Biomolecule Patterning on Analytical Devices: A Microfabrication

Publication Date (Web): March 26, 2010 ... native recognition biomolecules can be introduced into the host scaffold downstream from all compatibility ...
0 downloads 2 Views 3MB Size
pubs.acs.org/Langmuir © 2010 American Chemical Society

Biomolecule Patterning on Analytical Devices: A Microfabrication-Compatible Approach ,^

)

Guillaume Suarez,*,†,

Neil Keegan,*,†,^ Julia A. Spoors,† Pedro Ortiz,† Richard J. Jackson,† John Hedley,‡ Xavier Borrise,§ and Calum J. McNeil†

)

† Diagnostic and Therapeutic Technologies, The Medical School, Newcastle University, Newcastle upon Tyne NE2 4HH, U.K., ‡School of Mechanical and Systems Engineering, Newcastle University, Newcastle upon Tyne NE1 7RU, U.K., and §Nanolithography Laboratory, Institut Catal a de Nanotecnologia, E-08193 Bellaterra, Spain. Current address: Environmental Biophysical Chemistry, EPFL/ENAC/IIE/ GR-SLV, Ecole Polytechnique F ed erale de Lausanne, CH-1015 Lausanne, Switzerland. ^These authors contributed equally to this work.

Received December 1, 2009. Revised Manuscript Received March 1, 2010 The present work describes a methodology for patterning biomolecules on silicon-based analytical devices that reconciles 3-D biological functionalization with standard resist lift-off techniques. Unlike classic sol-gel approaches in which the biomolecule of interest is introduced within the sol mixture, a two-stage scenario has been developed. It consists first of patterning micrometer/submicrometer polycondensate scaffold structures, using classic microfabrication tools, that are then loaded with native biomolecules via a second simple incubation step under biologically friendly environmental conditions. The common compatibility issue between the biological and microfabrication worlds has been circumvented because native recognition biomolecules can be introduced into the host scaffold downstream from all compatibility issues. The scaffold can be generated on any silicon substrate via the polycondensation of aminosilane, namely, aminopropyltriethoxy silane (APTES), under conditions that are fully compatible with resist mask lithography. The scaffold porosity and high primary amine content allow proteins and nucleic acid sequences to penetrate the polycondensate and to interact strongly, thus giving rise to micrometer/submicrometer 3-D structures exhibiting high biological activity. The integration of such a biopatterning approach in the microfabrication process of silicon analytical devices has been demonstrated via the successful completion of immunoassays and nucleic acid assays.

1. Introduction In the past decade, the growing effort to produce miniaturized analytical systems has placed silicon microfabrication technology at the heart of the biosensor community. In parallel, micrometerto nanometer-scale biological functionalization of localized regions of a solid substrate has become a widely explored research area. A prime example of a microfabricated sensor is the circular diaphragm resonator (CDR), which promises to yield mass sensors with high sensitivity and an inherent internal reference measurement.1,2 To aid commercial viability, wafer-level patterning of thousands of devices at once would greatly reduce the cost of microfabricated biosensors. Moreover, any wafer-level manufactured device could potentially benefit from the final microfabrication step yielding the requisite site-specific immobilization pattern. The most common patterning avenues available are soft lithography and photolithography. In the former, microcontact printing is prevalent, in which biomolecules are transferred from a polydimethylsiloxane (PDMS) stamp to a solid surface through conformational contact.3 Alternatively, dip-pen nanolithography can be used in which an AFM tip delivers biomolecules onto a *Corresponding authors. (G.S.) E-mail: [email protected]. Tel: þ41 21 693 3727. Fax: þ41 21 693 7080. (N.K.) E-mail: [email protected]. Tel/Fax: þ44 191 222 7991. (1) Ismail, A. K.; Burdess, J. S.; Harris, A. J.; McNeil, C. J.; Hedley, J.; Chang, S.-C.; Suarez, G. J. Micromech. Microeng. 2006, 16, 1487–1493. (2) Ismail, A. K.; Burdess, J. S.; Harris, A. J.; Suarez, G.; Keegan, N.; Spoors, J. A.; Chang, S.-C.; McNeil, C. J.; Hedley, J. J. Micromech. Microeng. 2008, 18, 1–10. (3) Bernard, A.; Renault, J. P.; Michel, B.; Bosshard, H. R.; Delamarche, E. Adv. Mater. 2000, 12, 1067–1070. (4) Hon, K. K. B.; Li, L.; Hutchings, I. M. CIRP Ann. Manuf. Technol. 2008, 57, 601–620.

