Chemical Modifications of Au/SiO2 Template Substrates for Patterned
Biofunctional Surfaces
Elisabeth Briand,† Vincent Humblot,‡ JessemLandoulsi,‡ Sarunas Petronis,† Claire-Marie Pradier,‡
Bengt Kasemo,† and Sofia Svedhem*,†
†Department of Applied Physics, Chalmers University of Technology, 412 96 G€oteborg, Sweden, and‡Laboratoire de R�eactivit�e de Surface, UMR CNRS 7197, Universit�e Pierre et Marie Curie - Paris VI, 75252
Paris Cedex 05, France
Received May 10, 2010. Revised Manuscript Received July 16, 2010
The aim of this work was to create patterned surfaces for localized and specific biochemical recognition. For thispurpose, we have developed a protocol for orthogonal and material-selective surface modifications of microfabricatedpatterned surfaces composed of SiO2 areas (100 μm diameter) surrounded by Au. The SiO2 spots were chemicallymodified by a sequence of reactions (silanization using an amine-terminated silane (APTES), followed by aminecoupling of a biotin analogue and biospecific recognition) to achieve efficient immobilization of streptavidin in afunctional form. The surrounding Au was rendered inert to protein adsorption by modification by HS(CH2)10CONH-(CH2)2(OCH2CH2)7OH (thiol-OEG). The surface modification protocol was developed by testing separately homo-geneous SiO2 and Au surfaces, to obtain the two following results: (i) SiO2 surfaces which allowed the grafting ofstreptavidin, and subsequent immobilization of biotinylated antibodies, and (ii) Au surfaces showing almost no affinityfor the same streptavidin and antibody solutions. The surface interactions were monitored by quartz crystalmicrobalance with dissipation monitoring (QCM-D), and chemical analyses were performed by polarization mod-ulation-reflexion absorption infrared spectroscopy (PM-RAIRS) and X-ray photoelectron spectroscopy (XPS) toassess the validity of the initial orthogonal assembly of APTES and thiol-OEG. Eventually, microscopy imaging of themodified Au/SiO2 patterned substrates validated the specific binding of streptavidin on the SiO2/APTES areas, as wellas the subsequent binding of biotinylated anti-rIgG and further detection of fluorescent rIgG on the functionalized SiO2
areas. These results demonstrate a successful protocol for the preparation of patterned biofunctional surfaces, based onmicrofabricated Au/SiO2 templates and supported by careful surface analysis. The strong immobilization of thebiomolecules resulting from the described protocol is advantageous in particular for micropatterned substrates forcell-surface interactions.
Introduction
An increasing number of applications, especially in biotechnol-ogy, requires surfaces with defined regions of different chemicalfunctionality to achieve site-specific attachment of one species insome areas while minimizing unwanted surface interactions inother areas.1 Some strategies to achieve such patterned surfacesinvolves the modification of homogeneous substrates by micro-contact printing techniques, through transfer of organic com-pounds on defined positions of the substrate,2,3 or by chemicalmodification of some parts of the surface, for example, viaelectron irradiation4 or extreme UV interference lithography,5
potentially followed by the replacement of the nonmodifiedmolecules by another species. Another approach is based on thecombination of different (inorganic) substrate materials for thecreation of the patterned template surface, commonly via photo-lithography. In this case, the patterned inorganic surfaces aredesigned for orthogonal chemical modifications to obtain thedesired final functionally patterned substrates, a procedure whichtakes advantage of the affinity of certainmolecular groups toward
specific materials. This last strategy has been termed ‘‘orthogonalself-assembly” by Laibinis et al.,6 “selective molecular assemblypatterning” (SMAP) by Michel et al.,7 and “substrate selectivepatterning” (SSP) by Bergkvist et al.8 Various chemical com-pounds and solid substrates have been used in such surfacemodification protocols, for example, TiO2/SiO2 templates mod-ified (sequentially) with alkylphosphates and PLL-g-PEG,7,9 Au/AlO3 templates functionalized with alkanethiols and alkanephosphates or carboxylic acids,6,10 andAu/SiO2 templatesmodifiedand with alkanethiol and poly(ethylene glycol) (PEG)-silanes.8
In this report, we have developed a protocol for biofunctionalmodification of Au/SiO2 templates by, first, amino silane(APTES) and, second, thiol-PEG. This choice was made to allowfor a thermal treatment after the silanization to stabilize thestructure of the silane,11-13 knowing that the alkylthiol layers are
*To whom correspondence should be addressed.(1) Schmidt, R. C.; Healy, K. E. J. Biomed. Mater. Res. 2009, 90A, 1252–1261.(2) James, L. W.; Amit, K.; Hans, A. B.; Enoch, K.; George, M. W. Nanotech-
(8) Bergkvist, M.; Niamsiri, N.; Strickland, A. D.; Batt, C. A. Surf. Sci. 2008,602, 2121–2127.
(9) Michel, R.; Lussi, J. W.; Csucs, G.; Reviakine, I.; Danuser, G.; Ketterer, B.;Hubbell, J. A.; Textor, M.; Spencer, N. D. Langmuir 2002, 18, 3281–3287.
(10) Burdinski, D.; Saalmink, M.; van den Berg, J. P. W. G.; van der Marel, C.Angew. Chem., Int. Ed. 2006, 45, 4355–4358.
(11) Guo, Z.; Guilfoyle, R. A.; Thiel, A. J.; Wang, R.; Smith, L. M. NucleicAcids Res. 1994, 22, 5456–5465.
