Bioinspired Microstructured Adhesives with Facile and Fast Switchability for Part Manipulation in Dry and Wet Conditions

The rapid growth in the miniaturized mechanical and electronic devices industry has created the need for temporary attachment systems that can carry out pick‐and‐place and transfer printing tasks for fragile and tiny parts. Current systems are limited by a fundamental trade‐off between adhesive strength and state‐changing trigger force, which causes the need for a rapidly switchable adhesive. In this study, an elastomeric microstructure is presented combining a trapezoidal‐prism‐shaped (TPS) and a mushroom‐shaped microstructure, which overcomes the trade‐off with the help of the TPS structure. The optimal design exhibits a strong adhesive strength of 87.8 kPa and a negligible detachment strength of <0.07 kPa with a low trigger shear stress of 10.7 kPa on smooth glass surfaces. The large tip‐to‐stem ratio (50 to 20 µm) enhances the suction effect, allowing the microstructure to maintain its adhesive performance even in wet conditions. Pick‐and‐place manipulation tasks of a single and an array of ultralight parts from micrometer to millimeter scales are performed to demonstrate the capability of handling fragile and tiny parts. Moreover, it demonstrates the ability to transfer parts across water and air interfaces. This proposed microstructure offers a facile solution for manipulating microscale fragile parts in dry and wet conditions.


DOI: 10.1002/adfm.202303116
due to their superior attachment performance, [1,2] such as strong yet controllable attachment, [3,4] insensitivity to surface conditions, [5][6][7] and simple actuation mechanisms. [8,9]Inspired by those attachment systems, there have been mainly two most popular bioinspired adhesives: the mushroom-shaped structure inspired by the ladybird beetle's footpad [10,11] and the wedge-shaped structure inspired by the gecko's footpad. [8,12,13]20] In addition, due to the limitations of the fabrication methods, the shape and size of the contact tips were limited, which reduced the attachment performance.On the other hand, although the wedge-shaped or anisotropic adhesive microstructures showed some switchability, [21][22] they had weak attachment forces due to the nature of the peeling process and showed difficulty in handling small parts due to the nature of shear-induced attachment, which requires a pair of adhesives oriented in opposite directions.Also, to maintain the attachment Figure 1.A) Illustration of the developed gecko-inspired micro-structured adhesive with the design principles along the direction of high switchability or strong adhesion (indicated by the arrow).The design is a custom mushroom-shaped structure integrated with a trapezoidal-prism-shaped (TPS) structure, which adds switchability while maintaining the adhesive performance.B) Scanning electron microscope (SEM) images of the positive square-packed bioinspired microstructured adhesives array fabricated by two-photon polymerization-based 3D-microprinted (Figure S1, Supporting Information).C,D) Working mechanisms of the bioinspired adhesive structure with controllable both weak and strong adhesion modes, respectively.
force, a strong tangential force must be maintained while grasping parts, which limits the size of the target parts and possibly damages the parts.
To overcome the shortcomings of dry adhesives in handling small and fragile parts, a variety of controllable attachment systems have been developed and studied with various actuation means.[28][29][30] However, those systems required strong trigger forces for switching adhesive states (in normal or tangential directions) and exhibited poor attachment performance.Hence, other actuation means such as light, [31][32][33] heat, [34,35] swelling, [36] and magnetic field [37][38][39] have been considered to control the adhesion.However, those actuation methods suffered from slow switching rates (10 -100 sec) [40][41][42] or showed difficulty in handling sensitive parts since the target parts are affected by the actuation mechanism. [37,39,43,44]Other attachment mechanisms, such as electroadhesion-based, [45] capillary forcebased, [46] and suction-based soft grippers, [47][48][49] have advantages for grasping larger parts but struggle to minimize their size for smaller parts.[52][53] Neverthe-less, due to the rapid growth of demands for the miniaturization of various mechanical systems and electronic devices in the industry, it is still required to achieve a strong yet gentle and controllable attachment system, which is robust and highly switchable between sticky and non-sticky states with an easy actuation mean.
In this work, we report a new type of adhesive microstructure capable of easily switching between attachment and detachment modes with strong attachment and negligible detachment forces (see Figure 1A).Throughout a full parametric characterization, we experimentally optimize the dimensions of the microstructure based on the trigger force, attachment force, and detachment force.The optimized microstructure exhibits a superior attachment force of 87.8 kPa when sheared in the grasping direction, negligible detachment force of <0.07 kPa (invisible within the noise level of 0.002 mN of the high precision load cell) when sheared in the releasing direction with a trigger force of 10.7 kPa on smooth glass surfaces.The unique working mechanism of the microstructure is thoroughly studied and explained by both simulation and experiment.With an array of the optimized structure, we present various pickand-place demonstrations with ultralight parts from micrometer to millimeter scales, such as mini-LEDs (multiple LEDs at the same time) and ultrathin glass and printed resin (thickness of 30 μm with a width and length of 1 cm) in dry and wet conditions.

