We report low threshold InAs/AlSb quantum cascade lasers emitting near 15 µm. The devices are based on a vertical design similar to those employed previously in far infrared InAs-based QCLs, whereas the doping level of the active core is considerably decreased. The lasers exhibit a threshold current density as low as 730 A/cm2 in pulsed mode at room temperature and can operate in this regime up to 410K. The continuous wave regime of operation has been achieved in these devices at temperatures up to 20°C. The cw regime is demonstrated for InAs-based QCLs for the first time at room temperature.
© 2016 Optical Society of America
The development of InAs-based quantum cascade lasers was initially motivated by the giant conduction band discontinuity of 2.1 eV between InAs and AlSb. In comparison with other materials, this large band offset allows hosting high energy intersubband transitions, necessary to fabricate short wavelength QCLs emitting below 4 µm . The shortest QCL emission wavelength of 2.6 µm has been obtained in the InAs-based devices . Another advantage of the InAs/AlSb material system is the small electron effective mass, which can be helpful to obtain high QCL gain . The benefit from the small electron effective mass in InAs can be fully exploited in long wavelength devices where the lasing transition levels are close to the bottom of the conduction band and the effect of nonparabolicity is weak. Even the very first InAs/AlSb QCLs operating above 15 µm exhibited lower thresholds and higher maximum operation temperatures (Tmax) compared with InP- and GaAs-based devices. We reported near room temperature (RT) operation of InAs-based QCLs emitting at 19 µm with a threshold current density (Jth) of 0.6 kA/cm2 at 78 K , and operation up to a temperature of 333 K (60°C) at a wavelength of 18 μm with a Jth of 1.6 kA/cm2 at 78 K , whereas InP-based lasers operated only at cryogenic temperatures in the same spectral range  and demonstrated RT operation only up to 16 μm . Tmax of 220 K and 100 K have been achieved in lasers based on GaAs at wavelengths of 15 μm and 23 μm, respectively . InP-based QCLs operating in pulsed mode up to 240K at 24 μm with a Jth of 5.7 kA/cm2 at 50 K have been reported . Laser emission at 3.8 THz from an InAs/AlAs0.16Sb0.84 quantum cascade structure was recently demonstrated . These first InAs-based THz QCLs with a metal-metal waveguide operated at liquid helium temperature under magnetic field.
The use of a plasmon-enhanced dielectric waveguide with doped InAs cladding layers instead of the commonly employed for far infrared QCLs metal-metal waveguide resulted in further improvement in performances of the long wavelength InAs/AlSb lasers. The devices with a dielectric waveguide operated in pulsed mode at 20 µm up to 353K with a Jth of 1.4 kA/cm2 at 78 K and 4.3 kA/cm2 at RT . Despite the substantial improvement compared with the metal-metal waveguide lasers, these QCLs exhibited quite high propagation losses originated from the intersubband absorption in the laser active region and free carrier absorption in the doped waveguide. Both loss mechanisms are proportional to the carrier concentration, whereas free carrier absorption is also a more than quadratic function of the wavelength . Several groups reported a linear decrease in the QCL threshold current density with reducing doping level [13,14]. In this work we fabricated and tested InAs/AlSb QCLs with the doping level decreased and the emission wavelength shifted down to 15 µm.
