This chapter presents an introductory review on quantum cascade lasers (QCLs). An overview is prefaced, including a brief description of their beginnings and operating basics. Materials used, as well as growth methods, are also described. The possibility of developing GaN-based QCLs is also shown. Summarizing, the applications of these structures cover a broad range, including spectroscopy, free-space communication, as well as applications to near-space radar and chemical/biological detection. Furthermore, a number of state-of-the-art applications are described in different fields, and finally a brief assessment of the possibilities of volume production and the overall state of the art in QCLs research are elaborated.

1. Introduction

Quantum cascade lasers (QCLs) are based on a fundamentally different principle to ‘classic’ semiconductor lasers, that is, they use only one type of charge carriers, electrons, using intersubband transitions, so they can be called unipolar lasers. QCLs were conceived in the early 1970s. First, Esaki and Tsu [1] fabricated the first one-dimensional periodic potential multilayer by periodically varying the composition during epitaxial growth (superlattice). Later, QCLs were proposed by Kazarinov and Suris [2] and finally first demonstrated at Bell Laboratories in 1994 by Faist et al. [3]. Using superlattices leads to both quantum confinement and tunnelling phenomena, the basic processes in QCLs operation.

A conventional laser diode generates light by a single photon that is generated from an electron interband transition; this means that a high-energy electron in the conduction band recombines with a hole in the valence band, being the energy of the photon determined by the band-gap energy of the material system used. However, QCLs do not use bulk semiconductor materials in their optically active region, but a periodic series of thin layers of varying material composition forming a superlattice, which leads to an electric potential that changes across the length of the device (one-dimensional multiple quantum well confinement), splitting the band-permitted energies into a number of discrete electronic subbands, making electrons cascade down a series of identical energy steps built into the material during crystal growth, and emitting a photon at every step, unlike diode lasers, which emit only one photon over the equivalent cycle. With an appropriate design of the thickness of these layers, population inversion is achieved between discrete conduction band-excited states in the coupled quantum wells by the control of tunnelling, making laser emission possible. Therefore, the position of energy levels is mainly determined by the thickness of the layers, rather than the material, and thus allowing tuning the emission wavelength of QCLs over a wide range in the same material system. Thus, one electron emits a photon during every intersubband transition within the quantum well (QW) in the superlattice, and then can tunnel into the next period of the structure where another electron can be emitted, leading QCLs to outperform diode lasers operating at the same wavelength by a factor even greater than 1000 in terms of power.

Classically, a QCL is made of a periodic repetition of active sections, which consist of tunnel-coupled quantum wells and injector, where a miniband is formed. As Figure 1 shows [4], from the injector miniband the electrons are injected into the upper laser energy level (4) of the active section, where the laser transition takes place. Afterwards, the lower laser energy level (3) is emptied by longitudinal optical emissions (LO emissions) and the electrons enter the next step by tunnelling.

Combining materials in the active region, QCLs could be designed to emit at any wavelength over a wide range of the spectrum [5], as the emission wavelength is determined by quantum confinement. Figure 2 shows a typical QCL in operation and some commercial examples. These structures are typically grown using either molecular beam epitaxy (MBE) or metal-organic chemical vapour deposition (MOCVD), being the most used growth mechanisms utilized to grow the alternated different semiconductor layers required for heterostructures fabrication on to a substrate. Ever since the first QCL was fabricated using InGaAs/InAlAs grown over InP substrate [1], other materials have been used in order to fabricate QCL structures, such as GaAs/AlGaAs, InGaAs/AlAsSb, InAs/AlSb, Si/SiGe and GaN-based materials, such as AlGaN/GaN and AlN/GaN.

First commercialized a decade later of their first demonstration [6], the key features of these lasers reside in the fact of their high optical power output and, on the other hand, their tuning range and room-temperature operation. Spectroscopy applications are related to gas detection and analysers (pollutants, components, etc.). Other practical uses include industrial control, plasma chemistry and detection, such as collision avoidance radar or poor visibility-driving condition aids. Finally, the 3 to 5μm atmospheric window would make QCLs perfect candidates for substitution of optical fibre in high-speed and free-space communications.