(12) Krasnoslobodtsev, V.; Smirnov, S. N. Langmuir 2002, 18, 3181.(13) Pasternack, R. M.; Rivillon Amy, S.; Chabal, Y. J. Langmuir 2008, 24,
known to be unstable when submitted to thermal treatment andto decompose when heated above 350 K.14 Thus, starting from Siwafers processed by photolithography into Au-coated substrateswith SiO2 spots (100 μm in diameter), we arrive at a simpleprocedure to obtain patterned surfaces exposing specificallyfunctionalized areas surrounded by a surface resistant to proteinadsorption. The surface modification protocol was designed toresult in a protein resistant surface, with biotinylated spots ontowhich streptavidin can be bound, followed by another biotiny-lated biomolecule of choice (Figure 1), here biotinylated anti-bodies. The biotin/streptavidin system was chosen for its well-known properties and the strong binding between the twocompounds, and the general strategy developed in this workcan be transposed easily to other applications involving otherchemicals. The aim of this work is to demonstrate the spatiallydefined immobilization of biofunctional proteins in templatedpatterns. The strong binding of compounds to the template sur-faces makes this protocol attractive for use in cell-substrateinteraction studies, for example, addressing issues related to long-termstability of thepattern templatewhenused for cell culture.15-18
Four different surface analytical techniques were used tocharacterize the successive steps of the presented surface mod-ification protocol: polarization modulation-reflection absorp-tion infrared spectroscopy (PM-RAIRS), x-ray photoelectronspectroscopy (XPS), quartz crystal microbalance with dissipa-tion monitoring (QCM-D), and fluorescence microscopy. Thesevarious techniques offer complementary structural and functionalinformation at the successive process steps.
Material and Methods
Chemicals. 3-(Aminopropyl)triethoxysilane (APTES), bioti-namidohexanoyl-6-aminohexanoic acid N-hydroxysuccinimideester (biotin-NHS), D-biotin, mouse monoclonal anti-rabbit IgGbiotin conjugate (biotinylated anti-rIgG), FITC-conjugated rab-bit IgG (FITC-rIgG), and phosphate buffered saline (PBS)(10 mM, pH 7.4) tablets were purchased from Sigma-Aldrich,Cy5-labeled streptavidin was obtained from GE Healthcare, andHS(CH2)10CONH(CH2)2(OCH2CH2)7OH (thiol-OEG) wasfrom Polypure SA (Oslo, Norway). Solvents were of analyticalgrade. Water was filtered and deionized using a Milli-Q unit.
Substrates. For PM-RAIRS and XPS measurements, Siwafers were coated with 200 nm of Au by e-beam evaporationat a rate of 1 A/s using an AVAC HVC600 general purposemultimaterial evaporator. The coated wafer was then cut intosamples of 1 cm �1 cm.
For QCM-D measurements, commercially available crystalswith gold electrodes (Q-Sense, Sweden) were used either withoutfurther modification or after a subsequent coating of 5 nm of Ti(as an adhesion layer) and 50 nm of SiO2 by e-beam evaporation(AVAC HVC600) at a deposition rate of 1-2 A/s. Auger experi-ments on themodified quartz crystals (data not shown) confirmedthe presence of SiO2
Patterned surfaces were prepared as follows: A double-side-polished30 0 Siwaferwas coatedwith photoresist (Shipley 1813) byspin coating (2000 rpm for 1 min). The resist then was soft-bakedona hot-plate for 2min at 110�Cand exposed toUV light (400nmwavelength, 10mW/cm2 intensity, 10 s exposure time) through thephotolithographic mask using a Karl S€uss MJB2 aligner. Thewafer was then dipped into a bath of MICROPOSIT MF-319developer for 1min to remove the illuminatedphotoresist areas.A5 nm adhesion layer of Ti and 70 nm of Au were then evaporatedon topof the substrate and the unexposed photoresist. Finally, theremaining photoresist with a top Ti/Au layer was removed by lift-off technique in a bath of acetone under ultrasonic agitation,leaving the patterned Au layer on the substrate.
The different origin of the SiO2 layer on the QCM crystals(physical vapor deposition) and the Si wafers (spontaneouslyformed oxide layer) was not expected to change its reactivitytoward silanes. The only significant differences between the twoSiO2 layers would concern the roughness of the substrate, whichwe do not take into account in this study.
Chemical Modification. Cleaning. Immediately beforeuse, the surfaces were cleaned twice by treatment in a UV-ozonechamber (15 min), followed by ultrasonication (5 min) in each ofacetone, isopropanol, and water. The samples were then driedunder a stream of nitrogen.
Silanization. The silanization protocol followed was the onedescribed byGuo et al.11 A solution ofAPTES at a concentrationof 2% v/v was prepared in a mixture of acetone and water (95/5).The silanes were allowed to hydrolyze for 2 h, and then thesubstrates were introduced into the solution for 30 min. Afterthorough rinsing with acetone, the surfaces were dried andannealed at 110 �C for 2 h. To modify the NH2 terminated silaneself-assembled monolayer (SAM), a solution of biotin-NHS inPBSat 1mg/mLwasdeposited on the surfaces for 1 h, followedbyextensive rinsing with water.
Thiolation. A 1 mM solution of thiol-OEG was prepared inethanol. The samples were dipped into the thiol solution for atleast 18 h before extensive rinsing with ethanol.
