Quantum cascade laser (QCL) is an important source of electromagnetic radiation in mid infrared region. Recent research in mid-IR QCLs has resulted in record high wallplug efficiency (WPE), high continuous wave (CW) output power, single mode operation and wide tunability. CW output power of 5.1 W with 21% WPE has been achieved at room temperature (RT). A record high WPE of 53% at 40K has been demonstrated. Operation wavelength of QCL in CW at RT has been extended to as short as 3µm. Very high peak power of 190 W has been obtained from a broad area QCL of ridge width 400µm. 2.4W RT, CW power output has been achieved from a distributed feedback (DFB) QCL. Wide tuning based on dual section sample grating DFB QCLs has resulted in individual tuning of 50cm−1 and 24 dB side mode suppression ratio with continuous wave power greater than 100mW.
© 2013 OSA
The quantum cascade laser  is a semiconductor laser where electromagnetic radiation is achieved by intersubband transitions between energy levels inside superlattice quantum wells. In comparison to conventional interband lasers in near infrared, the energy of intersubband transitions can be tuned over a very wide wavelength range from 3 to 15µm by tuning the compositions and thickness of wells and barriers. The intersubband lasers are free from Auger recombination which limited high temperature operation of conventional lasers in mid-infrared region. Lastly, multiple stages can be cascaded in series to achieve a large output power for various applications. Due to all these attractive properties, QCL is a compact semiconductor source of choice in the wavelength range of 3-5µm, corresponding to the first atmospheric window, inside mid-infrared spectrum. Specifically, a lot of research has taken place around 4.5-5µm which is technologically important for infrared countermeasure and spectroscopic application, and it has resulted in record room temperature (RT) continuous wave (CW) output power  and wall plug efficiency (WPE) . The wavelength 3-4µm, also known as the spectroscopic finger print region of many important hydrocarbons, also attracted the attention of the QCL community. The wavelength of CW operation at RT has been extended to 3µm , which is extremely attractive for spectroscopic detection of large number of molecules.
Highest Power CW wave, RT operation of QCL
One of the most impressive features of QCL is that, it is free from Auger recombination and has high characteristic temperature for threshold current which improves room temperature operation. In the relative wavelength range of 3-5 μm, lattice matched AlInAs/GaInAs on InP, which has a conduction band offset of ~0.52eV between barrier and well, is insufficient for confinement of upper electron state. The confinement can be improved significantly by growth of strained superlattice core region, which has higher band offset. Much of the success of mid-infrared QCL can be attributed to growth of high quality, dislocation free, strain-balanced superlattice, where in spite of strain in individual layers, the net strain per period is close to zero. Further improvements can be done through the principles of band structure engineering by reducing electron escape from higher laser level by larger escape potential of electrons and reducing scattering from the upper laser level. Different active region design has been explored in the course of high power QCL. Specifically, a QCL core having 5 different material compositions were was designed for high power, CW wave operation . As can be seen in Fig. 1(a), AlAs inserts were included as a part of barrier in the injector region to reduce electron escape to the continuum by raising the escape activation energy.The active region has 3 wells and can be nominally called a single photon resonance (SPR). However in contrast to conventional deep well design, this active region consists of a shallow quantum well followed by a short barrier, giving rise to a smaller conduction band offset on both interfaces of this barrier. Interface scattering rate is proportional to the square of the overlap weighted by the band offsets. The upper laser level has a large overlap with the interface, due to diagonal nature of transition, so an interface with a lower conduction band offsets will increase the upper laser state lifetime. The insertion of near lattice matched (NLM) material as the first well and barrier in the active region also resulted in the increase in the energy difference between upper laser level and the next high energy level from 80 meV in a conventional 3 well active region to 90 meV here. This suppresses carrier leakage through longitudinal optical phonon. The voltage defect i.e. the energy difference between the lowest laser level and ground state of the injector is ~180 meV, which reduces thermal backfilling of lower laser state. A 40 period QCL core instead of standard 30 period is used to increase the mode confinement factor and increase the slope efficiency and output power.
The QCL wafer was processed into double channel geometry of width 19µm and 8 µm for pulsed and CW operation, respectively. A laser with a cavity length of 3mm was tested under pulsed mode operation with a pulse width of 500ns and a duty cycle of 5%. It yielded a maximum pulsed WPE of 27%. Buried ridge processing was performed on 8µm wide ridge for CW operation and cavity length of 5mm was cleaved. A maximum output power of 5.1W and WPE of 21% is obtained in CW mode at RT. The power current voltage curve and WPE curve for a QCL under CW operation at RT are shown in Fig. 1 (b).
