Search

This tutorial paper focuses on the physical origin of thermal droop, i.e., the decrease in the luminescence of light-emitting diodes (LEDs) induced by increasing temperature. III-nitride-based LEDs are becoming a pervasive technology, covering several fields from lighting to displays, from automotive to portable electronics, and from horticulture to sensing. In all these environments, high efficiency is a fundamental requirement, for reducing power consumption and system cost. Over the last decade, a great deal of effort has been put in the analysis of the efficiency droop, the decrease in LED internal quantum efficiency (IQE) induced by high current density. On the other hand, an IQE decrease is observed also for increasing temperature, a phenomenon usually referred to as thermal droop. For commercial LEDs, the IQE decrease related to thermal droop can be comparable to that of efficiency droop: for this reason, understanding thermal droop is a fundamental step for making LEDs capable of operating at high temperature levels. In several fields (including street lighting, automotive, photochemical treatments, projection, entertainment lighting, etc.), compact and high-flux light sources are required: typically, to reduce the size, weight, and cost of the systems, LEDs are mounted in compact arrays, and heat sinks are reduced to a minimum. As a consequence, LEDs can easily reach junction temperatures above 85–100 °C and are rated for junction temperatures up to 150–175 °C (figures from commercially available LED datasheets: Cree XHP70, Osram LUW HWQP, Nichia NVSL219CT, Samsung LH351B, and LedEngin LZP-00CW0R) and this motivates a careful analysis of thermal droop. This paper discusses the possible physical causes of thermal droop. After an introduction on the loss mechanisms in junctions, we will individually focus on the following processes: (i) Shockley–Read–Hall (SRH) recombination and properties of the related defects; (ii) Auger recombination and its temperature dependence, including the discussion of trap-assisted Auger recombination; (iii) impact of carrier transport on the thermal droop, including a discussion on carrier delocalization, escape, and freeze out; (iv) non-SRH defect-related droop mechanisms. In addition, (v) we discuss the processes that contribute to light emission at extremely low current levels and (vi) the thermal droop in deep ultraviolet LEDs, also with reference to the main parasitic emission bands. The results presented within this paper give a tutorial perspective on thermal droop; in addition, they suggest a pathway for the mitigation of this process and for the development of LEDs with stable optical output over a broad temperature range. [ABSTRACT FROM AUTHOR]

Defects can significantly modify the electro-optical characteristics of InGaN light-emitting diodes (LEDs); however, modeling the impact of defects on the electrical characteristics of LEDs is not straightforward. In this paper, we present an extensive investigation and modeling of the impact of defects on the electrical characteristics of InGaN-based LEDs, as a function of the thickness of the quantum well (QW). First, we demonstrate that the density of defects in the active region of III-N LEDs scales with increasing thickness of the InGaN QW. Since device layers with high indium content tend to incorporate more defects, we ascribed this experimental evidence to the increased volume of defects-rich InGaN associated to thicker InGaN layers. Second, we demonstrate that the current–voltage characteristics of the devices are significantly influenced by the presence of defects, especially in the sub turn-on region. Specifically, we show that the electrical characteristics can be effectively modeled in a wide current range (from pA to mA), by considering the existence of trap-assisted tunneling processes. A good correspondence is obtained between the experimental and simulated electrical characteristics (I–V), by using—in the simulation—the actual defect concentrations/activation energies extracted from steady-state photocapacitance, instead of generic fitting parameters. [ABSTRACT FROM AUTHOR]

In a combined experimental and numerical investigation, we present the effects of trap-assisted tunneling on the sub-threshold forward bias characteristics of a blue InGaN/GaN single-quantum-well LED test structure grown on a SiC substrate. The different role of donor- and acceptor-like traps has been studied, for the information it can provide on the role played by point defects. Using the energy Et and trap density Nt as the only tunneling-related fitting parameters, the behavior of the measured I( V) curves is well reproduced by our model over a wide current and temperature range. The very good agreement between simulations and experiments suggests that trap-assisted forward tunneling is one of the most relevant contributions to the current flow below the optical turn-on of the diode. [ABSTRACT FROM AUTHOR]

A self-energy-based model for the calculation of the electron capture time through phonon emission in presence of carrier population is applied to nitride-based quantum wells (QWs). The role of plasma dynamical screening on the Frölich interaction is investigated, highlighting its central role in the dependence of capture time on carrier density. The proposed approach yields a closed-form expression for capture time as a function of the carrier densities in QW and barrier states. Its application to simple two-population rate equations allows to reproduce available experimental data with excellent agreement, offering an accurate yet practical alternative to the usual approximation of a constant capture time in modeling light-emitting diodes and lasers. [ABSTRACT FROM AUTHOR]

