Processing diamond surfaces in the rough

Artificial intelligence (AI) chips are hot stuff — literally and figuratively. While countries clamor to build AI infrastructure, engineers iterate on chips’ tendency to overheat. Diamond, the best-known natural thermal conductor, is a compelling built-in coolant. But achieving maximum thermal dissipation using diamond requires better diamond processing techniques — a difficult task, since diamond is also the hardest naturally occurring material.

Lian et al. used femtosecond lasers to alter the atomic structure of diamond surfaces. They found a fundamental engineering tradeoff: while high-power lasers increased the hydrophilicity of diamond — a crucial property for use in liquid-cooling environments — they also introduced more surface defects at the atomic level, which reduced the material’s thermal conductivity.

“In the past, people only paid attention to those macroscopic features, where wettability is the most important parameter,” said author Yang Lu. “We revealed that the surface atomic structure is another important parameter that cannot be ignored.”

To assess the atomic-scale impacts of femtosecond laser processing, the researchers tuned their laser beam to within micrometers of the diamond target. Then, they snaked the laser over the diamond’s surface atomic line by line and found more atomic defects with increasing laser power. Lastly, the team characterized the surface roughness of their sample and measured its thermal conductivity and hydrophilicity. Given the fundamental trade-off between wettability and thermal conductivity, the researchers determined the optimal laser power density was 3.77 Joules per square centimeter.

In the future, the researchers hope to remove laser-induced defects, enabling tuning diamond surfaces solely for wettability. Their “ultimate dream” is to design a fully diamond semiconductor that removes cooling requirements.

Source: “Tuning atomic structure of single-crystalline diamond surfaces by femtosecond laser for enhanced heat transfer,” by Yiling Lian, Zichen Zhang, Misheng Liang, Xun Zhao, Kefan Guo, Jiayi Li, Zheling Li, and Yang Lu, Applied Physics Letters (2026). The article can be accessed at https://doi.org/10.1063/5.0314498 .

This paper is part of the Thermal Properties of Graphene and Carbon Materials—From Physics to Applications in Thermal Management Collection, learn more here .