J. Phys. Chem. B 2006, 110, 23315-23320
23315
High-Efficiency Stepwise Contraction and Adsorption Nanolithography Li Tan,†,‡ Zhenqian Ouyang,§ Maozi Liu,† John Ell,† Jun Hu,†,§,| Timothy E. Patten,† and Gang-yu Liu*,† Department of Chemistry, UniVersity of California, One Shields AVenue, DaVis, California 95616, Department of Engineering Mechanics and Nebraska Center for Materials and Nanoscience, UniVersity of Nebraska, Lincoln, Nebraska 68588, Nanobiology Laboratory, Bio-X Life Science Research Center, College of Life Science and Biotechnology, Shanghai JiaoTong UniVersity, Shanghai 200030, P. R. China, and Shanghai Institute of Applied Physics, Chinese Academy of Sciences, P.O. Box 800-204, Shanghai 201800, P. R. China ReceiVed: May 17, 2006; In Final Form: September 4, 2006
A new miniaturization protocol is demonstrated using stretching and relaxation of an elastomer substrate. A designed microstructure is formed on the stretched substrate and subsequently becomes miniaturized when the substrate relaxes. More importantly, the miniaturized structures can be transferred onto a new substrate for further miniaturization or can be utilized as stamps for nanolithography of designated materials. As an example of this approach, an elastic mold was first cast from a Si mold containing periodic line arrays of 1.5-µm line width. Upon relaxation, line width is reduced to 240 nm. The new elastomer may be used as stamps for micro- and nanofabrication of materials such as proteins. The polymer surface roughness or wrinkling behavior at nanoscale is found to follow classic stability model in solid mechanics. This observation provides means to design and control the surface roughness to meet specific requirements.
Introduction Patterning materials on a micro- or nanometer scale is required in many scientific and engineering fields, such as gratings for wavelength-tunable laser generation in optics,1 narrow gate width transistors in electronics,2 and portable biosensors.3 Requirements for these applications include not only active components with small dimensions but also a large pattern density to reach high-yield and high-device efficiency. Photolithography continuously has demonstrated the potential to miniaturize feature sizes,4 however, the high cost of the equipment and the difficulties in fabricating various materials have limited its broad application. High-energy beam lithography,4 including electron beam and deep UV, and scanning probe lithography5-7 are the methods of choice for fabricating sub100 nm patterns in laboratories. However, these methods suffer the drawback of low throughput, and thus widespread usage is limited. Alternative techniques such as microcontact printing8-11 and imprint techniques12,13 generally share the characteristics of cost-effectiveness and versatility. Production of nanofeatures is still difficult at present as they are well-known as the 1× techniques with limited tunability on both feature size and density. Recently, we developed a new nanofabrication protocol, dubbed stepwise contraction and adsorption nanolithography (SCAN),14 to achieve both pattern miniaturization and pattern density increase (Figure 1A). The fabrication procedure includes: (i) the design and production of an original microstructure (P1) on a stretched rubber surface; (ii) the miniaturization * Author to whom correspondence should be addressed. Phone: 530754-9678. Fax: 530-754-8557. E-mail:
[email protected]. † University of California. ‡ University of Nebraska. § Shanghai JiaoTong University. | Chinese Academy of Sciences.
of the pattern using the following cycle: relaxing the elongated rubber to form a smaller pattern P2 and transferring the new structure to another stretched elastomer through contact; (iii) repeating this cycle until the final pattern size, Pn, is obtained; and (iv) the final transferring of the pattern, Pn onto desired surfaces. This technique allows for the fabrication of reduced multicomponent micro- or nanometer-sized patterns through repeated elastic substrate contraction and patterned structure adsorption.14 This method was used successfully to fabricate one-dimensional multicolored, high-density line arrays (with 1588 lines per inch), monocomponent line gratings (with line width less than 40 nm), and two-dimensional high-density arrays of dots (with 106 spots per square inch). Despite the successes of SCAN, the task of transferring patterned functional materials from one elastomer surface to another (step 3, Figure 1A) was not trivial. The main limitation is the requirement of repeated material transfer from one elastomer and another, which often leads to sacrifice of pattern fidelity, for example, broken features and nonsharp edges. We report a new procedure to avoid the primary difficulty of SCAN, that is, the requirement of repeated material transfer. A master pattern is imprinted into a stretched substrate elastomer containing a film of uncured elastomer. The top film is cured, and then the substrate elastomer is relaxed to reduce the feature size at the top. Replication of this substrate yields a new master that can be used in further reduction steps. We designate this new approach as high-efficiency stepwise contraction and adsorption nanolithography (hSCAN). Similar to soft lithography or imprint lithography,10 structures fabricated using the hSCAN platform can be used as a mold and can be applied to print various materials with further reduced pattern dimensions and increased pattern density. In other words, the miniaturization is accomplished simply through mold engineering, and material transfer only occurs at the final stage by stamping.
