Deep red lights are widely used in medicine, biology, and chemistry. A photosensitizer, a benzoporphyrin derivative (BPD), has been a potent generator of singlet oxygen upon activation at 692 nm^[1]. BPD analogs probably owe their relative efficacy in the eradication of virus in whole blood to their activation wavelength (692 nm), which penetrates whole blood significantly more effectively than the activation wavelength (630 nm) of Photofrin. An aza-BODIPY photosensitizer (SAB) nano-particle (NP) exhibited strong absorption around 701 nm^[2]. Under 701 nm excitation, SAB NPs presented two narrow fluorescence emission peaks near 769 nm and 834 nm. The S1 state of perylene bisimide (PBI) has an absorption band at ∼705nm corresponding to the transition of the excited electron to the higher excited state as explained later^[3]. An amplified femtosecond laser, Spirit One 1040-8, was directed to a noncollinear optical parametric amplifier (NOPA), Spirit-NOPA-3H (Spectra-Physics), to generate 705 nm femtosecond laser pulses for the PBI pump laser. However, the structure of the NOPA is complex. For photodynamic therapy (PDT) treatment, a 705 nm laser was applied to irradiate the tumor cells. Under 705 nm laser irradiation, simultaneous PDT and photothermal therapy (PTT) destroyed tumor cells to generate danger associated molecular patterns, such as pro-inflammatory cytokines and chemotactic factors, which significantly promoted dendritic cell maturation, improved the infiltration of effector T cells, and rapidly released cytokines^[4]. In addition, enhanced photoluminescence and emission intensity were shown under co-excitation with 708 nm and 532 nm lasers, which can be utilized to realize a new modality of far-field super-resolution imaging^[5]. A laboratory-built Ti:sapphire chirped pulse amplification (CPA) system delivered 708 nm pulses as the pumping source, a gain bandwidth of more than one octave for a BBO crystal was obtained, and 0.73–optical–cycle pulses with a pulse energy of 32 µJ at 1.8 µm were generated^[6]. Similar to the NOPA, the 708 nm laser with CPA has a complex structure. The visible band laser can be extended by frequency conversion^[7], but the structure was also complex.

Direct generation of laser radiation has become the main part of laser research due to its high conversion efficiency and compact structure. Pr3+:YLF crystals have been considered to be an ideal material for direct generation of visible laser radiation because the transitions between energy levels are mainly in the visible range. In 1977, the first Pr3+:YLF laser at room temperature was achieved in blue at 479 nm (corresponding to the P30→H34 transition) using a pulsed dye laser at 444 nm^[8]. Since then, research on Pr3+:YLF crystal lasers has continued through today. In 2014, a cyan laser of 70 mW with a slope efficiency of 7% was realized at 491 nm by an optically pumped semiconductor laser (2ω-OPSL) at 480 nm^[9]. In 2008, a green laser of 4.30 W with a slope efficiency of 45% was reported at 522 nm by two OPSLs at 479 nm^[10]. In 2016, a laser of 1.70 W with a slope efficiency of 49% was realized at 522 nm using a multi-mode InGaN diode laser (LD) at 444 nm^[11]. In 2022, an orange laser of 3.28 W with a slope efficiency of 34.2% at 604 nm was obtained by LDs^[12]. In 2020, an orange laser of 4.88 W with a slope efficiency of ∼49% at 607 nm was demonstrated at an absorbed pump power of 12.15 W^[13]. In 2021, a red laser of 8.14 W with a slope efficiency of 51.5% at 639 nm was achieved by LDs of 24 W^[14]. Furthermore, deep red lasers of 1.36 W, 3.11 W, 1.45 W, 4.50 W, and 1.90 W with slope efficiencies of 15%, 31.4%, 17.8%, 41.5%, and 21.8% were reported at 696.6 nm, 698.6 nm^[15], 718.5 nm, 720.8 nm^[16], and 729 nm^[17] by LDs, respectively. It was found that all the above lasers come from the P30 transitions with 479 nm and 491 nm from P30→H34, 522 nm from P30→H35, 604 nm and 607 nm from P30→H36, 640 nm from P30→F32, 696 nm and 698 nm from P30→F33, and 719 nm, 721 nm, and 729 nm from P30→F34. However, deep red lasers with transitions from I16 to F34 in Pr3+:YLF crystal have not been reported so far. This may be due to the fact that the population in I16 is less than that in P30, resulting in a weaker fluorescence intensity and a higher threshold. In addition, the peak emission lines from the I16 transition are not obvious and are ignored. We are interested in the study of the lasers from the I16 to F34 transition because it is helpful to expand the lasers by adding transitions between energy levels.

