A programmable material built from modular units is offering a new way to think about how robots adapt to their surroundings. Researchers at Duke University have demonstrated a system in which mechanical properties can be written, erased and rewritten within solid building blocks, allowing structures to shift stiffness and damping without changing their overall shape.
Bai, Y., Yuan, X., Weng, Y., Yin, K., Wang, H., & Ni, X. (2026). Digital composites with reprogrammable phase architectures. Science Advances, 12(4). https://doi.org/10.1126/sciadv.aed9698
The work, led by Assistant Professor Xiaoyue Ni and first author Yun Bai, a doctoral researcher in mechanical engineering and materials science, was recently reported in Science Advances. It outlines what the team describes as a digital composite material with reprogrammable phase architectures. Rather than relying on geometry alone to determine performance, the approach embeds change directly into the internal state of the material.
Assistant Professor Xiaoyue Ni from Duke University stated,
“Our goal is to eventually construct larger systems using the composite materials. We want to build flexible, programmable materials for robotics that can enable them to perform a wide variety of tasks in a wide variety of environments.”
At the core of the system are small cells filled with a composite of gallium and iron. At room temperature, this mixture can exist either as a solid or a liquid. By applying localized heating through electrical current, individual cells can be switched from solid to liquid in specific patterns. Once set, those patterns remain stable without continuous energy input. Cooling the structure resets the cells to a uniform solid state, allowing the process to begin again.
In two dimensional tests, the researchers fabricated thin sheets composed of hundreds of addressable cells. By selectively liquefying certain regions, they were able to tune stiffness and damping across the surface while preserving the overall geometry. Mechanical testing showed that the sheets could approximate the behavior of a range of conventional soft materials, from relatively rigid plastics to compliant elastomers. The result is a material whose effective properties are not fixed at manufacture but determined by how its internal phases are programmed.
The concept becomes more tangible in three dimensions. The team constructed cube shaped modules, each resembling a small lattice of 27 individually controllable cells. These cubes can be mechanically connected in different configurations, similar to interlocking blocks. When assembled into a column and attached to a motorized base, they functioned as a programmable tail for a robotic fish. With identical motor input, different internal solid liquid patterns in the cubes produced distinct swimming trajectories. In effect, the same actuator generated different behaviors depending on how the material was programmed.
This approach differs from soft robotics strategies that depend primarily on new polymers or complex external control systems. Here, the emphasis is on internal reconfiguration. By storing mechanical states in the material itself, the structure acts as both component and parameter set. The team likens the process to writing binary information into hardware, except the output is mechanical response rather than digital data.
The researchers suggest that future iterations could incorporate alternative metals with tailored melting points, expanding the operating temperature range. In principle, this could enable applications in environments where thermal constraints are strict, including biomedical contexts. The possibility of miniaturization is also under consideration. If scaled down, similar programmable composites might navigate confined spaces such as small fluidic channels or assist in adaptive medical devices that adjust stiffness after deployment.
The study builds on a broader trend in materials science that seeks to blur the boundary between structure and function. Additive manufacturing has made it possible to design complex geometries with predetermined properties, but modifying those properties typically requires reprinting the part. By contrast, this system aims to separate fabrication from configuration. The same set of blocks can be reused, with their mechanical characteristics altered through controlled phase changes.
Significant engineering challenges remain. Thermal management, response time and long term cycling durability will influence how such materials perform outside the laboratory. Integration with existing robotic platforms will also require attention to power delivery and control architecture. Even so, the demonstration points toward a class of robotic systems that rely less on swapping components and more on reprogramming matter itself.
For robotics research, the implication is straightforward. Adaptability does not have to come solely from software updates or new hardware modules. It can be embedded into the material substrate. The Duke team’s work suggests that future machines may adjust their mechanical identity as readily as they adjust their code.
Adrian graduated with a Masters Degree (1st Class Honours) in Chemical Engineering from Chester University along with Harris. His master’s research aimed to develop a standardadised clean water oxygenation transfer procedure to test bubble diffusers that are currently used in the wastewater industry commercial market. He has also undergone placments in both US and China primarely focused within the R&D department and is an associate member of the Institute of Chemical Engineers (IChemE).
