1 Introduction

High-power lasers in 2-μm-wavelength region have attracted much attention in recent years because of their potential applications in remote sensing, medical operation, lidar and commercial processing fields [1–3]. Besides, they can be further used as pump sources of OPO and OPA systems operating in the mid-infrared region. Tm- or Ho-doped material is a traditional choice for generating 2-μm laser. For Ho-doped materials, although there are diode lasers at 1.9 μm that can be used as the pump source, the power level of the commercially available laser diode is limited to tens of watts. In comparison, Tm-doped materials can be directly pumped by commercial laser diodes at around 800 nm. The existence of a cross-relaxation process in Tm ions can lead to a quantum yield of nearly two, thus guaranteeing a high-slope-efficiency well beyond stokes limit in this pumping scheme. Up to now, laser operations with hundreds of watts of output power have been achieved in Tm-doped fiber lasers and bulk lasers. Benefiting from the large ratio of surface to volume, the thermal management in fiber lasers was much simpler than in bulk lasers. Eight hundred and eighty five watts output power was achieved using a Tm-doped fiber in 2010 [4]. Using a MOPA configuration, the output power was scaled up to >1 kW [4]. However, the strong Raman effect and the nonlinearities limited the performances of high-power fiber devices. In comparison, bulk lasers have larger mode volume, which greatly reduce the nonlinearity in gain medium, and the maximum storable energy is larger. This is particularly advantageous for generating ultrashort pulses with high peak power [5]. Wang et al. [6] reported a 267-W cw Tm:YAG laser with a slope efficiency of 29.8 % with respect to the pump power in 2013. They used an LD side-pumped rod-shaped Tm:YAG as the gain medium and fabricated screw threads on the side surface to provide effective heat dissipation. However, a complex design of coupling system was needed to efficiently focus the pump laser into the gain medium. In comparison with the configuration of the LD side-pumped rod-shaped gain medium, LD end-pumped slab lasers have larger aspect ratio and more heat per crystal volume can be dissipated than in rod geometry, which provides the possibility of scaling to higher power levels than in rod lasers [7, 8]. Besides, optimized mode matching in end-pumping scheme leaded to a higher slope efficiency and a more efficient laser operation [9]. Li et al. [10] reported a 200-W Tm:YLF end-pumped slab laser operation at 486 W absorbed pump power in 2013, corresponding to a slope efficiency of 52.2 % with respect to the absorbed pump power. The asymmetric pump line combined with the spherical resonator necessarily leaded to an increase in the beam quality in one axis, while the beam quality of the other transversal beam axis was kept diffraction-limited. The poor laser beam quality on one axis could be improved by special resonator design such as the hybrid resonator, which was successfully used in Nd-doped laser systems [11]. Up to present, most efforts in pursuing high-power 2-μm output have focused on laser systems based on single crystals.

Transparent Tm:YAG ceramics have been fabricated with similar optical properties compared to the corresponding single crystals in recent years. Compared with lasers in 1-μm region, lasers in 2-μm region experience much lower Rayleigh scattering in ceramics, leading to exceptionally low propagation losses. For cw laser operation, slope efficiency as high as 65 % with respect to the absorbed pump power has been reported, which was even higher than the slope efficiency of 59 % with respect to the absorbed pump power that was achieved with single crystal [12, 13]. Recently, higher fracture stress has been observed in ceramics (386 ± 50 MPa) than in crystals (235 ± 16 MPa), which means that higher output power can be expected in ceramic laser systems (6.4 ± 0.6 kW/cm2 for ceramic vs 3.9 ± 0.3 kW/cm2 for single crystal) [14]. Furthermore, Tm:YAG ceramics have many other advantages over single crystals when applied in slab laser systems. For example, it is easier to manufacture ceramics with more function design freedom (such as multilayer ceramics with gradient doping concentration). The ceramic materials can also be produced with much larger aperture size using the advanced ceramic processing technology [15]. Besides, laser ceramics are more suitable for mass production because of the low cost and short period for fabrication [16, 17]. These characteristics have made laser ceramics very promising for application in high-power lasers, in which a large number of gain media with large size are of essential need for further power scaling [18]. Zhang et al. [19] have demonstrated an LD end-pumped Tm:YAG ceramic laser with an output power of 17.2 W and a slope efficiency of 36.5 % with respect to the absorbed pump power. Further power scaling of Tm:YAG ceramic lasers calls for ceramics with good optical quality within relatively larger size. In addition, appropriate thermal management is required because the heat accumulation becomes prominent for further increase in pump power.