In this study, Pr3+:YLF single-wavelength continuous-wave lasers from I16 to F34 transitions are reported for the first time. Maximum output powers of 2.44 W, 2.10 W, 2.01 W, and 2.42 W are obtained at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm, with slope efficiencies of 19.8%, 16.5%, 15.8%, and 19.4%, respectively. It is useful to obtain new laser radiation using additional transitions among energy levels.

The energy levels in Pr3+:YLF were plotted as shown in Fig. 1 according to Ref. [18]. Upon excitation at 444 nm, the H34 ground state populations were pumped to the excited state P32, followed by a radiation-free transition to the upper laser levels P31, I16, P30, then a radiation transition to the lower laser level F34, and finally a rapid decay to the ground state. The lasers at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm came from I16→F34 transitions, i.e., 21,401cm−1 to 6942cm−1, 21,401cm−1 to 7142cm−1, 21,100cm−1 to 6920cm−1, and 21,083 cm⁻¹ to 6983cm−1, respectively. Under thermal equilibrium conditions, according to the Boltzmann statistics, N(I16,21,083cm−1)/N(P30,20,860cm−1)=34% (at 300 K), the population distribution of the upper laser levels was mainly on the level of P30. Therefore, the main challenge of generating lasers from I16 transition came from the P30 level with more populations.

Based on the fluorescence spectrum with 444 nm excitation, the emission cross sections in Pr3+:YLF crystal were calculated using the F-L method^[19] as shown in Fig. 2. Obviously, the emission cross sections in the π direction are several times larger than those in the σ direction, and the lasing threshold in the π direction is much lower, making it easier to generate lasers, so we took the π direction wavelength as the object of study. The two wavelengths (698 nm, 721 nm) with larger emission cross sections are from the transitions of P30, while the four wavelengths (691.7 nm, 701.4 nm, 705.0 nm, 708.7 nm) with smaller emission cross sections are from the transitions of I16. The emission cross sections at 698 nm and 721 nm are 1.07×10−19cm2 and 1.78×10−19cm2. The emission cross sections at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm are 0.15×10−19cm2, 0.15×10−19cm2, 0.13×10−19cm2, and 0.15×10−19cm2, respectively. Apparently, the emission cross sections from P30 are 7–12 times larger than those from I16. Thus, the problem facing the operation of lasers with smaller emission cross sections is the suppression of excitation at 698 nm and 721 nm. This is consistent with the results of the energy level analysis.

In order to realize lasers with I16→F34 transitions, an experimental setup was constructed as shown in Fig. 3. An InGaN laser diode array with a maximum output power of 24 W and a central wavelength of 444 nm (linewidth of about 2 nm) was used as the pump source. The output of the pump source is a collimated beam of approximately 5.5mm×3.8mm with vertical polarization. The quality factor of the pump beam is approximately Mx2=47 and My2=16. To match the pump mode to the cavity mode, a convex lens (focal length 75 mm) was used to focus the pump beam to a waist radius of about 100 µm. To design a compact laser, a linear resonant cavity was selected for experiment. The cavity consisted of M1 and M2 (100 mm radius of curvature). M1 was a plane mirror with high transmittance (>98.3%) at 444 nm to improve pumping efficiency and high reflectance (>99.8%) for light at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm to reduce intracavity losses. M2 was a concave mirror with high reflectivity (>99.8%) at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm to reduce intracavity losses. The reflectance of M1 (M2) at 698 nm and 721 nm was 99.99% and 99.99% (99.6% and 97.4%), respectively. We had hoped that M2 would have a high transmittance at 698 nm and 721 nm to suppress its excitation. In fact, it is almost impossible for the M2 to achieve high reflectivity at 691 nm and 701 nm and high transmittance at 698 nm at the same time. To suppress the excitation at 698 nm and 721 nm, a Fabry-Perot etalon as a frequency-selective device was considered but quickly ruled out because of its low rejection ratio. Fortunately, the transmittance of a birefringent filter (BF) can theoretically vary from 0 to 100%^[20]. Therefore, a 2 mm thick BF with a high suppression ratio was chosen to suppress the excitation at 698 nm and 721 nm. To reduce the losses, the BF was inserted into the resonant cavity at a Brewster angle. To measure the power of the signal, a commercial blue light filter was used to filter the residual pump light.

