1. Introduction

The generation of ultrashort high-power laser pulses by modelocked lasers requires gain materials with broad emission bands and good thermo-mechanical properties at the same time [1]. A commercially available laser material combining these properties is the tetragonal, disordered crystal Yb³⁺:CaGdAlO₄ (Yb:CALGO) [2]. Since its first demonstration as a laser material [3], lasers with up to 73% of slope efficiency in continuous wave (cw) operation [4,5] as well as ultrafast lasers delivering Kerr-lens modelocked 30-fs pulses [6,7] have been realized with this material.

However, Yb:CALGO is also suitable for the inscription of waveguiding structures with fs-laser pulses which enable compact waveguide lasers [4]. After the first realization of waveguide laser operation in a directly laser inscribed waveguide in Er,Yb-doped glass in 2004 [8] there has been great progress in the research on direct laser-inscription of waveguide structures for laser applications in different materials. In particular crystals, possessing much better thermo-mechanical properties than amorphous materials like glass, enabled outstanding results [9,10]. The first waveguide laser in an Yb-doped YAG crystal utilized a depressed cladding structure 2005 [11]. In the following, double track structures were established in Yb:YAG ceramics [12]. This approach also enabled the first watt-level waveguide laser action in Nd:YAG [13] and up to now allowed for more than 5.6 W of output power at slope efficiencies in excess of 74% in Yb:YAG [14]. Femtosecond-laser-inscribed waveguide lasers have also been shown to be useful for the generation of ultrafast laser pulses. The first modelocked laser operation in an fs-laser-inscribed waveguide utilized a fiber-extended cavity [15]. Soon after, modelocked laser operation of fs-laser-inscribed waveguides in more compact cavities with higher repetition rates was achieved: In Nd:YAG, 16 ps pulses at 11 GHz repetition rate and an average output power 12 mW were obtained utilizing graphene as the saturable absorber [16]. More recently also modelocking of Yb:YAG waveguide lasers in a short free-space extended cavity at 2 GHz repetition rate and 2 ps pulse duration was obtained at an average output power of 322 mW using carbon nanotube saturable absorbers [17].

In our previous work, we demonstrated that Yb:CALGO is suitable for efficient fs-laser-inscribed waveguide lasers. We obtained a maximum cw output power of 2.4 W at a slope efficiency of 69% in double track structures inscribed with 150-fs pulses at 1 kHz repetition rate [4]. Such devices could serve as GHz-repetition rate sources, generating sub-50-fs pulses at high average power levels, since the gain material Yb:CALGO combines broad emission bands with good thermo-mechanical properties.

Up to now, most fs-laser-inscribed waveguide lasers in crystals have been fabricated with kHz-repetition rate lasers at inscription velocities below 1 mm/s. While for glasses lasing was obtained in waveguides inscribed at MHz repetition rates [18] and decreasing losses are reported for repetition rates in excess of 100 kHz [19], to the best of our knowledge, there is no previous report on fs-laser-inscribed waveguide laser structures in any crystal using MHz-repetition rates. An important step toward a fast production process of the required waveguide structures was realized by utilizing a 200 kHz repetition rate laser, enabling 17 mm/s inscription velocity [20]. Waveguide laser operation was also reported in Tm:Lu₂O₃ ceramics inscribed at 500 MHz [21], but at a lower inscription velocity of 5 mm/s. However, in none of these results waveguide laser output powers in excess of 250 mW were demonstrated.

Here we present, to the best of our knowledge, the first single-crystalline waveguide laser structures inscribed with fs-laser pulses at 1 MHz-repetition rate with inscription velocities of up to 100 mm/s. In systematic investigations with more than 100 inscribed structures, we tested the influence of several parameters including the inscription velocity. The inscribed Yb:CALGO waveguides enable efficient watt-level waveguide laser operation and deliver up to 3.9 W of cw output power at slope efficiencies up to 67%. This performance is similar to that of previous kHz-inscribed structures [4]. However, the new waveguiding structures were fabricated within milliseconds what is 4 orders of magnitude faster compared to previous Yb:CALGO waveguides [4]. Also compared to previous reports in other crystals this is an improvement of nearly an order of magnitude [10,20]. It should be noted that the inscription velocity was limited by the velocity of the stages used to translate the sample through the focus and not by the process itself, thus even faster inscription seems feasible. Considering that Yb:CALGO crystals of suitable dimensions are commercially available, this result represents a milestone toward mass production of integrated ultrashort pulse devices.