In this paper, we demonstrate a high-power LD end-pumped Tm:YAG slab ceramic laser. In our experiment, various measures were taken for efficient heat dissipation and stress release including the usage of micro-channel heat sinks and paddings. The influence of the wavelength of the pump source on the laser performance was investigated in detail. The highest laser output power achieved was 52 W at 2013 nm with 220 W pump power, corresponding to a slope efficiency of 27.8 % with respect to the incident pump power. As far as we know, this is the highest laser output power achieved in Tm:YAG ceramic laser systems. The result proves Tm-doped laser ceramic of large size to be a competitive candidate as the gain medium in high-power 2-μm slab lasers with its reliable laser performance and many other promising advantages.

2 Experimental setup

A laser diode stack consisting of five diode bars, with the central wavelength around 785 nm, was used as the pump source to utilize the cross-relaxation process in Tm:YAG systems. The spectral width (FWHM) of the laser diode was 2.8 nm. The beam qualities of the diode stack were \(M_{\text{x}}^{ 2} = 147\) and \(M_{\text{y}}^{2} = 78.2\). Two different pump coupling systems were adopted in pump wavelength tuning experiment and power scaling experiment, respectively, to focus the pump beam into different sizes. In the pump wavelength tuning experiment, two plano-convex cylindrical lens with perpendicular axes of curvature were used to focus the pump laser into a spot size of ~1.6 mm(x) × 0.6 mm(y), while for further power scaling, two cylindrical lens on the horizontal direction combined with a cylindrical lens on the vertical direction were used to focus the pump beam into a larger spot size of ~5.5 mm(x) × 0.54 mm(y), as shown in Fig. 1. The linear resonator was comprised of two flat mirrors. The rear mirror M1 was anti-reflection (AR) coated at 760–810 nm and high reflection (HR) coated at 1,890–2,170 nm, and the output coupler M2 had a transmission of 10 % at 1,890–2,170 nm. An external reflecting mirror M3 that was HR coated at 1,930–2,100 nm and AR coated at 760–810 nm was used to reflect the 2-μm output to the power meter. The Tm:YAG ceramic slab used in the pump wavelength tuning experiment was 1.29 mm thick (y), 10 mm wide (x) and 15 mm long with a doping concentration of 3 at. %. Both surfaces with dimension of 1.29 mm × 10 mm were AR coated at 760–810 nm and 1,750–2,050 nm. Micro-channel heat sinks made of oxygen-free copper were used to provide effective heat removal. The fin width and channel width were both 0.5 mm, with channel height and channel length to be 1 and 15.5 mm, respectively. The total cooling water flux was 5.3 L/min under a pump pressure of 0.6 MPa. The sample was sandwiched between two micro-channel copper heat sinks. Indium foil of about 0.1 mm thick was used to provide even thermal contact for efficient cooling. To release the mechanical stress and thermal stress in ceramic samples, two polished paddings were inserted into the space between heat sinks, as shown in Fig. 1. The thicknesses of the paddings were carefully chosen to evenly release stress in ceramic samples and guarantee an effective thermal contact between the sample and the heat sinks.

