Surface Modification of Poly(dimethylsiloxane) with a PerfluorinatedAlkoxysilane for Selectivity toward Fluorous Tagged Peptides

Dan Wang, Richard D. Oleschuk, and J. Hugh Horton*

Department of Chemistry, Queen’s UniVersity, Kingston, Ontario, Canada K7L 3N6

ReceiVed July 9, 2007. In Final Form: October 22, 2007

Poly(dimethylsiloxane) (PDMS) and similar polymers have proved to be of widespread interest for use in microfluidicand similar microanalytical devices. Surface modification of PDMS is required to extend the range of applicationsfor devices made of this polymer, however. Here we report on the grafting of perfluorooctyltriethoxysilane viahydrolysis onto an oxidized PDMS substrate in order to form a fluorinated microchannel. Such a fluorinated devicecould be used for separating fluorous tagged proteins or peptides, similar to that which has been recently demonstratedin a capillary electrophoresis system or in an open tubular capillary column. The modified polymer is characterizedusing chemical force titrations, contact angle measurements, and X-ray photoelectron spectroscopy (XPS). We alsoreport on a novel means of performing electroosmotic measurements on this material to determine the surface zetapotential. As might be expected, contact angle and chemical force titration measurements indicate the fluorinatedsurface to be highly hydrophobic. XPS indicates that fluorocarbon groups segregate to the surface of the polymer overa period of days following the initial surface modification, presumably driven by a lower surface free energy. Oneof the most interesting results is the zeta potential measurements, which show that significant surface charge can bemaintained across a wide range of pH on this modified polymer, sufficient to promote electroosmotic flow in amicrofluidic chip. Matrix-assisted time-of-flight mass spectrometry (MALDI-TOF MS) measurements show that afluorous-tagged peptide will selectively adsorb on the fluorinated PDMS in aqueous solution, demonstrating that thefluorinated polymer could be used in devices designed for the enrichment or enhanced detection of fluorous-labeledproteins and peptides.

IntroductionAs micro total analysis systems (µ-TAS) have become

increasingly popular for use in a variety of research fields,1-9

researchers have put more focus on the selection and developmentof new fabrication materials to replace conventionally used glass,quartz, and silicon, which are relatively expensive because ofrelatively complicated and time-consuming fabrication processes.Inexpensive polymers have come to attract more attention becauseof such attributes as being readily disposable, easily molded orembossed with microchannels, and thermally or adhesivelysealed.10Several different polymer types have been investigatedas substrate materials forµ-TAS applications, includingpoly(dimethylsiloxane) (PDMS),10-13 polymethylmethacrylate(PMMA),14,15polycarbonate (PC), polyethylene (PE), and poly-styrene.13 Among these, PDMS has attracted the most attention

as a material for constructing microfluidic devices in biologicaland water-based applications for a number of reasons: reproduc-ible features on the micrometer scale can be produced with highfidelity by replica molding, optical transparency down to 280nm, low-temperature curing, nontoxic, reversibly deformed andself-sealed, and the fact it can be readily tailored by a range ofwell-described surface-modification protocols.10,11

Unmodified PDMS, however, is not optimal for microfluidicapplications: the hydrophobic surface results in PDMS beingdifficult to wet with aqueous solvents, making microchannelfilling difficult. The lack of sufficient ionizable surface sites alsomeans that microfluidic chips cannot support strong electro-osmotic flow (EOF).16 A number of strategies have been carriedout to render the PDMS surface more hydrophilic. One of theeasiest means is an air plasma oxidation method used to oxidizethe Si-CH3 groups on the PDMS surface to Si-OH.10,17 Thishas been shown to increase the EOF rate significantly owing toan increase in the surface zeta potential,ú, arising fromdeprotonation to form SiO- sites.18 Such surfaces are unstable,however, with significant decreases in the number of ionizablesurface sites and consequently the EOF, taking place within 24h following oxidation. This has been attributed to the migrationof short-chain oligomers of PDMS to the surface, driven by theconcomitant decrease in surface free energy.10,19,20

* Author to whom correspondence should be addressed. E-mail: hortonj@chem.queensu.ca. Tel: (613)-533-2379. Fax: (613)-533-6669.

(1) Colyer, C. L.; Tang, T.; Chiem, N.; Harrison, D. J.Electrophoresis1997,18, 1733.

(2) Effenhauser, C. S.; Bruin, G. J. M.; Paulus, A.Electrophoresis1997, 18,2203.

(3) Yao, S.; Anex, D. S.; Caldwell, W. B.; Arnold, D. W.; Smith, K. B.;Schultz, P. G.Proc. Nat. Acad. Sci. U.S.A.1999, 96, 5372.

(4) Santini, J. T., Jr.; Cima, M. J.; Langer, R.Nature1999, 397, 335.(5) Burns, M. A.; Johnson, B. N.; Brahmasandra, S. N.; Handique, K.; Webster,

J. R.; Krishnan, M.; Sammarco, T.; Man, P. M.; Jones, D.; Heldsinger, D.;Mastrangelo, C. H.; Burke, D. T.Science1998, 282, 484.

(6) Kricka, L. J.Clin. Chem.1998, 44, 2008.(7) Ramsey, J. M; Jacobson, S. C.; Knapp, M. R.Nat. Med.1995, 1, 1093.(8) Freemantle, M.Chem. Eng. News1999, 77, 27.(9) Weigl, B. H.; Yager, P.Science1999, 283, 346.(10) Mcdonald, J. C; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H. K.;

Schueller, O. J. A.; Whitesides, G. M.Electrophoresis2000, 21, 27.(11) Ocvirk, G.; Munroe, M.; Tang, T.; Oleschuk, R.; Westera, K.; Harrison,

D. J Electrophoresis2000, 21, 107.(12) Duffy, D. C.; Mcdonald, J. C.; Schueller, O. J. A.; Whitesides G. M.Anal.

Chem. 1998, 70, 4974.(13) Rossier, J.; Reymond, F.; Michel, P. E.Electrophoresis2002, 23, 858.(14) Kricka, L. J.; Fortina, P.; Panaro, N. J.; Wilding, P.; Alonso-Amigo, G.;

Becker, H.Lab Chip2002, 2, 1.

