Research progress of GaAs-based 980nm high-power semiconductor lasers
INTRODUCTION
Lightweight, reliable, and efficient high-power semiconductor lasers are required in medical, industrial, and military applications. Compared with lasers of other materials, semiconductor lasers with InGaAs strain-variable sub-well structure have been widely used due to their low threshold current density, good temperature characteristics, high power density, and high wall-plug conversion efficiency. Table 1 shows the comparison of laser power and wall-plug conversion efficiency between semiconductor lasers and other materials in the wavelength range above 900 nm [14]. Although the strain structure improves the performance of the laser, the optoelectronic performance, heat dissipation performance and beam quality of the high-power laser need to be greatly improved, and there are bottlenecks in the design and preparation of the epitaxial structure, the improvement of the optical catastrophic damage (COD) threshold and the improvement of heat dissipation efficiency, and the disadvantages of low wallplug efficiency (WPE) still need to be solved. The research of high-power semiconductor lasers in China started late, and there is a gap between the international top level in optimizing the performance of lasers, developing new structures, and expanding application fields [5]. Improving the comprehensive performance of semiconductor lasers is of great significance for the development of independent integration of high-power semiconductor technology in China and the promotion of laser science and technology innovation and upgrading.
In this paper, the historical development of InGaAs lasers, the factors affecting the comprehensive performance of lasers and their improvement methods, the design of epitaxial structure, chip structure and heat sink packaging structure, and the development prospects of high-power InGaAs lasers are reviewed.
1 Historical development of high-power InGaAs quantum well lasers
In 1984, Laidig et al. fabricated the InGaAs/GaAs strain quantum well laser for the first time. In 1991, Welch et al. [6] realized a 980 nm wavelength InGaAs/GaAs vertical cavity surface emitting laser (verticalexternal cavit). ｙｓｕｒｆａｃｅｅｍｉｔｔ inglaser, VCSEL) at room temperature. In 1994, Fan et al. [7] achieved a 1.05W output of a multi-quantum well InGaAs laser. In 2004, Chilla et al. [8] designed a vertical external cavity surface-emitting laser structure with an output power of 980 nmCW. Up to 30W. In 2007, the American JSDU company developed the wavelength 910~ The laser array at 980 nm has an output power of 480 W and a WPE of 73% at room temperature [9]. In 2013, the German company Laserline developed a high-power optical fiber coupling product with a continuous output power of 45kW [4]. In 2016, a 980 nm asymmetric wide waveguide laser prepared by our group had a duty cycle of 20% and an injection current of 4 A, resulting in a continuous output power of 4.1 W per tube [10]. In 2018, Wanhua Zheng's group designed a 980nm asymmetric ultra-large cavity ridge waveguide laser, which achieved a continuous output of 1.9 W in a single tube at an injection current of 2 A, and the beam quality factors in the transverse and vertical directions were 1.77 and 1.47, respectively [9]. Figure 1 shows the research progress of single-tube output power of high-power lasers at home and abroad [11].
2 Effect of laser epitaxial structure on optoelectronic performance
At present, the electro-optical conversion efficiency of high-power semiconductor lasers is about 50%, and increasing the output power and improving the beam characteristics are the key issues in the development of high-power lasers.
2.1 Increase the output power of the laser
The ideas for increasing the output power of the laser are: improving the quantum efficiency and WPE of the device, reducing the Γ of the optical limiting factor, reducing the carrier loss, and increasing the COD threshold. Specific measures include optimization of the active zone structure, the use of non-absorption window technology, cavity surface passivation and coating, etc.
In 2000, Ke et al. prepared an InGaAs/GaAs single quantum well ridge waveguide laser with an external quantum efficiency of 0.31 W/A and a slope efficiency of 0.37 W/A [12]. In 2005, Gao et al. [13] designed a single quantum well with a slope efficiency of 1.14W/A and a conversion efficiency of 317%. The strain-compensated structure has a smaller strain, a deeper potential well, and a larger critical thickness, which further improves the carrier limiting capability. In 2020, Yuan et al. [14] designed a 9XXnm laser with a GaAsP barrier layer, with a maximum WPE of 71.8% and a slope efficiency of 1.21 W/A. In recent years, the aluminum-free active zone has attracted attention because it is not easy to oxidize, but the carrier leakage of aluminum-free structure is severe. Heavy and low-slope efficiency, and prone to mode-hopping and wavelength redshift during high-current injection [15], it is not yet possible to achieve high-power stable single-mode output. In 2019, the internal loss of the 980 nm QWDs coupled large optical cavity laser designed by Kornyshov et al. [17] was only 0.5 cm-1, with good far-field characteristics and stable wavelength, and a continuous output power of more than 13 W. In 2020, Su et al. [18] designed a DETA-doped WS2QWDs structure that increased the PL intensity by a factor of six, and the QWDs structure is expected to greatly increase the carrier injection efficiency, or become the development direction of the next generation of high-power and high-performance lasers.
