When making temperature measurements with infrared cameras and point detectors, we are always facing the question of the emissivity of our sample. In practice we would like to have a single number set in the IR device. To approach the question, let’s first recall some fundamental definitions.
Planck’s radiation law gives the energy density emitted by a blackbody as a function of wavelength and surface temperature. A blackbody is defined as a perfect absorber for all incident radiation, which in thermal equilibrium emits as much energy as it absorbs. However, no material is a perfect emitter. The ratio of the emitted radiation energy densities between the sample and the blackbody is defined as the emissivity of the sample. But, as mentioned earlier, the emitted radiation, and hence the emissivity, depends on the wavelength and temperature.
However, in the case of measuring temperature with an IR device, the situation gets fewer parameters when we remember that the detectors are sensitive only to some wavelength window. In the modern quantum well cameras the sensitivity range of the detector is 8 – 9 �m; for the so-called short wave cameras it is 3 – 5 �m; and for the so-called long wave cameras it is 8 -12 �m. Thus, in practice we are interested only in the wavelength for which our camera is sensitive. Also, the temperature that matters is the temperature of the measured object.
Typically in the data tables listing emissivities of various materials, the so-called total emissivity is given, representing the radiation integrated over the whole spectrum. The temperature is normally selected to be close to room temperature. When performing an IR measurement, we are getting information only related to the radiation in the measurement direction, which is usually close to the sample surface normal direction. It should be clearly understood that this value is not the value needed for numerical calculations involving radiation heat transfer; that is, the total hemispherical emissivity integrated over angles of a half-space. In some sources of reference, both normal and hemispherical emissivities are given.
A simple way to determine the emissivity setting needed for our sample is to measure it with the camera at a known elevated temperature and to adjust the camera emissivity setting so that the temperature reading is correct. When the actual measurements are made for temperatures not too far away from this calibration temperature, we can obtain a reasonably good accuracy for temperature measurements.
Another important issue to remember is that the emissivity is not really a material property but a surface property. As such, the value also depends on surface conditions. Consequently, we must ask ourselves whether the surface at which we are aiming the IR device is really the surface of the material that we think it is. It may easily be the case that the sample surface has a thin coating, which might be water, grease, or an oxide layer too thin to be seen by the naked eye but capable of affecting the emitted radiation. Thus, if we have previously measured some emissivity value of the sample, it might not be the same any more when we make the next measurements.
Therefore, common practice is to paint the sample black, which both stabilizes the emissivity and increases it. Typically the paint emissivity value is around 0.95. However, the problem remains that the heat transfer is changed when compared with the unpainted case. If the IR device allows a small spot size relative to the sample, a better solution is to use a black dot on some larger area of uniform temperature and, according to its known emissivity, to calibrate the emissivity of the untreated part.
Also, the emissivity value in the surface normal direction depends largely on the surface roughness, as the angular dependence of the radiation changes. Thus, there may be large local variations in the normal emissivity values determined by using the IR camera.
Silicon is a good example of a material whose emissivity depends largely on the wavelength: for the 8-9 �m range, the emissivity is 0.6; and for the 3-5 �m range, it is only 0.1 [1]. Also polymer materials dealt with in electronics are problematic as they are partially transparent to IR radiation, and their physical properties depend on the compound and manufacturing method used. A typical value for epoxies is 0.8 – 0.9. This is valid both for epoxy materials used as package encapsulants and for FR4. More generally, most plastics have emissivity values around 0.8 [2].
For most metallic materials the emissivity increases with temperature, and for non-conducting materials the behavior is typically opposite. Also, the absolute emissivity value increases with increasing electrical resistivity, as stated by the Hagen-Rubens emissivity relation. Some examples of tabulated total normal emissivities appear in Table 1 [1].
Table 1. Tabulated Total Normal Emissivities
|
References
- Touloukian, Y.S. and Ho, C.Y. (Eds.): Thermophysical Properties of Matter, Plenum Press, New York, 1972.
- Kreith, F. and Bohn, M.S.: Principles of Heat Transfer, 4th Edition, Harper & Row, New York, 1986.
