Photomodification of Polymer Microchannels Induced by Static and

This paper presents a study of polymer surfaces modified by laser ablation using poly(ethylene terephthalate) (PET) as a model system. The surface pro...
0 downloads 0 Views 340KB Size
Download PDF
Anal. Chem. 2001, 73, 3845-3853

Photomodification of Polymer Microchannels Induced by Static and Dynamic Excimer Ablation: Effect on the Electroosmotic Flow F. Bianchi,† Y. Chevolot,‡ H. J. Mathieu,‡ and H. H. Girault†

Laboratoire d’Electrochimie and Laboratoire de Me´ tallurgie Chimique, EÄ cole Polytechnique Fe´ de´ rale de Lausanne, CH-1015 EPFL Lausanne, Switzerland

This paper presents a study of polymer surfaces modified by laser ablation using poly(ethylene terephthalate) (PET) as a model system. The surface properties induced by static and dynamic ablation with the 193-nm pulsed radiation of an ArF excimer laser (4 × 107 W/cm2) in air have been successfully used to control the electroosmotic flow (EOF) in photoablated PET microchannels. Through the creation of well-defined static ablation patterns onto the walls of a trapezoidal channel, it was found that the resulting reduction in the EOF could be controlled. For example, a reduction of 25% in the EOF was observed in 42-µm-deep microchannels when using a static ablation pattern treating 50% of the total wall surface area. A numerical study describing the fluidic behavior induced by a static pattern is also presented. Moreover, X-ray photoelectron spectroscopy has been used to point out surface changes between static and dynamic ablation, thereby demonstrating an ability to create new functionalities in microchannels by laser treatment.

I. INTRODUCTION With the increasing popularity of micro-total analysis system (µ-TAS) 1-3 and micro-flow injection analysis (µ-FIA),4,5 commercial microchips performing diverse medical and biological analyses integrated into a single device are now available.6 There is also an increasing interest in mass-produced plastic microchips using low cost fabrication methods such as molding,7-9 imprinting,10 or †

Laboratoire d’Electrochimie. Laboratoire de Me´tallurgie Chimique. (1) Shoji, S. Topics in Current Chemistry 1998, 194, 163-188. (2) Vandenberg, A.; Lammerink, T. S. J. Topics in Current Chemistry 1998, 194, 21-49. (3) Schwarz, A.; Rossier, J. S.; Bianchi, F.; Reymond, F.; Ferrigno, R.; Girault, H. H. In Micro Total Analysis Systems ’98; Harrison, D. J., van den Berg, A., Eds.; Kluvert Academic Publishers: Dordrecht, 1998; pp 241-244. (4) Effenhauser, C. S. Topics Curr. Chem. 1998, 194, 51-82. (5) Haswell, S. J. Analyst 1997, 122, R1-R10. (6) Harrison, D. J.; Wang, C.; Thibeault, P.; Ouchen, F.; Cheng, S. B. In Micro Total Analysis Systems 2000; van den Berg, A., Olthuis, W., Bergveld, P., Eds.; Kluvert Academic Publishers: Dordrecht/Boston/London, 2000. (7) Callenbach, T. Chimia 1999, 53, 72-74. (8) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974-4984. (9) McCormick, R. M.; Nelson, R. J.; Alonso-Amigo, G. M.; Benvegnu, D. J.; Hooper, H. H. Anal. Chem. 1997, 69, 2626-2630. ‡

