1. Introduction

The LWIR (7-14µm) region contains an important atmospheric windows [1], and most of the molecules have strong absorption bands in this region including CH₄, SO_x, NO_x, NH₃, etc., which is very suitable for trace gas analysis [2 –4]. Quantum cascade lasers (QCLs) is a new type of laser source based on electron transition radiation between intersubbands, and its emission wavelength can cover the middle and long-wave infrared (MIR&LWIR) and even terahertz through bandgap engineering [5]. Owing to QCLs’ advantages of portability, high stability, continuous-wave (CW) operation at room temperature, wide tuning performance and high power, QCL gradually replaces other lasers and becomes an ideal MIR&LWIR light source for trace gas sensing. For commercial application, gas sensing systems usually need to minimize power dissipation under the condition of ensuring device size and performance, in other words, improve device high wall-plug efficiency (WPE) [6]. Higher WPE means less heat generation in the active region, better device stability, and also longer lifetime.

The WPE can be expressed as ηwpe = η_i ⋅ η_o ⋅ η_v ⋅ η_e, where η_i, η_o, η_v and η_e are internal quantum efficiency, optical efficiency, voltage efficiency and electrical efficiency respectively [7,8]. Generally, the longer wavelength of QCL, the smaller WPE is, especially for LWIR-QCL (7-14µm) and even far-infrared (14-30µm), which is mainly due to the following reasons [9]: Firstly, the free carrier absorption loss is proportional to the square of the wavelength, which will increase α_w and reduce η_o. Secondly, as the wavelength increases, the ratio of photon energy and voltage defect Δ_inj (defined as the difference in energy between the lower energy level of the current period and the upper energy level of the next period) decreases, and then η_v decreases. Thirdly, in order to reduce α_w and improve Γ, lower and thicker waveguides are introduced. However, thick waveguides are not friendly to heat dissipation, resulting in increased thermal accumulation in the core layer, increased thermal backfilling current, reduced slope efficiency and other degradation phenomena of device performance. From the above, lower Δ_inj, α_w and thermal optimization contribute to a higher η_wpe of LWIR [1].

In recent years, a lot of efforts have been made to improve the performance of LWIR devices, but the focus has been on the wavelength of 8-12µm [10 –13] and even >12µm [14,15]. However, the wavelength of 7-8µm region was also important for trace gas sensing, where strong molecular absorption peaks of CH₄ and N₂O are concentrated [2,4]. It is worth noticing that CH₄ molecular absorption spectral lines in the MIR are only concentrated in the range of 3.2-3.6µm and 7.2-8.1µm, the absorption spectral lines at 7.7µm are clear and easy to be recognized [3], which is very important for CH₄ trace gas detection. QCLs emitting in the 7.3 to 7.85 µm region were reported in 2010 [16,17], but none of these devices exceeded 1W in power and the maximum WPE was only 5%. Besides, Richard Maulini et al. introduced a model for backfilling of the lower laser level in 2011 [1], they believed that the traditional model strongly overestimates backfilling at low voltage defect, predicting an optimum LWIR voltage defect of ∼100 meV for achieving maximum WPE at room temperature, which was applied in a 7.1µm QCL, giving a maximum CW output power of 1.38 W and maximum WPE of 10.0%. Up to now, the highest WPE in CW operation in 7-8µm is obtained by a high differential gain QCL with a wavelength of 8µm reported by Wenjia Zhou and Quanyong Lu et al [18] in 2019, which reported a high WPE of 20.4% in pulsed mode operation and 12.8% in CW operation, and a CW output power of 2.0 W is achieved from one-side of a single-element QCL using epi-layer down bonding, and the front and back facets are anti-reflectance(AR) and high-reflectance(HR) coated respectively. Furthermore, there were some novelty works reported by Botez et al. They introduced active structure design called step-taper resonant extraction (STA-RE type) [19 –21] and obtained very high internal efficiency η_i of 81% at 7.79µm and 86% at 8.8µm. The single-facet power of 7.79µm device was up to 1 W in CW, at a heatsink temperature of 14 °C, and the maximum achieved single-facet CW wall-plug efficiency is 6%. Besides, the characteristic temperatures T₀ and T₁ were obtained to 228 K and 529 K, respectively, which were very high when compared with conventional QCL. However, the material epitaxial process of this structure is quite complex.

Therefore, a watt-level QCL emitting at 7.7 µm with a high WPE and operating in CW mode at 293K is reported in this paper. The strain-balanced design was used to reduce leakage current. By optimizing the number of active region stages and waveguide structure, we improve the Γ and reduce the α_w successfully. In addition, buried ridge structure and diamond heatsink are adopted to improve the heat dissipation performance of the device. These measures all contributing to a high WPE.