2. Experimental methods

Femtosecond laser direct inscription is a powerful technique to fabricate waveguides inside various transparent dielectrics. The nonlinear absorption processes limit the intensity dependent material modification to the focal area [9,10,22]. In crystalline material the inscription of double track structures, between which the light mode is guided (cf. Fig. 1) in a refractive index profile created by stress surrounding the inscribed tracks [23] proved to be the most efficient waveguide laser structure [9,10].

 

Fig. 1. Sketch of the waveguide inscription and the orientation of the waveguide samples to the optical c-axis. The waveguides are always written along the x-axis while the optical c-axis is aligned either parallel to the waveguides along the x-axis (||) (a) or perpendicular to the waveguides and the x-axis, along the z-axis (⊥) (b).

Download Full Size | PDF

For the experiments, we used a 1 MHz-repetition rate Fidelity HP High Energy laser supplied by Coherent Inc. to inscribe such double tracks. This laser provides an average output power of more than 10 W at pulse durations of (140 ± 20) fs and emits pulses at 1 MHz-repetition rate at a center wavelength of 1040 nm with a beam quality M² < 1.3. The beam diameter at the laser exit amounts to 1 mm. To adjust the laser power on the sample we used a combination of a rotatable λ/2-waveplate and a reflective thin film polarizer. We adjusted the polarization of the pump beam with a second waveplate behind the power adjustment. Both the focusing lens (NA = 0.55, 4.51 mm focal length) and sample were mounted on separate two axis tilt stages to adjust them perpendicular to the beam by back reflection. The tilt stages were mounted on computer-controlled Aerotech ABL1000 translation stages.

Due to the longer wavelength and the lower beam quality of the MHz-laser system compared to the kHz-system used in [4], the use of the same focusing lens does not result in the same properties of the focal volume. In the MHz-repetition rate experiments presented here, the radius at the beam waist was 0.78 µm in air (calculated by Gaussian optics), a factor 1.6 larger than in the kHz-experiments, whereas the Rayleigh-range is the same factor shorter. Thus, the resulting pulse density during inscription could not be compared to the one in the kHz-experiments only considering the higher repetition rate and the faster writing velocity.

The Yb(8.1%):CALGO crystals we used were grown in our labs [4,5]. We oriented the axes of the tetragonal crystal lattice using crossed polarizers and determined the exact doping concentration of the samples by transmission measurements using the absorption spectra from [4]. The samples were polished on both xy-surfaces (cf. Figure 1) in optical quality since the structuring laser had to be focused through these surfaces for the inscription. The inscribed tracks did not end directly at the yz-surface due to distortions of the beam at the edges. Thus, after inscription approximately 150 µm were lapped from the yz-surfaces and the facets were polished plane parallel to each other and perpendicular to the tracks.

As a uniaxial crystal, Yb:CALGO shows different material properties depending on the orientation of the crystal. This results in differences between the properties of waveguides inscribed in different directions in the crystal. Based on our previous results on Yb:CALGO waveguides [4], only two waveguide orientations pictured in Fig. 1 were investigated in the MHz-inscription experiments reported here.

The inscription parameter ranges of the MHz-system are listed in Table 1 for the respective orientations. Single-track waveguides, as well as double-track waveguides, were tested. For the tracks, according to Fig. 1(a) we furthermore investigated two polarization states of the inscribing laser beam: parallel (π) and perpendicular (σ) to the inscription direction (which is also the waveguide direction).