(15) Soper, S. A.; Henry, A. C.; Vaidya, B.; Galloway, M.; Wabuyele, M.;Mccarley, R. L.Anal. Chim. Acta2002, 470, 87.

(16) Wang, B.; Abdulali-Kanji, Z.; Dodwell, E.; Horton, J. H.; Oleschuk, R.D Electrophoresis2003, 24, 1442.

(17) Ren, X.; Bachman, M.; Sims, C.; Li, G. P.; Allbritton, N.J. Chromatogr.,B 2001, 762, 117.

(18) Wang, B., Horton, J. H., Oleschuk, R. D.Can. J. Chem.2006, 84, 720.(19) Morra, M.; Occhiello, E.; Marola, R.; Garbassi, F.; Humphrey, P.; Johnson,

D. J. Colloid Interface Sci.1990, 137, 11.(20) Murakami, T.; Kuroda, S., Osawa, Z.J. Colloid Interface Sci.1998, 202,

37.

1080 Langmuir2008,24, 1080-1086

10.1021/la702038t CCC: $40.75 © 2008 American Chemical SocietyPublished on Web 12/29/2007

Vickers et al.21 used a two-step process involving solventextraction of the oligomers followed by oxidation as one approachto solve this problem, making oxidized PDMS surfaces stablefor at least 7 days in air. Another approach is to use chemicalderivatization methods. This consists of a facile two-stage surface-modification process consisting of an oxidation step followed byreaction with a triethoxysilyl derivative. Using this scheme, PDMSsurfaces coated with both sulfonic acid and amine sites havebeen produced.16,18, 22Such modified surfaces are more stablewith respect to maintaining EOF and, by appropriate tailoringof the acid or base groups present, allow EOF experiments totake place over a wide pH range. More complex surface-modification schemes have also been attempted: Roman et al.23

used transition-metal sol-gel chemistry to directly coat the PDMSmicrochannels with variously derivatized inorganic coatings toobtain a durable modified surface supporting electroosmoticmobility over a period of 95 days. Wang et al.24 demonstratedthe modification of PDMS channels with citrate-stabilized goldnanoparticles after coating a layer of linear polyethylenimine.Such microchips could be used to separate simple moleculessuch as dopamine and epinephrine and had a long-term stabilityof up to 2 weeks. Finally, Seo et al.25 improved the wettabilityof PDMS by directly incorporating a nonionic surfactant (TX-100) into the PDMS. The concentration of the surfactant at thesurface could then be changed by surface migration upon exposureto various solvents.

Recent work has shown that a capillary electrophoresis systemcontaining fluorinated microspheres may be used to enrich andseparate different subsets of peptides and proteins that have beentagged with fluorous species.26 Perfluoroalkylated porous Sisurfaces have also been used to capture peptides containingfluorous affinity tags, which are then simply washed off andanalyzed using desorption/ionization mass spectrometry.27Here,we describe the surface modification of PDMS by oxidationfollowed by reaction with 1,1,2,2-tetrahydroperfluorooctyltri-ethoxysilane (PFO) to create a material that may be used to forma fluorinated channel within a microfluidic device. Such a devicecould potentially be used to contain a bed of fluorinated beadsor within an open-column capillary column for the separationof fluorous-tagged species. Previous workers have focused onthe properties of the mixture of a PDMS and PFO copolymersystem,28 including segregation into fluorous and hydrocarbondomains within the bulk. Using a system more similar to thatexplored here, Emmanuel et al. have studied cross-linked layersof PFO on an oxidized PDMS substrate using XPS, contact anglemeasurements, and ATR-FTIR to study the adhesion of yeastsand bacteria to fluoroalkyltrichlorosilane-modified PDMS pros-theses.29 However, our work is the first, to our knowledge, toexamine the surface charge of PFO-modified PDMS with thegoal of enhancing and stabilizing flow performance of PDMSmicrochips and the ability of this material to adsorb fluorous-tagged peptides selectively. A number of surface analyticaltechniquesareused tocharacterize the fluorinatedPDMSpolymer.

X-ray photoelectron spectroscopy (XPS)30is used to characterizethe optimum reaction conditions for fluorination and the stabilityof fluorinated surface. Contact angle measurements are also usedto gauge the extent of fluorination at the surface. Atomic forcemicroscopy (AFM) and chemical force spectrometric methods31,32

are used to characterize the chemical properties of functionalgroups appended on the PDMS surface and fluorine-fluorineand fluorine-methyl interactions. The selectivity of the fluori-nated PDMS toward retaining a fluorine-tagged peptide whenwashed with both water and methanol solvents was assessedusing matrix-assisted laser desorption ionization-time-of-flight(MALDI-TOF) measurements. Finally, we evaluate the elec-troosmotic flow performance and zeta potential of fluorinatedPDMS microchips over a range of pH conditions and comparethese with those of unmodified and oxidized PDMS.