Fig. 1
figure 1

Experimental setup for pump wavelength tuning experiment and power scaling experiment

Full size image

3 Experimental results

The absorption spectrum of a 3 at.% Tm:YAG ceramic was measured using a spectrophotometer (LAMBDA 950, PerkinElmer), and the result is shown in Fig. 2. As shown in Fig. 2, the absorption peak of Tm:YAG is centered at 785.6 nm. It should be noted that although there do exist literature data for the fracture limit of the ceramic, these reported values were measured in the absence of external forces. However, in our experiment, the ceramic sample was sandwiched between two micro-channel copper heat sinks, and the existence of external forces was inevitable. Experiments should be taken to verify the fracture limit of the ceramic sample under current experimental conditions. When the wavelength of the laser diode shifted away from the absorption peak, the gain medium experienced smaller total absorption of pump radiation, the heat load density was also reduced at the same time, and a pump laser with higher power could be employed without causing mechanical failure. Hence, it is necessary to choose a suitable pump wavelength to balance these two effects in order to achieve the best laser performance. In order to investigate the influence of pump wavelength on laser performance, the wavelength of the pump laser was tuned by changing the temperature of cooling systems for laser diodes. Because the thermal fracture properties of Tm:YAG ceramic with different pumping wavelength also needed to be investigated, the influences of possible fluctuations in properties of different pieces of samples should be excluded; thus, a relatively small pump spot size of ~1.6 mm × 0.6 mm was chosen to make sure that several investigations of laser damage performance of Tm:YAG ceramic can be conducted on one piece of sample.

Fig. 2
figure 2

Absorption spectrum of 3 at.% Tm:YAG ceramic

Full size image

Laser performances with different pump wavelengths are shown in Fig. 3. Fractures occurred at the highest pump levels of all three conditions. In order to exclude the influence of the total absorption, the output power was plotted as a function of the absorbed pump power. The wavelength spans for three conditions are the variation in the center pump wavelength with pump power. It can be seen from Fig. 3 that when the wavelength of the pump laser was around the absorption peak (784.9–785.5 nm), the highest slope efficiency versus the absorbed pump power of 34 % was obtained. Thermal fracture occurred when the absorbed pump power was 66.7 W, and the corresponding output power of 2-μm laser was 19.2 W at this point. The beam quality at around 28 W absorbed pump power was measured with a beam profiler (NanoScan, Photon Inc.). The M 2 factors were calculated to be 2.19 and 1.13 in x and y directions, respectively. As the wavelength of LD shifted away from the absorption peak, the slope efficiencies versus the absorbed pump power dropped to 32.5 % for condition 2 and 30.4 % for condition 3. The 4 % difference in slope efficiency appeared to be within error margins, and it revealed that the influence of the pump wavelength was mainly attributed to the change in absorption efficiency. The highest output powers for condition 2 and condition 3 were 20.2 and 18.4 W, respectively. The maximum absorbed pump power density was then calculated to be 1.43 × 104, 1.27 × 104, 1 × 104 W/cm3, for condition 1, condition 2 and condition 3, respectively. Although there were differences between these calculated values, it was reasonable to conclude that the fracture limit of the ceramic sample at our current experiment condition was around 1.2 × 104 W/cm3. The data reported here gave a good indication of the fracture limit of Tm:YAG ceramic, and this was useful for anticipating the fracture limit in our further experiment including the power scaling experiment.

Fig. 3
figure 3

Laser performances with three different pump wavelengths

Full size image

A typical photograph of the fracture surface in ceramic sample is shown in Fig. 4a, and the SEM picture of the fractured ceramic is shown in Fig. 4b. It can be seen from Fig. 4b that the fracture mode of the Tm:YAG ceramic sample was mainly transgranular and that no residual pores can be found.

Fig. 4
figure 4

a Photograph of the fracture in ceramic sample. b SEM picture of the fractured ceramic

Full size image

For further power scaling, a telescope imaging system consisted of two cylindrical lens on the horizontal direction was used, as shown in Fig. 1. The focal length of the two lens was 200 and 100 mm, respectively. Combined with a third cylindrical lens with a focal length of 70 mm in the vertical direction, the pump beam was focused into a pumping line with four sigma beam widths of about 5.5 mm × 0.54 mm at the waist. The total cavity length was approximately 40 mm. The temperature of the cooling water was set to be 15 °C.

