Quantum-Cascade Lasers: Recent advances extend spectral output of QC lasers

JEROME FAIST

More than a decade after its first demonstration, quantum-cascade laser technology is still progressing rapidly. These lasers have now been demonstrated across a broad range of wavelengths, from as short as 3 µm up to more than 300 µm. And even for specifications for which quantum-cascade laser performance was poor—such as wall-plug efficiency and total dissipation—significant improvements have been made. These results and the fact that they were achieved using a standard indium phosphide (InP) technology platform bode well for the future of quantum-cascade lasers as a widespread laser source for chemical-sensing applications.

In a quantum-cascade laser, the light is emitted by electrons that are making transitions between confined states created by quantum confinement in a multilayer semiconductor quantum-well structure.¹ The particularities of this system, as compared to the more commonly used interband transitions, are now well known. The extremely efficient optical-phonon scattering rate between subbands, which occurs on a subpicosecond timescale, means that the system has a very low radiative efficiency and makes population inversion difficult to achieve. So it might seem surprising that a high-performance laser can be fabricated using such a system. However, the same high scattering rate and its low temperature dependence mean that once laser action is achieved, the threshold current is only weakly temperature dependent and the ultrafast scattering rate enables strong electron transport, producing high optical powers. In fact, room-temperature continuous-wave optical powers up to one watt with wall-plug efficiencies close to 10% have recently been achieved at 4.8 µm. Fundamental considerations set the limits of the wall-plug efficiency to about 25% at this wavelength.^{2, 3} In sensing applications, low power dissipation is of paramount importance (see Fig. 1). Milliwatt optical powers are high enough for most detection systems.⁴

Another important feature of this material system is the flexibility it offers in the design of laser sources operating over a very wide range of frequencies—a specific transition energy can be attained by growing a structure with well-designed quantum-well thicknesses. The limitations on the maximum wavelength span of quantum-cascade lasers are different for the short and long end of the range. At the short end, the magnitude of the conduction-band discontinuity (ΔEc) between the two semiconductor materials is an obvious limitation because the energy of the excited state must be kept well below the top of the confinement barriers to prevent significant electron leakage by thermionic emission. This requirement does not apply only to the zone center valley Γ, but the confinement should be large enough to prevent intervalley scattering to the X or L lateral valleys of the well and barrier materials. Depending on the active-region design, this requirement restricts the emitted photon energy to typically less than half of ΔEc. At the long end of the wavelength range, the limit is not as directly related to the materials but stems instead from the growing strength (growing as λ2) of free-carrier absorption.

Significant progress has been achieved recently in the 3 to 4 µm region (see Fig. 2). Previously, high-performance devices based on InP could not be fabricated—especially for wavelengths shorter than 3.5 µm—because of the limited conduction-band discontinuity (to about 820 meV) of the strain-compensated indium gallium arsenide/aluminum indium arsenide (InGaAs/AlInAs) material system. One interesting alternative is to replace the AlInAs with aluminum arsenide antimonide (AlAsSb), which offers a much larger discontinuity of about 1 eV and the possibility of achieving such a value in a lattice-matched system with InGaAs and InP. Indeed, some very interesting new results have been achieved with this material. One remaining limitation however, is the position of the x minimum of InGaAs at approximately 0.5 eV of the Γ minimum for lattice-matched InGaAs, thus limiting the maximum excited-state energy.

A more radical option has been to use InAs for quantum wells and AlSb for quantum barriers. Pioneered originally by H. Ohno of Tokyo University (Japan), this approach has now enabled a group at the University of Montpellier (France) to achieve wavelengths as short as 3 µm with operation close to room temperature and excellent performance—watt-level optical output at 300 K for a wavelength of 3.3 µm.⁵ This result is especially important because, for wavelengths shorter than 3 µm, high performance has been achieved using antimony-based interband devices. Quantum-cascade lasers based on InAs/AlSb are not expected to operate at wavelengths much shorter than the current 3 µm limit because emission of photons with a higher energy will be hampered by both interband reabsorption and the presence of the L-minimum in InAs at 0.73 eV from the Γ point. Further developments in the area of short-wavelength quantum-cascade lasers will require the use of wider-bandgap materials such as gallium nitride/aluminum nitride (GaN/AlN) heterostructures or even II-VI heterostructures. Currently, these materials are still hampered by the relative lack of maturity of the technology.

