Since the first quantum cascade lasers (QCLs) were demonstrated in 1994^[1], a significant development in QCLs has been seen in terms of high power, high efficiency, broad spectrum range, and high beam quality^[2–5]. QCLs have played an important role in applications such as chemical detection, high-resolution spectroscopy, free-space optical communication, and medical diagnostics^[6–10]. Following the first report of continuous-wave (CW) mode operating at room temperature in 2002^[11], research on high-power QCLs operating in a CW mode at room temperature has gained considerable attention^[12,13]. In 2011, Bai et al. reported a QCL operating at 298 K with a peak power of 5.1 W at 4.9 µm in a CW mode. This was achieved using a “shallow well” active region design and high-AlAs barrier layers^[14,15]. The insertion of AlAs layers in the structure can minimize carrier leakage by enhancing the wavefunction confinement from the continuum bands^[16], thereby improving temperature stability and reducing leakage current^[17]. Later in 2020, Razeghi et al. employed a similar AlAs barrier structure to achieve a record CW power of 5.6 W at a wavelength of 4.9 µm and 293 K^[18]. Both devices were fabricated with a ridge width of 8 µm. While increasing ridge width can enhance output power by enlarging the optical cavity and therefore the volume of the active core^[19], it also presents difficulties like heat dissipation, which impacts the CW performance, and the excitation of the higher-order transverse modes, which degrades the quality of the laser beam^[20].

For practical applications, devices with high CW power at room temperature along with high beam quality are desired. Several methods can be adopted to achieve high beam quality, including reducing the ridge width, using the cavity ridge that is angled to the facet norm^[21], selectively increasing the loss of the higher-order modes with metal contacts on side walls^[22], and using a tapered waveguide geometry^[23,24]. Significant progress has been made with these methods for different wavelengths. In 2012, Lyakh et al. reported a CW power of 4.5 W with a single spatial mode (measured slightly above threshold) tapered waveguide geometry QCL at a wavelength of 4.7 µm at 283 K^[25]. In 2013, long wavelength (>8μm) QCLs were reported with the beam quality factor (M2) of 1.4–2.0 (1.6 W pulsed)^[26], 2.08–2.25 (2.5–3.8 W pulsed)^[27], and 1.8 (4 W pulsed)^[28]. In 2023, Fei et al. reported a QCL operating in a CW mode with a single transverse mode (M2=1.15) and an output power of 3 W emitting at 4.6 µm^[29]. However, high power (>4W) and single transverse mode have not been reported simultaneously for QCLs around 5 µm at room temperature.

In this work, we present a 5 µm mid-infrared QCL that achieves both high CW power and single transverse mode at room temperature. The active region is the same as that in Ref. [15]; the average doping of the laser core is increased to 4×1016cm−3, with AlAs quantum barrier layers and grown by home-made solid-source molecular beam epitaxy (MBE). At 298 K, the device achieved a peak power of 4.7 W in a CW mode with a maximum wall plug efficiency (WPE) of 14.7%. The M2 was measured to be below 1.5 with a power up to 4 W, remaining in single transverse mode operation. In a pulsed mode, the output power reached 10.2 W, with a maximum WPE of 22.7%.

In the important 4.5–5.0 µm spectrum range of QCLs, it is necessary to design a relatively narrow ridge (compared to the record 5.6 W device of 8 µm ridge width) to ensure a stable single spatial mode output^[30]. Single-mode operation is usually limited by the difference between the threshold gain of the fundamental mode and the higher-order mode. The threshold gain (gth) of the (i) lateral mode of a Fabry-Perot cavity ridge laser can be calculated by^[31,32]αw(i)=4πλIm(neff(i)),(1)αm(i)=12Lln(1R1R2(i)),(2)Γ(i)=∬coreϵ|E(i)(x,y)|2dxdy∬allϵ|E(i)(x,y)|2dxdy,(3)gth(i)=αi(i)+α(i)mΓ(i),(4)where αw and αm are the waveguide loss and the mirror loss, respectively, Γ is the optical confinement factor, neff is the effective refraction index of the waveguide, λ=5μm is the wavelength, L=7.5mm is the cavity length, and R1 and R2 are the reflectances of the two high-reflection (HR) and anti-reflective (AR) coated facets of the Fabry-Perot cavity, respectively. E(i)(x,y) is the electric field of the (i) eigenmode.

The transverse magnetic fundamental optical mode (TM00) and the transverse magnetic first-order mode (TM01) electric field strength distributions were found using COMSOL Multiphysics shown in Fig. 1, and initial results are summarized in Table 1. The waveguide structure is shown in Fig. 2(a). The refractive index values of the materials used in the COMSOL model are the same as in Ref. [30].