In order to avoid COD while increasing the output power, the first is to suppress the strain relaxation of the cavity surface and prevent the contraction of the band gap; The second is to reduce the carrier density of the cavity surface. There are two ways in which the absorption-free window technique can improve COD levels: quantum well hybridization and window-free injection region processes [19].
Quantum well hybridization counteracts band gap shrinkage by widening the band gap through interdiffusion. In 1998, Xu et al. [20] used an impurity-free vacancy induction technique. No absorption window was prepared by SiO2
and SrF2 The thin-film controls the degree of quantum well promiscancy, which increases the maximum output power of the laser by 36%. ２０１３ Zhou et al. [21] increased the COD level of InGaAs/GaAsP lasers by 16% through impurity-free vacancy induction. Compared to impurity-free vacancy induction, Si. In 2020, Ma Xiaoyu et al. used this technology to blueshift the wavelength by 93 nm [22], proving that impurity induction can be used to increase COD water Ping, but at present, the cost of this technology is high, and it is necessary to reduce the cost by doping and other methods. Window non-injection is the introduction of current into the non-injection area, reducing the current carrying capacity of the cavity surface. Sub-density techniques, including the addition of a dielectric barrier layer, ion implantation, and corrosion doping layers. In 1994, Sagawa and Hiramoto et al. [23] designed it .The 25 μm dielectric barrier increases the maximum output power of the laser from 350 mW to 466 mW. In 2003, Liu Bin et al. [24] used proton injection to prepare the non-injection region of cavity surface carriers, which increased the threshold power and maximum output power of COD by more than 50%. In 2014, Zhang et al. [25] introduced a lateral current non-injection zone through wet corrosion, and the COD power was increased from 18W to 22W at an injection current of 30A. In 2019, Arslan et al. [26] found that the non-injection window region can reduce the cavity surface temperature by 40% without affecting the high-power output, which is the application of this technology in high-power lasers. The basis is provided. The width and corrosion depth of the window non-injection zone are very small, so improving the accuracy of the etching process and reducing the sensitivity of the COD level to the volume of the non-injection zone are the development directions of the window non-injection process to improve the COD level.
Cavity surface passivation refers to the removal of impurities introduced by cavity surface cleavage and the formation of a passivation layer, thereby reducing the carrier density of the cavity surface. Cavity surface coating refers to the coating of different films on the passivated cavity surface to protect the cleavage cavity surface and change the reflectivity of the cavity surface.
Cavity surface passivation includes two methods: dry and wet. In 1983, Lindstrom and Tihanyi proposed the concept of vacuum ion beam dry passivation [27], and in 2003, Cheng et al. [28] used P2S5/NH4OH and (NH4)2Sx solution chemical passivation, and the COD power was increased from 600 mW to 1500 mW. In 2019, Zhao et al. [29] designed a vacuum cleavage passivation process, which increased the output power of the device by 23% compared with the laser passivated after cleavage in air, but the cost problem restricts the promotion of vacuum cleavage passivation technology. The cavity surface coating is composed of an anti-reflective coating (anti-reflection coating) on the front cavity surface and a high-reflectivity film (high-reflection coating) on the rear cavity surface, of which the reflectivity of the high-reflective coating is >90%, which is the main part of the cavity surface coating. In 2005, Shu Xiongwen et al. [30] used ion-assisted SiO2 plating
/TiO2 high reflection coating and Al2O3 anti-reflection coating increase the external quantum efficiency by 77%. In 2013, Liu Lei [31] designed Al2O3/TiO2. The /SiO2 high-reflection coating system increases the slope efficiency of the device from 0.49W/A to 0.91W/A, and the optical output power is increased from 0.70W to 1W. The function of the anti-reflection coating is to eliminate stray light and improve the light transmittance of the device. In 2016, Xu Liuyang [32] designed a plasma passivated AlN anti-reflection coating, and the COD threshold power was increased from 3.0W to 6.1W. In 2020, Cui et al. [33] found that the al2O3 anti-reflection coating in the tensile stress state was higher than that in the conventional compressive stress state, which proposed a new idea for the optimization of cavity membranes.