Structural Design
As shown in Figure 1A,B, the adhesive structure was designed by adding a trapezoidal-prism-shaped (TPS) structure to the intersection between the tip and the stem of a mushroom-shaped structure on one side in order to achieve the simple switchable adhesion.The concept evolved from the transition from a flat punch-shaped structure to a mushroom-shaped structure, which benefits from both structures: strong attachment from the mushroom-shaped structure and easy detachment from the flat punch-shaped structure. [27]The TPS structure was aimed at enabling the initiation of the edge peeling of the contact element(see Figure 1C), while the mushroom-shaped tip was aimed at achieving a suction effect for the strong attachment (see Figure 1D).Due to the geometric asymmetry from the TPS structure, the controllable adhesion could be achieved by adjusting the degree of the involvement of the TPS structure in the attachment.Thus, to fully understand the effect of the TPS structure on the attachment performance, the structure was characterized by three parameters: the inner height H 1 , the outer height H 2 , and the revolve angle , while keeping the dimensions of the mushroom-shaped structure constant (see Figure 2A).The mushroom-shaped structure has a tip diameter of 50 μm, similar to conventional gecko-inspired mushroom structures, and a tip thickness of 2 μm.This resulted in a large tip-to-stem diameter ratio of 2.5:1 (50 μm vs 20 μm) which is notably a higher ratio than conventional designs.Although the large tip-to-stem ratio could be beneficial for the adhesive performance, [54] a further increase of this ratio will result in larger unsupported structures during fabrication and larger deformation of the tip during the demolding process, which might lead to damage to the structure.Similarly, the current fabrication techniques limit achieving a tip thickness below 2 μm, which might improve the tip conformability and adhesion.The height of the overall structure was 32 μm and the center-to-center distance between the adjacent structures was 55 μm.

Loading Mode
To switch between various adhesion states, three loading modes, 1) load-pull (LP), 2) load-grasp-pull (LGP), and 3) load-release-pull (LRP), were investigated for the microstructured adhesive, as illustrated in Figure 2B.The LP mode yielded the pull-off force F LP , in the conventional manner, involving no drag motion prior to the Pull phase.However, when the drag motion was involved in the process, the pull-off force could be increased or decreased due to the presence of the TPS structure, depending on the direction of the drag motion.When a drag motion was applied against the TPS structure (LGP mode), the pull-off force, F G , could be higher than F LP .In the LGP mode, the initiation of the edge peeling from the TPS structure side was prevented, which ensures the strong attachment from the suction effect of the interior void.On the other hand, when a drag motion was applied toward the TPS structure (LRP mode), the pull-off force, F R , could be significantly reduced since the TPS structure promotes the crack initiation on the edge near the TPS structure, which results in the loss of the suction effect.Hence, the LRP mode could generate a negligible F R (i.e., within the signal noise level).
To investigate the effect of the dimensions of the TPS structure on the performance of the adhesive microstructure, we defined a criterion evaluating both the adhesion and the controllability of the adhesives using a dimensionless index, the adhesioncontrollability ratio, R = F G ⋅ F LP ∕S 2 R .Given that the minimum pull-off force, F R , of the adhesives could be negligible for all proposed designs, the value of F R was not informative.Instead, the tangential force applied in the releasing direction, S R , indicating the ease of achieving negligible pull-off force in the LRP mode in a negative manner, was included in the ratio as a denominator.In addition, both F LP and F G are important factors for evaluating the adhesive performance.F LP is the maximum force that the adhesive could generate to overcome the weight of the part in the air while transferring.F G is the maximum force that the adhesive could generate to overcome the adhesion between the part and the substrate while picking up the part.In this regard, the proposed criterion could comprehensively assess the performance of the adhesives during a complete pick-and-place task, involving all three loading modes.