2. Fundamentals and operating principles

The main particularity in QCLs is the fact that instead of using bulk semiconductor materials in their optically active region, a periodic series of thin layers are used, that is, superlattices, consisting of a number of quantum well—barrier system equally spaced, introducing a multiple quantum well (MQW) leading to one-dimensional confinement allowing an electric potential variation (band splitting) that results in a number of discrete electronic subbands. In order to achieve the population inversion required for laser emission, it is necessary for a proper thickness and composition layer design. These confined energy levels depend on the layer thickness, so the tunability of the emission within the same material relies, in principle, on thickness variation, although multiwavelength QCLs emit by means of different materials within the same structure and multiple resonators [7, 8].

In classical semiconductor laser diodes, electrons and holes recombine across the band gap, thus, generating one photon per e−-h+ pair recombined. However, in QCLs this is not the case, as an electron in the conduction band within a QW emits one photon whenever it undergoes an intersubband transition, that is, thermalization into lower-energy levels, emitting one photon. This electron could tunnel into the next period of the structure, where the mentioned transition happens again emitting another photon. This phenomenon of a single electron emitting multiple photons as it passes through different periods of the structure is called cascade, making the quantum efficiency of QCLs greater than unity and leading to their high optical output power.

QCLs are usually three-level lasers, whose level transitions are depicted in Figure 3. In these lasers, the wave function formation is faster than scattering between states; hence, the time-independent solutions of the Schrödinger equation can be applied, so the system can be modelled using rate equations. Each subband, i, is considered to have ni electrons, and a scatter between the initial and the final subband, f, will have a scattering rate, Wif, and lifetime, τif. If no other subbands are populated, the rate equations are described as extraction and injection of electrons, respectively (depicted in Figure 3), whose values are both equal in steady-state conditions, where time derivatives are zero. Hence, the general rate equation for electrons in a subband i within an N level system is where I = Iee = Iie. At low temperature, absorption is near zero: hence, τ32 > τ21, so W21 > W32, and n3 > n2 leading to the existence of population inversion.

where Iee and Iie are the extraction and injection of electrons, respectively (depicted in Figure 3), whose values are both equal in steady-state conditions, where time derivatives are zero. Hence, the general rate equation for electrons in a subband i within an N level system is

where I = Iee = Iie. At low temperature, absorption is near zero:

hence, τ32 > τ21, so W21 > W32, and n3 > n2 leading to the existence of population inversion.

The scattering rate between two subbands strongly depends upon the overlap of the wave functions and energy spacing between the subbands. In order to decrease W32, the overlap of the upper and lower laser levels is reduced. This is often achieved through designing the layer thickness such that the upper laser level is mostly localized in the left part of the well of the 3QWs active region, while the lower laser level wave function is made to mostly reside in the central and right part of the wells, leading to the so-called diagonal transition. A vertical transition is one which the upper laser level is rather localized in the central and right part of the wells, increasing the overlap and, therefore, W32, reducing the population inversion but increasing the strength of the radiative transition and hence the gain. On the other hand, in order to increase W21, the lower laser level and the ground level wave functions are designed to obtain a good overlap, and the energy spacing between the subbands is designed such that it is equal to the longitudinal optical (LO) phonon energy so that the resonant LO phonon-electron scattering can depopulate the lower laser level. For instance, the LO-phonon energy is around 36 meV in GaAs (which is comparable to the room temperature kBT value of around 26 meV) and around 91 meV in GaN [9].