Experimental Section

Fabrication protocols for PDMS microchips based on a Sylgard184 PDMS formulation have been described in detail elsewhere16

and will be only briefly outlined here. Sylgard 184 PDMS prepolymer(Dow Corning Corporation, Midland, MI) was mixed thoroughly ina 10:1 mass ratio of silicone elastomer to curing agent and pouredonto a glass substrate containing six identical electrophoretic devices(Micralyne, Edmonton, Alberta) in a “Twin-T” configuration thatacted as the negative master. The PDMS was polymerized at 65°Cfor 4 h and peeled off of the substrate. This was then used in a secondPDMS molding step to yield the microfluidic substrate. Followingthis, the PDMS substrate and the unpatterned cover were placed inan air plasma generator (Harrick Scientific Corporation, Ossining,NY) for 40 s (10.2 W,10 MHz rf level at 80 mTorr). The fluorinatedPDMS devices were produced by immersing the freshly oxidizedPDMS substrate and cover into a 20 mmol/L solution of perfluoro-1,1,2,2-tetrahydrooctyl-1-triethoxysilane (PFO, United ChemicalTechnologies, Inc., Horsham, PA) in toluene for up to 4 h. Allglassware used in this process was coated with an inert cross-linkedalkyl silane layer by immersing the glassware in a 10 mM toluenesolution of octadecyltrochlorosilane (OTS, Sigma-Aldrich) for 24h to prevent any competing adsorption by the PFO on the glasswaresurface. After modification was completed, the PDMS substrate andcover were dried in a stream of dry nitrogen gas. They were thenlaid on top of one another, forming a reversible air- and water-tightseal. Obtaining a good seal between substrate and cover is importantin preventing leakage after the microfluidic chip is filled. Thefluorinated PDMS microchips provided a more reliable but reversibleseal than did the unmodified PDMS microchips. However, they stilldid not realize the extremely leak-tight but irreversible sealingproperties that oxidized PDMS microchips exhibit.10,12,33 Someswelling of the PDMS was observed during this fluorination process,but the swelling reversed after the device had been left to dry forseveral hours.

The measurement of electroosmotic mobility (µeo) in the micro-channels was performed using current monitoring11,17,34,35with amicrofluidic tool kit (Micralyne, Edmonton, Alberta) at an appliedfield strength,E, of 3.5 kV. The microchannels were first filled witha low-ionic-strength phosphate buffer solution (5 mmol/L). Sub-sequently, the buffer reservoir was emptied and filled with a higher-ionic-strength phosphate buffer solution (30 mmol/L). Electrodeswere then placed in the buffer and waste reservoirs at either end ofthe microchannel, and the flow rate was determined by measuringthe time taken for the current to increase to a higher plateau valueas the microchannel was filled with the higher-ionic-strength buffer.

(21) Vickers, J. A.; Caulum, M. M.; Henry, C. S.Anal. Chem.2006, 78, 7446.(22) Wang, B.; Chen, L.; Abdulali-Kanji, Z.; Horton, J. H.; Oleschuk, R. D.

Langmuir2003, 19, 9792.(23) Roman, G. T.; Culbertson, C. T.Langmuir2006, 22, 4445.(24) Wang, A.-J.; Xu, J.-J.; Zhang, Q.; Chen, H.-Y.Talanta2006, 69, 210.(25) Seo, J.; Lee, L. P.Sens. Actuators, B2006, 119, 192.(26) Brittain, S. M.; Ficarro, S. B.; Brock, A.; Peters, E. C.Nat. Biotechnol.

2005, 23, 463.(27) Go, E. P.; Uritboonthai, W.; Apon, J. V.; Trauger, S. A.; Nordstrom, A.;

O’Maille, G.; Brittain, S. M.; Peters, E. C.; Siuzdak, G.J. Proteome Res.2007,6, 1492.

(28) Johnston, E.; Bullock, S.; Uilk, J.; Gatenholm, P.; Wynne, K. J.Macromolecules1999, 32, 8173.

(29) Everaert, E. P. J. M.; van der Mei, H. C.; Busscher, H. J.Colloids Surf.,B 1998, 10, 179.

(30) Turner, N. H.; Schreifels, J. A.Anal. Chem.1994, 66, 163R.(31) Noy, A.; Vezenov, D. V.; Lieber. C. M.Ann. ReV. Mater. Sci.1997, 27,

381.(32) Finot, M. O.; Mcdermott, M. T.J. Am. Chem. Soc.1997, 119, 8564.(33) Chaudhury, M. K.; Whitesides, G. M.Langmuir1991, 7, 1013.(34) Harris, D. C.QuantitatiVe Chemical Analysis, 4th ed.; W. H. Freeman

and Company: New York, 1995; pp 715-720.(35) Huang, X.; Gordon, M. J.; Zare, R. N.Anal. Chem.1988, 60, 1837.

Surface Modification of Poly(dimethylsiloxane) Langmuir, Vol. 24, No. 3, 20081081

The EOF was measured at various pH values from 3 to 10 using aphosphate buffer solution in each case. For any given pH, an averageelectroosmotic mobility value was obtained from three consecutivemeasurements on the same device.

Chemical force titration was used here to determine the adhesiveforces between the functional groups on the modified AFM tips andfluorinated PDMS surfaces. The data were obtained using a PicoSPM(Molecular Imaging, Tempe, AZ) and a Nanoscope IIE controller(Digital instruments, Santa Barbara, CA). All force titration datawere acquired on a PDMS film cast in a similar manner to that forthe cover plates used in microfluidic chip manufacture. The PDMSfilm underwent exactly the same synthesis and surface-modificationprocedures as for cover plates used in the manufacture of microfluidicchips. The functionalized tips were prepared from contact-modesilicon AFM tips (MikroMasch) coated by thermal evaporation witha 5 nm layer of chromium to promote the adhesion of the followinglayer of gold (10 nm). The tips were then immersed in a solutionof 10 mmol L-1 1-dodecanethiol, 12-thiohexadecanoic acid, orperfluorodecanethiol in ethanol for 24 h to obtain methyl-,carboxylate-, and perfluoro-terminated tips. The tip radius as quotedby the manufacturer was<10 nm. The probe tip and fluorinatedPDMS surface were immersed in a droplet of a given pH solution.Unbuffered, low-ionic-strength solutions (10-3 M) of hydrochloricacid and sodium hydroxide were freshly prepared and used to controlthe pH. Solutions at pH 2 and 12 were of higher ionic strengths (ca.10-2 M). The only ions in solution were those introduced by pHadjustment with NaOH and HCl. The adhesive force between thetip and sample was determined from the average of the well depthfrom the retraction portion of 140-300 force-distance curves ateach pH value. The reported values of the adhesive interaction arean average of all of the force curves obtained, whereas the reportederrors reflect the standard deviation of the data.36,37