Langmuir 2010, 26(8), 6071–6077

substrate with nanoscale resolution.4 Both of these techniques allow a monolayer of protein to be patterned directly onto the required surface or to be selectively reacted with a patterned monolayer of precursor molecules such as functionalized thiols on gold5 or functionalized silanes on silicon oxide.6 In the latter technique of photolithography, a polymeric resist stencil is generated at the substrate surface using a photomask (UV lithography),7 nanoimprint lithography (NIL),8 extreme ultraviolet lithography (EUVL),9 X-ray lithography (XRL),10 or electron beam lithography (EBL).11 The UV lithographic photomask approach is compatible with standard clean-room fabrication procedures, routinely generating micrometer-scale features, whereas the other techniques produce more specialized nanoscale features. All of the above patterning strategies have a common theme: the production of a 2-D monolayer of immobilized material. As an alternative, sol-gel chemistry and its 3-D approach allow a considerable increase in the amount of biomaterial immobilized per unit surface area. The abundant literature surrounding biological sol-gel glasses essentially refers to “first-generation” sol-gels that make use of inorganic silane (5) Disley, D. M.; Cullen, D. C.; You, H. X.; Lowe, C. R. Biosens. Bioelectron. 1998, 13, 1213–1225. (6) Yakovleva, J.; Davidsson, R.; Lobanova, A.; Bengtsson, M.; Eremin, S.; Laurell, T.; Emneus, J. Anal. Chem. 2002, 74, 2994–3004. (7) Blawas, A. S.; Reichert, W. M. Biomaterials 1998, 19, 595–609. (8) Guo, L. J. Adv. Mater. 2007, 19, 495–513. (9) Zoller, F. A.; Padeste, C.; Ekinci, Y.; Solak, H.; Engel, A. Microelectron. Eng. 2008, 85, 1370–1374. (10) Chou, M. C.; Pan, C. T.; Wu, T. T.; Wu, C. T. Sens. Actuators, A 2008, 141, 703–711. (11) Martiradonna, L.; Stomeo, T.; De Giorgi, M.; Cingolani, R.; De Vittorio, M. Microelectron. Eng. 2006, 83, 1478–1481.

Published on Web 03/26/2010

DOI: 10.1021/la904527s

6071

Article

Su arez et al.

Figure 1. Schematic procedure for 3-D patterning of biomolecules on analytical devices using conventional lift-off lithographic techniques. The microfabricated analytical device coated with a thin layer of resist (spin coating) is treated in the following manner: (1) lithographic step via UV or e-beam exposure; (2) in situ polycondensation of APTES (2%) in TEA leading to porous polysiloxane with a high primary amine content; (3) chemical release of sacrificial resist mask with acetone and curing (120 C) of 3-D blank host structures; (4) analytical device packaging; (5) a posteriori biofunctionalization of patterned host structures via incubation step with native biomolecule solution; and (6) standard blocking procedure for efficient nucleic acid or immunoassay-based biodetection. The red arrow (1 to 5) indicates steps that are fully compatible with standard microfabrication processes, and the green arrow (5 and 6) points to biologically compatible procedures.

precursors such as tetramethylorthosilicate (TMOS) and tetraethylorthosilicate (TEOS).12 However, those pure inorganic sol-gels present a number of disadvantages in terms of longterm alteration in protein conformation due to matrix compression.13,14 In response to this biocompatibility issue, recent work has focused on developing organic/inorganic hybrid sol-gel glasses, so-called ORMOSILS, in which the introduction of chemical functionality into the sol mixture leads to improved protein stability and better control of the physicochemical properties of the material.15 However, in the vast majority of these sol-gel processes, alcohol is used as a cosolvent in the sol mixture.16 Therefore, using a patterned photoresist mask layer to achieve sol-gel patterning through the lift-off technique is not advisible because alcohol present in the sol matrix would solubilize the photoresist and jeopardize the quality of the final patterned structures. A few alcohol-free sol-gel alternatives have been published on the basis of either a sodium silicate process,17 a glyceryl silicate precursor,18 or an electrochemically induced sol-gel process16 or on the postaddition of enzyme to a precursory sol in which ethanol has been removed by rotary evaporation methods.19 However, a more general issue discounts the use of protein-entrapped sol-gels in this instance; biomolecules encased in the sol-gel may not withstand the postpatterning clean-room processes that chronologically consist of sacrificial-layer removal (12) Gupta, R.; Chaudhury, N. K. Biosens. Bioelectron. 2007, 22, 2387–2399. (13) Gill, I.; Ballesteros, A. Trends Biotechnol. 2000, 18, 282–296. (14) Mansur, H. S.; Orefice, R. L.; Vasconcelos, W. L.; Lobato, Z. P.; Machado, L. J. C. J. Mater. Sci.: Mater. Med. 2005, 16, 333–340. (15) Collinson, M. M. Trends Anal. Chem. 2002, 21, 30–38. (16) Yang, S.; Jia, W.-Z.; Qian, Q.-Y.; Zhou, Y.-G.; Xia, X.-H. Anal. Chem. 2009, 81, 3478–3484. (17) Yu, D.; Volponi, J.; Chhabra, S.; Brinker, C. J.; Mulchandani, A.; Singh, A. K. Biosens. Bioelectron. 2005, 20, 1433–1437. (18) Gill, I.; Ballesteros, A. J. Am. Chem. Soc. 1998, 120, 8587–8598. (19) Ferre, M. L.; del Monte, F.; Levy, D. Chem. Mater. 2002, 14, 3619–3621.