10.1021/ac010262z CCC: $20.00 Published on Web 07/10/2001

© 2001 American Chemical Society

plasma etching.11 All of these different techniques, in addition to mask projection excimer laser ablation and the large choice of polymer substrates, result in very different electroosmotic flow (EOF) velocities in polymer microchannels.12-14 The control of the EOF in free-zone capillary electrophoresis applications is a fundamental condition necessary to enhance the reproducibility and the separation performance.15 To date, several physicochemical methods to modify the EOF have been presented in the literature.16 In addition, the control of electroosmotic flow by electronic means for capillary zone electrophoresis has been demonstrated using radial electric fields,17,18 and this method has been applied in a capillary microdevice.19 In microchannel structures, the electroosmotic flow can also be monitored by working with buffer solutions at different pHs or ionic strengths13 and by changing the proportion of the materials forming a composite microchannel;14 however, in capillary electrophoresis (CE) applications, the pH is mainly fixed by the conditions of the electrophoretic separation. The ionic strength is chosen to prevent the Joule heating effect and analyte walls interaction. UV laser ablation combined with a projection mask system has been shown to be a promising technique for the production of prototype polymer microchips thanks to its very flexible and fast fabrication procedure.3,11,13,14,20 In addition, in combination with surface blocking methods, the above technique can be used to pattern proteins on the surfaces of polymer substrates, such as PET.21 Moreover, recent studies on the surface physical chemistry (10) Martynova, L.; Locascio, L. E.; Gaitan, M.; Kramer, G. W.; Christensen, R. G.; Maccrehan, W. A. Anal. Chem. 1997, 69, 4783-4789. (11) Rossier, J. S.; Schwarz, A.; Bianchi, F.; Reymond, F.; Ferrigno, R.; Girault, H. H. In Micro Total Analysis Systems 2000; van den Berg, A., Olthuis, W., Bergveld, P., Eds.; Kluvert Academic Publishers: Dordrecht/Boston/ London, 2000, pp 159-162. (12) Locascio, L. E.; Perso, C. E.; Lee, C. S. J. Chromatogr. A 1999, 857, 275284. (13) Roberts, M. A.; Rossier, J. S.; Bercier, P.; Girault, H. Anal. Chem. 1997, 69, 2035-2042. (14) Bianchi, F.; Wagner, F.; Hoffman, P.; Girault, H. Anal. Chem. 2001, 73, 829-837. (15) Gas, B.; Stedry, M.; Kenndler, E. Electrophoresis 1997, 18, 2123-2133. (16) Monnig, C. A.; Kennedy, R. T. Anal. Chem. 1994, 66, R280-R314. (17) Lee, C. S.; Blanchard, W. C.; Wu, C. T. Anal. Chem. 1990, 62, 1550-1552. (18) Hayes, M. A.; Ewing, A. G. Anal. Chem. 1992, 64, 512-516. (19) Polson, N. A.; Hayes, M. A. Anal. Chem. 2000, 72, 1088-1092. (20) Rossier, J. S.; Schwarz, A.; Reymond, F.; Ferrigno, R.; Bianchi, F.; Girault, H. H. Electrophoresis 1999, 20, 727-731. (21) Schwarz, A.; Rossier, J. S.; Roulet, E.; Mermod, N.; Roberts, M. A.; Girault, H. H. Langmuir 1998, 14, 5526-5531.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001 3845

of PET substrates after UV laser ablation have pointed out differences between static and dynamic ablation, such as wetability and surface structure.22-24 When ablation is carried out under an ambient atmosphere, the formation of particles is enhanced. This results in a separation of the ablation products into rapidly ejected molecules, responsible for the formation of the shock/blast wave, and slow particulate products, which can redeposit onto the substrate.25,26 The deposition of strongly adherent debris27 can be considered as surface modification in the sense that the debris chemical composition is significantly different from the native polymer substrate.28,29 During dynamic ablation, the ejected debris is allowed to redeposit onto the channel surfaces. Because of the constant scanning of the substrate with respect to the beam, a fraction of the ejected particles accumulate homogeneously over the whole channel surface. The induced channel depth obtained by dynamic ablation can be calculated using eq 1.

dtot ) nh(Φ)

(1)

where n is the number of pulses and h(Φ) is the ablated thickness per pulse at the fluence Φ. The appropriate number of pulses, n, is obtained using eq 2.

n)

υa vscan

(2)

where υ is the pulse repetition rate; a, the spot length; and vscan, the scanning speed. In static ablation, each of the n pulses hits the same surface spot, and thus, debris cannot accumulate on the ablated surface. Only the debris of the last pulse will deposit on the irradiated section, whereas most of the debris will accumulate around the irradiated surface.26 The induced depth can be described using eq 1. In the present report, we will demonstrate the use of UV laser ablation to perform several operations during microchip fabrication. Static ablation is used to remove the layer of debris left in the microchannel during dynamic ablation. By controlling which areas of the microchannel are subjected to static ablation, welldefined periodic patterns can be created on the walls of the microchannel, which is used to control the EOF. The resulting surface modifications were then characterized using scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS). As an application to the present physical microchip modification, we present a T-intersection on a microchip, where pressure-induced flow in a field-free channel is demonstrated using a methodology similar to that presented by Ramsey and Ramsey.30 (22) Wagner, F.; Hoffmann, P. Appl. Surf. Sci. 2000, 154, 627-632. (23) Wagner, F.; Hoffmann, P. Appl. Phys. A, Mater. Sci. Process. 1999, 69, S841S844. (24) Rossier, J. S.; Bercier, P.; Schwarz, A.; Loridant, S.; Girault, H. H. Langmuir 1999, 15, 5173-5178. (25) Lazare, S.; Granier, V. Chem. Phys. Lett. 1990, 168, 593-597. (26) Kelly, R.; Miotello, A.; Braren, B.; Otis, C. E. Appl. Phys. Lett. 1992, 60, 2980-2982. (27) Burns, F. C.; Cain, S. R. 1996, 29, 1349-1355. (28) Srinivasan, R.; Mayne-Banton, V. Appl. Phys. Lett. 1982, 40, 576. (29) Rossier, J. S.; Girault, H. H. Phys. Chem. Chem. Phys. 1999, 1, 3647-3652. (30) Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 1997, 69, 1174-1178.