10.1021/jp0630323 CCC: $33.50 © 2006 American Chemical Society Published on Web 10/12/2006
23316 J. Phys. Chem. B, Vol. 110, No. 46, 2006
Tan et al.
Figure 1. Schemes of (A) SCAN (stepwise contraction and adsorption nanolithography) and (B) hSCAN (high-efficiency stepwise contraction and adsorption nanolithography).
Experimental Section Materials. Chemicals were purchased from Sigma-Aldrich and were used as received unless otherwise stated. Elastic substrates were cast from poly(dimethylsiloxane) (PDMS) kit (Dow Corning Silastic L RTV), and all pre-elastomer coatings used in mold replication were Sylgard 184 PDMS. The property of PDMS materials is well-known.10,15,16 Elastic films are usually stretched to 300-400% elongation. Stretching and relaxing of the elastic films are manually performed at room temperatures and the rate of such deformation is estimated to be 20 mm/s. Bovine serum albumin (BSA, Aldrich) was first dissolved in water (10 mg/mL) before drop-coated on PDMS stamps to make protein patterns. Mica surfaces (Mica New York Corp., clear ruby muscovite) were prepared by cleaving prior to casting or material deposition. Micro- and nanostructures were characterized using atomic force microscopy (AFM) (contact mode, MFP-3D, Asylum Research Corp.). Elastic Substrate Preparation. The elastic substrates were prepared by vigorous mixing of a 10:1 ratio of part A (5.0 g) and part B (0.5 g) (Dow Corning Silastic L RTV) in a plastic beaker followed by degassing in a vacuum desiccator. The mixture was then spread on microscope slides and was cured at 65 °C for 24 h. Cured elastic thin films were then cut into long strips (4.0 × 1.0 cm) and were used as a substrate for soft mold replication. Si Mold Preparation. A Mikromasch TGZ03 calibration grating (rectangular SiO2 steps on a Si wafer; a step height of (496-503) ( 6 nm, a 3.0-µm pitch, and an active area of 3.0 × 3.0 mm) was functionalized to decrease the interfacial surface tension required for the hSCAN process. The grating was first sonicated for 15 min in a series of washings with DI H2O, ethanol, and hexanes and was then O2 plasma etched (Technics Micro-RIE Series 800) at 180 W for 2 min. Subsequently, the Si mold was immediately immersed in a 20 mM solution of 1H,1H,2H,2H-perfluorodecyltrichlorosilane (FDTS) (0.116 g, 0.200 mmol) in 10 mL hexanes for 10 min with the patterned face down. The Si mold was then washed immediately with copious amounts of hexanes and was stored until subsequent use. Poly(acrylic acid) Solution Formulation. Poly(acrylic acid) (0.5 g, 0.001 mmol, Mv ) 4.5 × 105 g/mol) and 3 mL of ethanol were added to a scintillation vial that was sonicated for 1 h. The solution was further stirred and was sonicated for 30 min
and was allowed to sit for 24 h. New solutions were created every 3 months for consistent results. Results and Discussion 1. Basic Procedures of hSCAN. The strategy of hSCAN in dimension miniaturization and nanostructure formation is shown in Figure 1B. In this process, a synthetic polymer with high elasticity was used as the substrate and was coated with a thin layer of noncured elastomer precursors. A rigid Si-mold with designed microfeatures was brought into contact with this coated substrate (Figure 1B, step 1), and then this whole assembly was baked to cure the elastomer precursor layer (Figure 1B, step 2). Finally, the rigid mold was peeled off and the elastic substrate was allowed to relax, which resulted in an elastic mold (P1) with laterally miniaturized features (Figure 1B, step 3). Further replication of the P1 mold into rigid structures (P1′) allowed the cycling of this hSCAN process to form even smaller feature sizes (Figure 1B, steps 4 and 5). In principle, nanostructures with a feature size of Pn ) P1 × R-n can be reached by repeating the stretching-shrinking cycle multiple times (i.e., n times), where R is the contraction ratio of the elastomer. The SCAN process (Figure 1A) utilizes transfer of functional material along with its pattern to achieve a successive miniaturization. In the hSCAN process, however, miniaturization does not involve material and pattern transfers. Instead, the elastomer molds themselves undergo miniaturization, and the production of functional patterns occurs at the final step in which the elastomer is used as a stamp. Thus, hSCAN can be used to create patterned materials with high degrees of efficiency, control, and reliability. 