Biospecifically Patterned Surfaces. After cleaning of thesubstrates, they were functionalized by silanization followed bythiolation (as described above).After thorough rinsing by ethanoland drying under a stream of nitrogen, 100 μL of each of thefollowing five solutions in PBS was successively placed on top ofthe substrates: (i) biotin-NHS 1 mg/mL for 1 h, (ii) Cy5 labeledstreptavidin 20 mg/mL solution for 1 h, (iii) biotinylated anti-rIgG 100 mg/L for 1 h, and (iv) FITC labeled rIgG 40 mg/L for1 h. After each immersion, the surfaces were thoroughly rinsed
Figure 1. Schematic representation of the chemical functionaliza-tion strategy for theAu/SiO2 patterned template surface in order tospatially control the immobilization of bioactive molecules (not toscale).
(14) Bain, C. D.; Troughton, E. B.; Tao, Y.-T.; Evall, J.; Whitesides, G. M.;Nuzzo, R. G. J. Am. Chem. Soc. 1989, 111, 321–335.(15) Falconnet, D.; Csucs, G.; Grandin, H. M.; Textor, M. Biomaterials 2006,
27, 3044–3063.(16) Lensen, M. C.; Schulte, V. A.; Salber, J.; Diez, M.; Menges, F.; M€oller, M.
Pure Appl. Chem. 2008, 80, 2479–2487.(17) Selhuber-Unkel, C.; Lopez-Garcia, M.; Kessler, H.; Spatz, J. Biophys. J.
2008, 95, 5424–5431.(18) Kilian, K. A.; Bugarija, B.; Lahn, B. T.; Mrksich, M. Proc. Natl. Acad. Sci.
with PBS and water. All the concentrations used in this work arein the range of what is usually used in the elaboration ofimmunosensors.
Experimental Techniques. PM-RAIRSMeasurements.The Fourier transform infrared (FT-IR) instrument used in thiswork was a commercial NICOLET 5700 Nexus spectrometer.The external beamwas focused on the sample with a mirror at anoptimal incident angle of 85�. A ZnSe grid polarizer and a ZnSephotoelastic modulator, modulating the incident beam between pand s polarizations (HINDS Instruments, PEM 90, modulationfrequency =37 kHz), were placed prior to the sample. The lightreflected at the sample was then focused on a nitrogen-cooledMCT detector. The sum and difference interferograms wereprocessed andFourier-transformed to yield the differential reflec-tivityΔR/R=(Rp-Rs)/(RpþRs) which is the PM-RAIRS signal.A total of 128 scans were recorded at 8 cm-1 resolution for eachspectrum.
X-ray Photoelectron Spectroscopy. XPS analyses were per-formed using a SPECS (PhoibosMCD 150) X-ray photoelectronspectrometer (SPECS,Germany) equippedwith a nonmonochro-matized magnesium X-ray source (hν = 1253.6 eV) powered at10 mA and 15 kV, and a Phoibos 150 hemispherical energyanalyzer. The resulting analyzed area was 5 mm in diameter. Apass energy of 20 eV was used for the survey scan and 10 eV fornarrow scans. The samples were fixed on the support usingdouble-sided adhesive tape, and no charge stabilization devicewas used. The pressure in the analysis chamber during measure-ment was around 10-10 Torr or less. The photoelectron collectionangle between the normal to the sample surface and the analyzeraxis was 0�. The following sequence of spectra was recorded:survey spectrum, O 1s, C 1s, Au 4f, N 1s, S 2p, and Si 2p. Thebinding energy scale was set by fixing the C 1s component due tocarbon only bound to carbon and hydrogen at 284.8 eV. The datatreatment was performed with the Casa XPS software (CasaSoftware Ltd., U.K.). Unless stated otherwise, the peaks weredecomposed using a linear baseline, and a component shapedefined by the product of a Gauss and Lorentz function in a70:30 ratio, respectively. Molar concentration ratios were calcu-lated using peak areas normalized according to Scofield factors.19
QCM-D Measurements. QCM-D measurements were con-ducted using a commercial instrument (QCM-DE4, Q-sense AB,Sweden) at a temperature of 22 ( 0.1 �C. The device has beendescribed in detail elsewhere.20,21 Briefly, oscillations of AT cutquartz crystals at the resonant frequency (here 5 MHz) or at oneof its overtones (15, 25, 35, 45, 55, 65 MHz) are obtained whenapplying AC voltage. The drive circuit is then open-circuited, andthe exponential decay of the oscillation amplitude is monitored.The dissipation, D, is defined as the fraction of energy of theoscillation that is dissipated during one period of oscillation.Upon adsorption ofmaterial on the crystal surface, the resonancefrequency of the crystal (Δf) decreases and the dissipation shift(ΔD) reflects the viscoelastic properties of the adlayer. Solutionswere injected into the measurement cell using a peristaltic pump(Ismatec IPC-N 4) with a flow rate of 100 μL/min. All frequencyshifts were normalized with the overtone number. Prior to theprotein adsorption, a PBSbuffer solutionwas injected to establisha stable baseline.
Fluorescence Microscopy. Experiments were carried outusing a fluorescence microscope (BX61, Olympus, Germany)with a 20� water immersion objective (UMPLFLN-20XW).The samples were illuminated by a Hg lamp and analyzed usingtwo wide bandpass fluorescence filters (U-MWIB2 andU-MWG2,Olympus,Germany). The imageswere processedwithanalysis software (Soft Imaging, Olympus, Germany) in multiplefluorescence mode.