2.2 Beam characteristics improved
The beam characteristics of the laser are evaluated by the M2 factor of Eq. (1), where ω is the beam waist radius, θ is the far-field divergence half-angle, and λ is the wavelength. ω is affected by the refractive index guidance and anti-guidance effects on the fast axis, and the change is small, and it is related to the bar width on the slow axis. Therefore, θ reduction is the main way to improve beam quality [34], and its methods mainly include largeopticalcavity (LOC) and longitudinal photonic band waveguides
(longitudinalphotonicbandcrystal, LPBC) and insert mode extension layer, etc.
The LOC theory was proposed by Tsang and Olsson in 1983 [35], and this structure can reduce the vertical emission angle, but the waveguide layer is too thick
High-order lasing and carrier leakage. Because the hole has a stronger loss of light absorption, the thickness of the P layer should be reduced. In 2010, Hu et al. [36] designed a LOC structure with the removal of the P waveguide layer, and the far-field divergence angle was reduced to 16.1°(θ⊥)×10.2°(θ∥). Removing the P-waveguide layer results in mode contention, so it needs to be preserved appropriately. In 2013, Li Jianjun et al. [37] designed a 980nm laser with an N waveguide layer of 1.15 μm and a P waveguide layer of 0.85 μm, with a single-mode operating far-field divergence angle of 24°(θ⊥)×6.6°(θ∥). In 2017, Serin et al. [38] designed a 4.8μm coupled large optical cavity structure with a far-field vertical divergence angle of 14°, which is basically close to the diffraction limit. In 2018, Zhao et al. [39] designed a 5.2 μm thick ultra-large optical cavity waveguide layer, in which the P waveguide layer is 700 nm thick and its far-field divergence angle is 11.5°(θ⊥）×６．８°（θ∥）。
LPBC is composed of dielectric materials with different refractive indices growing alternately, which can effectively suppress the divergence of light waves and reduce the divergence angle of the device. In 2003, Maximov et al. [40] designed an LPBC laser with a far-field vertical divergence angle of 6°, and in 2008, Novikov et al. [41] designed a 16.5 μm LPBC laser with a far-field vertical divergence angle of less than 5° at 1.3 W continuous output. In 2013, Wang Lijie [42] designed a single-horizontal model-stable Bragg-reflecting PBC dual-beam laser as shown in Figure 3, with a single-beam laser far-field divergence angle of 7.2°×5.4° and a continuous output power of 2.6W. In 2019, Yoshida et al. [43] designed photonic crystal surface emitting lasers (PCSELs) with an output power of 10 W and a far-field vertical divergence angle of less than 0.3°, which fully reflects the advantages of photonic crystals in reducing the divergence angle and increasing the output power. At present, the domestic for. There is little research on PCSELs, and the development goal of LPBCs and PCSELs is to simplify the growth process and further optimize their structure and performance.
The threshold current of the LOC structure is relatively large, and the cost of LPBC is relatively higher. The mode expansion layer refers to the addition of a waveguide layer to the restriction layer to reduce the limiting factor and the far-field divergence angle by inducing the near-field light wave mode expansion [44]. In 1996, Yen and Lee introduced a 100 nm mode expansion layer in a single tube of InGaAs lasers [45], and in 2006, Wang et al. [46] designed an Al0.5Ga0.5As mode expansion layer with a far-field vertical divergence angle of 25° at an output power of 75 W. In 2018, Li et al. [47] introduced a low refractive index layer between the mode expansion layer and the central waveguide, so that the vertical divergence angle of the far field was 18° and the threshold current density was only 173A/cm2.
3. Laser chip structure design
Fig.3 SchematicdiagramofLPBCstructure[42]Although wide strip lasers can increase the output power by increasing the width of the ridge waveguide, they will face problems such as high-order mode lasing and COD. In order to achieve high power and high beam quality output, the following four new laser structures have gradually become research hotspots. ...
3.1 Distributed feedback semiconductor laser (DFB) DFB is a laser engraved with a Bragg grating in a cavity, which has a stable emission wavelength and a small far-field divergence angle. In 1997, Jeon et al. [48] prepared a 980 nm DFB laser using one-time epitaxy, with a far-field divergence angle of 11.7°×17.8° and a continuous output power of 70 mW. In 2005, Wenzel et al. [49] designed a large optical cavity waveguide structure DFB with a slope efficiency of 1W/A and an output power of 500mW. Secondary epitaxy is a technique for the epitaxial growth of overlay, upper limiting layer and ohmic contact layer on etched Bragg grating. In 2011, the FBH Institute prepared it by secondary epitaxial growth. The 976 nm DFB laser shown in Figure 4 has a WPE greater than 60%, a 10 W continuous output lifetime of 5000 h, and a spectral width of less than 1 nm [50]. The secondary epitaxial growth time is long, which not only damages the quality of the material, but also reduces the sensitivity of the laser. In 2017, Decker et al. [51] designed the Direct Out In 2019, Qiu et al. [52] designed a gain-coupled DFB structure without secondary epitaxy, which has better beam quality, but the output power is still lower than that of conventional secondary epitaxial DFB, and increasing the output power and threshold of primary epitaxial DFB is the development direction of DFB.
