The recent explosion in energy-intensive artificial intelligence projects presents (many) problems, including how to keep them cool. The data centers needed to run them produce massive amounts of heat, and require substantial A/C systems to maintain them at functional temperatures. Vapor compression refrigeration—the method still most often found in cars, buildings, and factories—is commonly used to meet these cooling demands. But vapor compression still frequently relies on environmentally harmful chemical refrigerants.
While there are a number of promising, recent advancements in refrigeration, thermogalvanic cells have not really been seriously considered. If anything, the method’s underlying principles are much better suited for heating than cooling. In thermogalvanic cells, heat produced by reversible electrochemical reactions is employed to generate electrical power by leveraging the entropy transport from a high temperature source to a lower temperature sink. While reverse thermogalvanic systems are feasible, experts long believed the option was simply too cost-ineffective and weak to justify its use.
But thanks to some recent chemical alterations, researchers believe thermogalvanic refrigeration isn’t just possible—it could find its way into everything from small, wearable cooling gadgets, to household A/C, to those massive AI data centers. The advances are detailed in a study published on January 30th in the journal, Joule, from a team at China’s Huazhong University of Science and Technology.
“While previous studies mostly focus on original system design and numerical simulation, we report a rational and universal design strategy of thermogalvanic electrolytes, enabling a record-high cooling performance that is potentially available for practical application,” Jiangjian Duan, the study’s senior author, said in a statement on Thursday.
Existing thermogalvanic cells rely on what are known as electrochemical redox reactions using dissolved iron ions. In the first step, iron ions lose an electron and absorb heat in the process. From there, they later gain another electron to subsequently release heat. As the first reaction’s power cools its surrounding electrolyte solution, a heat sink removes the accompanying warmth.
Duan’s team, however, opted for an electrochemical solution that uses hydrated iron salt containing perchlorate. Doing so allows the iron ions to better dissolve and dissociate—when dissolved in a nitrile-infused solvent instead of simply water, the team’s cell cooling power improved by around 70 percent. Past studies have achieved a thermogalvanic cooling factor of around 0.1 K (32.18 F), but Duan’s iteration allowed for a cooling factor of about 1.42 K (34.55 F). A couple degrees Fahrenheit might not seem like a lot in the grander scheme of things, but given the cost efficiency and scalability, the improvement represents a promising leap in thermogalvanic cooling potential.
“Though our advanced electrolyte is commercially viable, further efforts in the system-level design, scalability, and stability are required to promote the practical application of this technology,” Duan said.
Moving forward, the team intends to introduce new physical designs and advanced materials to improve overall cooling potential. They are also planning to begin work on functional refrigerator prototypes for actual use in manufacturing centers—potentially even those problematic AI data centers.