Table 1. Inscription parameters for the waveguide orientations (Fig. 1) tested with the 1 MHz-repetition rate laser system

View Table

In total, we inscribed 102 waveguide structures into the two different oriented Yb:CALGO samples (cf. Figure 1 and Table 1). All were examined with a light microscope in transmitted light mode. Simple waveguide transmission measurements were performed with a helium-neon laser at 632.8 nm to obtain the waveguide losses and the guided mode profiles.

For those waveguides with the lowest losses and best mode confinement, we performed waveguide laser experiments. As pump source, an optically pumped semiconductor laser (OPSL) with 8 W of output power at a wavelength of 967 nm was applied. The 11.4 mm long waveguides absorb more than 99% of the incoupled pump light at this wavelength. Since the structures guide only light polarized parallel to the z-direction (see Fig. 1) [4], we chose this pump polarization. A highly reflecting (HR) mirror, with a reflectivity of (99.977 ± 0.009)% in the range of 1000 nm – 1100 nm, was attached, without any index matching fluid, to the end-facet of the waveguide. Using a piezo mount, the remaining tiny airgap could be adjusted precisely. The resonator was formed between this mirror and the uncoated incoupling-facet of the waveguide with 9% Fresnel reflection only. The resulting output coupling degree is 91%. The laser output from the incoupling-side was separated from the pump light by a dichroic mirror. Figure 2(a) shows a sketch of the setup, and Fig. 2(b) a photograph of the lasing waveguide crystal.

 

Fig. 2. (a) Sketch of the laser setup, (b) photograph of the lasing waveguide crystal.

Download Full Size | PDF

We furthermore investigated the influence of the inscription velocity on the laser performance. Tracks with distances of (25 ± 1) µm inscribed with 1 µJ of pulse energy and velocities of 0.1, 1, 2, and 4 mm/s as well as a waveguide inscribed at 100 mm/s with 1.16 µJ energy were utilized for this purpose.

3. Results and discussion

Microscopic investigations of the MHz-repetition rate written tracks showed completely different material modifications than those produced with an inscription rate of 1 kHz in Fig. 3. At 1 MHz the tracks consist of dotlike chains which look very similar to the ones observed in glass at high-repetition-rate-writing [19] and are thus assigned to heat accumulation effects. The distance of the dots – these dots are melted zones, according to [19] – becomes larger with increasing writing velocities. This seems to boost the formation of larger cracks, which is stronger for the tracks written at velocities between 2 mm/s and 50 mm/s. At 100 mm/s no significant crack formation is visible, just as for the slowly written tracks. From the microscopic investigations it appears that the most promising inscription results are obtained either at low velocities of 0.1 mm/s – 1 mm/s or at velocities as high as 100 mm/s. However, further optimization of the pulse energy could decrease the crack formation for the other writing velocities.

 

Fig. 3. Microscopic images of fs-laser induced material modification in the xy-plane. Material modifications inscribed at 1 kHz and 25 µm/s compared to those inscribed at 1 MHz and different inscription velocities are shown.

Download Full Size | PDF

Due to absorption the transmission losses could not be measured at the pump wavelength. The measured losses are mainly scattering losses and coupling losses. Coupling losses could be estimated to be in the range of 10% in all cases [24,25]. Rayleigh scattering losses are proportional to λ⁻⁴ [26]. Thus, the measured losses at 632.8 nm overestimate the scattering losses at pump and laser wavelength, but still allow for a comparison of the properties of different waveguides. The guided mode is similar for the laser wavelength and single mode in all cases.

The single-track waveguides guided the light laterally to the track. They had very high losses and weak confinement in the waveguide transmission measurements, as reported previously for other oxide materials [12] and were thus not examined in detail.