Parameter Characterization
First, we analyzed the effect of the revolve angle  by varying the angle from 20°to 100°with a 20°increment.As shown in Figure 2C, F G decreased as the angle increased, while both F LP and S R decreased and eventually became stationary when the angle was greater than 40°.Hence, the highest R of 32.5 at the revolve angle of 40°indicated the optimal revolve angle that can maximize the controllability R with respect to the revolve angle.Using the optimal revolve angle of 40°, the characteristics of H 1 and H 2 were analyzed further with 19 experimental points from the combination of H 1 and H 2 .H 1 and H 2 ranged from 0 μm to 20 μm and H 2 was constrained with H 1 as the upper bound due to the fabrication issues for the overhung features.
As shown in Figure 2D, the maximum pull-off force of 3.78 mN in the LP mode was achieved by the adhesive without the TPS structure.F LP decreased monotonically with increasing H 1 (two-way analysis of variance (ANOVA), p = 6.08 × 10 −6 < 0.05) and there was no significant difference in F LP while changing H 2 (two-way ANOVA, p = 0.633 > 0.05).However, both H 1 and H 2 affected F G and S R as shown in Figure 2E,F.The highest pull-off force of 4 mN in the LGP mode was achieved by the structure with H 1 = 5 μm and H 2 = 0 μm.The tangential force needed to trigger the detachment S R decreased as the TPS structure became larger.The low S R of less than 0.35 mN was obtained when H 1 ≥ 10 μm or H 2 ≥ 10 μm.Thus, although a smaller additional TPS structure could achieve a better adhesive performance, it also needed a higher tangential force to trigger the detachment, which counteracts each other.However, considering all the values of F LP ,F G , and S R , R showed an optimal region with respect to H 1 and H 2 as shown in Figure 2G.The maximum R of 32.5 was achieved by the structure with H 1 = 10 μm and H 2 = 7.5 μm, which represents an intermediate size of the TPS structure.Hence, the optimal dimensions of the TPS structure were chosen for further characterization of the adhesive structure.
Figure 2H shows the full loading cycles for the optimal design structure under the three different loading modes.The optimal structure could achieve a pull-off force of ≈1.15 mN (42.2 kPa) in the LP mode.However, when the structure was dragged by 10 μm in the grasping direction, the F G of 2.39 mN (87.8 kPa) was achieved, which is approximately a 107 percent increase over F LP .It should be noted that although the maximum F G (2.39 mN) was lower than F LP of the mushroom structure without the TPS structure (3.8 mN), the minimum tangential force needed to achieve a negligible detachment force S R (0.29 mN) was much lower than that of the mushroom structure (1.45 mN) as shown in Figure 2I.With the very small tangential force S R obtained at 20 μm in the releasing direction, the optimal structure could demonstrate a negligible detachment force within the noise level (<0.002 mN).Thus, it was shown that the optimal structure has superior controllability to switch between attachment and detachment modes.In addition, as shown in Figure 2I, the tangential force S G needed to achieve the maximum F G , was ≈0.57mN.Although it seemed relatively high, the adhesive structure could hold S G because S G was lower than the maximum tangential force in the grasping direction (1.48 mN) and the maximum frictional force between the substrate and the target part is stronger than S G in general.It is worth noting that the maximum tangential force achieved with the optimal structure (1.48 mN) was comparable to the maximum tangential force with the symmetric mushroom structure (1.45 mN), which implies that the optimal adhesive structure has a strong attachment in the tangential direction.