Results
The aim of this work was to fabricate patterned biofunctionalsurfaces, and we have chosen to demonstrate spatially definedimmunorecognition reactions through the site-specific immobili-zation of a biotinylated antibody. The surface modificationprotocol was based on a patterned inorganic SiO2/Au template,whichwasmodified by a sequenceof chemical reactions, such thatthe antibodies were specifically immobilized onto SiO2 whereasAu was modified to resist protein adsorption.Immobilization of Biotinylated Antibodies to SiO2 Sur-
faces.As a first step in the development of surfaces with antibodypatterns, SiO2 films deposited onto QCM-D sensors were mod-ified with APTES followed by the covalent grafting of biotin-NHS. The further binding of streptavidin to this surface wasmonitored in situ by QCM-D, and the specificity of the interac-tion was tested by probing the adsorption of streptavidin whichhad been presaturated with biotin.22 The QCM-D frequency anddissipation shifts (Δf and ΔD) were monitored for the twoexperiments, and the results are displayed in Figure 2. The levelof nonspecific binding of the presaturated streptavidin on theAPTES and biotin-NHSmodified surfacewas very low, about-2Hz (note that negative frequency shifts correspond to massuptake). In contrast, when a solutionof nonsaturated streptavidinwas in contact with the biotinylated surface (Figure 2), a fre-quency shift of-24( 1 Hz was recorded. The Δf value obtainedby QCM-D for the layer of streptavidin was similar to previousstudies23-26 and suggested a full coverage of streptavidin (i.e., astreptavidinmonolayer) on the biotinylated surface.The recordedfrequency shift corresponded to a coverage of ∼400 ng/cm2,according to the Sauerbrey equation.21,27 The low level of dis-sipation (<0.2 � 10-6) indicated that the immobilized streptavi-din molecules formed a compact and rigid structure.24 Previousstudies using QCM-D together with complementary techniquesto characterize the layer of streptavidin adsorbed on surfacesshowed similar results, and concluded a proportion of coupledwater of∼50%25,26 and an optimal orientation of themolecules.25
A small fraction of the streptavidin (2 Hz) was removed uponrinsingwith detergent (10mMSDS solution), indicating that onlya small fraction of streptavidin (or biotin) was reversibly bound tosome part of the surface and that the streptavidin molecules weremostly strongly attached to the surface via specific binding withbiotin moieties. The streptavidin molecules bound to the surfacewere functional, since biotinylated IgG antibodies (anti-rIgG)could be immobilized on the surface (-13 Hz, see Figure 2) whilea solutionofBSA induced only a low frequency shift (-2Hz). Thelatter corresponds to about the sameamount ofmaterial desorbedafter SDS rinsing (see above) and is likely related to defects in thestreptavidin layer. The dissipation signal increased significantlyafter the binding of anti-rIgG, and Voigt-based modeling28 wasused to determine the amount of hydrated anti-rIgG adsorbed onthe surface. Assuming a protein density of 1.4, the amount ofhydrated anti-rIgG per geometrical area is 1200 ng/cm2. In aprevious study, the dry mass and the hydrated mass for antibody(IgG) bound to various proteins adsorbed on surfaces wascompared.29 For all experiments, the proportion of coupledwater
(19) Scofield, J. H. J. Electron Spectrosc. Relat. Phenom. 1976, 8, 129–137.(20) H€o€ok, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal.
Chem. 2001, 73, 5796–5804.(21) Rodahl, M.; H€o€ok, F.; Kasemo, B. Anal. Chem. 1996, 68, 2219–2227.
(22) Pradier, C.-M.; Salmain, M.; Zheng, L.; Jaouen, G. Surf. Sci. 2002,502-503, 193–202.
(23) Khin, M.; Aung, M.; Ho, X.; Su, X. Sens. Actuators, B 2008, 131, 371–378.(24) Larsson, C.; Rodahl, M.; H€o€ok, F. Anal. Chem. 2003, 75, 5080–5087.(25) Su, X.-L.; Li, Y. Biosens. Bioelectron. 2005, 21, 840–848.(26) Edvardsson, M.; Svedhem, S.; Wang, G.; Richter, R.; Rodahl, M.;
Kasemo, B. Anal. Chem. 2009, 81, 349–361.(27) Sauerbrey, G. Z. Phys. 1959, 155, 206–222.(28) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Phys. Scr. 1999, 59,
in the hydrated mass measured by QCM-D was around two-thirds of the totalmass.Here, this proportionwas used to roughlyestimate the binding ratio between biotinylated anti-rIgG andstreptavidin, taking into account the respective molecular massesof the two proteins. It was found that 1 biotinylated anti-rIgGmolecule bound per 1.5 streptavidin molecules. Considering thesizes of the proteins (immunoglobulin G is about twice as big asstreptavidin), this result is expected and the surface presents agood streptavidin/biotinylated-IgG binding ratio. In a subse-quent step, the antigens (rIgG) were successfully bound to theimmunoreactive sensor surface. The ΔD-Δf slope (Figure 2,inset) was identical for the binding of both molecules, indicatingsimilar viscoelastic properties. Since both molecules are immu-noglobulin G, we assume a similar water content for both layersand an antigen/antibody ratio of 0.75 can be estimated.Passivation ofGold. QCM-DResults.As a second step in
the development of surfaces with patterns of immobilized anti-bodies, the resistance to protein adsorption of (nonpatterned) Ausurfaces modified by thiol-OEG SAMs was evaluated. The SAMlayer was formed ex situ on Au-coated quartz crystal, after whichit was tested toward nonspecific adsorption of proteins usingQCM-D (Figure 3). Low amounts of streptavidin (-3 Hz) andbiotinylated antibodies (-4 Hz) and absolutely no antigen (0 Hz)could be detected on the substrates. The nonspecifically adsorbedmaterials were almost fully removed by rinsing with SDS; that is,they were weakly and reversibly bound.
