- Email: [email protected]
Protein profiled features patterned via confocal microscopy
Biosensors & Bioelectronics 15 (2000) 85 – 91 www.elsevier.com/locate/bios
Protein profiled features patterned via confocal microscopy Dan V. Nicolau a,b,*, Robert Cross a b
a Molecular Motors Group, Marie Curie Research Institute, Oxted, Surrey, RH8 0TL, UK Industrial Research Institute Swinburne, Swinburne Uni6ersity of Technology, PO Box 218, Hawthorn, Vic. 3122, Australia
Received 24 February 1999; accepted 9 December 1999
Abstract Protein patterns were printed using conventional microlithographic materials in a bilayer arrangement and unconventional exposure tools. The bilayer resist stack consisted of a lower poly(tert-butyl methacrylate) layer and an upper diazonaphtoquinone/ novolak layer. The protein features were printed in either ‘contact printing’, or ‘step and repeat’ mode. The latter printing mode can be managed in a flow-cell consisting of a standard microscope slide and cover slip, spaced apart by about 20 mm, as follows: (i) the exposure step is carried out in the cell using focused 488 nm beam of a confocal laser scanning microscope; (ii) the development step is performed by flowing the photoresist developer through the cell; (iii) the selective deposition of the protein (FITC-labelled avidin) is achieved via the flow of the protein solution through the cell until a desired contrast has been reached; (iv) the control of the process is assured using on-line monitoring of the photo-activated red fluorescence of the developing resist layer, and of the green fluorescence of the FITC-protein patterns, respectively. The protein printing technique uses equipment routinely available in biological laboratory. The ‘step and repeat’ patterning yields high and controllable resolution. The process can be applied in the fabrication of medical microanalysis devices. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Protein patterning; Bilayer resist; Confocal microscopy; Biosensors; Microlithography
1. Introduction The application of semiconductor technology in experimental biomedical science gained momentum in recent years. Microlithographic-like procedures have been used for light-assisted combinatorial chemistry pioneered almost a decade ago (Fodor et al., 1991) and employed to build high-density peptide libraries (Gallup et al., 1994) and oligonucleotide arrays (McGall et al., 1996; Gray et al., 1997). The use of microlithographic materials as functionalized scaffolds for building protein and peptide structures was also reported (Nicolau et al., 1998, 1999). In addition to the use of microlithography as a technological concept for printing high density features, and the use of microlithographic materials as scaffolds for bioactive architectures, the use of microlithographic technologies
would be a decisive advance towards the full integration of the semiconductor microlithography knowledge base into the biomedical research. The seamless integration of the semiconductor microlithography in biomedical research is precluded largely by the high cost of the high resolution exposure tools. The natural ‘technological’ solution would be to use either low cost exposure tools (e.g. contact printers), or better harness the optical tools that are already present in many biolaboratories (e.g. confocal microscopes). The present work tries to address both these avenues, using bilayer microlithographic technology, and optical tools present in most biolaboratories (i.e. deep-UV illuminators and confocal microscopes, respectively). 2. Materials and methods
2.1. Materials and layer deposition * Corresponding author. Tel.: +61-39-2145038; fax: + 61-392145050. E-mail address: [email protected] (D.V. Nicolau)
Because in biomedical research working with opaque substrates is uncommon, and therefore the use of reflec-
0956-5663/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S 0 9 5 6 - 5 6 6 3 ( 9 9 ) 0 0 0 7 2 - X
86
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
tion microscopes is unusual, we used microscopic cover slips as the base substrate. The glass of the cover slips is mildly hydrophobic assuring protein coverage. Poly(tert-butyl methacrylate), P(tBuMA), purchased from Aldrich was deposited as the bottom layer. The polymer is highly hydrophobic and has a molecular weight of :170 000. The :1 mm thick film was spin coated depositing a droplet of a 5% P(tBuMA) solution in xylene on cover slips; followed by spinning the substrate at 3000 rpm for 5 min using an in-house built spinner (an Eppendorf benchtop microcentrifuge); then soft-baked in a convection oven for 3 h at 90°C. The long baking time assured the avoidance of the interfacial mixing between PtBuMA and the top layer photoresist. The top layer consisted of a common g-line AZ type photoresist, i.e. AZ 1900, in the low viscosity formulation, purchased from Clariant. The 0.6 mm top layer was spin coated at 4000 rpm for 5 min and soft-baked for 1 h at 80°C. AZ type photoresists contain a photoactive compound, a derivative of diazonaphtoquinone (DNQ), a base resin (novolak) and a number of other minor chemical species (sensitisers). The DNQ:novolak weight ratio is :1:4.
2.2. Patterning procedures The patterning procedure consisted in (i) the patternexposure of the bilayer film; (ii) controlled development; (iii) surface treatment for protein attachment selectivity; and (iv) protein deposition. Several techniques for protein patterning are presented in Fig. 1.
2.2.1. Exposure E-beam patterned (2 mm lines connecting 5×5 mm squares) chromium masks (Hoya) were used to print features in the photoresist top layer in the ‘contact printing’ mode. The ‘contact printing’ exposure tool was an in-house built contact printer, adapted via the modification of a deep-UV illuminator. The deep-UV illuminator is a planar irradiation equipment commonly used in biolaboratories for exciting fluorofores bound to DNA samples. An optical filter limits the wavelength range of the exposure approximately between 290 and 450 nm. The chromium mask was placed on the deepUV illuminator, then the cover slip was placed with the photoresist facing the mask, and finally the whole ensemble was thoroughly contacted. This set-up, similar to a broadband UV exposure contact printing, was
Fig. 1. Protein patterning procedures using bilayer lithography.
