Comparative study of Yb:KYW planar waveguide lasers Q-switched by direct- and evanescent-field interaction with carbon nanotubes

: Both direct- and evanescent-field interactions with carbon nanotubes (CNTs) are applied to achieve stable Q-switched operation of Yb:KYW planar waveguide lasers. The performance characteristics were investigated in a same cavity configuration and analyzed in detail in the following three cases, CNTs deposited onto end mirror (M-coating), output coupler (OC-coating) and top surface of the planar waveguide (WG-coating). Maximum output powers, repetition rates, and minimum pulse durations are 61 mW, 1103 kHz and 215 ns for OC-coating, 39 mW, 1052 kHz and 275 ns for WG-coating, and 26 mW, 1119 kHz and 217 ns for M-coating, respectively. From the calculation of the configuration-dependent stability range, the beam size and the electric field distribution in the Yb:KYW and the calculated beam size at M and OC, we extract the maximum intracavity peak intensities at SA as ~47 kW/cm 2 for OC-coating and 860 kW/cm 2 for M-coating. The range of cavity length tolerance for Q-switched operation are measured to be 1.2 mm for OC-coating and 8.8 mm for M-coating. This results confirm well the calculated intensities at SA. interaction with CNTs. A detailed study is carried out by comparing the maximum intensity at the SA for each operation mode and the laser performance. From the experimental results, and the calculated intracavity beam size and electric field distribution, it is confirmed that the intensity at the SA, required for the evanescent-field interaction, is about 1500 times lower than that of the typical direct-field interaction in the presented waveguide lasers. We believe that this result is meaningful as the first quantitative analysis of evanescent-field interaction scheme in a Q-switched solid-state waveguide laser. In addition, the evanescent-field interaction scheme offers an efficient

rare-earth ion-doped monoclinic double-tungstate host material KY(WO 4 ) 2 (KYW), which exhibits large refractive indices, high third-order nonlinearity, large absorption and emission cross section, broad linewidths and minimal quenching effects, has been investigated for bulk and waveguide amplification [20]. Thanks to the superior characteristics of Yb:KYW and the waveguide geometry, Yb:KYW planar waveguide lasers operating near 1 μm have been successfully demonstrated in the cw and pulsed regimes [10,11,21]. However, the low refractive index contrast due to the low doping concentration of the active ions sometimes leads to a relatively thick planar waveguide. This thick planar waveguide structure may lead to low beam intensity and multimode operation in the waveguide. To overcome such a low refractive index contrast, Gd 3+ and Lu 3+ co-doped KYW host materials have been proposed and widely investigated [22][23][24][25][26]. By substituting the low electron density ions of Y 3+ for the higher electron density ions of Gd 3+ and Lu 3+ , refractive index contrast can increase while maintaining the doping ratio of active ions.
Saturable absorbers (SAs) applicable as passive optical switching devices are commonly used to realize more compact pulsed waveguide lasers. However, most SAs employed in waveguide lasers suffer from the damage problem due to the small beam size corresponding to the waveguide scale. To overcome this issue, the evanescent-field interaction scheme with bis(4-dimethylaminodithiobenzil)nickel (BDN) dye as SA deposited on top of waveguides was proposed [8,9]. Due to the low intensity of an evanescent-field at the SA, this scheme does not only avoid the optical damage and thermal problems of the SA, but is also suitable for a compact and integrated pulsed laser source, because the position of the SA does not disturb the cavity configuration. Recently, evanescent field interactions with SAs based on carbon nanotubes (CNTs) and 2D materials such graphene, tungsten disulfide (WS 2 ) and black phosphorous in waveguide lasers have been investigated [10][11][12][13][14][15][16]. However, most studies report the performance of the specific pulsed laser itself and comparative analysis of the direct-and evanescent-field interaction schemes has not been carried out, yet.
In this work, we investigated both evanescent-field and direct-field interaction schemes of the Q-switched Yb:KYW planar waveguide laser for a comparative study. CNTs are used as SAs, since they can be easily integrated at different positions, and their properties including linear and nonsaturable losses and modulation depth are precisely controllable. In the same cavity configuration, CNT-SAs are positioned on the HR-end mirror (M-coating) and output coupler (OC-coating) for direct-field interaction, respectively, and on the top surface of the Yb:KYW planar waveguide (WG-coating) for evanescent-field interaction. All configurations show stable Q-switched operation with typical tendencies for a saturable absorber Q-switched laser. Based on the calculation of the beam mode and the amplitude ratio of the electric field distribution within the Yb:KYW planar waveguide, maximum intensities at the SA in each case are compared. It turned out that the maximum intensity at the SA for the WG-coating is about 1500 times lower than that for the M-coating case.

