πŸ“… August 10, 2026  |  Technology

New Fuel Cell Breakthrough Could Help Power Energy-Hungry Data Centers

Engineers at Washington University in St. Louis have designed a carbon nanostructure that could make hydrogen fuel cells cheaper, longer-lasting, and powerful enough to ease the growing electricity demands of data centers.

The explosive growth of data centers across the United States is placing unprecedented strain on the nation's power grid. These sprawling facilities, which run the servers behind artificial intelligence, cloud computing, and streaming services, consume vast amounts of electricity not only to power their machines but also to keep them cool. According to the Electric Power Research Institute, data centers could account for as much as 9 percent of total U.S. electricity generation by 2030, up sharply from just 4 percent in 2023. That surge has researchers scrambling for ways to ease the pressure on an already strained grid.

One promising solution comes from a team led by Gang Wu, a professor of engineering at Washington University in St. Louis, who has developed a new approach to improve low-temperature hydrogen fuel cells. Fuel cells generate electricity by combining hydrogen and oxygen, producing only water and heat as byproducts. "If a data center is able to supply its electricity itself by using a fuel cell, it would directly convert hydrogen and other fuels into electricity, reducing the burden on the energy grid," Wu explained. The findings, published on August 6 in the journal Nature Nanotechnology, involved collaborators from Brookhaven National Laboratory, Lawrence Berkeley National Laboratory, Northeastern University, and the University of Pittsburgh.

At the heart of every fuel cell is a catalyst, a material that speeds up the chemical reaction while limiting energy losses. Platinum has long been considered one of the most effective catalysts, but because it is a rare and costly precious metal, engineers try to use as little of it as possible. One common strategy is to shrink platinum into nanoparticles, dramatically increasing its surface area so that even tiny amounts can drive powerful reactions. The catch is that these nanoparticles tend to be unstable: during operation, they can dissolve, drift out of position, or clump together, causing performance to steadily decline over time.

More recently, scientists have turned to platinum-cobalt intermetallic catalysts, prized for their superior activity and durability compared with conventional platinum alloys. Producing them, however, requires a difficult tradeoff. To keep the particles small and evenly spread, they are typically heated, or annealed, at temperatures below 700 degrees Celsius. But those temperatures are often too low to fully transform the disordered atomic structure into the highly ordered arrangement needed for peak performance. Wu's team addressed this problem by engineering a new type of carbon support: porous, hollow carbon spheres containing evenly spaced radial nanochannels. This architecture keeps the platinum-cobalt nanoparticles densely packed yet uniformly distributed, even when heated to far higher temperatures.

The results were striking. By heating the catalyst to 1,000 degrees Celsius, far above the usual limit, the researchers achieved a highly ordered atomic structure while keeping the nanoparticles smaller than five nanometers and evenly dispersed. In durability tests, the material retained 85 percent of its original performance after 150,000 severe voltage cycles, equivalent to roughly 25,000 hours of continuous operation. The carbon structure's open channels also proved useful in another way: they allowed protons, oxygen, and water to move more freely through the electrode, further boosting efficiency.

Wu believes the breakthrough could eventually help fuel cells power everything from vehicles to electricity generation, with data centers as a particularly promising application. "Through further development and collaboration with industry partners, we'll be able to solve the remaining catalyst problems and significantly advance fuel cell technologies for powering our future more efficiently and sustainably," he said. Wu has already filed a patent on the technology through his university's Office of Technology Management, and researchers hope that with continued refinement, fuel cells could soon offer data centers a practical way to generate their own clean electricity, taking pressure off an increasingly strained power grid.