2. Laser structure, fabrication and measurements
The device design is similar to that which we employed in . It was revised to shift the emission wavelength from 20 to 15 µm and the active zone also contained 55 stages. The basic parameters of the design are very similar: the lifetime at the upper level of the lasing transition (ul) is calculated to be τul = 0.24 ps for both structures, the lower level (ll) lifetime is 0.17 and 0.15 ps for the reference design and the new one, respectively. The oscillator strength of the transition f decreased from 80 to 66, which is due mainly to the increase in the transition energy. The next upper level in the active region is about 50 meV above the ul state and the width of the injector miniband ∆ is 100-105 meV for both structures under operating conditions. The employed vertical scheme providing a large f is generally used in InAs/AlSb QCLs to diminish the role of the interface scattering. The interface scattering is supposed to be an essential loss mechanism in this material system because of the giant band offset of 2.1 eV and the chemical nature of the interfaces without common atoms . The high oscillator strength itself does not reduce the interface scattering but makes its impact less significant because of the faster radiative transition. The doping level of 3-µm-thick InAs (Si) cladding layers of the dielectric waveguide was increased from 3x1017 cm−3 to 4x1017 cm−3 compared with the reference work in order to keep the same refractive index contrast at the shorter operating wavelength. The cladding layers were separated from the 5-µm-thick active region by 3-µm-thick undoped InAs spacers. The laser structure was grown by molecular beam epitaxy on a (100) n-InAs substrate in a Riber 412 solid source machine equipped with cracker cells for the group V elements. All InAs layers of the structure were grown at a rate of one monolayer per second (3 Å/s), whereas AlSb barriers were grown at 1 Å/S. X-ray diffraction measurements on the grown wafer showed the deviation of the QCL period length from the targeted value of 88.9 nm to be less than 1%. The intentional doping of the active region was decreased by 6 times compared with the reference QCL . The residual doping of InAs layers was estimated to be 5x1015cm−3, whereas it was 2 times higher in the lasers from  grown in another MBE machine. The total electron sheet density in the structure ns is considered to be 8x1010 cm−2 per period compared with 4x1011 cm−2 in .
The grown wafer was processed into deep mesa ridge lasers with a ridge width varied between 8 and 20 µm using standard contact photolithography and wet etching. Hard baked photoresist was used for electrical insulation. Electrical contacts to the devices were made using non-alloyed Ti/Au metallization. 3.6-mm-long lasers with cleaved resonators and uncoated facets were soldered epi-side down onto copper heatsinks using indium. The fabricated devices were mounted in a LN2 flow cryostat and tested at temperatures between 80 and 420K. In pulsed mode the lasers were driven with 100-ns-long current pulses at a repetition rate of 40 kHz. The emitted light was collected with a f/1 off-axis parabolic mirror and then analysed using a Fourier transform infrared spectrometer (FTIR) Bruker Vertex 70 equipped with a pyroelectric detector. For optical power measurements the laser beam was collimated with another f/1 off-axis parabolic mirror onto a Melles Griot 13PEM001 power meter. An additional 30 Hz current modulation was applied for pulsed measurements of light-current curves with the slow FTIR detector. The pulsed optical power of the lasers was calibrated using the power meter at a 5% duty cycle (100ns/500kHz). The presented data on optical power take into account the measured loss in the cryostat window (35%) and the collection efficiency of the used f/1 optics estimated to be 70% for both experimental configurations.
Tested QCLs exhibited significantly better performances compared with the reference devices . The lasers operated near 15.1 µm at RT with pulsed threshold current densities depending on the ridge width: 1.22, 0.89, 0.8 and 0.73 kA/cm2 for the 8-, 12-, 16-, and 20-µm-wide devices, respectively. The higher Jth in narrow devices is due to a larger overlap of the optical mode with absorbing dielectric on the mesa walls . The maximum available current density, corresponding to the misalignment of the injector and the upper level of the lasing transition at high electric fields, was close to 1.5 kA/cm2 for all of them. Voltage-current and light-current characteristics of a 20-µm-wide laser are shown in Fig. 1 for different temperatures.
The peak output power from one facet was measured to be 100 mW at 80K, 50 mW at RT and ≥ 5 mW at 400K. In pulse mode the maximum operation temperature Tmax of this device was 410K, 16-µm-wide lasers operated up to Tmax = 400K and for narrower devices the maximum operation temperature further decreased due to the higher thresholds and thus narrower current dynamic range.
Both the threshold current density and the slope efficiency of the lasers change slowly at low temperature but beyond 200K their evolution is nearly exponential with characteristic temperatures T0 = 165K and T1 = 260K for the threshold current and the slope efficiency, respectively. The characteristics are quite similar for 16- and 20-µm-wide QCLs. Figure 2 displays the threshold current density and the initial slope of the light-current curves of the devices as a function of temperature, as well as emission spectra of the 16-µm-wide device measured at RT and at 400K. The emission spectra of the lasers were centred near 14.7 and 15.1 µm at 80K and RT, respectively. Almost no shift of the spectra with temperature was observed between 280 and 400 K. The initial slope efficiency of the lasers is 135 mW/A per facet at 80K and 90 mW/A per facet at 300K (Fig. 2). The pulsed light-current curve from  measured at 300 K shows a slope of 3.5 mW/A per facet. To compare this value with the corresponding data from this work it is necessary to take into account the collection efficiency and multiply this value by the ratio of the QCL wavelengths. This comparison shows a large - of the order of the reduction in the threshold current density - improvement in the RT slope efficiency of the new lasers.
