3 W Continuous-Wave Room Temperature Quantum Cascade Laser Grown by Metal-Organic Chemical Vapor Deposition

Abstract

In this article, we report a high-performance λ ~ 4.6 μm quantum cascade laser grown by metal-organic chemical vapor deposition. Continuous wave power of 3 W was obtained from an 8 mm-long and 7.5 μm wide coated laser at 285 K. The maximum pulsed and CW wall-plug efficiency reached 15.4% and 10.4%, respectively. The device performance shows the great potential of metal-organic chemical vapor deposition growth for quantum cascade material and devices.

1. Introduction

In the mid-infrared region, quantum cascade lasers [1] (QCL) has proven to be a powerful and versatile light source owing to their advantages in practicality including high power, wide tunability, small size and easy operation. In recent years, great efforts have been devoted to developing device performance, in terms of low power consumption [2,3,4], high efficiency [5,6,7,8,9] and high power [10,11]. The increasing demands of infrared countermeasures and the rapid development in long-range free-space communication [12,13] have served as major drivers to improve the power of quantum cascade lasers.

Since the first demonstration of QCL, molecular beam epitaxy (MBE) has mainly been used for the growth of QCL cores [14,15]. The precise control of layer thickness and the sharp interface in MBE are essential for the growth of ultrathin QCL layers. Previous efforts have focused on practically important λ ~ 4.6 μm QCLs to avoid atmospheric absorption. Room temperature watt-level power has been achieved so far. In particular, more than 3 W output power from a single facet was independently presented by two groups [16,17,18]. However, the cost of high-power QCLs chips is several orders of magnitude higher than that of near-infrared semiconductor laser chips, in part due to the high cost of the molecular beam epitaxy process [19]. Metal-organic chemical vapor deposition (MOCVD) is a well-established technique to grow an optoelectronic structure in the industrial field. It has the advantage of excellent uniformity and repeatability, but still poses challenges with respect to the growth of QCL cores. Significant research has been made to improve the performance of MOCVD-gown QCLs in the past two decades [20,21,22,23]. In 2010, Yu Yao reported a continuum-to-continuum design. The reported room temperature power and WPE reached 5 W and 23% in pulsed mode [24], which are comparable to the state-of-the-art MBE-grown QCLs. However, the CW power dropped rapidly to only 1 W at room temperature due to severe carrier leakage and backfilling. Most recently, record high power of 2.6 W emitting at λ ~5 μm at 288 K have been demonstrated [7] by employing a variety of active region materials of different compositions to suppress carrier leakage. However, the promotion of device performance comes at the cost of increasing growth difficulty. The adopted 11 different compositions place higher requirements on material growth, which also limits the efficiency of epitaxial growth. Despite the aforementioned achievements, the power of MOCVD-grown QCLs is still less than MBE-grown QCLs. This is mainly because the MBE system adopts an ultra-high vacuum environment, in which the interface switching speed is fast. In addition, the concentration of background impurities is low. In contrast, in the MOCVD system, the source gas flow is inevitably delayed during the interface switching process [25]. A smoother interface and lower background impurities contribute to the low interface roughness and low optical loss in MBE-grown QCLs [17]. However, we believe it is possible to obtain considerable high-quality material for MOCVD-grown QCLs. We have recently achieved high-quality long-wave MOCVD-grown QCL by carefully optimizing the growth conditions, particularly the growth interface [6]. QCLs emitting at λ ~8.5 μm with a record high efficiency of 7.1% and 1 W CW power at room temperature were demonstrated. Here, we follow our previous method and extend the wavelength to the mid-wavelength. Combined with the dedicated energy band design, it is reasonable to speculate that the performance of mid-wave MOCVD-grown QCLs can be further improved.

In this paper, we report the growth and fabrication of 4.6 μm strain-balanced In_0.66Ga_0.34As/In_0.36Al_0.64As using low-pressure MOCVD. Based on the slightly diagonal single phonon resonance design, a high-quality quantum cascade structure was obtained. An 8 mm-long, 7.5 μm-wide device was capable of delivering 5.8 W peak power in pulse mode (500 ns/2%) at 285 K. The threshold current density, slope efficiency and maximum WPE were 1.5 kA/cm², 4.35 W/A and 15.4%, respectively. The CW power of the device reaches 3 W at 285 K with a maximum WPE of 10.4%.

2. Materials and Methods

The laser core consisted of 40 periods of strain-balanced InGaAs/InAlAs quantum well and barrier pairs based on the single phonon resonance (SPR) design similar to the structure in Ref [26]. For the reference design, the upper lasing state is highly localized in the first thin well, which enhances the coupling between the injector region and the upper lasing state [14,27]. In addition, the optical transition is slightly diagonal, which increases the lifetime of the upper state. Figure 1 shows a comparison between the reference design and the one optimized. The thickness of the active region was slightly modified to target the wavelength of 4.5μm. In addition, the compositions of Al in InAlAs and Ga in InGaAs were changed to 0.64 and 0.34, respectively. As a result, the conduction band offset was increased from 708 meV to 782 meV. Therefore, thermally activated leakage of electrons from the upper laser level into the continuum can be suppressed. Another notable feature is the increased energy difference between the upper lasing state (E₄) and the excited state (E₅). The calculated energy difference E₅₄ is increased from 70 to 91 meV.

The QCL structure presented in this article was grown on a 2-inch n-InP substrate (S, 2E18) using a low-pressure (100 mbar) MOCVD with a close-coupled showerhead (3 × 2″) reactor. The group III precursors were trimethylindium (TMIn), trimethylgallium (TMGa), and trimethylaluminum (TMAl). The group V precursors were purified arsine (AsH₃) and phosphine (PH₃). Silane (SiH₄, 0.02% in H₂) was used as the n-type dopant. The total flow rate was 8 L/min. The growth temperature was read using a thermocouple mounted under the graphite disk. Before growth, the InP(S, 2E18) substrates were loaded into the chamber and deoxidized at 760 °C for 5 min in PH₃ ambient. The growth was initiated by the deposition of a 500 nm InP buffer layer followed by a 3 μm n-InP (4E16) lower waveguide at 700 °C. Then, the temperature was increased to 720 °C to grow the InGaAs/InAlAs laser core region. There were no growth interrupts at the interfaces of InGaAs-to-InAlAs and InAlAs-to-InGaAs. The TMIn flow rate was kept constant during InGaAs and InAlAs growth. Then, the temperature was decreased to 700 °C to grow a 3 μm InP upper waveguide (4E16) which was covered by a 1 μm InP cap layer (2E18).

Once the wafer growth was achieved, the material was characterized by high-resolution X-ray diffraction (HRXRD) and atomic force microscopy (AFM) to investigate the material quality. Figure 2a compares the measured and simulated HRXRD results. The experimental result is almost fully in line with the simulated one. Multiple high-order satellite peaks can be clearly observed, with extremely sharp and narrow FWHM of about 15 arc seconds. The observation of multiple satellite peaks at HRXRD indicates excellent in-plane uniformity in terms of layer thickness and chemical composition. To characterize the surface morphology of QCL, one piece of the wafer was etched by HCl to remove the upper InP waveguide. Figure 2b shows the surface image of the active core characterized by AFM (5 μm × 5 μm scan in contact mode). It can be seen that for such hundreds of strained quantum well/barrier pairs, step-flow growth is still achieved. The calculated surface root-mean-square (RMS) is only 0.15 nm. The average distance between different steps is 0.25 μm.

The fabrication process of QCLs is as follows. One piece of the wafer was processed into a double-channel geometry with a ridge of 16 μm to characterize the intrinsic physical parameters of the wafer, for example, waveguide loss and modal differential gain. Another piece of the wafer was processed into a buried hetero-structure (BH) configuration with a ridge width of 7.5 μm as shown in the Appendix A, Figure A1. Firstly, the wafer was patterned into stripes by using a 300 nm-thick SiO₂ mask and wet-etched through the active region into a double-channel geometry. The etchant is HBr:HNO₃:H₂O (1:1:10) which is nonselective and isotropic. After etching and rinsing, Fe-doped semi-insulating InP was selective-area regrown by MOCVD to planarize the channels. Then a SiO₂ mask was removed using a buffered oxide etch (BOE), and a new 300 nm-thick SiO₂ layer was deposited by plasma-enhanced chemical vapor deposition as an electrical isolation layer. The current injection windows were opened on the top of the ridge by lithography. Besides, top contact was made, including a Ti/Au layer by e-beam evaporation and a 5 μm-thick Au layer by electroplating. The backside of the wafer was thinned down to 120 μm and polished. An AuGeNi/Au layer was then evaporated as the bottom contact. The processed wafer was annealed at 370 °C and cleaved into 8 mm bars. A high-reflectivity coating was applied on the back facet with the front facet left uncoated. For packaging, the chips were soldered epi-down on diamond submounts to improve the heat removal efficiency. Then the submounts with chips were mounted on the copper heat sinks with indium solder followed by wire bonding.

