucose assay. A photographic imageof the magnets and the magnetic particles kept inside the Teflontube by the pair of the magnets is shown in Figure 2. Visualobservation of the magnetic particles inside the tube indicatedthat they were distributed slightly toward the left from the centerof the magnets. This can be explained because the mobile phaseflowed into the magnet from the right side. It was noticed that, atflow rates higher than 250 µL/min, the magnetic particles werewashed away from the magnets.

The effect of the flow rate of a mobile phase on the conversionefficiency of glucose to H2O2 by the GOx MMP is shown in Figure5. The conversion efficiency was obtained by comparing the peakheight of glucose with that of standard H2O2 (peak height wasused since both peaks have about the same half-widths). It wasenhanced as the flow rate decreased because of a longer contacttime of the sample and the enzymatic microreactor. A flow rateof 5 µL/min was usually used in this study. The calibration curveof the glucose sample obtained from the peak height response atthe glassy carbon electrochemical detector was linear between2.5 × 10-6 and 5 × 10-4 mol/L (R ) 0.9995), and the detection

(19) Longworth, G.; Tite, M. S. Archaeometry 1977, 19, 3-14.(20) Kamyshny, A.; Feldman, A.; Baszkin, A.; Boissonnade, M. M.; Rosilio, V.;

Magdassi, S. J. Collloid. Interface Sci. 1999, 218, 300-308.

Table 3. Particle Size and Saturation Magnetization ofMaghemite Supported on Silica Gel

nominal pore diameter (nm)

3 7 10 15 30 50 80

particle size (nm) 4.5 4.2 6.2 6.9 9.0 9.5saturation 0.24 1.85 2.43 6.95 7.58 8.46 7.53magnetizationat 15 kOe (Am2/kg)

Figure 4. TEM image of silica gel (a) and magnetic silica gel (b).Nominal pore diameter of the silica gel, 50 nm.

Analytical Chemistry, Vol. 76, No. 18, September 15, 2004 5501

limit was 1 × 10-6 mol/L and 0.36 ng injected sample (S/N ) 3).It is worth recalling that other modified working electrodes mayoffer better sensitivity and selectivity for hydrogen peroxide thanthe glassy carbon electrode.21

A typical response profile for the three repeated injectionsusing a glucose concentration of 5 × 10-4 mol/L is shown inFigure 6. The repeatability obtained for injections of glucose withthe same microreactor gave an RSD of 1.5% (n ) 6). Thereproducibility of glucose injection with different injected microre-actors was 2.4% (n ) 6). This higher value may result from therelatively fast microparticle deposit formation in the syringe forinjection. Work is under investigation to improve the MMPsuspension. Interestingly, renewal of the MMP can be performedwhen needed by simply applying a flow rate above 250 µL/min.

The platinum electrode was operated at +0.650 mV during threemonths of continuous use with no special surface treatment otherthan a daily cleaning by a flow of methanol/water (1:1 v/v) in theFI manifold (without the EMMP). The shelf life of the GOx MMP(kept at 10 °C in a refrigerator) was studied over a period of eightmonths. Response fluctuations were observed, but they were notsignificantly higher than those corresponding to the experimentalerror (RSD ) 2.4%). This suggested a high stability of the GOxMMP. The FI was evaluated for the assay of glucose in afermentation broth. High dilution was allowed in order to adjustthe glucose concentration to the linear domain of the calibrationcurve and in order to minimize interferences and protein foulingat the platinum electrode. The flow rate was adjusted to 50 µL/min. A glucose concentration of 72.58 g/L was obtained, whichcompared favorably with the value of 72.53 obtained by a referencemethod using the Cobas Integra setup (Roche Diagnostics, Basel,Switzerland). Under our experimental conditions, no interferingsignal was noted, yet applications to more complex samples, withinterfering species, may be readily achieved by subtracting thesignal obtained without the EMMP from the signal obtained withthe EMMP in the FI setup.

CONCLUSIONSSilica-based magnetic microparticles with a high pore density

offer an attractive environment for the immobilization of a highdensity of GOx molecules and should be equally suitable fornumerous other enzyme and biocomponent immobilizations formicrobioreactor construction. The relatively small size (5 µm) ofthe investigated particles may be exploited in various microflowsystems. The enzyme-immobilized magnetic microparticles canbe readily injected into the flow analysis manifold, and themicroreactor can be located in proximity to the detector forminimizing dead volumes. It may be flushed away when requiredsimply by removing the permanent magnets or by increasing theflow velocity. The minute amount of microbioreactor consumedmakes the methodology economically interesting. The presentwork utilized an electrochemical detector, but optical detectorswould equally fit into the microflow setup.

ACKNOWLEDGMENTThis research was financially supported by the Ministry of

Economy, Trade and Industry, and the Ministry of Environmentof Japan.

Received for review April 2, 2004. Accepted June 28,2004.

AC049489V(21) Serradilla Razola, S.; Aktas, E.; Vire J.-C.; Kauffmann, J.-M. Analyst 2000,

125, 79-85.

Table 4. Immobilization of GOx on the Surface of Silica Gel

nominal pore diameter (nm)

15 30 50 80

bare magnet bare magnet bare magnet bare magnet

AP (mmol/g) 0.953 0.791 0.570 0.639 0.495 0.612 0.263 0.296GA (mmol/g) 0.966 0.873 0.566 0.538 0.486 0.530 0.261 0.264GOx (µmol/g) 0.061 0.071 0.178 0.175 0.182 0.167 0.206 0.238GA/AP 0.963 1.108 1.034 0.879 1.044 0.916 1.026 0.920GOx/GA (×10-6) 63 82 320 325 375 315 790 901

Figure 5. Effect of flow rate on the conversion efficiency of glucoseto H2O2 by GOx IMMP.

Figure 6. Example chart of the response by electrochemicaldetector repeated 3 times of injection of the sample. Sample: 5 ×10-4 mol/L â-D-glucose, 2 µL.

5502 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004