Terahertz output

Limits on the design and operation of quantum-cascade lasers in the terahertz regime have a very different physical origin. There is the problem of achieving population inversion between subbands spaced by a photon energy (about 10 meV) that is significantly less than the thermal energy at room temperature (26 meV). Another challenge is the decreasing ratio between the intersubband gain cross section and the “free-carrier” loss cross section—the latter increases roughly as the wavelength squared. In recent work, we have demonstrated that this limit can be circumvented by a redesign of the active region. The whole idea of this design is to redistribute the oscillator strength of the injector reabsorption, shifting it to higher photon energies farther away from the laser wavelength. The active transition occurs between a bound state and a miniband that transports the electrons toward the doped injector (see Fig. 3). The latter, is formed by a coupled quantum-well pair, in which the doping is also located.

The energy step between the miniband and the lower state of the injector has several functions. First, it reduces the number of electrons thermally excited from the bottom of the injector to the lower state of the laser transition. Second, most of the reabsorption of the injector now occurs between the lower state of the injector and the miniband, and is designed deliberately at a frequency much higher than the design photon energy. As a result, the absorption at the lasing energy is reduced. Lasers designed using this concept have been demonstrated between 1.2 and 2 THz.⁶

An important feature of quantum-cascade lasers, inherent to the atomic-like nature of the optical transition, is the fact that, in contrast to the interband case, the active medium is transparent on both the blue and red side of the transition. As a result, it is possible to design the active region with a very broad gain spectrum. Such a broad spectrum can be conveniently exploited by inserting the active region in a grating-coupled external cavity. By using a combination of two active regions tuned to two different wavelengths, a total tuning of 265 cm–1 was achieved. Recently, by using a refinement of the design and processing of the device, we achieved a total tuning range of 295 cm–1 with peak power of 800 mW at the center of the tuning range.

Another important consequence of the transparency of the active material on both sides of the gain spectrum is that it enables the gain to be combined with an intracavity nonlinear process. Two recent experiments exploited this effect. In the first, a group from Harvard University (Cambridge, MA) used an active region designed to simultaneously generate two colors in the mid-infrared (mid-IR) and at the same time produce a resonant nonlinear susceptibility.⁷ The two mid-IR beams are therefore mixed intracavity, generating terahertz radiation at the difference frequency. In contrast to the direct generation by quantum-cascade lasers, which is still limited to low cryogenic temperatures, this nonlinear generation process, albeit much less efficient, is only limited by the temperature dependence of the mid-IR devices, which can work far above room temperature.

In the second experiment, wavelength conversion from the terahertz to the near-IR was demonstrated by a group from the University Paris Diderot (Paris, France) using a terahertz quantum-cascade laser engineered to phase-match the GaAs bulk nonlinearity at telecom wavelengths.⁸ In this nonlinear process, the terahertz radiation was transferred as an optical sideband onto an optical carrier. In a further experiment, the terahertz radiation was modulated at microwave frequencies, and the resulting modulation carried by the optical fiber. These ingenious experiments demonstrated that the very-low-loss transmission capability of an optical fiber (that even includes the possibility of regenerating the signal with optical amplifiers) can be leveraged to carry terahertz signals over long distances.

REFERENCES

1. J. Faist et al., Science 264, 553 (1994).

2. A. Evans et al., Appl. Phys. Lett. 91, 071101 (2007).

3. J. Faist, Appl. Phys. Lett. 90, 253512 (2007).

4. www.alpeslasers.com

5. J. Devenson, O. Cathabard, R. Teissier, and A. N. Baranov, Appl. Phys. Lett. 91, 141106 (2007).

6. C. Walther, M. Fischer, G. Scalari, R. Terazzi, N. Hoyler, and J. Faist, Appl. Phys. Lett. 91, 131122 (2007)

7. M.A. Belkin, et al., Nature Photonics 1, 288 (2007).

8. S.S. Dhillon et al., Nature Photonics 1, 415 (2007).

Tell us what you think about this article. Send an e-mail to [email protected].

JEROME FAIST is a professor at the Institute for Quantum Electronics of the ETH Zurich, Wolfgang-Pauli-Str. 16, 8093 Zürich, Switzerland; e-mail [email protected]; www.iqe.ethz.ch.