Tunable semiconductor lasers can be produced by using multiple resonators or multisection injection devices [10–12]. In fact, tunable QCLs have been demonstrated [7] some time ago. Moreover, a QCL with a heterogeneous cascade containing two substacks previously optimized to emit at 5.2 and 8.0 μm, respectively, was presented by Gmachl et al. [13]. On the other hand, single-mode tunable QC distributed feedback lasers emitting between 4.6 and 4.7μm wavelength have been reported [14]. These lasers were pulsed, continuously tunable single-mode emission and were achieved from 90 to 300 K with a tuning range of 65 nm and a peak output power of approximately 100 mW at room temperature—so the lasers described by Pellandini et al. [12], Gmachl et al. [13] and Köhler et al. [14] were laser sources for the mid-infrared region. In order to realize near-infrared QCLs, optical non-linearity in intersubband lasers has been used to design such lasers emitting at 4.76 μm, with third harmonic and second harmonic generation at 1.59 and 2.38 μm, respectively [15].

3. Materials used and growth methods

The first QCL in 1994 used InAlAs as the cladding layers due to its low refractive index of around 3.20. The core region, which includes active and injector regions, usually has 500 stacking layers consisting of alternative InGaAs and InAlAs layers with total thickness about 1.5–2.5 μm. The average refractive index in this region can be calculated according to the volume fraction of these two constituent materials and is often around 3.35, which is clearly higher than the cladding layers [16]. The confinement factor with typical Np of approximately 30 is usually around 0.5. To reduce the optical loss, the cladding layers are usually doped with a low concentration of 5 × 1016 cm−3 and the separate confinement heterostructures (SCHs) are implemented with InGaAs of high refractive index to increase the optical confinement factors. In later designs, the InAlAs cladding layers were replaced by InP because it has a lower refractive index of around 3.10 and a higher thermal conductivity for better heat treatment, which is a critical step for the device performance [16].

So far, the emission wavelength of QCLs has been extended from the near infrared (around 100 THz) to terahertz regimes. While the longest demonstrated wavelength is 1.6–1.8 THz with GaAs/Al0.1Ga0.9As system at 80 K under continuous-wave (CW) operation [17], the shortest wavelength has been extended to 3 μm with In0.53Ga0.47As/AlAs0.56Sb0.44 system at 300 K under pulsed operation [18]. The highest output power of short-wavelength QCL (4.6 μm) under CW operation at room temperature has been demonstrated (100 mW and maximum temperature 105°C). All of these state-of-the-art QCLs have been grown by MBE so far, but for industry MOCVD is preferable for mass production. As a result, attempts have been made to grow QCLs by MOCVD. Until now, there are only three groups successfully reporting the growth of QCLs by MOCVD: an InP-based QCL (λ ≈ 8.5 μm) operating in pulsed mode at room temperature, with low-temperature threshold current density in the region of 1500 A/cm2 [19], an In0.53Ga0.47As/In52Al0.48As QCL (λ ≈ 8.5) operated in continuous-wave operation at room temperature with an output power of 5.3 mW [20], and Diehl et al. [21] managed to grow a QCL working in continuous-wave mode above 370 K, with an optical output power of 312 mW at room temperature and an emission wavelength of 5.29 mm. More effort is still required for MOCVD growers to develop more advanced techniques to compete with MBE. Having said that, room-temperature operation in InAs/AlSb QCLs has been achieved at 4.5 μm [22]; more recently, MBE-grown InAs/AlSb on n-InAs (100) substrate QCL operating at 8.9 μm has been reported, with a maximum operating temperature of 305 K [23].

QCLs provide one possible method of realizing high-efficiency light emitters in indirect band gap materials such as silicon. Electroluminescence at terahertz frequencies from Si/SiGe intrasubband transitions has been demonstrated [24], and recently silicide low-loss (down to 2 cm−1) waveguides were designed [25, 26]. Figure 4 shows a transmission electron microscopy (TEM) image from a Si/SiGe QCL.

