Dry adhesion observed in the feet of various small creatures has attracted considerable attention owing to the unique advantages such as self-cleaning, adaptability to rough surfaces along with repeatable and reversible adhesiveness. Among these advantages, for practical applications, proper detachability is critical for dry adhesives with artificial microstructures. In this study, we present a microstructured array consisting of both asymmetric rectangle-capped tip and tilted shafts, which produce an orthogonal anisotropy of the shearing strength along the long and short dimensions of the tip, with a maximum anti-shearing in the two directions along the longer dimension. Meanwhile, the tilt feature can enhance anisotropic shearing adhesion by increasing shearing strength in the forward shearing direction and decreasing strength in the reverse shearing direction along the short dimension of the tip, leading to a minimum anti-shearing in only one of the two directions along the shorter dimension of the rectangular tip. Such a microstructured adhesive with only one weak shearing direction, leading to well-controlled attachment and detachment of the adhesive, is created in our experiment by conventional double-sided exposure of a photoresist followed by a moulding process.
Because Van der Waals force was found to be responsible for dry adhesion observed in the feet of various small creatures [1,2], biomimetic dry adhesives using micropillars arrays have attracted considerable attention over recent years. These biomimetic adhesives have been found to have unique advantages, including self-cleaning capability, adaptability to rough surfaces along with repeatable and reversible adhesive properties [3,4]. Among these advantages, for practical applications as dry adhesives, proper detachability for these artificial microstructures from various surfaces is critical. Geckos are an excellent example of creatures able to achieve controlled adhesion . When loaded in forward direction, gecko's adhesive toe pads stick with a strong force to the surface, however, they can also be easily peeled off from the surface when loaded in the reverse direction . This directional or anisotropic adhesive behaviour that allows smart or controllable surface attachment and detachment has inspired many applications such as climbing robots , clean transportation systems [8,9] and micro-transfer printing mechanisms . Rapid switching of the gecko's toe pads between attachment and detachment can be attributed to millions of inclined seta shafts and to the corresponding loading response of the toes of the gecko [11–15].
Numerous methods and processes have been reported in literature to emulate this particular anisotropic adhesion produced using inclined micropillar arrays [16–20]. Jin et al. studied adhesion and friction properties of micropillar samples at various tilt angles and found that the tilt angle of the micropillars can lead to significant anisotropic friction. Anisotropy of the static friction force increases as the tilt angle decreases . Lee et al. also observed that an array of tilted microfibres exhibited anisotropic lateral resistance. Using this property, they demonstrated sliding of a clean glass surface against and along the tilted microfibres without applying any external normal force and strong anisotropy for shear stresses was observed . Friction and adhesion forces for tilted rectangular flaps were also investigated by Yu et al. The structures were found to have strong friction and adhesion forces when sliding along the tilt direction, and low friction and adhesion forces when sliding against the tilt direction . However, these artificial adhesives with inclined micropillars demonstrated a relatively lower magnitude of interfacial friction and adhesion. This was because the micropillars did not have a laterally bulged, flat top tip, which has been regarded as an effective mechanism for enhancing adhesion and friction [4,21–24]. To overcome this, Murphy et al. proposed a micrometre-scale structure using an array of mushroom-shaped fibres, with titled stalks and laterally extended tip ends. Using this structure high normal interface strength and high shearing anisotropy was achieved . Jeong et al. presented an approach for fabricating inclined nanohair structures with a bulged tip ending. These nanohair surfaces showed strong anisotropic interfacial strength . The bulged flat tips of the micropillars presented in these attempts were approximately round in their planar shape, and were mostly intended to enhance the interfacial strength itself by an increased actual contact area of the microstructured adhesive on the target surface . As a matter of fact, the shape of the tip endings alone can also affect the anisotropy of the artificial adhesive, as has been discussed in reports presented in the literature by other research groups [25,26]. For example, Kwak et al. presented a new microstructure with a triangular-shaped tip and vertical pillars. These microstructures were found to have anisotropic shear adhesion properties with relatively high normal adhesion. This property can be attributed to the anisotropic nature of the bulged triangular tips . Bin Khaled et al. demonstrated that defects introduced into the overhanging caps could also produce anisotropic adhesion. This implies that the shape and defects of the caps affected both the magnitude and anisotropy of the total adhesion force . Thus, a combination of a tilted shaft and asymmetric cap shape for the micropillars can be a reasonable route to further enhance the anisotropy of the shearing adhesion.
