TL;DR: A study published in Nature on November 11, 2026 reports that 3D-printed polymer metamaterials built from disordered, multiscale beam networks withstood sudden impacts better than conventional ordered lattices and shielded fragile objects embedded inside them.
The paper, "Scale-rich network-based metamaterials", is open access and carries DOI 10.1038/s41586-026-11191-2. Its authors are based at MIT's Department of Mechanical Engineering, Northeastern University's Network Science Institute and Utrecht University in the Netherlands. MIT News reported the results on November 11, 2026, in a story by Jennifer Chu.
The finding reverses a long-standing convention. Metamaterials take their unusual properties from internal architecture, not from chemistry, and the field's standard method is to design one building block — a unit cell — then copy it across the material. The MIT and Northeastern team instead let beam length, beam thickness and connectivity differ by orders of magnitude within a single sample.
In physical tests, those samples reached a broad, adjustable stiffness range, held up against sudden impacts better than ordered metamaterials, and cushioned fragile objects embedded within the material.
What a metamaterial is, and why order ruled
Metamaterials are synthetic materials whose internal structure gives the whole object properties the base material lacks.
Because geometry does the work, designers have treated precision as the source of performance: one unit cell, repeated exactly, yields a periodic microstructure and, with it, better mechanical behavior.
The Nature paper describes how deeply that assumption runs. Crystalline solids are lattices of uniform bonds, and even amorphous microstructures show a narrow spread of connectivity and bond strength. Architected materials inherit the preference. Unit-cell designs enforce uniform connectivity and geometric compatibility, and plate- and shell-based architectures depend on fixed thicknesses and uniform pore sizes. The upshot is a material governed by one or a few discrete length scales.
Carlos Portela, an associate professor of mechanical engineering at MIT, has built metamaterials that way himself, repeating a single cell into square lattice structures and honeycomb architectures.
Why identical cells give way
Ordered patterns often do improve strength and resilience, Portela has found.
The weakness appears after damage. Crack or puncture an ordered lattice and the remainder tends to collapse quickly.
His explanation is uniformity. Deform one region and the cells around it, being identical, respond the same way, so the deformation travels from cell to cell. Portela says years of work on periodic microstructures turned up many limitations and non-ideal responses, that localization effect among them.
"We wanted to break free of that notion and use network science to design these materials," he says in the MIT News account.
What the team printed
The researchers made their samples with 3D printing, depositing polymer into patterns of tiny interconnecting beams. Each was a large square about the size of a credit card, with internal beams that ran from microns to centimeters.
They call the result a scale-rich metamaterial: a structure whose features vary widely in size and whose beams branch and join into one connected web rather than a repeated grid.
Two dials, a wide design space
The heterogeneity is specific rather than arbitrary. Ligament lengths, thicknesses and connectivity all vary by orders of magnitude inside the same structure, and the paper reports that two parameters are enough to produce that spread.
The authors describe the resulting model as minimal and universal — a compact set of rules that still ranges across structural features by orders of magnitude, which they present as a route past constraints that have limited metamaterial design.
Bringing network science into a solid
Network science studies the nodes and links inside a system and how those elements relate to one another and to the whole. Its subjects run from food webs, evolutionary trees and neural connections to financial markets, cellular networks, social networks and the internet.
Albert-László Barabási, a professor of physics at Northeastern University, has worked at the center of that field, mapping complex networks across disciplines and asking where they are vulnerable and how they can be made more robust. Nearly all of it involved conceptual networks mapped onto systems that already exist.
Barabási wanted to know whether network science could instead specify a physical material with properties nature does not provide, and he joined forces with Portela, whose MIT group engineers materials from the ground up.
The Nature paper argues that real networks look nothing like crystals. Protein interaction maps, social networks and global infrastructure such as the internet show node degrees spanning orders of magnitude and interaction strengths spread across broad ranges, and that heterogeneity underpins their robustness and resilience. Before this work, the authors write, no generative framework existed for architected materials in which length, thickness and connectivity all span orders of magnitude at once.
How the algorithm grows a material
Turning a graph into a solid adds constraints a network diagram never faces: the number and thickness of beams, how densely they pack, and where they may intersect.
The algorithm places a point at a random coordinate inside a domain sized to the finished material. A line runs through that point at a random angle until it reaches the domain boundary or an existing beam, forming a junction. The program then picks another point and angle, repeats the operation, and keeps adding beams into the gaps of the growing skeleton, with each generation shorter and thinner than the last.
Two parameters steer the output: the thickness of the initial beam and the rate at which thickness falls from one step to the next. Adjust them and the network changes, which gave the researchers a way to work toward different stiffness and deformation behavior.
What the tests showed
The team ran compression and impact tests alongside conventional ordered metamaterials. MIT News reports that the network-inspired samples proved more resilient to both.
The Nature paper states the measured behavior more technically: deformation that spreads through the structure instead of concentrating, with high energy absorption; more energy dissipated when loads arrive dynamically; elastic anisotropy that is widely tunable; and programmable control of elastic waves. Anisotropy here means stiffness varies with direction; delocalized deformation means strain is not confined to one collapsing band.
The protection test is easier to picture. Researchers embedded fragile objects in the material, and the network samples cushioned and protected them. MIT News also released a video comparing an ordered pattern with a network-inspired disordered one, in which the network samples held up better under compression and impact.
The biological precedent for uneven structure
The design draws on systems where uneven dimensions are functional. The paper notes that multiscale organization is a hallmark of biological systems and shapes their mechanical and transport properties, then lists examples: dragonfly wings with vein lengths from 0.1 mm to roughly 50 mm, more than three orders of magnitude; leaves whose venation spans about 10⁻¹ mm to 10³ mm; spider webs strung with threads from about 1 mm to 10³ mm; and bone, built from rod- and plate-like elements 100 μm to 1,500 μm long and 40 μm to 400 μm thick.
MIT News traces the inspiration to a leaf's vasculature, the veins in a dragonfly's wings and the virtual connections in a social network.
Where the design could be used
According to MIT News, the procedure can be used to design resilient, impact-absorbing materials such as helmets and armor, and acoustic lenses — materials whose microstructure efficiently focuses sound or vibrations. MIT's summary also names packaging among the uses where impact absorption could improve.
Those are potential applications. The experiments in the paper used printed squares, and the demonstrations centered on mechanical testing and embedded-object protection.
What the authors say is new
Nature published a companion News & Views commentary, "Extraordinary materials shaped by the principles of complex networks," by Marco Paggi on November 11, 2026, alongside the research.
Portela describes the shift as a change in approach rather than in degree. "This is a completely new way of thinking for metamaterials," he says in the MIT News report. "Unlocking this network design for materials really frees us to think of new microstructures and mechanical benefits that haven't been possible before."
The paper closes by noting that physical networks raise questions abstract ones cannot, because link length and thickness carry consequences — producing effects such as anisotropic stiffness and stress propagation — and it positions scale-rich materials as a new area of inquiry for network science as well.
By the numbers
- November 11, 2026: publication date of the study in Nature
- 2: parameters that control the generated network
- Microns to centimeters: range of beam sizes in the printed samples
- 0.1 mm to about 50 mm: dragonfly wing vein lengths cited in the paper
- 100 μm to 1,500 μm: bone element lengths cited in the paper
Related: Stanford's metamaterial reactors for electrified chemistry.