- 8,834 visits since January 2015
Optoelectronic Device NewsMy Tweets
Connecting Theory and Practice in Optoelectronics
Auger recombination inside the light-emitting InGaN quantum wells (QWs) was recently identified as major cause of output power limitations in GaN-based blue light-emitting diodes (LEDs) which are the core of many modern light sources. In this electron-hole recombination process, the released energy is transferred to another carrier (electron or hole) without light emission. The Auger recombination rate rises strongly with the QW carrier density and therefore intensifies with stronger current injection into the LED.
In contrast to LEDs, GaN-based blue laser diodes are expected to suffer less from Auger recombination, based on the popular opinion that the QW carrier density does not rise with increasing current injection above lasing threshold. Shuji Nakamura, who received the 2014 Nobel Prize in physics for his pioneering work on GaN-LEDs, stated in his Nobel lecture that “Auger recombination, with the resulting efficiency droop, does not appreciably occur in blue laser diodes”. We dispute this claim based on our numerical analysis of high-power InGaN/GaN laser measurements.
The energy efficiency is the fraction of the electrical input energy that is emitted as laser light. It is usually given as power conversion efficiency (PCE) and it is surprisingly low for GaN-based lasers. OSRAM just announced a record number of PCE=43% at SPIE Photonics West. This is certainly a remarkable achievement, considering the struggle to break the mysterious 40% limit. However, 43% is far below the record PCE of 84% reported for GaN LEDs. The inherently low hole conductivity and large series resistance on the p-doped side of GaN lasers are usually blamed for the efficiency deficit. However, the series resistance is known to shrink with rising temperature, which can be attributed to the increasing density of free holes in p-doped layers. Thus, one would expect that the PCE improves at elevated temperatures. But the OSRAM paper reported that the measured PCE drops with higher ambient temperature despite the shrinking series resistance. Ergo, there seems to be an even stronger loss mechanism involved.
One of the key rules of semiconductor laser physics relates to the carrier density inside the active layer. As long as I can remember, this rule states that the carrier density remains constant when the injection current rises above the lasing threshold. The reason lies in the stimulated emission of photons which consumes all additional carriers injected above threshold. The threshold carrier density delivers the threshold optical gain that compensates for the optical loss, which is usually not dependent on the injection current. Thus, the threshold carrier density should also remain constant. However, my recent analysis of high-power lasers yields different results (see picture). Read more of this post
There is an increasing interest toward the development of simple and compact comb laser sources. One promising application is the use of a InAs/GaAs Quantum Dot or InAs/InP Quantum Dash single section Fabry-Perot lasers. Many experiments on these devices have demonstrated the possibility of generating a wide optical spectrum of lasing longitudinal modes that are phase-locked. The phase locking is demonstrated by the very narrow RF line at the beat note frequency and by the possibility of getting pulses directly at the laser output or after group delay dispersion compensation with a proper length of dispersive optical fibre. There is however a lack of modelling work providing physical explanations on the capability of the QD lasers of generating phase locked lasing lines.
Two long sought-after goals for the semiconductor community have been (i) to develop long-wavelength semiconductor lasers on GaAs substrates, to enable exploitation of vertical-cavity architectures as well as monolithic integration with GaAs-based high-speed microelectronics, and (ii) to realise uncooled operation of semiconductor lasers, whereby the external cooling equipment typically required to maintain operational stability in long-wavelength devices can be removed in order to significantly reduce energy consumption without degrading the device performance.