|
One or more heat dissipation or cooling components are typically attached to a
chip in order to cool it. In the typical air-cooled thermal solution a chip or
its package would use a heat sink to conduct heat away from the chip and a fan
to then transfer that heat to the ambient. Other cooling solutions approaches
could employ a cold plate, heat pipe, or other technology to remove the chip’s heat.
Regardless of the cooling device used, it is not advisable to simply place it on
the chip. The reason is that there is no practical and feasible way of making
both the chip surface and the cooling device surface perfectly flat enough or
free enough of small scale surface asperities. These microscopic irregularities
prevent the surfaces from making complete and continuous contact; in fact the majority
of the surfaces will end up separated from each other:

By default, these spaces will fill with air which is thermally insulative. The
result is that, no matter how good the thermal performance of the cooling device
the thermal resistance in this gap will render the overall solution intolerably poor.
The solution is to displace the air in the gap with something that is a better thermal
conductor—a Thermal Interface Material or TIM. While a TIM may serve other
purposes such as attaching or adhering the surfaces together its primary purpose
is to provide complete wetting between the chip to be cooled and the rest of the
thermal solution (such as a spreader or heat sink.) Wetting is the ability of the
material to completely fill the interstices, displacing all air. The better the wetting
of the TIM, the less air trapped, and the lower the thermal resistance.
Increasingly chips are being packaged with heat spreader lids; current examples
include the Intel Pentium and AMD Athlon64. These lids, known as Integrated Heat
Spreaders (IHS), provide both thermal spreading ability and mechanical strength
to protect the die. (Some of this protection is required due to the increasingly
high clamping pressures used to attach heat sinks; the increased pressure tends
to minimize the gap and increase the conformity between the surfaces.) These
packages use a TIM, called TIM1, between the die and the IHS:

The typical air-cooled solution for this kind of package would use another TIM,
called TIM2 between the IHS and the heat sink base.
The importance of a TIM is not proportional to its volume in the overall thermal
solution. About 40% of the thermal resistance in the overall solution lies in the
TIM1 layer between the die and the lid:

Source: "Materials for Thermal Management", Dr. Nancy Dean,
Advanced Packaging, March 2003, p.16
Many current designs are thermally inferior, exacerbating the problems of designing a
feasible back-end (heat sink and fan) thermal solution. Steve Pawlowski, Director of Intel’s
Microprocessor Lab stated at a developers’ conference in Taipei recently: "There's a
certain [thermal] transfer coefficient between the die and the package. If you can't
come up with a better package that will allow that heat transfer to occur from the die,
no matter what you do on the outside, it’s going to be difficult." Addressing the thermal
transfer bottleneck at the first interface, at the die level, is a critical challenge for
today’s thermal engineers and scientists.
The selection of the various components within a thermal solution has a huge
impact on the constraints imposed (or relaxed) on the other components and on the
overall results. The use of an extremely high-performing TIM can allow the use of a
simpler and cheaper heat sink, or even the elimination of the fan. The slightly
increased cost of a better TIM may be more than recovered by cost reductions elsewhere
in the overall thermal solution.
It is cheaper and more efficient to cool a component the closer the cooling solution
is to the heat source. For example, you can produce a chilled glass of lemonade more
efficiently and at less cost by putting ice in it than by air-conditioning the
surrounding room. Similarly, cost and efficiency measures in cooling a chip are improved
by increasing the thermal performance of the material directly adjacent to that chip—the TIM.
For all these reasons the choice of a TIM has perhaps the largest impact of any component
selection on the performance, cost, and other attributes of the overall solution.
TIMs include greases, gels, pads, pastes, waxes, phase-change materials, and
particle-filled materials. Using a solder would provide the improved thermal
performance of an all-metal heat path, but the problem of mismatched CTE between die
and lid material makes this choice difficult. Intel has recently started using an Indium
solder as a TIM1 in some of its chip products. Others have experimented with Low Melt
Alloys—some commercial products use them today, but their adoption has been limited due
to concerns about the reliability of those formulations.
Find out how Indigo compares to other thermal interface materials.
|