Active control of bioinspired dry adhesion via magneto-responsive microstructures and PDMS composite backing

Abstract

Bioinspired dry adhesives require a balance between flexibility for conformal surface contact and stiffness for stable load-bearing adhesion. Here, we present a magnetically tunable dry adhesive that addresses this challenge by integrating 10 µm cylindrical polydimethylsiloxane (PDMS) microstructures with a composite backing containing ∼20 wt% Fe₃O₄ particles. The adhesion behavior of the fabricated samples is investigated both experimentally and using finite element analysis. The results reveal that application of an external magnetic field induces localized stiffening within the backing layer, actively reinforcing the adhesive interface. This magnetic activation suppresses premature microstructure bending, reduces interfacial stress concentrations, and enhances load transfer during adhesion. The optimized magnetically activated configuration reached its maximum shear strength at a displacement approximately 22% lower than that of the control and exhibited a 94% increase in peak adhesive stress during peel-off testing, reaching 0.062 N/cm². Numerical simulations corroborated the experimentally observed stress redistribution and adhesion enhancement mechanisms. These findings demonstrate a practical strategy for achieving magnetically tunable, high-performance dry adhesion through real-time magnetic control, with potential applications in soft robotics, advanced manufacturing, and intelligent gripping systems.

Bioinspired dry adhesives require a balance between flexibility for conformal surface contact and stiffness for stable load-bearing adhesion.

1. Introduction

Bioinspired dry adhesives have attracted considerable attention due to their ability to achieve strong and reversible adhesion without chemical bonding.^1,2 Inspired by biological systems such as geckos, spiders, and insects, these surfaces rely on micro and nanoscale fibrillar architectures that maximize van der Waals (vdW) interactions through increased contact splitting and conformal surface adaptation.^3–6 Among the materials used for synthetic dry adhesives, polydimethylsiloxane (PDMS) remains one of the most widely adopted because of its flexibility, low surface energy, and ease of microfabrication.⁷

Despite significant progress, a fundamental challenge remains. High adhesion requires a compliant interface capable of conforming to surface irregularities and maximizing real contact area.⁸ However, under external loading, highly compliant microstructures are susceptible to bending, rotation, and premature detachment, which reduce load-bearing performance. As a result, conventional dry adhesive systems often face a trade-off between contact formation and mechanical stability. To address this limitation, recent studies have explored stimuli-responsive dry adhesives that actively modify their mechanical properties in response to external inputs such as heat, light, pneumatic actuation, or magnetic fields.^9–11 Among these approaches, magnetically responsive systems offer several advantages, including rapid actuation, remote operation, low energy consumption, and the ability to penetrate opaque materials without significant attenuation. Magnetorheological elastomers (MREs), consisting of magnetic particles dispersed within an elastomeric matrix, can undergo reversible field-induced stiffening when exposed to an external magnetic field, providing a promising route for active control of adhesive performance.

Although magnetic reinforcement has been shown to increase the stiffness of elastomeric systems, its integration with microstructured dry adhesive surfaces remains relatively unexplored. In particular, the interaction between field-responsive backing layers and surface microstructures, and the extent to which this coupling can enhance adhesion performance, have not been systematically investigated.

In this study, we develop a magnetically tunable dry adhesive comprising 10 µm cylindrical PDMS micropillars supported by a composite backing containing approximately 20 wt% Fe₃O₄ particles. The design combines a compliant adhesive interface with a field-responsive support layer capable of localized magnetic reinforcement. Through experimental characterization and finite element analysis, we investigate the independent and combined effects of surface microstructures and magnetic activation on shear and peel adhesion. The results demonstrate that magnetic stiffening of the backing layer stabilizes the micropillar foundations, suppresses stress concentrations, and significantly enhances load-bearing performance. This work establishes a simple and effective strategy for achieving magnetically tunable dry adhesion and provides new insights for the design of advanced adhesive systems for soft robotics, intelligent gripping, and advanced manufacturing.

2. Materials and method

2.1. Materials

Polydimethylsiloxane (PDMS) and the corresponding curing agent (Sylgard 184 Silicone Elastomer Kit, Dow Corning) were obtained from Revolution Industrial (Victoria, Australia). A commercially available UV-curable epoxy, Norland Optical Adhesive (NOA-61), was sourced from AusOptic (Macquarie Park, NSW, Australia). For negative mold replication, epoxy resin (Type 105 Resin) and hardener (Type 205) were obtained from West Systems (Molendinar, QLD, Australia). To facilitate demolding, 1H,1H,2H,2H-perfluorooctyl-trichlorosilane (PTCS) was obtained from Sigma-Aldrich (North Ryde, NSW, Australia). The ferromagnetic filler phase consisted of magnetite nanopowder (Fe₃O₄, 98% purity, 100 nm average particle size) supplied by US Research Nanomaterials, Inc. (Houston, TX, USA). Silicon master templates with cylindrical pillars (10 µm in diameter) were used to produce the bioinspired surface structures.

