Nội dung

Biomedical Microdevices 6:2, 117±123, 2004 # 2004 Kluwer Academic Publishers. Manufactured in The Netherlands. BioMEMS Materials and Fabrication Technology Section: Microfabricated Quill-Type Surface Patterning Tools for the Creation of Biological Micro/Nano Arrays Juntao Xu,1* Michael Lynch,1 Janice L. Huff,1 Curtis Mosher,1 Srikanth Vengasandra,1 Guifu Ding,3 and Eric Henderson1,2 1 BioForce Nanosciences Inc., 2901 South Loop Drive, Suite 3400, Ames, Iowa 50010 E-mail: jxu@bioforcenano.com 2 Iowa State University, GDCB Department, Ames, Iowa 50011 3 Institute of Micro/Nanometer Sciences and Technology, Shanghai JiaoTong University, Shanghai, P.R. China Abstract. Novel quill-type cantilever-based surface patterning tools (SPTs) were designed and constructed for use in controlled placement of femtoliter volumes of biological molecules on surfaces for biological applications. These tools were fabricated from silicon dioxide using microelectromechanical systems (MEMS) techniques. They featured a 1 lm split gap, ¯uidic transport microchannels and self-replenishing reservoirs. Experimental trials were performed using these tools on NanoArrayerTM molecular deposition instrumentation. Cy3-streptavidin was loaded as a biological sample and patterned on an amine-reactive dithiobis-succinimidyl undecanoate (DSU) monolayer on gold. Results showed these tools were capable of generating high quality biological arrays with routine spot sizes of 2±3 lm. The spot size could potentially achieve sub-micron dimensions if these SPT designs are reduced in size by more precise microfabrication techniques. The geometric designs of these tools facilitated sample replenishment from the local reservoir on the cantilever which allowed printing of large numbers of spots without sample reloading. Key Words. surface patterning tool, quill-type, microcantilever, microarray, nanoarray Introduction With the rapid development of nanotechnology, a variety of novel microfabricated tools are required for exploitation and manipulation of the nano-scale world (Fujita et al., 2001; Kakushima et al., 2001). Devices that can deliver femto- or even attoliter volumes of materials onto a substrate to form micro- and nanometer sized arrays and patterns constitute one important category of such tools. Once it is possible to routinely pattern surfaces on the micron- and sub-micron spatial scales with biomaterials, the door will be open for development of a vast spectrum of ultraminiaturized bioanalytical tests and devices (Mitchell, 2002). To date most bio-patterning experiments in the micron to submicron spatial scale have been carried out using microcantilever-based atomic force microscopy (AFM) probes (Jaschke and Butt, 1995; Piner et al., 1999; Amro et al., 2000). This approach is attractive because AFM probes are readily available and the microcantilever design facilitates force control between the tool and surface. However, since AFM probes are speci®cally designed for imaging purposes, their structure, material and geometry are not optimized for materials patterning, especially for biological materials. With an AFM probe, the sample forms a thin molecular layer on the tip surface and around its base area. When placed in contact with a surface, these molecules migrate from the tip onto the surface and the time of surface contact directly correlates with the amount of material transferred. This process works well for small organic species, but is problematic for larger biomolecules including proteins, large nucleic acids and biomolecular ensembles. Furthermore, since the reservoir of material is limited to the surface layer, the sample depletes, resulting in the requirement for reloading during protracted patterning. Finally, the geometry of a sharp AFM tip is not suitable for transportation of macromolecules from the base area to the end of the tip where surface contact is made. As a consequence, biomolecular arrays created with AFM probes can result in spots that are devoid of the patterning material in the center in our previous experiments (results not shown). A less frequently used scanning probe for bio-patterning is derived from the scanning near ®eld optical microscope (Meister et al., 2002). In *Corresponding author. 