High performance heat exchangers are used to lower the operating junction temperature of power devices so that their steady state output, as well as their short duration on demand output, is significantly improved. Use of better thermal interface materials were evaluated using a test vehicle similar to an actual design. While the thermal conductivity of thermal interface materials (TIMs) are always reported by the manufacturer, the value alone is not sufficient to determine which TIM would be better for any particular application. Other parameters may also affect the thermal resistance, thereby influencing the effectiveness of the heat transfer as much, or more than just the thermal conductivity of the TIM. The interface surface roughness, wettability, area, and pressure all affect how well a TIM performs in a particular application. While the manufacturer tests and reports a value of thermal conductivity using optimal conditions for their material, actual use of the material may provide substantially different results. By manufacturing a test vehicle modeled from an actual application, most of the interface parameters can be held constant and the cooling effectiveness of different TIMs can be directly compared and contrasted.

This test vehicle design uses a thermal interface material between the heat source and the coldplate (Figure 4-1).

Fourier’s Law states that the flow of heat is proportional to the temperature gradient and the cross sectional area normal to the heat flow direction. For a one dimensional heat flow at steady state, this can be expressed as:


The thermal conductivity (k) is an intrinsic property of how the bulk material internally conducts heat. It is not dependent on the size or shape of the material and more importantly, does not include any effects from the thermal interface.

Thermal resistance (R) is not an intrinsic material property and should be determined for each configuration according to this equation:


Thermal resistance more accurately predicts thermal performance than thermal conductivity since the affects at the thermal interface are
not ignored.

When two surfaces are mated under pressure, the contact is not perfect, even for highly polished flat surfaces. Surface irregularities prevent
intimate contact of large areas between the mating surfaces. Solid contacts only occur between the high points of the two mating surfaces leaving a large number of voids between the low lying areas. Most of the heat transfer takes place via these solid contact points, but is restricted since the contact areas are very small. Heat transfer also occurs through the air entrapped in the irregular voids, but is extremely low since the thermal conductivity of air is very low compared to the metals that are in direct contact. In order to eliminate the air gaps and improve thermal transfer, a thermally conductive material is used. This material conforms to the surface peaks and valleys and displaces the air, providing more area for heat to flow and reducing the thermal resistance of the interface.

The thermal resistance at the interface is usually much greater than the overall bulk resistance of the two mating bodies and therefore provides the biggest barrier to increasing the heat transfer rates. This thermal resistance between two heat conducting surfaces depends on several factors such as:

• Geometry/flatness
  
• Surface finish of mating surfaces
  
• Hardness
  
• Modulus of elasticity
  
• Contact pressure
  
• Thermal conductivity
• Length of heat conducting path

There are four primary parameters that can be changed to enhance the conducted heat flow rate of any system (k, A, DT, and L). Once a suitable material with high thermal conductivity is selected, the other three parameters can be improved to enhance the heat removal rate.

To increase the effective area of heat transfer (A), the voids created by the imperfect surfaces must be filled with suitable highly conductive thermal interface materials. Many different approaches have been adopted by the industry to fill in these voids. Thermal greases, soft metal films, soft metal plating, better machining, and surface finishing techniques are some of the commonly adopted approaches.

An experimental setup was created based on a coldplate’s thermal management performance in cooling a simulated semiconductor device.Instead of using the actual semiconductor device which produces heat during operation, three planar 600 W, 3 W resistors were used to simulate the actual power levels predicted (Figure 4-2).

Several thermal interface materials were tested to determine their effectiveness.

• Thermal Greases (2)
  
• Soft Metal Foil
  
• Phase Change Metal Alloy (PCMA)
  
• Thermal Pad
  
• No TIM

Figure 4-3 shows the temperatures obtained when the best of the TIM materials were tested at the same flow rate, input water temp, and power input. Each TIM has a bar triplet that indicates the temperature at the center of zones 1, 2, and 3 (purple, red, and green, respectively). Under these conditions, the two thermal greases provided similar results, but the testing was sensitive enough to always discriminate one from the other. The soft metal foil TIM was almost as good. This material was developed as a compressible metallic shim for thermal applications under power devices. Rather than being flat, it has an embossed pattern that provides contact with both sides of an interface, even though surface irregularities exist. At the time of these experiments, samples were only available in a two inch width, much narrower than the heat spreader. While placed at the hottest portion of the zone, improved results would be expected if the foil covered the full width of the heat spreader. Also shown in this graph, is the higher temperatures obtained when no TIM was used.

The phase change metal alloy provided a temperature performance similar to the soft metal foil, however, regions of melting and flow occurred that allowed some of the material to move out of the interface. To properly test this material, the experimental design would need modifications to keep the TIM in place.

The thermal pad material produced the highest temperatures, but was the only TIM that was not maintained at a 0.004 inch bondline. Since it is constructed with three 0.002 inch layers (thermal grease/aluminum/ thermal grease), the resulting bondline was greater than the other TIMs. The higher heat transfer path length resulted in a lower heat flow.

Fine differences in performances of thermal interface materials can be experimentally determined if the test vehicle is matched to the actual application. This study indicated that while a standard thermal grease may perform well in a particular application, other TIMs should also be considered. The soft metal foil provided similar thermal control without the careful application processes needed to apply thermal grease. To achieve a uniform coating and repeatable bondline control with thermal grease, an investment in fixturing tools and maintenance is required. The foil offers a more manufacturable, easier to apply, easier to rework, repeatable method for achieving cooling in high power devices.

The EMPF can perform thermal interface materials testing. For more information contact Ken Friedman, at 610.362.1200, extension 279 or via email at kfriedman@aciusa.org.