3846

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

T-junctions can be used for mass spectrometry injection31,32 or for electric field decoupling.33 II. EXPERIMENTAL SECTION 1. Microchannels Prototyping. The microstructures were fabricated using ArF excimer (Lambda Physik, LPX 205-i, Germany) laser ablation at 193 nm and sealed using a low-temperature lamination technique.13 The ablation setup is a standard maskprojection arrangement in which the image of a free-standing molybdenum (Goodfellow, U.K.) mask with a rectangular hole is projected onto the substrate. The rectangular hole of the projection mask (10 × 0.5 mm) was obtained by laser-cutting a molybdenum foil (20 × 20 × 0.05 mm) using an experimental prototype laser system (Nd:YAG, type SLAB pumped by flash lamps). A poly(ethylene terephthalate) (PET) substrate (thickness, 100 µm; Melinex, grade S; ICI; U.K.) was used. The substrate is placed on computer-controlled x-y translation stages (Physik Instrumente; Germany) and scanned under the excimer laser beam (pulse repetition rate, 50 Hz; fluence, 850 mJ/cm2) to produce 2-cm-long geometrically well-defined trapezoidal microchannels.14,34 The patterning process is performed immediately after the dynamic channel fabrication with the same laser conditions. The microrectangular patterns, schematized in Figure 1, are produced in the microchannels by static ablation through the same projection mask that is used for the dynamic channel fabrication. This produces a well-defined pattern, as illustrated in Figure 2a. The floor roughness observed in Figure 2b after static (surface S) and dynamic (surface D) ablation at fluences well above the ablation threshold on commercial PET (Melinex, grade S) has been already reported in previous publications.22,23 It can be attributed to stresses in stretched PET foils.28,35 Once the patterning is finished, the channels are first rinsed with distilled water and then sealed by a lamination machine (Morane; U.K.). The set of microchannels studied were produced with a range of different surface ratios, sr ) sstat/stot, where sstat is the statically ablated surface and stot, the total surface of the microchannel. The distance, lp, between two static patterns is restricted by the debris distribution around the static shots.26 The minimum lp value was estimated by SEM imaging (not shown here) to be ∼250 µm for the projection mask used in this work. Consequently, the maximum value of the ratio sr is ∼0.5 for 42µm-deep channels, considering that the sealing roof (L in Figure 1) cannot be patterned. The different structure dimensions presented in the caption for Figure 1 have been measured by inspection using a 2-D metrology microscope equipped with a digital camera (Nikon Coolpix; Japan). 2. Electroosmotic Flow Measurements. The electroosmotic flow in the microchannels was investigated using Huang’s current monitoring method.36 The flow measurements were performed using the same procedure already described in previous publications.13,14,37 The running acetate buffer with a pH 4.66 (acetic acid (31) Lazar, I. M.; Ramsey, R. S.; Sundberg, S.; Ramsey, J. M. Anal. Chem. 1999, 71, 3627-3631. (32) Lazar, I. M.; Ramsey, R. S.; Jacobson, S. C.; Foote, R. S.; Ramsey, J. M. J. Chromatogr. A 2000, 892, 195-201. (33) Huang, X.; Zare, R. N. Anal. Chem. 1990, 62, 443-446. (34) Rossier, J. S.; Ferrigno, R.; Girault, H. J. Electroanal. Chem. 2000, 492, 15-22. (35) Bahners, T.; Kesting, W.; Shollmeyer, E. Appl. Surf. Sci. 1993, 69, 12-15. (36) Huang, X.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1988, 60, 1837-1838.

software v.2 (Kratos). The sensibility factors were 0.278 and 0.711 for carbon C 1s and oxygen O 1s, respectively. Transmission was near 0.144 and 0.139 for carbon C 1s and oxygen O 1s. The individual components were taken as mixed Voigt functions and fitted using commercial data processing software Igor Pro 3.14 (WaveMetrics Inc.; Oregon). To facilitate the detection alignment, the irradiated samples (static and scanned) were produced with a projection mask of 1 cm2. All of the samples were introduced in the analytical system immediately after preparation to limit contamination and aging effects. The reference PET was also cleaned with ethanol and distilled water prior to analysis. Figure 1. Schematic representation of the microchannel and the patterning process. The channel dimensions are depth d, 42 µm; bottom section g, 30 µm; lamination side e, 60 µm; and double depth induced by the static ds, 3 µm. The perimeter section is given by f ) e + g + 2[(e - g/2)2 + d2]1/2. The surfaces represented here are the dynamic ablated surface, D; the static ablated surface, S; and the lamination surface, L. The lengths a (100 µm) and lp are, respectively, the spot length and the distance between two static patterns.

