Thermal characteristics of quantum-cascade lasers by micro-probe optical spectroscopy V. Spagnolo, G. Scamarcio, D. Marano, M. Troccoli, F. Capasso, C. Gmachl, A.M. Sergent, A.L. Hutchinson, D.L. Sivco, A.Y. Cho, H. Page, C. Becker and C. Sirtori Abstract: The facet temperature profile and the thermal resistance of operating quantum-cascade lasers (QCLs) have been assessed using a microprobe band-to-band photoluminescence technique. Substrate-side and epilayer-side-mounted QCLs based on GaInAs/AlInAs/InP and GaAs/AlGaAs material systems have been compared. The dependence of the thermal resistance on the CW or pulsed injection conditions and its correlation with the output power have been studied. These results were used as inputs for a two-dimensional heat-diffusion model which gives the heat fluxes and the thermal conductivity of the active regions, in order to design QCLs with improved thermal properties. 1 Introduction Quantum-cascade lasers (QCLs) are based on transitions between quantised states in GaInAs/AlInAs and GaAs/Al- GaAs multiple-quantum-well structures [1, 2]. Laser emis- sion has been reported in a wide range of mid-infrared wavelengths (3.5–24 mm) [1, 2]. Laser action in the terahertz range (67–80 mm) has been also demonstrated [3–5]. Excellent performance in terms of peak power (. 2 W) and maximum operating temperature (, 470 K) has been achieved in the pulsed mode [6–8]. Continuous wave (CW) operation up to 320 K has been demonstrated for a buried GaInAs/AlInAs/InP (InP-based) QCL [9]. The highest operating temperature in a CW for GaAs/AlGaAs (GaAs-based) QCLs is 135 K [10]. On the other hand, the lifetime for CW operation at room temperature is much lower than that of quantum-well diode lasers. This is due to the high electrical power needed to achieve the laser threshold. The cavity losses in the mid-IR ð10 ÿ 30 cm ÿ1 Þ lead to threshold currents in the kA=cm 2 range. The cascading scheme inherently requires applied threshold voltages of the order of 10 V. Moreover, the device thermal resistance is strongly increased by the ternary alloy nature of employed materials, the presence of a large number of interfaces and the associated phonon interference effects [11, 12]. These characteristics lead to the temperature of the active material being much greater than the heatsink. Thus, the attainment of CW operation in a wider range of wavelengths and the enhancement of the maximum operating temperature requires a deeper knowledge of the key physical phenomena controlling thermal dissipation in QCLs. In this paper, we present our recent results on the determination of the local lattice temperature and the thermal resistance in CW and pulsed operating QCLs, both in InP- and GaAs- based structures. We used a microprobe spectroscopy technique based on the analysis of the thermal induced shift of band-to-band photoluminescence (PL), similar to that already successfully used for conventional diode lasers [13]. We have compared substrate-side and epilayer-side mounted devices with identical epilayer structures. By using a heat dissipation model, we have estimated the heat flow configurations and the active region thermal conductivity. In conventional diode lasers, the facet temperature can be significantly higher than that in the device core because of nonradiative surface recombination processes. This effect is absent in QCLs, since they are unipolar devices and the emitted photon energies are well below those of recombination processes. Thus, we can use the facet temperature as a close estimate of the internal device temperature. 2 Investigated samples 2.1 InP-based QCLs The GaInAs/AlInAs structure consists of three-well QCLs designed for emission at 8 mm [14]. The active region is composed of a , 0.53-mm-thick stack of twelve active regions with interleaved injector regions sandwiched between two 0.5-mm-thick Ga 0:47 In 0:53 As waveguide core layers. The top cladding layer is formed by an inner 2.5 mm thick Al 0:48 In 0:52 As layer doped to n ¼ 1 ÿ 2  10 17 cm ÿ3 and an outer 0.5-mm-thick Ga 0:47 In 0:53 As layer heavily doped to n ¼ 5  10 18 cm ÿ3 for plasmon-enhanced confine- ment. The InP-substrate, doped to n ¼ 2  10 17 cm ÿ3 ; acts as a lower waveguide cladding layer. The devices were processed into 2.5-mm-long, 11-mm-wide, deep-etched q IEE, 2003 IEE Proceedings online no. 20030610 doi: 10.1049/ip-opt:20030610 V. Spagnolo, G. Scamarcio and D. Marano are with INFM, Dipartimento Interateneo di Fisica di Bari, Via Amendola 173, 70126 Bari, Italy M. Troccoli, F. Capasso, C. Gmachl, A.M. Sergent, A.L. Hutchinson, D.L. Sivco and A.Y. Cho are with Bell Laboratories, Lucent Technologies, 600 Mountain Avenue, Murray Hill, NJ 07974 H. Page, C. Becker and C. Sirtori are with Thomson-CSF, Laboratoire Central de Recherches, 91404 Orsay, France Paper first received 17th October 2002 and in revised form 2nd April 2003 IEE Proc.-Optoelectron., Vol. 150, No. 4, August 2003 298