117 118 Xu et al. this approach a nanometer-sized aperture (* 100 nm) is made at the end of a hollow pyramidal or conic probe tip. Materials are transferred onto the substrate through this aperture. Although it is logical to correlate the patterning spot size to the diameter of the aperture, to date spot sizes achieved by this approach are still well above 1 mm (Meister et al., 2002). In addition, the methods used for fabrication of these apertures, such as focused ion beam (FIB) (Veerman et al., 1998), are serial in nature, precluding facile high volume production. Furthermore, loading a sample into the tiny probe tip constitutes yet another obstacle to the utility of these tools. A related strategy uses cantilevered micropipettes (tapered hollow core ®ber) to pattern chemicals (Lewis et al., 1999). This method can achieve continuous sample loading through the delivery tube, and spot sizes reach sub-micron dimensions. However, these devices are subject to blockage by small particulates, their use is complex, and the method used to fabricate them serial in nature. We have investigated a different approach that exploits both the advantages of microcantilever devices and the quill pin tool designs that have been used successfully in the microarray industry (Heller, 2002). The quill pin tool principle is analogous to that used in a fountain pen. Capillary force in a narrow hydrophilic channel facilitates the delivery of liquid from a reservoir to the end of the pin. When the pin tool contacts a surface, liquid ¯ows from the channel onto the surface. Pin tools speci®cally designed for microarrays produce minimum spot sizes in the range of tens to hundreds of microns. In the current study, we sought to ultraminiaturize this design principle in the construction of a quilltype, cantilever-based surface patterning tool (SPT). We showed that it was possible, using advanced microfabrication techniques, to shrink the tool size, thereby reducing the spot diameter down to a few microns, with potential for nanoscale dimensions. Another recent report demonstrates the use of microcantilever devices for creation of biological arrays (Belaubre et al., 2003). However, the tool cantilever geometry in this study is relatively large and the minimum spot size produced is approximately 30 mm. The ®rst generation devices described in our study featured integrated self-replenishing sample reservoirs and ¯uid transportation microchannels, which addressed limitations inherent in the use of AFM probes for molecular patterning. When tested on NanoArrayerTM instrumentation (described in Materials and Methods section) for use in molecular patterning applications, these new designs allowed reliable patterning of large molecular species, reduced reloading requirements, and featured back-loading, a process that will facilitate future parallelization of the process and elimination of washing steps. Design and Fabrication Surface patterning tool design Cantilevers for use in AFM have stringent design requirements to ensure regulation of surface contact force and resonance frequency. These parameters are somewhat relaxed when designing cantilevers for use in molecular patterning. Consequently, the design space for the selection of cantilever materials, structure and geometry used in this study was correspondingly increased. Our strategies were calculated to exploit these additional freedoms. Although relaxed relative to AFM, the force constant of these devices could not be totally ignored. If the force constant is too large the probability of scratching the surface becomes too great. Conversely, if the force constant is too low, the cantilever is likely to stick to the surface by electrostatic or capillary forces. Therefore, we designated a desired range of force constants to be 0.03± 0.3 N/m (Zhang et al., 2002). To make a 1 mm deep channel on the cantilever, the total thickness of the quilltype SPT had to be much thicker than that of a conventional AFM probe. The force constant k of a diving board cantilever with Young's Modulus E, length l, width w and thickness t is given as (Zhang et al., 2002): kˆ Ewt3 : 4l3 1† Thickness and length are the main parameters impacting the force constant. To obtain an appropriate force constant, our cantilevers had to be relatively long. However, as the length increases so does the ¯uid transportation distance, which could result in increased chances of blockage or other forms of restriction causing reduced ¯uid ¯ow. Therefore, for this particular application, we endeavored to ®nd the optimally compromised values for cantilever width w† and cantilever material (Young's modulus E) with