Figure 1. Every heat-up and cool-down flexes the interface slightly - over many cycles, that motion can pump thermal grease out and leave voids behind.
Pump-out is the gradual loss of thermal grease from an interface caused by repeated mechanical movement. Each time a device heats up and cools down, the component and heat sink expand by different amounts and the package may bow, so the gap between them opens and closes slightly. Over hundreds or thousands of cycles, that motion squeezes grease out toward the edges and leaves voids behind, and the thermal resistance of the interface rises. Pump-out is most likely with large or flexible packages, wide temperature swings, low clamping pressure (parts pressed together only lightly) and thicker bond lines. It is tested with thermal cycling, power cycling or accelerated mechanical cycling, while tracking thermal resistance and voids.
A product passes its thermal qualification with margin, ships, and a year later field returns show components running hotter than they did on day one. The heat sink and the fan are fine. When the assembly is opened, the grease has thinned out in the middle of the interface and collected at the edges.
That pattern is pump-out. It is one of the most common reasons a thermal interface that measured well at the start loses performance over time - and because it develops slowly, it is easy to miss in short qualification tests.
How Pump-Out Happens
Figure 2. Hot: the package bows and grease is pushed outward. Cool: the gap reopens but grease does not fully return. After many cycles: voids in the center, grease at the edges.
The interface moves. The component, its package and the heat sink are made of materials with different coefficients of thermal expansion. When the device heats up, they expand by different amounts; packages and lids can also bow. The gap between the surfaces changes slightly - both sideways and in thickness.
The grease is sheared and squeezed. As the gap narrows and widens, grease is pushed outward and does not fully flow back.
The effect accumulates. Each cycle moves a little more grease to the edges. Voids open in the interface, often starting where the movement is largest.
Thermal resistance rises. Air in the voids conducts heat poorly, so the component temperature creeps up over time.
Pump-out is different from dry-out. Pump-out is mechanical - grease is physically moved out of the gap. Dry-out is a material change - the carrier separates from the filler at elevated temperature and the paste hardens. The two can happen together, but they have different causes and fixes.
Symptoms
Component temperature that rises gradually after many power or thermal cycles, under the same load and cooling.
On disassembly: a thin or bare area in the middle of the interface and excess grease collected at the edges.
Grease that has migrated beyond the interface onto surrounding surfaces, especially in vertical mountings.
Voids visible in acoustic microscopy images of the interface that grow with cycle count.
What Makes Pump-Out More Likely
| Factor | Why it matters |
|---|---|
| Wide temperature swings | Larger swings mean larger relative movement each cycle. |
| Large or flexible packages | Bigger parts and thin lids or baseplates bow more, so the gap changes more. |
| Frequent cycling | Power-cycled products - inverters, servers under varying load - accumulate cycles quickly. |
| Low clamping pressure / thick bond line | In an accelerated study at Purdue University, one grease's thermal resistance rose about four-fold at lower squeezing pressure, against about 1.2-fold at higher pressure, under the same mechanical cycling. |
| Vertical orientation | Gravity adds to the outward movement of a soft material. |
| Material rheology | The same Purdue study linked better resistance to voiding with a higher ratio of elastic modulus to viscosity - a material that holds its structure flows out less. |
What Studies Show
In a study by the U.S. National Renewable Energy Laboratory (NREL) with DuPont, five thermal greases were cycled between −40°C and 125°C in a fixture built from two metals with different thermal expansion, mounted vertically. Acoustic imaging showed changes in contact area and voiding in several greases after only 30 cycles, and grease movement increased further by 100 cycles. Some greases also separated into their constituents.
The Purdue study showed that the damage depends on both the material and the assembly: lower clamping pressure (a thicker bond line) produced more voids and pump-out, and voids near the heated edge of the interface affected thermal resistance far more than voids in a cooler area. Good resistance to voiding alone did not guarantee the best thermal stability - conductivity and stiffness mattered too.
The practical lesson: initial thermal resistance does not predict long-term behavior. Materials need to be compared after cycling, in conditions close to the real application.
How to Test for Pump-Out
Figure 3. Acoustic microscopy can track voids and grease movement through an intact assembly as cycles accumulate. Illustration - not measured data.
| Method | What it does | Trade-off |
|---|---|---|
| Power cycling | Switches the real device on and off | Most representative, but slow and costly |
| Thermal cycling | Moves the whole assembly between two temperatures | Standard and repeatable; a test fixture may not flex like the real package |
| Mechanical cycling | Opens and closes the gap mechanically while heat flows | Fast screening (hours, not months); correlation to field life still being studied |
What to measure: thermal resistance or component temperature before and after cycling; void area and grease position using scanning acoustic microscopy (C-SAM) without opening the assembly; and the grease distribution on disassembly at the end of the test.
Test in the real orientation - vertical if the product is mounted vertically - and at the real clamping pressure and bond line.
How to Reduce Pump-Out
Control the bond line and clamping. A thin, evenly applied layer under adequate, uniform pressure moves less. Follow the component manufacturer's mounting guidance.
Reduce relative movement where the design allows: stiffer lids or baseplates, more uniform clamping, smaller temperature swings.
Ask for cycling data. Compare candidate materials by their thermal resistance after cycling, not only their initial value or W/m·K.
Consider the material form. Thicker, more structured materials and curing gels respond to movement differently from a thin grease. Verify any alternative with the same cycling test.
Re-check after cycling. Include a post-cycling thermal measurement in qualification, not just an initial one.
Next Step: Compare Material Forms
Whether a grease, putty, gel or pad is the best fit depends on the gap, the clamping force and whether parts are reworked - as well as on how the material behaves under cycling. See Thermal Putty vs Thermal Grease vs Thermal Gel vs Thermal Pad: How to Choose.
Temperatures Creeping Up in the Field?
Tell us your cycling profile, package size, mounting orientation and what you saw on disassembly. Our engineers can help you identify whether pump-out is the cause and what to test next.
References
NREL - Degradation Characterization of Thermal Interface Greases (preprint)
Purdue University - A New Accelerated Method to Characterize Degradation of Thermal Interface Materials
Electronics Cooling - Reliability Testing of Thermal Greases
Published by: Fong Yong Chemical Co., Ltd. - specialty chemical and thermal management materials manufacturer, Taiwan. Established 1983.
