Visible solid-state lasers have significantly advanced in power and beam quality, owing to the development of pump sources and crystals^[1,2]. The commercially available laser diode (LD)-pumped sources, having the advantages of compactness, high power, and cost-effectiveness, are favored by researchers. Of the many approaches to generate visible lasers, ${\mathrm{Pr}}^{3+}$-doped materials, one of the successful candidates for high-power efficient solid-state lasers that provide rich laser transitions have aroused much interest in recent years. Various ${\mathrm{Pr}}^{3+}$-doped materials such as the fluoride crystals^[3–10] (${\mathrm{Pr}}^{3+}\text{:}{\mathrm{LiYF}}_4$, ${\mathrm{Pr}}^{3+}\text{:}{\mathrm{LiGdF}}_4$, ${\mathrm{Pr}}^{3+}\text{:}{\mathrm{BaY}}_2{\mathrm{F}}_8$, ${\mathrm{Pr}}^{3+}\text{:}{\mathrm{KY}}_3{\mathrm{F}}_{10}$, ${\mathrm{Pr}}^{3+}\text{:}{\mathrm{LiLuF}}_4$, etc.), and the oxides crystals^[11–15] (${\mathrm{Pr}}^{3+}\text{:}{\mathrm{YAlO}}_3$, ${\mathrm{Pr}}^{3+}\text{-}{\mathrm{Mg}}^{3+}\text{:}{\mathrm{SrAl}}_{12}{\mathrm{O}}_{19}$, etc.) have been extensively researched. Fluoride crystals have lower crystal field strength and phonon energy compared to oxides, which have greater potential for the development of laser operation. As one of the most common materials for visible lasers, the ${\mathrm{Pr}}^{3+}\text{:}{\mathrm{LiYF}}_4$ (YLF) crystal has been shown to feature excellent optics, reasonable mechanical properties, and is a well-known laser gain medium with abundant laser transitions in the visible spectral region^[16–21].

In particular, the lasers at 696 and 698 nm in the deep red region have served many different purposes. For instance, the method of obtaining ultraviolet (UV) laser by direct frequency doubling of the deep red laser is more compact and has a higher conversion efficiency than the two-stage nonlinear process to generate a UV laser, which is more suitable for obtaining higher-power UV lasers^[22–26]. Additionally, the 696 nm lasers can be used for the spectral analysis of the methanol chlorophyll equation^[27], the enzymatic reaction steps of the photoreduction of prochlorophyll-ester^[28], and the improvement of the picture resolution of the Cr(III)tris-oxalato complex^[29]. The lasers at 698 nm are important not only for research into high-precision atomic clocks^[30], but also for nondestructive, real-time, or portable potato quality measurements^[31].

To the best of our knowledge, the maximum output power at around 698 nm was reported to be 1.5 W using an optically pumped semiconductor laser (OPSL) as the pump source^[32]. However, there are currently few reports of $\sim 696\mathrm{nm}$${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ lasers, due to the difficulty of generating lasers with relatively small stimulated emission cross sections. The output power reported at present was relatively lower, with the highest output power of only 116 mW^[20,33,34]. Therefore, the deep red lasers based on ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ crystals still have the potential to achieve power scaling.

In this paper, we demonstrate a compact and efficient InGaN LD-pumped ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ crystal to generate deep red lasers at 698.6 nm ($\mathrm{\pi }$-polarization) and 696.6 nm ($\mathrm{\sigma }$-polarization). To obtain the lasers in both polarization directions, a Lyot filter was utilized as a birefringent filter to suppress the oscillation of other lasers in the cavity. The maximum output power at 698.6 nm was up to 3.11 W, with a slope efficiency of 31.4%. For the $\sigma$-polarization 696.6 nm laser, a maximum output power of 1.36 W was achieved, with a slope efficiency of 15.0%. Notably, both output powers are significantly increased compared to previous reports, and without saturation. This is the first time to date that the output power of the LD-pumped ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ 696 nm lasers broke into the watt level. Furthermore, theoretical models of the input–output power characteristics were built to explain the results for both wavelengths.

The schematic setup of the LD-pumped ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ continuous-wave (CW) laser is depicted in Fig. 1. A blue InGaN laser diode with a maximum output power of 24 W was used as the pump source, giving a peak wavelength of approximately 444 nm. As calculated from the relevant parameter manuals, the ${\mathit{M}}^2$ factors of the pump source were 46.9 and 15.5 in the $x$ and $y$ directions, respectively. The pump beam was focused into the laser crystal by a plano–convex focusing lens with a focal length of 50 mm. The laser gain medium is a 15-mm-long a-cut ${\mathrm{Pr}}^{3+}$-doped YLF crystal with $3\mathrm{mm}\times 3\mathrm{mm}$ polished facets (Jiadong Optical Inc., Hefei, Anhui, China). It used the Czochralski method of growth, and the doping concentration is ∼0.2% (atomic fraction). A typical plano–concave cavity consisted of an input plane mirror (IM) and an output plano–concave coupler (OC) with a radius of curvature $R=100\mathrm{mm}$. For the 698 nm laser, the length of the physical cavity was optimized to 98 mm, while at 696 nm, it corresponded to 96 mm.