On the patterned surfaces, silanization proceeded before thestep where thiols were added to passivate the Au regions inbetween the SiO2 spots. Therefore, the very same sequence ofsteps was performed on the (nonpatterned) Au substrates inorder to investigate the orthogonality of the reactivity ofAPTES and thiol-OEG on Au. PM-RAIRS (Figure 4) andXPS analyses (Figure 5, Tables 1 and 2) were performed on threetypes of samples: Au after exposure to APTES, Au after thiolinteraction (thiol-OEG), and Au after both exposures (APTESþthiol-OEG).
PM-RAIRS Results. The PM-RAIRS spectra are displayedin Figure 4 for two regions of interest: the one corresponding tothe symmetric and asymmetric C-H stretching of the CH2 andCH3 moieties (3100-2800 cm-1) and the one corresponding tothe region comprised between 900 and 1900 cm-1 (for the othermoieties of the siloxane and thiol molecules).
On Au substrates exposed to APTES (lower spectrum), the IRspectrum indicates the presence of the silane.30 The peak at 2964cm-1 is characteristic of the asymmetric C-H stretching in CH3
moieties, and the CH2 symmetric and asymmetric stretchingbands appear at 2857 and 2927 cm-1, respectively. The broad
Figure 2. Δf (black) and ΔD (gray) recorded on a SiO2 þAPTESþ biotin-NHS coated quartz crystal exposed to a solution ofpresaturated streptavidin (20 mg/L), streptavidin (20 mg/L), bio-tinylated anti-rIgG (100 mg/L), and rIgG (30 mg/L). BSA solu-tions at 100mg/Lwere used to test the level of nonspecific binding.Inset: ΔD-Δf plots for the binding of streptavidin (black), bioti-nylated anti-rIgG (gray), and rIgG (light gray).
Figure 3. Δf (black) andΔD (gray) recorded during physisorptionof streptavidin (20 mg/L), biotinylated anti-rIgG (100 mg/L), andrIgG (30 mg/L) on a thiol-OEG-OH coated Au QCM-D crystal.
Figure 4. PM-RAIRS spectra of Au substrates after functionali-zation by thiol-OEG, by APTES, and by APTES then thiol-OEG.
(29) H€o€ok, F.; V€or€os, J.; Rodahl, M.; Kurrat, R.; B€oni, P.; Ramsden, J. J.;Textor, M.; Spencer, N. D.; Tengvall, P.; Gold, J.; Kasemo, B. Colloids Surf., B2002, 24, 155–170.
(30) Kim, J.; Seidler, P.; Wan, L. S.; Fill, C. J. Colloid Interface Sci. 2009, 329,114–119.
peak at 1655 cm-1 can be attributed to amine deformation, whilethe IR band at 1150 cm-1 can be attributed to Si-O-C asym-metric stretching vibrations.
On Au surfaces modified with a SAM of thiol-OEG, we canobserve (i) the CH2 stretching band at lower wavenumbers,indicating a densely packed layer (νasymCH2
at 2919 cm-1)31,32
and (ii) the presence of intense bands at 1645 and 1550 cm-1,corresponding to the CdO stretching and to the N-H bendingcoupled with the C-N stretching, respectively, characteristic ofthe amide bond of the thiol. One may note the presence of a bandat 1720 cm-1, characteristic of νCdO in COOH groups. Weexplain this band by the presence on the surface of mercaptoun-decanoic acid molecules, likely remaining after the synthesis ofthiol-OEGmolecules, which would not alter the order of the thiollayer.33 Characteristic vibrations of OEG chains in both theamorphous and the crystalline states are alsoobserved.34-36EtherCH2 wagging (1355 cm-1) and twisting (1290 cm-1) vibrationsand the COC stretching band (1145 cm-1) are characteristic forOEG chains in an amorphous state. The presence of a secondpeak at 1120 cm-1 (COC stretching) and at 1265 cm-1 (ether CH2
twisting) indicates the presence of OEG chains in a helicalconfiguration. The latter band may include the contribution ofthe amide III band.
On Au surfaces exposed to both APTES and thiol-OEG, thespectrum bears a close resemblance to the one observed on theAufunctionalized with the thiol SAM except for some aspects. First,the peak position of the νasCH2
is at higherwavenumber comparedto the one recorded in the previous case (2927 vs 2918 cm-1).Second, shoulder was observed on the νas band that could beattributed to a small contributionofνasCH3
.Moreover, the specificfeatures of theOEGmoieties on the substrates were not as intenseas those on the thiolated surface. The broad peak between 1120and 1170 cm-1 clearly demonstrates the presence of both mole-cules coexisting on the surface (1150 cm-1 for APTESSi-O-C and1110 cm-1 for OEGC-O-C).
