The pump-out effect is a well-known and annoying problem when using thermal compounds on CPUs, GPUs and other electronic components subject to high thermal loads. It describes the displacement of the thermal paste from the contact area between the processor and heat sink, for example, typically as a result of temperature cycles or mechanical loads. This leads to a reduction in heat transfer capacity and can jeopardize the functionality and service life of the affected systems in the long term or even lead to total failure.
The mechanisms behind this effect are closely linked to the thermal and mechanical properties of the paste and are largely unknown to the average user. This is exactly what I would like to change, especially as these questions keep coming up in connection with my thermal paste database and the durability estimates made there. In this article, I will systematically analyze the causes of the pump-out effect, supplemented with possible measures to avoid it and an explanation of my test setup for estimating the possible behavior of such a paste.
Causes of the pump-out effect
The pump-out effect is based on a complex interplay of thermal, mechanical and chemical factors. A central reason is the different thermal expansion of the components of the thermal compound. The matrix of the paste, usually consisting of silicones or polymers, has a significantly higher coefficient of thermal expansion than the filler particles embedded in it or the solid metal contact surfaces. With repeated temperature cycles, this leads to stresses that push the paste to the edge of the contact area.
Another important aspect in this context is the outgassing of the thermal paste. Many pastes consist of volatile organic compounds that can evaporate at higher temperatures. This process, also known as “outgassing”, leads to a change in the chemical and physical properties of the paste. Over time, this can reduce thermal conductivity and leave residues on adjacent components. These residues can have a potentially corrosive effect or impair the function of sensitive electronic components. Possible air inclusions also play a role here, which can occur due to poor production or improper filling. This is always a good indicator of a paste with limited long-term durability.
In addition to thermal stresses, mechanical movement between the contact surfaces also plays a role. During the operation of processors or GPUs, vibrations or minimal relative movements can cause shear forces that displace the paste sideways. The matrix, which reduces its viscosity at higher temperatures, offers less resistance to these forces and is therefore pushed out of the contact area more easily.
Another factor is the lack of adhesion of the paste to the contact surfaces. Differences in chemical interactions between the matrix and the surface materials promote displacement. Another indicator of good adhesion is the contact resistance, i.e. the so-called interface resistance, which I always determine. This can be used to derive how well the surface of the material “clings” to the contact surfaces (IHS, heatsink). These values are also easily comparable and meaningful in the database, as they are always the same calibrated reference blocks. I have explained how to determine this value in detail often enough in the linked basics, so I won’t go into that here. But it is the value that can have a major influence with very low BLT. Here you can see an excerpt from my database:
If the matrix loses volatile components as the vapor pressure increases, there is a loss of volume, which further reduces the mechanical stability. The layer geometry of the paste also influences the effect. Thin layers, which are necessary to keep the thermal resistance low, are particularly susceptible to thermal stresses and mechanical displacement. Ageing of the paste also contributes to the pump-out effect. Repeated temperature cycling can change the chemical structure of the matrix, reducing its elasticity and adhesion. Such processes reinforce the degradation mechanisms and increase the susceptibility to the pump-out effect over longer periods of time.
47 Antworten
Kommentar
Lade neue Kommentare
Mitglied
1
1
Mitglied
Urgestein
1
Mitglied
Veteran
1
Mitglied
1
Veteran
Urgestein
1
Mitglied
Urgestein
Veteran
Urgestein
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →