Real laboratory tests of thermal pastes on igor’sLAB – Part 1 – Measurement setup and methodology

1 - Introduction and basics 2 - Thermal conductivity and disruptive factors 3 - From thermal resistance to thermal conductivity 4 - How the TIMA5 works with ASTM D 5470-17

Not all thermal conductivity is the same!

What the suppliers usually state in their data sheets is an idealized thermal conductivity (λ) or the so-called bulk thermal conductivity and it is measured (if at all) under ideal conditions or subsequently extrapolated, but without taking into account any influences of macroscopic heterogeneities or structural irregularities. In the end, however, you don’t really get very far. Let’s first compare the terms in order to understand the differences.

General or bulk thermal conductivity (λ)
Bulk thermal conductivity (often simply referred to as thermal conductivity) is a physical property of materials that indicates how well a material conducts heat. It is a measure of a material’s ability to transport thermal energy through itself and is therefore an important parameter in many thermal applications and technologies. Thermal conductivity (λ or k) is defined as the amount of heat that flows through a unit area of a material per unit time when a temperature gradient of one Kelvin per meter (K/m) is maintained along the direction of heat flow.

It is used for homogeneous, isotropic materials where the thermal conductivity is the same in all directions and over the entire material. The thermal conductivity of a material is measured under controlled conditions, usually at a constant temperature. Sometimes it is also referred to as bulk thermal conductivity, although this term should normally only be used for homogeneous materials. Alternatively, the material is idealized and simplified as being homogeneous with a defined, fixed mean value.

The measurement is typically taken in a controlled environment with a uniform temperature distribution and without taking external factors into account. However, what is missing in this value is the so-called heat transfer resistance for solid-liquid transitions (e.g. IHS-> paste and paste-> heatsink). This takes into account the way in which the heat is transferred between and not in the media involved and depends on the thermal properties of the media involved and the geometry of the transition area.

Effective thermal conductivity (λ_eff)
Effective thermal conductivity is a concept used to describe the thermal conductivity of inhomogeneous or anisotropic materials or material systems, such as those found in thermal pastes and pads. These materials usually consist of different components, each of which has different thermal conductivities. The effective thermal conductivity takes into account the influence of these different phases and their spatial distribution. The effective thermal conductivity is either measured or calculated using models or empirical formulas that take into account the geometry, distribution and properties of the individual components of the material. It is also what I will be measuring in real and complex terms from now on. The effective thermal conductivity of a material can therefore be significantly lower than the thermal conductivity of its individual components. This is due to various factors that influence heat transfer in heterogeneous materials. This is exactly what I’m coming to now.

Reasons why the effective thermal conductivity is often so much lower

• Phase boundaries and interfacial resistances:
  In multiphase materials such as pastes and pads, there are numerous interfaces between the different phases. These interfaces pose a resistance to heat transport as they often have atomic irregularities and inconsistencies in the crystal structure that impede heat flow.
• Inclusions:
  Some thermally conductive materials contain pores or air bubbles, usually filled with air or another gas. Since air and most gases have a much lower thermal conductivity than solids, they significantly reduce the effective thermal conductivity of the material.
• Geometric arrangement:
  The arrangement of the different phases in the material plays a major role. If the phases are arranged in such a way that the heat flow often alternates between materials with high and low thermal conductivity, the overall thermal conductivity is reduced. For the usual pastes with corundum and zinc oxide, the degree of grinding must be sufficiently optimized.
• Influences of defects and impurities:
  Defects, impurities and contaminants in the materials can further reduce the thermal conductivity as they disrupt the orderly flow of phonons (particles carrying thermal energy). I have already found water in a paste that was caused by an outdated and partially defective system. This is not so rare.
• Microstructure:
  The microstructure of the material, including the size, shape and distribution of the different phases, influences the effective thermal conductivity. Complex and irregular microstructures can significantly hinder the flow of heat (see also picture above). Here again, the degree of grinding comes into play, but not only.
• Thermal barriers:
  Materials made up of different components often have thermal barriers at the interfaces between these components. These barriers can be caused by differences in thermal expansion coefficients, mechanical stresses or chemical incompatibilities and contribute to a reduction in effective thermal conductivity. This fact is very important and also explains why, for example, applied thermal paste should not be allowed to dry!
• Economies of scale:
  In nanoscale materials or structures, the laws of thermal conduction can work differently than in macroscale materials. The effective thermal conductivity can be further reduced by quantum effects, interfacial resistances and other nanoscale phenomena. Ok, this is getting too detailed, but let’s just take note of it.

In practice, the effective thermal conductivity is often determined experimentally by testing the material and measuring the temperature difference and heat flow. These measured values are then used to calculate λeff. This is also exactly what I will implement in my measurements.