μ „λ ₯ μ†Œλͺ¨κ°€ 큰 데이터센터λ₯Ό ꡬ할 μƒˆλ‘œμš΄ μ—°λ£Œμ „μ§€ 기술
μ›Œμ‹±ν„΄λŒ€ν•™κ΅ μ„ΈμΈνŠΈλ£¨μ΄μŠ€μΊ νΌμŠ€ 연ꡬ진이 κ°œλ°œν•œ νƒ„μ†Œ λ‚˜λ…Έκ΅¬μ‘°κ°€ μˆ˜μ†Œ μ—°λ£Œμ „μ§€λ₯Ό 더 μ €λ ΄ν•˜κ³  였래 μ§€μ†λ˜κ²Œ λ§Œλ“€μ–΄, κΈ‰μ¦ν•˜λŠ” λ°μ΄ν„°μ„Όν„°μ˜ μ „λ ₯ μˆ˜μš” 문제λ₯Ό ν•΄κ²°ν•  μ‹€λ§ˆλ¦¬κ°€ 될 수 μžˆλ‹€.

λ―Έκ΅­ μ „μ—­μ—μ„œ 데이터센터가 폭발적으둜 λŠ˜μ–΄λ‚˜λ©΄μ„œ μ „λ ₯망에 μ „λ‘€ μ—†λŠ” 뢀담을 μ£Όκ³  μžˆμŠ΅λ‹ˆλ‹€. 인곡지λŠ₯, ν΄λΌμš°λ“œ μ»΄ν“¨νŒ…, 슀트리밍 μ„œλΉ„μŠ€λ₯Ό λ’·λ°›μΉ¨ν•˜λŠ” μ„œλ²„λ₯Ό μš΄μ˜ν•˜λŠ” 이 κ±°λŒ€ν•œ μ‹œμ„€λ“€μ€ μž₯λΉ„λ₯Ό κ°€λ™ν•˜λŠ” 데뿐만 μ•„λ‹ˆλΌ 냉각을 μœ μ§€ν•˜λŠ” 데도 λ§‰λŒ€ν•œ μ „λ ₯을 μ†ŒλΉ„ν•©λ‹ˆλ‹€. μ „λ ₯μ—°κ΅¬μ†Œ(EPRI)에 λ”°λ₯΄λ©΄, λ°μ΄ν„°μ„Όν„°λŠ” 2030λ…„κΉŒμ§€ λ―Έκ΅­ 전체 μ „λ ₯ μƒμ‚°λŸ‰μ˜ μ΅œλŒ€ 9%λ₯Ό μ°¨μ§€ν•  수 있으며, μ΄λŠ” 2023λ…„μ˜ 4%μ—μ„œ 크게 λŠ˜μ–΄λ‚œ μˆ˜μΉ˜μž…λ‹ˆλ‹€. μ΄λŸ¬ν•œ 급증은 이미 뢀담이 큰 μ „λ ₯망의 압박을 μ™„ν™”ν•  방법을 μ—°κ΅¬μžλ“€μ΄ μ„œλ‘˜λŸ¬ 찾도둝 λ§Œλ“€κ³  μžˆμŠ΅λ‹ˆλ‹€.

μœ λ§ν•œ ν•΄κ²°μ±… ν•˜λ‚˜λŠ” μ›Œμ‹±ν„΄λŒ€ν•™κ΅ μ„ΈμΈνŠΈλ£¨μ΄μŠ€μΊ νΌμŠ€μ˜ 곡학 ꡐ수 κ°•μš°(Gang Wu)κ°€ μ΄λ„λŠ” μ—°κ΅¬νŒ€μ—μ„œ λ‚˜μ™”μŠ΅λ‹ˆλ‹€. 이듀은 μ €μ˜¨ μˆ˜μ†Œ μ—°λ£Œμ „μ§€λ₯Ό κ°œμ„ ν•˜λŠ” μƒˆλ‘œμš΄ 방식을 κ°œλ°œν–ˆμŠ΅λ‹ˆλ‹€. μ—°λ£Œμ „μ§€λŠ” μˆ˜μ†Œμ™€ μ‚°μ†Œλ₯Ό κ²°ν•©ν•΄ μ „κΈ°λ₯Ό μƒμ„±ν•˜λ©°, λΆ€μ‚°λ¬Όλ‘œλŠ” λ¬Όκ³Ό μ—΄λ§Œ λ°œμƒν•©λ‹ˆλ‹€. 우 κ΅μˆ˜λŠ” "데이터센터가 μ—°λ£Œμ „μ§€λ₯Ό μ΄μš©ν•΄ 슀슀둜 μ „λ ₯을 곡급할 수 μžˆλ‹€λ©΄, μˆ˜μ†Œ λ“±μ˜ μ—°λ£Œλ₯Ό 직접 μ „κΈ°λ‘œ μ „ν™˜ν•΄ μ „λ ₯망의 뢀담을 쀄일 수 μžˆμ„ 것"이라고 μ„€λͺ…ν–ˆμŠ΅λ‹ˆλ‹€. 이 연ꡬ κ²°κ³ΌλŠ” 8μ›” 6일 ν•™μˆ μ§€ 'λ„€μ΄μ²˜ λ‚˜λ…Έν…Œν¬λ†€λ‘œμ§€'에 κ²Œμž¬λ˜μ—ˆμœΌλ©°, λΈŒλ£©ν—€μ΄λΈκ΅­λ¦½μ—°κ΅¬μ†Œ, λ‘œλ ŒμŠ€λ²„ν΄λ¦¬κ΅­λ¦½μ—°κ΅¬μ†Œ, λ…ΈμŠ€μ΄μŠ€ν„΄λŒ€ν•™κ΅, ν”ΌμΈ λ²„κ·ΈλŒ€ν•™κ΅μ˜ ν˜‘λ ₯ μ—°κ΅¬μžλ“€μ΄ ν•¨κ»˜ μ°Έμ—¬ν–ˆμŠ΅λ‹ˆλ‹€.

λͺ¨λ“  μ—°λ£Œμ „μ§€μ˜ 핡심은 ν™”ν•™ λ°˜μ‘μ„ μ΄‰μ§„ν•˜λ©΄μ„œ μ—λ„ˆμ§€ 손싀을 μ€„μ΄λŠ” 물질인 μ΄‰λ§€μž…λ‹ˆλ‹€. λ°±κΈˆμ€ μ˜€λž«λ™μ•ˆ κ°€μž₯ 효과적인 촉맀 쀑 ν•˜λ‚˜λ‘œ μ—¬κ²¨μ‘Œμ§€λ§Œ, ν¬κ·€ν•˜κ³  κ°’λΉ„μ‹Ό κ·€κΈˆμ†μ΄κΈ° λ•Œλ¬Έμ— μ—”μ§€λ‹ˆμ–΄λ“€μ€ κ°€λŠ₯ν•œ ν•œ 적은 양을 μ‚¬μš©ν•˜λ € ν•©λ‹ˆλ‹€. ν”νžˆ μ“°μ΄λŠ” μ „λž΅μ€ λ°±κΈˆμ„ λ‚˜λ…Έμž…μžλ‘œ μͺΌκ°œμ–΄ ν‘œλ©΄μ μ„ 크게 늘렀, μ•„μ£Ό 적은 μ–‘μœΌλ‘œλ„ κ°•λ ₯ν•œ λ°˜μ‘μ„ μΌμœΌν‚€λŠ” κ²ƒμž…λ‹ˆλ‹€. λ¬Έμ œλŠ” μ΄λŸ¬ν•œ λ‚˜λ…Έμž…μžκ°€ λΆˆμ•ˆμ •ν•˜λ‹€λŠ” μ μž…λ‹ˆλ‹€. μž‘λ™ 쀑에 λ…Ήκ±°λ‚˜ μœ„μΉ˜λ₯Ό μ΄νƒˆν•˜κ±°λ‚˜ μ„œλ‘œ 뭉쳐 μ„±λŠ₯이 점차 μ €ν•˜λ  수 μžˆμŠ΅λ‹ˆλ‹€.