The gain medium was a Pr3+:YLF crystal that was optically polished at both ends. Pr3+:YLF crystals of different lengths and doping concentrations were used in the experiments, such as 3 mm × 3 mm × 7 mm with 0.5% (atomic fraction) doping and 3 mm × 3 mm × 20 mm with 0.15% doping. Unfortunately, the 3 mm × 3 mm × 7 mm crystal with 0.5% doping was damaged due to severe thermal effects. Therefore, lower doping concentrations (0.15%) of Pr3+:YLF were chosen to reduce thermal effects. To increase the absorption efficiency, a 20 mm length Pr3+:YLF crystal was used. The crystal with π polarization absorption efficiency of about 65.4% was wrapped on the side by indium foil and placed in a water-cooled copper block to dissipate heat. The temperature of the water-cooled copper block was kept at 16°C throughout the experiment.

When the cavity length was optimized to 96 mm, the average laser spot radius was about 94 µm. Considering the negative thermal lens of the crystal, it was possible that the laser mode matches the pump light well at this point so that the best output was obtained as shown in Fig. 4. The full width at half-maximum (FWHM) of the spectrum was measured to be 0.05 nm at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm using an Advantest Q8384 optical spectrum analyzer (resolution of 0.05 nm), as shown in Fig. 4(a). Since the spectral linewidth is equal to the resolution of the optical spectrum analyzer, the actual linewidth may be less than 0.05 nm. Such a narrow spectrum may be attributed to the use of the BF, the competition among modes in the cavity, and the laser operating at several times the threshold power. At the same time, the maximum output power of the laser was obtained when the absorbed power of the crystal was about 15.6 W (corresponding to a pump power of 24 W). Using an optoelectronic power meter (Thorlabs S425C-L), output powers of 2.44 W, 2.10 W, 2.01 W, and 2.42 W were demonstrated at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm as shown in Fig. 4(b), with slope efficiencies of 19.8%, 16.5%, 15.8%, and 19.4%, respectively. The corresponding threshold powers were 3.40 W, 3.40 W, 3.40 W, and 3.78 W, respectively. In addition, both M1 and M2 have high reflectivity for the laser, so the reflected output on both sides of the BF is the output of the laser.

To assess the stability of the power, the output power was recorded over 150 min (10 min interval) as shown in Fig. 5. Minimum powers of 2.41 W, 2.08 W, 1.99 W, and 2.39 W, maximum powers of 2.54 W, 2.20 W, 2.11 W, and 2.46 W, and average powers of 2.47 W, 2.14 W, 2.05 W, and 2.42 W were obtained at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm, respectively. From Fig. 5, standard deviations of 0.045 W, 0.046 W, 0.041 W, and 0.021 W and peak-to-peak power fluctuations of 2.47±0.06W (5.3%), 2.14±0.06W (5.6%), 2.05±0.06W (5.8%), and 2.42±0.03W (2.9%) were calculated.

To evaluate the beam quality, the beam diameters in the horizontal and vertical directions were measured at different positions on both sides of the beam waist by a charge coupled device (CCD). The beam quality factors M2 of the laser were calculated using ISO-11164. As shown in Fig. 6, beam quality Mx2 and My2 factors were estimated to be 2.29 and 2.03 at 691.7 nm, 2.23 and 1.86 at 701.4 nm, 2.31 and 2.08 at 705.0 nm, and 2.41 and 2.04 at 708.7 nm. It can be inferred from the M2 factors that the output may include some higher order modes. The laser spots captured by the CCD were embedded as shown in Figs. 6(a)–6(d), from which it can be seen that the output beams are approximately circular Gaussian light.

Continuous wave lasers with transitions from I16 to F34 were demonstrated in Pr3+:YLF crystal for the first time. The lasers at 691.7 nm, 701.4 nm, 705.0 nm, and 708.7 nm were successfully achieved, with output powers of 2.44 W, 2.10 W, 2.01 W, and 2.42 W and slope efficiencies of 19.8%, 16.5%, 15.8%, and 19.4%, respectively. Besides the careful tuning in the experiments, the following points are worth mentioning. Faced with the problem of small emission cross sections and high thresholds at new wavelengths, a high-power pump source (24 W LD array) was used, which inevitably brought about serious thermal effects. Experiments showed the use of crystals with low doping concentration to improve the damage resistance threshold of the crystal, which is the first point worth noting. The second point is the design of a closed cavity with a simple structure consisting of two mirrors to reduce losses. The third point is the use of the BF with a high suppression ratio in the case where the coating makes it difficult to suppress the 698 nm, 721 nm excitations. In conclusion, the method of carefully designing an additional energy level transition to excite the generation of laser radiation with new wavelengths has been successful. We expect that this method will be widely used to develop lasers with a possibility to generate radiation with new wavelengths.