XPS measurements were performed using a Thermo InstrumentsMicrolab 310F surface analysis system (Hastings, U.K.) underultrahigh vacuum conditions and an Al KR X-ray source (1486.6eV) at 15 kV anode potential and 20 mA emission current. Scanswere acquired in fixed analyzer transmission (FAT) mode at a passenergy of 20 eV and a surface/detector take off angle of 75°. Allspectra were calibrated to the O 1s line at 532.0 eV; minor chargingeffects were observed, ranging from 1.0 to 2.0 eV. Spectra werebackground subtracted using a Shirley fit algorithm and a Powellpeak-fitting algorithm within the spectrometer software. The PDMSsubstrates used in XPS analyses were made using the same prepolymerand curing agent process noted above, but before curing, spin casting(at 3000 rpm for 40 s) was used to transfer the mixture onto a Petridish such that PDMS films of<0.5 mm thickness were obtained.These relatively thin polymer substrates minimized charging effectsduring XPS measurements. Further surface treatment on PDMS wascarried out in the same manner as described above.

Contact angle measurements were made using a model VCAOptima XE -3000S (AST Products, Inc., Billerica, MA) to assesschanges in the hydrophobic character of the modified PDMS surfaces.The values were determined using deionized water and the averagecontact angle from a minimum of three different droplets measured.

Mass spectrometric measurements were carried out using aVoyager DE-STR MALDI-TOF system (Applied BiosystemsCorporation, Foster City, CA). The fluorinated peptide derivativeused was a single-tagged cortactin derivative (F-CTN). Syntheticphosphocortactin (5µL of a 500 pmol solution; pCTN; LHKHCSP-QVDSVR) was reacted with a 3:1 DMSO/ethanol solution (5µL),saturated Ba(OH)2 solution (4.6µL), and 500 mM NaOH (1µL).To this solution, 0.7µL of fluorous thiol tag CF3(CF2)5CH2CH2SH(Fluorous Technologies, Inc.) was added. The reaction mixture wasmaintained at 37°C for 60 min, at which point it was quenched bythe addition of trifluoroacetic acid. The molecular weight of theresultant F-CTN, LHKHCSFQVDSVR was calculated to be 1721g/mol, and the calculated isotope number ratios were consistent

with the MALDI-TOF mass spectrum of a standard. In a typicalMALDI-TOF experiment, 2µL of a 2.5µmol/L F-CTN solutionwas spread onto PDMS or suitably modified PDMS substrates andallowed to dry for a period of 1 h. The substrate was then washedwith 2 µL of water to remove any unbound peptide, followed by awash with 2µL of methanol. The methanol extract was then mixedwith 2 µL of a sinapinic acid matrix (sinapinic acid dissolved in 1:1water/acetonitrile solution) and spotted onto the MALDI sampleplate. An F-CTN standard was made by directly combining theinitial F-CTN solution with the sinapinic acid matrix.

Results and Discussion

XPS and Contact Angle Characterization.XPS and contactangle measurements were carried out on a set of PDMS samplesexposed to various degrees of plasma oxidation and exposure tosolutions of PFO in order to determine the optimum conditionsto maximize the quantity of fluoro groups at the polymer surfaceand to characterize the fluorinated interface fully. The O 1s/C1s XPS peak area ratio of the PDMS film was found to saturateby 40 s of exposure to the plasma oxidation process describedabove. Previous workers38,39have also used XPS to examine thestability of oxidized PDMS and have generally found thatoxidation times between 30 to 180 s, under similar conditionsto those used here, gave the most stable layer; that is, polymerprepared in such a way remained hydrophilic for the longesttime periods following oxidation. However, regardless of theoxidation exposure time, oxidized PDMS generally reverts to itshydrophobic state in less than 48 h.

The emission intensity of photoelectrons for a subsurfacespecies is generally attenuated in an exponential fashion as afunction of overlayer thickness, with a decay constant equal tothe escape depth of the photoelectrons. Using the published escapedepth value of 2.37 nm40 for Si 2p photoelectrons at a kineticenergy of 1400 eV, the attenuation of the Si 2p peak suggestsan SiOx layer thickness of approximately 2.4 nm. This is somewhatlower than the values in other reports, which range from 7 to160 nm.38,41-44

Because the oxidation of PDMS leads to the formation ofSi-OH sites on the polymer surface,16which may undergo furtherhydrolysis with triethoxysilyl derivatives, further exposure tothe fluorinating agent was carried out on samples that had beenoxidized for periods of 40-120 s. Non-oxidized PDMS was alsoexposed to a solution of PFO in toluene as a control case. Figure1a shows a plot of the F 1s/Si2 p peak area ratio as a functionof exposure time to the PFO solution subsequent to various degreesof oxidation. The F 1s and Si 2p signals consisted of a singlepeak at binding energies of 688.5( 0.5 eV and 101.9( 0.3 eV,respectively, regardless of preparation conditions. The Si 2pbinding energy is consistent with previously published valuesfor silica gel materials. It should be noted that while the bindingenergy of the Si 2p peak did not shift significantly following theexposure of oxidized samples to PFO solution, the peak widthdid increase, from 2.0 to 2.5 eV, following plasma oxidation.38,41,45

This is consistent with a range of silicon oxide sites beingintroduced into the polymer surface region during the oxidation

(36) Mengistu, T. Z.; Goel, V.; Horton, J. H.; Morin, S.Langmuir2006, 22,5301.

