2 μm lasers continue to be of interest in medicine based on absorption by water in the human body [1] or by ${\mathrm{CO}}_2$ in remote-sensing applications [2] and are essential in the detection of wind vortices using lidar [3]. 2 μm lasers also have the added benefit of being eye safe [4]. The dual-wavelength operation of the 2 μm laser, if feasible, would allow for the efficient generation of terahertz waves with multihundred-watt peak power via a difference-frequency generation process [5].

Laser power scaling with good beam quality can be achieved via implementation of resonant pump/lase schemes and designing for optimal thermal management. This may include cryo-cooling, especially where efficiency is the most critical requirement [6]. While the Ho:YAG 2.1 μm laser meets the criteria of eye safety and emission in an atmospherically transparent window, ${\mathrm{YVO}}_4$ can be a thermally better choice of host for ${\mathrm{Ho}}^{3+}$ ions over YAG [7], especially for cryo-cooled applications. This was recently demonstrated in [8], where a fiber-coupled InGaAsP/InP diode laser module at $\sim 1930\mathrm{nm}$ was used to in-band pump a cryo-cooled ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ single crystal to produce 1.6 W of continuous wave (CW) output power at 2053 nm. Following this result, a room-temperature laser version of the ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ laser was demonstrated with resonant pumping by a much brighter Tm:YAP laser [9]. This laser emitted at 2041, 2053, or 2066 nm but was never shown to emit at multiple wavelengths simultaneously.

In this Letter, we present the experimental results of a cryo-cooled, resonantly pumped (at 1966 nm) ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ laser that emits simultaneously at 2053 and 2068 nm with a nearly quantum-defect-limited optical-to-optical efficiency of $\sim 92\%$. To the best of our knowledge, this is the highest demonstrated slope efficiency of any ${\mathrm{Ho}}^{3+}$-doped laser.

The ${\mathrm{Ho}}^{3+}(2\%)\text{:}{\mathrm{YVO}}_4$ gain element was cut from a single crystal grown by the Czochralski method. A section of a boule was selected based on minimal observable stress birefringence. This section was then cut and polished to form a $2.4\mathrm{mm}\times 5\mathrm{mm}\times 25\mathrm{mm}$ laser slab with the $c$-axis orientation perpendicular to the $5\mathrm{mm}\times 25\mathrm{mm}$ face. The two $2.4\mathrm{mm}\times 5\mathrm{mm}$ faces were antireflection (AR) coated for the wavelength range 1900 to 2100 nm. The reflectance at 2050 nm was measured by the vendor to be $\sim 0.35\%$ per surface. The gain element was mounted to the bottom of a standard liquid-nitrogen-cooled cryostat using highly thermally conductive indium cushioning.

The $c$-axis of the slab was chosen to coincide with the thin dimension of the gain medium to facilitate cooling as the thermal conductivity of ${\mathrm{YVO}}_4$ is at least 50% higher along the $c$ axis versus the $a$ axis ($60\mathrm{W}/\mathrm{m}*\mathrm{K}$ versus $40\mathrm{W}/\mathrm{m}*\mathrm{K}$ at 77 K) [7].

The simplified optical layout of the experiment is depicted in Fig. 1. The cryostat was set between a pair of cavity mirrors: a plano–concave (radius of curvature $[\mathrm{ROC}]=2.5\mathrm{m}$) dichroic $M1$ ($\mathrm{AR}\sim 1900$, $\mathrm{HR}\sim 2100\mathrm{nm}$) and a plano–concave ($\mathrm{ROC}=0.75\mathrm{m}$) output coupler $M2$ (reflectivity 71% to 88% at 2050 nm). The two mirrors were set 125 mm apart and formed a stable cavity. The cryostat windows were AR coated (1900 to 2100 nm). The transmission loss per cryostat window at 2050 nm was measured to be $\sim 0.1\%$. Transmission losses (2 μm) of the 25 mm long ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ gain medium were measured to be $\le 1\%$ per pass. We estimate that the overall round-trip cavity losses were less than 3%.

 

Fig. 1. Simplified layout of the experimental laser setup. All the components are properly listed in the body of the paper.

Download Full Size | PDF

In order to assess the ultimate laser potential of the ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ gain medium, the laser slab was end-pumped at 1966 nm by a Tm fiber laser collimated and focused by a 20 mm (L1) lens and a 200 mm (L2) lens, respectively. The pump beam at focus had a measured beam quality factor, $M^2=3.0\pm 0.1$, with a radius of $w_p=125\mathrm{\mu m}$. As the laser cavity mode radius $w_L=400\mathrm{\mu m}$ and the Rayleigh length was 5 mm, we calculated that the pump beam energy was fully utilized.

The $\pi$- and $\sigma$-polarized absorption cross sections are nearly equal at $\lambda \sim 1966\mathrm{nm}$ ($\sim 5*{10}^{-21}{\mathrm{cm}}^2$) [8]. This pump wavelength was chosen to properly utilize the unpolarized pump source. The relatively low absorption cross section favors more even pump power distribution along the length of the slab.