3.2 Distributed Braggreflector (DBR) DFB and DBR both have built-in Bragg grating, the difference is that the DFB grating is in the resonator, while the DBR grating is at one or both ends of the resonator and does not use secondary epitaxy, so the manufacturing process is simpler. Parke and O′Brien designed an edge-emission DBR at a wavelength of 971.9 nm in 1992 with a continuous output power of 110 mW and a slope efficiency of 0.35~0.47 W/A [53]. In order to increase the output power while maintaining a stable wavelength, the structure of the DBR has been improved. In 2009, Fiebig et al. [54] designed a tapered DBR with a full expansion angle of 6°, an operating wavelength of 980nm, a maximum output power of 12W, and a far-field vertical divergence angle of less than 15°. In 2011, Reddy designed a dual-wavelength DBR with two DBR etching gratings in the center of the gain conduction band, which can output two stable wavelengths of 968 nm and 976.8 nm simultaneously, laying the foundation for the high-power multi-wavelength output of DBR lasers [55]. In 2020, Paoletti et al. [56] designed a 10-spatially multiplexed multi-emitter DBR, which can output multiple wavelengths in the 920nm range, with a power of up to 100W and a simple preparation process, and multi-wavelength output DBR lasers will become the development trend of high-power DBR lasers.
3.3 Vertical cavity surface-emitting lasers
As shown in Figure 5, VCSELs have two configurations, top-emission and bottom-emission, which have the advantages of low threshold current, high wall-plug conversion efficiency, and small divergence angle. In 1990, Ceels and Coldren achieved a continuous output of 980 nm VCSEL at room temperature with a threshold current of 0.7 mA [57]. In 2001, Miller et al. [58] prepared a VCSEL array composed of 19 single devices with a continuous output power of 1.08W at room temperature, which proved the characteristics of high power and stable output of VCSEL array. In 2008, Princeton Optronics designed a 5mm VCSEL array with a maximum WPE of 51% and a maximum output power of 231W, the highest level in the world at the time [59]. In 2018, Warren et al. [60] fabricated a 150-cell VCSEL array using back-emitting VCSEL array technology, with a peak output power of 400 W and a far-field vertical divergence angle of 15°. In 2017, Czyszanowski et al. proposed an improved scheme for semiconductor-metal subwavelength gratings to replace P-DBR. Subwavelength gratings have the functions of optical coupling and electrical injection, and their thickness is smaller, and the polarization characteristics and thermoelectric properties are better, but they have the problems of high threshold and large self-absorption [61]. Better beam quality of vertical outer cavity surface-emitting lasers, 2020 Zhang Jianwei et al. [62] designed a vertical outer cavity surface-emitting laser with a far-field divergence angle of 9,2°×9.0°, and the spot was uniformly distributed, but the output power was not as good as that of VCSELs. At present, companies such as PhilipsPhotonics have commercialized VCSELs in the fields of smartphones and sensors, and Accelink has achieved the practical application of VCSELs in optical communications, but in the field of consumer electronics, domestic manufacturers have not yet achieved mass production [63].
4 Heat sink package design and its impact on device performance
High power output requires not only excellent laser chips, but also reasonable design of heat sink packages. According to Shen et al., the thermal resistance of the chip accounts for only 35.7% of the total thermal resistance of LD, and about 64% of the thermal resistance comes from heat sink, soldering, and packaging [64]. The waste heat affects the performance of the device and accelerates the aging of the device, and the heat sink package structure with low thermal resistance and high heat dissipation efficiency generates less heat, which can efficiently dissipate heat for the chip and improve the power and service life of the laser.
