TIM Pump-Out and Dry-Out: Causes and Prevention
Short answer: TIM pump-out is the mechanical migration of thermal interface material out of the bond line, driven by repeated thermal cycling and the resulting shear stress; TIM dry-out is the loss of the liquid carrier phase through evaporation, oxidation, or bleed, leaving a dry, high-resistance residue. Both raise interface resistance — typically from an initial 0.05–0.15 °C·cm²/W to 0.5 °C·cm²/W or worse over hundreds to thousands of cycles. Prevention combines low-bleed, high-viscosity or phase-change materials, controlled bond-line thickness (50–100 µm typical), adequate mounting pressure, matched CTE, and heat sink flatness held to ±0.005 mm on machined interfaces.
What Exactly Is TIM Pump-Out?
Pump-out is a mechanical failure mode, not a chemical one. When a heat sink and its heat source expand and contract at different rates, the gap between them opens and closes with every power cycle. The thermal interface material (TIM) sitting in that gap is squeezed and sucked like a bellows. Over time, the material migrates from the center of the die or device footprint toward the edges — and eventually off the interface entirely.
The physics is straightforward. A silicon die or copper slug has a coefficient of thermal expansion (CTE) near 2.6–17 ppm/°C depending on material, while an aluminum heat sink base sits around 23 ppm/°C. Clamp a thin layer of grease between them and cycle the junction 40 °C, and the bond line breathes by microns each cycle. Grease has essentially no elastic memory, so it does not return to where it started.
Pump-out shows up most aggressively in three situations:
- Large-area interfaces where the total expansion across the footprint is greatest.
- High power density with fast thermal transients — think GPU dies, IGBT modules, and LED arrays driven by PWM dimming.
- Thin bond lines under high clamp pressure, where the material is already near its flow threshold.
The telltale failure signature is a gradual, monotonic rise in junction temperature over weeks, with a visible dry ring or pumped ridge of material around the perimeter of the device when the heat sink is removed.
What Causes TIM Dry-Out, and How Is It Different?
Dry-out is chemical and physical degradation of the TIM itself. Most thermal greases are a silicone or hydrocarbon oil carrier loaded with ceramic, metal oxide, or metal particles. The carrier does the wetting; the filler does the conducting. When the carrier leaves, only a crumbly filler skeleton remains.
There are four common pathways:
1. Evaporation of volatile fractions. Low-molecular-weight oils boil off at sustained temperatures above roughly 100–125 °C. This is why "high-temperature" greases are formulated with phenyl silicones or heavily refined base oils.
2. Oxidative hardening. Oxygen diffuses into the bond line and cross-links the polymer, turning a paste into a rubbery or chalky solid.
3. Oil bleed and separation. The carrier migrates into porous surfaces — bare aluminum, anodized layers, or unfilled polymer substrates — starving the bulk material.
4. Filler settling and agglomeration. Over years at temperature, dense fillers settle or sinter, reducing effective contact area.
The practical distinction matters for diagnosis. Pump-out leaves material somewhere else — you find it around the edges. Dry-out leaves material in place but dead — the residue is still centered on the device, just non-functional. Many field failures are both at once: the material pumps to the perimeter and the thin film left behind dry-outs first because it has the least carrier volume per unit area.
Why Do Some Designs Fail in Months and Others Last a Decade?
The difference is almost always mechanical and thermal design discipline rather than the TIM brand. The table below summarizes the dominant drivers.
| Driver | Accelerates pump-out / dry-out | Mitigates |
|---|---|---|
| CTE mismatch | Aluminum base on silicon die, large footprint | Copper or CuMo base, matched CTE spreader |
| Thermal swing ΔT | >40 °C per cycle, fast ramp rates | Slower ramp, thermal mass, duty-cycle shaping |
| Cycle count | Continuous on/off or PWM dimming | Steady-state operation, hysteresis control |
| Bond-line thickness | <30 µm or >150 µm | 50–100 µm controlled gap |
| Mounting pressure | Uneven or excessive clamp force | Balanced spring load, torque spec |
| Surface flatness | Concave/convex base, >0.05 mm bow | Machined flatness ±0.005 mm |
| Substrate porosity | Bare aluminum, rough casting | Anodized or plated interface, filled polymer |
| TIM chemistry | High-bleed grease, low-viscosity paste | Low-bleed, phase-change, or cured gap filler |
Note that the two most controllable variables — flatness and bond-line thickness — are manufacturing outputs, not material properties. A heat sink base that bows 0.1 mm across a 50 mm footprint will create a wedge-shaped bond line. The thin end dry-outs first, the thick end pumps first, and the failure accelerates from both directions.