The fabricated lasers were tested in the continuous wave (cw) regime. The optical power of wide lasers with a ridge width of 16 and 20 µm reaches 60 mW/facet at 80K. The temperature behaviour of the devices operating in the cw regime depended on the quality of their soldering on the submounts. The best devices soldered under optimized pressure were able to operate continuously at temperatures up to 20°C. Voltage-current and light current characteristics of a 16-µm-wide laser measured in the cw regime at different temperatures are shown in Fig. 3. The cw maximum output power of this device was measured to be 5.5 mW at 0°C and 0.3 mW at 20°C, whereas the threshold current density increased from 0.9 to 1.25 kA/cm2 in the sametemperature range. The slope efficiency and the threshold current density of this laser measured in the cw regime are shown in Fig. 4 as a function of temperature. In the cw regime the lasers demonstrated single frequency emission, except at conditions of coexistence of several spatial modes, corresponding to the zones of kinks in the light-current curves. Such behaviour has already been noticed in InAs-based QCLs fabricated using a similar technology and was explained by stabilization of the main longitudinal mode due to the saturable absorption in the dielectric around the ridge edges .
The lasers studied in this work exhibited much better performances compared with QCLs emitting at 20 µm presented in the reference paper. Their threshold current density at RT has been reduced down to 730 A/cm2 from 4.3 kA/cm2 for the devices that we reported in . It should be pointed out that the cw regime of operation at RT is demonstrated for the first time for InAs-based QCLs. One of the reasons of this progress can be attributed to the reduction of optical losses both in the waveguide itself and in the laser active region due to the decreased doping and shorter operating wavelength and thus weaker free carrier absorption. Oursimulations using the Drude model with experimental material parameters  to calculate free carrier absorption show that the “empty” waveguide loss (αw) for these devices is lowered down to 3.1 cm−1, in spite of the higher doping of the cladding layers, in comparison with 5.1 cm−1 in . However, the decreased waveguide loss alone cannot explain the observed significant improvement in the threshold current density. For further analysis, we can use the same common formula for the threshold conditions including the transparency current Jth  as it was done in :
A rough estimation of Jleak can be made from analysis of the shape of voltage-current characteristics. The initial part of voltage-current curves before the steep voltage increase can be attributed to a leakage current that passes through the structure before the alignment of the injector and the upper level of the lasing transition. The leakage contribution is obvious in the V-I characteristics of the reference devices. In the lasers studied in this work It is visible in the pulsed V-I curve at 400K (Fig. 2) and, because of the better measurement accuracy, in the cw data in Fig. 3. A comparison shows that, in the temperature range 80-300K, Jleak is responsible for at least (0.3-0.5) of (1.3-4.3) kA/cm2 in the threshold current density in the reference QCLs, and less than 0.1 kA/cm2 in this work. This improvement is a straight consequence of the decreased doping if Jleak is due to nonresonant electron injection from the last injector states directly into the lower level of the lasing transition via elastic and inelastic tunnelling processes . It should be noted that this leakage channel bypassing the ul state is more probable in the employed design compared with diagonal schemes including a thin quantum well separating the transition region and injector . Reduction of such leakage is however also not sufficient to produce the observed more than 5-fold improvement in Jth.