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
used for multi-step, ex situ protein patterning. The optimum exposure energy was 80 mJ/cm2. For high resolution, in situ lithography, a ‘projection printing’ arrangement was devised comprising a flow cell and a confocal laser microscope. The flow-cell was built (i) attaching the cover slip with the photoresist on top on a microscope glass slide; (ii) attaching parallel lines of rubber spacers on the microscope glass slide leaving a longitudinal inlet and an outlet for fluid flow; (iii) attaching a transparent cover slip on the spacers (the thickness of the spacers assures the presence of a gap — about 20 mm — between the photoresist surface and the top cover slip). The scanning confocal laser microscope was a BioRad MRC600 having a KrAr mixed gas laser. The 488 nm line of the 3 mW laser was used for exposure of the photoresist. Neutral density filtration was at 1% for the exposure. The exposed pattern could be visualised using the 568 nm (green) line of the laser to excite the fluorescence of the photoresist. The patterned resist was visible as bright red fluorescence (unexposed resist) on a darker red background (exposed resist), as a result of photoactivation of the DNQ fluorescence. Higher power exposure bleached the fluorescence of the resist. The layout of the patterns was designed using Adobe Illustrator and rasterised by importing the file into Photoshop (both from Adobe Systems). The rasterised file was then translated into a sequence of PARK commands. The PARK (x,y) command is intrinsic to the BioRad COMOS software and has the effect of driving the beam steering mirrors to a particular X,Y position in the field. A sequence of PARK commands, as used here, exposes each pixel in the pattern for about 20 ms. The microscope used for the control of the exposure was a Nikon Optiphot II, and used a 60x differential interference contrast (DIC) planapo oil immersion objective with a numerical aperture (NA) 1.4.
2.2.2. De6elopment The developer used was a common AZ 351 (purchased from Clariant). The contact printing mode required only minor developing control. The development can be also performed using an aqueous fresh solution of 0.05 n NaOH. 2 mm feature were resolved using an immersion development for 50 s. After development, the surface of the resist was thoroughly washed with deionised water. The confocal microscope generates high exposure energies compared to semiconductor lithography. Despite the fact that the confocal microscope was tuned to 488 nm (instead of the 465 nm of the g-line), and despite the attenuation used (only :1% of the exposure was delivered to the surface of the photoresist), a very diluted developer (1:10) had to be used for in situ protein patterning for a controlled development. Moreover, short development times had to be used (in the range of 20 – 30 s). The diluted developer was flown through the
87
gap of the flow-cell using a syringe. After development, the surface of the resist was washed with deionised water in the same flow-cell. The developed patterns were observed using DIC optics in a Nikon Eclipse 800 upright microscope, again using NA 1.4 oil-immersion optics, but with additional magnification to create an : 25 mm square field at the camera faceplate. The camera was a Hamamatsu C4880-01 chilled CCD, with a 512× 512 pixel chip. Images from the camera were grabbed using a Snapper card and Improvision software (from Openlab).
2.2.3. Surface treatment In all protein printing modes the hydrophobicity of the patterned features was manipulated to assure the highest protein contrast. The chemical processes responsible of hydrophobicity manipulation are presented in Fig. 2. Contact printing. In the ‘protein-on-top’ mode, the surface treatment aimed at making the photoresist lines as hydrophobic as possible and making the bottom of the channels (i.e. the ‘spaces’) as hydrophilic as possible. This was achieved through (i) the flood exposure of the patterned and developed features; (ii) the immersion of the patterned substrate in a solution of HCl:H2O2 at 90°C for 1/2 h, following a procedure described as an alternative Image Reversal technique (Moritz and Paal, 1986). In this mode, the bottom P(tBuMA) layer is actually redundant, and the scheme equally works with a single layer of photoresist (the treatment with the oxidizing solution renders the glass hydrophilic). In the ‘protein-on-bottom’ mode, the surface treatment aimed at making the photoresist lines as hydrophilic as possible and making the bottom of the channels (i.e. the ‘spaces’) as hydrophobic as possible. This was achieved through the simple flood exposure with a high energy and at higher wavelength (no UV light). Projection printing. The features produced in-situ via ‘projection’ printing were appropriate only for ‘proteinon-bottom’ mode. Therefore the surface treatment limited to a flood, green light, high energy exposure with the confocal microscope. 2.2.4. Protein deposition The protein patterning consisted in the hydrophobicity-controlled attachment of a fluorescent protein (i.e. FITC avidin) either on lines (‘protein-on-top’ mode) or on the bottom of the patterned and developed channels (‘protein-on-bottom’ mode). The patterned substrates were (i) flooded with a solution of FITC-avidin at a concentration of 25 mg/ml; (ii) incubation for several minutes in the flow cell at ambient temperature; and (iii) washing with deionised water followed by soft drying. The contrast of the protein patterning was controlled on line, i.e. via the monitoring of the fluorescence contrast
88
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
Fig. 2. Physico-chemistry of protein patterning using the bilayer process.
Fig. 3. The 2 mm protein lines printed via positive tone, ‘protein-on-top’ lithography. Red fluorescent, hydrophobic resist features (left); resist lines (2X magnification) in contrast enhancement mode (middle); and green fluorescent FITC-protein lines (right).
using the confocal microscope tuned to detect green fluorescence. The same setup used for monitoring the development was used for fluorescence observations following flood exposure of the pattern to fluorescently-labelled streptavidin. In this case epifluorescence illumination from a mercury lamp was used to excite the labelled streptavidin using the standard fluorescein epifluorescence dichroic filter cube of the microscope.