This is why we hold machined interface surfaces to ±0.005 mm and verify flatness with a coordinate measuring machine rather than a straight edge. You can read more about how that tolerance translates into real interface performance in our guide to heat sink flatness specifications.
How Do You Detect Pump-Out Before It Becomes a Field Failure?
Because the failure is gradual, condition monitoring works well. The most reliable indicators:
- Thermal resistance trending. Log junction-to-ambient or junction-to-case resistance at a fixed power and ambient. A 20–30% rise over the first 1,000 hours is an early warning.
- Fan speed or duty creep. In closed-loop systems, the controller compensates by spinning the fan faster. Rising RPM at constant load is a proxy for TIM degradation.
- Case-to-sink delta. Measure the temperature drop across the interface directly with a pair of thermocouples. A healthy greased interface shows a small, stable delta; degradation widens it.
- Post-mortem inspection. On teardown, photograph the footprint. Pumped material around the perimeter, a dry center, or a shiny oil film on the heat sink base all confirm the mechanism.
For qualification, accelerated thermal cycling between the minimum and maximum operating temperatures — typically 500 to 1,000 cycles — with in-situ resistance measurement is the industry-standard screen. The key is to measure during cycling, not only at the end, so you capture the slope rather than a single endpoint.
What Material Choices Actually Prevent These Failures?
There is no universal answer, but the selection logic is consistent. The table below maps application profiles to suitable TIM classes.
| Application profile | Recommended TIM class | Typical bond line | Notes |
|---|---|---|---|
| High-power GPU/CPU, large die | Phase-change material or metal TIM | 25–75 µm | Phase change resists pump-out at operating temp |
| IGBT module, industrial | Cured silicone gap filler or PCM | 100–200 µm | Cured materials cannot pump |
| LED array, constant on | Low-bleed high-viscosity grease | 50–100 µm | Low volatility critical at 100 °C+ |
| Consumer electronics, low power | Standard silicone grease | 50–100 µm | Acceptable if ΔT is small |
| Vibration + thermal cycling | Gap pad with controlled compression | 0.5–2 mm | Compliant, no migration path |
| Extreme reliability, no service | Sintered or soldered interface | <50 µm | Eliminates organic TIM entirely |
Two rules of thumb are worth remembering. First, the more the interface moves, the more you want a material that either cures in place or changes phase — both resist migration far better than a grease that stays viscous forever. Second, the hotter and longer the duty cycle, the lower the volatile content you should accept. A grease with 2% volatiles at 125 °C for 10,000 hours will lose meaningful mass; one with 0.3% will not.
If you are still narrowing down chemistry, our breakdown of thermal grease selection criteria covers viscosity, filler loading, and bleed testing in more detail.
Design and Manufacturing Practices That Extend Interface Life
Prevention is mostly about removing the mechanical freedom that lets TIM migrate. Practical steps:
Control the bond line deliberately
Specify a target gap and hold it with standoffs, a machined pocket, or a controlled spring load. Do not let the bond line be whatever the tolerance stack happens to produce. A 50 µm target with ±15 µm control is far better than an uncontrolled 20–200 µm range.
Balance the mounting load
Four-point or perimeter clamping with springs or shoulder screws distributes pressure evenly. Over-torquing one corner bows the base and creates a wedge. Specify torque and, where it matters, verify with pressure-sensitive film.
Match the base material to the die
For large, high-power footprints, a copper or copper-composite spreader reduces the CTE gap and spreads heat so the interface sees a gentler gradient. Copper bases are heavier and cost more, but they often pay for themselves in reduced field returns.