2.2. Fabrication

The fabrication of the eight experimental samples was conducted in three distinct stages to create a magneto-responsive system, as shown in Fig. 1. Initially, a composite was prepared by mixing 5 g of Fe₃O₄ nanoparticles (100 nm, 98% purity, US Research Nanomaterials, Inc.) into 10:1 PDMS prepolymer. The mixture was degassed and cast into a PMMA template with a 16 × 16 array of square cavities, each measuring 1.2 mm × 1.2 mm and 1 mm deep. Following thermal curing at 80 °C, the composite was demolded to yield a substrate consisting of independent Fe₃O₄-PDMS square macropillars.

Fig. 1. Schematic illustration of the fabrication process for magneto-responsive dry adhesives.

To produce the “Smooth” baseline samples (1,2), neat 10:1 PDMS was poured into the interstitial spaces surrounding the previously fabricated macropillars. The system was cured until the neat PDMS levelled with the pillar tops, resulting in a smooth surface with an embedded magnetic skeleton. This planarization step ensured a consistent interface for either direct adhesion testing of the smooth controls or for the subsequent integration of the secondary microtopography.

The “Pillar” samples (3,4) were then produced by bonding a prefabricated 75 µm thick microstructured PDMS membrane, featuring 10 µm diameter pillars, onto the smooth surface of the Stage 2 samples. This micron-sized pillar membrane was fabricated via the replica molding technique described in previous work.⁷ A thin layer of uncured PDMS served as the adhesive agent to fuse the layers, resulting in a final hierarchical device with a total thickness of approximately 2.5 mm.

2.3. Structural and morphological characterization

We evaluated microstructural morphology, particle dispersion quality, and micropatterned array geometry using optical microscopy and optical profile tensiometry. The spatial dispersion and micro-agglomeration state of Fe₃O₄ within the PDMS was investigated using an optical microscope (Nikon Eclipse Ei) operated in transmitted bright-field and dark-field scattering modes. Transmitted bright-field imaging evaluated macroscale spatial dispersion, while dark-field illumination and high-magnification edge cross-sections assessed localised particle clustering and matrix boundary encapsulation. Additionally, the cross-sectional profile, pillar height, draft angles, tip morphology, and interfacial bonding continuity of the micropillar arrays were characterized using an optical tensiometer (Attension Theta, Biolin Scientific) configured with a telecentric lens and monochromatic backlighting for high-contrast side-profile silhouette analysis.

2.4. Adhesion measurement

All fabrication steps and mechanical adhesion measurements were performed under strictly controlled ambient conditions at a stabilized laboratory temperature of 23 ± 1 °C and a relative humidity of 45 ± 5%. To ensure cross-batch consistency, we monitored the multi-material hierarchical samples using digital micrometre callipers to maintain a stringent total sample thickness tolerance of ±15 µm across all tested groups. Adhesion was measured using an ElectroPuls E10000 tensile testing machine from Instron Systems at a constant speed of 0.5 mm/minute, as shown in Fig. 2. For shear tests, samples measuring 2 × 2 cm² were brought into conformal contact with a clean plastic substrate under a standardized, automated preload force of 1.0 N (corresponding to a nominal contact pressure of approximately 2.5 kPa). This preload was maintained for a fixed contact time of 30 s prior to testing to ensure uniform interfacial wet-out and maximize initial van der Waals interactions across the contact plane. Following the dwell period, the samples were pulled sideways at a constant speed of 0.5 mm per minute until they detached. The same preload force (1.0 N) and contact time (30 s) protocol was applied uniformly to the 90° peel-off sequences using the wheel fixture to ensure consistency across both loading modes. For peel tests, 90° peel strength measurements were performed using a custom CNC-machined aluminum wheel fixture mounted to the lower base of the testing machine (Fig. 2). The fixture comprises a central axle supported by low-friction bearings, allowing the wheel to rotate freely. As the upper tensile grip executes vertical displacement at a constant separation rate of 10 mm min⁻¹, the passive rotation of the wheel maintains a continuous, precise 90° peel angle relative to the tangential detachment point without introducing parasitic lateral shear loads. To eliminate viscoelastic substrate stretching and compliance artifacts during peeling, a non-stretchable Kapton film (thickness: 0.1 mm) was rigidly bonded to the back of the flexible composite substrate using pressure-sensitive acrylic transfer tape.

