High-Brightness 1940 nm Gallium Antimonide Diode Lasers with External-Cavity Spectral and Polarization Beam Combining

Abstract

In this paper, we present a beam-combining technique to boost GaSb-based 1940 nm diode laser output power through spectral beam combining (SBC), spatial beam combining, and polarization beam combining (PBC). Four spectral beam-combining (SBC) configurations were developed using commercially available standard bars. The four SBC configurations were paired to perform PBC after spatial beam combining. The total output power of the 1940 nm laser reached 23.4 W, with the beam qualities of the combined beam achieving 10.6 in the slow axis and 11 in the fast axis. The brightness of the combined laser reached 5.4 MW·cm⁻²·sr⁻¹.

1. Introduction

Lasers in the wavelength range of 1900 to 2300 nm have critical applications in many fields, including gas spectroscopy, medical diagnosis and treatment, material processing, and infrared countermeasures [1,2,3,4,5]. Diode lasers made of AlGaInAsSb/gallium antimonide (GaSb) quantum materials are well suited for this wavelength range and offer advantages because of their small size, long lifetime, and low cost. Their application, however, has been limited by low power and poor beam quality. Several research groups are optimizing the output power and divergence angle of GaSb-based single emitters and bars, including improvements in their growth process, laser diode structure, and heat dissipation. A continuous wave (CW) power of 1.96 W based on a GaSb-based broad-area (BA) waveguide single-chip laser was achieved in 2006 [6] and the fast-axis divergence was improved from 67° to 44°. The CW power of the bar and array reached 16.9 W in the same year [1]. In 2016, our team proposed a fishbone-shaped microstructure to enhance the brightness of the GaSb single emitter. We achieved a notable 55% decrease in lateral divergence, and the output power increased by almost 19% [7]. We demonstrated another method to improve the brightness based on the on-chip microstructure of the turbulent scattering cavity, and the single-chip output power exceeded 1.3 W [8] with a lateral far-field improvement of 45.6%. Overall, these advances are not yet sufficient to support practical applications. Therefore, the demand for higher power and brightness requires the use of beam combining.

Beam-combining techniques, which include both coherent and incoherent beam combining, are often employed to boost the laser brightness. Coherent beam combining with diode lasers at long wavelengths has rarely been reported because of the complex structure and low efficiency [9]. Incoherent beam-combining technology with a simple structure is the primary method used to achieve power scaling. Among the numerous incoherent combining techniques, polarization beam combining (PBC), wavelength combining, and spectral beam combining (SBC), which Daneu et al. proposed and demonstrated in 2000, are all methods that do not cause beam quality to degrade. SBC has achieved significant improvements because of the significantly improved power and high-quality beams. The output power of SBC can be increased in two primary ways: (1) by increasing the quantity of beam-combining emitters as much as possible, and (2) by improving the combining efficiency of SBC. In 2017, Haas et al. reported kilowatt-level SBC. Their experiment constructed an external cavity based on a thin-film filter. The total SBC incorporated 230 emitters [10]. In the same year, Meng et al. reported 970 nm SBC based on three mini-bar stacks, achieving an output power of 579.2 W. The measured beam quality M² factors were 18.7 in the fast axis and 14.7 in the slow axis [11]. In 2019, Lin et al. reported 940 nm SBC with an output power of 310 W. Their experimental setup incorporated eight commercial BA semiconductor laser arrays. They measured the fast-axis and slow-axis M² factors to be 13.55 and 10.27, respectively [12]. To improve the combining efficiency of SBC, in 2022, Zhang et al. reported SBC with a high combining efficiency of 95.5%. In their experiments, they implemented a polarization multiplexing external-cavity configuration and achieved a 6% improvement in beam-combining efficiency compared with conventional SBC [13].

