Practical engineering notes on thermal interface materials — selection, calculation, failure modes and application specifics. Written for engineers who have to make the interface work, not to sell a material.
All resources
TIM1 for Chip Packaging
The hardest layer in the thermal stack: 180 W/m·K through-thickness, below 0.05 °C·cm²/W, 1500 h aging verified, solid with no pump-out.
Read →How to Choose a Thermal Pad
Four variables decide every interface: power density, gap, mounting pressure and service life. Worked examples throughout.
Read →Thermal Pad vs Thermal Grease
Why grease wins on day one and loses by year three — with calculated resistance at real bond-line thicknesses.
Read →How to Calculate Thermal Resistance
R = L/(k×A), contact resistance terms, a full junction-to-ambient budget, and the five mistakes engineers make.
Read →What Is TIM Pump-Out?
How CTE mismatch and thermal cycling squeeze grease out of the joint — and why sealed assemblies must avoid it.
Read →GPU Server Thermal Management
Why TIM2 is the largest controllable resistance in a liquid-cooled AI rack, with a worked 800 W budget.
Read →IGBT & SiC Power Module Interfaces
Power cycling, 150 °C continuous operation, baseplate bow, and how interface resistance maps to cycle life.
Read →EV & ESS Battery Cold Plate Interfaces
Gap variation, cell swelling, fast-charge flux and cell-to-cell temperature uniformity.
Read →What Does W/m·K Mean?
Through-plane vs in-plane, reference values, and why a big conductivity number can still perform badly.
Read →Graphene Thermal Interface Materials
What graphene really contributes to a TIM, and six questions to ask before believing a conductivity claim.
Read →AI Cooling Trends 2026
Chip power, the five-stage TIM roadmap and where the bottleneck moves when packages pass 1000 W.
Read →AI / GPU Thermal Material Guide
H100, H200, B200, MI300X and Ascend 910B power classes, with three proven cooling architectures.
Read →Graphene vs Copper vs Aluminum
Five dimensions of comparison, and a clear rule for when each material is the right answer.
Read →IGBT / SiC Thermal Design Guide
Thermal resistance calculation, a worked example and reference stacks for common SiC modules.
Read →Not sure which one applies to you?
Use this shortcut:
| If you are… | Start here |
|---|---|
| Picking a pad for the first time | Selection Guide |
| Choosing between grease and a pad | Pad vs Grease |
| Working out whether your design will hold temperature | Thermal Resistance Calculation |
| Debugging a field failure or thermal drift | TIM Pump-Out |
| Cooling AI servers or GPU clusters | GPU Server Thermal Management |
| Designing traction inverters or converters | IGBT & SiC Modules |
| Building EV or ESS battery packs | Battery Cold Plates |
| New to the units and terminology | W/m·K Explained |
| Evaluating graphene specifically | Graphene TIM Guide |