3. Results
3.1. ‘Protein-on-top’ microlithography The ‘protein-on-top’ contact mode protein lithography produced features with good definition of the fluorescent image (Fig. 3, left). However, the aggressive treatment of the resist features resulted in a poor definition of the resist lines (Fig. 3 middle and Fig. 3 right,
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
fluorescence and DIC image of the profiled resist patterns, respectively). Subsequently, the protein attached with great selectivity of this poorly defined template.
3.2. ‘Protein-on-bottom’ microlithography The ‘protein-on-bottom’, broadband, contact printing mode protein lithography produced features with very good contrast of the protein image (Fig. 4 left). The quantification of the density of the protein is rather difficult using fluorescence-based relative methods of measuring the surface concentration of the adsorbent. However, the fluorescent image of the protein (in Fig. 4) proves that all avidin is present on the features only. No trace of protein could be detected outside the patterns (bottom of the channels). The milder lithographic procedure allowed a good definition of the patterned resist lines (Fig. 4, middle; and Fig. 4, right; fluorescence and DIC image of the profiled resist patterns, respectively), which were subsequently reproduced with good accuracy by the protein attachment. The quality of the patters indeed varies from one method to another (conclusion concisely presented in Table 1). This good contrast lithography allowed the progress to high resolution, in situ lithography using the projection printing mode. The synergy between the high resolution of the confocal microscopy exposure, coupled with the flexibility of designing the patterns and the control of the development and the attachment of the protein, allowed the printing of complex features, with high resolution and good definition (Fig. 5).
89
4. Disscusion
4.1. Bilayer lithography The multilayer lithography (several schemes reviewed in Dammel, 1993) was proposed to ‘push’ the limits of the microlithographic systems beyond what would be otherwise optically achievable in certain difficult technological conditions (e.g. highly nonplanar substrates). It was recently shown (Nicolau et al., 1998) that proteins can be patterned on exposed and undeveloped DNQ/novolak resists using photo-generated carboxylic groups as anchoring sites on the resist surface. Although this procedure gives high contrast, positive tone protein features, it requires a rather long processing time. Hence, this method might be inappropriate for a ‘step-and-repeat’ patterning. On the other hand, it was also recently shown (Nicolau et al., 1999) that proteins can be patterned on exposed and undeveloped P(tBuMA-co-MMA) resist using e-beam (or deep-UV) lithography. Although this method has the advantage of a fast and very selective attachment mechanism of the protein on the unexposed areas (i.e. negative tone, hydrophobicity controlled, lithography), it has the disadvantage of the requirement of high cost exposure tools (either e-beam of deep-UV). Moreover, although both methods allow the printing of protein features in both positive and negative tone, the patterning is limited to ‘surface printing’. To this end, it was considered that a bilayer scheme using both P(tBuMA) and DNQ/novolak resists would combine the advantages of the two materials, also bringing new benefits: (i) fast dynamics for the protein
Fig. 4. The 2 mm protein lines via negative tone, ‘protein-on-bottom’ lithography (contact printing only). Red fluorescent, hydrophilic resist lines (left); resist lines in contrast enhancement mode (2X magnification, middle); and green fluorescent protein lines (right).
Table 1 Relative advantages and disadvantages of techniques for printing proteins Technique
Resolution
Speed
Pattern definition
Dimension control
Process
Contact printing, ‘protein-on-top’ Contact printing, ‘protein-on-bottom’ Projection printing, ‘protein-on-bottom’
Neutral Neutral Favourable
Neutral Neutral Neutral
Unfavourable Favourable Favourable
Unfavourable Neutral Favourable
Unfavourable Neutral Favourable
90
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
Fig. 5. High resolution resist lines printed via confocal microscopy lithography. The minimum line is 0.5 mm in width. The slightly porous area is caused by the undesired development due to the parasitic exposure (in green light for imaging). The ‘holes’ are artifacts caused by accidental exposure during aligning of the patterning exposure.
attachment (hydrophobicity-controlled mechanism); (ii) capability of printing in both positive and negative tone; (iii) capability of printing on profiled features; (iv) capability of using either contact printing or projection printing modes.
4.2. Chemistry of the bilayer scheme The P(tBuMA) has been the ‘sensitivity-enhancer’ in some e-beam resist formulations (Miller and Brault, 1981) and more recently as 193 nm deep-UV resist (Allen et al., 1994). However, in the present study P(tBuMA) has been used solely for its high hydrophobicity, and therefore its radiation chemistry was not used for protein patterning (as described in a previous work, Nicolau et al., 1999). The DNQ/novolak system (extensively reviewed in Dammel, 1993) was, and still is to some extent, the workhorse of the semiconductor lithography for decades. In the context of the present study, DNQ photochemistry was used only for obtaining the profiled features and to manipulate the surface hydrophobicity of the resist surface. The photo-generated
carboxylic groups were not used to link the protein molecules on the exposed patterns (as described elsewhere, Nicolau et al., 1998). In general, the photoresist surface was hydrophilised in order to reject the attachment of the protein, because the hydrophobicity of the unexposed photoresist is not sufficient to promote an effective adhesion. The only exception to this rule was when the resist was exposed (hence rendered hydrophilic, allowing the diffusion of aqueous solutions) and subsequently presented to a H2O2/HCl solution which crosslinked the novolak resin (rendering the resist hydrophobic). The bottom surfaces, either PtBuMA — as in Fig. 2 — or glass, are hydrophilised by the oxidizing solution. The tBu moiety can be readily hydrolyzed on the surface (Miller and Brault, 1981), whereas the glass is assumed to be surface-functionalised with an increased number of OH groups.