(37) Wang, B.; Oleschuk, R. D.; Horton, J. H.Langmuir2005, 21, 1290.

(38) Hillborg, H.; Ankner, J. F.; Gedde, U. W.; Smith, G. D.; Yasuda, H. K.;Wikstrom, K. Polymer2000, 41, 6851.

(39) Flitsch, R.; Raider, S. I.J. Vac. Sci. Technol.1975, 12, 305.(40) Ebel, H.; Pohn, C.; Svagera, R.; Wernle, M. E.; Ebel, M. F.J. Electron

Spectrosc. Relat. Phenom. 1990, 50, 109.(41) Owen, M. J.; Smith, P. J.J. Adhes. Sci. Technol.1994, 8, 1063.(42) Mirley, C. L.; Koberstein, J. T.Langmuir1995, 11, 1049.(43) Ouyang, M.; Yuan, C.; Muisener, R. J.; Boulares, A.; Koberstein, J. T.

Chem. Mater.2000, 12, 1591.(44) Hillborg, H.; Tomczak, N.; Olah, A.; Schonherr, H.; Vancso, G. J.Langmuir

2004, 20, 785.(45) Peterson, S. L.; McDonald, G. A.; Sasaki, P. L.; Darryl, Y.J. Biomed.

Mater. Res. A2005, 72, 10.

1082 Langmuir, Vol. 24, No. 3, 2008 Wang et al.

process. The C 1s spectra were considerably more complicated,as will be discussed further below. In Figure 1B, we plot the XPSpeak area ratio of F 1s with respect to the substrate component(methyl group of PDMS at 283.9 eV) of the C 1s signal, againas a function of exposure time to PFO solution.

The peak area ratio graphs in Figure 1 suggest that in the caseof the oxidized samples the relative amount of F at the surfacesaturates after 3 to 4 h ofexposure. Figure 1 also shows that asmall amount of F signal is observed even when the non-oxidizedPDMS is exposed to PFO solution, a point that we will returnto when discussing the detailed C 1s XPS data. It should be notedthat the F 1s/C 1s (methyl PDMS at 283.9 eV) area ratio shouldbe more sensitive to the relative amount of overlayer depositedbecause the signals here are derived from what are exclusivelyoverlayer (F) and substrate (C methyl) groups. The data in Figure1A should be less sensitive because the Si 2p peak consists ofsignal from both the substrate PDMS and the siloxane groupsof the cross-linked PFO overlayer, which could not be distin-guished within the XPS spectrum. In either case, the area ratiodata suggest that the cross-linked PFO layer has reached itsmaximum growth by about 4 h ofexposure. Although there maybe some variation between samples, the data in Figure 1 showthat the highest F 1s/C1s or F 1s/Si 2p area ratios are obatinedafter 2-4 h of exposure to solution and have achieved a saturatedO 1s intensity by 40 s of exposure to plasma oxidation. Giventhis result, we chose to perform most of the remaining experimentsunder conditions of 40 s of oxidation, followed by 4 h ofexposureto PFO solution.

Figure 2 shows the C 1s and Si 2p spectra for four differentPDMS samples. In the case of unmodified PDMS, a single peakat 284.1 eV is observed, consistent with C in a methyl environment

in the PDMS polymer.46This undergoes only limited broadeningupon oxidation. Again, this is consistent with previous studiesof oxidized PDMS and suggests that the oxidation of Si sites,as opposed to C, predominates during the plasma oxidationprocess.19,47 When an oxidized sample is exposed to PFO,significant changes in peak shape take place. Figure 2C showsdata for a sample exposed to the PFO solution for 4 h following40 s of plasma oxidation. This was the set of conditions that ledto the largest F 1s/C 1s (methyl) area ratio, although similarspectra were collected for other combinations of oxidation andexposure time. Four statistically significant peaks can be observedin the XPS spectrum. The methyl peak from the bulk PDMSsubstrate is at 283.9 eV, the same as for unmodified PDMS,within experimental error. This peak has increased in width ascompared with that of unmodified or oxidized PDMS, and ashoulder peak at 285.4 eV may be fit. The remaining two peakslie at much higher binding energies of 290.4 and 292.6 eV. Thesebinding-energy values are consistent with those reported byprevious workers who examined a copolymer of PFO and PDMS28

and a layer of 1,1,2,2-tetrahydoxyperfluorooctyltrichlorosilanedeposited on PDMS.29 The two higher-binding-energy peaksmay be attributed to CF2 and the terminal CF3 portions of thePFO respectively, as shown in Figure 2. The peak at 285.4 eVarises from the two methylene groups present at the base of thePFO moiety. Emmanuel et al.,29upon depositing the trichlorosilylderivative, did not see any evidence of PDMS methyl groups intheir XPS spectra, presumably because the more reactive natureof the chlorosilyl led to a thicker overlayer layer on the surface,which is considerably larger than the escape depth of C 1sphotoelectrons.

The attenuation observed for the C 1s signal of the methylgroup of PDMS at a binding energy of 284.1 eV upon exposingthe 40 s oxidized PDMS sample to PFO solution for 4 h was25.5%. The previously published escape depth of the C 1s

(46) Zhao, S.; Denes, F.; Manolache, S.; Carpick. R. W.Proceedings of theSEM VIII International Congress and Exposition on Experimental and AppliedMechanics2002, 162.

(47) Chaudhury, M. K.; Whitesides, G. M.Science1992, 255, 1230.