Manipulation Mechanism
To further investigate the underlying mechanisms of the three loading modes, a finite element analysis (FEA) was performed using COMSOL simulation.The simulations were carried out with the neo-Hookean hyperelastic model for the adhesive materials, the cohesive zone model (CZM) for the interfacial contact, and a rigid substrate assumption.In addition, the suction effect was considered in the simulation by calculating the projected area of the internal volume formed between the contact tip and the substrate until the crack at the interface is open (see Experimental Section).Along with the simulation results, the experimental observation of the real contact area during the operation was carried out to validate the simulation model with the suction-effect hypothesis.
In the LRP mode as shown in Figure 3A and Movie S1 (Supporting Information), the crack started at the edge on the releasing side (nearby the TPS structure) as soon as the shear force was applied in the releasing direction.As the shear displacement increased, the crack propagated toward the center of the stem without the formation of any internal voids, thereby eliminating the suction effect.As the crack developed further, the contact area dropped to a point so that the pull-off force could be negligible.Despite the fact that the crack in the experiment started at the edge of the releasing side and continued to grow, the experimen-tal result generally showed good agreement with the simulation result, as can be seen in Figure 3A.
On the other hand, when the microstructured sample was sheared in the grasping direction (the LGP mode, see Figure 3B; and Movie S2, Supporting Information), the bottom view from the simulation showed that the crack was initiated near the center with a slight offset away from the TPS structure side.As the microstructure was further sheared, the crack grew and formed a void of a large volume, which shows a great match with the experimental results.Unlike the LRP mode, the internal void was successfully maintained at the center in the LGP mode, which resulted in a large suction effect.From the cross-sectional view, we could observe that the volume of the interior void is large during the pull-off phase.It was also observed that at the moment right before the crack was opened up, the crack was almost in a circular shape with negligible offset away from the stem center.In addition, the cross-sectional view clearly showed that the microstructure could form a larger internal void of 1585 μm 3 in the LGP mode (Figure 3B) than that of 1344 μm 3 in the LP mode (Figure 3C), which resulted in a larger suction effect.
Although the shear-induced attachment or detachment can be achieved when the target parts are in contact with the substrate, the pull-off force achieved by the shear-induced attachment cannot be maintained while holding the parts in the air.Hence, it is important to understand the cracking behavior in the LP mode as illustrated in Figure 3C and Movie S3 (Supporting Information).The crack in the LP mode started at the center, but it grew faster toward the TPS structure side due to the greater stiffness in the presence of the TPS structure.In this manner, the adhesive structure showed a smaller maximum interior void and thus a lower pull-off force than the mushroom-shaped structure of the same dimensions.In addition, although the simulation only showed the center cracking mode (see the top row in Figure 3C), both the edge cracking (left-bottom in Figure 3C) and the center cracking (right-bottom in Figure 3C) modes were observed in the experiment, possibly due to the defect in the structure or the misalignment issue.
The optimized adhesive microstructure was further investigated to find the optimal displacements that can maximize the pull-off force F G for grasping and minimize the tangential force S R for releasing.As shown in Figure 3D, the pull-off force was negligible when the microstructure was sheared 20 μm in the releasing direction.The tangential force S R reached 0.29 mN when the structure was sheared 10 μm and beyond in the releasing direction, which represents the minimum required force to initiate the edge cracking in the LRP mode.It is worth noting that the shear displacement needed for the negligible pull-off force is small compared to the tip size (only ≈40% of its contact tip diameter).Moreover, the trigger tangential force required only ≈20% of the maximum tangential force.On the other hand, the pulloff force F G reached its maximum when the microstructure was sheared 10 which is smaller than the tangential displacement for the negligible detachment.
It was postulated that the microstructure could exhibit superior attachment performance thanks to the suction effect from the void.This can be further supported by both the simulation and the experimental results.As can be seen in Figure 3D, the trend of adhesion as a function of the tangential displacement measured in the experiment showed good agreement with the trend of the suction force calculated from simulation results.These results support the notion that the suction effect is dominant in the adhesive structure proposed in this study.In contrast to previous work, [19] where the suction effect played an insignificant role due to a small and short-lived interior cavity, our structure features a large tip-to-stem ratio, resulting in a relatively larger cavity.The interior cavity starts at an early shear phase in LGP mode and an early pull phase in LP mode, leading to a longer-lasting suction effect.Consequently, our structure can benefit from the suction effect even at low retracting speeds compared to others. [55]Furthermore, in contrast to the polyurethane utilized in the other gecko-inspired adhesives, our structure employs polydimethylsiloxane (PDMS), which is known for its lower surface free energy.Scaling up the microstructure may result in changes to the adhesive performance due to alteration in the dominance of the suction effect.Therefore, it is worth emphasizing that the significance of the suction effect in adhesive systems can be dominant or negligible depending on fac-tors such as structural shape, material selection, and working conditions.

