• There has been much focus on optimizing TIMs for CPUs, conversely little optimization is typically performed on TIMs in VRM/VRD applications (“A” TIM is used, not necessarily the “Best” TIM).
• In 1994 the TIM resistance accounted for only 5% of the total CPU thermal resistance for the 66MHz Pentium (Advanced Packaging).
• In 2000, the TIM accounted for 50% of the CPU thermal resistance for the Athlon processor.
• Similarly, as power levels and the associated power densities continue to increase in VRM/VRD’s, selecting the optimum TIM is of equal importance to selecting the optimum heat sink.
• There are two key factors that impact the performance of TIM’s:
  • 1.Bulk thermal conductivity of the material
  • 2.Surface wetting/conforming characteristics of the material

• The goal of a TIM is to fill in all of the voids between the faying surfaces.

As can be seen in the graph below, the Gap pad A3000 has the best thermal characteristics and the highest hardness (lowest conformability)

Material Class	Bulk K	Surface Wetting	Surface Conforming	Vendors
Grease	0.3-2W/mK Al Filled 6+	Very Good	Very Good	Hi-K Grease Shinetsu Bergquist
Gel	0.3-2W/mK	Very Good	Very Good	Thermoset-Lord MG Series
Adhesive	0.3-1.3W/mK	Very Good	Very Good	Dow Corning 3M
Tape	0.7-1.5W/mK	Moderate	Poor	Bergquist 3M Dow Corning
Phase Change	0.8-1.5W/mK	Very Good	Very Good for irregularities <0.002" Very Poor for large irregularities or bowing	Bergquist 3M Dow Corning
Pads	0.8-4W/mK	Poor	Very Good for large irregularities Very Poor for small irregularities	Bergquist 3M Dow Corning

Selecting the optimum material for a VRM/VRD application involves several factors:

• Electrical isolation (DirectFET MOSFETs, and flange mounted D-PAKs, etc.)
• Attachment to a PCB (back-side cooling VRM’s) - very difficult to achieve a high required pressure without deforming the PCB.
• Cost vs. performance, $0.04 to $0.53 for a 4-Phase DirectFET VRM 1U, but how to predict performance?

• Proper selection of the optimal “class” of TIM enabled a significant improvement in performance The higher performing TIM was an adhesive with a lower thermal conductivity than the competing interface pad
• Performance improvement gain was based on area of contact and thickness

Testing of interface materials in the application is critical

• Initial first order material selection can be based on the data sheet, however to optimize performance, empirical testing must be performed

• Vin = 12V, Vout=1.35V
• 4 phase, 500KHz
• 6604+6618 per phase
• VRM10 heat sink
• Imax based on Tcase=100 °C

• Customer implemented a thin, high thermal conductivity pad because it had the best “metrics”
• In the application it only makes contact with <20% of the area – resulting in non-repeatable poor performance
• When the screws are tightened, the board flexes and comes out of contact with the PCB
• Should use thicker and softer pad or a conformable material like Gel