2. Laser active region, waveguide design and experiment

The λ ∼ 7.7 µm high wall-plug efficiency watt-level CW QCL operating at room temperature reported here was based on a slightly-diagonal bound-to-continuum(BTC) design, and the strain-balanced material In_0.47Al_0.53As/In_0.58Ga_0.42As with ∼0.36% strain were used in the active region to increase the conduction band offset to ∼608 meV and prevent carriers’ from escaping into the continuum, which was different with the previous reported structure emitting at 8µm [22]. The BTC active design has several advantages: high fault tolerance for material growth, high temperature performance, broadband gain [5], large oscillator strength and good lifetime ratio of upper and lower energy level contributing to a large internal quantum efficiency η_i. Therefore, the design applied here is particularly suitable for LWIR devices.

The active region of the laser was grown by molecular beam epitaxial (MBE) equipment, and the specific thickness of the In_0.47Al_0.53As/In_0.58Ga_0.42As had been modified based on the Ref18 published previously with a lattice-matched structure emitting at 8µm. The layer sequence of one period of structure in nm, left to right and starting from the injection barrier, was 4.1/1.8/0.92/5.3/0.92/4.9/1.0/4.1/1.8/3.4/1.74/2.94/1.94/2.83/2.86/2.73. The In_0.47Al_0.53As layers are in bold font, and the In_0.58Ga_0.42As wells are in normal font, and the underlined layers are doped to n=1.5 E17cm⁻³.The conduction band diagram and wave-functions of different energy levels of the active region under an applied electric field of 53KV/cm are showed in Fig. 1. Introduction of low strain increases the conduction band offset and allowed higher ΔE_C5∼155meV and ΔE₅₄∼70meV, which effectively suppressed the leakage current and then reduced the threshold current density Jth. A slightly-diagonal transition was used to reduce the overlap of the wave functions between the upper and lower levels, which effectively increased the carrier lifetime of the upper maser level and reduced the probability of carriers’ thermal backfilling contributing to high η_i and temperature performance. The laser transition matrix element, transition lifetime between the upper laser level 4 and the lower laser level 3, the upper laser level lifetime τ₄ and the lower laser level τ₃ were calculated to be Z₄₃=1.7nm, τ₄₃=2.97ps, τ₄=1.18ps, τ₃=0.18ps, respectively. The ratio of τ3 and τ₄₃ was calculated to be as lower as 0.061. Besides, a low voltage defect of Δ_inj = E₃-E_g= 96meV was applied here to ensure a higher η_wpe, which was very close to the optimal voltage defect value of Δ_inj∼100meV in LWIR region reported in [1]. Because high differential gain of active region can compensate for an increase in thermal backflow due to a low Δ_inj [17].

 

Fig. 1. Band structure and wave functions of relevant energy levels of a bound to continuum and low voltage defect low strain-balanced In_0.47Al_0.53As/In_0.58Ga_0.42As LWIR QCL active region design for emitting at λ ∼7.7µm under an applied electric field of 53 kV/cm. The upper (level 4) and lower laser levels (level 5) are shown in red and blue, respectively. The ground state of the injector (level g) is shown in cyan. The red arrow marks the radiative transition.

Download Full Size | PDF

Detailed layer structures from the InP substrate (Si, ∼3 E17cm⁻³) consists of a 4.5µm InP buffer layer (Si, ∼3E16 cm⁻³), a 50-stage laser core (Si, ∼1.5E17 cm⁻³) sandwiched between two 300 nm InGaAs (Si, ∼4E16 cm⁻³) layer to improve the optical confinement of active regions, a 4.5 µm InP cladding layer (Si, ∼3E16 cm⁻³), a 0.85 µm graded doped InP layer (Si, ∼1.5E18 cm⁻³) and 0.15µm highly doped InP cap layer (Si, ∼5E18cm⁻³). InP buffer layer and cladding layer are thicker than previously reported references for λ between 7∼8µm, which contribute to a lower α_w and higher Γ. Two InGaAs layers also help to improve the Γ. The metal organic chemical vapor deposition (MOCVD) is used for all layers’ growth except for the active region and two InGaAs layer, which are grown by MBE.