Machine the interface flat
This is the single highest-leverage manufacturing control. A base machined to ±0.005 mm flatness and a fine surface finish gives the TIM a uniform, predictable bond line. Cast or as-extruded surfaces rarely achieve this without a secondary operation. Our CNC-machined heat sinks are produced with the interface as a controlled feature, not an afterthought.
Protect against substrate bleed
Bare aluminum and porous castings wick oil out of grease. Anodizing, chromate conversion, or a thin plated layer closes the surface and reduces bleed. Where the interface must stay bare for conductivity reasons, choose a low-bleed formulation.
Qualify with cycling, not just initial resistance
A TIM that looks excellent on day one can be the worst performer after 500 cycles. Always include thermal cycling in qualification, and measure resistance in situ.
When Should You Switch Away From Grease Entirely?
Grease remains the best choice for many cost-sensitive, moderate-power applications. But consider moving away when any of the following apply:
- The device cannot be serviced and must last 10+ years at elevated temperature.
- Power density exceeds roughly 50 W/cm² at the die.
- Thermal cycling exceeds 1,000 cycles with ΔT above 60 °C.
- The assembly sees vibration or mechanical shock that could redistribute a viscous material.
- Field returns are already showing elevated junction temperatures.
In those cases, phase-change materials, cured gap fillers, or metal-based interfaces (solder, sinter, or a well-bonded metal TIM) eliminate the migration mechanism rather than slowing it. The trade-off is process complexity and, for metal TIMs, rework difficulty — which is why we also publish guidance on heat sink rework and repair for teams that need serviceability.
Frequently Asked Questions
Q: How long does it take for TIM pump-out to cause a noticeable temperature rise?
A: It depends on thermal swing and cycle rate. In aggressive cases — large die, ΔT above 60 °C, thousands of cycles per month — measurable degradation can appear within 500 to 1,000 hours. In gentler consumer applications, it may take years or never become significant. Trending interface resistance is the only reliable way to know your specific rate.
Q: Is pump-out the same as dry-out?
A: No. Pump-out is mechanical migration of material out of the bond line, driven by thermal expansion and contraction. Dry-out is chemical degradation — evaporation, oxidation, or bleed — that leaves the material in place but non-functional. They often occur together, and both raise interface thermal resistance, but the corrective actions differ.
Q: Can a thicker layer of thermal grease prevent pump-out?
A: Generally no, and it can make things worse. A thicker bond line has higher bulk thermal resistance from the start and gives the material more volume to migrate. The better approach is a controlled thin bond line (50–100 µm typical) with a material formulated to resist flow, or a phase-change or cured material that cannot pump.
Q: Does heat sink flatness really affect TIM reliability?
A: Yes, significantly. A bowed or uneven base creates a wedge-shaped bond line: the thin region dry-outs first while the thick region pumps first. Holding the interface to ±0.005 mm flatness keeps the bond line uniform, which slows both mechanisms and makes thermal performance predictable across units.
Q: What is the best way to qualify a TIM for a long-life product?
A: Run accelerated thermal cycling between minimum and maximum operating temperatures for at least 500 cycles, measuring interface thermal resistance in situ rather than only at the endpoints. Combine that with a high-temperature soak (1,000 hours at maximum operating temperature) to screen for dry-out. Together these two tests cover both failure mechanisms.
Related Resources
- About BQUQ and our Dongguan manufacturing footprint: /about/
- Heat sink product families, including extruded and machined options: /heat-sinks/
- Extruded aluminum heat sink profiles: /extruded-heat-sinks/
- Industry trends in thermal management: /industry-dynamics/
- Technical articles and engineering guides: /bquq-blog/
- Frequently asked questions: /faq/
- Case studies and application examples: /case/
- Contact the engineering team: /contact/
Authored by the BQUQ Engineering Team. BQUQ (Dongguan) runs CNC machining (±0.005 mm), metal stamping, custom springs, and heat sink production in one ISO9001 factory. Source-direct from Dongguan, China — quote in 12 hours: sc@bquq.com | WhatsApp +86 13713157787 | www.bquq.com