Here, we present a microstructured array with asymmetric rectangle-capped tips and tilted shafts. These structures can produce an orthogonal anisotropy of the shearing strength along the long and short dimensions of the rectangle-capped tip with maximum anti-shearing in the two directions along the longer dimension. Meanwhile, the tilted feature can enhance anisotropic shearing adhesion by increasing shearing strength in the forward shearing direction and decreasing strength in the reverse shearing direction along the short dimension of the rectangular tip. This results in low anti-shearing in only one of the two directions along the shorter dimension of the rectangular tip. Such a microstructured adhesive with only one weak shearing direction, generating well-controlled attachment and detachment of the adhesive, is created in our experiment simply by conventional double-sided exposure of a photoresist followed by a moulding process. To explain the lateral anisotropy of such a biomimetic adhesive, the stress distribution at the contact interface between the micropillar cap and a flat surface is numerically analysed to spot the crack initiation. Meanwhile, owing to the presence of capped tips and a tilt angle, slanted pillars achieve higher normal adhesion strength at lower preload compared with vertical capped pillars and ordinary pillars (without any caps).
2. Experimental section
2.1. Fabrication of capped micropillars
Capped (laterally bulged and flat) micropillars can be fabricated by conventional double-sided exposure of a photoresist followed by a moulding process. The fabrication process flow has been discussed in detail in our previous publication . Figure 1 illustrates the procedure for fabricating rectangle-capped micropillars. The photoresist is spin-coated on a transparent glass slide and then exposed using UV light with a mask with an array of rectangular microholes from the topside at a certain angle θ (as shown in figure 1a). Second, the photoresist is exposed from the bottom using a flood exposure for a short time. This affects a thin layer of the photoresist near the surface of the glass slide. The development process has to be precisely controlled to achieve an undercut microstructure (figure 1b), because the development time determines the lateral undercut of the exposure photoresist at the bottom. Next, PDMS (Sylgard 184, Dow Chemicals) is used to duplicate the structure on the developed photoresist (figure 1c). Finally, rectangle-capped pillars are generated after chemically sacrificing the photoresist (figure 1d). Vertical pillar arrays can be also obtained by controlling the tilt angle θ to be 90° during the first exposure from the top. (A more detailed description of the fabrication method is included in electronic supplementary material, S1.)
Figure 1e,f shows SEM images of the vertical and tilted pillars 60 μm in length with rectangle-capped tips generated for our subsequent experiment. The width (measured on the short dimension) of the tips is about 15 μm, and the total height of the pillars is about 18 μm. Also, the tip cap stretches out of the shaft by 5 μm, and has a thickness of about 3 μm. The tilted pillars have a tilt angle of approximately 60° for their shafts.
2.2. Adhesion measurements
Adhesion performance for these micropillar arrays was tested against a silicon wafer by recording load–displacement curves on a home-built apparatus in two ways: normal and shearing. This measurement apparatus mainly consists of a motorized stage (PI GmbH M-531.DB with a high resolution of 0.1 μm) and an eddy current sensor (Waycon with a high resolution of 30 nm) for detecting cantilever deflection as a response of the adhesive to the force. Data collection was performed with LabVIEW and a sample frequency of 200 Hz. To avoid large deflection angles of the cantilever, a double cantilever beam was used for the measurement. To improve accuracy, adhesion data were measured four times and an average of these four values was reported. The silicon wafer was cleaned with ethyl alcohol and distilled water before the test, and all experiments were carried out in a cleanroom environment.