3. Optical and thermal simulation

An effective way to improve WPE is to increase the number of active stages(N_S). Principally, it can reduce the threshold current density J_th by increasing the modal confinement factor and decreasing the waveguide loss, according to the expression of the $ {J_{th}} = {J_{t{h_0}}} + {J_{t{h_{leak}}}} + {J_{t{h_{bf}}}}$, $\; {J_{t{h_0}}} = \frac{{({{\alpha_w} + {\alpha_m}} )}}{{\textrm{g}\Gamma }}$, where ${J_{t{h_{leak}}}}$ and ${J_{t{h_{bf}}}}$ are the leakage current density and backfilling current density at threshold, respectively [8]. Qualitatively speaking, increasing N_S is equal to increasing the thickness of the active region, which means that the light is more restricted to the core layer, and the free carrier absorption outside of the core layer reduced. As a result, the α_w is reduced and less gain is required to for lasing [6]. Therefore, we numerically simulated the influence of N_S on α_w and Γ, the results are shown in Fig. 2. Results confirmed that with the increase of N_S, the α_w decreases and the Γ increases.

 

Fig. 2. The calculated waveguide loss α_w and confinement factor Γ of the 7.7µm QCL as a function of the numbers of QCL-stages from 30 to 70.

Download Full Size | PDF

However, the active region of the device is consisted of a multi-layer quantum well/barrier pairs, which leads to a significantly lower thermal conductivity in the epitaxial direction than the bulk material. In other words, there is a great heat accumulation in the core layer when the laser works. The increase of N_S aggravates the heat accumulation and increases the temperature difference between the core and heatsink(ΔT), which in turn worsens the performance of the device, such as the decrease of WPE and the service lifetime. So it is important to balance the advantages and disadvantages of increasing the N_S and find a suitable value.

To this end, we also numerically simulated the maximum temperature(T_max) and average temperature(T_ave) of the core layer change with N_S, the results were shown in Fig. 3. And the buried ridge waveguide structure introduced in this simulation will be described later. Combined with the information obtained in Fig. 2, we can see that α_w changes slow and ΔT increases more when N_S increases to more than 60. We finally chose the N_S = 50, and a low α_w of 1.15cm⁻¹, high Γ of 66.5% and ΔT of 77.3k were calculated, which contributed to an extremely low J_th and good high temperature performance and were showed in the next part.

 

Fig. 3. (a) The detailed structure of the device and temperature distribution of a buried ridge QCL epi-down bonded on a diamond heatsink for CW operation. (b) The maximum (Tmax) and average (Tave) core temperature of the buried ridge QCL under CW operation as a function of Ns (T_heatsink=300K).

Download Full Size | PDF

4. Fabrication, packaging and characterization of the device

The ridge-type waveguide structure of the laser restricts almost all the heats generated in the active region to the inside of the ridge. The increase of ridge width is not conducive to the lateral dissipation of the heat in the active region [23]. However, a narrow ridge will reduce the device power and increase the far-field divergence angle. Considering the above reasons and combining with previous experiences, we chose an 8µm ridge width for the device. A buried heterogeneous(BH) structure was introduced into this device: Semi-insulated InP: Fe with high thermal conductivity was grown on both sides of the ridge through selective epitaxy by MOCVD, replacing SiO₂ insulation layer with extremely low thermal conductivity. As a result, the BH structure greatly increased the lateral heat dissipation and reduced the temperature of the active region, which was very beneficial for the fabrication of high-power CW-QCL operating at room temperature.

To further improve the heat dissipation and achieve higher power output, the device was epi-side-down mounted on the diamond heatsink with indium and cleaved into chips of 4mm, and the back facet was high-reflectance(HR) coated with Al₂O₃/Ti/Au/Ti/Al₂O₃. After installing the thermistor, the laser was fixed on the water cooling platform with thermoelectric cooler (TEC) for temperature control, and the output light is directly measured by the pyroelectric power meter (Molectron, EPM1000) without lens collimation. The power-current-voltage (P-I-V) characteristics and WPE as function of current of a 4mm×8µm (cavity length × ridge width) HR coated QCL at different temperature in continuous-wave and pulsed operation are shown in Fig. 4(a) and Fig. 4(c), respectively. The maximum WPE and the maximum peak power in pulsed mode was obtained to 13.5% and 2.3W at 293K. We also analyzed the characteristic temperature of the device based on the pulsed P-I-V test results. As shown in Fig. 4(b), the characteristic temperature of T₀ and T₁ were 151K and 290K, respectively. Furthermore, we also simulated the junction temperature of the device in CW mode operation, the junction temperature was calculated to be 381K, with the heatsink temperature of 303K and power injection of 13.12W (corresponding to the maximum power injection of the device in CW operation under 303K). It's worth noting that our pulse power supply does not match well with the device, so the optical power measured under the set duty cycle is lower than it actually is.