6072 DOI: 10.1021/la904527s

with acetone, wafer dicing, and device packaging (temperatures in excess of 100 C). In contrast, the present work describes a methodology that reconciles 3-D biological functionalization with standard resist lift-off techniques for the simple wafer-level patterning of siliconbased analytical devices. The initial polymer scaffold described can be patterned at the wafer level and activated with biomolecules after the processes that may affect protein stability have been accomplished, as shown in Figure 1. Under the conditions employed, electrostatic interactions are sufficient to hold the biological species (e.g., capture antibodies) within the 3-D polymer scaffold even after extensive washing steps, with little evidence of leaching. Both the porosity of the generated scaffold and the soft nature of electrostatic interactions give rise to structures with high biological activity capable of performing both protein and nucleic acid assays.

2. Experimental Methods 2.1. Materials. 3-Aminopropyltriethoxysilane (APTES), triethylamine (TEA), bovine serum albumin (BSA), Triton, phosphate-buffered saline (PBS), Tris, and ethylenediaminetetraacetic acid (EDTA) were all purchased from Sigma-Aldrich. DNA sequences were all synthesized by MWG (Germany). Mouse anticarcinoembryonic antigen (anti-CEA) monoclonal antibody was a kind gift from Fujirebio Diagnostics AB (Sweden). Antimouse IgG Alexa647 was purchased from Invitrogen. The amine-reactive Alexa 647 kit for anti-CEA labeling was purchased from Pierce, and labeling was achieved per the manufacturer’s protocol. 2.2. Instruments. Fluorescence studies were carried out using a Nikon 80i epi-fluorescence microscope containing the following filter blocks: polymer autofluorescence viewed using an FITC filter block (excitation filter, 478-495 nm; emission filter, 500540 nm). The Alexa fluorophore was viewed using a Cy5 filter block (excitation filter, 590-650 nm; emission filter, 663-738 nm). Langmuir 2010, 26(8), 6071–6077

Su arez et al. Confocal fluorescence microscope imaging was performed with a TCS SSP2 UV X40 Plan Apo Na. 0.85 lens confocal microscope (Leica Microsystems Heidelberg). Polymer autofluorescence was imaged using an excitation laser at 488 nm and emission collection from 500 to 540 nm. The Alexa fluorophore was captured using an excitation laser at 633 nm/emission collection from 640 to 740 nm. An analysis of the relative fluorescence intensity of micrographs was performed using improvision license server 1.2.1 Volocity imaging software. The Volocity software provides the fluorescence intensity output in arbitrary units. Scanning electron microscopy (SEM) samples were examined using a Stereoscan S40 SEM within the Electron Microscopy Research Services Unit, Newcastle University. The samples were mounted onto sticky carbon discs and coated with 15 nm of gold using a Polaron SEM coating unit prior to analysis. 2.3. Lithography. For electron-beam lithography, a 10  10 mm2 Si die was cleaned with acetone and isopropanol (IPA) and then spin coated with a 200 nm layer of PMMA 950K. Exposure was carried out in a LEO 1530 field-emission SEM with a Raith Plus attachment for lithography purposes. Conditions of exposure were dose = 110 μC/cm2 for the 200 nm lines and 0.05 pC/cm2 for the dots; accelerating voltage = 10 kV; and WD= 10 mm. Development was carried out in a standard MIBK/IPA (1:3) solution over a period of 30 s and stopped with IPA for 30 s. For UV lithography, either a 100 mm Æ100æ virgin silicon wafer was used to produce development patterns or a silicon wafer of micromachined CDR analytical devices was used. In the development stage, a low-cost photomask (emulsion on glass, JD Photo, U.K.) was used whereas a standard chrome-on-glass photomask (Compugraphics, U.K.) was used as an integral part of the fabrication process of CDR devices. The wafers were spin coated with a 7 μm SPR220 positive photoresist layer and exposed to UV through the photomask containing the pattern of interest. After baking, the wafer was developed using MF-26A to remove the exposed resist. All patterning procedures were undertaken in a class 1000 UV-free clean-room environment using an EVG101 spin coater, EVG103 developer, EVG620 mask aligner/UV lamp.