In addition, some teams have investigated the optimization and quality-influencing factors of SBC in recent years. In 2018, Pierscińska systematically analyzed temperature monitoring and thermal processes in semiconductor lasers, clarified thermal degradation mechanisms, and proposed improved thermal detection technologies [14]. Also in 2018, Sun et al. found that, while lower output coupler (OC) reflectivity improves SBC stability, excessively low reflectivity causes system instability, accompanied by side lobes in the far field and reduced efficiency [15]. The smile effect in semiconductor lasers can induce spectral degradation in SBC. To address this, Zhu et al. proposed a Beam Transformation System (BTS) in 2014 to modify traditional SBC structures, mitigating both the smile effect and slow-axis divergence [16]. In 2024, Zhang et al. developed a coupling cavity resonance model based on cavity light feedback and multi-beam interference theory. Their study demonstrated that increasing the Fourier lens focal length and reducing the incident angle optimizes spectral locking, while high-line-density gratings compress spectral width [17]. For spectral linewidth compression, Zhang et al. in 2022 utilized a V-shaped blazed grating arrangement for multi-path superposition, reducing the SBC spectral width to one-third of its original value through third-order diffraction [18].

In this study, we used SBC, PBC, and spatial beam combining to combine four 1940 nm BA semiconductor laser arrays. The brightness, beam quality, and power efficiency of the system were investigated. Our results demonstrate that this method effectively increased the overall output power.

2. Methods

Figure 1 shows a schematic diagram of the experimental setup. This setup included four diode laser arrays (Bar1, Bar2, Bar3, and Bar4), and each array was collimated by a BTS and a slow-axis cylindrical mirror (SAC). The system also included five half-wave plates (HWPs), four Fourier transform lenses (FTLs), four 1st-order blazed gratings, four external-cavity output couplers (OC), five prisms, and PBC. To preserve the diffraction efficiency of the grating, we inserted an HWP (not shown in Figure 1) behind each collimated bar. The BTS consisted of a cylindrical lens and a Dove prism array that matched the period of the emitters in the laser array. The lasers emitted from the laser diodes were collimated by a cylindrical lens in the fast axis, and then the fast and slow axes were swapped by the Dove prism array so as to reduce the influence of the smile effect on the SBC.

To achieve stable wavelength locking, we coated the front facet of each emitter with an anti-reflection (AR) coating of less than 1% and coated the rear facet with a high-reflection (HR) coating of greater than 98%. We placed the bar and grating on the front and rear focal planes of the FTL. The collimated beams, which were parallel to each other, converged on the grating at different angles. We placed the OC, which provided partial feedback, in the diffraction direction of the grating. The rear facet of the laser, FTL, grating, and OC collectively constituted the external cavity of the SBC system. Each emitter of the bar was forced to operate at a specific wavelength that followed the diffraction equation of the grating. The lasers emitted from the bar were diffracted at the same angle, combined, and then output through the OC. The wavelength of each element ${\lambda }_i$ can be calculated as follows:

$${\lambda }_i=\mathsf{\Lambda }\ast \left\{\sin \left({\theta }_{\mathrm{littrow}}\right)+\sin \left[{\theta }_{\mathrm{littrow}}+\arctan \left({\displaystyle \frac{i\cdot p}{f}}\right)\right]\right\},$$

(1)

where f is the focal length of the FTL, Λ is the grating period, θ_littrow is the Littrow angle of the grating in use, and p is the distance between adjacent emitters.

Based on commercial 1940 nm bars, we established four SBC configurations (SBC bar1, SBC bar2, SBC bar3, and SBC bar4) and divided them into two groups (SBC1 and SBC2). SBC1 included SBC bars 1 and 2, and SBC2 included SBC bars 3 and 4. We spatially merged SBC bars 1, 2, 3, and 4 in the fast-axis direction. We used PBC to combine SBC1 and SBC2 to maintain optimal beam quality.

For the realistic setup of the diode arrays, we used GaSb-based commercial bars with a center wavelength of 1940 nm.

Table 1. Structure parameters of the 1940 nm bars.