Figure 1. (A) F 1s/Si 2p and (B) F 1s/C 1s XPS peak area ratiosfor PDMS substrates following exposure to a 20 mmol/L solutionof perfluoro-1,1,2,2-tetrahydrooctyl-1-triethoxysilane for varyingtimes. Prior to exposure to the fluorinating agent, substrates wereexposed to plasma oxidation as noted in the legend. The O 1s/Si 2parea ratio as a function of plasma oxidation time is shown in theinset.

Figure 2. X-ray photoelectron spectra of the C 1s and Si 2p regionsfor (A) unmodified PDMS, (B) PDMS exposed to plasma oxidation,(C) PDMS exposed to plasma oxidation followed by exposure toPFO solution to form a fluorinated surface, and (D) PDMS exposedto PFO solution without previous oxidation. In the case of spectrumC, the higher binding peaks noted are assigned to the various Cenvironments within the grafted PFO surface species.

Surface Modification of Poly(dimethylsiloxane) Langmuir, Vol. 24, No. 3, 20081083

photoelectrons is 1.58 nm at a kinetic energy of 1200 eV, yieldinga PFO overlayer thickness of 2.2 nm. Because the PFO moleculeis rougly1.3 nm in length, this suggests that we have an overlayerthat is about two molecules thick at the PDMS surface.

Contact angle data for all relevant samples are summarizedin Table 1. The maximum contact angle observed was 119( 5°for the reaction conditions that resulted in the largest F/C signalarea ratio (40 s of oxidation followed by 4 h ofexposure to PFOsolution), although contact angle measurements on samplesoxidized for 40 s and exposed to PFO for other shorter timeperiods gave values within or close to the error limits of thisvalue. Previous researchers have measured contact angles rangingfrom 113 to 123° for perfluorinated siloxane layers29,48 andpolytetrafluoroethylene,49,50consistent with our measurements.The contact angle measurements then are also consistent withthe grafting of PFO to the PDMS substrate and the presence offluorinated hydrocarbons on the sample surface. The contactangles observed were also larger than that for unmodified PDMS(114( 5°) and significantly greater than those for PDMS samplesthat had undergone only oxidation (105( 5°). In the latter case,as the sample was tested some 6 h following plasma oxidationsome hydrophobic recovery may have occurred, and the contactangle may be even lower after a shorter interval between oxidationand the measurement.

Figure 2 also indicates that upon exposure of a non-oxidizedPDMS sample to PFO for up to 5 h the C 1s signal does notundergo any significant change in position or shape, indicatingthat no PFO has been grafted onto the surface. Similar resultswere observed for other exposure times studied. However, asnoted above, under these conditions a small F 1s signal couldstill be observed and, unlike in the oxidized cases where PFOis certainly grafted, the F signal is still increasing after 5 h ofexposure to the PFO solution. The most likely explanation forthis observation is that there is either physisorption of a smallamount of PFO onto the substrate or that PFO diffuses to someextent into the bulk of the PDMS. This latter explanation seemslikely given that we have observed that the toluene solvent usedhere leads to the swelling of the PDMS after several hours ofexposure.

XPS was also used to characterize the stability of the fluorinatedPDMS to aging. The F 1s/C 1s area ratio was monitored overa period of 7 days for a series of PDMS samples that hadundergone 40 s of plasma oxidation followed by 4 h ofexposureto PFO solution. Each sample was stored in air, without anyspecial precautions taken to minimize exposure to humidity.Note that each data point was collected for a different samplethat, other than the period of time elapsed before performing theXPS experiment, was prepared in an identical fashion. The relativestrength of the F 1s signal increased slightly, by 12%, relativeto the C 1s signal. The C 1s XPS spectra did not show anysignificant change in peak shape or position upon aging, in

comparison to that of a newly prepared sample. Contact anglemeasurements on the aged samples gave values of 120( 3° after1 day and 121( 3° after 7 days, which is, within experimentalerror, equivalent to the value for a newly prepared sample andalso suggests that the surface retains its fluorinated nature overreasonable time periods.

Such behavior is quite different than that observed for othermodified PDMS surfaces. Oxidized PDMS reverts to its originalhydrophobic nature in less than 24 h if stored in air, whereassurfaces modified by grafting aminopropyltriethoxysilane or2-(4-chlorosulfonylphenyl)ethyltrimethoxysilane to produce amineor sulfonic acid-terminated surfaces, respectively, also begin toreduce the density of these functional groups at the surface within48 h. In the case of oxidized PDMS, the loss of oxide sites atthe surface has been attributed to the diffusion of short-chainoligomers of PDMS, broken up during the initial oxidationprocess, diffusing to the surface region. The main driving forcebehind this is presumably the reduction in free energy affordedwhen hydrophobic methylsiloxane chains replace more hydro-phobic siloxy groups at the surface. Surfaces on which hydrophilicgroups have been grafted also exhibit an increase in hydropho-bicity over time, although the speed at which this occurs is reducedpresumably because a cross-linked layer of the grafted ethoxy-silane derivative at the surface restricts diffusion. However, witha fluorophilic layer grafted onto the PDMS, the surface freeenergy is lower than that of the unmodified substrate, and thedriving force for oligomer diffusion to the surface is now absent.

Zeta Potential Measurements.In addition to successfullyproducing a stable fluorinated material to incorporate within amicrofluidic device, the surface must also support electro-osmoticflow if liquids within the device are to be pumped electroki-netically. Thus, the electroosmotic flow rate for devices madefrom fluorinated PDMS was measured and compared to that fordevices made from both unmodified and oxidized PDMS. Theelectroosmotic flow rate was measured using the currentmonitoring method at various pH values. The flow rate allowsus to obtain the electroosmotic mobility,µeo, and hence the zetapotential,ú, at the polymer surface

whereη is the solution viscosity,εo is the electrical permittivityof vacuum, andεo is the dielectric constant for water. The resultingzeta potential values (in this case, all negative) for various PDMSsurfaces are plotted in Figure 3.