Fig. 2. Experimental setup showing the N45 magnet held by a custom-made CNC arm. The arm positions the magnet at a specific angle to focus the magnetic field on the sample during testing.

Active magnetic stimulation was provided by a high-grade N45 Neodymium cylinder magnet (22 mm diameter, 25.4 mm height) with a surface flux density of 6000 Gauss. The magnet was secured and precisely oriented using a multi-axis, in-house made CNC-machined aluminum arm. During active testing, the N45 Neodymium magnet was held at an air gap of 2.0 mm above the sample. The magnetic flux density map across the 2×2 cm² active sample area was measured using a calibrated transverse Hall-effect Gaussmeter, yielding a peak surface field of 600 ± 12 mT at the center with a uniform axial field gradient of ∇B = 45 T/m. The field variation across the active center remained within ±4.5%, providing uniform field exposure across the macropillar array. This targeted spatial configuration ensured maximum dynamic impedance matching, driving localized stiffening of the internal Fe₃O₄-PDMS macropillars at the precise moment of interface separation.¹²

The samples are named based on their surface type and whether the magnetic field was used as shown in Table 1. For the shear tests, S1-smooth is a flat surface tested with the magnet off, while S2-smooth-mag is the same flat surface tested with the magnet on. S3-pillar has the 10 µm micropillars on the surface tested with the magnet off, and S4-pillar-mag combines these pillars with an active magnetic field. Similarly, for the peel tests, P1-smooth and P2-smooth-mag represent the flat surfaces without and with the magnetic field, respectively. P3-pillar features the micropillars tested with the magnet off, and P4-pillar-mag uses both the pillars and the magnetic field targeted at the peeling point. These names are used consistently in the results graphs to allow for easy comparison between the passive and active states of the adhesive.

Table 1. Summary of experimental samples and testing conditions.

Sample name	Surface type	Magnetic field	Test type
S1-smooth	Flat PDMS	OFF	Shear
S2-smooth-mag	Flat PDMS	ON	Shear
S3-pillar	10 µm pillars	OFF	Shear
S4-pillar-mag	10 µm pillars	ON	Shear
P1-smooth	Flat PDMS	OFF	Peel
P2-smooth-mag	Flat PDMS	ON	Peel
P3-pillar	10 µm pillars	OFF	Peel
P4-pillar-mag	10 µm pillars	ON	Peel

2.5. Finite element modeling of smooth composite adhesive

To elucidate the interfacial stress distribution and structural reinforcement mechanisms within the stimuli-responsive smooth adhesive system, a three-dimensional macroscale finite element model (FEM) was constructed using ANSYS. A 3 × 3 pillar subarray was implemented as a Representative Volume Element (RVE) with symmetric boundary conditions to capture interpillar mechanical interactions and stress-field overlap representative of the full 16 × 16 array while optimizing computational efficiency. The simulation framework utilized a macroscale continuum mechanics approach combined with a virtual cohesive zone model (CZM) to investigate bulk load transfer and stress concentrations across the pillar bases. The simulation domain replicates the exact multi-component geometric architecture utilised during experimental characterisation, capturing the hyperelastic deformation profiles of the embedded magnetic composite skeleton under localised mechanical loading. A macroscale continuum modeling framework was strategically prioritized over a computationally intensive multiscale simulation.¹³ While microscale contact-splitting phenomena are addressed separately via analytical formulations, this macroscale framework is specifically designed to isolate and track macroscopic edge-stress concentrations, bulk deformation mechanics, and structural detachment fronts across the active composite boundaries.¹⁴

2.5.1. Geometric and mesh model construction

The geometric domain consists of a layered multi-material architecture (Table 2). The structural foundation is a 0.1 mm thick ground substrate supporting a 3 × 3 array of square macropillars.¹⁵ Each individual macropillar possesses cross-sectional dimensions of 1.2 mm × 1.2 mm, a height of 1 mm, and an interpillar edge-to-edge spacing of 1 mm. Both the ground substrate and the macropillars are modeled as a homogeneous Fe₃O₄-PDMS nanoparticle composite.

Table 2. Geometric dimensions, material compositions, and functional roles of the multi-layered finite element model components.