Laser experiments were conducted using a range of output coupler reflectivities, $R_{\mathrm{OC}}=75\%$, 81%, and 88%. As shown in Fig. 2, the results demonstrate high incident pump-to-laser output conversion efficiencies with all used output couplers. The most notable observation is the nearly quantum-defect-limited slope efficiency (92%) laser performance with dual-wavelength operation in a wide range of pump powers using the $R_{\mathrm{OC}}=81\%$ output coupler. Proper care was taken to separate the transmitted pump light from the laser output by introducing a 1° angular deviation between the pump and laser axis. It was found that under lasing conditions, nearly 99% of the pump power at 1966 nm was absorbed in the slab.

 

Fig. 2. Output power versus incident pump-power dependence for the cryo-cooled ${\mathrm{Ho}}^{3+}(2\%)\text{:}{\mathrm{YVO}}_4$ laser resonantly pumped at 1966 nm, measured with different output coupler reflectivities $R$: a) $R=75\%$, (b) $R=81\%$, and (c) $R=88\%$.

Download Full Size | PDF

The dual-wavelength laser operation of the cryo-cooled, resonantly pumped ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ laser was analyzed using a Yokogawa AQ6375 optical spectrum analyzer (OSA). Beginning with the case in Fig. 2(b), the laser starts lasing at 2053 nm in $\pi$ polarization at low pump power (2 to 5.5 W), transitions to a dual-wavelength regime with 2053 nm ($\pi$) and 2068 nm ($\sigma$ polarization) outputs in the pump power range 5.5 to 12 W, and ends with pure 2068 nm emission beyond 12 W. The polarization of the 2053 and 2068 nm wavelengths was determined by a diffraction grating and a polymer thin-film polarizer. No discernible difference in slope efficiency was observed in transition from a mixed $2053/2068\mathrm{nm}$ laser output to one of pure 2068 nm emission. We measured the output beam quality at full power ($>8\mathrm{W}$) and determined that the beam quality factor, $M^2$, was $\sim 5.5$ along both $a$ and $c$ axes. No astigmatism was observed.

In contrast, in the case in Fig. 2(a), $R_{\mathrm{OC}}=75\%$, the laser starts and maintains lasing at 2053 nm in the full range of available pump power. For the case in Fig. 2(c), $R_{\mathrm{OC}}=88\%$, the laser emits at 2053 nm only at threshold and quickly transitions to lasing at 2068 nm only.

An OSA trace of the CW dual-wavelength lasing is presented in Fig. 3. The trace was obtained using the outcoupler with 81% reflectivity and OSA resolution of 0.05 nm. The 2053 nm laser line was found to be significantly spectrally wider than the 2068 nm laser line, usually $\sim 0.17\mathrm{nm}$ versus $\sim 0.05\mathrm{nm}$ (resolution-limited), respectively, for the entire range of pump power. The OSA trace shown in Fig. 3 is stable in time, but the trace is the result of signal averaging over the period of time of $\sim 3\sec$. By simultaneous time-resolved power measurement at each wavelength, we observed that the power at each of the two wavelengths undergoes rapid fluctuations (on a millisecond time scale) in such a manner that both powers nearly anticorrelate in time.

 

Fig. 3. Laser output spectrum of the resonantly pumped, cryo-cooled ${\mathrm{Ho}}^{3+}(2\%)\text{:}{\mathrm{YVO}}_4$ laser at the onset of dual-wavelength lasing. Presented spectral distribution corresponds to output coupler reflectivity of 81% [Fig. 2(b)] and incident pump power $\sim 7\mathrm{W}$.

Download Full Size | PDF

While spectroscopy at this point does not allow us to predict exactly when the transition of lasing from 2053 to 2068 nm occurs, we can anticipate this laser behavior. As observed in our experiments, a strong correlation exists between the gain-medium temperature and output-coupler transmission value at the point when the output power at 2053 nm equals that at 2068 nm. For equality to hold, the laser gain at 2053 nm (${\sigma }_{g/2053}$) must equal the laser gain at 2068 nm (${\sigma }_{g/2068}$):

In conclusion, we have demonstrated the CW dual-wavelength operation of a cryogenically cooled, resonantly (in-band) pumped ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ laser with nearly quantum-defect-limited performance. The ${\mathrm{Ho}}^{3+}(2\%)\text{:}{\mathrm{YVO}}_4$ gain element, maintained at approximately 80 K and pumped by a Tm fiber laser at 1966 nm, emitted at the wavelengths of either 2053 or 2068 nm, or both simultaneously, as a function of the outcoupler transmittance and the incident pump power. It was observed that the onset of 2068 nm lasing is correlated with the heating of the gain medium and the outcoupler transmission value. We observed ${\mathrm{Ho}}^{3+}\text{:}{\mathrm{YVO}}_4$ laser operation at a mixture of 2053 and 2068 nm with a nearly quantum-defect-limited slope efficiency of $\sim 92\%$, which is, to the best of our knowledge, the highest slope efficiency demonstrated for any ${\mathrm{Ho}}^{3+}$-doped laser.