The Thermal Management Equation
03/01/2007
By Meredith Courtemanche, assistant editor
Theories on thermal management rival, in number and variety, strategies for winning the Super Bowl or baking the perfect loaf of bread. Will running the ball wear down the defense? Should you knead the dough with one hand or two? Will a focus on design preempt debate on substrate materials? Every thermal management approach is designed to mitigate excess heat generated by electronic components crowding into denser, smaller packages and devices. Speaking to members of IMAPS New England chapter and the American Ceramics Society (ACerS), Jonathan Margalit, Ph.D., developer of new business for thermal management-related applications at H.C. Starck Inc. (Goslar, Germany), presented an elegant equation to sum up thermal management and ground Starck’s advances in the field with refractory metals. Margalit focused on molybdenum (Mo) and tungsten (W), materials that he saw as a good match for silicon (Si).
 Jonathan Margalit, Ph.D., developer of new business at H.C. Starck, Inc. |
Refractory metals, by definition, withstand high heat, due to their high melting temperatures. Low thermal-expansion rates with these dense metals make them an attractive match for glass and ceramics. Tungsten offers a melting point of 3400°C; molybdenum melts at about 2625°C. Since both are fairly good electrical conductors, they make good thermal conductors. But the most compelling argument for a thermal management product based on refractory metals is the thermal compatibility factor: the coefficient of thermal expansion (CTE) for silicon (3.5) matches closely with those of tungsten (4.0) and molybdenum (5.45). Using these metals in packages, argues Margalit, gives designers a passive, rigid heat spreader that will coexist with ceramics, Si, GaAs, and other component materials, reducing the risk of warpage and breakage in high heat.
For practical application of refractory metals, Margalit outlined the equation that served as a basis for thermal research. The thermal conductivity of the heat spreader (λ) divided by the difference in CTE between the die and substrate (∆ CTE) renders the thermal compatibility of materials, as well as the efficiency of the heat spreader. Along with Starck colleagues Dincer Bozkaya, Baerbel Kloss, and Jonathan Tuck, Margalit manipulated the capabilities of refractory metals to integrate with this equation, creating a targeted, passive, adaptable heat spreader.
Starck introduced copper (Cu) into the equation to create a layered Cu/Mo/Cu structure similar to an insulated metal substrate (IMS). Copper, with much higher conductivity than refractory metals (close to 400, compared to the 100-200 range of Mo and W), channels heat away from the package; molybdenum, with a high melting point, prevents overexpansion and warpage. In this form the limitations of refractory metals become apparent. Molybdenum is an inherently dense element - the layered stack is hard to flatten into a thin, even heat spreader viable for use in advanced packages. The company is developing a product, Viatherm, to address this issue, and to give manufacturers more precise control over thermal management. The product builds on the Cu/Mo/Cu concept, introducing chemically etched holes - vias - through the molybdenum, which in turn fill with copper. The result is a thinner, less dense, striated product with copper “chambers” supported by molybdenum. This system, which could apply with other materials to make Cu/W/Cu or another layered heat spreader, maximizes the high conductivity (λ) of copper and the desirable CTE of molybdenum. It is controllable, allowing designers and manufacturers to etch more vias over a hot spot within the package. The product is also produced with known and established materials and processes, as Margalit pointed out, an aspect that brings the theory of striated, layered metals for thermal management into an appealing light in terms of ramp, testing, and commercial availability.
This kind of advanced thermal management that manipulates material properties and physical structure to maximize cooling in a passive heat spreader targets high-power, dense electronics in high ambient temperatures. Consider high-brightness LEDs (HB LEDs), a market sector expected by analyst firm Strategies Unlimited to steadily grow by 15-20%, reaching $8.3B in 2010. Margalit looked favorably upon applications for refractory-metals-based thermal management in large-scale display and automotive HB LEDs, where ambient temperatures surrounding packaged LEDs reach extremes. Refractory metals offer useful strength at temperatures as high as 1000°C, which seems high unless one considers usage in a massive cluster of HB LEDs forming a display in the Las Vegas desert, or in HB-LED headlamps operating for long periods close to a V8 engine in a car. Refractory metals aren’t the only approach to thermal management, but the range of applications and adaptability of the elements suggest that these heat spreaders will be part of the thermal management equation as Moore’s law squeezes chips to even smaller nodes.