Machine Learning‐Based and Experimentally Validated Optimal Adhesive Fibril Designs

Abstract

Setae, fibrils located on a gecko's feet, have been an inspiration of synthetic dry microfibrillar adhesives in the last two decades for a wide range of applications due to unique properties: residue‐free, repeatable, tunable, controllable and silent adhesion; self‐cleaning; and breathability. However, designing dry fibrillar adhesives is limited by a template‐based‐design‐approach using a pre‐determined bioinspired T‐ or wedge‐shaped mushroom tip. Here, a machine learning‐based computational approach to optimize designs of adhesive fibrils is shown, exploring a much broader design space. A combination of Bayesian optimization and finite element methods creates novel optimal designs of adhesive fibrils, which are fabricated by two‐photon‐polymerization‐based 3D microprinting and double‐molding‐based replication out of polydimethylsiloxane. Such optimal elastomeric fibril designs outperform previously proposed designs by maximum 77% in the experiments of dry adhesion performance on smooth surfaces. Furthermore, finite‐element‐analyses reveal that the adhesion of the fibrils is sensitive to the 3D fibril stem shape, tensile deformation, and fibril microfabrication limits, which contrast with the previous assumptions that mostly neglect the deformation of the fibril tip and stem, and focus only on the fibril tip geometry. The proposed computational fibril design could help design future optimal fibrils with less help from human intuition.

Enhanced Flexible Mold Lifetime for Roll‐to‐Roll Scaled‐Up Manufacturing of Adhesive Complex Microstructures

“…The latter specific microstructure was the optimal adhesive structure with a tip diameter of 50 µm and fiber height of 50 µm found in the study of machine‐learning‐based optimal adhesive microstructures. [ 35 ] As shown in Figure 5b,c, the adhesive microstructures were successfully printed up to 1000 cycles for the first time.…”

“…Silicone resins were poured onto the silanized master mold, and the final negative molds were prepared from such master. As for the adhesive microstructure from the two‐photon polymerization, the similar shape to the optimized adhesive microstructure [ 35 ] was chosen and directly 3D‐printed by a commercial two‐photon polymerization system (Photonic Professional GT, Nanoscribe GmbH) with a rigid commercial photoresin (IP‐S, Nanoscribe GmbH). After the printed microstructured sample was silanized through the aforementioned steps, it was used as a positive master mold.…”

Wireless soft millirobots for climbing three-dimensional surfaces in confined spaces

“…Figure 3 and movie S2 show that the robot footpads could be designed on demand to allow the robot to climb 3D dry surfaces with different roughness, while the control signal could be adapted to allow the robot to climb complex curved surfaces. We use mushroomshaped dry adhesives (27) for dry surfaces as the adhesion could be widely tuned by designing the mushroom-shaped fibril structures (28)(29)(30)(31). We compare the robot footpad without any adhesive (design 1) and that with mushroom-shaped dry adhesives (design 2) as illustrated in Fig.…”

Machine Learning-Enabled Prediction of 3D-Printed Microneedle Features

“…(E) The general workflow of the present study: After (i) designing the MNAs, and (ii) fabricating them using 3D printing followed by etching, (iii) their geometrical specifications and etching parameters processed using AI methods (iv) in order to predict the outcome of new prints, followed by (v) testing the fabricated MNAs for their skin perforation capability. microstructure to obtain desired mechanical properties [28], optimizing the energy expenditure in the 3D printing process [29], designing 3D printed surrogates that imitate the implementation of a target setup [30], maximizing adhesion of 3D printed biomimetic microfibrillar adhesives by optimizing their designs [31], and prediction of drug-releasing process from 3D printed medicines [32]. In this regard, AI techniques can also be trained with the geometrical details of MNs and their 3D printing process parameters for predicting and tuning the desired product; a realm not explored by the research teams so far.…”

“…ML is a prominent tool to understand the effect of different 3D printing parameters on the final product [ 26 ]. For example, ML was used for 3D-printability analysis in order to detect possible errors in the design [ 27 ], tuning microstructure to obtain desired mechanical properties [ 28 ], optimizing the energy expenditure in the 3D printing process [ 29 ], designing 3D printed surrogates that imitate the implementation of a target setup [ 30 ], maximizing adhesion of 3D printed biomimetic microfibrillar adhesives by optimizing their designs [ 31 ], and prediction of drug-releasing process from 3D printed medicines [ 32 ]. In this regard, AI techniques can also be trained with the geometrical details of MNs and their 3D printing process parameters for predicting and tuning the desired product; a realm not explored by the research teams so far.…”