Semiconductor lasers in the visible red light band of 600–700 nm have a high utilization value. In the field of military science, they are used in visible night vision, high-light illumination, and scope ranging systems. As an excellent and stable pump source, red semiconductor lasers can pump and multiply specific nonlinear crystals to produce target light for the detection and prediction of climate change. In medical treatment, red lasers with appropriate power and brightness can accelerate the absorption of mitochondria, thus speeding up the repair of damaged cells and capillaries, which helps to inhibit the continuous bleeding of wounds and effectively accelerate wound healing. In terms of cosmetology, red light laser devices can improve cell metabolism and then accelerate the growth of collagen, which ultimately achieves the effect of beauty and anti-aging. In industrial processing, red light and blue light composite welding has huge advantages; its efficiency is several times higher than other bands of lasers. So far, the high-power red semiconductor lasers on the market are mainly packaged through the spatial combining technology, but the subsequent application of the efficiency will be reduced due to low beam quality. Dense spectral beam combining (DSBC) can simultaneously improve the output power and beam quality, which ensures the brightness of the output laser. Thus, it can be seen that obtaining high-power red semiconductor lasers through DSBC has a very wide range of application prospects and research value.

DSBC was first realized in 1999 at the Lincoln Laboratory of the Massachusetts Institute of Technology (MIT)^[1,2]. In the following two decades, researchers all over the world have continued to deepen their research on spectral beam combining. From 2014 to 2017, TRUMPF of Germany increased the beam power of a 953 nm semiconductor laser from 350 W to more than 1 kW by continuously optimizing the structure of DSBC^[3–5]. In 2018, the China Academy of Engineering Physics also realized an infrared laser output of more than 1 kW by a DSBC experiment^[6]. From 2020 to 2022, Sichuan University, in collaboration with EVERBRIGHT, increased the output power of 976 nm semiconductor lasers from 350 W to more than 21 kW by DSBC systems with fiber coupling^[7–9]. Following this, the Institute of Semiconductors of the Chinese Academy of Sciences (CAS) successfully increased the beam combining efficiency of a 960 nm laser bar strip to more than 100% by the DSBC method^[10]. In a recent study, the Beijing Institute of Technology successfully realized spectral beam combining for solid-state lasers with a final output pulse energy of more than 170 mJ^[11].

By summarizing the existing spectral beam combining results, we can find that the current DSBC output wavelength is mainly used in the infrared band, and there are scarcely any reports on the visible red light band. The existing research is basically on fixed beam spectral width, while there are almost no reports on tunable spectrum DSBC experiments that can be adapted to different application scenarios.

In this study, a 650 nm spectrum-tunable DSBC system based on the structure of a dual-grating DSBC (DG-DSBC) module was built for the first time, to the best of our knowledge. The DG-DSBC system has two main advantages over the previous conventional structure. First, in the spectrum of the previous conventional structure, we can find that the combined beam spectrum generates a large feedback crosstalk, and the spectrum is very unstable during operation. These problems are detrimental to the subsequent application of DSBC systems in the visible red band. The new system has very stable and non-crosstalk beam spectra, which is very conducive to subsequent applications such as pumping of emerald-green lasers that require very high spectral stability. Second, the new structure can realize continuous tuning of the beam spectrum using only gratings with identical parameters and other supporting optical components, which greatly reduces the cost of subsequent applications and can pump different solid-state laser working substances, realizing multi-purpose use of one machine. This study summarizes and analyzes three different output spectra with the corresponding output power and beam quality results. (This experiment only takes these three spectral widths as an example, and other spectral widths can also be realized when conditions such as gain spectral width or experimental size allow.)

Figure 1 shows diagrams of the experimental structure and the DG-DSBC principle. The actual structure of the experimental system is shown in Fig. 1(a).