2.4. Synthesis and Characterization of a Three-Dimensional Polymer. It is important to state here that the polycondensation reaction was carried out within a standard clean-room environment that is under strict humidity control (45 ( 5% maximum). Under such conditions, the silane polycondensation reaction was not affected by the variability of outdoor atmospheric conditions. A solution of 3-aminopropyltriethoxysilane (APTES), 2% in TEA, was left to polycondense for between 1 and 5 h. The resulting polycondensate material was cured overnight at 120 C to improve inorganic cross-linking and to allow the evaporation of any residual TEA. The chemical structure of the 5 h polysiloxane network was elucidated using 13C and 29Si solidstate NMR.

2.5. Generation of a Three-Dimensional Polymer Pattern. The patterned resist was exposed to an organic solution of 2% 3-aminopropyltriethoxysilane (APTES) in triethylamine (TEA) for 1-5 h, as previously described. Once in situ polycondensation was completed, the resist stencil was removed by ultrasonication in acetone for 1 min at 130 kHz and 50% power in an Elma Transsonic TI-H-10 ultrasonic bath and then cured overnight at 120 C. For initial development work, flexibility was the key, so wafer dicing occurred at the stage of resist patterning and the polymer was deposited at the die level. For massproduced silicon devices, the deposition was carried out at wafer level.

2.6. Biological Activation of a Three-Dimensional Polymer Pattern. 2.6.1. Immunoassay Activation. Depending on the nature of the experiment, 3-D host structures were activated with biomolecules by incubation with 20 μg/mL of native anti-CEA mouse monoclonal antibody or anti-CEA labeled with Alexa 647 (in carbonate buffer at pH 9.6) for 1 h at room temperature. That was followed by three 5 min washes using Langmuir 2010, 26(8), 6071–6077

Article 0.05% Triton/PBS and a final rinse in buffer alone. In the control experiments, no capture antibody was loaded into the polymer. As in standard immunoassay experiments, nonspecific binding was minimized by using standard blocking procedures such as 5% BSA/PBS at pH 7.4 for 1 h at room temperature, followed by three 5 min washes using 0.05% Triton/PBS at pH 7.4. 2.6.2. ssDNA Activation. When carrying out DNA-based experiments, 1 μM probe DNA (50 -TACATAGAGATGGGAATCCAT-30 ) in 1 M phosphate buffer at pH 7.4 was loaded into the 3-D matrix for 1 h at room temperature. All of the surfaces were then washed with 1 M phosphate buffer at pH 7.4 in 0.05% Triton for 5 min, and the washing process was repeated three times. Surfaces not activated with probe DNA were used in negative control experiments. All surfaces were blocked with 10 μg/mL herring sperm DNA for 1 h at room temperature, and the washing steps were repeated.

2.7. Biodetection Assays. 2.7.1. Immunoassay Experiment. A host polymer structure activated with anti-CEA mouse monoclonal antibody and blocked as described above was compared with a negative control, which had simply been blocked. The immunoreactions were initiated by incubating the polymer samples in a solution of mouse IgG target antibody labeled with Alexa 647 (1:10 000/PBS at pH 7.4) for 1 h at room temperature. The final step in the process was three washing steps of 5 min with 0.05% Triton/PBS at pH 7.4 and a final rinse in buffer alone. After the final washing step, the samples were analyzed by fluorescence microscopy. 2.7.2. ssDNA Experiment. The probe (50 -TACATAGAGATGGGAATCCAT-30 ) and target (30 -AGCAGATGGATTCCCATCTCT-AlexaFluor647-50 ) have a 16 base pair complementary region with a Tm of 50 C. Polymer structures activated with probe DNA and then blocked were compared with a negative control, which had simply been blocked. A solution of 1 μM complementary target DNA-Alexa647 in hybridization buffer (10 mM Tris, 1 M NaCl, 1 mM EDTA) was then incubated over the surfaces for 1 h at room temperature. After a final washing step, the target analyte recognition was demonstrated by fluorescence microscopy.

3. Results and Discussion 3.1. Patterning of Polymer Scaffold Structures via Standard Photolithography. The patterning of polymer scaffolds on specific regions of silicon surfaces during microfabrication processing has been tested. Both development silicon wafers with a simple photoresist pattern and fully processed silicon analytical device wafers, which have in excess of a thousand devices per wafer, have been employed in this work. The silicon analytical devices are CDR, and as previously reported, the site-specific biofunctionalization of selected regions of the CDR diaphragm (60-150 μm diameter) leads to a disruption of the axisymmetric mass distribution, hence the resonance frequencies of the different modes are split by an amount proportional to the added mass.1,2 A Maltese cross will cause this type of disruption and has been used for the majority of patterns in this article because it will be useful in relation to CDR in future publications. For such devices, the fragility of the diaphragm and the requirement for a methodology that is robust and compatible with routine microfabrication technology make conventional soft and mask lithographic approaches difficult to reconcile. As described in Figure 1, the originality of the present work lies in the generation of 3-D polysiloxane networks with a high primary amine content under experimental conditions that are fully compatible with resistmask lithography techniques. In situ polycondensation of a hybrid inorganic-organic silicate, namely, aminopropyltriethoxy silane (APTES), takes place on the exposed regions of a masked metal oxide substrate in pure organic base triethylamine (TEA). It is important to recognize that the use of APTES to functionalize DOI: 10.1021/la904527s