Highest WPE QCL
The WPE is the most important figure of merit, which determines the overall quality of the QCL. A QCL with higher WPE will convert more input electrical power into useful electromagnetic radiation. This will lead to lesser demand on thermal packaging, and ensure higher temperature operation. It was a challenge to realize a QCL which has a WPE higher than 50% at any temperature i.e. a QCL which produced more light than heat . In conventional QCLs, the injector acts as a reservoir of electrons from which electrons are selectively injected with the correct energy into the upper laser level of the next stage. Generally the injector is a chirped superlattice so that there is a large energy difference between the ground state of the injector and the lower laser level, so that at RT, there is a large thermal backfilling from the ground state into the lower laser level. This thermal backfilling causes increase in laser threshold, due to decrease in population inversion. However at sufficiently low temperatures the backfilling is negligible. Instead, a large voltage defect may occur, which reduces the voltage efficiency i.e. the percentage of applied voltage that is dropped by actual laser transition from upper to lower level. The voltage defect can be reduced by using a single well as an injector instead of multi well injectors. As the differential gain of this wafer is high, a short cavity length does not show large increase in threshold current density, but greatly enhances the slope efficiency; thereby a large increase in WPE is obtained. The maximum WPE is obtained for a cavity length of 2mm. Below 80K, the thermal backfilling is negligible and the threshold current increases with the rise in temperature. However, the threshold voltage decreases with the rise in temperature as it become easier for electrons from ground state of the injector to populate the upper laser level in the next stage. A minimum of voltage defect is obtained at 150K. Due to these counteracting effects, the WPE has a highest value of 53% at 40K.
Figure 2 (a) shows the conduction band edge and wave function diagram. Figure 2 (b) shows the WPE vs current characteristic of a λ~5µm QCL having ridge width of 6µm and cavity length of 2 mm. Inset of (a) shows the graph of maximum WPE vs heat sink temperature, which clearly demonstrates rapid backfilling of lower laser level with temperature.
Watt level CW operation of 3.76 um QCL at RT
One of the ways to implement 3-4um QCL is to apply the knowledge of QCLs obtained from the higher wavelength side of 4-5 µm. Similar to its longer wavelength counterpart, 3 well active region or single phonon resonance was used to design the active region since in comparison to double phonon resonance or bound to continuum designs, the upper laser level is lower down in terms of absolute energy in quantum well system. This creates better confinement of electrons and increases the activation energy of electron escape through optical phonon emission and thermal emission to the continuum. The strain of the superlattice was increased and Ga0.24In0.76As/ Al0.76In0.24As superlattice was grown, to accommodate a large optical transition at shorter wavelength .
The QCL structure was grown in a single growth run on n- doped (Si, ~2E17cm−3) InP substrate, in a gas source molecular beam epitaxy reactor. The layer sequence and average doping levels are as follows: 1µm buffer layer acting as the lower cladding (Si, ~2E16cm−3), 30 period laser core (Si, ~2.8E16cm−3), 3µm upper cladding (Si, ~2E16cm−3), and 1µm cap layer (Si, ~1E19cm−3). The experimental and simulated (X’Pert Epitaxy) x-ray diffraction curves for the laser core are shown in Fig. 3. Excellent agreement has been found between the two curves, confirming the material compositions. The increase in strain in the superlattice was accompanied by sharp interface as was determined by low background and sharp higher order superlattice peaks in x-ray, with the smallest full width at half maximum (FWHM) of the satellite peak being 21.2 arc sec.
The wafer was processed into buried ridge geometry with ridge width of 8.3µm and a device with cavity length of 4mm was cleaved. The maximum RT, CW output power was 1.1 W, with a threshold current density of 1.67KA/cm2 and slope efficiency near threshold of 2.16 W/A. A maximum RT, WPE of 6% and 10% was obtained in CW and pulsed operation, respectively. Figure 4 represents the P-I-V curve and corresponding WPE for of a HR coated 8.3µm wide and 4mm QCL, at RT.