μ΅œκ·Όμ—λŠ” 기쑴의 백금 ν•©κΈˆλ³΄λ‹€ λ›°μ–΄λ‚œ ν™œμ„±κ³Ό 내ꡬ성을 κ°€μ§„ 백금-μ½”λ°œνŠΈ κΈˆμ†κ°„ν™”ν•©λ¬Ό 촉맀가 μ£Όλͺ©λ°›κ³  μžˆμŠ΅λ‹ˆλ‹€. κ·ΈλŸ¬λ‚˜ 이λ₯Ό μ œμ‘°ν•˜λ €λ©΄ κΉŒλ‹€λ‘œμš΄ 절좩이 ν•„μš”ν•©λ‹ˆλ‹€. μž…μžλ₯Ό μž‘κ³  κ· μΌν•˜κ²Œ λΆ„μ‚°μ‹œν‚€λ €λ©΄ 보톡 섭씨 700도 μ΄ν•˜μ—μ„œ μ—΄μ²˜λ¦¬, 즉 어닐링을 ν•΄μ•Ό ν•©λ‹ˆλ‹€. ν•˜μ§€λ§Œ 이 μ˜¨λ„λ‘œλŠ” λ¬΄μ§ˆμ„œν•œ μ›μž ꡬ쑰λ₯Ό 졜고의 μ„±λŠ₯에 ν•„μš”ν•œ κ³ λ„λ‘œ μ •λ ¬λœ ꡬ쑰둜 μ™„μ „νžˆ μ „ν™˜ν•˜κΈ° μ–΄λ ΅μŠ΅λ‹ˆλ‹€. 우 ꡐ수 μ—°κ΅¬νŒ€μ€ 이 문제λ₯Ό ν•΄κ²°ν•˜κΈ° μœ„ν•΄ μƒˆλ‘œμš΄ ν˜•νƒœμ˜ νƒ„μ†Œ 지지체λ₯Ό μ„€κ³„ν–ˆμŠ΅λ‹ˆλ‹€. μ΄λŠ” λ°©μ‚¬ν˜•μœΌλ‘œ κ· μΌν•˜κ²Œ λ°°μ—΄λœ λ‚˜λ…Έμ±„λ„μ„ κ°€μ§„ 닀곡성 쀑곡 νƒ„μ†Œ ꡬ체둜, 훨씬 높은 μ˜¨λ„λ‘œ 가열해도 백금-μ½”λ°œνŠΈ λ‚˜λ…Έμž…μžκ°€ μ‘°λ°€ν•˜λ©΄μ„œλ„ κ· μΌν•˜κ²Œ λΆ„μ‚°λœ μƒνƒœλ₯Ό μœ μ§€ν•˜κ²Œ ν•΄μ€λ‹ˆλ‹€.

κ²°κ³ΌλŠ” λ†€λΌμ› μŠ΅λ‹ˆλ‹€. 촉맀λ₯Ό κΈ°μ‘΄ ν•œκ³„λ₯Ό 훨씬 λ›°μ–΄λ„˜λŠ” 섭씨 1,000λ„κΉŒμ§€ κ°€μ—΄ν•¨μœΌλ‘œμ¨, 연ꡬ진은 5λ‚˜λ…Έλ―Έν„°λ³΄λ‹€ μž‘μ€ λ‚˜λ…Έμž…μžλ₯Ό κ³ λ₯΄κ²Œ λΆ„μ‚°μ‹œν‚€λ©΄μ„œλ„ κ³ λ„λ‘œ μ •λ ¬λœ μ›μž ꡬ쑰λ₯Ό μ–»μ–΄λƒˆμŠ΅λ‹ˆλ‹€. 내ꡬ성 ν…ŒμŠ€νŠΈμ—μ„œ 이 μ†Œμž¬λŠ” 15만 회의 κ·Ήμ‹¬ν•œ μ „μ•• 반볡 후에도 μ›λž˜ μ„±λŠ₯의 85%λ₯Ό μœ μ§€ν–ˆμœΌλ©°, μ΄λŠ” μ•½ 25,000μ‹œκ°„μ˜ 연속 μž‘λ™μ— ν•΄λ‹Ήν•˜λŠ” μˆ˜μΉ˜μž…λ‹ˆλ‹€. 이 νƒ„μ†Œ ꡬ쑰의 μ—΄λ¦° 채널은 또 λ‹€λ₯Έ μž₯점도 μžˆμ—ˆμŠ΅λ‹ˆλ‹€. μ–‘μ„±μž, μ‚°μ†Œ, 물이 전극을 톡해 더 자유둭게 이동할 수 있게 ν•˜μ—¬ νš¨μœ¨μ„ ν•œμΈ΅ λ†’μ˜€μŠ΅λ‹ˆλ‹€.