Parameters	Value
Central wavelength/nm	1940
Chip width/mm	10
Cavity length/mm	3
Number of emitters	19
Fill factor (%)	20
Emitter width/μm	100
θ fast 90% power content/(°)	44
θ slow 90% power content/(°)	19
θ fast after collimation/(°)	0.25
θ slow after collimation/(°)	2.85
Front facet reflectivity	2.5%
Rear facet reflectivity	>95%
Degree of TE polarization	92%
Conversion efficiency (%)	>20%
Threshold current (ITH) (A)	<6
Operating current (IOP) (A)	<46
Operating voltage (VOP) (V)	<1.20

shows the corresponding parameters. The focal lengths of the BTS, SAC, and FTL were 0.034 mm, 4.25 mm, and 500 mm, respectively. It is worth noting that, in this system f_fast, f_slow ≪ f_FTL, so we can ignore the effect of the focal length of the BTS and SAC.

The thermal effect will have a certain impact on the overall system, including the redshift of the spectrum, the decrease in beam quality, and the overall power reduction. In order to reduce the influence of heat in the experiment, we set up a water-cooling system at the bottom of the lasers to control the experimental temperature and maintain the water-cooling temperature at 20 degrees Celsius to ensure the universality and repeatability of the experiment.

All these lenses were AR coated with an R < 0.5% reflectivity. We used a grating with a groove density of 600 lines/mm and a first-order diffraction efficiency of 92% at 1940 nm (Thorlabs). The prisms were right-angle prisms with K9 glass substrates. The hypotenuse was coated with a high-reflectance film with a reflectivity of 98%. All the HWPs used were zero-order HWPs with a transmittance of 99.3% at 1940 nm. The reflectivity of the output coupling mirror was 20%. The four laser bars were connected in series.

3. Experimental Results

Figure 2 illustrates the output power of the four bars of SBC1, SBC2, and PBC under free-running operation. At a current of 44A, each bar delivered nearly 10 W of output power, which resulted in a total output of 38.5 W for the four bars. As the current was increased from 8 A to 44 A, the output power of SBC1 and SBC2 gradually increased, achieving 12.6 W and 11.8 W at 44 A, respectively. Figure 2 shows the combined efficiency of the four SBC configurations and PBC. The combined efficiency of each SBC configuration was approximately 60%, with an average efficiency of 61.8% of the four SBC configurations at maximum current injection. The power of the four SBC configurations was 14.1 W lower than that of the four bars under free-running conditions. We attributed most of the power loss to the grating and OC because of the suboptimal diffraction efficiency and partial reflection. As the power of SBC1 and SBC2 increased, the power of the PBC increased to 23.4 W, with a beam-combining efficiency of approximately 96%. We obtained all these power measurements using an Ophir FL500 power meter at a controlled test temperature of 20 °C.

The inset in Figure 2 shows the far-field beam pattern after PBC. It includes two ellipses with their primary axes parallel to each other. The optical field’s horizontal and vertical dimensions were approximately equal.

Figure 3 shows the fast-axis beam quality of the single emitter, four SBC configurations, SBC1, SBC2, and PBC, with an injection current ranging from 8 A to 44 A. We determined the beam width of the combined laser according to ISO11146. We measured M² using Thorlabs M2MS. Figure 3 also shows the detailed beam quality M² in the fast-axis direction. The fast-axis M² values of the four SBC configurations increased slowly with an increase in the current and remained close. Under 44 A current injection, the fast-axis M² of the four SBC configurations was close to 3.5, greater than that of the single emitter. The beam quality degradation was caused by two primary factors: (1) aberrations in the FTL which resulted in incomplete overlap of the beams at the grating, and (2) a smile effect which was induced by thermal dissipation in the bar structure. The fast-axis M² of SBC1 and SBC2 increased gradually along with the rising current, reaching values of 9.4 and 9.8, respectively, at the maximum injection current of 44 A. The fast-axis M² of SBC1 exceeded the combined values of SBC bar1 + SBC bar2 by 2.6. Similarly, the fast-axis M² of SBC2 showed a 2.8 times larger value than the SBC bar3 + SBC bar4. The PBC demonstrated a fast-axis M² of 10.6, which exceeded those of SBC1 and SBC2 by 1.2 and 0.8, respectively.