cantilever length l†. Figure 1 shows the general structure of our quill-type Fig. 1. Schematic of quill-type cantilever-based SPT. Microfabricated Quill-Type Surface Patterning Tools SPTs. These tools consisted of a cantilever with length in the range of 200±300 mm, width of 20±40 mm and thickness of 2±3 mm. At the distal end of the cantilever, there was a split gap with width of * 1 mm and length of * 40 mm. At the ®xed end of the cantilever, there was a 10 mm deep rectangular reservoir on the supporting substrate. A 1±10 mm wide microchannel on the cantilever connected the reservoir and the split gap. The depth of the channel was about 1 mm. Silicon dioxide, SiO2, was chosen as the tool material as it offers several signi®cant advantages for molecular patterning. SiO2 has a smaller Young's Modulus than either Silicon, Si, or Silicon Nitride, Si3N4 (SiO2: 70 GPa (Kim, 1996), Si: 190 GPa (Zhang et al., 2002), Si3N4: 385 GPa (Zhang et al., 2002)), and thus reduced the requisite cantilever length. SiO2 also facilitates cantilever thickness control during the ®nal releasing process by KOH etching because no critical time thickness control or boron doping process is required. SiO2 is biocompatible and highly hydrophilic, which facilitates ®lling of the microchannel with ¯uid. Finally, SiO2 is 119 transparent to visible light, enhancing its utility for in situ sample loading control through optical visualization. Based on the cantilever material, structure and geometry we chose, the force constant of the quill-type SPTs was in the range of 0.1±1 N/m, well within the optimal range designated for our purposes. Surface patterning tool fabrication The SPT is a disposable device. Therefore, the most advantageous microfabrication process will be cost ef®cient with a high yield. For these reasons, we selected only simple and robust fabrication processes in our initial trials. Figure 2 shows the main process ¯ow used for cantilever construction in this study. Starting material was a 3 inch double-side polished n-type f100g silicon wafer. In step one, both sides of the wafer were thermally grown with 2±3 mm SiO2 (wet oxidation). The front side SiO2 layer was used for construction of cantilevers. The backside SiO2 layer was used as a mask for release of cantilevers in a ®nal silicon anisotropic etching step. In step two, the front side SiO2 layer was patterned to de®ne Fig. 2. Schematic diagram of the fabrication processes for quill-type SPT. (a) Double side wet oxidation of 300 silicon wafer, SiO2 thickness was about 2±3 mm. (b) De®nition of cantilever, gap and reservoir by RIE, 500 nm electroplating Ni was used as hard mask. (c) De®nition of microchannel on cantilever by overlay lithography and RIE. (d) Release of cantilever by backside KOH etching. 120 Xu et al. the cantilever shanks, split gaps and reservoirs. This fabrication step was critical for the success of this novel SPT. It is not trivial to fabricate a gap with 1 mm feature size and 3 : 1 aspect ratio using conventional UV photolithography. To accomplish this step, a negative photoresist pattern was transferred into a 500 nm thick nickel positive pattern by mask electroplating, and then the nickel metal layer served as a hard mask for underneath SiO2 etching by anisotropic reactive ion etching (RIE). Reactive gases were a mixture of CHF3 (50 SCCM) and SF6 (1 SCCM) at 50 mTorr pressure. 50 W RF power was used. Since the etching selectivity of SiO2 to nickel was much higher than to photoresist, it allowed more precise control of the gap geometry than if we had used a thick photoresist as a mask for SiO2 etching. After cantilevers were de®ned, in step three, a 1 mm deep microchannel was fabricated by overlay photolithography followed by RIE. In the ®nal step, the backside SiO2 window was opened and the cantilever released by KOH anisotropic etching. Concentration of the KOH was 35 wt% in water, processing temperature was about 80  C. Materials and Methods All fabricated quill-type SPTs were tested using a commercial nano-patterning instrument called a NanoArrayerTM (Figure 3) (BioForce Nanosciences, Inc., Ames, IA) that is based on a patented technique (Henderson and Mosher, 2003). Brie¯y, this instrument Fig. 3. Schematic of the NanoArrayerTM instrumentation for molecular deposition and patterning. uses a precision motion control system in an environmentally regulated chamber for surface patterning using microfabricated tools like those described