0.1 M, sodium acetate 0.1 M; Merck; Germany), was first diluted to the operating concentrations (15 mM and 12 mM) and was then filtered and degassed before use. For each set of experiments, the microchannel fabrication and the EOF measurements were performed the same day to limit aging effects. The fluorescence detection and microscopic investigation was performed using a confocal microscope (Axiovert 25, lamp HPO 100 W; Zeiss; Germany) with a high-sensitive CCD black and white camera (CF 8/4, Kappa; Germany) interfaced to an 8-bit flash analog-to-digital converter (ADC) monochrome IMAQ board (PCI1408, National Instruments). To improve the clarity, the fluorescence images were postprocessed using an in-house custom program in IMAQ Vision for Labview (National Instruments). Rhodamine B (Acros Organics; Belgium) diluted in 10 mM phosphate buffer, pH ) 7.2, was used for imaging. 3. XPS Analysis. XPS analyses were performed using an imaging Kratos Axis Ultra (U.K.) X-ray photoelectron spectrometer equipped with a conventional hemispherical analyzer in spectrum mode. The X-ray source was monochromatized Al KR (1486.6 eV) operating at 150 W. The vacuum in the analytical chamber was kept near 10-9 Torr. The experiments presented in this report were carried out using a takeoff angle of 90° (angle between the direction of the observed photoelectrons and the sample surface). The analyzed area was 110 µm2. Charge compensation was performed by a self-compensating device (Kratos patent) using field-emitted low-energy electrons (0.1 eV). Peaks are referenced to the carbon C 1s peak at 284.7 (aromatic carbon), which is considered as the reference.38 All reported binding energies are corrected by this amount. The pass energies were 80 and 40 eV for wide-scan and high-resolution elemental scans. The full width at half-maximum height (fwhm) of the main C 1s core level component of the carbon peak was 1.01 eV for the reference PET. Although almost no degradation was observed on this reference, 40 eV pass energy was chosen to diminish X-ray exposure and avoid possible degradation of the debris. The element quantification was performed with vision processing

III. RESULTS AND DISCUSSION 1. Numerical Study. The double depth induced by static ablation can be considered as a surface defect able to alter the EOF by hydrodynamic (pressure) effects. In a recent publication, Long et al.39 presented a mathematical model describing the diminution on the electroosmotic flow induced by surface defects in a capillary as due to the recirculation flow induced by the pressure effect. Through numerical simulations, Long et al.39 demonstrated that if the channel dimensions are much larger (1 order of magnitude at least) than the defect size ds, then the geometric defect affects the flow only weakly, and there is no appreciable recirculation, with most velocities pointing in the same direction as the slip induced by the defect.39 If d is the channel depth and ds the size of the defect (see Figure 1), the ratios d/ds of the microchannels studied here are superior to 14. This means that the size of the defect, ds, is at least 1 order of magnitude lower than the channel depth, and so no significant recirculation effects counteracting in the EOF direction are expected.39 Geometrically, the patches induced by the static ablation can be schematized in a 2-dimensional (2-D) model, as depicted in Figure 3, where only one double depth is represented to reduce the size and the time of the simulations. The simulations describing the flow behavior in the 2-D model of Figure 3 are based on the numerical study already formulated in detail in a previous publication,40 which describes pressure effects induced by EOF at the T-junction of three microchannels. The simulations were carried out with numerical software Flux Expert (Flux Expert-Simulog, 60 rue Lavoisier, 38330 Montbonnot, St. Martin) on a UNIX workstation (Silicon Graphics, Indigo 2 Solid Impact 10 000 with 640 Mb RAM). The dimensionless electroosmotic flow velocity profiles, Vz, are reported for different ζ-potential ratios, ζr ) ζstat/ζscan, in Figure 4a (output section) and b (midgroove section), in comparison with the classic plug profile velocity, Vref, (full thick line) simulated in a channel without double depth. As a first observation, the flow profile at the output section for ζr ) 1 is weakly distorted compared to Vref (Figure 4a). For decreasing ζ-ratios, that is, 0 e ζr < 1, the flow profiles at the output section present an increasing concave curvature resulting from pressure effects induced by the inhomogeneous ζ-potentials distribution in the channel. The flow profiles result from a combination between a flat and a parabolic profile, that is, an electroosmotic profile perturbed by the pressure evolution in the pattern section (see Figure 4a,b), as already observed in a recent simulation work.40 In a recent publication,

(37) Wagner, F.; Hoffmann, P. Proc. SPIE 2000, 4088, 322-325. (38) Beamson, G.; Briggs, D. High-Resolution XPS of Organic Polymers; John Wiley & Sons: Chichester, New York, Brisbane, Toronto, Singapore, 1992.

