The use of light-emitting diodes (LEDs) has become very popular in decorative lighting and signalization applications but a breakthrough in general lighting has not been achieved yet. Main obstacles are the moderate system efficiency and luminous flux. Furthermore, optical and electrical characteristics are strongly dependent on the diode junction temperature, which in turn is determined by the forward current, heat sink, and ambient temperature.
The spectra radiant flux is the primary LED optical characteristic, determining luminous flux and color. A lot of effort has been attributed to model the LED spectral flux and its variation with junction temperature.
One of the first attempts consisted of a rather simple LASTIP simulation of previously measured LED and laser spectra. In 2005 however, a Gaussian spectrum model appeared more successful and this model simplified incorporation of the peak wavelength variation with junction temperature. Very soon afterwards, a double Gaussian model was presented, which is nowadays still used for lighting calculations by the International Commission on Illumination (CIE). A more general approach based on the double Gaussian model was published in 2006. In 2008 a variant of the split Gaussian function with a different exponential behavior on each side of the maximum was reported to be a more simplified approach, still containing the junction temperature as a free parameter. More recently an evaluation of several LED spectrum approximations has been published but due to the large number of fitting parameters an easy-to-use modeling approach is not provided. Furthermore, most spectral radiant flux models that have been published so far are purely mathematical and link with underlying physical principled is missing. Moreover, integration of the junction temperature variation into the initial model parameters often results in piece wise defined and complex models.