Characterization in Wet Conditions
To further evaluate the adhesive performance with various materials and wet conditions, the adhesive characterization as well as the underwater pick-and-place task were performed.As shown in Figure 3E, the target substrates were classified into two categories based on their hydrophobicity.The hydrophilic glass and steel substrates showed that the adhesive forces were slightly increased under wet conditions compared to dry conditions.Compared to the conventional mushroom-shaped structure, which has a significant drop in adhesive force under wet conditions, [56,57] our structure acted similarly to the octopusinspired adhesive structure with a strong attachment force. [58,59]he enhancement in wet conditions indicates the existence of the suction effect in the microstructure.The pressure difference between the interior and exterior of the structure is generated by the interior shape deformation during the shear or pull phase, resulting in the formation of an interior void that enables the suction effect.During various loading phases, the size of the interior space changes in a similar way to suction structures such as suckers or pores. [57]However, it should be noted that the design and geometry of the structure affect the effectiveness and efficiency of underwater adhesion.However, the adhesive force was weaker in both dry and wet conditions for hydrophobic materials like polytetrafluoroethylene (PTFE) and polyamide (PA).In addition, the pull-off force values between the microstructured adhesives and various substrates in the dry condition were equivalent to those in the underwater condition, except for the PTFE substrate, as shown in Figure 3E.The hydrophilic substrates showed higher force values in underwater conditions, while the hydrophobic substrates showed higher force values in dry conditions.

Ultralight Part Pick-and-Place
To demonstrate the capability of the optimal adhesive microstructure, pick-and-place demonstrations of a wide range of ultralight parts ranging from micrometer to millimeter scale were carried out as shown in Figure 4. Owing to the high controllability of the microstructure, the target parts could be picked up and released on the same substrate, allowing for a greater variety of substrate materials.Moreover, the microstructure array of the same size could handle parts of various scales, which demonstrates the adhesive microstructure's versatility in addition to its strong attachment and switchability.
As shown in Figure 4A-i and Movie S4 (Supporting Information), the square ultrathin glass with a width of 10 mm and a thickness of 30 μm was tested for the pick-and-place demonstration.When dragged in the grasping direction, most of the microstructures were in contact with the glass, and only a few were not due to the alignment issue.Once the ultrathin glass was picked up in the air, the real contact area between the microstructures and the glass was invisible since it moved away from the focal plane.During the releasing process, after the drag motion was .Pick-and-place demonstrations of ultralight parts in dry conditions.A) Pick-and-place process of (i) ultrathin glass, (ii and iii) mini-LEDs, and (iv) resin plate.The left column illustrates the parts with labeled dimensions.Five photos of each part show five stages of the process: the approach phase, the drag phase in the grasping direction, the transfer phase in air, the drag phase in the release direction, and the release phase.B,C) SEM images of the printed resin plates.D) Normal force and shear force curves of picking up the resin plates from a PDMS substrate.
applied to the releasing direction, the real contact area showed clear evidence of the edge peeling, which resulted in the loss of suction effect and thus the negligible detachment force.Hence, the glass part could stay on the substrate, showing that the adhesive structure is capable of handling ultrathin and ultralight parts.
The microstructure could also handle mini-LEDs arrays with two different dimensions all within 1 mm, as shown in Figure 4Aii,iii and Movie S5 (Supporting Information).The taller mini-LEDs had a height of 550 μm in a 1D array of three elements, while the flatter one had a height of 200 μm in a 1D array of two elements.The first rectangles in the series showed that the leftmost mini-LEDs for both types were partially contacted.Furthermore, they had higher average roughness values (R a ) of 0.18 and 0.41 μm and heavier weights of 29 and 69.5 mg compared to the ultrathin glass sample of 0.75 mg.Hence, the pick-and-place demonstrations of the mini-LEDs arrays could show the high tolerance for handling the mini-LEDs even without precise and accurate alignment.
To further confirm the capacity of the adhesive microstructure, a micrometer-scale square resin plate with a width of 100 μm, a thickness of 30 μm, and a weight of 0.33 μg was used as the target part.The SEM images shown in Figure 4B,C indicated that the plate had a smooth surface, which resulted in a high adhesive force with the substrate.Despite the light weight of the plate, the attachment force needed to separate the resin plate from the PDMS substrate was ≈0.75 mN, which was much larger than the weight of the plate due to the adhesion between the target part and the substrate, as shown in Figure 4D and Movie S6 (Supporting Information).Nevertheless, since the F G from four microstructures was 1 mN (from 2.39 mN/9 structures x 4 structures), the resin plates could be lifted up in spite of the strong adhesion to the substrate.Since the gripper made a flat-to-flat contact with the resin plates, the gap between the uncontacted microstructure and substrate was only 30 μm while the area is large.Therefore, during the pull phase with drag in the releasing direction, a slight electrostatic force in normal direction was captured.The fourth photo in Figure 4A-iv showed that the edge cracking of the adhesive structures was initiated after a tangential displacement was applied in the releasing direction, and thus the resin plates could be released with a negligible detachment force.Moreover, the orientation of both the resin plates was successfully maintained after performing the pick-and-place task.It should be noted that the adhesive structures showed some detachment force because some microstructures were in contact with both the resin plate and the substrate, which resulted in an incomplete deactivation of the attachment.
The adhesive characterization showed that the microstructure array possesses an adhesive strength exceeding 85 kPa, which is comparable to the adhesive strength of gecko toe pads and common commercial adhesives (typically ≈100 kPa).This level of adhesive strength is sufficient for handling ultrathin glass and micro-LEDs, as demonstrated in our experiments.Notably, the most critical factor in pick-and-place tasks for fragile parts is switchability, which is excellently exhibited in our structure, allowing for efficient changes in adhesive states to attach or release parts.

