NC State Uses Machine Learning and Laser to Make Solvent-Free Super-Repellent Elastic Materials

A surface that repels water is useful. A surface that repels virtually any liquid - concentrated acids, organic solvents, bases - while also stretching to five times its original length without losing that property is considerably more interesting to engineers. The challenge has been manufacturing such materials without harsh chemical processes and without the surface coatings delaminating under mechanical stress.

Researchers at North Carolina State University have combined machine-learning optimization with CO2 laser ablation to solve both problems simultaneously. Their approach produces superomniphobic siloxane elastomers - materials that retain near-perfect liquid repellency at 400% strain and across more than 5,000 stretch-release cycles. No chemical solvents are needed. The work appears in Matter.

The delamination problem and how previous work addressed it

Most superomniphobic surfaces are made by spray-coating a substrate with a solvent containing hydrophobic nanoparticles. The coating creates the nano-scale surface roughness that gives the material its repellent properties. The problem is adhesion: when the substrate is stretched, the coating cracks and peels away. Standard spray-coated superomniphobic surfaces typically fail beyond 100% elongation.

Arun Kumar Kota's group at NC State had previously addressed this by introducing microprotrusions - tiny pillars 10 to 100 micrometers across - onto the substrate surface before spray coating. When stretched, the coating between pillars delaminated, but the coating on top of the pillars remained intact, preserving repellency. That approach extended the working range to five times the initial length.

The new work eliminates the spray coating entirely. Instead, laser ablation creates both the microprotrusions and the rough surface texture that generates superomniphobicity in a single step.

Where machine learning enters the process

Laser ablation involves three key parameters that affect surface texture: laser power, scan speed, and spatial frequency - how many times the laser pulses per unit length. The interactions between those parameters and the resulting surface properties form a multidimensional space with millions of possible combinations. Finding the optimal settings through manual trial-and-error would be prohibitively slow.

The team trained a machine-learning algorithm on the three laser parameters together with the desired sliding angle - a measure of how easily liquids roll off the surface. The algorithm predicted which combination would produce a superomniphobic surface without requiring exhaustive experimental search. Guided by those predictions, the researchers experimentally verified a set of conditions that worked.

"In this work, instead of spray coating, we use laser ablation to create both the microprotrusions and the rough surface that creates superomniphobicity," Kota said. "We have created a platform for creating stretchable superomniphobic materials without the use of chemical solvents and without needing hundreds of thousands of trial-and-error experiments."

Material specifications and testing

The substrate was a siloxane elastomer modified with a fluorocarbon silane - siloxane for its stretchability, fluorocarbon modification for its inherent hydrophobic character. After laser ablation under machine-learning-determined parameters, the resulting surface retained superomniphobic properties at 400% strain - four times the initial length - and maintained that performance through more than 5,000 stretch-release cycles. The material also held its properties under diverse deformation modes beyond simple uniaxial stretching.

Systematic experiments characterized how elongation affected contact angles, breakthrough pressures, and sliding angles - the three key metrics for assessing repellency strength and practical utility.

Applications and limitations

Superomniphobic stretchable materials serve multiple niches. Soft robots operating in chemically harsh environments need surfaces that can handle unexpected fluid contact. Wearable wound dressings benefit from barrier properties against bodily fluids. Stretchable electronics need protection from corrosive environments. The solvent-free manufacturing approach also offers environmental advantages over spray-coating processes that use and dispose of organic solvents at scale.

Current limitations include the substrate choice - the siloxane elastomer with fluorocarbon modification may not be appropriate for all applications, and whether the laser ablation parameters transfer to other elastomeric materials requires additional testing. The durability under repeated chemical exposure, not just mechanical cycling, was not the primary focus of this publication. Scaling the laser processing to large substrate areas for manufacturing also needs further development.

The research was supported by the NSF (award 2245427), NIH (R21EB033960 and R01HL166724), and the Congressionally Directed Medical Research Programs (HT94252310663).