Greater than 400 mW of RT, CW output power at 3.4-3.55 µm QCL
Many important hydrocarbons have their fundamental modes of vibration in the 3-3.6 um electromagnetic spectrum range. As a result, an efficient source of radiation within this wavelength range is essential in spectroscopic detection of trace hydrocarbons. The 3-3.6um QCL has to accommodate a large energy transition within the quantum well. At present there are three competing technologies at short wavelength, namely InAs/AlSb on InAs, GaInAs/AlAs(Sb) on InP, GaInAs/AlInAs on InP. The InAs based material with high conduction band offset of ~2.1eV and large separation of and L conduction bands are attractive for short wavelength QCLs. However, large broadening due to interface scattering, and incompatibility of InAs material growth and waveguide technology with the existing InP technology are negative points. The GaInAs/AlAs(Sb) on InP is attractive due to its large conduction band offset of 1.6eV and compatibility with InP technology. But Sb based materials are not standard component in an MBE, and heat extraction from active region becomes difficult due to poor conductivity of Sb compound. The GaInAs/AlInAs on InP has a conduction band offset of only 0.52 eV in lattice matched condition. This value can be increased to as high as 1.2 eV by incorporating strain in the active region so that the net strain per period is close to zero. Two QCLs with cores of Ga0.2In0.8As/ Al0.82In0.18As superlattice were grown with slight difference in their layer thicknesses . Intervalley scattering is a major challenge in this wavelength as activation energy of electron escaping from the upper level to the lowest L valley state in the quantum well is reduced. Degradation of high temperature operation is a direct effect of this phenomenon. At present, this effect is minimized by increasing the Indium concentration inside GaInAs, which pushes up the lower L valley state. Thus, highly strained GaInAs reduces the intervalley carrier escape. No dislocation is found in the core region even at this high strain. To verify the material quality of the superlattice at this large strain, λ~3.56µm electroluminescence (EL) sample was prepared and tested with 200ns pulses at 10 kHz frequency. The linewidth is found to be 47.4 meV, showing a good material quality. Figure 5 shows the pulsed EL spectrum of the QCL at RT.
The two wafers were processed into double-channel laser with widths of 10.5µm and 8.6 µm for λ~3.56µm and λ~3.39µm QCL, respectively. Maximum RT, CW powers of 437 mW and 403mW were obtained at λ~3.56µm and λ~3.39µm respectively. Figure 6 shows P-I-V curve for CW operation at different heat sink temperatures. The inset shows the respective emission spectrum.
CW, RT operation of 3-3.2 µm QCL
For 3-3.2µm QCL, even higher conduction band offset is required for electron confinement. This is achieved by increasing Al composition in the AlInAs barrier to 85% and 89% for emission at 3.2 and 3 µm, respectively . The Ga composition inside GaInAs well was 23% and 21% respectively. However, extremely high tensile strain level in AlInAs can cause relaxation in the core region, if the global strain is not zero or the thickness of individual layer is larger than the critical thickness. In order to neutralize the net strain per period, Al0.6In0.4As/Al0.89In0.11As composite barriers was used as in the injector region instead of incorporating more Indium inside GaInAs, as pseudomorphic growth of extremely high strained GaInAs is very difficult. In this approach the coupling of the wavefunction in the injectors was ensured by choosing proper thickness of the composite barrier, while the addition of tensile strain was minimized by tuning the composition of lesser strained AlInAs composition. Conduction band edge and wave function diagram for a 3µm QCL is shown in Fig. 7.
A 3µm broad area QCL having ridge width and cavity length of 200 µm and 5 mm, respectively, shows a maximum pulsed output power of 10 W at RT (Fig. 8(a)). For RT, CW operation, a combination of buried ridge processing, narrow ridge width, epi-down bonding on diamond submount is done. As short wavelength QCLs have a small differential gain, to reduce the threshold current, the front facet is partially high reflection (PHR) coated and back side is high reflection (HR) coated. The PHR has a reflectivity of 85%, hence reduces the mirror loss from 1.4cm−1 for a HR coated back and uncoated front facet, to only 0.21cm−1 for HR coated back and PHR coated front facet. A 3.2 µm QCL having a ridge width and cavity length of 7µm and 5 mm, respectively, shows a maximum CW output power of 20mW at RT (Fig. 8(b)). The emission spectrum at a driving current of 0.5A is shown, in inset, to have a wavelength of 3.23 µm. A 3µm QCL having a ridge width and cavity length of 3µm and 5 mm, respectively, shows a maximum CW output power of 2.8 mW at RT (Fig. 8(c)). The emission spectrum at a driving current of 0.4A is shown, in inset, to have a wavelength of 3.02 µm. The minimization of lasing volume, efficient heat removal, and reduction of threshold current density by partial reflective coating on the front facet to decrease the mirror loss enable CW operation at RT. The 3.02 µm emission is the shortest wavelength demonstration of CW operation at RT, for any QCL.
The composite barriers have been generally used in short wavelength QCLs. The pros of using them in longer wavelength QCLs are larger conduction band offset in the active region and decrease in thermally activated carrier escape from the upper laser level into the continuum. The cons are that the higher strained material is more difficult to grow than milder strained ones and may lead to rougher interfaces. This may lead to broadening of the transition, due to higher interface scattering, and a decrease in the differential gain.