6073

Article

Figure 2. Blank cross scaffold structures obtained at different polycondensation times as visualized by fluorescence microscopy and SEM. (a) Fluorescence micrographs showing polymer scaffold autofluorescence under the following conditions: a 20 objective lens, 1 s of exposure time; filter blocks of 478-495 nm (excitation) and 500-540 nm (emission). (b) Low- and (c) high-magnification SEM images 881 and 6370, respectively, using an acceleration voltage of 8 kV and a working distance of 17 mm. The complementary techniques both confirm the control of patterned scaffold structure thickness by varying the polycondensation reaction time. Scale bars on SEM images indicate (b) 50 and (c) 5 μm, respectively.

silicon surfaces with primary amine groups has already been abundantly recorded in the literature. From an examination of the existing literature, the standard protocols for APTES silanization onto silicon oxide consist of reacting between 0.4 and 5% APTES diluted in toluene, acetone, ethanol, or water or a mixture of the components, with a reaction time from 30 s to 24 h at room temperature.6,20-29 In all of those cases, the APTES thickness does not exceed 9 nm, in accordance with the declared intention to covalently attach only a monolayer or a thin film of APTES to the substrate. In contrast with this situation, the use of a highly alkaline organic solvent with no nucleophilic character, such as TEA, enabled the establishment of siloxane bonds between the initial covalently attached APTES monolayer at the substrate surface (horizontal polycondensation) and then with the molecules from the solution (vertical polycondensation) in order to obtain a siloxane polymer with a thickness from 100 nm to a few (20) Vandenberg, E. T.; Bertilsson, L.; Liedberg, B.; Uvdal, K.; Erlandsson, R.; Elwing, H.; Lundstrom, I. J. Colloid. Interface Sci. 1991, 147, 103–118. (21) Heiney, P. A.; Gruneberg, K.; Fang, J. Langmuir 2000, 16, 2651–2657. (22) Wieringa, R. H. Ph.D. Thesis, University of Groningen, The Netherlands, December 2000; Chapter 2. (23) Zhang, G. J.; Tanii, T.; Zako, T.; Funatsu, T.; Ohdomari, I. Sens. Actuators, B 2004, 97, 243–248. (24) Zheng, J.; Zhu, Z.; Chen, H.; Liu, Z. Langmuir 2000, 16, 4409–4412. (25) Britland, S.; Perez-Arnaud, E.; Clark, P.; McGinn, B.; Connolly, P.; Moores, G. Biotechnol. Prog. 1992, 8, 155–160. (26) Wang, Z. H.; Jin, G. Colloids Surf., B 2004, 34, 173–177. (27) Sorribas, H.; Padeste, C.; Tiefenauer, L. Biomaterials 2002, 23, 893–900. (28) Weiping, Q.; Bin, X.; Lei, W.; Chunxiao, W.; Danfeng, Y.; Fang, Y.; Chunwei, Y.; Yu, W. J. Colloid Interface Sci. 1999, 214, 16–19. (29) Weiping, Q.; Bin, X.; Danfeng, Y.; Yihua, L.; Chunxiao, W.; Fang, Y.; Zhuhong, L.; Yu, W. Mat. Sci. Eng. C 1999, 8-9, 475–480.

6074 DOI: 10.1021/la904527s

Su arez et al.

