Thermal grease usually wins on day one and often loses by year three. Thermal pads usually lose on day one and win over the product's life. Understanding exactly why helps you pick the right one — and in high-power industrial designs the answer is less obvious than the datasheet suggests.
Why grease looks better on paper
Grease wins the initial comparison for one reason: bond-line thickness. A properly applied grease layer sits at 25–75 μm, while even a thin pad is 200–500 μm. Since thermal resistance scales directly with thickness, a 3 W/m·K grease at 50 μm can beat a 6 W/m·K pad at 500 μm.
That is a 5× difference in the grease's favour — which is why it remains the default for CPUs and GPUs in serviceable consumer hardware.
Where the comparison breaks down
The calculation above assumes the grease stays where you put it, at the thickness you applied it, for the life of the product. In a device that is opened and repasted every two years, that assumption roughly holds. In a sealed industrial or automotive assembly, it does not.
Three failure mechanisms that only affect grease
- Pump-out. Every thermal cycle expands and contracts the interface. With a CTE mismatch between die and heat sink, the gap width changes and viscous grease is progressively squeezed outward. Over thousands of cycles the centre of the interface starves and resistance climbs. This is the dominant failure mode in power cycling applications, and it is covered in detail in our pump-out article.
- Dry-out and oil separation. The carrier fluid in silicone grease slowly separates from the filler, especially above 100 °C. What remains is a dry, cracked, high-resistance film.
- Migration and contamination. Grease that escapes the gap travels. On optical assemblies this is fatal; on connector pins and relay contacts, silicone outgassing produces insulating deposits.
Head-to-head comparison
| Factor | Thermal grease | Graphene thermal pad |
|---|---|---|
| Initial thermal resistance | Lower (thin BLT) | Higher unless conductivity is very high |
| Typical conductivity | 3–8 W/m·K | Up to 180 W/m·K through-plane |
| Bond line thickness | 25–75 μm | 0.2–2.0 mm |
| Pump-out resistance | Poor | Not applicable — solid material |
| Dry-out / oil bleed | Yes, especially >100 °C | No |
| Contamination risk | Silicone outgassing, migration | None |
| Assembly throughput | Slow — dispense, screen or stencil | Fast — die-cut part, peel and place |
| Thickness control | Depends on pressure and dispense | Controlled by pad construction |
| Rework | Messy, needs cleaning | Lift and replace |
| Field service | Requires skilled technician | Requires skilled technician |
| Cost at volume | Low material, high labour | Higher material, low labour |
How a high-conductivity pad changes the trade-off
The reason grease keeps winning is the thickness term, not the conductivity term. If conductivity rises far enough, the thickness penalty disappears.
| Interface | k (W/m·K) | BLT | R (°C·cm²/W)* |
|---|---|---|---|
| Standard grease | 3.5 | 0.05 mm | 0.014 |
| Performance grease | 8 | 0.05 mm | 0.006 |
| Filled silicone pad | 6 | 0.5 mm | 0.083 |
| Graphene pad | 180 | 0.5 mm | 0.003 |
*Bulk resistance only, R = BLT / k. Real assemblies add contact resistance at both faces, which is pressure-dependent and often dominates at low mounting force.
At 180 W/m·K the pad's bulk resistance falls below even a good grease layer, while remaining a solid that cannot pump out, dry out or migrate. That is the specific case where the pad stops being a compromise and becomes the better engineering answer.
When to choose which
Choose grease when
- The gap is genuinely tiny and well controlled (<0.1 mm)
- The device is designed to be serviced and repasted
- Mounting pressure is very low and you need extreme conformability
- Unit volume is low and labour cost is not a constraint
Choose a solid pad when
- The assembly is sealed and will never be opened
- Service life exceeds 3 years, or the design is warrantied
- Continuous operating temperature is above 100 °C
- The product experiences frequent power or thermal cycling
- Optical surfaces, connectors or relays are nearby
- Production volume makes dispense and cleanup expensive
- Gap variation across the assembly is large