XPS Results. The surface chemical composition determinedby XPS on the functionalized Au surfaces is given in Table 1.Results obtained on the thiol-OEG sample showed that the S/Auratio was in the same order to that obtained previously insuccessful procedures (S/Au = 0.05).37 The sulfur was entirelybound to gold (components at about 162.8 eV, XPS peak notshown), as a significant amount of unbound sulfur would give apeak at lower binding energy and the presence of oxidized thiol, inthe form of sulfate, would lead to a peak around 168 eV.38 The S/Cmolar ratio obtained, the attenuation of sulfur signal beingneglected, was sensibly equal to the theoretical one (S/C=0.03) computed using the thiol-OEG molecule depicted inFigure 1. The APTES sample shows the presence of silicon in asignificant amount and an increase of the concentration ofnitrogen (Table 1). On the APTES þ thiol-OEG sample, themolar concentration of silicon is still high (Table 1). The S 2ppeak revealed the presenceof sulfur bound toAu (peakpositionatabout 162.8 eV, not shown), suggesting the self-assembly of thiol-OEG molecules despite the presence of silanes. However, theS/Au molar ratio was about 3 times lower than that obtained onAu-thiol-OEG samples (Table 1).
Typical C 1s and O 1s XPS peaks recorded on the threedifferent Au functionalized samples are presented in Figure 5.These peaks can be safely decomposed on the basis of literaturedata regarding polymers and other materials with biologicinterest.35 The C ls peak was decomposed in four componentswith the same fwhm (full width at half-maximum): (i) a compo-nent at 284.8 eV due to [C-(C,H)], (ii) a component at about286.4 eVdue to [C-(O,N)], (iii) a component near 288.0 eVdue to[CdO] (in particular in amide group) and [O-C-O] (acetal orhemiacetal), and (iv) the last component near 289.1 eV attributedto (CdO)-O-R. The O1s peak decomposition depended on thestudied sample. On the thiol-OEG sample, two components withthe same fwhm were used: a component at about 531.3 eVattributed to [CdO] and a component at 533.1 eV due to[C-O-H] or [C-O-C]. For both APTES and APTES-thiol-OEG samples, a contribution appeared clearly at about 532.1 eVdue to [Si-O] (Figure 5). The N 1s peak (not shown) presented acomponent at 399.5 eV (Nnonpr) attributed to amide or aminefunctions and a component at higher binding energy, about400.7 eV, only noticeable in the presence of APTES, indicatingthe presence of protonated amines (Npr). The molar fractionsassociated with these components are given in Table 2 in the formof mole concentration ratios with respect to the total carbonconcentration.
The C 1s peak showed a high contribution of the component at286.4 eV due to [C-(O,N)] in the presence of thiol-OEGcompared to APTES (Figure 5). Furthermore, the contribution
Figure 5. Decomposition of O 1s and C 1s peaks recorded onAu-APTES, Au-thiol, and Au-APTES-thiol .
Table 1. Surface Concentration (mole fraction in% computed over all
elements except hydrogen) of Elements Determined by XPS
(31) Dubois, L. H.; Zegarski, B. R.; Nuzzo, R. G. J. Chem. Phys. 1993, 1, 678–688.(32) Laibinis, P. E.; Whitesides, G. M.; Allara, D. L.; Tao, Y.-T.; Parikh, A. N.;
Nuzzo, R. G. J. Am. Chem. Soc. 1991, 113, 7152–7167.(33) Briand, E.; Humblot, V.; Pradier, C.-M.; Kasemo, B.; Svedhem, S. Talanta
2010, 81, 1153–1161.(34) Harder, P.; Grunze, M.; Dahin, R.; Whitesides, G. M.; Laibinis, P. E. J.
at 533.1 eV in the O 1s peak was much more pronounced on thesample surface modified with thiol-OEG. These findings supportthe presence of a PEG chain originating from the thiol-OEGmolecules. Indeed, XPS data, presented in Table 2, showed anexcellent agreement between C286.4/C and (2� O533.1 þNnonpr þNpr)/C for both thiol-OEG and APTES-thiol-OEG samples. Thepresence of carboxyl or ester function [(CdO)-O] in a smallamount may originate from unreacted thiol-COOH.
The interplay of the presence of silanes in a significant amountand of the thiol-OEG assembly on the Au surface may influencethe thickness of the adlayer. The apparent concentration ratio[C]/[Au] may be computed using the equation described in theAppendix. The adlayer average thickness estimated for thedifferent samples are given in Table 2. Silanized Au samples(APTES) lead to a thickness considerably higher than that of anAPTES monolayer (about 0.8 nm on the silica surface39). Thisresult may be explained by the polymerization of APTES, leadingto the formationof aggregates at the adsorbed state. In contrast tothe silica surfaces, the interaction of silanes on Au surfaces israrely described.40,41 It has been reported that the liquid-phase deposition may lead to the adsorption of prepolymerizedsilanes.41
It must be kept in mind that XPS data do not provideinformation about the organization of silanes on the Au surface;the value of the adlayer thickness depends on (i) the size ofthe presumable aggregates and (ii) the fraction of the surfacecoverage.
The presence of thiols on the silanized surface (sample APTESþ thiol-OEG) resulted in a slight increase of the adlayer thickness(from 2.0 to ca. 2.5 nm, Table 2). It is clearly shown that silanes,presumably in the form of aggregates, did not prevent the
attachment of thiols; however, they lead to a significant decreasein the amount of assembled thiol molecules. As the thicknessesobtained on APTES and thiol-OEG samples were almost thesame, the substitution of silanes by thiols cannot be observed,however not excluded. The increase of the adlayer thicknessobserved on the APTES þ thiol-OEG sample may be attributedto the disorderof assembled thiols, as indicated in the PM-RAIRSresults (see previous section).Demonstration of Orthogonal Functionality of the Pat-
terned Surface. The efficiency of the functionalization processon patterned substrates for the elaboration of immunoarrays wasfirst probed by the binding of fluorescently labeled streptavidin toSiO2/Au template substrates where the diameter of the SiO2 spotswas 100 μm. A characteristic micrograph recorded during theseexperiments is displayed in Figure 6, where the SiO2 areas becomered due to the binding of the labeled streptavidin. The contrastbetween the different parts of the pattern indicated that thestreptavidin was almost exclusively bound on the SiO2 patches,and very low fluorescence due to nonspecific binding of strepta-vidin or biotin to the regions in between the SiO2 spots wasseen.