Fabrication of waveguide and CNT-SAs
The Yb:KYW planar waveguide was fabricated by the liquid phase epitaxy (LPE) method as described in [26,27]. First, the undoped KYW substrate was grown by the top-seeded solution growth (TSSG) slow-cooling method. It was cut parallel to the (010) natural face and polished to laser grade quality. The active layer with a composition of KY 0.59 Gd 0.185 Lu 0.21 Yb 0.015 (WO 4 ) 2 (1.5 at.% Yb) was grown by LPE from a solution with solute/solvent ratio of 7/93 mol%. The precursors used were Y 2 O 3 , Gd 2 O 3 , Lu 2 O 3 , Yb 2 O 3, WO 3, and K 2 CO 3 . Potassium ditungstate, K 2 W 2 O 7, was used as a solvent. In this case, Gd 3+ and Lu 3+ ions were co-doped to reduce the lattice mismatch, increase the refractive index difference and decrease the thickness of planar waveguide, because Gd 3+ and Lu 3+ exhibit large atomic numbers compared to Y 3+ . Subsequently, the cleaned 2-mm-thick KYW substrate was vertically dipped into the solution, rotating it at 10 rpm. Once the growth process was completed, the sample was slowly extracted, fixed 3 mm above the solution and cooled to room-temperature at a rate of 1.5 K/h to avoid any cracks and thermal stress between the substrate and the epilayer. At this time, we did not deposit a top-cladding-layer on the planar waveguide. The lateral side was cut perpendicular to the N g axis of the beam propagation direction and each end facet and top surface of the planar waveguide were polished to laser quality. After polishing and cutting, we prepared a 10.7-μm-thick, 5.75-mmlong Yb:KYW planar waveguide. The overall loss arising from coupling, absorption and other propagation losses through the waveguide was measured to be about 0.95 dB.
The fabrication process of CNT-SAs follows the one described in [28]. CNTs synthesized by the arc-discharge method exhibiting broadband absorption around 1 μm are chosen as starting material. These CNTs are dissolved in 1,2-dichlorobenzene (DCB) and the welldispersed upper portion of CNTs is extracted after centrifugation. The CNT-DCB solution is then mixed with poly(methyl methacrylate) (PMMA). The prepared CNT/PMMA mixture in DCB is finally spin-coated on output coupler, end mirror and top surface of the waveguide. Subsequently, the SA layers were dried on a hot plate. To characterize saturable absorption properties of CNTs such as saturation fluence, modulation depths and nonsaturable loss, we performed fluence-dependent nonlinear transmission measurements of CNTs spin-coated on a quartz substrate. Here, the layer thickness and the concentration of CNTs are kept to be same for the case of OC-coating. From this measurement near 1 μm, we extract saturation fluence of 0.65 μJ/cm 2 , modulation depth of 0.67% and nonsaturable loss of 0.91%. Note that the coating conditions for each scheme are further optimized to achieve stable Q-switched operation. In the case of waveguide coating, the end facet of the waveguide is protected from the CNT coating to avoid any undesired side effects and disturbances on the laser operation.