2. Proof-of-Concept Experiments for hSCAN. Figure 2 outlines an experiment used to demonstrate the feasibility of hSCAN, where the key feature sizes are summarized in Table 1. A PDMS substrate with a thickness of 0.2 mm was stretched to 400% elongation before coating with a thin layer (thickness ∼ 0.1 mm) of the PDMS elastomer precursor. PDMS was chosen as the elastic material for both the substrate and topographic layers, because the resulting elastomer has a good elasticity and can be stretched 300-400% elongation without cracking or failure.17 The miniaturization procedure was performed as described above and topographic images of the resulting patterns after curing and substrate relaxation are given in Figure 2. The Si mold was first treated with 1H,1H,2H,2Hperfluorodecyltrichlorosilane to reduce interfacial tension18 and
High-Efficiency Adsorption Nanolithography
J. Phys. Chem. B, Vol. 110, No. 46, 2006 23317
Figure 2. AFM topographic images and corresponding cursor profiles of a Si grating mold (A and B), a primary PDMS mold, P1 (C and D), and a secondary PDMS mold, P2 (E and F).
TABLE 1: Key Feature Sizes Produced in the hSCAN Experiment Shown in Figure 2 trench width plateau width pitch/period feature height (nm) (nm) (nm) (nm) Si-mold P1 P2
1500 400 240
1500 800 260
3000 1200 500
500 250 80
to allow the mold to be released after the curing of the PDMS pre-elastomer. Figure 2A shows a topographic image of the Si grating, where the corresponding cursor profile (Figure 2B) reveals the dimension of flat 1.5-µm wide plateaus spaced 1.5µm apart. Intimate contact between the mold and pre-elastomer was achieved by applying an external pressure of ca. 40 KPa. Subsequently, the whole assembly was baked at 65 °C for 2 h to fully cure the PDMS pre-elastomer coating (Figure 1B, step 2).10 Finally, the Si mold was removed, and the PDMS substrate was allowed to relax, forming a PDMS replica (P1) with a negative pattern of the Si grating. The P1 mold has a line width and spacing of 800 and 400 nm, respectively, and an overall pitch size of 1.2 µm (see Figure 2C,D). We observed (Figure 2C) that the line width and the separation of the original Si mold were not miniaturized by the same ratio after substrate relaxation: the trenches and plateaus were reduced 3.8 and 1.8 times, respectively. This difference is most likely due to a
variance of topographic layer compression from the substrate layer upward. Provided that adhesion between the substrate and topographic layers is good, the contraction of the topographic layer closest to the substrate layer must be the same as that of the substrate layer. Because the volume of PDMS in the topographic layer must be conserved, its contraction is relaxed by increasing the amount of the distance from the interface of the layers. This relaxation creates a small but measurable “bulging” of the features which translates into less miniaturization for features further from the interface (i.e., plateaus) than those closer (i.e., trenches) to the substrate. Consequently, we would predict that hSCAN features miniaturized in more steps of smaller reduction increments would more faithfully reproduce the shape of the master than hSCAN features miniaturized in fewer steps of larger reduction increments, as the former procedure would show less pronounced “bulging” distortions at each step. The successive reduction of the dimension on P1 is made possible by replicating the patterns in P1, as shown in Figure 1B (steps 4 and 5), using a rigid, thermoplastic material to form a new master, P1′. Furthermore, a thermoplastic with strong hydrophilicity, such as poly(acrylic acid) (PAA), would be released easily from the secondary mold (P1) after casting. Thus, P1′ (step 5, Figure 1B) can be used for subsequent tertiary mold
23318 J. Phys. Chem. B, Vol. 110, No. 46, 2006
Figure 3. AFM topographic images of patterned BSA patterns from the PDMS elastomer mold (P1), which was characterized in Figure 2C. Arrays of BSA lines were produced at 160 kPa (A) and 40 kPa (B), respectively.
(P2) casting to reach even smaller pitch sizes. During the P1′ mold casting process, a PAA solution in ethanol (2.0 wt %) was spread over the P1 mold surface, and the solution was evaporated under a stream of nitrogen producing a PAA mold with negative features of the P1 stamp (Figure 1B (step 4)). The PAA mold was then used to create the tertiary PDMS mold (P2) that upon relaxation yielded a stamp which laterally reduced the dimensions of the line features. Specifically, the line width was reduced from 800 nm (line pitch size of 1.2 µm, Figure 2C) to 240 nm (line pitch size of 500 nm, Figure 2E). Continued replication by following steps 4-5 on the P2 mold is possible and should reduce the pattern dimension to even a smaller scale. In the aforementioned contraction process, very shallow topography features (a few hundred nm or 10-7 m in depth) are formed atop a thick PDMS cover layer (∼0.1 mm or 10-4 m); we envision a uniform thickness increase outward from the elastic substrate and, accordingly, little change on feature depth during the elastic substrate contraction. However, feature depth data in Table 1 suggests otherwise, where depth loss between each generation of molds is clear. We infer that the interfacial dewetting of PDMS from a hydrophilic PAA surface (step 1), or vice versa (step 4), during the pattern copying process might have prohibited a faithful replication in feature depth. This possibility is also supported by previous reports19-21 on shortfalls of PDMS in feature replication and can be improved by selection of less hydrophilic thermoplastics, by an electric-field assisted replication,19 or by replacing PDMS with innovated elastic materials.20
Tan et al. A distinct difference between hSCAN (Figure 1B) and the previously reported SCAN platform (Figure 1A) is that there is no material transfer involved throughout the multistep miniaturization. Thus, complications due to inefficient transfer are avoided. Similar to the microcontact printing (µCP), the fabricated elastic stamp (P2) from the hSCAN platform can be applied to print various materials, with miniaturized pattern dimensions and an increased pattern density relative to current µCP techniques. 3. Production of Protein Nanostructures Using hSCAN. Protein patterns were produced using the elastic mold (P1) characterized in Figure 3A. A drop of BSA (0.1 wt %) solution was deposited on elastomer mold (P1) (dimension characterized in Figure 2C,D). The aqueous solution formed a thin layer on the surface and then vaporized under a stream of nitrogen. Since the PDMS mold and the elastic substrate underneath were very thin, a rigid glass slide was used as support to avoid stamp distortions in the subsequent printing step. The shrinking process of PDMS in the lateral direction changed the contour of the protrusions, for example, the cross-sectional profile showed a shape change from the original square-wave to Lorenzian-shape (described in section 2 as “bulging”). This feature, along with the softness of PDMS, adds a bonus to hSCAN, that is, the mold contact area in the final material transfer process is regulated by varying the contact pressure. Consequently, the BSA coated elastomer (P1) was brought into contact with a glass support. BSA line arrays with a width of 500 (Figure 3A) and 200 nm (Figure 3B) were obtained at an external pressure of 160 and 40 kPa, respectively. The protrusions of the elastomer were flattened to a defined size at given contact pressure, which was utilized to control the line width. The inhomogeneous local flattening and incomplete material transfer are likely to be responsible for the observed saw-tooth-shaped edges in Figure 3A,B. The protein molecules form near a monolayer coverage in both cases. 