UCLA engineers have added a new ingredient to solve the problem of overheated computer processors: boron arsenide.
The novel semiconductor material, successfully integrated into high-power computer chips by the UCLA team, increases energy efficiency in computers and offers high levels of thermal protection.
The research, published recently in Nature Electronics , was led by Yongjie Hu, an associate professor of mechanical and aerospace engineering at the UCLA Samueli School of Engineering.
Hu and his team began their work in 2018, developing defect-free boron arsenide in their lab . The chemical compound, the researchers discovered in those early days, performed more effectively in drawing and dissipating heat than other known metal or semiconductor materials such as diamond and silicon carbide.
Now, Hu and the team have successfully demonstrated the material’s effectiveness by integrating the compound into high-power computer chips — devices with wide bandgap transistors made of gallium nitride known as high-electron-mobility transistors (HEMTs).
When running the processors at near maximum capacity, chips that used boron arsenide as a heat spreader demonstrated a maximum heat increase from room temperatures to nearly 188 degrees °F.
Chips that use diamond to spread heat, by contrast, raise temperatures to approximately 278 degrees °F. Silicon carbide, similarly, leads to a heat increase of about 332 °F.
The successful integration opens up a path for industry adoption of the technology, according to the lead researcher.
“These results clearly show that boron-arsenide devices can sustain much higher operation power than processors using traditional thermal-management materials,” Hu said. “And our experiments were done under conditions where most current technologies would fail."
According to Hu, boron arsenide is ideal for heat management because the material not only exhibits excellent thermal conductivity but also displays low heat-transport resistance.
“The key feature in our boron arsenide material is its very low thermal-boundary resistance," said Hu. "This is sort of like if the heat just needs to step over a curb, versus jumping a hurdle.”
In a short Q&A with Tech Briefs below, Hu explains why the integration could lead to a variety of new electronics opportunities.
Read Hu's edited responses below.
Tech Briefs: What inspired you to try boron arsenide?
Yongjie Hu: In general, I like the research that creates fundamental novelty and transforms to real technologies. For this particular study on boron arsenide, we made progress step by step to achieve the goal, building on tremendous efforts from my research group at UCLA and other groups in our field.
Tech Briefs: What makes this material novel, especially compared to the technologies available currently?
Yongjie Hu: These novel materials are not widely studied, and their high-quality crystals do not exist in nature. In our previous work (Science 361, 575, 2018; Nano Letters 17, 7507, 2017), we developed the experimental synthesis of boron arsenide (BAs) and boron phosphide based on ab initio [from-the-beginning] theory predictions, and measured record-high thermal conductivity up to 1300 W/mK.
In particular, due to the unique band structure, BAs have the highest thermal conductivity among common semiconductors and metals — over 3 times that of the current industrial standard for high thermal conductivity materials (copper or silicon carbide). In addition, our systematic characterizations (Applied Physics Letters 115, 122103, 2019) show that the semiconducting nature of BAs provides compatible properties and advantages in device manufacturing and integration, as exemplified in our recent studies for cooling wide-bandgap high power electronics (Nature Electronics 4, 416, 2021) and demonstrating flexible thermal interfaces for wearable electronics and flexible robots (Nature Communications 12, 1284, 2021).
Tech Briefs: How much can this material cool down a device?
Yongjie Hu: The thermal management performance using BAs exceeds the state-of-the-art technologies, such as those based on diamond or silicon carbide cooling substrates. For the same transistor operation power, the device hot spot temperature is much lower: For example, experimental measurement shows that operational AlGaN/GaN high-electron-mobility transistors (HEMTs) using boron arsenide as a heat spreader has a significantly lower hot-spot temperature rise of 60 °C, while diamond and SiC devices shows substantially higher temperature rise of 110 °C and 140°C respectively. This study shows that BAs devices can sustain much higher operation power than other traditional thermal-management technologies, which allows us to push the performance limit for high power and high frequency devices.
Tech Briefs: How do you integrate this boron arsenide material into a device? Is it simple? Challenging?
Yongjie Hu: We have developed multiple approaches, including epitaxial growth and chip-level bonding. The major challenge is to improve the interface quality and reduce thermal boundary resistance, with the consideration of varied crystal structures across the electronic junctions. In the reported specific case, BAs is used as a cooling substrate that holds GaN HEMTs chips with a low thermal boundary resistance of 4 GW/mK — over 8 times better of typical interface of GaN with diamond or SiC.
Because of the low thermal boundary resistance and high thermal conductivity, the integrated devices demonstrated the very high cooling performance beyond other technologies. The heterogeneous interface is characterized by high-resolution transmission electron microscopy and shows atomic-level uniformity (see the device interface with atomic resolution to the left).
Tech Briefs: Where will you test this next? What are you working on now?
Yongjie Hu: We have many exciting on-going efforts for both new materials development and device applications. For example: How to tailor these new building blocks for different device applications and frameworks? Another recent work successfully developed assembling these materials as flexible thermal interfaces for wearable electronics and flexible robots (Nature Communications 12, 1284, 2021). We have also been looking for new candidates by theory (Phys. Rev. B 103, L041203, 2021) and how to experimentally realize them. Importantly, we are always looking for new opportunities in interdisciplinary areas and welcome collaborations.
What do you think? Share your questions and comments below.