(39) Long, D.; Stone, H. A.; Ajdari, A. J. Colloid Interface Sci. 1999, 212, 338349. (40) Bianchi, F.; Ferrigno, R.; Girault, H. H. Anal. Chem. 2000, 72, 1987-1993.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3847

Figure 2. SEM pictures (a and b) of a static ablation in a scanned microchannel. The white lines are added to help visualization. Figure b points out the difference between scanned (D) and static (S) ablated surfaces.

Herr et al.41 observed this flow behavior in cylindrical coupledcapillaries with nonuniform wall surface charge distributions using the caged-fluorescence imaging technique. The reduction of the electroosmotic flow can be monitored as a function the ζ-ratio, ζr, as depicted in Figure 4c, where the ratio, Ivz/Ivref is reported as a function of ζv, where Ivz is the integral of the flow profile on the section X shown in Figure 4a. First we observe that the ratio Ivz/Ivref is equal to 1 if ζstat ) ζscan, thus confirming that pressure-induced recirculation counteracting the electroosmotic flow can be neglected, as predicted by Long et al.39 In the range of 0 e ζr < 1, the diminution of the flow velocity is proportional to the decrease of the ζ-potential in the static pattern, and a reduction of 8% in the EOF is observed for ζr ) 0. It has been demonstrated that in this dimension range, d/ds > 10, the double depth does not perturb the absolute value of the electroosmotic flow; however, the shape of the velocity profile, V, is affected in the pattern section by the pressure effect induced by the double depth (see outlet section of Figure 4) and by the inhomogeneous distribution of the ζ-potentials between the static and scanned ablated surfaces (see Figure 4a,b). Strong deviations from the flat plug profile, Vref of Figure 4a, are observed in Figure (41) Herr, A. E.; Molho, J. I.; Santiago, J. G.; Mungal, M. G.; Kenny, T. W.; Garguilo, M. G. Anal. Chem. 2000, 72, 1053-1057.

3848

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

4b. As a perturbation system, the static pattern will bring a loss in the separation efficiency due to the Taylor dispersion42 in electrokinetic-driven separation techniques, such as capillary electrokinetic chromatography and capillary zone electrophoresis. The loss in resolution can be estimated using the mathematical model presented by Andreev et al.,43 recently implemented to describe the loss in separation efficiency, in terms of theoretical plates heights, observed in composite polymer photoablated microchips.14 In Figure 4d, the theoretical plate heights resulting from axial diffusion and Taylor dispersion, Hdiff and HTaylor, respectively, are plotted on a logarithmic scale as a function of the ratio ζr. The Taylor dispersion, induced only by the double depth, when ζr ) 1, is 1 order of magnitude greater than for the classic plug profile (square marker in Figure 4d) but in the same order of magnitude as the diffusion contribution. For 0 e ζr < 1, the Taylor dispersion increases until HTaylor reaches a value of 2 orders of magnitude higher than for the classic plug profile and 1 order of magnitude greater than Hdiff, which does not evolve a lot in the range of 0 e ζr e 1. These results should be considered only qualitatively, because the diffusion and Taylor dispersions (42) Taylor, G. I. Proc. R. Soc. London 1953, A 219, 186-203. (43) Andreev, V. P.; Dubrovsky, S. G.; Stepanov, Y. V. J. Microcolumn Sep. 1997, 9, 443-450.

of microchannel, the electroosmotic velocity, shown in Figure 5a, can be described by the Schmoluchowsky equation.

veo ) -

ζobsE η

(3)

where  and η are, respectively, the permittivity and the viscosity of bulk electrolyte, and E is the electric field strength along the capillary. The symbol ζobs is the observed ζ-potential, which varies for each surface ratio sr as shown in Figure 5 (right axis). The linear behavior induced by the laser treatment on the EOF shown in Figure 5 allows the formulation of the ζobs as a linear combination (y ) mx + b) of the various surface contributions, ζscan, ζlam, and ζstat (the ζ-potential of the static ablated surface), as described in eq 4. where the only unknown parameter is the

Figure 3. Schematic representation of the 2-D geometric model used to simulate the flow behavior induced by a double depth and by the surface modification. Compared with the experimental model of Figure 1, only two ζ-potentials are used in the simulation, ζscan and ζstat.