Pick-and-Place Across the Water-Air Interface
An underwater pick-and-place task of a 1-inch square wafer (≈1 g) was performed, and the wafer was transferred across the water to the air as shown in Figure 5 and Movie S7 (Supporting Information).The microstructures were attached to a flat tip, whose position could be controlled by two linear stages (see details in Experimental Section).As shown in Figure 5ii, the microstructures first made contact with the wafer and a preload was applied with drag in the grasping direction.After picking up the wafer in water, it was pulled away from the water by the attachment force, which was working against the surface tension of the water, as shown in Figure 5iv.The microstructures could provide enough adhesive force to maintain the attachment of the wafer during the changes in interface from water to air.Since the suction effect was involved in the attachment force, the adhesive performance was not affected much in wet conditions.In addition, the wafer was also able to be transferred back into the water.When the wafer touched the bottom of the petri dish, the microstructures were dragged in the releasing direction (see Figure 5vi), allowing the microstructures to release the wafer.In addition to exhibiting similar underwater adhesion properties to those of other axisymmetric mushroom structures, [60,61] our microstructures maintained their high switchability even in underwater conditions.To avoid the significant drag forces that could have interfered with the purpose of demonstrating underwater adhesion (Movie S7, Supporting Information), we used a constant moving speed of 5 mm s −1 in our demonstration of the pick-and-place task under such conditions.The demonstration showed that the microstructure could function seamlessly in both dry and wet conditions unless the drag force became dominant.

Conclusion
In this study, we proposed a unique design of switchable adhesive microstructure by introducing a TPS structure beneath the tip of a mushroom-shaped microfiber.The proposed microstructure could manipulate fragile parts with a wide range of length scales in both dry and wet environments.The optimized adhesive microstructure yielded a superior attachment strength of 87.8 kPa when sheared 10 μm (1/5 of the tip diameter) in the grasping direction.Moreover, the adhesive could generate a negligible detachment strength of <0.07 kPa within the noise level of 0.002 mN when sheared 20 μm in the releasing direction, which corresponds to the trigger shear stress of 10.7 kPa.The involvement of the suction effect in the attachment of the adhesive microstructure was clearly confirmed by experimental data and finite element analysis models on shear-induced switching and cracking modes in adhesive strength in these systems.Thanks to the suction effect, the microstructured adhesive was able to perform pick-and-place tasks of various ultrathin and ultralight samples in both dry and wet conditions.These characteristics pave the way for the future transfer printing and pick-and-place of ultralight and fragile parts with gentle, reliable, and easy mechanical actuation.

