A revolutionary active region is the driver behind the record single-facet output powers emanating from Pranalytica’s quantum cascade lasers (QCLs). This advance will spur the launch of compact, lightweight, multi-watt, mid-wave infrared lasers, say the company’s Richard Maulini, Arkadiy Lyakh, Alexei Tsekoun and Kumar Patel.

QCLs are a novel class of laser that can plug critical gaps in the mid-wave and longwave infrared spectral regions that are currently served by very few continuous wave (CW), room temperature solidstate sources. QCLs can operate in this spectral range because their emission is not based on conduction band to valence bands transitions that govern the emission of conventional laser diodes. Instead, they generate laser emission from transitions between confined intersubband states formed within a superlattice of alternating layers of materials with lower and higher bandgaps, known as quantum wells and barriers. The emission wavelength is then dictated by properties of the superlattice, such as the thickness of the wells and barriers, and this opens up a range of wavelengths that can be reached through bandgap engineering.

The fundamental idea behind the QCLs is not new and dates back to the early 1970s. However, practical realization of this device took nearly 25 years, due to the extreme demand that the laser structure puts on epitaxial quality. Even after the first working QCL was produced, this class of laser remained little more than a laboratory curiosity for a decade. Initial performance was poor, and the first generation of QCLs were available only in the form of individual chips, or chips on carrier assemblies. Consequently, integrating this class of laser into a system required expertise in QCL handling, powering and packaging. In addition, early designs had to be cooled to cryogenic temperatures - CW, room-temperature performance was only realized in 2002.

At Pranalytica, our mission has been to improve the performance of QCLs and their packaging, so that they can make the transition from laboratory devices to commercial lasers that can serve a host of applications. Thanks in part to funding from the US Defense Advanced Research Projects Agency, we have made significant strides in this direction, including a recent roomtemperature demonstration of 3W, CW output from one single facet of a 4.6 μm laser. This record-breaking laser, which has a wall-plug efficiency of 13%, was the result of multiple advances that span the entire QCL production chain, from fundamental design of the active region through to thermal management of the chip.

Beckoning applications

Thanks to these improvements, our QCLs are now attractive candidates for real world applications. In the defense market space, they are being explored for protection of military and civilian aircraft, and high-power handheld devices are being tested as target illuminators. In addition, several non-defense QCL applications are imminent, including free-space optical communications, ultra-sensitive trace-gas sensing based on photo-acoustic spectroscopy and other detection techniques and remote sensing.

There is no denying that it has taken the QCL community a long time to get to the stage where its lasers are commercially viable. That’s partly because this class of laser has a relatively complex design, consisting of hundreds of superlattice layers, each with a thickness of just a few nanometers. Imperfections in the heterointerfaces can cause undesirable carrier scattering, and in the worst case, distortion of the shape of a given quantum level, driving the design away from the optimum.

In addition, QCL performance can be compromised by small deviations in growth uniformity, both across the wafer and in the timings of the growth process. Since carriers traverse the superlattice structure sequentially, any thickness deviations within the structure will degrade device performance. So it is no surprise that advances in MBE held the key to practical realization of the first QCLs.

Most QCLs are made from a combination of InGaAs wells and InAlAs barriers, grown on an InP substrate. This material system is popular because it is well understood, thanks to its use in numerous telecom lasers. But that’s not the only reason for selecting this particular material system – the pairing of In0.52Al0.48As and In0.53Ga0.47As is lattice-matched to InP, simplifying the epitaxial growth of very thick QCL structures. This combination produces a conduction band offset in excess of 0.5 eV, so it is possible to construct QCLs emitting at 6 μm and beyond. Shorter wavelengths can be reached by increasing the depth of the quantum wells. A higher conduction band offset is then needed, which can be realized through increasing the indium concentration in InGaAs, along with the aluminium concentration in InAlAs. However, compositional adjustments pay the penalty of adding strain into the superlattice, because these ternary compositions are no longer lattice-matched to InP substrates. Strain can be partially ameliorated through careful selection of alternating compressively strained and tensile strained layers of appropriate thickness. By optimizing this approach, we have made record-breaking 4.6 μm lasers that contain about 1 percent strain.