depend on the value of the dimensionless number Peclet (PE), and by choosing different electroosmotic velocities or diffusion coefficients from those used here, the values presented in Figure 4d can change considerably. 2. Electroosmotic Flow Measurements. The electroosmotic flow measurements shown in Figure 5 (left axis) are represented as a function of the ratio sstat/stot. They have been normalized by the reference value of the electroosmotic flow velocity, veof,ref, in a microchannel without static patterning but with the same section perimeter f as the patterned microchannels, where f is defined in Figure caption 1. As a first observation, the electroosmotic velocities shown in Figure 5 decrease linearly in the range of the ratio sstat/stot values studied, and from the parameters of the linear fit (full line, Figure 5), a maximal reduction of about 50% can be predicted for a surface ratio sr ) 1. However, this result should be considered with care because of the complexity of the structure (trapezoidal rather than rectangular) and because the effect on the static patterning is altered for different channel depths or geometries, which was not investigated here. The microchannels are composed of 2 different polymers, photoablated PET and lamination PE, the relative ratios of which also strongly affect the electroosmotic velocity.14 The ζ-potentials ζscan and ζlam, which are the ζ-potentials of the scanned photoablated PET and the lamination PE, respectively, were determined using a fitting procedure on the experimental data derived from a mathematical model described by Andreev et al.43 For each type

ζ-potential of the static ablated surface ζstat, The other geometric parameters are described in Figure 2a. In Figure 5, the ζobs is reported as a function of the ratio sr in order to determine the ζstat by fitting with eq 4. The results of the fit give ζ-potential values of -84, -24, and -53 mV for scanned, ablated PET; lamination PE; and static, ablated PET, respectively. These values (the first two) are in very good agreement with previous results.14 Comparatively, the ζ-potential resulting from the static ablated surface, ζstat, in the specified experimental conditions of Figure 5 is 34% lower than the ζscan value and 55% higher than the value for ζlam. 3. XPS Analysis. As demonstrated by EOF measurements, the surface properties of photoablated PET microchannels change after dynamic (D) or static (S) ablation. A net change in the capillary strength has pointed out that the static ablation process results in more hydrophobic surfaces, as compared to the dynamic mode. Moreover, surface analysis by Raman spectroscopy has underlined the disparity between static and dynamic ablation, but unfortunately, a quantitative analysis of the carbon bounds could not be achieved because of the high fluorescence background of the nonirradiated polymer surface.24 Since the first development in the 1980s, the photoablation mechanism of polymers has been a rather controversial issue. Although the relative contribution of the mechanism remains unknown,44 it is commonly suggested that a photochemical process with subsequent thermal degradation of fragment particles and molecules can describe the phenomena. The XPS spectra shown in Figure 6 correspond to the C 1s region (left side) and O 1s region (right side) for the reference PET substrate (Figure 6a), the static ablated PET surface (Figure 6b), the scanned ablated PET surface (Figure 6c), and the scanned ablated surface rinsed with distilled water (Figure 6d), respectively. Figure 7 displays the experimental atomic ratio O/C, the decrease in which observed from the reference PET to the statically photoablated PET confirms that, during the ablative (44) Krajnovich, D. J. J. Appl. Phys. 1997, 82, 427-435.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3849

Figure 4. Simulated flow profiles at the output section (a) and at the pattern section (b) for different ζ-potential ratios. (c) Diminution of the electroosmotic velocity induced by the surface modification, as compared with the classic plug profile. (d) Effect of the surface modification on the height equivalent theoretical plate Hdispersion ) HTaylor + Hdiffusion (for more details, see text).

Figure 5. Normalized experimental electroosmotic flow velocities (left axis) and observed ζ-potentials (right axis) as a function of the surface ratio sr. Both linear regressions (full lines) gave a coefficient correlation of R ) 0.998. Physical parameters: µ ) 0.001 kg m-1 s-1, 0 ) 8.85 10-12 C2 J-1 m-1 and  ) 78. The error bars are the RSD of 12 consecutive measurements (see text for more details).

decomposition, oxides of carbon are expelled while the carbon radicals remain attached to the surface predominantly by C-C recombination.45 Moreover, the XPS spectra of statically photoablated PET washed with water (see Supporting Information) produced the same result as that in Figure 6b, and the O/C ratio was found to equal the native statically ablated surface presented in Figure 7. From an initial value of 0.38 for the scanned sample, which is similar to the value observed for the reference PET, the O/C ratio of the water-cleaned surface D values is 0.22 and is even lower (45) Lazare, S.; Hoh, P. D.; Baker, J. M.; Srinivasan, R. J. Am. Chem. Soc. 1984, 106, 4288-4290.