The resonant optical absorption that is the main contribution to αAR is usually attributed to the thermal population of the lower transition level nth = ns exp(-∆/kT), where ns is the sheet doping density of the injector, ∆ is the width of the injector miniband, k is the Boltzmann constant and T is electronic temperature . Another temperature effect leading to increase in the population of the ll level is the thermal excitation of electrons into the states above the upper level of the transition followed by their scattering into the lower level . These effects on Jth can be quite strong or even dominating at RT and beyond in non-optimized QCL structures but they should be negligible at 80K . However, If the thermal effects are crucial at RT in the reference devices, their Jth-Jleak values at 80K (~1 kA/cm2) should be comparable with those of the QCLs from this work (<0.3 kA/cm2), which is not the case. Obviously, there is another mechanism supplying extra electrons to the lower transition level, which is less temperature sensitive than the thermal excitation and governed by the injector doping. We believe that this can be the tunnelling/scattering from the injector states delivering the discussed above leakage current. This current path will increase population of the ll level, which in turn is responsible for the additional resonant absorption in the laser active region. Operation of InAs-based QCLs with the discussed vertical design is very sensitive to the electron concentration in the ll state because of the short τul lifetime and for this reason it is strongly affected by the two-fold effect of the leakage through the injector barrier. In a weakly doped active region an efficient injection into the upper transition level and hence weaker resonant loss are achieved at low currents, which results in the observed improvement of the threshold and power characteristics of the new lasers compared with the reference devices.
It is interesting to compare our lasers with the best results published to date on QCLs operating at similar wavelengths. The most recent InP-based QCLs emitting at 15 µm  and 14 µm  exhibited RT threshold current densities of 3.5 and 2 kA/cm2, respectively, for devices with high reflection (HR) coated facets. The corresponding slope efficiencies are 346 mW/A for 55 cascades in  and 375 mW/A for 70 cascades in , all for HR-coated devices. These devices employ different injection schemes but both are based on a diagonal design with longer τul. As discussed above, the leakage current due to the tunneling/scattering from the injector states should be much smaller in such structures and the influence of excessive population of the ll level on the device performance is in general much weaker than in the case of vertical transitions. However, despite the short τul the very large oscillator strength of the lasing transition, due to the small InAs effective mass and a large wavefunction overlap, results in the lower threshold current density in our devices. It should be noted however that the slope efficiency of our lasers is about 2 times worse in comparison with the cited InP-based devices. This difference can be partially explained by a larger electron leakage from the ul level to the next state above it due to the smaller distance between them, which amounts to 50 meV in our lasers and 65 meV in the devices from . Further design optimization of the InAs/AlSb QCLs is required to increase the output power.
An important factor that contributes to the observed performance improvement is a higher quality of the QCL material grown in the RIBER 412 MBE machine. This new equipment with big effusion cells provides better flux stability, of the order of 1%, compared with the smaller RIBER Compact 21 machine, used previously to grow the reference wafer , where the InAs growth rate tended to decrease during the many hours long growth, resulting in mistuning of the QCL cascades and thus in a smaller gain. The lasers studied in this work could also benefit from the lower residual doping of InAs in the new machine. The cleaner environment results in decreased free carrier absorption in the laser waveguide.
Continuous wave performances of semiconductor lasers are strongly influenced by the efficiency of heat dissipation, characterized by their thermal resistance. The thermal resistance (Rth) of the tested lasers can be estimated from the rollover of the optical power in the cw regime (Fig. 3). The light-current curve measured at 20°C intersects the axis of current at 0.81-0.82 A (V = 12.4 V), where the device stops to lase due to overheating of its core. We can consider that at this point the average core temperature of the laser reaches its maximum operating temperature measured in pulsed mode, i.e. 400K, which gives Rth = 10.5 K/W for this device. The thermal resistance can also be found from a comparison of the temperature behaviour of the threshold current measured in pulsed and cw modes (Figs. 2,3). This method gives Rth values of 8 K/W at low temperatures and 10 K/W near RT. The normalized thermal conductance Gth, defined as Gth = 1/RthA, where A is the area of the laser ridge, obtained from these data is 170 W/Kcm2. A thermal conductance of 340 W/Kcm2 has been reported for InP-based ridge QCLs . This better Gth is due mainly to the higher thermal conductivity of 68 W/mK of InP cladding layers in these lasers compared with 27 W/mK for InAs. Heat dissipation in our devices can be improved by using thick metallization of the ridge walls insulated with a thin layer of a suitable dielectric material.