The leading materials candidates for wavelengths shorter than about 3.8 μm are III-V antimonides, which have larger conduction band offsets. Present efforts have focussed on either the InGaAs/AlAsSb or InAs/AlSb systems. QCLs built from these pairings of materials hold significant promise for short wavelength emission, but room-temperature, CW performance is yet to be realized.

 

Superior active regions

QCLs are unipolar devices, with emission governed by intersubband transitions that do not depend on the intrinsic properties of the material, but rather on the thickness and depth of the quantum wells and barriers that make up the gain medium. The challenge for designers of QCLs is to simultaneously optimize all the quantum cascade structure parameters influencing laser performance.

 



Fig.1. A typical QCL design employs a twophonon active region. Longitudinal optical phonons are needed for transitions between levels 3 and 2, and 2 and 1 (left). Pranalytica uses an alternative approach with a non-resonant extraction active region that vastly increases the freedom of QCL design (right)

 

Most of today’s QCLs are designed using the two-phonon resonance approach (see Fig.1). The lower laser level 3 is depopulated by two consecutive non-radiative transitions to the levels 2 and 1, which are each spaced by roughly the longitudinal optical (LO) phonon energy ELO in the material (In the case of InGaAs, ELO is about 35 meV). With this design, the lower laser level is rapidly depopulated, thanks to fast resonant, phonon-assisted scattering. But this advantage has to be weighed against the shackles of the two-phonon QCL design. Once the two phonon resonance condition is met, there are not sufficient degrees of freedom remaining in the design to optimize its other aspects. For example, with this design it is difficult to increase the energy spacing E54 between the upper laser level 4 and the active region level 5, which ultimately suppresses parasitic carrier injection into the latter state.

We have regained design flexibility for the QCL by removing the two-phonon resonance condition and turning to a non-resonant extraction approach. Our design replaces a single, resonant final state with several closely spaced final states separated from the state above by substantially more than ELO. Even though the transition to each of the new final states is slower than that in the resonant case, carrier lifetime in the state above is reduced thanks to the introduction of several parallel extraction paths.

 



Fig. 2. Pranalytica’s 4.6 μm QCLs feature a nonresonant extraction active region and can deliver a recordbreaking CW output of 3W at 293

 

MBE or MOCVD?

The first QCLs were produced by MBE, a technique that is adept at producing precise growth of thin layers with abrupt heterointerfaces. This form of epitaxy dominated the growth of QCLs for a decade, but notable improvements to MOCVD technology during the 1990s have enabled process engineers to now have a choice of deposition techniques. MOCVD’s potential advantages include a faster growth rate - a particular cost advantage for the very thick QCL structures - and nominally lower reactor maintenance.

The first MOCVD-grown QCL was demonstrated in 2005 by researchers at the University of Sheffield, UK, and since then this approach has been gaining traction. As of today there is no consensus in the QCL field regarding fundamental superiority of MBE or MOCVD, and we keep an open mind, producing lasers with both techniques.

We have produced a portfolio of high-quality, QCL epistructures for emission in the medium-wave infrared by optimizing our growth process for strained structures containing hundreds of nanometer-thick layers. QCL quality is normally assessed through measurements of the spontaneous emission spectrum’s full-width at halfmaximum: our 4.6 μm structures have a value of just 26 meV at room temperature, 20 percent less than that of previous growths of the same design.

To simplify systems integration of our QCLs, we have developed advanced, high-reliability, self-contained packages that employ well-proven telecom practices. These require only electrical power and heat sinking to operate.

QCLs run in CW mode generate a substantial amount of heat – typically 10 MW/cm3 – and we have addressed these thermal issues with a buried-heterostructure geometry. The epitaxial laser structure is etched to form near-vertical ridges defining the side-walls of the laser cavity, and valleys are overgrown with a material providing superior thermal conductivity to that of the active region superlattice. This additional material, MOCVD-deposited iron-doped InP in the case of InGaAs/InAlAs QCLs, is transparent to the lasing wavelength and electrically insulating. At the package level, we have pioneered the use of epi-side mounting of QCLs for efficient thermal management. Thanks to optimized thermal management, we have realized a ratio of pulsed-to-CW output power of just 1.5 for a 3W QCL attached to a diamond submount. This type of submount is widely used to report results, because it is very efficient at extracting heat from QCLs, but its thermal expansion mismatch to the thermal expansion of the QCL material impairs long-term reliability. In the case of diamond substrates, to prevent damage to the laser, QCLs are soldered to the submounts with soft indium solder, but this leads to solder electro-migration at high temperatures and/or high currents densities.