3850 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

than the O/C ratio of the PET surface S (see Table 1). The water cleaning procedure on the surface D has been accomplished to reproduce surface properties more similar to the one used for the EOF measurements. Comparatively, the O/C ratio of 0.22 observed in water-cleaned PET surface D is very similar to the value obtained by Lazare et al.46 for PET irradiated with a lamp at 185 nm (5 h, ∼ 45 mJ) under vacuum (see Figures 3c and 8 of ref 45). This result sounds controversial in the sense that surface D shows a higher surface potential than surface S, as demonstrated by EOF measurements, but may be explained by the presence of new functional groups (see below). (Note: In refs 45 and 46, the relative O/C surface ratio of nonirradiated PET values is 1.2, which is 3× greater than the valid PET atomic surface ratio value of 0.4. For further comparisons, all of the O/C ratios values of refs 45 and 46 have been divided by 3.) The spectra analyses, the full width at middle height (FWMH), and the relative area of each peak are given in Table 1. The XPS spectra of PET not irradiated (6a) and statically photoablated (6b) are in full agreement with data previously reported.38,46 The functional groups, shown in Chart 1, have been considered for the modeling of PET photolysis46 (a, b of Chart 1) and the PET solidstate polycondensation process47,48 (c, d, e, f of Chart 1). Their binding energies (see Chart 1) can be used as a base for the peak deconvolution in Figure 6c,d. The peaks’ deconvolution shown in Figure 6c,d have been accomplished by taking into account the binding energies presented in Chart 1 and by choosing FWMH < 1.5. (46) Lazare, S.; Srinivasan, R. J. Phys. Chem. 1986, 90, 2124-2131. (47) Devotta, I.; Mashelkar, R. A. Chem. Eng. Sci. 1993, 48, 1859-1867. (48) Ravindranath, K.; Mashelkar, R. A. J. Appl. Polym. Sci. 1990, 39, 13251345.

Figure 6. XPS multiplex spectra of freshly prepared samples. Sample preparations correspond to the laser conditions used to produce the microchannels given in the Experimental Section. Native PET was washed with water before analysis (see text for more details).

Figure 7. O/C ratio of the different PET surfaces presented in Figure 6 (see text for details).

In the deconvolution, it was chosen not to discriminate between some of the C 1s contributions. For example, C-OH (aromatic or aliphatic) and C-O-C were gathered. Similarly, carbonyl and

1,1 diols were not discriminated. This may explain the increased peak fwhm compared to the reference PET. Charging is not expected to be the reason for this, because no problems were observed on the reference PET. Carboxylic acid (a) and phenolic groups (b) have been observed by XPS measurements on the surface of the PET substrate after plasma treatment.49 In the present case (deconvolution), however, the carbon C 1s 3 (carboxylic acid) at 289 eV was not discriminated from the carbon C 1s 3 present in the PET monomer (ester) at 288.7 eV (see Chart 1). Similarly, the phenolic-oxygen-bound 2 at 533.6 eV is difficult to distinguish from the oxygen-bound 2 at 533.2 eV present in the PET monomer and carboxylic acids (see Chart 1). (49) Practical Surface Analysis: by Auger and X-ray Photoelectron Spectroscopy; Briggs, D.; Seah, M. P.; Briggs, D., Ed.; John Wiley & Sons: Chichester, New York, Brisbane, Toronto, Singapore, 1983; pp 388-391.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3851

Table 1. Data from Curve-Fitted Spectra Displayed in Figure 6 element-1s C 1

2

O 3

4

1

FWMH, eV area, %

1.01 61

PET Not Treated 1.1 0.86 1.18 22 17 49

FWMH, eV area, %

1.18 73

PET After Static Ablation 0.9 0.91 1.24 13 14 46 PET Scanned, Not Cleaned 3-3

FWMH, eV area, %

1.18 56

1.33 20

1.42 9

1.43 15

1.18 9

2

3

4

1.35 51

-

-

1.4 54

-

-

2-2 1.38 20

1.2 23

1.39 52

1.25 27

1.5 45

PET Scanned, Water-Cleaned 3-3 2-2 FWMH, eV area, %

1.29 71

1.5 18

1.24 4

1.34 7

1.22 7

1.5 15

Chart 1

However, an evident contribution of the C 1s peak at 287.7 eV and the plausible presence of O 1s at 532.3 eV reveals the presence of aldehydes and ketones functional groups (c in Chart 1) on the PET surface, which are excepted as well during PET polycondensation. In both spectra of Figure 6c,d, the peak areas at 287.7 and 532.3 eV match within less than the 10% error, which is a result of the probable presence of the functional group d in Chart 1. Moreover, the well-known formation of aliphatic alcohol (e in Chart 1) and ether (f in Chart 1) groups observed during PET polycondensation support the presence of the peak at 532.9 eV. 3852 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

Figure 8. (a) CCD image at the T junction, filled with a buffer solution, of a sidearm channel (2 cm) with a main channel (4 cm) on T microchip. Imaging of 200 µM of Rhodamine B in 10 mM phosphate buffer, pH ) 7.2, was obtained by applying a positive potential at the top channel relative to the sidearm and using (b) all native photoablated surfaces and (c) static patterning on the sidearm channel, sr ) 0.5. Imaging and high-voltage control was performed using the same system described in ref 14.