Experimental Section
Fabrication of the Microstructure: A square-packed bioinspired dry adhesive microstructure array mold master was 3D microprinted using two-photon polymerization (2PP) with a gap distance of 55 μm among the structures.The microstructure had a 50 μm tip diameter, 2 μm in tip thickness, 20 μm in stem diameter, 30 μm in stem height, and a fillet of 5 μm at the connection point between tip and stem.The optimal design of the microstructure had a TPS structure with 10 μm in inner height H 1 , 7.5 μm in outer height H 2 , and a revolve angle of 40°, as shown in Figure 2A.For characterization, nine structures were used in three by the array with a center-to-center distance of 55 μm.All the molds were prepared by the 3D microprinter using 2PP (Photonic Professional GT, Nanoscribe GmbH) using a rigid IP-S commercial photoresist (Nanoscribe GmbH) on a silicon wafer (Figure S1, Supporting Information).The master molds were then developed for 40 min in propylene glycol monomethyl ether acetate (Sigma-Aldrich Inc.) and washed for 3 min in isopropyl alcohol.The molds were then placed in a vacuum desiccator for 40 min with 0.2 ml of (heptadecafluoro-1,1,2,2-tetrahydrodecyl)trichlorosilane (Sigma-Aldrich Inc.) in a glass vial after being exposed to plasma (Tergeo Plasma Cleaner, PIE Scientific LLC) for 2 min to activate the surface.This silanization process allowed the master molds to prevent bonding between the mold and the PDMS counterpart.The molds were then placed in an oven and baked for 40 min at 90 °C.The 10:1 PDMS mixture (Sylgard 184, Dow Corning) was applied to the mold and degassed for an hour in a desiccator.The PDMS mixture was cured at 90 °C for an hour in the oven, and the cured PDMS with the negative pattern was removed from the master mold.It was proceeded through the identical plasma, salinization, and baking procedures for the negative PDMS molds.The negative molds were filled with the 5:1 PDMS mixture, which was then allowed to degas for an hour.The redundant PDMS solution was scraped off the top surface of the negative molds using a razor blade before curing at 90 °C in an oven for an hour.The PDMS mixing ratio of 5:1 was chosen to induce elastic modulus mismatch, which facilitates demolding from the PDMS mold made of 10:1 mixing ratio.
Finite Element Method Simulations: The finite element analysis was performed by the commercially available software COM-SOL Multiphysics 6.0 (COMSOL lnc.).A single unit of the adhesive microstructure was modeled with the same aspect ratio as the sample in the adhesion testing setup, which could eliminate the size effect.The adhesive microstructure was modeled by implementing the neo-Hookean hyperelastic model with a Poisson's ratio of 0.495, while the glass counterface was modeled as a rigid plain surface.All simulations were conducted at the same speed to exclude the rate-dependent effects.As the microstructure was symmetrical about the midplane, half of the microstructure was simulated to reduce the model complexity for convergence.The geometries in the simulations were meshed with more than 5200 triangular elements, with extremely fine meshes toward the corner part to handle the stress singularity.
The CZM with the bilinear traction-separation law was implemented to simulate the adhesive separation between the adhesive microstructure and the glass counterface.In order to simulate van der Waals interaction with the CZM, a damage initiation distance  o of 30 nm and a maximum separation distance  f of 100 nm were used in the linear separation criterion of the CZM.The maximum interfacial stress  0 , which determines the adhesive strength, was tuned to match the experimental result.The suction force was applied as a boundary force at the contact interface with the assumption that the interior pressure is equal to the full vacuum condition.The suction force was defined as F suction = A i × 101 kPa where the interior area A i is the integral of the tip surface at which separation from the contact interface is larger than 30 nm.Once the edge of the tip no longer sealed the interior void, the suction force was removed from the simulation.All the simulations were performed following the same loading conditions and sequences used in the experiments.First, the rigid counterface was moved in the normal direction toward the adhesive microstructure until the normal load of 0.003 mN was achieved.The counterface was then moved tangentially under the same normal load until a predefined tangential displacement was reached.For the load-pull testing sequence, the tangential displacement was not applied.Last, the counterface was withdrawn from the contact in the normal direction until complete detachment.
Adhesion Characterization Setup: The adhesive characterization in both dry and wet conditions was conducted on a customized experimental setup (Figure S2A, Supporting Information).The setup consisted of two linear motor stages (LTS150/M, Thorlabs Inc.) and two load cells (GSO-25, Transducer Techniques LLC; and LSB200, FUTEK Inc.) in order to measure the normal force and tangential force and to apply various loading conditions.The hemispherical contact probe (10 mm diameter, measurements were performed when no air bubble was trapped in the array of microstructures.The noise level of the signal from the load cell was determined by calculating the root-mean-square value of the 50 data points.Each test was repeated five times. Pick-and-Place Demonstration Setup: For the pick-and-place task of the ultrathin glass and the resin plates, the same setup as for the adhesive characterization was utilized.The ultrathin glass commercially available (D263Teco, Schott AG), and the resin plates were printed by 2PP 3D printer (Photonic Professional GT, Nanoscribe GmbH) using a rigid IP-S commercial photoresist (Nanoscribe GmbH).The microstructured adhesive (0.39 × 0.88 mm) was attached on the probe with a flat acrylic plate facing downward.The backing substrate for the ultrathin glass was a glass slide, whereas the backing substrate for the resin plates was a PDMS sheet.The setup for the pick-andplace task of mini-LEDs (15404085BA470, Würth Electronics; SML-P12VTT86R, ROHM Semiconductor) and wafer (Figure S2B, Supporting Information).The setup consisted of two linear stages (LTS150/M, Thorlabs Inc.) to apply normal and tangential displacements.The microstructured adhesive (1.1 × 1.8 mm) was attached to the tip of the holder, which was connected to the normal direction motion stage by an extension plate.When pickand-place of the wafer in wet conditions, the petri dish was filled with deionized water and the wafer was placed on the bottom of the box.

