CU Boulder researchers built it from interlocking staple shaped particles. A stronger vibration unlocks the network; a gentler, tuned one re locks it. Engineers can now design for temporary structures.
A free-standing arch, roughly the size of a coffee cup, sits on a lab bench at the University of Colorado Boulder. Made from hundreds of small plastic staples, each shaped like the letter "U" with legs splayed like a crown, the arch behaves like a single rigid object. Tilt the table and it does not slump. Pick it up and it stays whole. Apply a stronger vibration at a different frequency, and the arch dissolves into a pile of loose, sliding pieces.
The demonstration, from the lab of mechanical engineer Francois Barthelat at CU Boulder's Paul M. Rady Department of Mechanical Engineering, is the most tangible expression of a new design principle in granular materials: a substance whose strength is not a fixed property but a choice, written into the geometry of its particles and into the vibration an operator applies.
Barthelat's group, working with PhD students Youhan Sohn and Saeed Pezeshki and co-author V. Fouquet, has spent several years engineering two-legged, staple-shaped particles that maximize the way neighboring pieces tangle. The shape was not arbitrary. The team used Monte Carlo simulations to screen candidate particle geometries, then validated the best performer in physical, likely 3D-printed specimens in "pickup" tests that became the experimental core of the work. The result, reported in the peer-reviewed Journal of the Mechanics and Physics of Solids and in a companion study in Granular Matter, is a granular solid that combines high tensile strength with high toughness, a property pair conventional monolithic materials rarely deliver together.
What makes the new material different from a bag of paper clips is control. Gentle vibration, tuned to a specific frequency and amplitude, lets the particles settle into a deeply interlocked, stiffer network. Stronger vibration at a different setting forces the same particles to slide past one another and unravel into a flowing pile. The researchers describe the in-between state as "not quite solid, not a liquid," a regime an arXiv preprint from December 2024 previewed before the journal versions appeared.
That reversibility is the actual story, not the swarm robotics. The press release from CU Boulder, summarized by ScienceDaily on June 15, 2026, gestures at recyclable bridges, buildings that dismantle themselves, and swarms of small robots that entangle to perform a task and then come apart. Barthelat has reached for a vivid analogy, comparing the long-term vision to the T-1000 liquid-metal antagonist from Terminator 2. None of those applications exist as engineered systems. The demonstrated work is at particle scale, in physical specimens roughly the size of a coffee cup. Cost, scaling, and integration into real structures or robots remain open problems the researchers themselves flag.
The mechanism, in other words, is genuinely new. The use cases are not yet real. A reader should hold those two facts separately.
The design principle the work points toward is closer to a shift in mental model than a product roadmap. Today, a structural engineer's job is to choose a material that will hold. A future built from the kind of granular metamaterials Barthelat describes would let an engineer specify, on the drawing, whether a structure is meant to hold indefinitely, to hold for a season, or to come apart on a particular signal. Temporary shelters that disassemble for transport, deployable bridges that release under remote command, robotic swarms that change shape by re-engaging rather than rewiring: each becomes a problem of particle geometry and frequency tuning rather than a search for a new alloy.
The lab is already iterating. The current staple, two-legged, was the winner of the Monte Carlo screen. The next candidate is a burr-shaped, multi-legged particle, named for the spiky seedpods that cling to clothing, designed to tangle more aggressively than a U-shape can. If the principle scales, the question will shift from whether engineers can build a material that comes apart on cue to whether they can build one that comes apart only on cue, with the timing, the trigger, and the recovery all specified in advance.
For now, what sits on a Boulder bench is a small arch of crown-leg staples that holds, releases, and re-locks under a researcher's hand. That is the proof. The buildings are still drawings.