Figure 2 displays the transmission curves of the IM and OC. The IM has high transmission from 400 to 630 nm and high reflection ($>99.9\%$) from 690 to 800 nm. The coating of the OC used in the experiment was designed and fabricated in our laboratory using plasma direct-current sputtering technology. To suppress 640 nm laser emission, the IM and OC were coated with high transmissivity of 96.3% and 79.5% at 640 nm, respectively. The OC has a low transmissivity of 2.2% at 696 nm and 2.1% at 698 nm. To protect the crystal as well as laser cavity stability from thermal effect, it was wrapped in indium foil and placed in a copper block, which was kept at 16°C by a water-cooled cooler. The laser crystal was placed close to the IM. The full width at half-maximum (FWHM) of the pump laser spectrum was $\sim 2.2\mathrm{nm}$, resulting in a relatively low absorption efficiency of 47% for the crystal.

We inserted a 2-mm thickness Lyot filter with a Brewster angle into the laser resonator to select the wavelength. The operation performance of the single-wavelength 696.6 and 698.6 nm laser was optimized, respectively. To measure the output power, filters with high red transmittance and high blue reflectance were placed behind the output to filter out the pump light. The laser spectra in the experiment were measured by an optical spectral analyzer (Ocean-Optics HR4000+), with a resolution of about 0.25 nm.

The ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ crystal has various emission lines, for which partial energy level scheme is illustrated in Fig. 3. The 444 nm emission peak of blue InGaN LD matches the ground state absorption of ${{}^3\mathrm{H}}_4\to {{}^3\mathrm{P}}_2$. The lasers at 696.6 and 698.6 nm were based on a four-level system operation. The population is excited from the ground-state energy level (${{}^3\mathrm{H}}_4$) to the ${{}^3\mathrm{P}}_2$ level by the pump source, and then rapidly relaxes to the metastable ${{}^3\mathrm{P}}_0$ level (denoted by ${\mathrm{\Gamma }}_1$). The ${{}^3\mathrm{F}}_3$ energy level is split into several Stark sublevels due to crystal field effect^[35]. Following the electric-dipole selection rules, the lasers at 696.6 nm ($\mathrm{\sigma }$-polarization, ${\mathrm{\Gamma }}_1\to {\mathrm{\Gamma }}_{3,4}$) and 698.6 nm ($\mathrm{\pi }$-polarization, ${\mathrm{\Gamma }}_1\to {\mathrm{\Gamma }}_2$) are generated by the transition from the ${{}^3\mathrm{P}}_0$ level to the ${{}^3\mathrm{F}}_3$ different sublevels.

Figure 4 depicts the laser output characteristics. The spectra were measured at the maximum output power with emission peaks at 698.6 and 696.6 nm, respectively [see Figs. 4(b)–4(d)]. According to the previous studies on spectral analysis,^[4] the emission cross sections at 696 and 698 nm are smaller relative to 721 nm (${{}^3\mathrm{P}}_0\to {{}^3\mathrm{F}}_4$). In addition, the positions at 696 and 698 nm are relatively close to the position at 721 nm in the spectrum. Thus, it is difficult to suppress the laser transition at 721 nm with the mirror manufactured in our laboratory. The Lyot filter with a Brewster angle was inserted into the cavity and rotated by a certain angle to obtain the $\mathrm{\pi }$-polarized 698.6 nm laser [see Fig. 4(a)]. The maximum output power was 3.11 W with a threshold power of 0.9 W; the slope efficiency was 31.4%. The ${\mathit{M}}^2$ factors were measured to be 3.1 and 2.3 in the $x$ and $y$ directions, respectively [Fig. 5(b)]. Similarly, by tuning the Lyot filter to the optimal position, we achieved the 696.6 nm laser in the $\mathrm{\sigma }$-polarization direction with a maximum output power of 1.36 W, a slope efficiency of approximately 15.0%, and a threshold power of 2.44 W, as shown in Fig. 4(c). The beam ${\mathit{M}}^2$ was 1.8 and 2.2 in the $x$ and $y$ directions, respectively [Fig. 5(a)]. The embedded output beam images were captured by a CCD. It is worth emphasizing that due to the insertion of the Lyot filter in the resonant cavity, we measured the laser power as a triple-end output. The tail-end output power is 1.23 W (696.6 nm) and 1.33 W (698.6 nm) under the maximum input power, respectively. As can be seen in Figs. 5(c) and 5(d), the stability of the output power was recorded every 5 min for a monitoring time of 1 h. The corresponding power fluctuations at 696.6 and 698.6 nm were 0.46% and 0.52%, respectively. In addition, the maximum output powers are the highest known and have good linearity with no saturation.

