“`html
Although it may appear to be science fiction, researchers are striving to create nanoscale molecular devices that can be tailored for numerous uses, such as “intelligent” medications and substances. However, similar to all machines, these minute devices require a source of energy, just as electronic devices utilize electricity or living cells depend on ATP (adenosine triphosphate, the universal biological energy currency).
Investigators in the lab of Lulu Qian, a bioengineering professor at Caltech, are crafting nanoscale devices constructed from synthetic DNA, leveraging DNA’s distinctive chemical bonding characteristics to create circuits that process signals akin to tiny computers. Functioning at billionth-of-a-meter dimensions, these molecular devices can be engineered to develop DNA robots that organize cargo or act like a neural network capable of learning to recognize handwritten numerical symbols. Nevertheless, one significant challenge has persisted: how to design and energize them for various functions.
Now, Qian and former postdoctoral researcher Tianqi Song (currently an assistant professor at the University of North Carolina Greensboro) have created a method to power DNA circuits using thermal energy. Their framework refreshes itself when heated, leading to a reusable, rechargeable system that can be adapted for a variety of calculations. A paper detailing the study is set to publish in the journal Nature on October 1, 2025.
“Unlike specific fuels, heat is ubiquitous and readily available,” Qian notes. “With appropriate design, it can recharge molecular devices repeatedly, allowing them to maintain activity and continuously interact with their surroundings. Moreover, unlike chemical batteries, this recharge generates almost no waste—merely the remnants of the input signals themselves, which, in a natural habitat, would simply be recycled over time.”
The heat-recharging technique builds on a phenomenon known as a kinetic trap. Springs serve as a classic illustration of a kinetic trap—compressing a spring accumulates energy, which is then released when the spring rebounds. In a comparable manner, the DNA strands that compose the team’s system are engineered to bond in such a way that heating them allows for energy storage within the molecular connections themselves.
“Visualize two DNA strands designed to latch together, like puzzle pieces, but one is restrained by a third strand that hinders the reaction,” Song explains. “It’s akin to a spring compressed and held in place—the energy is accessible, waiting. The introduction of a catalyst strand releases the obstruction, prompting the spring to suddenly release and the DNA strands to swiftly join, unleashing the stored energy to propel the system forward. When you heat a test tube of DNA and subsequently cool it, the molecules do not always arrange themselves in their most stable configuration. Instead—and especially when they possess strong folded structures—the process of heating and cooling can reset them back to spring-loaded states, primed to release energy once more.”
By building on these two concepts—kinetic traps serving as energy reservoirs and heat as a reset mechanism—the team examined whether heat could act as a universal power source for intricate molecular circuits. In their design, the circuits perform their functions at room temperature, utilizing the energy stored in kinetic traps, like molecular “springs.” Upon task completion, the system can be recharged with a heat pulse, resetting it for the next input.
The pair demonstrated that this rechargeable approach can be utilized to energize vastly different system behaviors; in this instance, both as a neural network and as a logic circuit. These two systems represent foundational models of classical computing.
Significantly, the concept of reusability through kinetic traps is not restricted to heat. “In theory, any energy source—light, salt, or acid gradients akin to those across cell membranes—could fulfill the same function, provided it can break weak bonds among molecules, allowing them to naturally revert into their traps,” Qian states. “With this type of sustainable computation, we can begin to design molecular systems capable of not just performing tasks once but demonstrating long-term behaviors similar to those of living systems—such as learning and evolution.”
“Ultimately, such continuously functioning molecular devices—particularly those with self-directed learning and evolving capabilities—could ‘exist’ within everyday materials,” she elaborates. “Imagine a coating applied just once to an airplane, perpetually sensing stress and mending cracks to ensure passenger safety year after year. Or a pair of contact lenses that you purchase once, that rehydrate themselves and adjust to correct your vision regardless of how it changes over time. Or even a smart medication taken just once, that continues to learn to combat new diseases for a lifetime. What currently appears to be pure imagination could transform into reality if others build upon our proof-of-concept and advance this work in the forthcoming decades.”
The paper is titled “Heat-rechargeable computation in DNA logic circuits and neural networks.” Qian and Song are the authors of the study. Funding was received from Schmidt Sciences, LLC, and the National Science Foundation.
“`