After the cleaning procedure, a net decrease of the C 1s (ester/ carboxylic acid) 289 eV and of the C 1s at 287.7 eV (carbonyl) (see Table 1) are observed relative to the C-C contribution. It suggests the eventual presence of hydrophilic entities such as acid carboxylic/carbonyl groups (a) soluble in water. The C 1s at 286.2 decreases only slightly. The decrease of the carbonyl contribution (287.7 eV) could be associated with the reduction in the O 1s region of the peaks at 532.3 eV (as expected) and suggests the presence of hydrophilic aldehydes or ketones; however, due to the overall O 1s peak shape, one must be careful when interpreting the O 1s. One interpretation could be that long-carbon-chain alcohols remain at the surface, but shorter chains containing carbonyl may be solubilized in water. In summary, the XPS measurements have pointed out that the surface D is defined by the redeposition of a layer of debris that has a different chemical composition from the surface S, and all the functional groups present in Chart 1 have been potentially observed by the XPS measurements. After cleaning with water, surface D shows a clear decrease in the O/C ratio, but new CO functionalities (Figure 6c) are still observed, which explains that the surface D after cleaning shows a higher surface potential than surface S, as observed by the electroosmotic measurements; however, the relative decrease of the oxygenated contributions to C-C after washing may play a role in the EOF with time. 4. Microchip Application. In practice, surface properties resulting from the static ablation process are now used in our laboratory to produce on-line microhole arrays for the decoupling part of a planar CEEC microchip similar to the one described by Rossier et al.34 To facilitate visualization, Figure 8 represents a T junction (Figure 8a) on a microchip, featured by Ramsey and Ramsey,30 where similar induced electroosmotic pumping is reproduced. Briefly, the microchip is composed of 3 microchannels: the separation channel (sep), the sidearm channel, and the field-free (ff) channel. By applying a potential in a T microchip with photoablated surfaces, as schematized in Figure 8b, the electroosmotic flow, which can be visualized by Rhodamine B dye at pH ) 7.2, follows the electric field gradient as presented in a recent simulation work.40 When the sidearm channel surface is modified

by the photoprocess described in this paper, the difference in surface potential between the separation and sidearm channels generates a pressure drop at the T junction.30,40,50 This effect is then responsible for the pressure-induced flow observed in the ff channel (see Figure 8c) The mass transport toward the free gradient field channel is increased for lower ζ-ratios than ζstat/ζscan ) 0.64, but the geometric parameters can also be adjusted to improve the mass transport.40,50 The fabrication of the T microchip depicted in Figure 8c was accomplished in a single-step process in less than 20 min, surface photo modification included. IV. CONCLUSION A UV laser was used to fabricate simple microchannels and complex microchip structures and has been demonstrated as a promising method to prototype microchips on polymer substrates. Moreover, the possibility of controlling the surface modification in a microchannel by alternating dynamic and static ablation procedures has been demonstrated by XPS measurements. The new functional groups created have been shown to be soluble in water but are also strongly adherent on the polymer surface. It has also been demonstrated that well-defined static patterning in a microchannel can significantly reduce the electroosmotic flow; however, it may be pointed out that the use of such surface treatments in electrokinetic-driven separation could result in a loss (50) Culbertson, C. T.; Ramsey, R. S.; Ramsey, M. J. Anal. Chem. 2000, 72, 2285-2291.

of separation efficiency because of Taylor dispersion. Certainly, this fabrication patterning can be used as an alternative to postchemical treatments to modify locally the surface potential. The utilization of this fabrication system can enlarge the investigation mode of the microfabrication by its short time processing and its large flexibility. ACKNOWLEDGMENT This work has been achieved in collaboration with the Laboratoire d’Optique Applique´e (EPFL). Drs. F. Wagner and P. Hoffmann are gratefully acknowledged for the SEM pictures and for their very helpful discussions on the subject of laser ablation procedures. The authors thank, as well, N. Xanthopoulos for the XPS measurements, Dr. R. Ferrigno for the mathematical procedure of the numerical study, and Dr. Alastair Wark for revision of the manuscript. F.B. thanks l’EÄ cole Polytechnique Fe´de´rale de Lausanne (EPFL) for a doctoral fellowship. SUPPORTING INFORMATION AVAILABLE XPS multiplex spectra of freshly prepared samples. This material is available free of charge via the Internet at http:// pubs.acs.org. Received for review March 5, 2001. Accepted May 9, 2001. AC010262Z

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3853