Figure 4 shows the slow-axis beam quality of the single emitter, four SBC configurations, SBC1, SBC2, and PBC, depending on the injection current ranging from 8 A to 44 A. The single emitter laser demonstrated the optimal beam quality, with the M² factor increasing from 5.3 to 8.8. The deterioration of the slow-axis beam quality was caused predominately by the excitation of higher-order modes under high currents. Among the four SBC configurations, SBC bar 4 demonstrated the worst beam quality, with a slow-axis M² factor rising from 7.5 to 9.8, which was only marginally higher than that of the single emitter. The slow-axis M² factor of SBC1 increased from 7.6 to 10.4, and that of SBC2 rose from 8.0 to 10.1. The slow-axis M² factor of PBC increased from 8.8 to 11. Under maximum current conditions, the slow-axis M² of PBC was 2.3 higher than that of the single emitter.

The spectral characteristics of the four combined bars were investigated at 40 A, as presented in Figure 5. Each SBC configuration exhibited 19 distinct peaks, with occasional minor subpeaks observed between the primary emission lines. All the SBC configurations demonstrated comparable spectral bandwidths of approximately 29 nm, which were centered around 1940 nm. The external cavity locked all the single emitters, and occasional crosstalk between lasers was observed.

We systematically conducted brightness characterization, as shown in Figure 6, as defined by the radiometric relationship shown in Equation (2):

$$B={\displaystyle \frac{P}{{\lambda }^2{M}_x^2{M}_y^2}},$$

(2)

where $M_x^2$ and $M_y^2$ represent the beam quality in the slow and fast axes, respectively, P is the output power, and λ is the central wavelength. The brightness of the PBC increased progressively with the rising current, with a maximum value of 5.4 MW·cm⁻²·sr⁻¹ at 44 A.

4. Discussion

In mid-infrared wavelengths, achieving high output power through beam combining is significantly more challenging compared to in near-infrared systems, primarily due to lower inherent laser power. To enhance the stability of SBC systems, researchers are compelled to use longer-focal-length FTLs and OC with high reflectivity. However, excessively long focal lengths result in larger spot sizes incident on the diffraction grating. This increases the likelihood of crosstalk during wavelength locking, as diffracted light feedback through the OC introduces phase interference between adjacent channels. Additionally, high-reflectivity OC inherently reduce the overall combining efficiency.

A critical bottleneck is the grating performance at mid-infrared wavelengths. Commercially available gratings are currently limited by technical constraints, namely, insufficient groove density (typically ≤600 lines/mm) and low diffraction efficiency (≤94% at 1940 nm), both of which further degrade system efficiency. Therefore, promoting laser manufacturing technology and developing optical components of specific wavelengths are the key issues to be faced in subsequent research.

Secondly, although a BTS is used in this experiment to suppress the smile effect by beam shaping, the thermal effect at high current operation still leads to the expansion of the smile effect. This not only reduces the beam quality but also reduces the output power. Therefore, improving thermal management—specifically by actively reducing the temperature of the water-cooling system—has a certain influence on increasing output power in future designs.

Lastly, the spectral linewidth measured in this experiment was 29 nm, which is broader than that of near-infrared semiconductor lasers using SBC. While mid-infrared semiconductor lasers have broad applications in environmental monitoring, LiDAR, and medical diagnostics, the wider spectrum limits their suitability for scenarios requiring narrow linewidths. In future optimizations, increasing the grating groove density or integrating Rayleigh optical systems could be employed to compress the spectral width.

5. Conclusions

We combined four GaSb-based 1940 nm diode arrays by using SBC, spatial beam combining, and PBC. We enhanced the efficiency and output power of each SBC configuration. The asymmetry in beam quality between the fast-axis and slow-axis directions was reduced. The output power of the system was 23.4 W, and the beam qualities of the combined beams were M² = 10.6 and 11 in the fast and slow axes, respectively. The brightness of the combined laser was 5.4 MW·cm⁻²·sr⁻¹. The beam-combining efficiency of each SBC configuration reached approximately 60%, and that of PBC approached 97%. It is believed that further enhancement of the output power and brightness of the laser can be achieved by increasing the number of beam-combining units and improving the brightness of the single emitter.