Broad Area QCL with 190 W of pulsed output power at RT
High power, CW operation at RT, is possible only with narrow ridge widths. However, in some applications where very high peak power is required, the QCL peak power can be scaled up by increasing the ridge width, thus increasing the laser gain volume. This trend can be observed for scaling of power from 50µm to 400µm . Both the threshold current and roll over current, increases linearly with ridge width. However, WPE decreases with ridge width from 16% at a width of 50µm to 12% at a width of 200µm and then saturates. This behavior may be attributed to heating which increases with ridge width and then saturates when the ridge width approaches the thickness of the substrate. P-I-V curve for a BA QCL with a ridge width of 400µm wide and cavity length of 3 mm is shown in Fig. 9 (a).A high peak power of 116 W is obtained. The lasing far field consists of distinct dual lobes at angles around 38°C which suggest that near field is composed of, periodic local maxima. This also indicates that the near field is stable and there is no random formation of filament, which creates many narrow, bright spots at the near field. This is of course expected, as the linewidth enhancement factor which is one of the main reasons for filament formation, is much less than conventional interband lasers. To test if even more peak power can be achieved, another BAQCL with a ridge width of 800µm and cavity length of 3mm was processed from a separate λ~4.5µm QCL wafer and tested with 200 ns pulses of 0.02% duty cycle at RT. Power was measured by placing the thermopile directly in front of the laser facet. The peak power was 190W (Fig. 9 (b)). The kinks at the high current were due to change in the lateral modes.
2.4W RT, CW power output of distributed feedback grating QCL
Symmetric, single mode with large mode suppression ratio along with single lobed far field distribution is desirable in many industrial applications. To achieve these objectives distributed feedback grating (DFB) is incorporated in QCL waveguide . To get a single lobed far field, the fundamental TM00 transverse mode should have lower loss than higher order modes, so that it is easier for it to reach the threshold first. To achieve a single frequency operation, the DFB coupling needs to be strong so that only a single longitudinal laser cavity mode is selected. Lastly, the grating cavity feedback needs to be sufficiently stronger than the mirror feedback, so that the lasing spectrum is determined primarily by the DFB section and it can ensure high output power. For a grating duty cycle of σ = 0.7, in the grating depth range between 100 and 160nm, the interaction between the first two DFB modes and the first order surface plasmon mode, leads to mode discrimination between the two DFB modes. A grating depth of 120 nm will generate a coupling coefficient (κ) of 1.37cm−1 and modal loss discrimination of 0.4 cm−1. The small value of κ ensures high output coupling efficiency. The comparatively small depth of grating introduces only 0.9% increase in waveguide loss. The gratings are etched on InP waveguide cap layer by plasma etching. The back facet was HR coated, while the front facet was Anti reflection (AR) coated. The AR coating increases the mirror loss to 3.39 cm−1 from 1.36 cm−1 for a 5mm cavity device. Although the threshold current increases, slope efficiency and output coupling efficiency increase. Thus output power and WPE also increases. Figure 10 shows P-I-V curve at RT for 11µm wide and 5mm long DFB QCL. 2.4W of RT, CW power at 10% WPE is obtained. Right inset shows single lobed far field operation as function of current between 1.1 and 1.7A. Left inset shows the tuning of lasing spectrum from 2087 cm−1 to 2084 cm−1 as a function of current between 1 and 1.7A. Single mode emission with side mode suppression ration (SMSR) of 30 dB in the entire current range is observed.
Widely Tunable, dual section, sampled grating DFB (SGDFB) QCL
In many applications such as spectroscopy, along with high power, high side mode suppression ratio (SMSR), a broad tunability is also necessary. However, conventional DFB QCL has a tuning range of only ~5cm−1, which limits detection of molecules around a small wavelength range. Though external cavity QCL is capable of a wide tunability, requirement of precision optics, susceptibility of this system to mechanical vibrations and slowness during wide tuning due to large grating mass are some of its limitations. Dual section laser with SGDFB, with a proven performance in telecom wavelengths, was alternatively explored for wide tuning . In this approach, sampled gratings with two different periods Z1 and Z2 are etched on the two different sections on the waveguides, with two different contacts. The current injected into the front and back section If and Ib are independent of each other. The reflectivity spectrum becomes comb like, with multiple supermodes, whose spacing is inversely related to the sampling period. As the two different sampling periods Z1 and Z2 are combined within a single waveguide, they produce two reflectivity frequency combs. However, in static condition only one reflectivity peak from each comb strongly overlap at one wavelength. As the current in one of the section is increased, the refractive index in the section increases and the supermode shifts by a fixed amount. As a result, the frequency comb overlaps in another wavelength giving rise to Vernier type step tuning. Once a supermode is chosen the tuning inside one supermode is done by increasing and decreasing current in both sections, which shifts peak wavelength uniformly for both supermodes. The comb spacing for each section is designed to be less than ~5cm−1, the spectral tuning of a single DFB QCL, so that a continuous tuning can be achieved over a wide range.