3. Results and discussion
3.1. Enhanced normal adhesion by slanted capped micropillars
Figure 2a shows the schematic of a silicon wafer with 1 mm2 glued to the free end of the precisely calibrated cantilever. The adhesive sample was placed on a stage with two rotary degrees of freedoms (DOFs), which are needed to compensate for possible misalignment between the silicon wafer and the adhesive sample. During the test for normal adhesion, the sample is first moved in an upward direction at a constant rate of 5 μm s−1 into contact with the silicon surface to provide a compressive preload and is held in position for about 10 s (figure 2b). Then, the motorized stage moves downward at a constant rate of 5 μm s−1 until peeling-off occurs. Results for normal adhesion of these dry adhesives (defined as a normal peeling-off at P in the figure 2b) are shown in figure 2c. Normal adhesion strength of rectangle-capped structures increases with preload owing to an increase in the actual contact area of the adhesive with the silicon. Adhesion strength saturates after the top of the tips for micropillars comes into a full contact with the silicon surface. In comparison, ordinary micropillars without caps show smaller normal adhesion strength, regardless of the preload. This observation indicates that capped micropillars can effectively enhance normal adhesion, which is in agreement with other theoretical and experimental studies [21,22,28,29]. Significantly, increase in normal adhesion for capped micropillars is owing to the laterally bulged tips, which increase the contact area while maintaining flexibility of the structures while also increasing tolerance for tiny defects owing to the overhanging portion of the caps. The contact tip of capped micropillars can reduce or even eliminate edge stress concentration of the stalk , which is useful to prohibit crack initiation of the tip from the silicon surface and crack propagation, leading to an enhanced adhesive strength.
Although, vertical and slanted micropillars with identical rectangle-capped tips have similar saturated adhesion strength at a large preload, slanted micropillars demonstrate better adhesion at low preloads (less than 0.6 N cm−2), compared with vertical micropillars. This observation can be explained using a mechanical model of the micropillar cantilever presented by Autumn et al. . In general, the approximate vertical deflection Δapprox of the micropillar cantilever can be expressed by the following equation: 3.1where E is the elastic modulus, I the area moment of inertia, L the height of the micropillars, F the normal load applied, V the additional shear load and ϕ the tilt angle. Force V is generated when the micropillar cantilever is dragged along the tip surface during compressive preloading and is equal to zero in our experiments where no lateral sliding during preloading was applied. Hence, the equation can be simplified as follows 3.2Obviously, displacement of the micropillar cantilever increases with a decline in the tilt angle ϕ (0 < ϕ < π/2). This results in an improved surface compliance and is desirable for maintaining better contact and enhancing adhesion at low preload. At high preload, full contact of the micropillar tips with the silicon surface can be expected regardless of the tilt angle. This can be used to explain the significant improvement in adhesion of the micropillars with a tilt angle at a low preload.
3.2. Improved anisotropic anti-shearing for slanted capped micropillars
In addition to improved normal adhesion properties, slanted capped micropillars also demonstrate better anisotropic shearing strength compared with their vertical counterparts. As illustrated in figure 3a, a smooth silicon surface with an area of 4 mm2 was held in contact for 30 s with the PDMS dry adhesive with an area of 1 cm2 glued to the stage with two rotary DOF by a normal preload supplied by a digital force gauge, and then fully released from the preload. In our experiment, a preload of 1 N cm−2 was applied, which was large enough to lead to complete contact between the sample and silicon surface, based on the discussion in the normal test section. After preloading, the motorized stage is vertically moved at a constant rate until peeling-off occurs between the sample and silicon surface, which is attached to the calibrated cantilever by a string. In addition, to allow for reasonable comparisons, the actual contact area is used to calculate shear strength.
Strengths along three shearing directions were obtained for these samples to demonstrate their anisotropic shearing adhesion. One direction is parallel to the long dimension of the rectangular tip (denoted as 90° in figure 3d), and the other two opposite directions are along the short dimension of the rectangular tip (denoted as 0° for the reverse shearing and 180° for the forward shearing in the case of titled micropillars, respectively). Measurements were taken of the slanted pillars in the 180° direction; for an example, figure 3b shows typical shearing strength versus time with peeling-off occurring at P after the preload is removed, and figure 3c demonstrates the relationship between shearing strength and test speed. Measured values were found to be more accurate for lower test speeds, because the test system can detect a pull-off occurring at lower speeds more accurately. Shearing strength only increased by 5% from 4.37 to 4.59 N cm−2 with the test speed increasing from 2 to 10 μm s−1, as shown in figure 3c, so speed was set to 5 μm s−1 for the following measurements. Figure 3d shows the test results for two types of adhesives in three directions.