For a device with a 8 µm ridge width, the threshold gain difference between the TM00 mode and the TM01 mode is calculated at 0.49cm−1, while for the 7.5/7/6.5 µm device is 0.65/0.75/0.95cm−1, respectively. We selected 6.5 µm as the ridge width in order to obtain a single transverse mode and steady beam quality, since the threshold gain difference between the TM00 and TM01 modes of the 6.5 µm device is almost double that of the 8 µm device.

Figure 2(a) shows the schematic of the device structure. The epitaxial regrowth was employed to achieve a buried heterostructure for better thermal dissipation. The active region is done by MBE, while the others are fabricated by metal-organic chemical vapor deposition (MOCVD), including a 0.5 µm InP buffer layer, a 3 µm low-doped (2×1016cm−3) InP cladding layer, a 0.04 µm InGaAs confinement layer, the active region (a 30-stage InGaAs/InAlAs/AlAs quantum cascade structure), a 0.02 µm InGaAs confinement layer, a 2 µm lowly doped (2×1016cm−3) InP cladding layer, a 2 µm doped (4×1017cm−3) InP cladding layer, and a 0.65 µm highly doped (2×1019cm−3) InP contact layer. After photolithography and wet chemical etching to obtain the 6.5 µm ridge, regrowth was performed by MOCVD for Fe-doped InP to bury the structure. The insulating Si3N4 (300 nm) layer was deposited by plasma-enhanced chemical vapor deposition (PECVD) and the Ti/Pt/Au (30/30/300 nm) metal contact was deposited by e-beam evaporation; the refractive values of different materials are shown in Table 2.

See Fig. 2 (b) for a cleaved cross-section of the device. The front and rear facets of the device’s cleaved surfaces are coated with anti-reflective (AR) and high-reflective (HR) films, utilizing Y2O3 (420 nm thickness) and Ti/Au as the coating materials, respectively. The device was epi-down bonded using Au80Sn20 solder to an AlN heat sink to enhance thermal dissipation.

The MBE growth was optimized for best surface quality and characterized by X-ray diffraction (XRD) and an atomic force microscope (AFM). The sample with InP layers grown by MOCVD was loaded into the MBE system and underwent heating for deoxidation treatment in an arsenic environment. By observing InP surface reconstruction patterns via reflection high-energy electron diffraction (RHEED) and optimizing the deoxidation temperature, we established an optimal initial surface for QCL active region growth. At the same time, by optimizing the growth conditions such as growth temperature and V/III ratio for the InGaAs/InAlAs/AlAs superlattice structure, we achieved a flat surface in the active region.

Figure 3(a) shows the XRD result of the MBE active region material. The diffraction satellite peaks are clear and sharp, with the full width at half-maximum (FWHM) near the substrate peak region measured to be 17.7s±1.2s. This indicates the uniform growth across the 30 stages of the active region material, and a good quality of the steep heterojunction interfaces. Figure 3(b) shows the AFM image of the sample surface after MBE, with the surface roughness root mean square (RMS) value of 0.160 nm, showing that the active region surface is smooth and flat.

The device was characterized at different temperatures with thermoelectric cooling (TEC). For CW modes, a TDK Lambda model Z36-24-LAN power supply was used as the driver. For pulsed modes, a driver power supply (model HLD-300 from Jingji Technology Co., Ltd.) was employed. Power acquisition was performed using a Gentec power meter, and voltage data were collected with a Keysight voltage oscilloscope. Figure 4(a) shows the light-current-voltage (L-I-V) characteristics in the temperature range of 288–328 K in pulsed modes. At 288 K, the maximum power reached 10.2 W, and the maximum WPE reached 22.7%. The threshold current density and the slope efficiency were 1.89kA/cm2 and 3.46 W/A, respectively. At 328 K, the threshold current density increased to 2.34kA/cm2, and the slope efficiency dropped to 3.26 W/A. The maximum power dropped to 8.55 W and the maximum WPE to 18.5%. Figure 4(b) shows the results in a CW mode. The peak power reached 4.7 W with 14.7% WPE, with the threshold current density and slope efficiency being 2.23kA/cm2 and 2.73 W/A, respectively. Compared to a pulsed mode, the device exhibited a higher threshold current density and lower slope efficiency in a CW mode. This was due to the heat accumulation in the CW mode and therefore a higher active region temperature, which limited the output power of the device.