The expansion coefficient is quite different from that of GaAs. In 2013, Zhao et al. [65] introduced a microchannel AlN secondary heat sink between the copper heat sink and the chip, which reduced the thermal resistance by 80% Although AlN can reduce the thermal resistance, the preparation process is more complex than that of alloys, and in 2019, Shi Linlin et al. [66] designed the CuW secondary heat sink and graphite-assisted heat sink structures, reducing the size of the transition heat sink from 2.0 mm to
At 0.6 mm, the junction temperature is reduced from 338.9 K to 334.9 K. For higher-power semiconductor lasers, the ceramic heat sink is not very effective in heat dissipation, and the alloy material is easy to corrode, so the researchers introduced a diamond heat sink with higher thermal conductivity and more stability. In 1996, Weiss et al. [67] used CVD diamond film as a laser heat sink for the first time, and the slope efficiency of the device was 1W/A. In 2003, Pan Cunhai et al. [68] designed a heat sink for a diamond film/Ti/Ni/Au metallization system, and the thermal resistance was 40% of that of the AlN heat settle. In 2016, Parashchuk [69] found that the diamond heat sink can increase the slope efficiency of the laser by 1.5~2 times, and the operating current range by 2~3 times. In 2019, the Cu/diamond(Ti) composite designed by Wang Luhua [70] has a thermal conductivity of up to 811W/(m·K) and a thermal expansion coefficient very close to that of GaAs, which is expected to become a new generation of heat sink packaging materials.
4.2 Heat sink package structure design
Packaging includes chip packaging and protective case packaging, which can not only form a closed circuit and protect the chip, but also an important way for the chip to dissipate heat. Chip packaging is to directly solder the chip to the heat sink, mainly including Cmount and Fmount packages [71], and the structure is shown in Figure 6. Cmount has low cost and mature technology, but it is large in size and small in power. The Fmount package assembles the chip, the sub-heat sink and the copper heat sink vertically, which has a shorter heat dissipation path and higher heat dissipation efficiency. In 2010, Zhang Yanxin et al. [72] designed the Fmount structure with a WPE of 55%; In 2015, Bezotosnyi et al. [73] applied Fmount to a 980nm laser, and the COD threshold was increased from 20W to 28W, and the WPE reached 65% at 15W continuous output. In 2020, Wu et al. [74] designed a "sandwich" heat sink composed of two layers of copper plates and a high thermal conductivity interlayer, which reduced the thermal resistance by 27.4% compared with the conventional Fmount.
Protective case packaging is a structure that solders a chip to a certain part of the heat sink and seals and protects it, such as a microchannel heat sink. In 1988, Mundinger et al. [75] invented the silicon microchannel heat sink, and in 2005, Liu Yun et al. [76] improved it to a five-layer oxygen-free copper microchannel heat sink with a thermal resistance of 4.982×10-3K·cm2/W. In 2015, Fan Siqiang [77] designed a microchannel structure with a micro-evaporation chamber and a multi-channel capillary layer, using R134a phase change refrigeration, and the maximum temperature of 100W bar was only 305K. In 2019, Shen et al. [78] designed a hybrid heat sink of microchannel and jet slit to reduce the thermal resistance of the laser array by 15% and achieve a WPE of 64.2% through jet shock. For example, in 2009, RNIN successfully dissipated heat for a 150kW high-energy laser using spray cooling and heat storage technology. In 2010, John et al. [79] used InGa liquid alloy as a coolant, which improved the thermal conductivity by 28 times, but there were problems such as immature technology and poor universality.
5 Conclusion and outlook
In this paper, the historical development of 980nm InGaAs high-power quantum well lasers and the problems affecting the performance of InGaAs lasers and the improvement methods are reviewed. In terms of epitaxial structure, strain-compensated quantum wells, no absorption window, cavity surface passivation and coating solve the problems of lattice mismatch and carrier leakage in the active region, so that the output power and electro-optical conversion efficiency of the laser are continuously improved. The development of large-cavity asymmetric waveguides, photonic crystals and mode expansion layers improves the symmetry of the beam and the quality of the light output. The new chip structure improves the output power and beam characteristics of the laser; The development of heat sink packaging technology has enhanced the reliability of lasers and promoted the integration and industrial production of high-power lasers.
In the future, the development direction of high-power 980nm semiconductor lasers mainly includes: (1) improving chip structure and epitaxial growth technology. Develop a new laser structure to reduce intracavity losses and threshold currents, improve output power and conversion efficiency, and simplify the fabrication process. (2) Develop the heat sink packaging technology of lasers. Develop heat sink materials with higher thermal conductivity and more stable performance of the package structure to avoid the heat generation of the package structure affecting the reliability and output power of the device. (3) Development of optical correction technology. Achieve the same quality of light output in the fast and slow axes, resulting in a smaller divergence angle, a better symmetrical spot, and a more collimated beam. (4) Develop highly integrated technology to make high-power lasers lightweight.
