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.
Existing semiconductor lasers operating at the 1300 and 1550 nm optical communication wavelengths typically incorporate InGaAs(P) or (Al)InGaAs quantum wells (QWs) in the active region of the device, and are grown on InP substrates. Despite the wide deployment of InP-based QW lasers in optical communication networks, these devices suffer fundamentally from low efficiency and require power-hungry external cooling. For example, Auger recombination accounts for approximately 80% of the current at threshold in a 1550 nm InGaAsP/InP QW laser, leading to strong temperature sensitivity and placing a low ceiling on the overall device efficiency. These issues are fundamentally related to the band structure of the materials employed in the active region of semiconductor lasers, and so any attempt to overcome them must begin by addressing the nature of the band structure in III-V semiconductor alloys.
Dilute bismide alloys, containing small fractions of substitutional bismuth (Bi) atoms, have emerged as a promising candidate material system with which to develop GaAs-based long-wavelength lasers while simultaneously addressing the problems associated with Auger recombination . Incorporation of a small fraction of Bi into a conventional III-V semiconductor such as (In)GaAs leads to a rapid decrease in the band gap and increase in the spin-orbit-splitting energy, both of which are characterised by strong, composition-dependent bowing. This unusual behaviour arises due to the fact that Bi, which is much larger and more electropositive than the As atoms it replaces, acts in dilute concentrations as an isovalent impurity in (In)GaAs. Theoretical analysis we have undertaken has confirmed that Bi incorporation strongly perturbs the valence band structure in (In)GaBiAs alloys, leading to a range of interesting material properties as well as the potential to exploit the impact of Bi to facilitate band structure engineering in photonic and photovoltaic devices.
For sufficiently high Bi compositions (> 10% in GaBiAs) it has been demonstrated both theoretically and experimentally that a band structure can be obtained in which the spin-orbit-splitting energy exceeds the band gap. It is well known that the dominant Auger recombination pathway in 1300 and 1550 nm QW lasers is the so-called “CHSH” process, whereby an electron-hole pair recombining across the band gap excites a hole into the spin-split-off band (as opposed to emitting a photon). In addition to reducing the populations of electrons and holes available to contribute to the lasing mode, the highly energetic holes produced by the CHSH Auger recombination process dissipate their excess energy in the crystal lattice, leading to the generation of waste heat.
If the band structure of the material forming the active region of the laser has a spin-orbit-splitting energy that exceeds the band gap, then the CHSH process is suppressed by conservation of energy: an electron-hole pair recombining across the band gap will not have sufficient energy to excite a hole to the spin-split-off band. The crossover to a band structure of this type occurs in GaBiAs alloys having a band gap close to 0.8 eV (1550 nm), so that GaBiAs offers the potential to achieve 1550 nm semiconductor lasers grown on GaAs substrates in which Auger recombination is suppressed.
This unique band structure, and its potential to bring about highly efficient and uncooled semiconductor laser operation at 1550 nm, has stimulated significant research interest in dilute bismide alloys. Despite challenges associated with the epitaxial growth of this metastable material system, significant progress has been made over the past number of years. The first electrically pumped dilute bismide QW lasers, based on GaBiAs QWs containing approximately 2% Bi, were demonstrated in 2013. The Bi composition in such devices has since been extended up to approximately 8%, as the community continues to push towards realising high quality GaBiAs QWs containing > 10% Bi. From a theoretical perspective, this progress in materials growth and device engineering has mandated the development of suitable models to describe the electronic structure of GaBiAs and related alloys, as well as the optical properties of dilute bismide heterostructures.
In this work – a collaboration between the University of Bristol (U.K.), Tyndall National Institute (Ireland), the University of Surrey (U.K.), and Philipps-Universität Marburg (Germany) – we (i) outline the theoretical models we have developed to describe the band structure of GaBiAs alloys, (ii) provide an overview of the impact of Bi incorporation on the theoretical performance of GaBiAs QW lasers, (iii) quantify the potential of this material system for the development of 1550 nm devices, and (iv) compare the results of our theoretical calculations directly to experimental measurements of the spontaneous emission and optical gain spectra for first-generation GaBiAs devices , .
These experimental measurements – undertaken at the University of Surrey on devices grown at Philipps-Universität Marburg – are the first of their kind for this emerging material system. The spontaneous emission and optical gain spectra computed using our theoretical model are in excellent agreement with experiment , verifying the understanding we have developed of the electronic and optical properties of GaBiAs alloys and heterostructures, and highlighting the potential of our theoretical framework for use in the design and optimisation of future dilute bismide semiconductor lasers and related devices.
Full details of our work on dilute bismide semiconductor lasers will be presented at the NUSOD 2016 conference in Sydney, Australia: talk ThB4 will focus on theory vs. experiment for GaAs-based dilute bismide QW lasers, while poster MP09 will describe theoretical calculations which highlight InGaBiAs alloys grown on InP substrates as a promising material system for the realisation of highly efficient, mid-infrared diode lasers operating at wavelengths > 3000 nm.
 C. A. Broderick, P. E. Harnedy, and E. P. O’Reilly, “Theory of the electronic and optical properties of dilute bismide quantum well lasers”, IEEE J. Sel. Topics Quantum Electron. 21, 1503313 (2015)
 I. P. Marko, C. A. Broderick, S. R. Jin, P. Ludewig, W. Stolz, K. Volz, J. M. Rorison, E. P. O’Reilly, and S. J. Sweeney, “Optical gain in GaAsBi/GaAs quantum well diode lasers”, Scientific Reports 6, 28863 (2016)