4.3. Comparison of the protein patterning techniques The advantages and disadvantages of the presented biomicrolithographic techniques are compared in Table
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
1. In general, and provided that a confocal microscope can be retrofitted for protein patterning, the projection mode patterning seems to be the champion at all per-formance categories. The rather lower speed of the step and repeat mode is counterbalanced by the possibility to process all steps in one single flowcell. Future work is directed towards the use of better resolution optical tools, i.e. two-photon confocal microscopy, better filtering of the exposure to improve the control of the development step.
91
5. Conclusions Protein patterns were printed using conventional microlithographic materials in a bilayer resist arrangement using both ‘contact printing’ and ‘projection printing’ mode. The patterning exposure tools were either a modified UV illuminator or a confocal laser microscope, respectively. The ‘contact printing’ technique has the advantage of simplicity, while the ‘projection printing’ one has the advantage of high resolution and fully controlled process. Both techniques have the advantage of perfect portability in a biomedical environment, with potential applications in the fabrication of medical microanalysis devices.
4.4. Strength of the attachment and bioacti6ity of the protein features Acknowledgements The attachment of the streptavidin on the polymer surface was strong enough to withstand extended periods of thorough washing. In fact, the projection printing method uses a flow cell with continuous ‘washing’ of the patterns after their formation. Other proteins might have different hydrophobicity ‘window’ for an optimum attachment. However, we have proven that the methods presented are able to tune the hydrophobicity of the surface of both polymers, i.e. the photoresist (diazonaphtoquinone/novolak) and poly(tert-butyl-methacrylate), respectively, from hydrophilic/protein repelling to hydrophobic/protein attractive modes. Therefore, the attachment of a different protein will be easily accommodated using slightly different technology settings. Moreover, the projection printing method presented here offers the capability to observe the both the development and the formation of the (fluorescent) protein features in real time, and hence it offers the opportunity to stop the respective processes when an optimum result is achieved. Finally, commercial photoresists (here AZ 1900) as well as the bottom polymer (here PtBuMA) are available in wide range of initial hydrophobicities that might be more appropriate for a particular protein. It was reported (e.g. Pritchard et al., 1995) that avidin which was attached via hydrophobicity-induced mechanisms retains its bioactivity versus biotin. Although the hydrophobicity-induced attachment is less selective than chemically-controlled one (i.e. the chemical linkage of protein’s amino groups on carboxylic functionalised polymer surfaces, e.g. Nicolau et al., 1999), and because it is entirely a physically controlled mechanism, it is able to better preserve the biotin–avidin bioactivity of the protein. Future work is directed towards the optimisation of the photoresist used (the residual attachment of the proteins is kinetically rather than thermodynamically suppressed). .
We wish to thank Nick Carter for processing the pattern layouts into exposure routines. One author (DVN) was sponsored by a Fellowship awarded by the Australian Academy of Science/Royal Society of UK.
References Allen, R.D., Wallraff, G.M., Hinsberg, W.D., Simpson, L.L., Kunz, R.R., 1994. Methacrylate terpolymer approach in the design of a family of chemically amplified positive resists. In: Thompson, L., Willson, C.G., Tagawa, S. (Eds.), Polymers for Microelectronics, ACS Symp. Series, 537. American Chemical Society, Washington, DC. Dammel, R., 1993. Diazonaphtoquinoe-Based Resists. SPIE Press, Bellingham. Fodor, S.P.A., Read, J.L., Pirrung, M.C., Stryer, L., Lu, A.T., Solas, D., 1991. Light-directed, spatially addressable parallel chemical synthesis. Science 251, 767 – 773. Gallup, M.A., Barrett, R.W., Dower, W.J., Fodor, S.P.A., Gordon, E.M., 1994. Applications of combinatorial technologies to drug discovery. 1. Background and peptide combinatorial libraries. J. Med. Chem. 37, 1233 – 1251. Gray, D.E., Case-Green, S.C., Fell, T.S., Dobson, P.J., Southern, E.M., 1997. Ellipsometric and interferometric characterization of DNA probes immobilized on a combinatorial array. Langmuir 13, 2833 – 2842. McGall, G., Labadie, J., Brock, P., Wallraff, G., Nguyen, T., Hinsberg, W., 1996. Light-directed synthesis of high-density oligonucleotide arrays using semiconductor photoresists. Proc. Natl. Acad. Sci. USA 93, 13555 – 13560. Miller, L.J., Brault, R.G., 1981. A method for rapidly screening polymers as electron beam resists. J. Electrochem. Soc. 128, 1158 – 1161. Moritz, M.A., Paal, A.A., 1986. US Patent 4 007 047. Nicolau, D.V., Taguchi, T., Taniguchi, H., Yoshikawa, S., 1998. Micron-size protein patterning using on diazo-naphto-quinone/ novolak polymeric films. Langmuir 14, 1927 – 1936. Nicolau, D.V., Taguchi, T., Taniguchi, H., Yoshikawa, S., 1999. Positive and negative tone protein patterning using conventional deep-UV/e-beam resists. Langmuir 15, 3845 – 3851. Pritchard, D.J., Morgan, H., Cooper, J.M., 1995. Micron-scale patterning of biological molecules. Angew Chem. Int. Ed. Engl. 34 (1), 91 – 93.