here. Surface contact force is controlled via an optical lever detection system similar to those employed by an atomic force microscope (Meyer and Amer, 1988). It is important, however, to make the distinction between the NanoArrayerTM and an AFM. This instrument does not scan or acquire images, and features no cantilever oscillation. Likewise, the SPTs described in this report would not function as AFM probes for a number of reasons, not the least of which is lack of an imaging ``tip'' or protuberance. A high magni®cation vision system is used to monitor sample loading and the patterning process. The entire process can be automated or controlled manually using custom designed software (NanowareTM) with an integrated graphical user interface (GUI). Sample preparation and patterning Puri®ed Cy3-streptavidin (Jackson ImmunoResearch Laboratories, West Grove, PA) in phosphate-buffered saline (PBS) with 10% glycerol was used exclusively for the experiments described in this report. Glycerol was added to prevent evaporation and to keep the protein hydrated and bioactive. The Cy3 ¯uorophore allowed monitoring of the streptavidin by ¯uorescence microscopy after patterning. Before loading with Cy3-streptavidin, the SPT was UV/O3 treated using a TipCleanerTM (BioForce Nanosciences, Inc., Ames, IA) for 20 minutes. This process cleaned the surface of the microchannel and split gap and rendered them hydrophilic, thus facilitating ®lling by the sample. Cy3-streptavidin sample solution (1 ml) was delivered into the reservoir on the backside of cantilever by micropipette. It spontaneously ®lled the microchannel by hydrophilic capillary force. Then the back-loaded SPT was mounted on the NanoArrayerTM for patterning. Dithiobis-succinimidyl undecanoate (DSU) (Dojindo, Japan) treated gold surfaces were used as deposition substrates (Wagner et al., 1996). DSU forms a self-assembling monolayer on gold surfaces through the strong chemical interactions between sulfur and gold as well as the hydrophobic packing of the alkyl chains. The exposed amine-reactive succinimide group enabled the covalent binding of Cy3streptavidin to the deposition substrate. All experiments were performed under ambient conditions with a relative humidity of 35±40% and temperature of 23±24  C. After patterning, a Nikon TE 2000U inverted microscope equipped with a 40 6 oil objective and Chroma Technology (Vermont) ®lter set for Cy3TM(#41007a) was used to visualize the Cy3streptavidin patterned on DSU/gold. Images were Microfabricated Quill-Type Surface Patterning Tools 121 acquired with a Hamamatsu (Japan) ORCA ER cooled CCD camera. A similar procedure was used to directly image the ¯uorescent sample in the micro-channel and gap of the microfabricated SPT. Fluorescent array images were analyzed for net intensity, diameter, area, and coef®cient of variance with the Array Pro Analyzer v4.5 software package from Media Cybernetics (Carlsbad, California). Results and Discussion Quill-type SPTs with micron sized split gaps and microchannels on the cantilever have been successfully fabricated. Figure 4 shows SEM images of SPTs, which include both single cantilever (see Figure 4a) and multiple cantilever (see Figure 4b) designs with 40 mm long split gaps and 10 mm wide microchannels. The cantilevers all appear straight and ¯at, with no undesirable warping or bending that can sometimes occur in the ®nal releasing process when internal stress and the stress gradient of cantilever materials are excessive (Liu and Gamble, 1998). We conclude, therefore, that thermally grown silicon dioxide has low internal stress and is suitable for constructing cantileverbased SPTs. In addition, the cantilever releasing process is robust. Since the etching selectivity of Si to SiO2 in 35% aqueous KOH at 80  C is about 300 : 1, this process is tolerant of some overtime etching without signi®cantly affecting cantilever geometry. It is suitable for mass fabrication. The yield rates in our ®rst fabrication trial are as high as 75%. Figure 4c and d show, respectively, magni®ed SEM images of the microchannel and gap on a cantilever. The microchannel is clean and smooth. No etching residue or Fig. 4. SEM pictures of fabricated quill-type SPTs. (a) Single cantilever SPT. (b) Multiple cantilever SPT. (c) 1 mm deep microchannel on cantilever. (d) 1 mm wide split gap at the end of cantilever. Fig. 5. Optical images of quill-type SPTs with different geometric combinations between gap