micrometers, depending on the incubation time employed. Figure 2 shows the details of this process via SEM imaging at different polycondensation reaction times. The pattern created is an exact representation of what is required to be patterned onto a CDR diaphragm to create a sensor. It is clear that at the end of the process only the regions of the wafer that were exposed to the reaction solution through the resist mask are patterned with the polycondensate structures. All subsequent depositions reported in this article employed a 5 h deposition time. 3.2. Proposed Mechanism for APTES Polycondensation in TEA. As expected, 13C NMR analysis indicated that propyl chains remained unaffected during the polymerization process, displaying characteristic signals for C1, C2, and C3 (11.8, 27.3, and 45.16 ppm, respectively).30 Moreover, the absence of a signal above 50 ppm suggests that there were no residual ethoxy groups left and that the hydrolysis of APTES was total. The presence of three signals on the 29Si NMR spectrum clearly showed that the polymer is highly cross-linked with a high content of tridentate (RSi-[OSi-])3, 68 ppm, 57%) and bidentate (RSi-[OSi-]2OH, 60 ppm, 38%) silicon atoms whereas a low proportion of monodentate bonding is observed (RSi-[OSi-]OH2, 51 ppm, 5%).31 Interestingly, no signal was found for either Si-N bonds (2033 ppm) or free residual APTES molecules (-45.5 ppm).32 Because no Si-N bond was seen in the 29Si NMR, it seems unlikely that the primary amine group reacted with the silicon atom through nucleophilic substitution. Furthermore, the peaks obtained at 354-356 ppm from improved 15N NMR are likely to originate from primary amines. (See Supporting Information Figures S1, S2, and S3 for NMR data.) The combined information indicates that the primary amine group has taken no part in the polymerization process and is still bonded to C3 of the propyl chain. Hence, the primary amine remains available for the electrostatic attraction of biomolecules or even future covalent modification if deemed desirable. From the solid-state NMR analysis, a reaction mechanism can be proposed for the polycondensation of APTES under highly alkaline and nonaqueous (but not anhydrous) conditions. It seems likely that two concomitant reactions involving the silicon atom take place simultaneously, namely, partial, very slow hydrolysis and nucleophilic substitution of ethoxy groups. As previously mentioned, traces of water (hydroxyl groups in TEA) attack the Si atoms of APTES, liberating ethoxy groups in solution. Once again, under such alkaline conditions, the silanolates formed through hydrolysis are very nucleophilic and tend to react on adjacent Si atoms (electrophilic), establishing siloxane bonds. In the most probable scenario, in an early stage of the reaction, the polycondensation process takes place simultaneously at the substrate surface and in bulk solution. The initial polymer material at the substrate surface forms a covalent bond with the silicon surface, per conventional silane monolayers. During further reaction, dots generated from bulk polymerization start to react covalently with the surface polymer, forming islands; meanwhile, their size increases with time, which is so-called Ostwald ripening.33 Despite this mechanism being hard to control at a molecular level, two essential reasons make this in situ polycondensation process fully compatible with highly defined submicrometer-scale features. In the first place, polymer is (30) Chiang, C.; Lui, N.; Koenig, J. L. J. Colloid Interface Sci. 1982, 86, 26–34. (31) Heitz, C.; Laurent, G.; Briard, R.; Barthel, E. J. Colloid Interface Sci. 2006, 298, 192–201. (32) Ek, S.; Liskola, E. I.; Niinist€o, L.; Vaittinen, J.; Pakkanen, T. T.; Root, A. J. Phys. Chem. B 2004, 108, 11454–11463. (33) Brinkler, C. G.; Scherer, G. W. Sol-Gel Science: The Physics and Chemistry of Sol-Gel Processing; Academic Press: New York, 1990

Langmuir 2010, 26(8), 6071–6077

Su arez et al.

Article

Figure 3. Confocal microscopy images of a biomolecule-loaded cross structure (150 μm diameter) using different filter blocks to visualize both (a, b) polymer autofluorescence and (c, d) Alexa647anti-CEA. Panels a and c are 2D snapshots of 20 confocal layers taken from the top to the bottom of the sample as viewed from above. Panels b and d use the same information and display it as a slice through the sample to depict the layer depth and the distribution throughout the layer accurately. The slice was taken through the center of one of the arms of the cross pattern. The depths of the polymer and anti-CEA-Alexa647 layers were ∼4.75 and ∼4.2 μm, respectively, and are representative of 5 h scaffold samples.

covalently deposited at the surface and the level of coverage is controlled by the length of incubation. Second, the size of the polymeric dots that react with the surface polymer to create the polymeric interface is controlled by the size and shape of the exposed regions on the sacrificial mask. 3.3. Biological Activation of Patterned Structures. The loading of the patterned host structures with capture biomolecules such as antibodies and ssDNA has been achieved via a simple incubation step. Specifically, the patterned die were incubated for 1 h at room temperature with a solution of anticarcinoembryonic antigen (anti-CEA) previously labeled with Alexa647 dye in PBS at pH 7.4 or a capture DNA sequence labeled with Alexa647 in 1 M phosphate buffer at pH 7.4. Fluorescence microscopy, using specific filter blocks for the polymer autofluorescence and Alexa647, was used to visualize the polymer and immobilized biomolecules. Micrographs a and c in Figure 3 indicate that the biomolecules did indeed take on the shape of the polymer as expected; however, there was no indication regarding biomolecule location in the xz axis. To prove the packing improvements over the current state-of-the-art monolayer APTES approach, confocal microscopy was used to visualize the immobilized biomolecules within the 3-D polymer structure. The vertical cross-sectional analysis (xz axis) of the sample was performed using a confocal laser scanning microscope, and 20 parallel slices from the top to the bottom of the sample were collected using 2 different filter sets to image both the autofluorescence of the polycondensate structures and the loaded biomolecules. The images in Figure 3 show that the capture biomolecules assume virtually identical 2-D and 3-D patterns to those of the host scaffold at the silicon surface. In addition, a quantitative measure of loading reproducibility between polymer scaffolds was performed and analyzed using Volocity software. The fluorescence intensity of four independent polymer scaffolds loaded with anti-CEA-labeled Alexa 647 (20 μg/mL) was assessed, resulting in a mean fluorescence intensity (arbitrary units) of 77.28 ( 3.60% (n = 4). These experiments demonstrate the Langmuir 2010, 26(8), 6071–6077

