Direct measurement of hot-carrier generation in a semiconductor barrier heterostructure: Identification of the dominant mechanism for thermal droop

Abstract

Energy measurements of electrons emitted from a semiconductor can reveal internal physical processes hitherto elusive. Signatures of hot-electron processes in heterostructures have been observed from cesiated, light-emitting and p-in diodes. In pin devices with AlGaN barriers, a high energy peak was measured and ascribed to a trap-assisted Auger recombination process. Temperature dependent measurements of light-emitting diodes with AlGaN electron blocking layers also show such hot carriers when electrons thermally reach these barriers, identifying carrier escape as the mechanism of thermal droop and demonstrating the efficacy of such barriers to partially mitigate thermal droop.

Evidence for trap-assisted Auger recombination in MBE grown InGaN quantum wells by electron emission spectroscopy

“…The EDCs show a low energy peak, consistent with the expected energy position of electrons emitted from the -valley conduction band minimum, and a high energy peak ~1 eV above the -valley peak, the expected energy position of the SV peak, thus demonstrating that a hot electron generating process must be occurring within the device. Unlike previous studies on high efficiency devices, 1,4,27 the presence of an SV peak cannot be an indicator of interband Auger recombination due to the lack of an appreciable efficiency droop. We believe that TAAR is a strong candidate for the efficiency limiting mechanism in these devices because of the efficiency behavior of these devices in conjunction with the direct measurements of hot carriers emitted from the p-type surface indicating some hot carrier generating mechanism is occurring.…”

“…A similar TAAR mechanism occurring in AlGaN heterostructure barriers has recently been observed. 27 The devices measured in this paper consist of LEDs grown by ammonia MBE. The epitaxial structure is as follows: 3 µm GaN on a sapphire template450 nm n-GaN [Si] = 2.510 18 cm -3 10 nm UID GaN barrier3 nm InGaN QW12 nm UID GaN barrier80 nm p-GaN [Mg] = 310 19 cm -3 a 5 nm p ++ contact layer.…”

Disorder effects in nitride semiconductors: impact on fundamental and device properties

“…However, so far, no full microscopic model exists yet to describe optical and transport phenomena as four challenges exist: (1) modeling requires accurate microscopic descriptions of alloy heterostructures, with some open questions on alloy randomness, interface abruptness and composition variation along the growth direction; (2) the high extended and point defect densities existing in nitride materials can add another level of complexity for the description of the physical system. For instance, the large difference between optoelectronic performance of molecular beam epitaxy (MBE)-grown materials compared with metal-organic chemical vapor deposition (MOCVD)-grown ones is still mysterious, although part of the explanation could be due to the presence of Ca impurities in MBE material acting as a killer impurity [2]; another such major effect is the curing of some nonradiative (NR) recombination centers by the growth of superlattices or underlayers before growing the active LED layers [3], recently attributed to the trapping by these structures of surface defects in GaN [4]; in MOCVD materials grown under optimal conditions, NR defects appear in selected layers such as AlGaN electron-blocking layers (EBLs) or in higher In content layers for green LEDs [5][6][7][8]; (3) for the electronic quantum description, a number of phenomena need to be further explored such as electronhole carrier localization and tunneling, Coulomb interactions, … Simulations of basic optical properties requires the computation of numerous energy levels, energy relaxation toward emitting levels, computation of the carrier population, … Simulations of LEDs require the additional computation of transport coefficients taking disorder and localization into account, both for perpendicular transport (I-V characteristics of LEDs, unipolar barrier transport) and in-plane transport. All these tasks require huge computational resources; (iv) finally, for comparisons with simulations, experiments need to determine accurate parameters, avoiding systematic errors.…”

Quantitative correlation of hot electron emission to Auger recombination in the active region of c-plane blue III-N LEDs

“…Peaks with a semiconductor origin have energies that increase with the applied voltage due to the voltage drop from p-contact to the sample surface. 7,9,11 The high energy thresholds (HETs) of the peaks can be extracted from the negative parts of the derivative of the EDCs, and equal to the extrapolated x-intercepts of the high energy slopes. 7,9,11 The extrapolated HETs at the expected turn-on voltage of 2.67 V should correspond to the bulk valley minimum as shown in Fig.…”

“…2(b). 7,9,11 Based on the extrapolated values, we assign the measured peaks in increasing energies in the following order: Au PE and Pd PE (both peaks are due to diode light),Γ and first side valley at ∼0.9 eV higher energy than Γ. 9-14 A low energy shoulder appears on Γ valley at high current density, which may be ascribed to thermalization or subbandgap PE through Franz-Keldysh absorption in the band-bending region.…”