The temperature of the active region crystal lattice, frequently referred to as the junction temperature, is a critical parameter. The junction temperature is relevant for several reasons. Firstly, the internal quantum efficiency depends on the junction temperature. Secondly, high - temperature operation shortens the device lifetime. Thirdly, a high device temperature can lead to degradation of the encapsulant. It is therefore desirable to know the junction temperature as a function of the drive current.

Heat can be generated in the contacts, cladding layers, and the active region. At low current levels, heat generation in the parasitic resistances of contacts and cladding layers is small due to the I2R dependence of Joule heating. The dominant heat source at low current levels is the active region, where heat is created by non-radiative recombination. At high current levels, the contribution of parasitics becomes increasingly important and can even dominate.

There are several different ways to measure the junction temperature, which include micro - Raman spectroscopy (Todoroki et al., 1985), threshold voltage (Abdelkader et al., 1992), thermal resistance (Murata and Nakada, 1992), photothermal reflectance microscopy (Epperlein, 1990), electroluminescence (Epperlein and Bona, 1993), photoluminescence (Hall et al., 1992) and a non-contact method based on the peak ratio of a dichromatic source (Gu and Narendran, 2003). Most methods are indirect methods that infer the junction temperature from an easily measurable parameter. In this chapter, we discuss a method based on the shift of the peak emission wavelength with the temperature and a method based on the shift of the diode forward voltage with temperature. We also discuss the carrier temperature as inferred from the high-energy slope of the emission spectrum.