In the ever-evolving landscape of materials science, a team of researchers from Duke University, Southwest Jiaotong University, Tsinghua University, and other institutions has made a groundbreaking leap. Their recent study, published in Science Advances on January 23, 2026, introduces a “digital composite” that redefines the possibilities of material adaptability. Unlike traditional composites with fixed phase architectures, this innovative material can be reprogrammed at the voxel level to switch between solid and liquid states, opening doors to a new era of flexible, functional materials for soft robotics, adaptive structures, and beyond.
The Limitation of Traditional Composites: Why We Needed a Change
For decades, composite materials have been the backbone of countless industries, from aerospace to biomedical engineering. Their performance is largely determined by the spatial arrangement of their phases—regions with distinct properties like composition or physical state. However, there’s a major flaw in conventional composites: once their phase architecture is formed during manufacturing, it’s fixed. This rigidity limits their ability to adapt to changing environments or functional needs.
Existing strategies to reconfigure materials often rely on geometric deformation, which can only achieve limited adjustability. Most fail to provide voxel-addressable, non-volatile control in 3D structures—a critical requirement for bulk materials that need multidirectional tuning. This gap left a significant void in materials science, waiting for a solution that could merge programmability, stability, and scalability.
The Innovation: A Digital Composite That “Writes” Its Own Phase Architecture
The research team’s answer to this challenge is a digital composite that operates like a high-tech hard disk. Each tiny voxel (a 3D pixel) in the material is an elastomeric chamber filled with a liquid metal composite (LMC)—a mix of 95.2% gallium and 4.8% iron particles. What makes this LMC special is its ability to switch between solid and liquid states in seconds using electrical signals, with no need for continuous energy to maintain its phase. It’s a true “erase-write-store” cycle, just like saving or deleting data on a hard drive.
The Science Behind the Switch
The LMC’s phase transitions are controlled by a carefully engineered thermal hysteresis window. By adjusting the iron content and the surface chemistry of the voxel’s chamber (made of gold-coated polydimethylsiloxane, or PDMS), the researchers tuned the LMC’s freezing and melting points. This ensures the material stays stable in either solid or liquid form at room temperature, while still allowing fast, energy-efficient phase changes when needed.
To control each voxel individually, the team integrated a flexible, stretchable electrode array on the surface of the PDMS chambers. This array acts as a resistive network, enabling voxel-specific Joule heating (for melting) and temperature sensing. A multiplexed circuit—powered by components like PWM drivers, MOSFETs, and a Raspberry Pi—delivers precise control, with the minimum reprogramming time clocking in at just 3.02 ± 0.44 seconds.
Scalability: From 2D Arrays to 3D Structures
One of the most exciting aspects of this digital composite is its scalability. The researchers demonstrated that 2D arrays (such as a 9×9 voxel grid) can be stacked into 3D structures using modular building blocks. Each basic 3×3×3 voxel unit is connected using a thin, reversible PDMS adhesive (150 μm thick) that can withstand 100 assembly-disassembly cycles without damage. This modular design allows for the creation of large, free-form structures—like a 6×6×6 cube or a 3×3×30 beam—opening up possibilities for real-world applications.
Experimental Results: Proving Programmability and Performance
The team’s experiments left no doubt about the material’s capabilities. They tested thousands of configurations, proving that the digital composite can be fine-tuned to exhibit a wide range of mechanical properties.
Reprogrammable Viscoelasticity
In 2D tests, the researchers characterized 60,475 different configurations of a 9×9 voxel sample in just one week—with 10,115 of these being nearly isotropic (uniform in all directions). They were able to independently adjust key viscoelastic parameters: Young’s modulus, Poisson’s ratio, shear modulus, and loss factor. The experimental results matched finite element analysis (FEA) predictions with a maximum deviation of only 14.6%, and the material could even mimic the viscoelastic behavior of six commercially available materials.
Reversible Plasticity
Unlike traditional materials, which suffer permanent deformation after reaching their yield point, the digital composite exhibits “reversible plasticity.” After stretching a sample to 6% strain, the team simply melted all voxels to reset the material to its original state. Even after 100 cycles, the stress-strain curves remained consistent—proving the material’s durability.
Custom Stress-Strain Curves
Using an inverse design approach, the researchers programmed the composite to generate specific stress-strain responses. By cycling through 40 different phase architectures, they created conventional behaviors (elastic, plastic, viscoelastic) and even unconventional ones—like a heart-shaped curve. The experimental results matched the target curves with a maximum stress deviation of just 11.8%.
Strain Localization Mitigation
The composite also proved effective at reducing strain localization—an issue that weakens materials under load. By optimizing voxel configurations, the team reduced the maximum principal strain by over 20% under pure tension, pure shear, and combined loading—all while keeping the material’s elastic properties unchanged.
Real-World Application: A Reprogrammable Robotic Fish Tail
To showcase the material’s practical potential, the researchers integrated a 3×3×30 voxel beam into a robotic fish as a tail appendage. By reprogramming the beam’s voxel configuration, they achieved three distinct swimming behaviors:
- Straight-line swimming: When solid voxels were positioned far from the tail’s actuation point, the fish moved efficiently forward.
- Turning: Eccentrically placed solid voxels created asymmetric vibration, making the fish spiral in a circular path.
- No movement: A helical arrangement of solid voxels converted propulsive energy into torsion, leaving the fish stationary.
This demonstration highlights how the digital composite could revolutionize soft robotics—allowing machines to adapt their movements on the fly without complex mechanical modifications.
The Future: Expanding the Horizons of Reprogrammable Materials
While the current prototype relies on an external cooling system for reprogramming, the team has plans to integrate active cooling modules to make the material fully untethered. They also aim to extend the technology to other phase-change materials, tailoring the thermal hysteresis window for different operating temperatures.
In the long term, the digital composite could become a platform for data-driven materials research. By combining it with AI, researchers could accelerate the inverse design of complex mechanical behaviors. The material’s programmability could also be extended beyond mechanics—incorporating optics or magnetics to create multifunctional systems. Imagine adaptive buildings that adjust their stiffness to withstand storms, or medical devices that change shape to fit different body parts—all made possible by this breakthrough.