Figure 2 .
Figure 2. A) Geometrical parameters of the proposed adhesive structure.B) Three loading modes, load-pull (LP), load-grasp-pull (LGP), and loadrelease-pull (LRP), which yield the pull-off forces, F LP , F G , and F R , respectively.The tangential forces due to the drag motion are denoted as S G for the LGP mode and S R for the LRP mode.C) Performance of the adhesives under three loading modes, F LP , F G , S R , and R, characterized as a function of the revolve angle  of the microstructure where H 1 = 10 μm, H 2 = 7.5 μm.D,E) Pull-off forces F LP and F G , F) the tangential force S R , and G) the adhesion-controllability ratio R are presented as a function of H 1 and H 2 of the microstructure where  = 40°.H) Normal forces measured under the three loading modes for the optimal geometrical parameters, H 1 = 10 μm, H 2 = 7.5 μm, and  = 40°.I) Shear forces measured under optimal tangential displacement for various microstructure designs and loading modes.All data presented in Figure 2C-I are experimental results.

Figure 3 .
Figure 3. Working mechanism of the adhesive microstructure with the finite element analysis of the mechanism of A) the LRP, B) the LGP, and C) the LP modes.The top rows show the sectional view of the microstructure under the corresponding loading phase, where the color map represents the von Misses stress.The middle rows show the bottom view of the microstructure, where the color map represents the distance between the contact substrate and the mushroom tips.For a distance greater than 100 nm, the region was assumed to be a crack and set as transparent.The bottom rows show the optical images of the evolution of the contact area.The real contact area is seen as a darker region, whereas the lighter region bounded by the dashed red lines denotes the crack at the interface.D) Normal and shear forces measured from experiments and suction force computed by the finite element model as a function of the tangential displacements.E) Forces of F LP , F G , and S R measured between the microstructured adhesive and the substrates of glass, steel, polytetrafluoroethylene (PTFE), and polyamide (PA) in dry and wet (underwater) conditions.

Figure 4
Figure 4. Pick-and-place demonstrations of ultralight parts in dry conditions.A) Pick-and-place process of (i) ultrathin glass, (ii and iii) mini-LEDs, and (iv) resin plate.The left column illustrates the parts with labeled dimensions.Five photos of each part show five stages of the process: the approach phase, the drag phase in the grasping direction, the transfer phase in air, the drag phase in the release direction, and the release phase.B,C) SEM images of the printed resin plates.D) Normal force and shear force curves of picking up the resin plates from a PDMS substrate.

Figure 5 .
Figure 5. Underwater pick-and-place demonstration of a silicon wafer and its transfer across the water-air interface.The red dashed square in the highlighted region in (i) shows the area with microstructure from the bottom side (1.1 × 1.8 mm 2 ).