In conclusion, we have demonstrated InAs-based quantum cascade lasers with the lowest threshold reported to date for this material system. The devices are based on a vertical design similar to those employed previously in far infrared InAs-based QCLs. Due to the short upper level lifetime the performance of QCLs based on a vertical design is very sensitive to the population of the bottom level, which in turn is enhanced in this laser scheme by the electron tunnelling/scattering from the doped injector. Very low injector doping of the lasers studied in this work resulted in significant reducing the threshold current density compared with the previous results. The lasers exhibit a threshold current density as low as 730 A/cm2 in pulsed mode at RT, which allowed us, for the first time for InAs/AlSb QCLs, to achieve the cw regime of operation up to 20°C in these devices emitting at 15 µm. To our knowledge, the longest emission wavelength of RT cw operation for QCLs fabricated from other materials is 12.4 µm. These InP-based single mode lasers from the catalog of Alpes Lasers SA deliver an optical power of 4 mW at 10°C . The results obtained in this work can serve for the performance improvement of InAs/AlSb QCLs operating in other spectral regions.
“Investment for the Future” program (EquipEx) (EXTRA, ANR-11-EQPX-0016); Agence nationale de la recherche (ANR) (Delta ANR-11-NANO-020); Fonds européen de développement économique régional (FEDER) (47851, 49793).
References and links
1. A. N. Baranov and R. Teissier, “Quantum cascade lasers in the InAs/AlSb material system,” IEEE J. Sel. Top. Quantum Electron. 21(6), 1200612 (2015). [CrossRef]
2. O. Cathabard, R. Teissier, J. Devenson, J. Moreno, and A. N. Baranov, “Quantum cascade lasers emitting near 2.6,” Appl. Phys. Lett. 96(14), 141110 (2010). [CrossRef]
3. E. Benveniste, A. Vasanelli, A. Delteil, J. Devenson, R. Teissier, A. Baranov, A. M. Andrews, G. Strasser, I. Sagnes, and C. Sirtori, “Influence of the material parameters on quantum cascade devices,” Appl. Phys. Lett. 93(13), 131108 (2008). [CrossRef]
4. D. Chastanet, G. Lollia, A. Bousseksou, M. Bahriz, P. Laffaille, A. N. Baranov, F. Julien, R. Colombelli, and R. Teissier, “Long-infrared InAs-based quantum cascade lasers operating at 291 K (λ= 19 μm) with metal-metal resonators,” Appl. Phys. Lett. 104(2), 021106 (2014). [CrossRef]
5. D. Chastanet, A. Bousseksou, G. Lollia, M. Bahriz, F. H. Julien, A. N. Baranov, R. Teissier, and R. Colombelli, “High temperature, single mode, long infrared (λ = 17.8 μm) InAs-based quantum cascade lasers,” Appl. Phys. Lett. 105(11), 111118 (2014). [CrossRef]
6. K. Unterrainer, R. Colombelli, C. Gmachl, F. Capasso, H. Y. Hwang, A. M. Sergent, D. L. Sivco, and A. Y. Cho, “Quantum cascade lasers with double metal-semiconductor waveguide resonators,” Appl. Phys. Lett. 80(17), 3060–3062 (2002). [CrossRef]
7. M. Rochat, D. Hofstetter, M. Beck, and J. Faist, “Long-wavelength (λ≈16 μm), room-temperature, single-frequency quantum-cascade lasers based on a bound-to-continuum transition,” Appl. Phys. Lett. 79(26), 4271–4273 (2001). [CrossRef]
8. J. Ulrich, J. Kreuter, W. Schrenk, G. Strasser, and K. Unterrainer, “Long wavelength (15 and 23 μm) GaAs/AlGaAs quantum cascade lasers,” Appl. Phys. Lett. 80(20), 3691–3693 (2002). [CrossRef]
