A liquid-metal gripper gently moves fish embryos and flicks out a hair in milliseconds

A deformable blob of liquid metal can handle delicate specimens and actively eject a tiny object. Researchers including teams at Soochow University and the University of Science and Technology of China published their LiMU gripper in Nature Communications on March 6, 2026. A preprint had appeared on February 4, 2025. The distinctive idea is a gripper that can switch between gentle handling and active release.

Robots have long had alternatives to fingers. In 2010, Cornell, the University of Chicago and iRobot demonstrated a balloon filled with ground coffee. The bag conformed around an object; removing air jammed the particles together, and releasing the vacuum softened it again. That precedent matters: changing a material's behavior to obtain a versatile grip is not itself a new invention.

LiMU addresses the other half of the operation. At very small scales, successfully picking something up does not guarantee that it will detach when wanted. Electric and magnetic fields regulate the functional liquid metal's surface tension and stiffness. Changing its surface state enables active release. Think of surface tension as a tendency for the droplet's surface to pull inward: the material supplies the recoil rather than a separate mechanical spring.

Soochow describes a hair, 70 micrometers in diameter, accelerating to 696 millimeters per second within six milliseconds, with peak acceleration of 42 g. In a separate, gentle-handling demonstration, manipulation of 200 zebrafish embryos achieved 99 percent success. These are distinct tests. Rapid hair ejection should not be confused with the way embryos were transferred, and the embryo result does not establish safety for every biological specimen.

The paper reports contact pressure around 10 pascals and operation on differently shaped targets in both air and liquid. The interesting engineering combination is conformity, retention and controlled release, rather than simply a stronger mechanical grip. For miniature assembly, releasing an item correctly can be as consequential as holding it securely.

Practical adoption still needs evidence about repeated operation, contamination, switching between objects and release placement. Those are questions for deployment, not capabilities assumed from a demonstration. The prospect is compelling because a programmable material need not imitate the human hand to solve manipulation problems that fingers handle poorly.

Preprint first posted February 4, 2025; journal paper published March 6, 2026; Chinese team account published March 10.