As shown in Fig. 1(a), the light sources in the experiment consisted of 15,650 nm COS (chip on submount) semiconductor single emitters arranged along the slow axis stepwise. (The light source number determination method will be explained later in detail.) The spatial intervals of each single emitter in the fast- and slow-axis directions were 1 and 10 mm. The 650 nm COS single emitters are the latest version, self-developed by the Weifang Academy of Advanced Opto-Electronic Circuits, whose central wavelength (λ0) is around 650 nm. The output parameters of the 650 nm COS single emitters are shown in Fig. 2. Compared with the first generation of 650 nm single emitters used in the first 650 nm DSBC experiment^[12] realized by our group in 2024, the thermal saturation working current of the latest emitter is increased from 2.1 to 3.0 A. Therefore, the maximum output power of each single emitter is also increased from about 1.2 W to more than 1.6 W. However, since the output power of the single emitter decreases rapidly after the working current exceeds 3.0 A, in order to ensure the stability of the whole experimental system, the maximum working current of the experiment was set to 2.7 A. Thus, the maximum output power of the single emitter is about 1.55 W, and the gain spectral width is about 11 nm. The output coupler (OC) is made of fused silica with a reflective coating of 10% at 600–700 nm on the front surface. The effective focal length of CL is about 250 mm. The materials and parameters of other components in the system are the same as those used in the first realization of the traditional 650 nm DSBC experiment^[12] by our group in 2024; they will not be expanded here.

The principle of beam combining in the DG-DSBC module is shown in Fig. 1(b). In the figure, we set up five levels of COS single emitters as an illustration, which are represented by violet (-k), blue (-g), red (0), yellow (g), and green (k) beams. Each beam is incident in the same direction on grating 1, which has been placed in the Littrow structure. The initial angles of incidence and diffraction are α0 and β0. The angles of incidence and diffraction are derived from the grating Eq. (1) to be approximately 54.6°: α0=β0=arcsin(λ02·Λ).(1)

Afterwards, all beams are incident in the same direction onto grating 2 (grating 2 is shortened in the figure mainly to make the schematic more aesthetically pleasing) and then continue to be output at the same diffraction angle. So far, the beams have not yet been combined. After adding the OC, part of the red beam from the center zeroth single emitter (red) is partially reflected back to grating 2. For this feedback light (red dotted line in the figure), grating 2 diffracts it back to grating 1 at a different diffraction angle. Then, grating 1 diffracts it back to each level of the single emitters to establish the wave-locking oscillation corresponding to each emitter. This process is the initial establishment of the DG-DSBC cavity feedback, which forces the beam output from each single emitter to pass through grating 1 again at a different diffraction angle and converge on the same area of grating 2. (As an example, the kth single emitter corresponds to a diffraction angle of βk—same for other levels.) Each level of the beam passes through grating 2 at a different angle of incidence. Finally, the DG-DSBC external cavity feedback is established by the partial feedback of the OC and combined into a single laser beam (pink arrow in the figure).

After the entire external cavity is established, the schematic of the tunable spectrum principle is shown in Fig. 3(a). From the figure, we can see that since the direction of the center zeroth beam (red) is always kept constant, the diffraction angle βk of the laser output from the kth single emitter (green) after passing through grating 1 changes accordingly in the case of decreasing (blue grating 2′) and increasing (violet grating 2″) the vertical pitch between the two gratings. When the vertical interval decreases, the diffraction angle βk′ increases compared to βk. Using the grating equation corresponding to grating 2, we know that an increase in the diffraction angle βk′ leads to an increase in the locked wavelength λk (the derivation process is the same as the traditional spectral beam combining, not to be expanded here), which ultimately leads to a larger locked spectral width. Similarly, when the vertical interval increases, the diffraction angle βk″ decreases compared to βk, which ultimately leads to a smaller locked spectral width. The principle is the same for all other levels. Finally, we can derive the group of equations corresponding to the vertical interval Δ and the total spectral width λtotal of the combining beam from the graphical geometrical relationship: λtotal=2·(λk−λ0),(2)Δ=(7·dcosα0)·{tan[arcsin(λkΛ)−sinα0]−tanα0},(3)where d is the fast-axis interval of each single emitter.