9. K. Ohtani, M. Beck, and J. Faist, “Double metal waveguide InGaAs/AlInAs quantum cascade lasers emitting at 24 μm,” Appl. Phys. Lett. 105(12), 121115 (2014). [CrossRef]
10. M. Brandstetter, M. A. Kainz, T. Zederbauer, M. Krall, S. Schönhuber, H. Detz, W. Schrenk, A. M. Andrews, G. Strasser, and K. Unterrainer, “InAs based terahertz quantum cascade lasers,” Appl. Phys. Lett. 108(1), 011109 (2016). [CrossRef]
11. M. Bahriz, G. Lollia, A. N. Baranov, and R. Teissier, “High temperature operation of far infrared (λ ≈20 µm) InAs/AlSb quantum cascade lasers with dielectric waveguide,” Opt. Express 23(2), 1523–1528 (2015). [CrossRef] [PubMed]
12. R. M. Culpepper and J. R. Dixon, “Free-carrier absorption in n-type indium arsenide,” J. Opt. Soc. Am. 58(1), 96–102 (1968). [CrossRef]
13. S. Lee, M. Giehler, R. Hey, T. Ohtsuka, A. A. Wacker, and H. Grahn, “Dependence of lasing properties of GaAs/AlxGa1−xAs quantum cascade lasers on injector doping density: theory and experiment,” Semicond. Sci. Technol. 19(4), S45–S47 (2004). [CrossRef]
14. T. Aellen, M. Beck, N. Hoyler, M. Giovannini, J. Faist, and E. Gini, “Doping in quantum cascade lasers. I. InAlAs-InGaAs/InP midinfrared devices,” J. Appl. Phys. 100(4), 043101 (2006). [CrossRef]
15. P. Laffaille, J. C. Moreno, R. Teissier, M. Bahriz, and A. N. Baranov, “High temperature operation of short wavelength InAs-based quantum cascade lasers,” AIP Adv. 2(2), 022119 (2012). [CrossRef]
16. Y. B. Li, R. A. Stradling, T. Knight, J. R. Birch, R. H. Thomas, C. C. Phillips, and I. T. Ferguson, “Infrared reflection and transmission of undoped and Si-doped InAs grown on GaAs by molecular beam epitaxy,” Semicond. Sci. Technol. 8(1), 101–111 (1993). [CrossRef]
17. S. Slivken, A. Evans, W. Zhang, and M. Razeghi, “High-power, continuous-operation intersubband laser for wavelengths greater than 10 μm,” Appl. Phys. Lett. 90(15), 151115 (2007). [CrossRef]
18. C. Sirtori, F. Capasso, J. Faist, A. L. Hutchinson, D. L. Sivco, and A. Y. Cho, “Resonant tunneling in quantum cascade lasers,” IEEE J. Quant. Electron. 34, 1722 (1998). A. Wittmann, A. Hugi, E. Gini, N. Hoyler, and J. Faist, “Heterogeneous high-performance quantum-cascade laser sources for broad-band tuning,” IEEE J. Quantum Electron. 44, 1093 (2008).
19. A. Wittmann, A. Hugi, E. Gini, N. Hoyler, and J. Faist, “Heterogeneous high-performance quantum-cascade laser sources for broad-band tuning,” IEEE J. Quantum Electron. 44, 1093 (2008).
20. D. Botez, C.-C. Chang, and L. J. Mawst, “Temperature sensitivity of the electro-optical characteristics for mid-infrared ( λ = 3–16 μ m)-emitting quantum cascade lasers,” J. Phys. D Appl. Phys. 49(4), 043001 (2016). [CrossRef]
21. K. Fujita, M. Yamanishi, T. Edamura, A. Sugiyama, and S. Furuta, “Extremely high T0-values (∼15µm), low-threshold-current-density quantum-cascade lasers based on the indirect pump scheme,” Appl. Phys. Lett. 97, 201109 (2010). [CrossRef]
23. A. Evans, J. Nguyen, S. Slivken, J. S. Yu, S. R. Darvish, and M. Razeghi, “Quantum-cascade lasers operating in continuous-wave mode above 90 °C at λ∼5.25 μm,” Appl. Phys. Lett. 88(5), 051105 (2006). [CrossRef]
24. Alpes Lasers, “List of CW-RT-DFB-QC lasers as of 2016-05-08,” http://www.alpeslasers.ch/lasers-on-stock/sbcw6519.pdf