To investigate the variation in internal quantum efficiency in InGaN structures, we measure the differential carrier lifetime of an InGaN/GaN double-heterostructure light-emitting diode under varying electroluminescence injection conditions. By coupling this measurement to an internal quantum efficiency measurement, we determine the carrier density and the radiative and nonradiative contributions to the lifetime without making any assumptions on recombination processes. We find that droop is caused by a shortening of the nonradiative lifetime with current. The observed shortening of both radiative and nonradiative lifetimes with current is found to be in excellent agreement with an ABC model including phase-space filling.

We study the carrier distribution in multi quantum well (multi-QW) InGaN light-emitting diodes. Conventional wisdom would assume that a large number of QWs lead to a smaller carrier density per QW, enabling efficient carrier recombination at high currents. We use angle-resolved far-field measurements to determine the location of spontaneous emission in a series of multi-QW samples. They reveal that, no matter how many QWs are grown, only the QW nearest the p layer emits light under electrical pumping, which can limit the performances of high-power devices.

We study differential carrier lifetimes in InGaN light-emitting diodes ͑LEDs͒ of varying wavelengths. Increase in wavelength is correlated with an increase in lifetime, due to the impact of the polarization fields on carrier overlap. This effect explains the early onset of droop in longer-wavelength LEDs.Efficiency droop is a nonthermal process of disputed origin which quenches the internal quantum efficiency ͑ IQE ͒ of GaN-based light-emitting diodes ͑LEDs͒ at high current density ͑J͒. One point of particular interest is the relationship between emission wavelength ͑͒ and droop. Early on, it was noticed that in LEDs emitting at longer the onset of droop occurs at lower J. 1 This is an important limitation for longwavelength high-power LEDs, especially in the green range. It has been suggested that this could be due to increased carrier leakage 2 with In content or to a carrier delocalization 3 from In-rich areas.We have recently shown evidence that droop is a bulk phenomenon occurring even in photoluminescence in bulk GaN. 4 We have also shown that droop is related to a shortening of the nonradiative carrier lifetime, manifesting the onset of a nonradiative recombination phenomenon behaving roughly ϳn 3 where n is the carrier density. 5 In this letter, we explore the relationship between and carrier lifetime, and show how the aforementioned effects can largely be explained by the influence of the polarization fields on carrier wave functions.We study a set of five ͑0001͒ InGaN LEDs with multiquantum-wells ͑MQW͒ of varying , whose characteristics are indicated in Table I. The first four samples have the same nominal active region ͑AR͒ design and a varying In content, while the fifth sample has nominally the same In content as the fourth one but slightly different AR layer thicknesses, yielding a longer . We measure IQE for each sample as described in Ref. 5 ͓Fig. 1͑a͔͒. We note that its peak value peak is similar for all samples, while the onset of droop happens at lower J for longer-wavelength samples. We stress the importance of comparing samples of similar IQE , because variations in crystal quality influence the shape of the efficiency curve-namely, larger Shockley-Read-Hall ͑SRH͒ recombinations push the curve toward higher J which can mask a more fundamental variation in droop. Figure 1͑b͒ shows the same curves, normalized to unity and with a linear J axis, making the effect of droop onset more obvious.We perform differential carrier lifetime ͑͒ measurements on these samples, as described in Ref. 5. By integrating , we obtain the carrier density n. By coupling this measurement to that of IQE , we access the radiative and nonradiative lifetimes R and NR , and the corresponding recombination rates G R and G NR . As in Ref. 5, we can now derive three quantities: a = G NR / n, b = G R / n 2 , and c = G NR / n 3 . These are useful because they enable comparison to an ABC model, where the coefficients A, B, and C stand for SRH, radiative, and Auger coefficients, respectively. 6 Namely, in an ideal ABC model, a sho...

No abstract

A multi-color light emitting diode (LED) using two distinct active regions connected via a tunnel junction was grown by rf-plasma-assisted molecular beam epitaxy. The LED is contacted through n-type layers, one of which provides efficient contact to a p-type layer via a second tunnel junction. The tunnel junctions used in the structure use the high polarization fields found in the III-nitrides material system to narrow the depletion width of the junction. Two peaks at 405 nm and 490 nm are observed in the electroluminescence spectrum of the LED, and current-voltage characteristics indicate a turn-on voltage of roughly 7.5 V and dynamic series resistance of 19 ª at 80 mA for a ¿¼¼ µm square device.