Figure 4. Fluorescence visualization of biological activity in host polymer structures using (a-d) immunoassay and (e-h) DNA biodetection formats. The left-hand panel show the autofluorescence of the four blank host structures investigated, and the right-hand panel shows the presence of the Alexa647-labeled biomolecule. In the immunoassay format, panel b shows the positive probe antibody-target reaction and panel d shows the negative control (no probe antibody immobilization prior to target detection). In the DNA format, panel f shows the positive probe-target DNA recognition and panel h shows the negative control (no probe DNA immobilization prior to target detection).

ability to create a 3-D biomolecular structure that can easily be patterned onto solid surfaces. The 3-D structure of the patterned surface clearly facilitates a higher biocomponent loading capacity over that of conventional monolayer-immobilization procedures. The polymer-based recognition component shows continuous fluorescence over 4 μm, with this new volume element of the polymer-immobilization procedure demonstrating the increased loading capacity. A well-formed comparative protein monolayer would cover the same surface area but would have only a single-layer height on the order of 10 nm. To put it another way, hundreds of monolayers could fit into the 4 μm height. Finally, a very limited stability study showed that a polymer patterned surface could be stored at room temperature for 50 days without any deterioration in its ability to act as a patterning scaffold for bioimmobilization. 3.4. Detection of Biological Analytes. A series of experiments were carried out to establish the compatibility of the 3-D biomolecule patterning approach with standard immunoassay and nucleic acid detection protocols. In both cases, the efficiency of the detection system is notably dependent on high activity, namely, the ability to maintain the maximum specificity of the biorecognition element while minimizing the nonspecific adsorption DOI: 10.1021/la904527s

6075

Article

Figure 5. Shakespeare’s verse “written” with 200-nm-width lines on a silicon substrate. The different stages of the patterning procedure were visualized using (a, b) bright light and (c, d) fluorescence microscopy (magnification, 50; exposure time, 500 ms). (a) The sample obtained by e-beam lithography is exposed to APTES solution (2% in TEA) in which in situ polycondensation takes place. Once the PMMA stencil is removed, the patterned host structures (b, c) are revealed as an accurate replica of the resist stencil, which can then be loaded with fluorescently labeled DNA (d).

of biomolecules. In the case of the immunoassay experiment, the host structures were loaded with mouse anti-CEA probe molecules, followed by washing steps and blocking with BSA. At this stage, the immunoreactions were initiated by incubating the biofunctionalized dies with target analyte labeled with the Alexa647 fluorophore (antimouse IgG Alexa647). In contrast, the control experiments had no capture antibody loaded into the host structure and the samples were blocked only with BSA. After a final washing step, the samples were analyzed by fluorescence microscopy as shown in Figure 4a-d. Very clearly, only the specific response shows an intense fluorescence that embraces the shape of the polymer whereas the negative control shows very limited background fluorescence. In other words, the capture antibody remains specific once loaded into the 3-D structure and nonspecific adsorption is efficiently limited by blocking the active volume using conventional techniques. Similar experiments were carried out with nucleic acid sequences as biorecognition elements. Two 21-base sequences were manufactured: the probe and target, which share a 16 base pair complementary region. Standard 150 μm host structures were loaded with 1 μM probe sequence and then washed and blocked with herring sperm DNA. Control host structures were loaded only with herring sperm DNA blocking solution. Finally, all surfaces were exposed to 1 μM complementary target DNA labeled with Alexa647, in hybridization buffer, and washed again. The specific target analyte recognition was demonstrated by fluorescence microscopy, as shown in Figure 4e-h. Further polymer patterns were used to assess sample matrix effects, in conjunction with immunoassay and nucleic acid detection assay. On this occasion, fetal calf serum (FCS) instead of PBS was spiked with the target molecules. The extra material present in the serum sample had no detrimental effect on probe-target recognition, and the fluorescent images followed the trend seen visualized in Figure 4. Hence, the overall strategy for probe immobilization remains site-specific, controllable, and appropriate when FCS is used as the sample matrix. 6076 DOI: 10.1021/la904527s

Su arez et al.

Figure 6. (a) SEM images in detail from Shakespeare’s verse were (b) quotation marks and (c) array of dots showing submicrometerscale blank host structures patterned onto silicon. Acceleration voltage, 20 kV; WD, 6 mm.