Protein profiled features patterned via confocal microscopy Dan V. Nicolau a,b,*, Robert Cross a b
a Molecular Motors Group, Marie Curie Research Institute, Oxted, Surrey, RH8 0TL, UK Industrial Research Institute Swinburne, Swinburne Uni6ersity of Technology, PO Box 218, Hawthorn, Vic. 3122, Australia
Received 24 February 1999; accepted 9 December 1999
Abstract Protein patterns were printed using conventional microlithographic materials in a bilayer arrangement and unconventional exposure tools. The bilayer resist stack consisted of a lower poly(tert-butyl methacrylate) layer and an upper diazonaphtoquinone/ novolak layer. The protein features were printed in either ‘contact printing’, or ‘step and repeat’ mode. The latter printing mode can be managed in a flow-cell consisting of a standard microscope slide and cover slip, spaced apart by about 20 mm, as follows: (i) the exposure step is carried out in the cell using focused 488 nm beam of a confocal laser scanning microscope; (ii) the development step is performed by flowing the photoresist developer through the cell; (iii) the selective deposition of the protein (FITC-labelled avidin) is achieved via the flow of the protein solution through the cell until a desired contrast has been reached; (iv) the control of the process is assured using on-line monitoring of the photo-activated red fluorescence of the developing resist layer, and of the green fluorescence of the FITC-protein patterns, respectively. The protein printing technique uses equipment routinely available in biological laboratory. The ‘step and repeat’ patterning yields high and controllable resolution. The process can be applied in the fabrication of medical microanalysis devices. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Protein patterning; Bilayer resist; Confocal microscopy; Biosensors; Microlithography
1. Introduction The application of semiconductor technology in experimental biomedical science gained momentum in recent years. Microlithographic-like procedures have been used for light-assisted combinatorial chemistry pioneered almost a decade ago (Fodor et al., 1991) and employed to build high-density peptide libraries (Gallup et al., 1994) and oligonucleotide arrays (McGall et al., 1996; Gray et al., 1997). The use of microlithographic materials as functionalized scaffolds for building protein and peptide structures was also reported (Nicolau et al., 1998, 1999). In addition to the use of microlithography as a technological concept for printing high density features, and the use of microlithographic materials as scaffolds for bioactive architectures, the use of microlithographic technologies
would be a decisive advance towards the full integration of the semiconductor microlithography knowledge base into the biomedical research. The seamless integration of the semiconductor microlithography in biomedical research is precluded largely by the high cost of the high resolution exposure tools. The natural ‘technological’ solution would be to use either low cost exposure tools (e.g. contact printers), or better harness the optical tools that are already present in many biolaboratories (e.g. confocal microscopes). The present work tries to address both these avenues, using bilayer microlithographic technology, and optical tools present in most biolaboratories (i.e. deep-UV illuminators and confocal microscopes, respectively). 2. Materials and methods
2.1. Materials and layer deposition * Corresponding author. Tel.: +61-39-2145038; fax: + 61-392145050. E-mail address: [email protected] (D.V. Nicolau)
Because in biomedical research working with opaque substrates is uncommon, and therefore the use of reflec-
0956-5663/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S 0 9 5 6 - 5 6 6 3 ( 9 9 ) 0 0 0 7 2 - X
86
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
tion microscopes is unusual, we used microscopic cover slips as the base substrate. The glass of the cover slips is mildly hydrophobic assuring protein coverage. Poly(tert-butyl methacrylate), P(tBuMA), purchased from Aldrich was deposited as the bottom layer. The polymer is highly hydrophobic and has a molecular weight of :170 000. The :1 mm thick film was spin coated depositing a droplet of a 5% P(tBuMA) solution in xylene on cover slips; followed by spinning the substrate at 3000 rpm for 5 min using an in-house built spinner (an Eppendorf benchtop microcentrifuge); then soft-baked in a convection oven for 3 h at 90°C. The long baking time assured the avoidance of the interfacial mixing between PtBuMA and the top layer photoresist. The top layer consisted of a common g-line AZ type photoresist, i.e. AZ 1900, in the low viscosity formulation, purchased from Clariant. The 0.6 mm top layer was spin coated at 4000 rpm for 5 min and soft-baked for 1 h at 80°C. AZ type photoresists contain a photoactive compound, a derivative of diazonaphtoquinone (DNQ), a base resin (novolak) and a number of other minor chemical species (sensitisers). The DNQ:novolak weight ratio is :1:4.
2.2. Patterning procedures The patterning procedure consisted in (i) the patternexposure of the bilayer film; (ii) controlled development; (iii) surface treatment for protein attachment selectivity; and (iv) protein deposition. Several techniques for protein patterning are presented in Fig. 1.
2.2.1. Exposure E-beam patterned (2 mm lines connecting 5×5 mm squares) chromium masks (Hoya) were used to print features in the photoresist top layer in the ‘contact printing’ mode. The ‘contact printing’ exposure tool was an in-house built contact printer, adapted via the modification of a deep-UV illuminator. The deep-UV illuminator is a planar irradiation equipment commonly used in biolaboratories for exciting fluorofores bound to DNA samples. An optical filter limits the wavelength range of the exposure approximately between 290 and 450 nm. The chromium mask was placed on the deepUV illuminator, then the cover slip was placed with the photoresist facing the mask, and finally the whole ensemble was thoroughly contacted. This set-up, similar to a broadband UV exposure contact printing, was
Fig. 1. Protein patterning procedures using bilayer lithography.