Layer/component	Material	Geometry/dimensions	Role in simulation
Top backing plate	Polyethylene terephthalate (PET)	Thickness: 0.1 mm	Loading vehicle/displacement application
Surrounding matrix	Pure PDMS	Height: 1.1 mm (flush with pillars)	Smooth contact interface matrix
Embedded pillars	PDMS-Fe3O4 composite	3 × 3 array; 1.2 mm × 1.2 mm × 1 mm	Active mechanical skeleton (stiffened phase)
Ground substrate	PDMS-Fe3O4 composite	Thickness: 0.1 mm	Foundation layer

To model the co-cured smooth configuration, the entire 3 × 3 pillar array is completely submerged within a neat (pure) PDMS matrix block with a total height of 1.1 mm, resulting in a flush, planar top interface. Finally, a 0.1 mm thick rigid plastic backing layer (polyethylene terephthalate, PET) is bonded directly to the upper surface of the smooth matrix layer to serve as the loading vehicle, matching the experimental testing protocol. This backing plate acts as a kinematic loading vehicle to ensure uniform displacement translation across the detachment front, directly matching the constraints of the experimental setup.¹⁶

To ensure accurate numerical solutions and establish grid independence, we performed a systematic mesh convergence analysis. The multi-material domain was discretized using lower-order, 8-node structural solid elements (ANSYS SOLID185) configured with a mixed µ–P (displacement–pressure) formulation to effectively prevent volumetric mesh locking under incompressible hyperelastic loading.¹⁷ The mesh density was progressively refined across four independent grid variations by an element size reduction factor of 1.5. Convergence was evaluated based on the stability of the peak interfacial shear and normal stresses. A finalized mesh architecture with a localized element size refined down to 15 µm at the critical material interfaces was selected. This configuration guaranteed that further mesh refinement yielded a variation in peak stress values of less than 1.1%, establishing a fully converged, mesh-independent predictive model.

2.5.2. Constitutive material models

The polymeric phases within the model exhibit distinct mechanical characteristics

2.5.2.1. Neat PDMS matrix

Modeled as an isotropic, incompressible hyperelastic material governed by a two-parameter Mooney–Rivlin strain energy density function:W = C₁₀(Ī₁ − 3) + C₀₁(Ī₂ − 3), where C₁₀ and C₀₁ are the material constants derived from experimental tensile testing of pure Sylgard 184 (C₁₀ = 0.054 MPa and C₀₁ = 0.013 MPa), and Ī₁ and Ī₂ are the first and second deviatoric strain invariants.

2.5.2.2. Fe₃O₄-PDMS composite skeleton

To simulate the passive and actively stimulated states, the composite macropillars and base substrate are modeled with an elevated hyperelastic stiffness profile relative to the pure matrix. Under active magnetic stimulation, the material parameters (C₁₀ = 0.145 MPa, C₀₁ = 0.038 MPa) are modified to reflect the magnetic-field-induced structural rigidification (the “active skeleton” state) observed experimentally. Rather than implementing a fully coupled magneto-mechanical solver which introduces massive computational expense without changing macroscopic stress trajectories, the magnetic constitutive coupling is handled by executing field-dependent property modulation. In the passive state (no magnetic field), the composite matrix constants are defined at a baseline of C₁₀ = 0.082 MPa, C₀₁ = 0.021 MPa.

2.5.2.3. Plastic upper layer

Modeled as a standard linear elastic isotropic solid (E = 2.5 GPa, v = 0.34) to guarantee uniform displacement translation without structural damping from the loading fixture.

2.5.3. Contact formulation, boundary conditions, and loading

Perfect interfacial bonding is prescribed across all internal material boundaries, specifically between the composite macropillars, the surrounding neat PDMS matrix, and the lower substrate using a bonded contact formulation to simulate a perfectly cross-linked monolithic structure.

The bottom surface of the 0.1 mm composite ground substrate is assigned a fixed boundary condition, completely constraining all translational degrees of freedom (U_x = U_y = U_z = 0). Mechanical testing is simulated via displacement-controlled loading applied uniformly to the upper surface of the 0.1 mm plastic backing layer. For shear simulation, a lateral displacement is prescribed parallel to the interface, whereas peel-off behavior is captured by applying a vertical normal displacement vector, allowing for direct tracking of interfacial stress concentrations and reaction force evolution.

2.5.4. Simulation methodology and loading configuration

The model geometry, comprising the 3 × 3 macropillar array embedded within the neat PDMS matrix and topped with the rigid PET backing, is illustrated in Fig. 3. This configuration ensures that the structural contribution of the magnetic composite skeleton is fully captured during the simulated deformation.

Fig. 3. Finite element model geometry showing the 3 × 3 macropillar array embedded within the PDMS matrix and the rigid PET loading plate.