A SGDFB QCL with front and rear section of ~1.6mm and ~1.4mm is fabricated. Both sections use 30 period grating which is sampled 3 times, where the grating period is 753nm.The wafer is processed into double channel ridge waveguide with ridge width of 10µm and cleaved into a device cavity length of 3mm. The tuning characteristic, output power and SMSR is shown in Fig. 11. Single mode tuning of more than 50cm−1 is obtained. More than 100 mW of CW power, with a mean SMSR of 24 dB is obtained at RT over the tuning range.
In summary, InP based quantum cascade laser technology has demonstrated its versatility as a source in the mid-infrared. However, many challenges remain which limit the performance of QCLs. For example, non-idealities such as intervalley scattering, carrier escape to the continuum, heat removal from the core region, and interface scattering still effects the RT, performance of QCL, especially at short wavelengths. Some of these can be lessoned or removed by better gain medium and waveguide design strategies, while the quality of strained material growth can be improved to reduce linewidth of transition. QCLs with spatially and spectrally pure single mode, and wide tuning, are also being explored for practical applications, by borrowing some successful ideas from the near infrared laser technology. Thus InP based QCL technology is currently observing a rapid progress to achieve even more efficient compact, room temperature, spectrally pure sources for different applications
The authors would also like to acknowledge the encouragement and support of Dr. K.K. Law from the Naval Air Warfare Center, Dr. T. Manzur from the Naval Undersea Warfare Center, Dr. R. LaMarca from Naval Air Systems Command, Dr. W. Holloway Jr from Naval Air Systems Command, Dr. D. Pavlidis from National Science Foundation, Dr. J. Zavada from National Science Foundation.
References and links
1. M. Razeghi, “High-Performance InP-based Mid-IR Quantum Cascade Lasers,” IEEE J. Sel. Top. Quantum Electron. 15(3), 941–951 (2009). [CrossRef]
2. Y. Bai, N. Bandyopadhyay, S. Tsao, S. Slivken, and M. Razeghi, “Room temperature quantum cascade lasers with 27% wall plug efficiency,” Appl. Phys. Lett. 98(18), 181102 (2011). [CrossRef]
3. Y. Bai, S. Slivken, S. Kuboya, S. R. Darvish, and M. Razeghi, “Quantum cascade lasers that emit more light than heat,” Nat. Photonics 4(2), 99–102 (2010). [CrossRef]
4. N. Bandyopadhyay, Y. Bai, S. Tsao, S. Nida, S. Slivken, and M. Razeghi, “Room temperature continuous wave operation of λ ~ 3-3.2 μm quantum cascade lasers,” Appl. Phys. Lett. 101(24), 241110 (2012). [CrossRef]
5. N. Bandyopadhyay, Y. Bai, B. Gokden, A. Myzaferi, S. Tsao, S. Slivken, and M. Razeghi, “Watt level performance of quantum cascade lasers in room temperature continuous wave operation at λ ~ 3.76 μm,” Appl. Phys. Lett. 97(13), 131117 (2010). [CrossRef]
6. N. Bandyopadhyay, S. Slivken, Y. Bai, and M. Razeghi, “High power continuous wave, room temperature operation of λ~3.4 µm and λ~3.55µm InP-based qantum cascade lasers,” Appl. Phys. Lett. 100(21), 212104 (2012). [CrossRef]
7. Y. Bai, S. Slivken, S. R. Darvish, A. Haddadi, B. Gokden, and M. Razeghi, “High power broad area quantum cascade lasers,” Appl. Phys. Lett. 95(22), 221104 (2009). [CrossRef]
8. Q. Y. Lu, Y. Bai, N. Bandyopadhyay, S. Slivken, and M. Razeghi, “2.4W room temperature continuous wave operation of distributed feedback quantum cascade lasers,” Appl. Phys. Lett. 98(18), 181106 (2011). [CrossRef]
9. S. Slivken, N. Bandyopadhyay, S. Tsao, S. Nida, Y. Bai, Q. Y. Lu, and M. Razeghi, “Sampled grating, distributed feedback quantum cascade lasers with broad tunability and continuous operation at room temperature,” Appl. Phys. Lett. 100(26), 261112 (2012). [CrossRef]