MYSTERY of NITRIDE LEDs PIONNERING EXPERIMENT its energy toward the bottom of that valley, the L point, but may also transition back to Γ. These intra- and intervalley relaxations occur through the emission of phonons, primarily longitudinal polar-optical( LO) phonons with an energy of 91 meV in GaN. In all cases, if the initial electron energy is sufficient( that is, higher than the energy separation between the Γ and L valleys plus the 91 meV of the LO phonon), electrons will populate both valleys. If the time required for the Auger-created hot electrons to diffuse to the surface is shorter than the complete relaxation of all electrons from L to Γ, then some electrons with the energy characteristic of the L valley can be emitted into vacuum. In practice, things are a bit more complex: the band profiles near the surface are not flat but bent, so the energy of the Γ and L valleys decreases near the surface, and hot electrons can occupy a range of energies between the valley positions deep in the material and those at the surface( 6 in Fig. 1). To observe such hot electrons, one simply needs to inject electrons and holes into the quantum wells with concentrations large enough to induce the Auger effect. This is done by forward-biasing an LED. As shown in Figure 1, this forward bias allows electrons to be injected from the n-type region into the quantum wells but it does not give them high kinetic energy: they are injected in GaN and InGaN and relax to the Γ minimum of the main conduction band. If they traverse the
Figure 3. Electron emission spectra for various LED currents showing peaks related to the surface( 3), Γ( 2) and L( 1) valleys.
active region( through leakage), they will be emitted with the energy corresponding to the Γ valley of the p-side. In contrast, if recombination occurs in the InGaN well through the Auger process, electrons will populate the higher-energy valleys accessible to them. A collaboration carried out experiments first at the École Polytechnique in Palaiseau, and later at the University of California, Santa Barbara. The LED, engineered with perforations in its upper contact to allow electrons to pass through, was placed facing an electron spectrometer( shown in the photo at the top of the article) in an ultra-high-vacuum chamber. A single atomic layer of cesium was then deposited on
Figure 4. Evolution of the L-valley peak and emitted light intensity as a function of diode current. The extrapolated emission in the absence of droop is shown, defining the“ missing current”( left). This missing current is then used to track the intensity of the L-valley peak( right).
Photoniques 137 I www. photoniques. com 27