To evaluate the adhesive performance, two distinct loading scenarios were applied to the upper surface of the PET layer. In the shear loading configuration, a lateral displacement was prescribed parallel to the interface, as shown in Fig. 4A. This loading condition evaluates the lateral stiffness of the integrated composite skeleton and its ability to suppress pillar bending under transverse force. Concurrently, a peel loading scenario was applied via a vertical normal displacement, as detailed in Fig. 4B. This loading configuration allows for the assessment of interfacial crack propagation and the stabilizing effect of the magnetic backing during normal detachment. To evaluate static equilibrium stress distributions while avoiding numerical instability and inertial artefacts, a displacement rate of 10 µm min⁻¹ was applied to the model. This quasi-static velocity isolates elastic load-transfer mechanics across the pillar array. Although lower than the experimental velocity (0.5 mm min⁻¹), both rates operate within the quasi-static regime for PDMS, ensuring the simulated stress profiles accurately reflect the equilibrium load distribution before crack initiation.

Fig. 4. Schematic representation of the simulated loading configurations. (A) Shear loading configuration showing lateral displacement applied to the PET backing layer. (B) Peel-off loading configuration showing vertical displacement applied to the PE.

3. Results and discussion

Bioinspired dry adhesives are pivotal for advanced applications such as climbing robotics, microgrippers, and medical interfaces, where residue-free, reversible attachment is paramount. Conventional optimization strategies have traditionally focused on static geometric parameters such as the aspect ratio and density of micropillar arrays to predefine adhesion levels. However, these passive designs are inherently limited because they cannot adapt to varying mechanical loads or surface conditions once fabricated.

This study addresses these limitations by utilizing polydimethylsiloxane (PDMS) for its superior conformal contact capabilities, integrated with a ∼20 wt% Fe₃O₄ magneto-responsive backing to enable active, on-demand control. Through a comprehensive evaluation of the eight experimental groups established in the methodology, we characterize the synergistic interaction between surface microarchitecture and localized magnetic stimulation. Our results demonstrate that the S4-Pillar-Mag and P4-Pillar-Mag configurations provide a unique, magnetically tunable performance profile that significantly transcends the capacity of standard passive adhesives.

3.1. Microstructural characterization and array morphology

Optical microscopy across representative backing regions reveals the particle loading distribution and agglomeration state of the Fe₃O₄ nanoparticles. Transmitted light micrographs (Fig. 5A) confirm a continuous, uniform filler network with minimal unmixed polymer domains across macroscale areas. Dark-field scattering analysis (Fig. 5B) reveals that while magnetic interactions lead to localized sub-micron to low-micron scale clusters, these microagglomerates are homogeneously distributed throughout the elastomeric volume without macroscale phase separation or sedimentation prior to curing. Furthermore, high-magnification edge cross-sections (Fig. 5C) show consistent particle encapsulation up to the matrix boundary layer, ensuring mechanical stability and isotropic magneto-responsive performance.

Fig. 5. Optical microscopy analysis of the Fe₃O₄-PDMS magnetic composite backing layer (A) transmitted bright-field micrograph; (B) dark-field scattering image; (C) high-magnification edge profile.

Complementary backlit optical tensiometry enables non-destructive, high-contrast structural validation of the micropillar array and its supporting substrate interface (Fig. 6). Planar macroscopic analysis (Fig. 6A) reveals a periodic, square-grid arrangement of micropillars with uniformity and minimal defect density across the field of view. Telecentric side-profile silhouettes (Fig. 6B) yield distinct boundary contrast, confirming a mean pillar height of 100 µm, uniform sidewall draft angles, and distinct microtextured tip topographies. Crucially, cross-sectional profile analysis demonstrates uninterrupted structural continuity across the micropillar-substrate boundary, establishing a fully integrated, monolithic interface free of microvoids, boundary fractures, or delamination defects.

Fig. 6. Structural and morphological characterization of the micropatterned array. (A) Macroscopic top-view image. (B) Side-view backlit optical tensiometer silhouette.

3.2. Field-dependent mechanical properties of composite backing

The magnetorheological response of the 20 wt% Fe₃O₄-PDMS composite backing was quantified under magnetic flux densities ranging from 0 to 600 mT. Static shear modulus (G) increased from 0.21 MPa in the zero-field state to 0.43 MPa under a 600 mT applied field, representing a 105% magnetically induced modulus increase. This field-dependent stiffening originates from interparticle dipole interactions that restrict chain mobility within the PDMS matrix, establishing the structural foundation for active interface reinforcement.

