Measuring Thermal Diffusivity of Thin Ceramics
Until now, determining the thermal diffusivity of extremely thin and thermally highly conductive ceramic substrates has been error-ridden, but this could be the optimal approach.
This article is published by EEPower as part of an exclusive digital content partnership with Bodo’s Power Systems.
Only with a perfect thermal design can power modules function reliably over their entire service life under all operating conditions. To achieve this, it is necessary to determine the thermal properties of all components as precisely as possible. An important parameter here is the thermal conductivity of the active metal-brazed ceramic substrate (AMB) used as a carrier which requires a reliable measuring method of thermal diffusivity of the ceramic. The thermal diffusivity is generally determined quickly and non-destructively using the laser flash or light flash method (LFA). In this process, a short light or laser flash heats a sample on one side while an infrared sensor measures the temperature rise on the opposite side. The thermal diffusivity can be calculated using the half-rise time of the signal and the sample thickness.
Image used courtesy of Adobe Stock
The measurement of very thin and highly conductive materials places special demands on the sample preparation and the measuring device used. A team from Rogers Germany GmbH, in collaboration with colleagues from the Fraunhofer Institute for Ceramic Technologies and Systems IKTS, scrutinized the common measurement method for determining thermal diffusivity using 0.32 millimeter thin Si3N4 ceramics. It is the material of choice for a wide range of power modules due to its fracture toughness and thermal conductivity.
Better Dipping Than Spraying
For a meaningful measurement, the sample must not allow light to pass through it and must show sufficient absorption and emission behavior. The usual preparation method essentially consists of first applying an opaque layer of gold by sputtering and then spraying thermally conductive graphite over it. However, the thinner the sample, the greater the influence that the type and quality of the coating has on the measurement result. The sputtering process is largely undefined, and the manual spraying of the graphite also has only low reproducibility in terms of its layer thickness and homogeneity.
The team, therefore, also investigated two alternative preparations with automatic dip coating. They used an automatic dip coater from Ossian to dip the samples into a graphite-isopropanol solution. Four 10-by-10-millimeter squares were cut from a single silicon nitride plate for each method. Four samples were prepared in the usual way. Four were coated with gold and automatically dipped into the graphite solution. Four were finally dip-coated with graphite only. The team then determined the thermal diffusivities of all the samples using two different LFA measuring devices from Netzsch Gerätebau GmbH, which will be discussed in more detail later.
The samples prepared with the standard method showed the lowest values (74 W/mK at 25°C), while the gold-coated substrates dipped in graphite achieved the highest thermal diffusivities (82 W/mK at 25°C). The results of the ceramic plates dipped only in graphite were between the two extremes (78 W/mK at 25°C).
For the measurement of highly conductive and thin materials, the team recommends the new dip coating method for applying graphite. It eliminates the manual influence, offers the best reproducibility, and allows fine adjustment of the amount of graphite deposited.
Figure 1. Measured values of the individual LFA models in the 25 - 150°C temperature range. Image used courtesy of Bodo’s Power Systems [PDF]
Pulse Width Is Decisive
The higher the thermal diffusivity of a sample, the steeper the rise of the signal in the LFA measurement. For reliable measurement of a thin and well-conducting material, the light pulse must be so short that the infrared sensor on the opposite side only detects a rise in temperature when the light is switched off again.
For their investigations, the team used three different LFA models from Netzsch-Gerätebau GmbH, which differ among others in their minimum pulse width. For the LFA 427, it is 0.1 milliseconds; for the LFA 447, 0.06 milliseconds; and 0.01 milliseconds for the LFA 467 Hyper-Flash. On samples sputtered with gold and dip-coated with a thin graphite layer, the LFA 427 determined the thermal diffusivity to be 70 W/mK, the LFA 447 came to 75 W/mK, and the LFA 467 Hyper-Flash measured 80 W/mK, in each case at 25°C.
The investigations show that not all LFA devices are suitable for measuring thermal diffusivity on thin ceramic substrates with high thermal conductivity. The smaller the pulse width, the faster the increase in the temperature of the sample can be. Reliable measurement of thin and highly conductive ceramics requires a measuring device with fast data acquisition and a sufficiently small pulse width.
This article originally appeared in Bodo’s Power Systems [PDF] magazine.