4. Control Surface Roughness during hSCAN. In the relaxation of an elongated elastic thin film, the length reduction in the x-direction is nearly always22 accompanied with an expansion in the z- and y-directions.23 Very little wrinkling or wavelike features on surfaces are expected for a homogeneous elastomer, that is, no casting material on the surface. However, when a different material covers the elongated substrate, the upper layer may restrain vertical deformation in the z-direction of the substrate. This constraint introduces localized bulging or wrinkling. The morphology of the casting layer in hSCAN was found to follow the classical theory in solid mechanics in terms of deformation or formation of wrinkles. This simple calculation allows us to predict and control the wrinkling by selecting designated polymer materials and casting thickness. Quantitative analysis of surface roughness follows. The extent of contractioninduced restraint can be predicted using Volynskii et al.’s analysis24 of elastic stability theory in solid mechanics. Generally, this restraint is dependent upon the mechanical properties of the upper and bottom layers, as well as the amount of lateral deformation. If the amount of lateral deformation (x) on the elastic substrate falls into the range of a linear deformation (c) described by eq 1, the restraint can be ignored and this doublelayer structure will contract uniformly in the direction of z-axis. In eq 1, c stands for the critical strain limit for a linear deformation; Ef and Es stand for Young’s modulus of the upper film and bottom substrate, respectively; υf and υs represent Poisson’s ratio of each film. Because similar elastic materials (Ef ) Es and υf ) υs ) 0.5) are used for both layers in this
High-Efficiency Adsorption Nanolithography
J. Phys. Chem. B, Vol. 110, No. 46, 2006 23319 (step 2), a careful control of the upper layer thickness is necessary and the aforementioned theory can be utilized to minimize the pattern distortion because of the wrinkling. This can be attained by carefully manipulating the extent of deformation on the elastic substrate to be below the critical strain limit (c), and thus a uniform contraction will be guaranteed throughout the entire hSCAN process. Alternatively, the thickness can be tuned to ensure that the wavelength in wrinkles is much greater than the pitch size of features. In this case, periodic wrinkling features themselves may be used for nanolithography, although this research direction is beyond the scope of this paper. Summary
Figure 4. Controlling of local surface roughness in the two component polymers. (A) Wave generation depends on stretching ratio during relaxation of an elongated substrate; (B) wavelength depends on the thickness of cover layer (inset: illustration of the cover layer structure on elongated substrate); and (C) AFM images of elastomer surface with cover layer at three selected thickness points as indicated in B. Scale bar is 20 µm.
process, the critical strain limit for a linear deformation can be estimated at 79%. However, when the elastic substrate undergoes large deformation, as is the case in aforementioned experiments (x > 200%), the restraint will appear as a periodic height variation in the z-direction, resulting in wrinkling on the surface.24,25 This process is visualized in the data shown in Figure 4A, where wavelength of wrinkles is consistent with the model expressed in eq 2:
c )
x
9Es2
3
64Ef2 (1 - υf2)(1 - υs2)2
λ ) 2πh
x 3
(1 - υs2)Ef
3(1 - υf2)Es
(1)
(2)
in which h stands for the thickness of the upper elastic layer. The wavelength is quantified from the topographic images of AFM. The correlation between the wavelength and casting layer thickness is shown in Figure 4B. The thinner upper layer gives rise to a shorter wavelength in wrinkles (Figure 4C). The presence of this wrinkling feature may introduce additional topographical background to patterned upper layers. If the upper layer is not composed of a uniform layer but of a layer with height contrasts such as the grating mold shown in Figure 1B