Laser experiments with different output couplings at 120-W pump level were conducted. The slope efficiencies for output couplers of 8 and 10 % were 29.3 and 30 %, respectively. According to the Caird analysis, the cavity loss was calculated to be 1.2 % for a round-trip, and the propagation loss of this gain medium at the laser wavelength was conservatively estimated to be 0.0023 cm−1.

The laser characteristic with an output coupler of 10 % transmission is illustrated in Fig. 5, and the corresponding beam profile of the output laser is shown in the Fig. 6. A maximum output power of 52 W was achieved at an incident pump power of 220 W, with a threshold pump power of 22 W. The slope efficiency with respect to incident pump power was 27.8 %, and the optical-to-optical conversion efficiency was 23.6 %. The maximum absorbed pump power density was then calculated to be 1.1 × 104 W/cm3, which was close to the fracture limit. The corresponding laser spectrum was monitored by an optical spectrum analyzer (AQ6375, Yokogawa), and the result is shown in Fig. 7. The central wavelength was 2,013.6 nm with a full width at half maximum (FWHM) of about 0.9 nm. As to our knowledge, this is the highest power reported for Tm:YAG ceramic laser up to present. It can be clearly seen that the slope efficiency experiences a slight drop when the pump power exceeds 200 W. For pump power lower than 200 W, the slope efficiency was calculated to be 29.1 %, and for higher pump power, the slope efficiency dropped to 14 %, leading to a total calculated slope efficiency of 27.8 %. As the pump wavelength at this power level was around the absorption peak of 786 nm, the degradation of laser performance was mainly attributed to the high temperature rise of Tm:YAG ceramic sample. Due to the laser transition terminating in the ground energy level of Tm ions, high temperature will lead to stronger reabsorption of 2-μm laser and a reduction in the population inversion, which will contribute to the deterioration of laser performance. By using heat sinks that can provide more efficient cooling, such as micro-channel heat sinks with more optimized parameters, the laser performance could be further improved and higher output power could be anticipated [20]. Besides, mode matching between the pump and signal field also has a great influence on the laser performance. Optimized design of the resonator is needed for further improvement of the slope efficiency. Optimizing the output coupling ratio and the size of the pump beam will also contribute to the improvement of laser performance. The M 2 factor of the output laser at a pump power of 70 W was measured. The M 2 factors were measured to be \(M_{\text{x}}^{2} = 94.6\) and \(M_{\text{y}}^{2} = 1.06\). As eigenmodes of higher orders will oscillate at higher pump level, the beam quality will deteriorate at higher pump levels. By increasing the beam size on the horizontal direction from 5.5 mm to around 12 mm and using a dual-end-pumping scheme, a maximum output power surpassing 100 W was feasible.

Fig. 5
figure 5

Output power versus incident power

Full size image
Fig. 6
figure 6

Beam profile of the output laser

Full size image
Fig. 7
figure 7

Laser spectrum of Tm:YAG ceramic slab laser

Full size image

4 Conclusion

In conclusion, we have successfully demonstrated an efficient diode-end-pumped high-power Tm:YAG ceramic slab laser. Careful thermal management was employed to ensure good extraction of waste heat and low-temperature excursions and to relieve the stress in ceramics. Various investigations have been taken to find the optimum conditions for high-power laser operation of such a quasi-three-level system. A maximum output power of 52 W at 2,013.6 nm was obtained at an incident pump power of 220 W, corresponding to a slope efficiency of 27.8 % with respect to incident pump power. As far as we know, this is the highest output power achieved in a Tm:YAG ceramic laser. The result proves that laser ceramic is a competitive candidate as the gain medium for high-power laser applications.