Imagine sand that holds footprints, or crumpled paper with creases - some materials retain imprints of their history. Researchers at Yale University have shown that networks of flowing water and other liquids can also remember the past. The work was published in the journal Physical Review X.
In the laboratory of Professor Amir Pahlavan, a small chip with many microscopic channels was created. Each channel is about half the thickness of a human hair. Flexible microfibers that obstruct the fluid flow were placed inside the channels. The flow pressure can switch the fibers between two stable states, and this determines how the liquid passes through.
This is how memory of past flows forms: the current state of each fiber depends on its history, and, when fibers are coupled, also on the states of its neighbors. The two stable configurations of a fiber can be compared to zero and one - like a bit (the smallest unit of information) in computer memory. Computers usually store data in electronic devices, but here information is written in the position of flexible fibers that switch passively when the flow changes.
"Unlike traditional microfluidics (the technology of controlling fluids in microchannels), which uses active elements, this system employs microfibers for passive information storage," says lead author Abhinith Rajput, a graduate student in Pahlavan's laboratory. "The fibers can be tuned so that they interact and mutually influence each other's physical state, giving rise to complex memory of the entire system."
One possible application is soft robotics for environments where conventional electronics do not work. Fluid flow and pressure could replace electric current, fibers could replace valves, and channels could replace connections. "In many such environments, electronics can fail due to electromagnetic interference or other problems," explains Pahlavan, an associate professor of mechanical engineering. "So we want to create an all-fluidic controller that can be embedded in a soft robot and used as its brain."
A more distant question is whether such fluidic networks can be trained to perform useful tasks. This would move the work from simple information storage to physical learning, where training changes the system's response to achieve a desired result. The researchers also hope this will help understand memory and learning in some living organisms: fungi, plant roots and slime mold - a single-celled creature without a brain that can find the shortest path through a maze.
For now, this is a laboratory demonstration, not finished devices. The authors emphasize that their conclusions concern information storage in passive fibers, while the possibility of learning and use in robots remains a subject for future testing.