3.3. Shear adhesion

The shear adhesion results, illustrated in Fig. 7, reveal the mechanisms by which the combination of surface microarchitecture and magnetic actuation influences the stress-displacement behavior of the adhesive system. The baseline sample, S1-Smooth-NoMag, reached a peak shear stress of approximately 0.48 N per cm² at a displacement of 3.0 mm. When a localized magnetic field was applied to the smooth surface (S2-Smooth-Mag), the peak stress showed a negligible increase to 0.49 N per cm², failing at nearly the same displacement. This suggests that while the internal magnetic macropillars increase the bulk stiffness of the base, they do not significantly enhance the shear strength of a flat interface without surface topography.¹⁸ In a smooth configuration, the interface remains susceptible to macroscopic edge-stress concentrations that the internal magnetic reinforcement cannot effectively mitigate.¹⁹

Fig. 7. (A) Plot of stress versus displacement recorded for the shear test. Adhesion strength is estimated from this plot. (B) Plot of peel stress versus displacement recorded for the peel-off test. Peel strength is determined from this plot.

The introduction of surface microstructures in S3-Pillar-NoMag resulted in a peak shear stress of 0.51 N per cm². Although this represents only a modest increase over the smooth controls, the S3 curve is steeper at low displacements than S1. This indicates a higher initial shear stiffness provided by the pillars, a classic demonstration of the contact splitting principle, where dividing the contact area into multiple discrete microcontacts enhances the interface's sensitivity to small displacements.^20,21

The most significant observation is the performance of the synergistic sample, S4-Pillar-Mag. This sample showed the steepest initial stress increase of all tested groups, reaching a peak of 0.53 N cm⁻² at a much lower displacement of 2.33 mm. The empirical evidence confirms that the localized magnetic field specifically enhances the performance of the micropillars. While the passive pillars in S3 required a displacement of ∼2.9 mm to reach their limit, the magnetic reinforcement in S4 allowed the system to reach peak strength much faster.^12,22

This accelerated activation indicates that the magnetic probe stiffens the macropillars within the composite base. This, in turn, provides a more rigid foundation for the micropillars, preventing elastic instability, specifically premature bending or tilting that usually occurs under lateral loads.²³ By stabilizing the base of the microfeatures, the magnetic field ensures that the shear stress is focused more efficiently at the interface, allowing the system to reach maximum load-bearing capacity with minimal displacement. This “active reinforcement” effectively resolves the compliance paradox by providing on-demand structural integrity without sacrificing the initial flexibility required for surface mounting.²⁴

3.4. Peel adhesion

The peel adhesion characteristics, illustrated in Fig. 7, demonstrate how surface topography and magnetic fields influence the detachment energy and peak stress at a 90° angle. The baseline sample, P1-Smooth-NoMag, displays a characteristic plateau after initial detachment, maintaining a steady-state peel stress of approximately 0.032 N per cm² at a displacement of 22.5 mm.

The introduction of the magnetic field to the smooth surface (P2-Smooth-Mag) results in significantly different behavior. While the initial peel stress remains lower than the baseline for much of the displacement, a sharp peak of ∼0.053 N per cm² is observed at approximately 20.2 mm. This transient spike indicates that the localized magnetic field stiffens the internal macropillars, creating a temporary resistance to the advancing peel front before rapid failure occurs.²⁵

The microstructured sample, P3-Pillar-NoMag, shows a gradual, continuous increase in peel stress throughout the test, reaching approximately 0.043 N per cm² at 25.4 mm. Unlike the smooth samples, the presence of micropillars prevents a flat plateau, as the individual pillars undergo sequential stretching and detachment, contributing to more sustained energy dissipation.²⁶

The most notable performance is observed in the synergistic sample, P4-Pillar-Mag. This specimen shows the highest rate of stress increase from the onset of the test. The combination of surface micropillars and magnetic reinforcement allows the sample to sustain a significantly higher stress profile across the entire displacement range. The empirical evidence culminates in a distinct peak stress of 0.062 N/cm² at 22.3 mm. To evaluate peel performance rigorously, two distinct regimes were analyzed from the stress-displacement profiles: the transient peak initiation stress (σ_peak, representing resistance to crack nucleation) and the steady-state propagation stress (σ_ss). The active magnetic composite (P4) exhibited a 94% increase in peak initiation peel stress relative to the passive smooth reference (P1) (0.032 N/cm² → 0.062 N/cm²). This pronounced transient peak in P4 occurs at a displacement of ≈22.3 mm due to localized magnetic pre-clamping and structural resistance prior to macrocrack propagation. Following crack initiation, the post-peak propagation stress for P4 stabilizes at ≈0.033 N/cm², returning near the steady-state plateau level of P1 (≈0.032 N/cm²). Reporting both initiation peaks and propagation regimes provides an uncompromised evaluation of active peel mechanics, demonstrating that magnetic pre-actuation primarily enhances the initial force required for detachment initiation.