We circumvent all these issues by: utilizing AlN submounts with a thermal coefficient similar to that of the laser; bonding the submount to the QCL with hard AuSn solder; and optimizing device geometry and facet coatings for room-temperature, CW operation. This has enabled a maximum CW output of 2.9W at 293K. We have also studied the performance of our QCLs without thermoelectric cooling (often called “uncooled” operation) and found that they produce a maximum average power of 1.2W, and a CW power in excess of 1W. Recently, thanks to further improvements in thermal management, we have raised the bar for average power output for “uncooled” operation to 2.0W.

It is worth noting that our output power and wallplug efficiency figures are given for single-ended emission. As with all edge-emitting semiconductor lasers, as-cleaved QCLs emit light equally from both facets, and many researchers report the combined output from both facets as the output power. But the vast majority of applications demand single-ended output, a requirement that is fulfilled by depositing a high reflectivity coating on one of the facets. This is a daunting task for high-power QCLs – optical power density on the facet of a 2W laser can exceed 10 MW/cm2. However, we have risen to the challenge of producing a reliability coating operating in the mid-infrared and developed QCLs emitting 1W or more that can deliver many thousands of hours of degradation-free operation (see Fig.3).

 



Fig.3: A robust facet coating has aided development of reliable, highpower midinfrared QCLs. This includes 2.5W QCLs that show no signs of degradation during a 100- hour predelivery test

 

To facilitate the integration of our QCL chips into various applications and ensure long-term reliability, these devices are installed into custom-designed butterfly type packages containing a thermoelectric cooler and collimation optics.

 



QCLs can be mounted in butterfly packages containing thermoelectric coolers and collimation optics

 

This complete system, which is hermetically sealed in nitrogen atmosphere, is very compact – its mass is less than 100 g and it has a volume below 50 cm3. An additional appeal of these hermetic packages is that by providing a well-defined electrical and optical system interface, they free system designers from needing to become QCL experts, thereby dramatically reducing the risk and time for developing QCL applications.

 



Output powers in excess of 1W can be produced from an uncooled package (left). A table-top variant has also been produced (above). This version can also drive the laser with pulses exhibiting rise and fall times of less than 5 ns

 

Some applications require higher powers, and to meet these needs we have developed specially packaged, cryogenic QCLs that deliver a CW output in excess of 7W at 80K. However, the increase in the output power of this chip has the downside of a more expensive, larger and heavier system.

Any systems integrator requires infrastructure in addition to the laser package, including drive electronics and package thermal management. These tasks for room-temperature, CW QCLs are challenging, because this class of laser requires significantly higher drive voltages (12-16V) than its diode cousin, as well as a more capable form of external thermal management. To reduce the barrier to entry for QCL  integration, we have developed appropriate drivers, heatsinks and controllers that represent the entirety of equipment necessary for operating a QCL in a customer’s system. Such systems, which are now commercially available in several different versions with a CW output power in excess of 2.5W, should help to unlock the door to deployment of lasers for protection of military and civilian aircraft, gas sensing, and a host of other important applications.

 



Pranalytica manufactures commercial, turn-key, high-power QCL systems that operate off standard AC power and can deliver CW output powers in excess of 2.5W (above). The laser head houses the QCL and its thermal management system, and a controller provides all of the necessary supply, command, control and safety functionality. This version can deliver more than 2W of collimated radiation at 4.6 μm. For applications requiring projection of the QCL beam over several kilometers, the laser head can be coupled to an external objective (right)

 

Further reading

Jérôme Faist et al., Science 264, 553–556 (1994).

Mattias Beck et al., Science 295, 301–305 (2002).

Jérôme Faist et al. IEEE J. Quant. Electron. 38, 533 (2002).

Arkadiy Lyakh et al. , Appl. Phys. Lett. 95, 141113, (2009).

Richard Maulini et al., Appl. Phys. Lett. 95, 151112, (2009).

Robert Curl et al., Chem Phys. Lett. 487, 1-18 (2010).