The internal quantum efficiency (ηi) and the waveguide loss (αw) were obtained from the mirror-loss study by fitting the inverse external quantum efficiency and the inverse mirror loss^[33]. The cavity length and the inverse of the external differential quantum efficiency are related as^[33]1ηext=2αwLηiln(1/R1R2)+1ηi,(5)with the external quantum efficiency ηext defined as ηext=ηseNħω,(6)where L is the cavity length, N is the active region stage number, ħ is the reduced Planck constant, ω is the angular frequency, and R1 and R2 are the reflectivities of two cavity surfaces, which are equal when neither is coated. The calculation using the Fresnel formula is as follows: R1=R2=(n−1n+1)2,(7)where n=3.16 is the effective refractive index of the waveguide mode and the refractive index of the air is unity. A linear fit for the quantum efficiency is shown in Fig. 5(a), where the fitting result shows that the internal quantum efficiency ηi=71.7%, and the waveguide loss αw=0.89cm−1, consistent with the simulation results in Sec. 2.

By analyzing the threshold current densities and slope efficiencies obtained from L-I-V measurements under pulsed modes across temperatures ranging from 288 to 328 K, we fitted two characteristic temperatures for the device. The temperature characteristics are shown in Fig. 5(b) for the threshold current density and the slope efficiency in pulsed modes. The characteristic temperatures are defined and fitted as^[33]Jth=J0exp(T/T0),(8)ηs=η0exp(−T/T1),(9)where T is the heat sink temperature. The fitting result shows the characteristic temperatures T0 and T1 for the threshold current density and slope efficiency being 204 and 384 K, respectively. Compared to quantum cascade lasers reported in the literature with a similar profile^[16], this device had a higher characteristic temperature; therefore, its threshold current and slope efficiency were less temperature sensitive, resulting in better performance when operating in CW modes at room temperature.

Figure 6 shows the device lasing spectrum at different operating currents at room temperature (298 K). Just above the threshold current (1.10 A), a single peak was measured at 5.061 µm. As the output power increased, the spectrum broadened to multiple longitudinal modes, and the centroid wavelength exhibited a redshift. When operating at a high power of 4 W, the centroid wavelength was 5.142 µm.

In order to obtain the “far-field” divergence angle of the device, we set up a related far-field testing optical path. We used the Vigo photonics LabM-I-10.6 type HgCdTe infrared detection module fixed on a rotating platform to scan the divergence angle in the horizontal direction (perpendicular to the epitaxial growth direction) and the vertical direction (along the epitaxial growth direction), respectively, and obtained the beam intensity distributions in both directions, as shown in Figs. 7(a) and 7(b). The “near-field” measurement results are shown in Fig. 7(c). The details are shown in Table 3. As the driving current increases, the horizontal direction far-field divergence angle increases by 9.9%, and the vertical direction far-field divergence angle increases by 10.5%. The beam quality factor M2 for different axes is calculated from Mx2=πw0xθxλ,(10)My2=πw0yθyλ,(11)where w0x and w0y are the beam waists of the horizontal and vertical directions, and θx and θy are the beam divergence half-angles of the two directions. Due to nonlinear effects, the beam quality of QCLs tends to degrade with increasing output power^[34]. When the drive current increased from 1.40 to 2.80 A, the M2 of the vertical direction increased from 1.27 to 1.48, and the M2 of the horizontal direction increased from 1.13 to 1.43. For the near-field spot profile, as the output power increased, the near-field spot size slightly increased. Although the degradation of the beam quality was observed with the increase of the laser beam power, the quality factor M2 remained below 1.5, and no side lobe was observed in the near- or far-field profile, so single transverse mode lasing was verified.

A high-power (>4W), single transverse mode QCL operating in CW modes at room temperature (298 K) has been demonstrated. By designing a low-leakage-current quantum cascade structure with AlAs quantum barriers, optimizing the MBE growth conditions for the surface and interface quality, employing epitaxial regrowth InP thermal dissipation materials and epi-down packaging, we successfully achieved the highest CW power 5 µm QCLs at room temperature (298 K), with a threshold current density and a slope efficiency of 2.23kA/cm2 and 2.73 W/A, respectively. The device exhibited a high characteristic temperature with T0=204K and T1=384K. At a 4 W high-power output, it maintained a single transverse mode and high beam quality (M2<1.5). The comparison between the power in pulsed modes and CW modes confirmed that the heat management would be the next key step for further improving QCL quality. Our result confirmed that it was sufficient to achieve good beam quality and high power with a narrow ridge design and verified a promising recipe for mass manufacturing high-brightness QCLs for academic and industrial applications.