To further probe the function of the patterned surface and theactivity of locally bound streptavidin on the patterned SiO2
surface, solutions of biotinylated anti-rIgG and FITC-rIgG weresuccessively deposited on the surfaces. After rinsing with PBS, thesurfaces were again analyzed by fluorescencemicroscopy, and thecombined fluorescence micrograph is presented in Figure 6.Again, the fluorescently labeled antigen (FITC-rIgG) is mainlypresent on the areas covered by streptavidin, since no specificgreen fluorescence is observed in the surrounding areas and thesmall red spots observed after Cy5-labeled streptavidin remainedunchanged after binding of the fluorescent antigens. Moreover,on the SiO2 areas where the antigen recognition process tookplace, the combined fluorescence gave a yellow color. One canobserve areas where the red component was mainly present,
Table 2. Surface Chemical Composition (mole concentration ratio with respect to carbon)Determined from the XPS peak Components as Assigned
Figure 6. Fluorescence micrograph of patterned surface after (A) binding of Cy5-labeled streptavidin and (B) detection of FITC-rIgG onpatterned surfaces previously functionalizedwithCy5-labeled streptavidin and biotinylated anti-rIgG. The disc areas correspond to the SiO2
surfaces, with a 100 μm diameter.
(39) Libertino, S.; Giannazzo, F.; Aiello, V.; Scandurra, A.; Sinatra, F.; Renis,M.; Fichera, M. Langmuir 2008, 24, 1965–1972.(40) Kurth, D. G.; Bein, T. Langmuir 1995, 11, 3061–3067.(41) Wang, W.; Vaughn, M. W. Scanning 2008, 30, 65–77.
meaning that the antigen binding efficiency was locally poor inthese areas; however, the specific shape of these areas (e.g., thestraight line on the top circle) seemed to indicate defects inthe immersion procedure (maybe trapped microbubbles duringthe deposition of the 100 μL on top of the substrates) during theelaboration of the immunoarray rather than a loss of activityof the streptavidin.
Discussion
The current approach for the functionalization of inorganicphotolithographically patterned SiO2/Au surfaces fulfilled theinitially set requirements, since we observed localized binding ofstreptavidin to SiO2 regions which had been chemically modifiedwith biotin moieties. Furthermore, subsequent localized bindingof active biotinylated antibodies on the SiO2 regions of thepatterned surfaces and good resistance to protein adsorption onAu substrates modified with a thiol-OEG SAM were achieved(Figure 6). TheΔf value obtained byQCM-D for the bound layerof streptavidin on nonpatterned SiO2 surfaces modified forspecific binding suggests a full coverage of the molecules onthat surface (Figure 2). Moreover, we observe no increase ofthe dissipation after the binding of the streptavidin molecules,which suggests a strong and compact binding of the proteinadlayers. According to the QCM-D analysis, the thiol-OEGSAM on Au had good repelling properties toward adsorp-tion of proteins, with a low level of nonspecifically boundstreptavidin and antibody molecules. No FITC-rIgG wasdetected on these functionalized Au surfaces, indicating thatthe amount of adsorbed proteins was too low to allow antigendetection via nonspecific adsorption of streptavidin and bioti-nylated anti-rIgG, or that the adsorbed proteins (streptavidinand biotinylated anti-rIgG) were not adsorbed in their activeforms.
