engineering approaches used for modeling III-nitride LEDs are based on a drift-diffusion model of carrier transport in heterostructures. This model has an intrinsic drawback: it predicts remarkably overestimated operation voltage of LEDs with a large number of deep InGaN QWs. The reason for that can be*
understood from Fig.8a where the band diagram is plotted for an MQW LED structure emitting light at*
515 nm. Because of large band offsets at the QW interfaces, every well is surrounded by potential barriers where the electron/hole concentration becomes extremely small,
*and lower, next to the*
barrier/QW interfaces. A low electrical conductivity of these regions results in a ladder-like conduction*
and valence band alignment shown in Fig.3a and, eventually, in unrealistically high operation voltage,
greater than 5 V at 30 mA (Fig.3c). Besides, the band alignment like that shown in Fig.8a implies a*
strong ballistic electron leakage to occur in such a structure [42]. *
First of all, some important materials properties of III-nitride compounds, like conduction and valence*
band offsets, spontaneous electric polarization, deformation potentials, ionization energies of donors and*
acceptors, hole effective masses, Auger recombination coefficients, etc., are still known with insufficient*
accuracy. In addition, the necessity of using advanced physical models requires knowing additional parameters related, for instance, to TDs or localized electron/hole states. The properties of accompanying*
materials used for LED fabrication also need a careful evaluation. For example, the optical properties of*
metals used as electrodes have a strong dispersion in the visible and UV spectral ranges, which should be*
accurately considered to predict properly LEE from an LED die. Another example is ITO having interrelated electrical and optical properties. Making simulations, one should specify the electrical conductivity*
of the ITO film, as well as the electron mobility, in order to estimate the materials optical characteristics*
depending on both parameters [30]. *
*Not only the reliable materials properties but also some additional information on the LED structure is*
frequently needed for accurate simulations. First, the LED band structure is quite sensitive to distribution*
of polarization charges depending on particular composition profile and strain. Because of indium surface segregation, the InGaN QW profile may by much different from the nominal one, providing considerable broadening and smoothing of the QW interfaces and penetration of indium in the QW barriers*
[45,46]. Second, the strain in the well may depend on both the composition profile and stress relaxation*
occurring in the layers preceding the QW. And, considerable Mg redistribution next to the active region*
during growth of various LED structures has been reported, eventually affecting the emission efficiency*
[47,48]. All the above effects are crucial for LED operation and can be allowed for in simulation on the*
basis of additional experimental information or advanced theoretical studies. *
4 Conclusion As discussed above, there is still a big room for improvement of the existing physical*
models and building up new models accounting for specific properties of III-nitride materials and devices. In addition, much effort should be made to understand better effects of III-nitride compound microstructure on the carrier transport and recombination processes. Such developments are expected to*
form the main stream in the future research. However, this will not provide any guideline for understanding the impact of technological factors and, first of all, of heterostructure growth conditions on the operation and characteristics of III-nitride LEDs, despite its evident importance. On the other hand, it is hard*
to expect a rapid progress in experimental studies of such issues, as strain relaxation and impurity redistribution in LED structures during growth, modification of the QW composition profile and instability of*
the QW thickness caused by indium surface segregation, as well as other technological factors. In this
respect, coupled modeling of heterostructure growth and device operation would be a solution to this*
problem. The analysis of indium incorporation during epitaxial growth and optical transitions in strained*
InGaN/GaN materials and QWs [49] provides an example of a small first step forward in this direction.*
Alternatively, detailed experimental information has to be invoked to highlight the contribution of technological factors to formation of the composition, *doping, and defect density profiles in LED heterostructures
) light on the matter would be appreciated, and illustrative graphs doubly so.