우 κ΅μˆ˜λŠ” 이 기술적 도약이 ꢁ극적으둜 μžλ™μ°¨λΆ€ν„° μ „λ ₯ 생산에 이λ₯΄κΈ°κΉŒμ§€ λ‹€μ–‘ν•œ λΆ„μ•Όμ—μ„œ μ—°λ£Œμ „μ§€λ₯Ό ν™œμš©ν•˜λŠ” 데 도움이 될 κ²ƒμœΌλ‘œ 보며, 데이터센터가 특히 μœ λ§ν•œ μ‘μš© λΆ„μ•Όκ°€ 될 수 μžˆλ‹€κ³  λ§ν•©λ‹ˆλ‹€. κ·ΈλŠ” "산업계 ν˜‘λ ₯μ‚¬λ“€κ³Όμ˜ 지속적인 개발과 ν˜‘λ ₯을 톡해 남아 μžˆλŠ” 촉맀 λ¬Έμ œλ“€μ„ ν•΄κ²°ν•˜κ³ , 우리의 미래λ₯Ό 더 효율적이고 지속가λŠ₯ν•˜κ²Œ λ§Œλ“œλŠ” μ—°λ£Œμ „μ§€ κΈ°μˆ μ„ 크게 λ°œμ „μ‹œν‚¬ 수 μžˆμ„ 것"이라고 λ§ν–ˆμŠ΅λ‹ˆλ‹€. 우 κ΅μˆ˜λŠ” 이미 μžμ‹ μ˜ λŒ€ν•™ κΈ°μˆ κ΄€λ¦¬μ²˜λ₯Ό 톡해 이 κΈ°μˆ μ— λŒ€ν•œ νŠΉν—ˆλ₯Ό μΆœμ›ν–ˆμœΌλ©°, μ—°κ΅¬μžλ“€μ€ 지속적인 κ°œμ„ μ„ 톡해 μ—°λ£Œμ „μ§€κ°€ κ³§ 데이터센터에 슀슀둜 μ²­μ • μ „λ ₯을 생산할 μ‹€μš©μ μΈ 방법을 μ œκ³΅ν•˜μ—¬, 점점 뢀담이 μ»€μ§€λŠ” μ „λ ₯망의 압박을 λœμ–΄μ€„ 수 있기λ₯Ό κΈ°λŒ€ν•˜κ³  μžˆμŠ΅λ‹ˆλ‹€.


πŸ“š 였늘의 핡심 μ–΄νœ˜ (8~10개)

strain
λΆ€λ‹΄, μ••λ°•
catalyst
촉맀
nanoparticle
λ‚˜λ…Έμž…μž
durability
내ꡬ성
anneal
(μ—΄μ²˜λ¦¬λ‘œ) ν’€λ¦Όμ²˜λ¦¬ν•˜λ‹€
intermetallic
κΈˆμ†κ°„ν™”ν•©λ¬Όμ˜
disperse
λΆ„μ‚°μ‹œν‚€λ‹€, 흩어지닀
tradeoff
절좩, κ· ν˜•μ 
porous
λ‹€κ³΅μ„±μ˜, ꡬ멍이 λ§Žμ€
precious metal
κ·€κΈˆμ†