A schematic principle of the quantity of 650 nm single emitters for the experimental system is shown in Fig. 3(b). As shown in the figure, the total width of the light source involved in the combining process is related to the aperture of the grating in the diffraction direction by the Pythagorean Theorem. Therefore, through the geometrical relationship, we can deduce the maximum number of light sources that can be allowed in the grating under an ideal condition: N=P·sin(90°−α0)d,(4)where P represents the diffraction direction pass size of the gratings (the value of this size is 30 mm). Therefore, combined with the residual divergence angle that exists in the actual beam as well as the adjustment error, we determined the number of 650 nm single emitters to be 15 in order to ensure that there is no light leakage at all.

In this experiment, in order to give examples of spectral tuning in the range of gain spectral width, we preset three beam spectral widths of 5, 8, and 10 nm. We can calculate the vertical intervals corresponding to the three preset beam spectral widths of 52.15, 24.82, and 17.81 cm through Eqs. (2) and (3), respectively. It is worth noticing that the grating placement position will exceed the maximum boundary of the current experimental platform if the locked spectral width is less than 5 nm. Thus, smaller spectral widths were not preset in this experiment. The main purpose of setting the 10 nm spectral width is to maintain the same spectral width as the traditional non-crosstalk 650 nm DSBC experiment previously realized by our group. (Since the traditional DSBC structure has been reported in the literature, this paper will not expand the structure in detail and only gives the output power results directly in Fig. 4.)

Figure 4 shows the DG-DSBC system output results and beam efficiency curve. The results are measured in the spectral width of 10 nm. (The other two groups of spectral widths of the results are basically consistent and will not be expanded here.) Beam combining efficiency is defined as the ratio of the combined output power to the free-running power of the 15 single emitters. As can be seen from Fig. 4, when the working current is 2.4 A, the output power of the DG-DSBC system is 19.415 W, while the efficiency of the DG-DSBC system reaches the maximum value after stabilization, which is about 88.27%. When the working current reaches 2.7 A, the free-running output power of 15 COS single emitters is 23.195 W. The maximum output power of the DG-DSBC system combined beam reaches 20.385 W, and the DG-DSBC combined beam efficiency is about 87.89%. Further comparison with the results of the traditional non-crosstalk 650 nm DSBC experiment (pink curve) reveals that, in the case of 2.0 A working current, due to the increase in the quantity of single emitters participating in the combining system, the total output power is finally increased from 12.072 to 16.307 W with the same spectral width. The output power is increased by more than 35%. It should be explained that the traditional DSBC structure mentioned in Fig. 4 uses the transform lens with a focal length of 250 mm in order to match the spectral width (10 nm) of the new DG-DSBC structure, and that only 11 single tubes can participate in the beam combining system as the peak spacing becomes 0.93 nm in the limit of the 11 nm gain spectral width. The new structure can allow 15 single tubes to participate in the beam combining, so the power is greatly improved. The lower right corner of the figure (red photo) gives the actual combined beam spot under the maximum working current. From the result, it can be seen that the combined beam spot is basically the same as a single emitter spot without any side flap spot, which indicates that the beam combining result is excellent, with no crosstalk occurring.

The final experimental spectrum results for the three preset spectral widths are shown in Figs. 5(a)–5(c). The actual spectrum result corresponding to the preset spectral width of 5 nm is shown in Fig. 5(a). The spectrum ranges from 647.59 to 652.48 nm with a center wavelength of about 650.04 nm, and the total width is about 4.89 nm. The actual spectral result corresponding to the preset spectral width of 8 nm is shown in Fig. 5(b). The spectrum ranges from 645.78 to 653.82 nm with a center wavelength of about 649.80 nm, and the total width is about 8.04 nm. The actual spectral results corresponding to the preset spectral width of 10 nm are shown in Fig. 5(c). The spectrum ranges from 655.25 to 645.08 nm with a center wavelength of about 650.17 nm, and the total width is about 10.17 nm. There are inconsistencies in the intensity of the discrete wave peaks in the figure, which we discuss in two main cases. The first case is when the emitters are located in the central area of the gain spectrum. Although we have controlled the beam directivity error within 100 µm at a distance of 500 mm during the adjustment, the feedback beams from each emitter still have different degrees of deviation after the whole DSBC system. Therefore, the tiny deviations cause different beam combining effects, which directly lead to the differences in the intensity of the locked peaks in the central area of the spectrum. The second case is when the emitters are located at the edge of the gain spectrum. The locked peak wavelength corresponding to the emitter at this position is already located at the edge area of the gain spectrum and the gain spectrum curve is approximately the same as the Gaussian distribution. As a result, the gain intensity cannot be compared with the central area, which ultimately leads to a significant intensity reduction of the edge area emitters.