This enhanced performance in P4 suggests that the magnetic field provides a rigid structural support for the micropillars at the peeling tangent point. By stiffening the underlying Fe₃O₄ base, the magnetic probe ensures that the micropillars remain perpendicular to the substrate for a longer duration during the peel process, thereby maximizing the work required for detachment and increasing the overall adhesive strength.^27,28

A distinct phenomenological feature of the magnetic peel series (P2 and P4) is the occurrence of a sudden stress peak prior to detachment. This behavior is attributed to the synergistic interaction between the peel front and the localized magnetic field. As the adhesive is pulled from the CNC-machined wheel, the peel front advances into a region of concentrated magnetic flux. According to the experimental setup and supplementary observations, the magnetic probe induces a localized stiffening of the internal magnetic macropillars.^29,30

Unlike shear failure, where magnetic actuation promotes edge-initiated crack propagation, the curvature of the peel setup focuses the tensile force toward the center of the specimen.³¹ This prevents the immediate propagation of edge cracks, forcing the system to reach a higher energy threshold visualized as a sharp peak to drive the crack through the magnetically reinforced central zone.³² In the synergistic P4 sample, this effect is amplified by the micropillars, which maintain interfacial contact longer under the support of the stiffened base, resulting in the maximum recorded peel stress of 0.062 N per cm².

3.5. Statistical reliability and reproducibility analysis

To establish operational reliability and data transparency, we conducted a comprehensive statistical evaluation across all mechanical testing configurations (n = 5 independent samples per group). We confirmed dataset normality using the Shapiro–Wilk test (p > 0.05). To provide full visibility of experimental variance, individual raw data points (n = 5) are overlaid directly on the revised Fig. 6A and B values, standard deviations (SD), and calculated 95% confidence intervals.³³ The localized coefficients of variation remained strictly below 3% across all shear trials and below 4% across all normal separation runs, demonstrating high fabrication consistency.^34,35

A one-way Analysis of Variance (ANOVA) performed on the peak shear dataset (Fig. 8A) indicated highly significant variations among the distinct experimental groups (F(3,16) = 5.24, p < 0.001). A subsequent Tukey's Honestly Significant Difference (HSD) post-hoc test was utilized for pairwise multiple comparisons to isolate specific mechanical contributions.³⁶ The analysis confirmed that the transition from the smooth baseline to the smooth-magnetized state is statistically non-significant (p = 0.68), validating that bulk macropillar stiffening alone cannot bypass edge-stress concentrations on a flat interface. Conversely, the transition from the smooth baseline to the passive microstructured surface yielded a significant performance enhancement (p < 0.05). Most critically, the synergistic configuration showed a statistically significant increase in shear capacity over its passive microscale counterpart (p < 0.05), validating that field-induced dynamic impedance matching is a true physical phenomenon rather than a byproduct of random measurement fluctuations.³⁷

Fig. 8. Statistical reproducibility and raw data distribution of adhesive samples: (A) peak shear stress and (B) peak peel stress. Individual raw data points (n = 5, open circles) are overlaid on bar graphs.

The peel characteristics similarly showed strong statistical differentiation in a separate one-way ANOVA (Fig. 8B) (F(3,16) = 48.37, p < 0.001). Pairwise Tukey HSD evaluations established that the sharp transient peak stress induced by macropillar stiffening in the smooth-magnetized architecture was highly statistically significant relative to the steady-state baseline plateau (p < 0.001). Furthermore, introducing passive microscale pillars provided a statistically significant baseline enhancement over the smooth control (p < 0.01) due to sequential microscale contact splitting. Ultimately, the synergist microscale is structured-magnetized system achieved a highly statistically significant upward shift in peak peel stress over the passive microstructured configuration (p < 0.001). The presence of tightly bounded, non-overlapping 95% confidence intervals across these primary active-to-passive shifts confirms the mechanical testing platform's high fidelity, reproducibility, and reliability.

The operational reversibility of this active actuation is governed by the structural elasticity of the composite polymer matrix. Because the applied magnetic field deforms the composite strictly within its linear elastic regime, removal of the external field allows complete elastic recovery of the original surface morphology without permanent deformation or interfacial degradation. While this study focuses on the fundamental mechanics of active tuning under field application, the non-contact, fully elastic nature of the magnetic interaction provides a clear framework for repeatable, multi-cycle active adhesion.