Experimental results
The schematics of the planar waveguide laser configurations are shown in Fig. 1. A 980-nm single-mode laser diode emitting a maximum cw power of 750 mW (1999 CHP, 3S Photonics) is coupled to a polarization-maintaining fiber and collimator. To prevent possible damages of the laser diode by feedback from the waveguide laser, a Faraday isolator is used. A half-wave plate is placed after the Faraday isolator to match the pump polarization to the N m -axis of the planar waveguide for the efficient laser operation. Note that the absorption and emission cross sections along the N m -axis are much larger than those of other axes. To couple the pump beam into the planar waveguide efficiently, two cylindrical lenses with focal lengths of 150 and 30 mm are used. The pump beam size was measured by the knife edge method and the beam waists are 25.48 μm along the x-axis (parallel to the N m waveguide axis) and 5.07 μm along the y-axis (parallel to the N p waveguide axis). The latter value is well matched to the thickness of the used waveguide. The output coupler (OC) with 4% transmission at the laser operation wavelength was optimal for a comparative study of Qswitched operation in direct-and evanescent-field interaction schemes. An AR-coated aspheric lens with a focal length of 4.51 mm (C230TME-B, Thorlabs) is inserted between the planar waveguide and OC for compensation of the substantial difference of the beam propagation characteristics between the xz-and yz-planes. In this configuration, the role of the lens in the xz-and yz-planes is different. In the xz-plane, the lens increases the stability region and helps the overlap between the pump and cavity mode for efficient laser operation. In the yz-plane, the lens collimates the divergent beam after the waveguide and re-couples the returned beam to the waveguide.    The spectrum in stable Q-switched operation for each SA is shown in Fig. 3(a). The center wavelength and spectral bandwidth at FWHM are 1025 nm and 1.3 nm, 1026 nm and 1.4 nm, and 1026 nm and 1.8 nm for OC-, WG-, and M-coating, respectively. The beam profiles shown in Fig. 3(b) are the far-field profiles measured at a distance of 20 cm from OC for the three different cases and closely related to the near-field beam profiles. Along the vertical direction (y-axis), the divergence of the beam is collimated by an intracavity lens, while the beam in the horizontal plane (x-axis) is focused on OC. The difference of the beam profiles is mainly attributed to the cavity-length-dependent x-axis beam size at OC shown in Fig. 4(b). Note that, from the mode number calculation, the waveguide structure allows two modes in the vertical direction, but we could experimentally avoid spatial multi-mode operation by proper selection of the pump beam size and the intracavity lens as well as precise alignments of the lens position and the cavity length. Figures 3(c) and 3(d) show the pulse train measured for each case in different time spans and display stable Q-switched operation.

Discussion
To verify the direct-field interaction in the Q-switched Yb:KYW planar waveguide, the maximum intensity at the SA is calculated from the experimental results and the intracavity beam size. Figure 4(a) shows the stability region and the calculated beam size at M and OC. The vertical line denotes the actual cavity length where stable Q-switched operation is achieved in each case. The beam propagation in the xz-plane inside the cavity is shown in Fig. 4(b). As the cavity length decreases, the intracavity beam size in the waveguide increases. From the pulse duration, the pulse energy and the calculated beam size at M and OC, we extract the maximum intracavity peak intensities at SA as ~47 kW/cm 2 for OCcoating and 860 kW/cm 2 for M-coating. The range of cavity length tolerance for Q-switched operation are measured to be 1.2 mm for OC-coating and 8.8 mm for M-coating. This results confirm well the calculated intensities at SA. are the direct maximum inte electric field d m intensity at th t work is illus veguide layer a [26]. The thic Fig. 5(a). The c ed mode is sho howing spatial tical direction io, we see that b:KYW planar ve index differe e maximum in be 0.56 kW/cm is also reveale nt-field interac for the low inte am size in the sults and analys ability region and coupler (OC The M-coating corresponds to the general direct-field interaction scheme of waveguide lasers because the SA is exposed only to the small beam size corresponding to the waveguide size. Because of the relatively high intensity, the Q-switched operation itself is easily achieved in a broad range of the cavity length tolerance. But stable operation can only be achieved in a narrow range with low output powers due to thermal load due to the relatively high intensity of the pump beam. The OC-coating shows highest output powers among the three different schemes. By controlling the beam parameters via the cavity length, a suitable beam size at the SA with better overlap between the pump and cavity mode is achieved in our cavity configuration. In the case of WG-coating, the output power is higher compared to Mcoating with stable Q-switched operation. Because of the lower intensity in the SA, the WGcoating can suppress possible optical and thermal damages efficiently.

Conclusion
We demonstrate stable Q-switched operation of Yb:KYW planar waveguide lasers by both direct-and evanescent-field interaction with CNTs. A detailed study is carried out by comparing the maximum intensity at the SA for each operation mode and the laser performance. From the experimental results, and the calculated intracavity beam size and electric field distribution, it is confirmed that the intensity at the SA, required for the evanescent-field interaction, is about 1500 times lower than that of the typical direct-field interaction in the presented waveguide lasers. We believe that this result is meaningful as the first quantitative analysis of evanescent-field interaction scheme in a Q-switched solid-state waveguide laser. In addition, the evanescent-field interaction scheme offers an efficient compact solution for pulsed operation of waveguide lasers and can be further applied to various types of waveguide lasers for realization of more compact pulsed lasers.