IC Thermal Management Part 3: Underfill glass transition temperature, TG
1) The very first task is to obtain accurate information on which to act. We’d like to have accurate temperature readouts at relevant locations, which was the subject of the first entry.
2) The dissipated power must be transported away to the outside environment, often using some form of air mover. Constructing optimal cooling solutions for best user experience was the subject for the second entry. One has to balance the performance needs with the power requirements while delivering acceptable acoustics, as an example. Choice of air movers and heat sinks is largely driven by the form-factor of the application. This aspect of IC thermal management is a core mechanical engineering function.
3) Designing optimal cooling solution and choosing best control algorithms would necessitate good understanding of the critical thermal parameters governing the temperature regime of the CPU. Thermal interface materials which couple the dissipated power to the heat sinks are often the first major roadblock for the heat on the way to environment: its choice is of critical importance. For high flux density application it’s often responsible for the largest share of temperature budget. This aspect of IC thermal management is also a core mechanical engineering function.
Semiconductor Design and the effects
We now want to touch upon what actually determines the temperature limits for the IC design. While ultimately it’s a complicated subject, with the considerations that need to be given to circuit speeds and power draw as well as optimal user experience, the long term IC reliability is the starting point when initiating a mechanical design cycle. (For this blog, we neglect the underlying complexities of semiconductor device physics and materials science engineering.) Delivering reliably operating system requires coordination of multiple manufacturers in the production chain, from fab to packaging to system. While earlier manufacturing could be based on case-by-case, supplier-to-supplier interaction, the commonality of material sets and packaging form-factors for most applications led to development of industrial reliability specifications.
Among these a major challenge for industry is meeting reliability standards in temperature cycled environments. For IC packaging, the underlying conflict driving these issues is the difference in thermal expansion between the crystalline silicon, the material hosting the enabling circuitry, and the “spatial transformer”, the material used for delivering signals from micro- to macro-scale: the package substrate. These issues came to the forefront to become critical factors in the selection of packaging materials when IBM introduced flip-chip technology as a way of achieving high interconnect density IC packaging. Initially met with caution (as are most breakthroughs), the tipping point was the introduction of underfill, a material initially intended for electrical insulation between the solder chip-to-package electrical connectors. While certainly achieving this objective, the underfill with the right mechanical properties protected solder connectors from excessive repeated deformation and eventual failure: anyone who broke a metal wire by bending it multiple times is familiar with the phenomenon. By addressing this weakness of an early flip-chip technology, mass-production for consumer became possible.
Writing about this particular issue in an IC thermal management blog isn’t a coincidence. The underfill is often a first material that loses its mechanical stability as temperature increases above a certain point, a glass transition temperature. Choice of the transition temperature has bearing on stresses in increasingly weak interlevel dielectric materials (needed for high-speed operation), so finding an optimum underfill is a complicated balancing act.
One famous physicist was fond of saying when faced with a complex problem: “trivialize the issue”. Following his advice, we can narrow down the highest hotspot operating temperature to this parameter, underfill glass transition temperature, or Tg. For the purposes of IC thermal management, we can think of it as a parameter which collapses all the complexities of thermomechanical interactions in the chip to one identifiable rule: keep temperatures below Tg.
The engineering consultants at Glew Engineering can assist you with navigating through this and more related issues. Our mechanical engineers, electrical engineers, and materials science enghineers, have extensive experience it the above described problems and their solutions.