 

Fig. 4. (a) The P-I-V and WPE experimental results at different temperature and (b). characteristic temperatures fitting of a cavity length 4mm, 8µm ridge wide, HR-coated, epi-down device in pulsed operation with duty cycle of 2%. Inset in (a): pulsed emitting spectrum. (c). The P-I-V and WPE experimental results at different temperature in CW operation. Inset in (c): CW laser spectrum at a current of 0.6 A at 293K under water and TEC cooling. (d) The P-I results of uncoated device at different cavity length in pulsed mode. Inset in (d): Inverse external quantum efficiency as a function of cavity length L.

Download Full Size | PDF

To prevent the device from burning out, we did not add the current to the roll-over. A maximum CW optical output power of 1.17W, a threshold current density (Jth) of only 1.09 kA/cm² were measured at 293 K, and the maximum power still reached 950mW at 313K. The maximum WPE of 9.08% was also measured at 293K under a current of 0.85A and a little lower than that reported in Ref 1, which was mainly due to the fact that we did not have the process of proper AR coatings in the front cavity facet (Al₂O₃/Ge AR coatings we have mastered may absorb light of long-wave infrared, and are prone to disaster under high output power), and in order to ensure the optical power efficiency, the cavity length of the device could not be further increased. Besides, emission spectrum of QCL at 293K measured through Fourier transform infrared (FTIR) spectrometer was shown in the inset of Fig. 4(a), giving a central wavelength 7.7 µm at a current of 0.6 A. There were no AR coatings for the device, so the low threshold current was partly related to the small mirror loss α_m∼0.42cm⁻¹. Besides, P-I measurement results of uncoated devices with cavity length of 4mm,6mm,8mm were shown in Fig. 4. (d), giving the internal efficiency η_i of 64.1% and waveguide loss α_w of 1.18cm⁻¹ as inset of Fig. 4. (d) shown, which contributed to the low J_th.

By constructing the external-cavity quantum cascade laser (EC-QCL) system of Littrow, we obtained the spectra and peak output power under different blazed grating angles as shown in the Fig. 5., the device was fabricated to 4mm and emitting on both sides with only the front facet AR(Al₂O₃/Ge) coated for higher external cavity feedback. The results showed a broad tuning range from 1396.2cm⁻¹ (λ=7.16µm) to 1226cm⁻¹ (λ=8.16µm) covering all the CH₄ molecular absorption spectral lines in 7.2-8.1µm. Good tuning performance of the EC-QCL comes from the broad gain of BTC active structure, good optical collimation, the high collection efficiency of lens and high diffraction efficiency of blazed gratings. The maximum peak output power collected from the back facet (collimated through lens) of the 4mm external cavity tunable laser was 221mW corresponding to an emitting wavelength of 7.67µm (1302.98cm⁻¹), under the duty cycle of 2%(10kHz,2us) and current of 600mA. The average output power was 505mW at equal emitting wavelength when the current increased to 800mA.

 

Fig. 5. Tuning behavior and peak power of the EC-QCL of the device. Measurements were taken at room temperature with a duty cycle of 2% (10kHz,2 us) at 0.6A.

Download Full Size | PDF

The beam picture of the laser at pumping current slightly above threshold is shown in shown in Fig. 6. The picture was taken with a pyroelectric camera placed 25cm away from the laser, with lens(Lightpath,390037IR1) collimation. The beam consisted of a pure zero order mode TM₀₀ even when the current was increased to 1 A, which confirmed that the 8 µm-wide of the ridge was suitable to prevent higher-order transverse modes from lasing.

 

Fig. 6. The beam picture of the QCL measured in pulsed mode at room temperature under a current of 0.5 A with a duty cycle of 8%(40KHz 2us), The horizontal and vertical beam sizes are 2.71mm and 2.26mm, respectively.

Download Full Size | PDF

5. Conclusion

In conclusion, we demonstrate 9.08% WPE and 1.17W output power of a 7.7 µm HR-coated QCL in CW operation at 293K. The maximum WPE and the maximum peak power in pulsed mode at 293K were measured to 13.5% and 2.3W, respectively. The device used bound-to-continuum design, strain-balanced active region and low voltage of 96meV for achieving higher WPE. Ns of 50 and 4.5µm of both InP cladding and buffer layer had also been utilized, giving a low α_w of 1.18cm⁻¹ contributing to the high WPE and low J_th of 1.09 kA/cm² in CW mode. We believe that the maximum WPE in CW mode can exceed the previous result (10%) reported in Ref 1 by more proper AR coatings on the front facet, and the maximum output power will be higher by increasing the cavity length. Finally, a broad EC-QCL tuning range from 7.16µm to 8.16µm and peak power of 221mW were reported, which were friendly to trace gas sensing for CH₄ and N₂O.