From the above examples, it can be seen that the three preset spectral widths are consistent with the actual beam combining results, and the center wavelengths are also basically consistent with the intrinsic center wavelengths of the single emitters. This proves that the experimental structure can indeed realize the arbitrary spectral width tunable function under the conditions. The main reason for the slight deviation of the experimental results from the ideal preset value is the error in the calculation of the perpendicular distance between the two gratings, as well as the error in the placement and angle of the optical components during the actual experiment.

Figure 5(d) shows the life test results of this DG-DSBC experimental system. Since this system is only a preliminary experiment, we did not conduct a professional product-level life test on the system. The experiment tested the output power of the system at 10 nm for 1 h of operation. It can be found that the combined beam output power is basically stabilized at about 20 W, which proves that the DG-DSBC experimental system has good stability.

In Fig. 6(a), the beam quality measurements of the DG-DSBC system for three different spectral widths are shown under the maximum working current. When the working current is 2.7 A, the three preset spectral widths correspond to the fast-axis beam quality (MFast2) of about 1.833, 1.733, and 1.686, and the slow-axis beam quality (MSlow2) of about 23.211, 21.315, and 20.789, respectively. The beam quality deterioration of the combined beam as compared to the single emitter is shown in Fig. 6(b). Deterioration percentage is defined as the ratio of the beam quality difference between the combined beam and the individual single emitter to the individual single emitter beam quality. When the working current is 2.7 A, the three preset spectral widths correspond to the fast-axis beam quality deterioration of about 11.02%, 4.96%, and 2.12%, while the slow-axis beam quality deterioration is about 14.99%, 5.61%, and 2.99%, respectively.

Combining Fig. 6(a) with Fig. 6(b) shows that when the combined spectral width gradually decreases, the deterioration of the beam quality becomes more pronounced. The main reason is that as the spectral width decreases, the distance between the two gratings is getting bigger and bigger, which leads to a reduction in the accuracy of the dimming with a further increase in the parallel error of the two gratings. Because it is the first experimental attempt, the deterioration of the beam quality in the case where the vertical distance between the gratings is more than 52 cm (actual distance is more than 100 cm) has been controlled as much as possible to about 10%. In the future, we will further improve the dimming accuracy.

According to the beam brightness formula, when the working current is 2.7 A, we can obtain that the DG-DSBC system beam brightness of the three preset spectral widths of 5, 8, and 10 nm is about 113.405, 130.617, and 137.655MW·cm−2·sr−1, respectively. In the case of the 10 nm width, the final DG-DSBC system beam brightness improvement is more than 71% compared to the traditional non-crosstalk 650 nm DSBC experiment.

In this study, a spectrum-tunable 650 nm semiconductor laser DSBC system based on the DG-DSBC module has been constructed for the first time, to the best of our knowledge. Three spectrum-tunable examples are finally realized: the final locked spectra can be stably operated at 4.89, 8.04, and 10.17 nm, which are basically consistent with the theoretical calculations. The subsequent optimization of this experiment is planned in three ways. First, we will continue to optimize the internal structure of the 650 nm COS single emitter to further increase the gain spectral width, which will allow a larger spectral tunable range. Second, we will customize the diffraction grating with larger apertures to allow more COS single emitters to participate in the beam combining process, which will further improve the final output power. Finally, we will further optimize the optical dimming method to reduce the positional and angular errors of the experiment to achieve the best output beam quality. Despite the shortcomings of the experiment, this study is the first to realize spectrum-tunable visible red band semiconductor laser DSBC.