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
used for multi-step, ex situ protein patterning. The optimum exposure energy was 80 mJ/cm2. For high resolution, in situ lithography, a ‘projection printing’ arrangement was devised comprising a flow cell and a confocal laser microscope. The flow-cell was built (i) attaching the cover slip with the photoresist on top on a microscope glass slide; (ii) attaching parallel lines of rubber spacers on the microscope glass slide leaving a longitudinal inlet and an outlet for fluid flow; (iii) attaching a transparent cover slip on the spacers (the thickness of the spacers assures the presence of a gap — about 20 mm — between the photoresist surface and the top cover slip). The scanning confocal laser microscope was a BioRad MRC600 having a KrAr mixed gas laser. The 488 nm line of the 3 mW laser was used for exposure of the photoresist. Neutral density filtration was at 1% for the exposure. The exposed pattern could be visualised using the 568 nm (green) line of the laser to excite the fluorescence of the photoresist. The patterned resist was visible as bright red fluorescence (unexposed resist) on a darker red background (exposed resist), as a result of photoactivation of the DNQ fluorescence. Higher power exposure bleached the fluorescence of the resist. The layout of the patterns was designed using Adobe Illustrator and rasterised by importing the file into Photoshop (both from Adobe Systems). The rasterised file was then translated into a sequence of PARK commands. The PARK (x,y) command is intrinsic to the BioRad COMOS software and has the effect of driving the beam steering mirrors to a particular X,Y position in the field. A sequence of PARK commands, as used here, exposes each pixel in the pattern for about 20 ms. The microscope used for the control of the exposure was a Nikon Optiphot II, and used a 60x differential interference contrast (DIC) planapo oil immersion objective with a numerical aperture (NA) 1.4.
2.2.2. De6elopment The developer used was a common AZ 351 (purchased from Clariant). The contact printing mode required only minor developing control. The development can be also performed using an aqueous fresh solution of 0.05 n NaOH. 2 mm feature were resolved using an immersion development for 50 s. After development, the surface of the resist was thoroughly washed with deionised water. The confocal microscope generates high exposure energies compared to semiconductor lithography. Despite the fact that the confocal microscope was tuned to 488 nm (instead of the 465 nm of the g-line), and despite the attenuation used (only :1% of the exposure was delivered to the surface of the photoresist), a very diluted developer (1:10) had to be used for in situ protein patterning for a controlled development. Moreover, short development times had to be used (in the range of 20 – 30 s). The diluted developer was flown through the
87
gap of the flow-cell using a syringe. After development, the surface of the resist was washed with deionised water in the same flow-cell. The developed patterns were observed using DIC optics in a Nikon Eclipse 800 upright microscope, again using NA 1.4 oil-immersion optics, but with additional magnification to create an : 25 mm square field at the camera faceplate. The camera was a Hamamatsu C4880-01 chilled CCD, with a 512× 512 pixel chip. Images from the camera were grabbed using a Snapper card and Improvision software (from Openlab).
2.2.3. Surface treatment In all protein printing modes the hydrophobicity of the patterned features was manipulated to assure the highest protein contrast. The chemical processes responsible of hydrophobicity manipulation are presented in Fig. 2. Contact printing. In the ‘protein-on-top’ mode, the surface treatment aimed at making the photoresist lines as hydrophobic as possible and making the bottom of the channels (i.e. the ‘spaces’) as hydrophilic as possible. This was achieved through (i) the flood exposure of the patterned and developed features; (ii) the immersion of the patterned substrate in a solution of HCl:H2O2 at 90°C for 1/2 h, following a procedure described as an alternative Image Reversal technique (Moritz and Paal, 1986). In this mode, the bottom P(tBuMA) layer is actually redundant, and the scheme equally works with a single layer of photoresist (the treatment with the oxidizing solution renders the glass hydrophilic). In the ‘protein-on-bottom’ mode, the surface treatment aimed at making the photoresist lines as hydrophilic as possible and making the bottom of the channels (i.e. the ‘spaces’) as hydrophobic as possible. This was achieved through the simple flood exposure with a high energy and at higher wavelength (no UV light). Projection printing. The features produced in-situ via ‘projection’ printing were appropriate only for ‘proteinon-bottom’ mode. Therefore the surface treatment limited to a flood, green light, high energy exposure with the confocal microscope. 2.2.4. Protein deposition The protein patterning consisted in the hydrophobicity-controlled attachment of a fluorescent protein (i.e. FITC avidin) either on lines (‘protein-on-top’ mode) or on the bottom of the patterned and developed channels (‘protein-on-bottom’ mode). The patterned substrates were (i) flooded with a solution of FITC-avidin at a concentration of 25 mg/ml; (ii) incubation for several minutes in the flow cell at ambient temperature; and (iii) washing with deionised water followed by soft drying. The contrast of the protein patterning was controlled on line, i.e. via the monitoring of the fluorescence contrast
88
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
Fig. 2. Physico-chemistry of protein patterning using the bilayer process.
Fig. 3. The 2 mm protein lines printed via positive tone, ‘protein-on-top’ lithography. Red fluorescent, hydrophobic resist features (left); resist lines (2X magnification) in contrast enhancement mode (middle); and green fluorescent FITC-protein lines (right).
using the confocal microscope tuned to detect green fluorescence. The same setup used for monitoring the development was used for fluorescence observations following flood exposure of the pattern to fluorescently-labelled streptavidin. In this case epifluorescence illumination from a mercury lamp was used to excite the labelled streptavidin using the standard fluorescein epifluorescence dichroic filter cube of the microscope.