Although GaN-based materials have not been employed to fabricate QCLs, they are also promising materials to be used as such devices. III–V nitrides are known in their wurtzite structure to possess a large spontaneous polarization and piezoelectric constants. As a result, two-dimensional (2D) charges build up at nitride heterointerfaces where the polarization discontinuities occur, causing strong built-in electric fields [27]. Furthermore, GaN is a material with large LO-phonon energy, leading to a thermal population reduction of the lower laser state, a feature desirable for high-temperature operation of terahertz QCLs, as proposed by Diehl et al. [21]. On the other hand, ultrafast LO-phonon scattering in GAN/AlGaN QWs can be useful in order to rapidly depopulate the lower laser state [28, 29]. Lastly, the large LO-phonon energy can also increase the lifetime of the upper laser state by reducing the relaxation of electrons with higher in-plane kinetic energy via emission of a LO phonon. Using low-pressure MOCVD, GaN/AlGaN active regions for QCLs have been grown by Huang et al. [30]. GaN/AlGaN active layers are depicted in Figure 5.

Finally, typical phonon frequency redshift is a key indicator of good periodicity of MOVPE-grown GaN/AlGaN QCL structures, 822 cm−1 in the superlattice has been measured, indicating a redshift with respect to the single AlGaN layer [31].

AlN/GaN compound is another possible material to be used to fabricate quantum cascade structures. A suitable method of fabrication is hot-wall epitaxy (HWE) [32], a low-cost, convenient and scalable technique, where the epitaxial layers are grown under conditions as near as possible to thermodynamic equilibrium, allowing a minimum material loss. Inoue et al. [33] grew short-period superlattices consisting of five periods of GaN wells of 10 (or nine) molecular layers (MLs) each with 1 ML AlN barriers, which was designed to emit photons at a wavelength within the mid-infrared range (around 5 μm) [34, 35].

4. Applications

We could say that QCLs, in short, are eligible to those applications where a powerful and reliable mid-infrared source is required. For instance, most chemical compounds have their fundamental modes in the mid-infrared region (3–15 μm) of the electromagnetic spectrum, thus making this range of paramount importance for gas sensing and spectroscopy applications. The so-called two ‘atmospheric windows’ are two windows corresponding to the ranges 3–5 and 8–12 μm, at which the atmosphere happens to show a high transparency leading to remote sensing and detection in those windows. In this section, a brief description of the current applications of QCLs is elaborated, being summarized as subsequently.

4.1. Trace-gas detection by optical methods in the mid-infrared

Most trace gases of importance, from products of fossil fuel burning to constituents of human breath, have telltale absorption features in this wavelength range, that is, their ‘fingerprint’ region of the spectrum, as a result of molecular rotational-vibrational transitions [16]. Narrow-linewidth, tunable semiconductor lasers in this wavelength range are used to spectrally map out and qualitatively and quantitatively detect these trace gases, by a measurement technique called tunable infrared laser diode absorption spectroscopy (TILDAS) [36]. A schematic representation of a TILDAS is shown in Figure 6. The advantage of TILDAS is its high sensitivity and specificity, usually combined with the advantages of the solid-state device approach.

This technique has been successfully introduced in distributed feedback quantum cascade laser (DFB-QCL) structures [37]. DFB-QCLs were first introduced in 1997 [38], providing continuously tunable single-mode laser output, and were demonstrated for the first time in trace-gas sensing applications 1 year later [37]. DFB-QCLs operate as follows: a grating with period L is incorporated into the wavewide, lowering the threshold gain (by reducing the outcoming loss) for a different wavelength close to the Bragg wavelength:

λ=2 ⋅neff(T)⋅L$�=2 \cdot {�}_{\text{eff}}(�)\cdot �$E6

where neff is the effective refractive index of the waveguide mode, and is a function of temperature. As temperature increases, wavelength shifts to longer values. Changing the heat-sink temperature could control laser temperature; however, the process is slow due to the fact that it implies adiabatically temperature change of a large volume. Tuning rates increase with heat-sink temperature current from 0.3 to 0.4 nm K−1, for a laser emitting at approximately 5.2 μm, and from 0.4 to 0.65 nm K−1, for a laser whose emitting wavelength is around 8 μm [16].