Quantum Cascade Lasers (QCLs) [1] are important sources for coherent mid-wave infrared (MWIR) light that find widespread use in applications such as molecular spectroscopy, free-space optical communications, and defense countermeasures. These applications demand tunability of emission wavelength and high output powers. The conventional QCL is a two-terminal n-i-n device in which an externally applied bias establishes the necessary electric field in the cascade region for proper alignment of the quantum states, defines the emission wavelength, and also determines the current through the cascade structure. Because QCLs are 2-terminal devices, changing bias directly changes both wavefunction overlap and energy separation. The spectral location and magnitude of the gain peak thus has a strong dependence on applied voltage, impacting the capability to modulate. To overcome these limitations, the transistor-injected quantum-cascade laser (TI-QCL) has been proposed [2]. The TI-QCL is a three-terminal device in which the cascade region is placed within a heterojunction bipolar transistor. In forward-active mode the current injection is controlled by a forward-biased emitter-base junction, and the reverse bias on the base-collector junction is used to control the electric field for alignment of the quantum states. This novel structure enables the decoupling of emission wavelength and output power. This extra control positions the TI-QCL to impact a wide range of applications such as spectroscopy with its wide tunability, and free-space communication with its capability to separately modulate power and frequency.Previously, transistor characteristics and near-infrared (NIR) band-to-band base emission was demonstrated in a GaAs design [3, 4] before moving to an InP-based design [5]. We will review the progress made to push the TI-QCL concept towards full functionality, focusing on InP-based epitaxial designs. A self-consistent Schrodinger-Poisson solver and non-equilibrium Green’s function method is used to aid in the active-region design of the structure, which is optimized for a peak emission wavelength of 8.27 μm as shown in Figure 1 [6]. A wide wavelength tunability from approximately 7.8 μm to 8.8 μm is projected based on computational analysis. The InP-based structure is grown using molecular beam epitaxy by IntelliEpi; a cross-section of the material is depicted in Figure 2. It contains an active-region composed of 30 repeated periods of In0.41Ga0.59As/In0.36Al0.64As strain-balanced superlattice. The devices are fabricated using standard III-V processing. An image of a fabricated TI-QCL bar consisting of a sloped, 20 μm wide emitter mesa and evaporated SiO2 as the passivation layer is depicted in Figure 3. The fabricated devices are diced and cleaved into 2 to 4 mm x 500 μm die and mounted onto In-plated copper mounts followed by wire bonding. The devices are electrically characterized at room temperature, 77 K, and 18 K. Under cryogenic temperatures, the common-base characteristics exhibit negative differential resistance (NDR) in collector current at all emitter injection currents as shown in Figure 4, which signifies tunneling transport through the quantum cascade superlattice region. The base current also exhibits corresponding NDR behavior and switching in the band-to-band NIR light (1.57 m) output from the In0.53Ga0.47As base is experimentally observed when biased in a region where transport through the cascade region is rejected due to quantum state misalignment.
Read full abstract