PIONNERING EXPERIMENT MYSTERY of NITRIDE LEDs
The decisive physics experiment
Figure 2. Experimental set up with LED and electron analyzer. Blue LED in operation during measurement, emitting both blue photons and electrons.( photo credit: Ph. Lavialle, École Polytechnique)
efficiency LEDs, met with ambiguous results as they did not improve over state-of-the art LEDs, suggesting that the root cause had likely not been identified. An alternative and not entirely incompatible explanation also existed: the Auger effect( recently renamed the Auger – Meitner effect). During the recombination of an electron with a hole in an InGaN quantum well, the released energy can, instead of generating a photon, be transferred to another electron( or hole), giving it a large kinetic energy within its band( 4 in fig. 1). This excess energy is then dissipated as heat( phonons)( 5 in fig. 1). This process directly reduces luminous efficiency as an electron hole pair has disappeared in the process and becomes significant when carrier densities are high( its probability scales roughly with the cube of the charge density n 3) and thus under high current injection. This led to the emergence of the so-called ABC models, which describe LED efficiency as a function of current, incorporating contributions from defect-related recombination( A · n), radiative recombination( B · n ²), and Auger recombination( C · n ³). These models provided remarkably good fits to experimental data. However, the somewhat arbitrary choice of the A, B, and C coefficients( and later refinements to the model [ 1 ]) meant that the argument was not yet watertight. Moreover, proponents of the electron leakage theory claimed that leakage could itself be modeled with an n ³ dependence. Well known in small – bandgap materials, the Auger process was long considered negligible in wide bandgap materials like nitrides, where energy and momentum conservation make such interactions unlikely. For this reason, the hypothesis initially seemed implausible. In 2007, a precise analysis of the change in carrier lifetime in photoluminescence concluded that an n 3 mechanism, seen as a signature of an Auger effect, explained the data. Then, around 2011, Auger processes were revisited theoretically [ 2 ], taking into account various indirect mechanisms. Specifically, the involvement of phonons or changes in wavefunction extension caused by defects, alloy disorder, and other factors can relax momentum conservation, making the process 100 to 1000 times more probable than previously predicted. Doubt began to creep into the nitride LED community, exposed to several alternative explanations: could the droop indeed be linked to the Auger effect? What was missing was a direct, irrefutable experimental proof that could reveal the true origin of the droop. And that proof finally came through a decisive physics experiment were the hot electrons generated by an Auger event could be observed.
Electron emission is, historically, a technique from surface physics seemingly far removed from the internal physical mechanisms invoked to solve the issues surrounding LEDs. It is at the root of the photoelectric effect. Its basic principle is to observe the energy of electrons emitted from a surface when photoexcited as they retain the energy they had inside. To do this, electrons must first be given enough energy to escape the material. This gives a relation between the photon energy and the maximum energy of the outgoing electron which led Einstein to the photon hypothesis and Millikan to determine the value of the Planck constant by relating the maximum energy of the photoemitted electron from metals to the photon energy. The principle of the experiment carried out in LEDs is to use emitted electrons to measure electron energies inside the semiconductor. To study semiconductors, it is beneficial to lower the vacuum level, by coating the surface with a monolayer of cesium, which reduces that level so much that an electron in the semiconductor conduction band has a higher energy than the one in vacuum. Analyzing the energy of electrons emitted then probes the energy distribution of electrons above the conduction band near the surface. This turns out to be crucial for the identification of the Auger processes occurring inside materials. In the Auger effect, an electron initially thermalized at the bottom of the semiconductor conduction band, the so-called Γ conduction band, gains the energy of the electron-hole pair, the bandgap energy, either within that same band or in higher ones 4. The electron relaxes down its energy by returning to the minimum of the Γ valley, so-called Γ point, or by transferring to another higher lying conduction band( L band, 5 in Fig. 1). In that latter case, an electron in the L valley can relax part of
26 www. photoniques. com I Photoniques 137