As can be seen in Figure 3, the unmodified PDMS surfaceshows the slowest flow rate and hence the smallest zeta potential.Below a pH of 4.0, the flow rate was slow enough that themagnitude of the zeta potential was close to the sensitivity levelof our measurement technique. Above a pH of 4.0, the zetapotential was on the order of-37 mV, increasing to-50 mVat pH values above 8.0. We have previously observed similarbehavior on unmodified PDMS at isolated pH values of 3.0 and8.0. Unmodified PDMS thus supports electro-osmotic flow, albeitweakly, and this has been observed by ourselves and severalother workers.16 Unmodified PDMS should contain no surfaceionizable sites under these conditions and several models havebeen suggested to explain this observation, but this remains amatter of some controversy. Because the zeta potential is a directmeasure of the charge density at the surface of shear, it containscontributions not only from any ionized groups on the surfacebut also from any chemisorbed or physisorbed ions within theStern layer. It may well be that physisorbed species are an

(48) Cui, G.; Xu, H.; Xu, W.; Yuan, G.; Zhang, D.; Jiang, L.; Zhu, D.;Chem.Commun.2005, 277.

(49) Fang, Z.; Qiu, Y.; Luo, Y.J. Phys. D2003, 36, 2980.(50) Gumpenberger, T.; Heitz, J.; Baeuerle, D.; Rosenmayer, T. C.Appl. Phys.

A 2004, 80, 27.

Table 1. Contact Angles and F 1s/C 1s XPS Area Ratios forVariously Modified PDMS Samples

PDMS sample contact angle (deg) F/C area ratio

unmodified PDMS 114( 5oxidized PDMS 105( 5fluorinated PDMS 119( 5 33.2fluorinated PDMS aged for 1 day 120( 3 33.6fluorinated PDMS aged for 7 days 121( 3 37.2

µeo )εoεrú

η(1)

1084 Langmuir, Vol. 24, No. 3, 2008 Wang et al.

important contributor to the zeta potentials observed on theunmodified PDMS.

The magnitude of the zeta potential increased considerablyafter plasma oxidation of the PDMS surface, indicating theformation of more charged sites. The zeta potential is highlypH-dependent in this case, and the slope of the zeta potentialcurve changes notably at a pH of 4.0. This suggests that siteswith a surface pKa (pKa ) -log Ka, where Ka is the aciddissociation equilibrium constant) of 3.0-4.0 make a largecontribution to the overall zeta potential on the surface. Belowa pH of 4.0, these sites rapidly deprotonate with increasing pH,leading to a large increase in the magnitude of the zeta potential.Above a pH of 4.0, the surface is saturated with deprotonatedsites, and the slope of the pH-zeta potential curve is markedlydecreased. In all cases, the zeta potential of the oxidized PDMSis about 20 mV lower than that of the unmodified PDMS. Theincrease in zeta potential magnitude and the finding that thesurface pKa is about 4.0, indicating the presence of SiOH groups,is consistent with our previous chemical force titration results,together with electroosmotic mobility measurements at a morelimited range of pH values.

For PDMS substrates that underwent reaction with PFO toform a fluorinated surface, Figure 3 demonstrates that themagnitude of the zeta potential observed is considerably greaterthan that of unmodified PDMS. It is comparable to that observedfor oxidized PDMS. The higher flow rate demonstrates that thesurface charge density was not strongly affected by the fluorinemodification process. Aging of the sample, even for as much as7 days, did not significantly affect the measured zeta potentials.

The fact that modified samples did not undergo changes intheir zeta potential after as much as 7 days of storage is certainlyconsistent with the XPS measurements that indicated little changein the surface chemistry of the fluorinated polymer. However,the fact that the PDMS terminated with perfluoroalkyl groupshad zeta potentials comparable to those of oxidized PDMS isquite surprising. Thus, we carried out further experiments,described below, to determine if there was strong evidence forresidual ionizable sites on the perfluorinated PMDS. Regardlessof the mechanism, the continued enhanced zeta potential uponfluorination and, in particular, the marked stability of thesematerials in supporting electro-osmotic flow over time periodsof at least days to weeks is an important observation. Whereasother surface-modification schemes have been shown to providecharged surface that have lifetimes of more than a few days,21,23,24

this scheme is relatively less complicated than most and also

provides a surface that is both hydrophobic and supports electro-osmotic flow over a wide pH range. Such a material may be verypractical for constructing practical polymer-based microfluidicsystems.

Chemical Force Titrations. Figure 4 shows the adhesiveforce as a function of pH between a PDMS surface that hadundergone oxidation for 40 s followed by a 4 hexposure to PFOsolution and a Au-coated AFM tip terminated with a self-assembled monolayer of 12-thiododecanoic acid (COOH-terminated). Similar experiments were carried out using tipsterminated with dodecanethiol (methyl-terminated) and perfluo-rodecanethiol (fluoro-terminated). In the case of the methyl- andfluoro-terminated tips, the force titration profiles show a slightincrease in tip-sample adhesive interaction at a pH of about 7.0,with a stronger drop off in adhesive interaction at higher pHvalues. The average force observed with the methyl-terminatedtip was 28( 13.4 nN whereas that observed with the fluoro-terminated tip was lower at 8( 4.5 nN. The same experimentwas run on samples that had been allowed to age for 1 day andfor 7 days, and the force titration profiles and average forcesobserved were the same, within experimental error.