3. Results
3.1. ‘Protein-on-top’ microlithography The ‘protein-on-top’ contact mode protein lithography produced features with good definition of the fluorescent image (Fig. 3, left). However, the aggressive treatment of the resist features resulted in a poor definition of the resist lines (Fig. 3 middle and Fig. 3 right,
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
fluorescence and DIC image of the profiled resist patterns, respectively). Subsequently, the protein attached with great selectivity of this poorly defined template.
3.2. ‘Protein-on-bottom’ microlithography The ‘protein-on-bottom’, broadband, contact printing mode protein lithography produced features with very good contrast of the protein image (Fig. 4 left). The quantification of the density of the protein is rather difficult using fluorescence-based relative methods of measuring the surface concentration of the adsorbent. However, the fluorescent image of the protein (in Fig. 4) proves that all avidin is present on the features only. No trace of protein could be detected outside the patterns (bottom of the channels). The milder lithographic procedure allowed a good definition of the patterned resist lines (Fig. 4, middle; and Fig. 4, right; fluorescence and DIC image of the profiled resist patterns, respectively), which were subsequently reproduced with good accuracy by the protein attachment. The quality of the patters indeed varies from one method to another (conclusion concisely presented in Table 1). This good contrast lithography allowed the progress to high resolution, in situ lithography using the projection printing mode. The synergy between the high resolution of the confocal microscopy exposure, coupled with the flexibility of designing the patterns and the control of the development and the attachment of the protein, allowed the printing of complex features, with high resolution and good definition (Fig. 5).
89
4. Disscusion
4.1. Bilayer lithography The multilayer lithography (several schemes reviewed in Dammel, 1993) was proposed to ‘push’ the limits of the microlithographic systems beyond what would be otherwise optically achievable in certain difficult technological conditions (e.g. highly nonplanar substrates). It was recently shown (Nicolau et al., 1998) that proteins can be patterned on exposed and undeveloped DNQ/novolak resists using photo-generated carboxylic groups as anchoring sites on the resist surface. Although this procedure gives high contrast, positive tone protein features, it requires a rather long processing time. Hence, this method might be inappropriate for a ‘step-and-repeat’ patterning. On the other hand, it was also recently shown (Nicolau et al., 1999) that proteins can be patterned on exposed and undeveloped P(tBuMA-co-MMA) resist using e-beam (or deep-UV) lithography. Although this method has the advantage of a fast and very selective attachment mechanism of the protein on the unexposed areas (i.e. negative tone, hydrophobicity controlled, lithography), it has the disadvantage of the requirement of high cost exposure tools (either e-beam of deep-UV). Moreover, although both methods allow the printing of protein features in both positive and negative tone, the patterning is limited to ‘surface printing’. To this end, it was considered that a bilayer scheme using both P(tBuMA) and DNQ/novolak resists would combine the advantages of the two materials, also bringing new benefits: (i) fast dynamics for the protein
Fig. 4. The 2 mm protein lines via negative tone, ‘protein-on-bottom’ lithography (contact printing only). Red fluorescent, hydrophilic resist lines (left); resist lines in contrast enhancement mode (2X magnification, middle); and green fluorescent protein lines (right).
Table 1 Relative advantages and disadvantages of techniques for printing proteins Technique
Resolution
Speed
Pattern definition
Dimension control
Process
Contact printing, ‘protein-on-top’ Contact printing, ‘protein-on-bottom’ Projection printing, ‘protein-on-bottom’
Neutral Neutral Favourable
Neutral Neutral Neutral
Unfavourable Favourable Favourable
Unfavourable Neutral Favourable
Unfavourable Neutral Favourable
90
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
Fig. 5. High resolution resist lines printed via confocal microscopy lithography. The minimum line is 0.5 mm in width. The slightly porous area is caused by the undesired development due to the parasitic exposure (in green light for imaging). The ‘holes’ are artifacts caused by accidental exposure during aligning of the patterning exposure.
attachment (hydrophobicity-controlled mechanism); (ii) capability of printing in both positive and negative tone; (iii) capability of printing on profiled features; (iv) capability of using either contact printing or projection printing modes.
4.2. Chemistry of the bilayer scheme The P(tBuMA) has been the ‘sensitivity-enhancer’ in some e-beam resist formulations (Miller and Brault, 1981) and more recently as 193 nm deep-UV resist (Allen et al., 1994). However, in the present study P(tBuMA) has been used solely for its high hydrophobicity, and therefore its radiation chemistry was not used for protein patterning (as described in a previous work, Nicolau et al., 1999). The DNQ/novolak system (extensively reviewed in Dammel, 1993) was, and still is to some extent, the workhorse of the semiconductor lithography for decades. In the context of the present study, DNQ photochemistry was used only for obtaining the profiled features and to manipulate the surface hydrophobicity of the resist surface. The photo-generated
carboxylic groups were not used to link the protein molecules on the exposed patterns (as described elsewhere, Nicolau et al., 1998). In general, the photoresist surface was hydrophilised in order to reject the attachment of the protein, because the hydrophobicity of the unexposed photoresist is not sufficient to promote an effective adhesion. The only exception to this rule was when the resist was exposed (hence rendered hydrophilic, allowing the diffusion of aqueous solutions) and subsequently presented to a H2O2/HCl solution which crosslinked the novolak resin (rendering the resist hydrophobic). The bottom surfaces, either PtBuMA — as in Fig. 2 — or glass, are hydrophilised by the oxidizing solution. The tBu moiety can be readily hydrolyzed on the surface (Miller and Brault, 1981), whereas the glass is assumed to be surface-functionalised with an increased number of OH groups.