A new miniaturization protocol, referred to as high-efficiency stepwise contraction and adsorption nanolithography (hSCAN), is demonstrated using stretching and relaxation of an elastomer substrate. A thin film of elastomer precursor is first cast on the substrate in the stretching phase. A designed microstructure is then formed on the stretched precursor covered substrate, and subsequently the precursor layer is allowed to cure. The microstructure becomes miniaturized when the substrate relaxes. The miniaturized structures can be used as molds or stamps to transfer materials onto a designated support or to produce the structure on another stretched substrate for further miniaturization. The proof-of-concept experiments were demonstrated by miniaturizing microstructures on PDMS substrate into 240-nm feature size and by utilizing those features for production of protein nanostructures. During optimization of surface roughness and functionality for pattern transfer, we found that the polymer surface roughness or wrinkling behavior at nanoscale follows classic stability model in solid mechanics. This observation provides means to minimize surface roughness for hSCAN application or to produce near periodic nanostructures. Acknowledgment. This work was supported by the National Science Foundation (NER-DMI-0304345 and a SEED grant from the CPIMA at Stanford University) and by the University of California, Davis. M. Liu acknowledges the Tyco Electronics Foundation for a graduate fellowship in functional materials. References and Notes (1) Lawrene, J. R.; Turnbull, G. A.; Samuel, I. D. W. Appl. Phys. Lett. 2003, 82, 4023-4025. (2) Arias, A. C.; Ready, S. E.; Lujan, R.; Wong, W. S.; Paul, K. E.; Salleo, A.; Chabinyc, M. L.; Apte, R.; Street, R. A.; Wu, Y.; Liu, P.; Ong, B. Appl. Phys. Lett. 2004, 85, 3304-3306. (3) Su, M.; Li, S. Y.; Dravid, V. P. J. Am. Chem. Soc. 2003, 125, 9930-9931. (4) Levenson, M. D. Phys. Today 1993, 46, 28-36. (5) Xu, S.; Miller, S.; Laibinis, P. E.; Liu, G. Y. Langmuir 1999, 15, 7244-7251. (6) Liu, G. Y.; Xu, S.; Qian, Y. L. Acc. Chem. Res. 2000, 33, 457466. (7) Piner, R. D.; Zhu, J.; Xu, F.; Hong, S. H.; Mirkin, C. A. Science 1999, 283, 661-663. (8) Wilbur, J. L.; Kumar, A.; Kim, E.; Whitesides, G. M. AdV. Mater. 1994, 6, 600-604. (9) Kumar, A.; Abbott, N. L.; Kim, E.; Biebuyck, H. A.; Whitesides, G. M. Acc. Chem. Res. 1995, 28, 219-226. (10) Xia, Y. N.; Whitesides, G. M. Annu. ReV. Mater. Sci. 1998, 28, 153-184. (11) Kane, R. S.; Takayama, S.; Ostuni, E.; Ingber, D. E.; Whitesides, G. M. Biomaterials 1999, 20, 2363-2376. (12) Chou, S. Y.; Krauss, P. R.; Renstrom, P. J. J. Vac. Sci. Technol., B 1996, 14, 4129-4133. (13) Bailey, T. C.; Johnson, S. C.; Sreenivasan, S. V.; Ekerdt, J. G.; Willson, C. G.; Resnick, D. J. J. Photopolym. Sci. Technol. 2002, 15, 481486. (14) Ouyang, Z. Q.; Tan, L.; Liu, M.; Judge, O.; Zhang, X.; Li, H.; Hu, J.; Patten, T. E.; Liu, G. Y. Small 2006, 7, 884-887.
23320 J. Phys. Chem. B, Vol. 110, No. 46, 2006 (15) Wu, T.; Efimenko, K.; Genzer, J. Macromolecules 2001, 34, 684686. (16) Wu, T.; Efimenko, K.; Genzer, J. J. Am. Chem. Soc. 2002, 124, 9394-9395. (17) Zhu, X. Y.; Mills, K. L.; Peters, P. R.; Bahng, J. H.; Liu, E. H.; Shim, J.; Naruse, K.; Csete, M. E.; Thouless, M. D.; Takayama, S. Nat. Mater. 2005, 4, 403-406. (18) Tan, L.; Kong, Y. P.; Pang, S. W.; Yee, A. F. J. Vac. Sci. Technol., B 2004, 22, 2486-2492. (19) Kim, D. H.; Lin, Z. Q.; Kim, H. C.; Jeong, U.; Russell, T. P. AdV. Mater. 2003, 15, 811 ff.
Tan et al. (20) Rolland, J. P.; Hagberg, E. C.; Denison, G. M.; Carter, K. R.; De Simone, J. M. Angew. Chem., Int Ed. 2004, 43, 5796-5799. (21) Goh, C.; Coakley, K. M.; McGehee, M. D. Nano Lett. 2005, 5, 1545-1549. (22) Lakes, R. Science 1987, 235, 1038-1040. (23) Russell, T. P. Science 2002, 297, 964-967. (24) Volynskii, A. L.; Bazhenov, S.; Lebedeva, O. V.; Bakeev, N. F. J. Mater. Sci. 2000, 35, 547-554. (25) Volynskii, A. L.; Bazhenov, S.; Lebedeva, O. V.; Ozerin, A. N.; Bakeev, N. F. J. Appl. Polym. Sci. 1999, 72, 1267-1275.
