Fabrication of aligned microstructures with a single elastomeric stamp

Abstract

The fabrication of complex patterns of aligned microstructures has required the use of multiple applications of lithography. Here we describe an approach for microfabrication that encodes the two-dimensional spatial information of several photomasks onto a single elastomeric stamp by mapping each photomask onto distinct heights on the surface of the stamp. Pressing the stamp against a surface collapses the topography of the stamp such that each recessed layer contacts the surface in stepwise sequence; the greater the applied pressure, the larger the area of the stamp that contacts the surface. After contact of each new layer with the surface, we use techniques of soft lithography (microcontact printing, microfluidics, and patterning through membranes) to pattern the surfaces that contact the stamp and those that do not with inorganic, organic, or living materials. Microfabrication through the use of multilevel stamps provides a promising alternative to conventional lithography for the construction of multicomponent, aligned surfaces; these structures may find use as components of microfluidic devices or biological patterns.

Figure 1

Schematic outline of patterning with a five-level polydimethylsiloxane (PDMS) membrane. (A) The surface of the stamp consists of five distinct sets of regions, each located at a unique height. (B) In the top view of the stamp pattern, lighter shades indicate less recessed regions (Left). We label features that are not indented collectively as {1}, features that have the smallest indented depths as {2}, and so on. We denote the depths of the regions {I} by d_I (d₁ = 0). As in this example, holes in PDMS membranes are considered to have infinite depth and are denoted by {N} in an N-level membrane. (Right) In the series, black regions indicate contact with the substrate; when the stamp is placed on a surface and pressure is applied, the portions of the stamp that contact the surface are {1}, followed by {2}, and so on. The application of soft lithographic techniques (processes I, II, III, and IV) at each change in the masking pattern generates a distinct combination of processes experienced by each of the five regions on the substrate.

Figure 2

Micrographs of aligned patterns. (A) Inking a four-level stamp (d₂ = 1 μm, d₃ = 2 μm, d₄ = 50 μm) with three fluorescently labeled proteins and stamping onto a glass slide generated a pattern of those proteins. Each stage of compression prints onto the substrate blue-, green-, and red-labeled proteins that had been adsorbed onto {1}, {2}, and {3}/{4}, respectively. (B) Flowing solutions of proteins in channels between a three-level stamp (d₂ = 2 μm, d₃ = 50 μm) and a glass slide created a grid of proteins. First, we flowed a solution of red-labeled protein between the stamp and substrate in the open channels of {2} and {3}; the channels consisted of a grid of lines. We then washed the channels with PBS, collapsed {2} onto the slide, and flowed a solution of green-labeled protein through {3}; the channels were a parallel set of lines. The second protein (a rabbit anti-mouse antibody) bound to the first protein (a goat anti-rabbit antibody), yielding a region that fluoresces yellow. In the schematics that accompany the image, channels are represented by white regions. (C) A combination of etching and evaporation through a three-level membrane (d₂ = 7 μm) generated a pattern of silver and gold on a glass slide. We placed a membrane (≈150 μm thick) on a ≈50-nm-thick film of silver on glass and etched the exposed silver on {2} and {3} with a ferricyanide solution (4). We then rinsed and dried the sample and evaporated ≈100 nm of gold through the holes in the membrane onto {3}.

Figure 3

Multilevel patterning in continuous and reversible formats. (A) Flowing solutions of proteins under a stamp with corner-cube topology resulted in deposition of an arbitrary gradient onto a Petri dish. We placed the stamp onto a dish, applied moderate pressure to the back of the stamp, and flowed in a solution of BSA. Reducing pressure to the stamp while flowing in successively more dilute solutions of green-labeled protein (from 50 to 10 μg/ml in PBS) generated a gradient of protein next to the BSA-coated regions. We then flowed in a solution of BSA to passivate the uncovered surface, flowed in a solution of red-labeled protein, and allowed the stamp to lift from the surface. (B) Fabrication of an addressable array of wells. Here d₂ = 8 μm and d₃ = 50 μm; an external fence prevents {1} from contacting the substrate without compression. A sequential combination of compressions and flows trapped liquid selectively in certain wells: after partial compression of the stamp, a solution of Malachite green dye was captured in the wells enclosed by {1}; the fluid in the channel was replaced by a solution of Congo red; and further compression of the stamp captured this solution into the second array of wells. The remaining solution was washed out with water, and an image was captured. A second image was taken after the wells holding the solution of Congo red were unsealed, washed with water, and resealed, and fresh solution of Congo red was introduced into the channel.

Figure 4

Fabrication of patterned cellular cocultures. (A) Possible permutations of patterns mapped onto a multilevel stamp to build a coculture. The first row depicts a set of possible patterns generated by layer {1} placed against a substrate, and the second row depicts a set of possible patterns generated by compressing layer {2} onto the substrate. Each arrow indicates a choice of two patterns that could be used to generate a coculture. The dotted line delineates the patterns used in B. (B) Cocultures of labeled NRK cells and fibroblasts. The image combines fluorescence and phase-contrast views of NRK cells (red) and fibroblasts (green). The three-level membrane (d₂ = 1 μm, membrane thickness = ≈150 μm) was placed against a Petri dish, masking {1}. We coated regions corresponding to {2} and {3} with fibronectin, collapsed {2} of the membrane onto the dish, seeded NRK cells onto {3}, removed the membrane, coated {1} with 0.1% Pluronics F127 in DMEM to render it resistant to cell attachment, and seeded fibroblasts onto {2}. Removal of the membrane did not appear to disrupt the adsorbed layer of fibronectin, because protein does not print onto the hydrophobic stamp (6).