Shearing strength for vertical micropillars was the same in the 0° and 180° directions, which agrees with the structural symmetry in these two directions. Meanwhile, maximum shearing strength comes from the measurement in the 90° direction (along the long dimension of the rectangular tip). This means that the asymmetric tip of the micropillars alone can create anisotropic shearing adhesion. More interestingly, it can be concluded from the test results for slanted micropillars that a tilt angle for the capped micropillars can increase shearing strength in the 180° direction (i.e. the forward shearing), while further decreasing shearing strength in the 0° direction (i.e. the reverse shearing). Hence, the ratio between the maximum and minimum shearing strengths for vertical micropillars is approximately 1.5. However, the ratio reaches up to 2.2 for slanted micropillars. Therefore, enhanced anisotropic shearing adhesion was obtained by using rectangle-capped and tilted micropillars, which tend to have a weak anti-shearing strength along one single weak direction, leading to well-controlled attachment or detachment of the adhesive.
Many earlier reports suggest that the detachment mechanism of the capped structures is due to crack nucleation at the centre of the cap and its subsequent propagation in an outward direction [28,31]. This is in agreement with calculations of normal stress distribution at the contact interface between an adhesive and a flat substrate. For this reason, the anisotropic shearing strength of rectangle-capped micropillars can be explained by their mechanical and deformation behaviour. For modelling, a slanted micropillar was considered as the analysis object as shown in figure 4a. The bottom surface of the adhesive was fixed and the tip was assumed to be in perfect contact with the silicon smooth surface. Load (three directions: 0°, 90° and 180°) was applied to the silicon surface, and the displacement of silicon along the vertical direction was free and the slope angle of its upper surface was zero. Finite-element analysis was carried out using COMSOL v. 4.3 for a quantitative demonstration of the redistribution stress of the cap upper surface when the load was applied. The tilted rectangle-capped pillar model is based on SEM measurements, and boundary conditions were developed to best match the adhesion test method in this work.
As PDMS used in the fabricating of the adhesive is a hyperelastic material, a hyperelastic solid mechanics model is necessary to better account for the possibility of large deformations. Many constitutive models can be used to define the hyperelastic material behaviour and determine the strain energy function such as the Mooney–Rivlin model, neo-Hookean model and Ogden model, etc. [32–34]. Analysis, in this work, was performed using the second-order Ogden model to describe the mechanical behaviour of PDMS. The strain energy function, W, can be described as  3.3where μi and αi are material constants, λ1, λ2 and λ3 are the principal stretches. μi and αi parameters are referenced from the literature  and μ1, μ2, α1 and α2 are 63.49, 0.04 MPa, 6.37 × 10−10 and 3.81, respectively. Stress distribution of the adhesive at the contact interface can be expressed as 3.4where PPK is the first Piola–Kirchhoff stress tensor, n is the normal vector of the contact interface, σz is the stress normal to the contact surface, and τx and τy are the stresses tangential to the upper surface of the cap.
As shown in figure 4b, maximum tensile stress (σzmax) showed a nearly linear increase with increasing applied load for all three loading methods. σzmax is largest when loaded in the 0° direction and is least when loaded in the 90° direction for fixed applied loads. Here, compressive stress (σz < 0) was ignored because it is in the opposite direction to crack initiation and propagation. However, the magnitude of tangential stresses ( for 0° and 180° and for 90°) is considered in the direction of the shear stress, as compressive stress does not exist along these directions. As shown in figure 4c, like the tensile stress, the largest and least (the maximum magnitude of tangential stress) also occurs in the 0° direction and 90° direction, respectively. Taking a 2 μN load as an example, figure 4d,e shows distribution of σz and τx (or τy) at the upper surface of the cap. Based on the analytical result, σzmax was approximately 17.7, 10.8 and 8.5 kPa for the 0°, 180° and 90° directions, respectively. Meanwhile, was 8.2, 7.6 and 4.7 kPa for the 0°, 180° and 90° directions, respectively. Although, an accurate value of the maximum stress for crack initiation leading to adhesion failure is unknown, comparison of σzmax and can explain adhesive detachment, which readily occurs when loaded in the 0° direction. This is because σzmax and are largest in this direction for a fixed applied load. In other words, for larger magnitudes of stress, the probability of crack initiation is higher. Once the crack is formed and extended, adhesion failure will readily occur. Finite-element analysis results were found to be in agreement with experimental data: shear adhesion strengths are least in the 0° direction and largest in the 90° direction.