The best test for SiOH sites on the fluorinated PDMS surfaceis to obtain the force titration profile of a COOH-terminated tip.We and other workers16,22,51using hydroxylated tips on oxidizedPMDS clearly see a large change in the force titration profile ata pH of 4.0-5.5. The peak occurs at a pH where the maximumnumber of ionic H bonds can be formed, halfway between thesurface pKa of the SiOH-terminated sample (3.0) and that of theOH-terminated tip (5.0). Here, however, we cannot observe sucha peak, at least within the error range of the experiment. Thissuggests that residual SiOH sites, which could support electro-osmotic flow on the fluorinated sample, cannot be present insignificant quantities. Note that all the electro-osmotic flowexperiments were carried out at a constant ionic strength of 30mmol/L whereas the force titration experiments were carried outat a lower ionic strength of 1 mmol/L. Because the ionic strengthwas held constant at all but the highest pH of 12.0, this suggeststhat a possible explanation for the high surface charge on thefluorinated surface was the preferential adsorption of OH- anions.This effect must be more pronounced than on unmodified PDMSbecause in that case force titration profiles have not shown anysignificant variation in force over the entire pH range.

Retention of Fluorinated Peptides by Mass Spectrometry.Fluorous tags have become a popular method of targeting andtagging protein or peptides at specific sites (e.g., glycosylation).Once the biochemical species possesses a fluorous tag, it willbe retained or physisorbed onto a surface with a similar fluorous

(51) Song, J.; Duval, J. F. L.; Stuart, M. A. C.; Hillborg, H.; Gunst, U.;Arlinghaus, H. F.; Vancso, G. J.Langmuir2007, 23, 5430.

Figure 3. Zeta potential as a function of pH as determined byelectro-osmotic flow measurements on microfluidic chips containingmicrochannels of (A) unmodified PDMS, (B) PDMS exposed toplasma oxidation, (C) PDMS exposed to plasma oxidation followedby exposure to PFO solution to form a fluorinated surface, and (D)the same as for curve C after aging for 7 days.

Figure 4. Force titration profiles showing the adhesive interactionas a function of pH between a fluorinated PDMS substrate and AFMtips terminated with 12-thiohexadecanoic acid.

Surface Modification of Poly(dimethylsiloxane) Langmuir, Vol. 24, No. 3, 20081085

character. To assess the ability of the fluorinated PDMS toselectively retain a fluorinated peptide, a fluorinated cortactinpeptide was physisorbed from aqueous solution onto variousPDMS substrates.

The PDMS substrates were then washed with water, followedby a wash with methanol to extract any residual fluorous-taggedpeptide from the surface. Methanol is commonly used as a stronglyeluting solvent in fluorous chromatography. Figure 5 shows theMALDI-TOF spectra of the methanol wash fraction from theunmodified PDMS, oxidized PDMS, and fluorinated PDMSsubstrates. Only in the case of fluorinated PDMS is there a seriesof peaks with the primary peak at the expectedm/z value of1721. The peak area profiles are consistent with those expectedfrom the isotopic pattern for a species of this elementalcomposition. This result demonstrates that fluorinated PDMS isable to retain the fluorinated peptide during the water washingstage, unlike the unmodified and oxidized substrates. This resultis consistent with that reported by Go et al.,27 who found thatfluorous-tagged glucose and tyrosine could be readily extractedfrom a fluorinated amorphous Si substrate using solutionscontaining more than 60% methanol in water but were selectivelyretained when the same substrates were washed with water.

Conclusions

The surface modification of PDMS using perfluoro-1,1,2,2-tetrahydrooctyl-1-triethoxysilane and the subsequent aging effecton this modified surface have been characterized by X-rayphotoelectron spectroscopy, electro-osmotic flow, and contactangle measurements. Functioning microfluidic devices have alsobeen constructed using the fluorinated PDMS polymer.

XPS showed that a layer of grafted PFO molecules could besuccessfully grown on an oxidized PDMS substrate. Contactangle and chemical force titrations also supported this conclusion.The F 1s XPS signal grew slightly in intensity, relative to theC 1s and Si 2p signals, when the modified PDMS was stored inair for up to 7 days, indicating that the diffusion of hydrophobic

dimethylsiloxane oligomers to the surface region, usuallyobserved in surface-modified PDMS, was blocked by the low-surface-energy fluorocarbon layer.

The fluorinated PDMS microchips showed excellent flowperformance at various pH values from pH 3 to 10, comparedwith unmodified and oxidized PDMS devices, and indicated thatthe surface supports a zeta potential of some-50 to -70 mVover this pH range. Because the zeta potential increases somewhatat higher pH and the tip-sample adhesive interaction also fallsoff at pH values above 8.0, this suggests that the preferentialadsorption of OH- from solution may be at least partiallyresponsible for this effect. Significantly, the fluorinated PDMSdevices did not show the aging effects that degrade the flowperformance of oxidized and other PDMS surface-modificationstrategies.The resultsdemonstrate that facile fluorinemodificationof the PDMS devices can significantly eliminate the aging effectand generate a surface with significant zeta potential. Therefore,the fluorinated modification process is an effective means ofimproving the flow performance and durability of surfacemodifications for microfluidic devices made from PDMS. Massspectrometric investigations also demonstrate that the fluorinatedPDMS substrate is able to selectively adsorb a fluorous-taggedpeptide in aqueous solutions but to release the same peptidewhen washed with methanol, showing that this surface-modification scheme is also potentially useful as a means oftargeting the enrichment of selected chemical species from morecomplex mixtures.

Acknowledgment. We acknowledge the Natural Sciencesand Engineering Research Council of Canada for financialsupport.

Supporting Information Available: Si 2p XPS spectra ofunmodified and oxidized PDMS. This material is available free of chargevia the Internet at http://pubs.acs.org.

LA702038T

Figure 5. MALDI-TOF spectra obtained from a methanol wash solution on both modified and unmodified PDMS substrates that had beenpreviously exposed to an aqueous solution of 2.5µmol/L F-CTN.

1086 Langmuir, Vol. 24, No. 3, 2008 Wang et al.