Choosing a thermal pad is not about picking the biggest W/m·K number on the datasheet. It is about matching four variables at once: how much heat you move, how much gap you must fill, how much pressure your hardware can apply, and how long the interface has to survive. This guide walks through all four, with the numbers you actually need.
Start with the four variables that define a TIM
Every thermal interface material decision comes down to four parameters. Get these right and the datasheet comparison becomes trivial; get them wrong and a 12 W/m·K pad will outperform a 180 W/m·K one.
| Variable | What to determine | Typical range | Why it matters |
|---|---|---|---|
| Power density (W/cm²) | Heat flux through the die or module footprint | 5–500 W/cm² | Sets how low your thermal resistance budget must be |
| Bond-line thickness (BLT) | Gap you must fill after assembly tolerance | 0.1–2.0 mm | Resistance scales linearly with thickness — the most under-managed variable |
| Mounting pressure | Force your hardware actually applies | 5–100 psi | Determines achievable BLT and long-term stability |
| Service life & temperature | Operating temp plus required field life | −40 to 200 °C; 3–20 yrs | Eliminates grease and phase-change options in most industrial cases |
Step 1 — Calculate how much thermal resistance you can afford
Before looking at any material, work out your budget. Take the maximum allowable junction temperature (Tj,max), subtract your worst-case coolant or ambient temperature, divide by power, then subtract the resistances you cannot control (die-to-case, spreader, heat sink to air).
Example: a 300 W GPU with Tj,max = 95 °C, coolant at 35 °C, die-to-lid resistance 0.05 °C/W, cold plate 0.08 °C/W.
Step 2 — Measure the real gap, not the nominal one
Engineering drawings give you nominal stack height. Real assemblies give you warpage, component height variation, and machining tolerance. Add them up:
- Flatness / warpage of the heat source (BGA substrates and power modules commonly bow 50–150 μm)
- Height variation between adjacent components on the same cold plate
- Machining tolerance on the heat sink or cold plate surface
- Assembly tolerance — screw torque variation across the mounting pattern
Add 20–30% margin on top. If that total is 0.5 mm, you need a pad that reliably compresses to 0.5 mm at your available pressure — not one that only reaches its advertised performance at 0.2 mm.
Step 3 — Check available mounting pressure honestly
This is where most industrial designs fail. High-performance pads often quote resistance figures at 40–100 psi, but a spring-clip heatsink on a plastic housing may only deliver 5–10 psi.
| Mounting method | Typical applied pressure | What this means for pad choice |
|---|---|---|
| Spring push-pins / clips | 5–15 psi | Need a soft, highly conformable pad; thickness must be generous |
| Screw-down cold plate (4–6 points) | 20–50 psi | Standard range — most pads reach spec performance |
| Bolted power module baseplate | 50–150 psi | Can use thinner, stiffer pads; watch for pad extrusion |
| Adhesive / tape retention only | <5 psi | Contact resistance dominates; consider phase change or grease |
Lenecold's graphene pad reaches its <0.05 °C·cm²/W figure at low mounting pressure, which is the practical reason it works in clip-mounted designs where a filled-polymer pad of similar nominal conductivity would not conform.
Step 4 — Decide what happens after 3 years
Datasheets describe day-one performance. Your product has to survive thermal cycling, humidity, and sustained high temperature. Ask:
- Does the interface dry out? Silicone grease loses carrier oil over time and its resistance climbs.
- Does it pump out? Repeated expansion and contraction physically walks grease out of the gap — see why pump-out happens.
- Does it contaminate anything? Siloxane outgassing from silicone materials can foul optical surfaces and cause contact failures in relays and switches.
- Does it crack at low temperature? Some filled pads stiffen and lose contact below −20 °C.
Quick comparison: common interface options
| Option | Conductivity (typical) | BLT | Long-term stability | Rework | Best for |
|---|---|---|---|---|---|
| Thermal grease | 3–8 W/m·K | 25–75 μm | Poor — dries, pumps out | Messy | Prototypes, serviceable consumer devices |
| Filled silicone pad | 1–6 W/m·K | 0.3–2.0 mm | Good | Clean | Low power, gap-filling, low pressure |
| Phase change material | 1–5 W/m·K | 25–75 μm | Fair | Needs reflow | High volume, controlled BLT |
| Solder TIM | 50–80 W/m·K | 25–50 μm | Excellent | Very hard | Flip-chip, extreme flux |
| Graphene pad | 180 W/m·K | 0.2–2.0 mm | 150 °C × 1500 h validated | Clean, die-cut | High power + long life + no service access |
A short selection checklist
Compute the resistance budget
From Tj,max, worst-case sink temperature and power. Write the number down.
Measure the worst-case gap
Including warpage and tolerance stack-up, plus 25% margin.
Confirm real mounting pressure
Measure or simulate it. Do not assume it from the drawing.
Filter by conductivity at that BLT and pressure
Not by headline W/m·K. Request R-versus-pressure curves.
Verify aging data
Ask for resistance drift after high-temperature soak, not just initial values.
Prototype and measure
Instrument the real assembly. Simulations of interface layers are notoriously optimistic.