To better understand the experimental results, we performed some theoretical simulations on the relationship between output power and absorbed pump power. For a four-level laser system, the input–output characteristics can be determined by^[36]$$P_{\mathrm{in}}=\frac{A_e\gamma I_{\text{sat}}}{{\eta }_p}{\left[{\int }_a\frac{\in (x,y,z)g(x,y,z)}{(2P_{\text{out}}/T{A}_e{I}_{\text{sat}})\in (x,y,z)+1}\mathrm{d}V\right]}^{-1},$$(1)where ${\eta }_p$ indicates the pumping efficiency, $\gamma$ indicates the logarithmic loss per pass, $A_e$ indicates the effective mode area of the lasing mode, and $I_{\mathrm{sat}}$ indicates the saturation intensity. The power transmission in the laser is denoted with $T$, and the normalized pump distribution inside the active material is denoted with $g$. $P_{\text{in}}$ and $P_{\mathrm{out}}$ are the input and output power, respectively. The equation for the dimensional mode distribution factor of the laser beam ($\in$) is $$e^{\frac{-2(x^2+y^2)}{{\omega }^2}},$$(2)where $\omega$ represents the laser waist radius. When considering $P_{\mathrm{out}}$ is limited to zero, we can write the equation as $$P_{\mathrm{th}}=\frac{A_e\gamma I_{\text{sat}}}{{\eta }_p{\int }_a\in g\mathrm{d}V},$$(3)where $P_{\text{th}}$ is the threshold power, and the integral ${\int }_a\in g\mathrm{d}V$ considers the spatial overlapping of pump and mode distribution.

For a circular Gaussian laser beam, by introducing the parameter $\alpha ={(\omega /{\varpi }_p)}^2$, some analytical solution can be obtained, $$P_{\text{out}}=\frac{T}{2\gamma }{\eta }_p\frac{\alpha (\alpha +2)}{{(1+\alpha )}^2}[P_{\text{in}}-P_{\text{th}}(\alpha )],$$(4)$$P_{\mathrm{th}}=\frac{\pi \gamma I_{\text{sat}}{\varpi }_p^2}{2{\eta }_p}(1+\alpha ),$$(5)where ${\varpi }_p$ is the average pump light waist radius ($\omega ={\varpi }_p$). Through Eqs. (4) and (5), the output powers and thresholds of the ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ lasers can be estimated and calculated based on the emission cross sections and lifetime as well as the laser geometer. The parameters we used in the simulations are listed in Table 1.

As can be seen from Table 1, the laser emission cross section at 698 nm is $9.79\times {10}^{-20}{\mathrm{cm}}^2$, which is much larger than $1.51\times {10}^{-20}{\mathrm{cm}}^2$ at 696 nm. But the threshold power of the two wavelengths only differs by a factor of over 2. This is owing to the saturation intensity of the 696 nm which is 7 times more than that of 698 nm. The 696.6 nm laser with smaller emission cross section requires a decrease in the average pump light waist radius to reduce the threshold power of the laser, and reducing the average pump size is beneficial in improving the output power. The difference in loss ($\gamma$) between the two wavelengths is mainly attributed to the different angles of the adjusted Lyot filter. Thus, the high emission cross section and low cavity loss are key parameters to achieve high power and high slope efficiency.

In summary, we reported LD-pumped ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ CW deep red lasers. By inserting a Lyot filter and optimizing the laser resonant cavity, the $\mathrm{\pi }$-polarized 698.6 nm laser and $\mathrm{\sigma }$-polarized 696.6 nm laser were obtained, respectively. For $\mathrm{\pi }$-polarization, the maximum output power was up to 3.11 W. The slope efficiency was 31.4%, and no output saturation power was observed. A similar trend also existed for the $\mathrm{\sigma }$-polarization: a maximum output power of 1.36 W was measured with a slope efficiency of 15.0%. To the best of our knowledge, the output power of these two wavelengths is by far the highest for an LD-pumped ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ crystal. This represents the first time to date that the output power of an LD-pumped ${\mathrm{Pr}}^{3+}\text{:}\mathrm{YLF}$ laser at 696 nm reached the watt level, which is 10 times higher than the previously reported. Moreover, theoretical simulations were performed to explain the relationship between the output laser power and the absorbed pump power, which were consistent with the experimental results. With the continuous development of solid-state deep red lasers, it is expected to expand the application prospects in the visible region and achieve high-power UV lasers by frequency doubling.