3.6. Numerical validation of base reinforcement

To further elucidate the role of the magnetic macropillars on the bulk adhesive behavior, a macroscale FEA model of the smooth composite configuration was developed (Fig. 4). The simulation results corroborate the experimental observations that internal magnetic reinforcement provides limited shear enhancement for smooth interfaces.

As visualized in the shear simulation (Fig. 9A), the stress distribution within the smooth composite adhesive remains dominated by interfacial edge concentrations. While the embedded macropillars elevate the global structural stiffness, they do not sufficiently alter the stress profile at the contact plane to mitigate these edge-initiated failure modes.³⁸ Similarly, the peel-off simulation (Fig. 9B) demonstrates that the rigid macropillars create a high-stiffness foundation that delays crack propagation; however, the absence of microtopography prevents the formation of a sustained energy dissipation zone.³⁹ These numerical findings confirm that the synergy observed in the experimental S4 and P4 groups is not merely an additive effect of base stiffness, but a direct result of the hierarchical interaction between the magnetically stabilized macropillars and the surface microfeatures.

Fig. 9. Numerical simulation of the smooth composite adhesive. (A) Stress distribution map under shear loading. (B) Deformation profile under vertical peel loading.

It is worth noting the effect of displacement rate on absolute stress values. Because PDMS is inherently viscoelastic, higher experimental loading rates (0.5 mm min⁻¹) yield slightly higher effective modulus and detachment forces due to viscous dissipation compared to the near-equilibrium state modeled in FEA (10 µm min⁻¹). Consequently, the numerical model predicts spatial stress concentrations and relative load-transfer trends rather than precise rate-dependent absolute magnitudes.

The observed numerical values for both shear and peel-off capacities fall within the tenths of a N/cm² range, whereas the empirical experimental values consistently occupy a lower magnitude threshold. This order-of-magnitude discrepancy is a well-documented outcome of the physical transition from idealized continuum finite element modeling to experimental contact testing. The macroscale FEA simulation assumes a perfectly planar, flaw-free interface with 100% nominal contact area under fully synchronized stress distribution. In contrast, real-world elastomeric contact interfaces are inherently governed by surface roughness, microscopic air entrapment, and spatial compliance variations. According to established contact mechanics theory, microscale surface roughness severely restricts the real contact area to a small fraction of the nominal geometric area, which directly degrades the effective macroscopic adhesion of thin elastomeric films.^40,41 Consequently, these interfacial contact realities drastically reduce the true load-bearing area compared to the continuous numerical mesh.

Furthermore, stochastic asperities induce localized multi-scale defect configurations that break contact sequentially in minute, abrupt movements rather than instantaneous, perfect sheet separation.^33,42 Consequently, while the FEA model accurately captures the qualitative stress pathways and reinforcement trends, it inherently overestimates the quantitative failure thresholds by bypassing these stochastic, interface-reducing microscale defects. It is important to note the multi-scale boundary of this computational approach. While the physical composite exhibits microscale surface features, the FEA framework operates on a continuum scale. Consequently, the model predicts macroscale stress redistribution and load-concentration profiles during shear and peel loading rather than resolving single-pillar contact-splitting dynamics at the submicron interface.

4. Conclusion

This study demonstrates a successful methodology for enhancing bioinspired dry adhesion through the synergy of surface microstructuring and magneto-responsive stimulation. By integrating 10 µm cylindrical micropillars with an ∼20 wt% Fe₃O₄-PDMS composite backing, we successfully resolved the “compliance paradox” inherent in soft adhesives. The proposed design effectively decouples the conformal contact provided by the surface microfeatures from the structural stability required for high load-bearing, which the magnetic composite base actively modulates.

Instron testing confirms that while microstructures provide a foundational increase in adhesion, the application of a localized magnetic field creates an “active mechanical skeleton” that maximizes this effect. These empirical observations were further validated through macroscale finite element analysis, which elucidated how magnetic stiffening of the internal macropillar skeleton suppresses interfacial edge-stress concentrations. The synergistic configuration (S4) achieved a peak shear strength of 0.53 N/ cm², while peel-off tests revealed an approximately 94% gain in peak stress over the smooth baseline and unique crack-arresting behavior.

By unifying numerical insights into stress distribution with experimental performance metrics, this work establishes a robust framework for developing magnetically tunable robotic grippers capable of high-load transport with low-energy, on-demand release. Future work will focus on optimizing magnetic nanoparticle distribution and exploring dynamic field orientations to further enhance the responsiveness and precision of the adhesive interface.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data supporting this study are included in the article.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d6ra06133h.