Recent experiments of electron emission spectroscopy (EES) on III-nitride light-emitting diodes (LEDs) have shown a correlation between droop onset and hot electron emission at the cesiated surface of the LED p-cap. The observed hot electrons have been interpreted as a direct signature of Auger recombination in the LED active region, as highly energetic Auger-excited electrons would be collected in long-lived satellite valleys of the conduction band so that they would not decay on their journey to the surface across the highly doped p-contact layer. We discuss this interpretation by using a full-band Monte Carlo model based on first-principles electronic structure and lattice dynamics calculations. The results of our analysis suggest that Auger-excited electrons cannot be unambiguously detected in the LED structures used in the EES experiments. Additional experimental and simulative work are necessary to unravel the complex physics of GaN cesiated surfaces. [ABSTRACT FROM AUTHOR]

What has turned into highly complex and somewhat misunderstood efficiency loss mechanisms occurring in light-emitting diodes (LEDs) based on the InN-GaN material system at high injection levels are discussed. Suggestions are made as to the dominant mechanism (s) in an open forum format as well as pointing out some of the shortcomings of the methodologies used and premises forwarded. It is unequivocally known that increased junction temperature would cause a reduction in radiative power due to mainly the reduction in radiative recombination efficiency. Another obvious mechanism is the asymmetry in doping in wide bandgap semiconductors, such as GaN, wherein the hole concentration lags well behind that of electrons in the active region. Because an electron and a hole are required for radiative recombination, the radiative efficiency cannot keep up with increasing carrier injection due to progressively lagging hole population. This results in either electron escape without radiative recombination or electron accumulation, which in turn changes the internal bias of the device, manifested as reduced internal forward bias, which reduces the rate of increase in light intensity. Some of the reports ascribe the efficiency loss at high injection levels to Auger recombination (mainly through indirect and recently reportedly direct deductions) as the main and or the only source of efficiency loss by in many cases simply relying on the temperature and injection independent (not well taken) A, B, C coefficients to fit a third order polynomial to the efficiency vs. injection current. As for the direct deduction, the spectroscopic analysis of Auger kicked hot electrons as they traverse through the G and L bands before being emitted into the vacuum by means of cesiated surface challenges the existing theories and some experiments regarding carrier scattering and G-L separation. Despite just a few reports to the contrary, the bulk of the resonant optical emission experiments do not support the Auger argument as being the main cause. In parallel, there exists a body of theoretical and experimental reports for electron overflow of ballistic/quasi-ballistic electrons traversing the active region to p-GaN, escaping recombination altogether in the active region. In fact and from the get go, the LED industry ubiquitously employed, and continues to do so, an (Al,In)GaN electron-blocking layer (EBL) to prevent electron escape for improved light output that in and of itself would more than suggest that the electron escape (overflow) does indeed occur. The only adverse effect of EBL is that it impedes hole injection due to the valence band offset between the p-type (Al,In)GaN EBL and p-GaN and also generates piezoelectric (if not lattice matched) and differential spontaneous polarization induced fields that pull down the conduction band edge at the interface reducing EBL's effectiveness. To at least reduce the aforementioned aggravating factors to some extent, the electron overflow and the associated efficiency loss can be reduced substantially (particulars of which depend on the active layer design) by inserting a stair-case electron injector (SEI) with a step-like indium composition to act as an "electron cooler" or by linearly graded cooler in some form or another prior to the active region. Use of multiple layered heterostructures for the active layers also plays the role of electron cooler, albeit not necessarily in the most optimum fashion. As if oblivious to the raging issues alluded to above, the LED industry has been moving along very successfully with a 2-prong approach. In one, dubbed the "high voltage LED", a set of LEDs (most likely configured in the form of a full-wave bridge rectifier) operating at low currents, where the efficiency is at its maximum, is used, which also has the added benefit of much reduced power supply complexity and weight. The other is the continual improvement of layer quality and optimum active layer design, taking technological parameters into consideration, which at the time of writing sported 63% wall-plug efficiency at an injection current of 350 mA for ~1.1 mm x 1.1 mm LEDs, which translates to approximately 75% efficiency once the voltage (about 3 V) and phosphor conversion efficiencies are taken into account. Assuming the same extraction and external quantum efficiencies, one obtains about 86% for each. It would not be an exaggeration to conclude that mid 90% internal quantum efficiencies are very likely in play here, which means that the purported inherent problems dominating the discussion are practically reduced to an academic exercise. Despite aggravating factors involving holes and hetero-barriers, commercial LEDs almost invariably use multilayer heterostructure designs for the active layers, loosely termed as multiple quantum wells, presumably to circumvent technological challenges and the ramifications of the polarization induced field even though double heterojunction varieties could be preferable from the point of view of hole transport. Eventually, the real limiting factors are the extent of hole supply (can be mitigated by increased hole concentration, which is well known and continually explored) and the proverbial thermal wall (can be mitigated by increased efficiency and efficient heat removal) of course. [ABSTRACT FROM AUTHOR]