One of the questions addressed in this study is whether theassembly of the thiols on the Au surface is affected by the prioradsorption of silanes, used to functionalize the SiO2 surface andform a thiol-OEG monolayer. To answer this question, surfacecharacterization of both molecules on the Au substrates wasnecessary. For the silanes adsorbed on the Au, relevant informa-tion was obtained on the basis of IR and XPS results: highdisorder of the adlayer (high wavenumber for νasymCH2
) and anaverage thickness clearly higher than the one expected for amonolayer (determined by XPS). These observations can beexplained by inspecting the chemical nature of the silanes. Theshort alkyl chain length of the molecule does not promoteinterchain interactions, which are the driving force to form adense packed monolayer.42,43 The presence of three R-O-Sigroups in the molecule, associated with no specific binding of thesilane at the Au surface, explains the polymerization of theAPTES. This polymerization, confirmed also by the XPS O532.1/Si ratio (about 1.8 indicating the major presence of Si-O-Si andSi-OH groups), suggests that silanes are mainly in the form ofaggregates when adsorbed on the Au surface. This explains thehigh value for the thickness of theAPTES layer. The formation ofsilane aggregates at the adsorbed state may lead to low accessi-bility of its chemical NH2 moieties. Indeed, some authors havereported41 that the APTES film morphology can affect theaccessibility of reactive amino groups, thereby altering the degreeof further immobilization. Thiol-OEG shows a very different
behavior when adsorbed alone on Au substrate. The S/Au molarratio determined by XPS suggests a fully assembled layer.Furthermore, the peak position of the νasCH2
vibration shows thattheSAMiswell ordered anddensely packed, due to the alkyl chainlength. The formation of the thiol layer on the substrate pre-functionalized with APTES was also investigated. PM-RAIRSresults indicate a high disorder of the adlayer. The influence of thepresence of remaining silane molecules upon the subsequentassembly of thiols-OEG was made clear. It leads to (i) a decreaseof the amount of assembled thiols (decrease of the S/Au molarratio, Table 1) and (ii) a high disorder in the adlayer as shown bythe IR results. This may be explained by the fact that alkyl chainsare not close enough to each other to promote strong interchaininteractions. This hypothesis is supported by the presence of ashoulder in the IR C-H stretching peaks, suggesting thepresence of CH3 moieties. Hence, the thiol-OEG moleculesdo not form a full monolayer, as APTES is still present at theAu surface in a significant amount. However, this did notprevent the patterned surface from fulfilling its requirement,namely, to passivate the gold substrates and make themresistant to protein nonspecific adsorption. Indeed, the fluor-escence micrographs present very well localized binding of thefluorescent streptavidin molecules. Very low fluorescent sig-nals are seen on the gold surrounding areas. The presence ofsome small red spots, associated with the size of the variousmolecules, made quenching of the streptavidin molecules bythe gold substrates unlikely. Since the thiol-OEG monolayerdisplayed a low amount of adsorbed streptavidin by QCM-D,it is difficult to link this streptavidin adsorption to the remain-ing presence of APTES on the substrate or to the thiol-OEGmolecules themselves. Moreover, the layer was resistant tobiotin adsorption, confirming the lack of accessibility to biotin-NHS molecules. These findings indicate that the formation ofan ordered thiol-OEGmonolayer is not the critical point in theelaboration of a protein-repelling surface and the imperfectlayer obtained (not ordered, presence of APTES) fulfills itsrequirements.
The surface presents good results concerning the immunosen-sing properties of the SiO2 areas, with the local detection of theantigen by the functionalized surfaces and almost no increaseof the fluorescein detection on the Au areas modified with thiol-OEG.
The flow distribution on the patterned substrates shouldnevertheless be improved in our setup to optimize thebinding of the different species, and the use of detergent couldhelp to reduce the low amount of nonspecific adsorption ofproteins.
These substrates represent thus a simple and convenientmethod to prepare patterned surfaces modified with func-tional immobilized proteins via strong interactions. This mayprevent the desorption that seems to occur upon interac-tion with cells and on polymer modified substrates whenattached via electrostatic interaction, as reported for PLL-PEG polymer.44
Conclusion
We have reported a successful elaboration of orthogonalassembly of APTES and thiol-OEG SAMs on patterns of SiO2
and Au, starting with the silanization step. Efficient patterning ofproteins in considerable amounts was achieved. Silanes, likely inthe form of aggregates, were still present on the Au surface after
(42) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M.Chem. Rev. 2005, 105, 1103–1169.(43) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem.
Soc. 1987, 109, 3559–3568.(44) Lussi, J. W.; Falconnet, D.; Hubbell, J. A.; Textor, M.; Csucs, G.
the assembly of thiol-OEG. However, this had little effect on theefficiency of our protocol and the further localized immobiliza-tion of biotin analogues onto the SiO2 parts on the templatesubstrate. The use of streptavidin as anchoringmolecule enables awide range of applications for this protocol, and in future workthese substrates are attractive for studies dealing with substrate-cell interactions, by allowing the localized binding of specificmolecules, such as peptides, carbohydrates, or proteins, involvedin cell attachment and cell differentiation.
Acknowledgment. The SSF (Swedish Foundation for Strate-gic Research) Biomics program, the VINNOVA (The SwedishGovernmental Agency for Innovation Systems) NanobioIT pro-gramproject #22578-2, and the EuropeanUnion Seventh Frame-work Programme under grant agreement no. NMP4-SL-2009-229292 (“Find&Bind”) are gratefully acknowledged for financialsupport.
Appendix
From the experiments performed with XPS, the apparentconcentration ratio [C]/[Au] (and thus the apparentthickness of the adlayer) may be computed using the
following equation:
½C�½Au� ¼
iAu
iC
σC
σAu
λAdC CAd
C 1- exp- t
λAdC cos θ
!24
35
λSuAuCSuAuexp
- t
λAdAu cos θ
!
cos θ is 1 because the photoelectron collection angle θ is equalto zero. iC and iAu are the relative sensitivity factors of the 1s Cand 4f Au levels, respectively, provided by the spectrometermanufacturer. The photoionization cross sections σ are equalto 1 for C 1s and 17.4 for Au 4f.19 The Ad and Su superscriptsdesignate the organic adlayer and the Au substrate, respec-tively. The approximate concentration of Au in the substrate isCAuSu = 86 mmol/cm3, and the amount of carbon in the organic
adlayer was determined according to the thiol and silanemolecules. The electron inelastic mean free paths (IMFPs)were calculated using the Quases programbased on the TPP2Mformula.45 Considering the energy of theMgKR excitation line(1253.6 eV) and the Au 4f binding energy of 83.6 eV, λAu
Su wasfound to be 1.4 nm and λAu
Ad 3.6 nm in both thiol and APTES.λCAd was also found to be almost the same, 3.1 nm, whether it is
computed in thiol or APTES matrix.(45) Tanuma, S.; Powell, C. J.; Penn, D. R. Surf. Interface Anal. 1997, 25, 25–35.