4.3. Comparison of the protein patterning techniques The advantages and disadvantages of the presented biomicrolithographic techniques are compared in Table
D.V. Nicolau, R. Cross / Biosensors & Bioelectronics 15 (2000) 85–91
1. In general, and provided that a confocal microscope can be retrofitted for protein patterning, the projection mode patterning seems to be the champion at all per-formance categories. The rather lower speed of the step and repeat mode is counterbalanced by the possibility to process all steps in one single flowcell. Future work is directed towards the use of better resolution optical tools, i.e. two-photon confocal microscopy, better filtering of the exposure to improve the control of the development step.
91
5. Conclusions Protein patterns were printed using conventional microlithographic materials in a bilayer resist arrangement using both ‘contact printing’ and ‘projection printing’ mode. The patterning exposure tools were either a modified UV illuminator or a confocal laser microscope, respectively. The ‘contact printing’ technique has the advantage of simplicity, while the ‘projection printing’ one has the advantage of high resolution and fully controlled process. Both techniques have the advantage of perfect portability in a biomedical environment, with potential applications in the fabrication of medical microanalysis devices.
4.4. Strength of the attachment and bioacti6ity of the protein features Acknowledgements The attachment of the streptavidin on the polymer surface was strong enough to withstand extended periods of thorough washing. In fact, the projection printing method uses a flow cell with continuous ‘washing’ of the patterns after their formation. Other proteins might have different hydrophobicity ‘window’ for an optimum attachment. However, we have proven that the methods presented are able to tune the hydrophobicity of the surface of both polymers, i.e. the photoresist (diazonaphtoquinone/novolak) and poly(tert-butyl-methacrylate), respectively, from hydrophilic/protein repelling to hydrophobic/protein attractive modes. Therefore, the attachment of a different protein will be easily accommodated using slightly different technology settings. Moreover, the projection printing method presented here offers the capability to observe the both the development and the formation of the (fluorescent) protein features in real time, and hence it offers the opportunity to stop the respective processes when an optimum result is achieved. Finally, commercial photoresists (here AZ 1900) as well as the bottom polymer (here PtBuMA) are available in wide range of initial hydrophobicities that might be more appropriate for a particular protein. It was reported (e.g. Pritchard et al., 1995) that avidin which was attached via hydrophobicity-induced mechanisms retains its bioactivity versus biotin. Although the hydrophobicity-induced attachment is less selective than chemically-controlled one (i.e. the chemical linkage of protein’s amino groups on carboxylic functionalised polymer surfaces, e.g. Nicolau et al., 1999), and because it is entirely a physically controlled mechanism, it is able to better preserve the biotin–avidin bioactivity of the protein. Future work is directed towards the optimisation of the photoresist used (the residual attachment of the proteins is kinetically rather than thermodynamically suppressed). .
We wish to thank Nick Carter for processing the pattern layouts into exposure routines. One author (DVN) was sponsored by a Fellowship awarded by the Australian Academy of Science/Royal Society of UK.
References Allen, R.D., Wallraff, G.M., Hinsberg, W.D., Simpson, L.L., Kunz, R.R., 1994. Methacrylate terpolymer approach in the design of a family of chemically amplified positive resists. In: Thompson, L., Willson, C.G., Tagawa, S. (Eds.), Polymers for Microelectronics, ACS Symp. Series, 537. American Chemical Society, Washington, DC. Dammel, R., 1993. Diazonaphtoquinoe-Based Resists. SPIE Press, Bellingham. Fodor, S.P.A., Read, J.L., Pirrung, M.C., Stryer, L., Lu, A.T., Solas, D., 1991. Light-directed, spatially addressable parallel chemical synthesis. Science 251, 767 – 773. Gallup, M.A., Barrett, R.W., Dower, W.J., Fodor, S.P.A., Gordon, E.M., 1994. Applications of combinatorial technologies to drug discovery. 1. Background and peptide combinatorial libraries. J. Med. Chem. 37, 1233 – 1251. Gray, D.E., Case-Green, S.C., Fell, T.S., Dobson, P.J., Southern, E.M., 1997. Ellipsometric and interferometric characterization of DNA probes immobilized on a combinatorial array. Langmuir 13, 2833 – 2842. McGall, G., Labadie, J., Brock, P., Wallraff, G., Nguyen, T., Hinsberg, W., 1996. Light-directed synthesis of high-density oligonucleotide arrays using semiconductor photoresists. Proc. Natl. Acad. Sci. USA 93, 13555 – 13560. Miller, L.J., Brault, R.G., 1981. A method for rapidly screening polymers as electron beam resists. J. Electrochem. Soc. 128, 1158 – 1161. Moritz, M.A., Paal, A.A., 1986. US Patent 4 007 047. Nicolau, D.V., Taguchi, T., Taniguchi, H., Yoshikawa, S., 1998. Micron-size protein patterning using on diazo-naphto-quinone/ novolak polymeric films. Langmuir 14, 1927 – 1936. Nicolau, D.V., Taguchi, T., Taniguchi, H., Yoshikawa, S., 1999. Positive and negative tone protein patterning using conventional deep-UV/e-beam resists. Langmuir 15, 3845 – 3851. Pritchard, D.J., Morgan, H., Cooper, J.M., 1995. Micron-scale patterning of biological molecules. Angew Chem. Int. Ed. Engl. 34 (1), 91 – 93.