Electroluminescence (EL) characterization of InGaN/GaN light-emitting diodes (LEDs), coupled with numerical device models of different sophistication, is routinely adopted not only to establish correlations between device efficiency and structural features, but also to make inferences about the loss mechanisms responsible for LED efficiency droop at high driving currents. The limits of this investigative approach are discussed here in a case study based on a comprehensive set of currentand temperature-dependent EL data from blue LEDs with low and high densities of threading dislocations (TDs). First, the effects limiting the applicability of simpler (closed-form and/or one-dimensional) classes of models are addressed, like lateral current crowding, vertical carrier distribution nonuniformity, and interband transition broadening. Then, the major sources of uncertainty affecting state-ofthe- art numerical device simulation are reviewed and discussed, including (i) the approximations in the transport description through the multi-quantum-well active region, (ii) the alternative valence band parametrizations proposed to calculate the spontaneous emission rate, (iii) the difficulties in defining the Auger coefficients due to inadequacies in the microscopic quantum well description and the possible presence of extra, non-Auger high-current-density recombination mechanisms and/or Auger-induced leakage. In the case of the present LED structures, the application of three-dimensional numerical-simulation-based analysis to the EL data leads to an explanation of efficiency droop in terms of TD-related and Auger-like nonradiative losses, with a C coefficient in the 10-30 cm6/s range at room temperature, close to the larger theoretical calculations reported so far. However, a study of the combined effects of structural and model uncertainties suggests that the C values thus determined could be overestimated by about an order of magnitude. This preliminary attempt at uncertainty quantification confirms, beyond the present case, the need for an improved description of carrier transport and microscopic radiative and nonradiative recombination mechanisms in device-level LED numerical models. [ABSTRACT FROM AUTHOR]

2-D numerical simulation is employed to assess a number of possible design approaches aimed at optimizing the internal quantum efficiency (IQE) of ZnO-based light-emitting diodes (LEDs) grown along the c-axis. First, the relative performance of similar ZnO-based and GaN-based LED structures is compared and discussed. Second, the effects on IQE of thickness, doping, and alloy composition of the MgZnO electron blocking layer (EBL) is studied in order to maximize the carrier confinement in the active region. The optimum number of quantum wells is also addressed, and different strategies for barrier doping are considered, showing that, if the EBL is doped p-type, a similar doping in the barriers is not required to compensate for the spontaneous and piezoelectric interface charges and to enhance hole transport. Different choices of the geometrical and doping parameters of the n-type access region are considered, and the impact of different values of the electron mobility is determined. Finally, the analysis of a ZnO/BeZnO LED structure suggests that the incorporation of BeZnO layers does not provide significant advantages. [ABSTRACT FROM PUBLISHER]

As a preparatory step toward establishing reliable numerical design tools for ZnO-based optoelectronic devices, we have reassessed the available information on material parameters relevant for the simulation of light-emitting diodes (LEDs) with active regions including ZnO, MgZnO, and BeZnO layers. The impact of different approximations for the electronic structure and the interface polarization charge on the optical properties of bulk ZnO and ZnO/MgZnO quantum wells has been evaluated, and a consistent set of parameters has been used not only for systematic comparison of ZnO/MgZnO and ZnO/BeZnO single quantum well structures but also for the first simulation of a realistic ZnO/BeZnO multiple quantum well LED. [ABSTRACT FROM AUTHOR]

Physical mechanisms causing the efficiency droop in InGaN/GaN blue light-emitting diodes and remedies proposed for droop mitigation are classified and reviewed. Droop mechanisms taken into consideration are Auger recombination, reduced active volume effects, carrier delocalization, and carrier leakage. The latter can in turn be promoted by polarization charges, inefficient hole injection, asymmetry between electron and hole densities and transport properties, lateral current crowding, quantum-well overfly by ballistic electrons, defect-related tunneling, and saturation of radiative recombination. Reviewed droop remedies include increasing the thickness or number of the quantum wells, improving the lateral current uniformity, engineering the quantum barriers (including multi-layer and graded quantum barriers), using insertion or injection layers, engineering the electron-blocking layer (EBL) (including InAlN, graded, polarization-doped, and superlattice EBL), exploiting reversed polarization (by either inverted epitaxy or N-polar growth), and growing along semi- or non-polar orientations. Numerical device simulations of a reference device are used through the paper as a proof of concept for selected mechanisms and remedies. [ABSTRACT FROM AUTHOR]

Direct interband and intraband Auger recombination due to electron-electron-hole and hole-hole-electron transitions in bulk InGaN is investigated by first-order perturbation theory including Fermi statistics, realistic electronic structures obtained by nonlocal empirical pseudopotential calculations, and their corresponding wavevector-dependent dielectric functions. Our results confirm that the intraband Auger coefficient is negligible in alloy compositions relevant for solid-state lighting and indicate that the resonant enhancement associated with interband transitions for wavelengths ranging from blue to green cannot account for the efficiency droop experimentally observed in GaN-based light emitting diodes. [ABSTRACT FROM AUTHOR]