and microchannel. blockage is found inside the channel or at the joint between the micro-channel and gap. The split gap is tapered from 10 mm at the channel to 1 mm at the very end of cantilever. The radius of curvature of the two fork arms is also about 1 mm. It is noteworthy that the minimum gap feature size is limited by the conventional UV photolithography process. Thus, this limitation can be reduced to the nanometer scale if high resolution lithography techniques such as E-beam lithography or phase-shift mask are used (Yin et al., 2000). We constructed a number of geometric variations between the gap and the micro-channel in the same microfabrication run. These were used to evaluate the performance and optimize the design of the microchannel and gap for molecular patterning. Figure 5 shows optical images of several of these design variations. Surface patterning tool performance After sample loading, quill-type SPTs were tested on a NanoArrayerTM using Cy3-streptavidin in a standard protein patterning application. Figure 6a shows a ¯uorescent image of a 10 6 10 Cy3-streptavidin array patterned on a DSU/gold surface after washing away unbound protein with ddiH2O. The distance between Fig. 6. (a) Fluorescent image of 10 6 10 Cy3-streptavidin array deposited by quill-type SPT on DSU/gold surface. (b) Letters formed by Cy3-streptavidin spots. Space between spots is all 5 mm. 122 Xu et al. spots is 5 mm, with a mean spot diameter of 2.2 mm and a coef®cient of variance (CV) of 4.1% (approx. 4 ¯ volume each spot). Spots are relatively round in shape and highly ¯uorescent, with a net intensity CV of 7.9%. With a single loading, it was able to print at least 3,000 spots (30 10 6 10 arrays were written in about one hour in our experiment) and all spots were ¯uorescent. Figure 6b shows ¯uorescent patterns in the form of letters made by Cy3-streptavidin spots. It was possible to print virtually any pattern using these tools. Again, the spots are all round in shape and highly ¯uorescent. However, it is noticeable that, in Figure 6, the spots are not all perfectly uniform in shape and size. The process of patterning liquid samples onto a substrate is controled by several factors including the SPT itself. Local DSU/gold surface uniformity, contact force and dwell time variation, as well as ambient humidity ¯uctuation all affect spot size. Systematic studies on the reproducibility of spotting by SPT will be conducted in our ongoing research. Sample ®lling Fluorescent images of quill-type SPTs after use for Cy3streptavidin patterning are shown in Figure 7a and b. The ¯uorescent intensity inside the channel and gap was much higher than elsewhere on the cantilever or substrate. Most of the Cy3-streptavidin sample was retained within the microchannel. This illustrates that the microchannels in these devices can ef®ciently conduct sample ¯uid from the reservoir to the gap through capillary force. Thus, sample replenishment, an important feature of a high volume or high throughput system, was possible with our designs. Some SPTs exhibited sample ¯uid outside the reservoir. In cases where pipetting was not the obvious cause, this appeared to result from sample ¯uid wetting from the reservoir to adjacent surfaces. Since the inside surface of the reservoir was silicon and the outside surface was the more hydrophilic SiO2, when sample was ®lled to the top edges of the reservoir there was no impediment for ¯uid Fig. 7. Fluorescent images of quill-type SPT after patterning. (a) Gap and microchannel ®lled with Cy3-streptavidin. (b) ``Spillwetting'' of Cy3-streptavidin to area outside of reservoir (as shown by arrows). to continue to wet onto the outside surface spontaneously, or onto the cantilever surface. Although this ``spill-wetting'' problem does not affect patterning (SPTs are mounted at a 12 angle to the deposition substrate such that only the tip end was in contact with substrate), it was a sub optimal feature since future designs with multiple channels will not tolerate crosscontamination between samples. One way to solve this problem is to modify the wetting properties of the surface, i.e., to make the surface inside the reservoir and channel more hydrophilic than the outside surface. This work is currently underway. In summary, the experimental results presented here demonstrate that our ®rst generation SPTs were capable of forming high quality biomolecular arrays with micronsized spots. Further, the