Researchers at Rice University have developed a method to grow diamond films directly onto silicon-based test structures, cutting operating temperatures by 23 degrees Celsius, or about 41 degrees Fahrenheit, in laboratory tests. The technique, published on February 23, 2026, uses selective-area microwave plasma chemical vapor deposition to place diamond heat spreaders only where they are needed, a departure from blanket-coverage approaches that risk damaging delicate chip components. Because heat remains a major bottleneck limiting processor speed and battery life in consumer electronics, the researchers say a scalable diamond cooling approach could eventually influence designs ranging from smartphones to high-power RF systems.
How Selective Diamond Growth Works
The Rice team’s approach starts with patterning diamond seeds onto a substrate using one of two methods: photolithography for fine-scale features and laser-defined masking for larger patterns. Once the seeds are in place, the wafers go into a microwave plasma chamber, where carbon-rich gas breaks down and deposits onto the seeds, building solid diamond layer by layer. By growing diamond only in targeted zones, the process avoids coating entire wafer surfaces, which reduces material waste and protects temperature-sensitive metal contacts and gate structures underneath.
“This approach allowed us to scale up to a full 2-inch wafer,” a Rice researcher said. That scale matters because semiconductor fabrication depends on wafer-level processing; a technique that works on a single chip but fails at wafer scale has little commercial value. The 2-inch demonstration is still small by industry standards, where 6-inch and 8-inch gallium nitride wafers are common, but it represents a concrete step toward integrating diamond cooling into wafer-level semiconductor processing. The researchers argue that because their patterned seeding and plasma conditions rely on equipment already familiar to many fabs, the learning curve for industrial adoption could be shorter than for entirely new materials systems.
The 41-Degree Temperature Drop in Context
The headline figure of a 23 degrees Celsius reduction in device operating temperature comes from tests on silicon substrates, not yet from high-power gallium nitride transistors where the thermal problem is most severe. That distinction is critical. Silicon runs cooler than GaN power amplifiers to begin with, so a 23-degree drop on silicon, while meaningful, does not automatically predict the same benefit on a high-power RF device dissipating tens of watts per millimeter of gate width. The result does, however, confirm that selectively grown diamond films maintain enough thermal conductivity to pull heat away from active device regions without the crystal-quality penalties that plague some low-temperature diamond recipes.
Separate research on GaN transistors has produced larger temperature swings under higher thermal loads. A study in Applied Physics Letters demonstrated that diamond heat spreaders grown directly onto completed GaN high-electron-mobility transistors reduced channel temperature by an average of 111 degrees Celsius at 24 watts per millimeter. An earlier effort using a combined dry and wet etching process to integrate polycrystalline diamond onto AlGaN/GaN structures reported a more modest 25-degree reduction at 20 watts per millimeter based on electro-thermal simulation calibrated with time-domain thermoreflectance measurements. These benchmarks suggest that the magnitude of cooling scales with both the quality of the diamond film and the severity of the thermal load, which means Rice’s selective-area technique could yield much steeper drops once applied to high-power GaN devices operating near their reliability limits.
Low-Temperature Growth Protects Sensitive Chips
One of the persistent obstacles to putting diamond on finished electronics is temperature. Traditional diamond CVD runs at 700 to 900 degrees Celsius, hot enough to melt solder joints and destroy gate metals. Research published in MRS Advances showed that introducing oxygen species during the plasma process enables high-quality diamond growth at 400 degrees Celsius, roughly half the conventional temperature. That reduction opens the door to depositing diamond after a chip’s metal layers and contacts are already in place, rather than requiring diamond to be bonded on as a separate substrate, a step that introduces thermal resistance at the bonding interface and adds manufacturing cost.
The Rice team’s selective-area method builds on this low-temperature foundation. By combining patterned seeding with oxygen-assisted growth at 300 to 400 degrees Celsius, the process can be performed on substrates that already carry functional transistor layers. That compatibility is what separates this work from earlier diamond cooling demonstrations that required either bonding a pre-grown diamond slab or depositing diamond before any device fabrication, both of which limit where and how diamond can be used in a real production flow. Lower process temperatures also reduce thermal stress between the diamond and underlying materials, improving adhesion and decreasing the risk of cracks or delamination during repeated heating and cooling cycles.
Why Consumer and Defense Electronics Stand to Benefit
“This matters because heat is what limits the battery life of your phone and the speed of your computer,” a Rice researcher said. By using diamond to cool electronic components, the technique targets a constraint that affects nearly every category of semiconductor device. In smartphones, thermal throttling forces processors to slow down during sustained workloads such as gaming or video recording, undermining the performance promised on spec sheets. In 5G base stations and military radar, GaN power amplifiers generate intense heat densities that degrade reliability and cap output power, forcing designers to oversize modules and rely on bulky heat sinks or active liquid cooling. Diamond’s thermal conductivity, roughly five times that of copper, makes it the most effective passive heat spreader known, but only if it can be deposited cheaply and without damaging the device it is meant to protect.
The gap between a 2-inch wafer demonstration and a product shipping in a phone or radar module remains wide. Manufacturers would need to validate diamond growth uniformity across larger wafers, confirm long-term reliability under thermal cycling, and justify the added process cost against alternatives like silicon carbide substrates or micro-channel liquid cooling. Still, the combination of selective patterning, low deposition temperature, and measurable temperature reduction addresses three of the main objections industry has raised to diamond integration. If future work shows similar or better cooling on GaN devices under realistic power densities, the technology could enable smaller, lighter radio units, longer-lasting batteries in handhelds, and higher compute performance in thermally constrained form factors such as ultrathin laptops and augmented-reality headsets.
Next Steps Toward Scalable Diamond Cooling
For now, the Rice results function as a proof of concept that diamond can be grown where it is most useful without compromising underlying circuitry. The next logical step is to move from silicon test structures to fully processed GaN power transistors and RF amplifiers, where the thermal stakes are higher and the benefits easier to monetize. Researchers will also need to refine patterning strategies so that diamond is placed not just over obvious hot spots but in geometries that spread heat laterally toward package-level heat sinks, an optimization problem that links device physics, materials science and thermal engineering. Integrating in-situ temperature sensing or thermoreflectance mapping into the development cycle could help correlate specific diamond patterns with measured reductions in peak junction temperature.
On the manufacturing side, questions remain about throughput, yield and compatibility with existing fab chemistries. Microwave plasma reactors must be scaled to handle larger wafers while maintaining uniform plasma density, and any new process steps must fit within contamination-control rules that govern advanced semiconductor lines. Yet the fact that selective-area diamond growth has already been demonstrated on wafer-scale samples, at temperatures aligned with prior low-temperature CVD studies, gives process engineers a concrete starting point rather than a purely theoretical roadmap. If those engineering challenges can be met, diamond films grown directly on chips may shift from laboratory curiosity to standard feature in high-value electronics, quietly pulling heat away so that future devices can run cooler, faster and longer.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