3. Results and Discussion

After device fabrication, testing was conducted by fixing the laser on a water-cooling platform with a chip facet close to a calibrated thermopile detector. For pulsed measurements, the voltage and pulsed width were measured using a voltage probe. The peak power in pulsed operation (pulse width 500 ns, repetition 40 kHz) was calculated by dividing average power by duty cycle (2%). The temperature was monitored by a thermistor and controlled by a thermoelectric cooler (TEC). Figure 3a shows the pulsed power-current-voltage (PIV) curves of QCL devices with 16-μm-wide and 2-, 4-, and 6-mm-long cavities. The heat sink temperature is kept at 295 K. Figure 3b shows the threshold current density dependence on mirror loss and inverse external quantum efficiency dependence on inverse mirror loss. The internal quantum efficiency, waveguide loss, transparency current density and differential gain were determined to be 60%, 0.66 cm⁻¹, 1.5 kA/cm² and 7.24 cm/kA. The waveguide loss is three times smaller than in Ref [7], which means a loss in slope efficiency can be negligible when the length is increased. Therefore, we can use longer cavities to achieve better performance. However, this waveguide loss is slightly larger than that of the state-of-the-art MBE-grown QCLs [8], implying that there is still room for improvement in material quality. In addition, the modal differential gain is 2.7 times larger than that in Ref [9], which features a diagonal design. The modal gain strongly depends on the overlap between the upper laser level and the lower laser level. In diagonal designs, increasing the lifetime of the upper laser level comes at the cost of lower differential modal gain as in Ref [9]. The slightly diagonal design in this paper can increase the lifetime of the upper laser level without much compromise on the differential gain.

Figure 4 shows the CW (solid lines) and pulsed (dash lines) PIV and WPE results at 285 K. The AR-HR coated device emits to 3 W (5.8 W) with threshold current density J_th = 1.76 kA/cm² (1.5 kA/cm²), slope efficiency η_s = 4 W/A (4.35 W/A), and WPE = 10.4% (15.4%), for CW (pulsed mode) operation. The L-I curves (red lines) show kink-free straight lines up to the saturation current, which indicates no mode hopping occurred. The inset shows the laser spectrum of the device at room temperature. The emission spectrum centered at 2160 cm⁻¹ (4.6 μm) was obtained by Fourier transformed infrared spectrometer (FTIR) in rapid scan mode with a resolution of 0.25 cm⁻¹.

Figure 5 shows the pulsed mode PIV results at different temperatures. As the temperature increases, threshold current density increases and slope efficiency decreases as expected. An experimental relationship can be established using the following equations:

$$\begin{gathered} J_{th}=J_0\exp (T/T_0) \\ {\eta }_s={\eta }_0\exp (-T/T_1) \end{gathered}$$

where T is the heat sink temperature. The characteristic temperatures T₀ and T₁ in the pulsed operation are identified to be 240 K and 264 K, respectively. Compared with previously reported QCLs in Ref [26], T₀ and T₁ are increased by 50 K and 124 K, respectively. We attribute the increase to the larger spacing between the upper laser level and the excited state, and to the continuum. The high T₀ indicates that the thermal backfilling is significantly suppressed. Furthermore, the T₀ of this device is comparable with that of the previously reported muti-composition design QCLs in Ref [7], which features a higher E₅₄ of ~100 meV. The relatively low T₁ can be improved by reducing the overlap between the upper state E₄ and the excited state E₅, as described in Ref [9,27].

Figure 6a shows the beam image at 1.9 A in CW operation. The transverse mode was obtained by placing a pyroelectric camera 1.5 m away from the collimated chip. The beam shape remains basically unchanged with some beam steering at the maximum current density. This beam steering results from the onset of the first-order lateral mode at a high injection current. Figure 6b shows the far field pattern at different currents. The far-field pattern was obtained by fixing the device on a motorized rotation stage with 0.1° resolution about 40 cm away from the mercury cadmium telluride detector. Furthermore, beam steering was suppressed thanks to the temperature-insensitive design [28]. In addition, we have measured the beam M² at 1.5 A in CW mode as shown in Appendix B,

Table A1. Relation between the width of light beam and distance.

Z/mm	Dx/mm	Dy/mm
300	3.04	2.57
500	3.24	3.02
700	3.57	3.55
1100	4.60	4.92
1300	5.25	5.84
1500	6.02	6.66

. The calculated beam M² in the x-axis and y-axis direction are 1.146 and 1.045.

In summary, we demonstrated high power 4.6 μm QCL using a single phonon resonance design. Room temperature CW power up to 3 W with a maximum WPE of 10.4% was obtained from an 8 mm-long, 7.5 μm-wide, AR-HR coated device. The device shows a characteristic that the threshold current is insensitive to temperature changes. In addition, the beam shape remains basically unchanged until the saturation current.