3.3. Optimization design for anisotropic shearing properties
To further discuss anisotropic shear strength and improve the anisotropic properties or obtain a larger ratio of shear strength between the 0° and 90° directions, other dimensions of slanted rectangular capped posts were taken into account. Figure 5a shows the simulation results of the ratio of the applied load between 90° and 0° for different lengths (along the longer dimension of the rectangle) varying from 30 to 210 μm in increments of 15 μm. Although, maximum normal tensile stress is unknown when pull-off occurs, a structure with better anisotropy by comparing the ratio of the applied load between 90° and 0° can be determine based on when this stress is the same in these two directions. In the simulation, stress is set to approximately 10 000 Pa in the 90° and 0° directions for different lengths, and the ratio of the applied load between 90° and 0° was varied, and is shown in figure 5a. The numbers in figure 5a are the applied load in 90° and 0° directions, and ratio of the applied load between the 0° and 90° directions was found to be largest for the capped-rectangle with a size of approximately 120 μm along the longer dimension. This means that this structure perhaps has better anisotropic shearing strength, and this conclusion agreed with experimental results. As shown in figure 5b, four types of adhesives were used as test samples to compare with simulation results. Lengths along the longer dimension of the capped-rectangle were 30, 60 (as presented in the above), 120 and 210 μm, and the lengths along the shorter dimension are nearly the same, reaching 15 μm (oblique views of SEM images of other three adhesives are included in electronic supplementary material, S2). As shown in figure 5c, the capped-rectangle of 120 μm in the longer dimension (noted C) has the best anisotropic property among the four types of adhesives, and the ratio of the shearing strength between 90° and 0° is largest up to approximately 4.0, which agreed with simulation results. From figure 5c, it can be also concluded that shear adhesion will enhance with decreasing length of the pillar. As the length decreases from 210 to 30 μm, shearing strength in the 90° direction increases from 3.8 to 6.4 N cm−2. This result is agreement with other studies [37,38], which indicated that effectiveness of dry adhesives improves with decreasing pillar diameter. In addition, as length decreases, the cap will play a more efficient role in surface adhesion.
Biomimetic microstructures for generating anisotropic dry adhesion effects consisting of slanted micropillars with rectangle-capped tips are presented in this work. Vertical rectangle-capped micropillars display anisotropic shearing adhesion in two orthogonal directions, along the longer and shorter dimension of the rectangular tip. Whereas the slanted micropillars can enhance anisotropic shearing adhesion by increasing shearing strength in the forward shearing direction and decreasing strength in the reverse shearing direction along the short dimension of the rectangular tip. It was determined from empirical and simulation results that rectangle-capped slanted micropillar arrays have a weak anti-shearing strength only on one single direction, leading to well-controlled attachment or detachment of the adhesive. By varying the length along the longer dimension of the rectangle, the adhesives were found to display the best anisotropic shearing strength when the length was close to 120 μm. In addition, owing to the presence of the capped tips and tilt angle, slanted pillars can attain higher normal adhesion strength at a low preload compared with vertical capped pillars and ordinary pillars (without any caps). The unique normal and shear adhesion capabilities of these biomimetic adhesives presented here would have a broad range of potential applications from climbing robots to biomedical patches.
This work was supported financially by the NSFC Major Research Plan on Nanomanufacturing (grant no. 91323303), NSFC Fund (51175417) and Programme for New Century Excellent Talents in University of Ministry of Education of China (NCET-13-0454).
- Received February 3, 2015.
- Accepted February 26, 2015.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.