3.5. Compatibility with Submicrometer-Scale Technology. The possibility of applying this biopatterning approach to submicrometer-scale-resolution lithographic techniques was also investigated. A proof of principle is shown in Figure 5, in which the verse by William Shakespeare “If it were done when ‘tis done, then ‘twere well it were done quickly” (Macbeth: act 1, scene 7) was “written” in polymer. Initially, 200-nm-width lines were created on a PMMA-resist-coated silicon oxide substrate using e-beam lithography, as depicted by bright-field microscopy in Figure 5a. After the in situ polycondensation of APTES, resist removal, and curing, the high fidelity of the submicrometer polycondensate structures was visible via bright-field and epifluorescence microscopy (polymer autofluorescence), as shown in Figure 5b,c, respectively. Finally, the 3-D pattern was loaded with ssDNA conjugated to the Alexa647 dye. The characteristic red fluorescence demonstrates the high resolution of the highlighted sentence with no visible background, as depicted in Figure 5d. The juxtaposition of the four pictures emphasizes the high degree of replication obtained with this biomolecule-patterning procedure, from the PMMA stencil to the immobilization of DNA on very confined regions of the sample. In addition, the high fidelity of the host-structure shape was revealed by SEM (Figure 6), making it conceivable that a few tens of nanometers is probably the limiting dimension for this patterning approach, considering the intrinsic dimensions of most biomolecules. 3.6. Wafer-Level Deposition on Silicon Analytical Devices. Polymer deposition was carried out at the wafer level, resulting in an entire batch of silicon CDR devices with a characteristic polymer host structure patterned at the sensor surface. See Figure 7 for optical images of this surface. It is our intention to use this patterning procedure in parallel with several other strategies to fully characterize the CDR sensors in an analytical setting. However, the relevance of this procedure as a tool for patterning biomolecules in three dimensions using a protocol that is compatible with standard cleanroom techniques is important to the wider surface-functionalization and sensor community in its own right. For example, biomolecule patterning on solid substrates leads to a large number of analytical applications, from screening tools for proteomics and pharmacology to biosensors for diagnostics and environmental monitoring. Langmuir 2010, 26(8), 6071–6077

Su arez et al.

Article

Figure 7. Bright-field microscope images of polymer scaffold patterning on the sensing regions (150-μm-diameter cross features) of the CDR MEMS devices at the wafer level: (1) sacrificial resist layer generated by UV lithography, where only the CDR sensing regions remain uncovered; (2) substrate exposure to APTES/TEA solution for 2 h at room temperature, where the resulting polymer layer covers the entire surface of the substrate; and (3) after lift-off, the host polymer pattern is revealed, which can be biologically activated at any time after dicing and packaging the dies.

4. Conclusions The 3-D biomolecule patterning procedure that is developed introduces a valuable solution to the limitations encountered in the existing state of the art for the biofunctionalization of microfabricated analytical devices. In particular, the technique reported can be used on mechanically sensitive surfaces/devices, which may not be advisible for soft lithography and direct write techniques, and this patterning technology can be configured in a high-throughput format because of compatibility with lift-off lithography. In contrast with monolayer chemistry in which the

Langmuir 2010, 26(8), 6071–6077

chemical discrimination of silicon-selective regions concerns only the proximity of the surface, this new procedure generates a highly defined 3-D network surrounded by a planar background. The ability of the host matrix to be activated with proteins or DNA throughout the structure gives rise to sitespecific 3-D biomolecule patterning, thereby increasing the bioactive loading compared with that of 2-D interfaces. The potential of this procedure is essentially due to its simplicity of use, robustness, and versatility in terms of the molecules that can be patterned into a universal scaffold, which in turn can be deposited onto any substrate presenting a hydroxyl functional group. Perhaps the greatest advantage of this new lithographic polymer-based approach is that biomolecules can be added after in situ polymerization has been completed as opposed to incorporation during the polymerization process. Therefore, the denaturation of biomolecules during clean-room processing, device packaging, or as a result of extended storage time need not be a limiting constraint. This adds value over currently available biomolecule entrapment techniques within sol-gels in which the biomolecule is introduced prior to polymerization. Finally, the nature of biomolecule immobilization via physical adsorption into a polymer matrix rich in primary amines, combined with the high definition of e-beam/UV lithography, has led to the creation of very active 3-D submicrometer structures that are able to recognize specific antigen targets or nucleic acid sequences. The utility of the polymer as a noncovalent patterning scaffold has been proven, and investigations of the potential of the scaffold for covalent modification are envisaged in future work. Investigations of the full bioanalytical utility of this patterning technique for clinical measurements using MEMS sensors are currently in progress. Acknowledgment. This work was funded by the European Commission within the SmartHEALTH Integrated Project (FP62004-IST-NMP-2-016817). We are grateful to Dr. Trevor Booth for his expertise in confocal laser scanning microscopy and to the Electron Microscopy Research Services Unit at Newcastle University. Supporting Information Available: 13C, 29Si, and 15N solidstate NMR spectra. This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la904527s

6077