designs facilitated sample replenishment at the tip, and allowed printing of large numbers of spots before (or precluding) sample reloading. The typical spot size routinely produced with these SPTs was about 2±3 mm. The occurrence of occasional sub-micron spots suggest that with further development this approach will be able to routinely produce nanoscale features. We believe that spot diameter is correlated with the width of the gap and the curvature radius of the two fork arms in these initial designs. Since conventional UV photolithography has a fabrication limitation of about 1 mm, the gap size and fork arm curvature radius of these tools are the minimum size we can reach using the relatively low cost, simple approach presented here. However, using high-resolution lithography techniques we believe the feature size on these devices can be reduced signi®cantly, permitting routine production of spot sizes in the sub-micron region. An additional parameter that will be further explored is the chemical nature of the deposition surfaces, the rational design of which should facilitate control of deposition domain size by minimizing undesirable surface wetting and liquid ¯ow. Conclusions We have successfully designed and fabricated quill-type cantilever-based SPTs for patterning macromolecules in micron and potential sub-micron domains. The fabrication process for these tools was designed to be cost ef®cient and robust with a high yield of usable devices per microfabrication run. Our ®rst generation SPTs generated high quality biological arrays with routine spot sizes of 2±3 mm. Several thousand spots were printed without reloading, indicating that the reservoir/microchannel system holds promise. We believe the minimum spot size with this type of SPT was mainly limited by its gap width. If a Microfabricated Quill-Type Surface Patterning Tools higher-resolution lithography technique is used, both gap and spot size can be reduced to submicron dimensions. To further improve sample loading ef®ciency and deposition control, cantilever surface modi®cation is needed to better con®ne sample ¯uid inside the microchannel and reservoir. All of these improvements are being addressed in the next generation of tools currently being designed and fabricated. Acknowledgments This work is supported by a grant from the Department of Health and Human Services and Public Health Services (R43-EB000613) and the Army BCRP fellowship DAMD17-00-1-0362 to JLH. The authors would like to thank Prof. Ding and staff members at the Institute of Micro/Nanometer Sciences and Technology (Shanghai JiaoTong University, China) for fabrication of the surface patterning tools. References 1. N.A. Amro, S. Xu, and G. Liu, Langmur 16, 3006±3009 (2000). 2. P. Belaubre, M. Guirardel, G. Garcia, J.B. Pourciel, V. Leberre, A. Dagkessamanskaia, E. Trevisiol, J.M. Francois, and C. Bergaud, Appl. Phys. Lett. 82(18), 3122±3124 (2003). 123 3. H. Fujita, H. Toshiyoshi, G. Hashiguchi, and Y. Wada, RIKEN Review 36(12) (2001). 4. M.J. Heller, Ann. Rev. Biomed. Eng. 4, 129±153 (2002). 5. E. Henderson and C. Mosher, US Patent No. 6,573,369 (2003). 6. M. Jaschke and H.J. Butt, Langmuir 11(4), 1061±1064 (1995). 7. K. Kakushima, M. Mita, D. Kobayashi, G. Hashigushi, J. Endo, Y. Wada, and H. Fujita, IEEE MEMS Proceedings (Interlaken, Switzerland, 2001) p. 294. 8. M.T. Kim, Thin Solid Films 283(1), 12±16 (1996). 9. A. Lewis, Y. Kheifetz, E. Shambrodt, A. Radko, E. Khatchatryan, and C. Sukenik, Appl. Phys. Lett. 75(17), 2689±2691 (1999). 10. C. Liu and R. Gamble, Sensors and Actuators A 71, 233±237 (1998). 11. A. Meister, S. Jeney, M. Liley, T. Akiyama, U. Staufer, N.F. DeRooij, and H. Heinzelmann, NanoTech. (Montreux, Switzerland, 2002) p. 26. 12. A. Meister, S. Jeney, M. Liley, T. Akiyama, U. Staufer, N.F. De Rooij, and H. Heinzelmann, Proc. MNE'02 (NEF-9/356, 2002). 13. G. Meyer and N. Amer, Appl. Phys. Lett. 53(12), 1045±1047 (1988). 14. P. Mitchell, Nature Biotech. 3(20), 225±229 (2002). 15. R.D. Piner, J. Zhu, F. Xu, S. Hong, and C.A. Mirkin, Science 283, 661±663 (1999). 16. J.A. Veerman, A.M. Otter, L. Kuipers, and N.F. Van Hulst, Appl. Phys. Lett. 72(24), 3115±3117 (1998). 17. P. Wagner, M. Hegner, P. Kernen, F. Zaugg, and G. Semenza, Biophys. J. 70, 2052±2066 (1996). 18. Y. Yin, B. Gates, and Y. Xia, Adv. Mat. 12(19), 1426±1429 (2000). 19. M. Zhang, D. Bullen, S. Chung, S. Hong, K. Ryu, Z. Fan, C.A. Mirkin, and C. Liu, Nanotechnology 13, 212±217 (2002).