The results of the microscopic investigations as well as the transmitted mode profiles of the structures with the lowest guiding losses are imaged in Fig. 4 for tracks inscribed in samples of || (top) and ⊥ (bottom) orientation of the c-axis to the waveguides, respectively. The extent of the material modifications is marked as white ellipses in the mode profiles shown in Figs. 4(a) and 4(b). Since the waveguides were polarization selective and only guided light polarized in z-direction (cf. Figure 4(a) – Fig. 4(d) the light of the guided mode was always polarized parallel to the z-axis. In Figs. 4(c) and 4(d) microscopic images of the cross-section in the yz-plane in transmitted light mode are shown. The track cross-sections in xy-plane in transmitted light mode are imaged in Figs. 4(e) and 4(f). The phase-contrast mode pictures in Figs. 4(g) and 4(h) indicate refractive index changes and the dark field image in Figs. 4(i) and 4(j) reveals scattering at the tracks. Besides their orientation the main difference between the waveguides shown in the top and bottom of Fig. 4 is their inscription velocity of 0.1 mm/s and 100 mm/s, respectively. We do not show the waveguides inscribed || to the optical axis at 100 mm/s because no significant differences are visible.

 

Fig. 4. (a) and (b) Transmitted waveguide mode profile measured with a helium-neon laser, extent of material modifications marked with white ellipsoids. (c)–(j) Microscope images of the cross-sections of the tracks with the lowest transmission losses: (c) and (d) cross-sections in yz-plane, (e) and (f) cross-sections in xy-plane: transmitted light, (g) and (h) phase-contrast, (i) and (j) dark field transmitted light. Inscription parameters: Both waveguides: 1 MHz-repetition rate, σ-polarization, 25 µm track distance, for (a), (c), (e), (g) and (i) (top): 1 µJ pulse energy, 0.1 mm/s inscription velocity, c-axis || WG; for (b), (d), (f), (h) and (j) (bottom): 1.55 µJ, 100 mm/s, c-axis ⊥ WG. The white scale bar in (j) denotes 50 µm and holds for all images.

Download Full Size | PDF

For both orientations the guided modes shown in Figs. 4(a) and 4(b) are well confined in y- and even slightly compressed in z-direction (stronger for the ⊥-oriented waveguides). The letter also exhibits stronger crack formation in Fig. 4(f) than the one inscribed with ||-orientation in Fig. 4(e). These cracks are different than the large ones observed in Fig. 3.

The loss variation with inscription velocity and pulse energy is low in the investigated range. In general, we observed lower losses for waveguides || to the c-axis. The minimum value here was determined to be 0.3 dB/cm as compared to 1.1 dB/cm for the other orientation. It should, however, be noted that the losses easily vary by a factor of two or more even for waveguides inscribed with very similar parameters. The high variation of the measured transmission losses is partially caused by difficulties of maintaining the same incoupling-conditions for waveguides inscribed with different parameter sets. Therefore, the stated losses should rather be seen as an upper limit. Much more statistic is needed to evaluate the optimum set of parameters. The lower losses for the waveguides || to the c-axis are beneficial for the laser experiments, as the gain cross-sections for high inversions are higher perpendicular to the c-axis, in which direction the guided light is polarized here.

In the laser experiments we observed laser thresholds between 0.8 and 1.1 W with slope efficiencies between 44% and 67% at track distances between 22 µm and 28 µm. Only for the lowest track distance, enabling waveguide laser operation of 20 µm, we observed a higher threshold of 1.2 W and a lower slope efficiency of 16%, indicating higher waveguide losses at such small track distances.

The results of the investigation on the influence of the inscription velocity on the laser performance are shown in Fig. 5 compared to the results of the kHz-repetition rate waveguide laser-inscribed at 0.025 mm/s from [4]. The best performance was obtained with the MHz-inscribed waveguides inscribed with 0.1 mm/s. In this case, the slope efficiency was as high as 67% and the threshold amounted to 0.9 W. These results are similar to the previous results using kHz-repetition rate inscribed waveguides in Yb:CALGO, shown on the grey background in Fig. 4, for comparison [4].

 

Fig. 5. Slope efficiency, laser threshold and waveguide losses vs. inscription velocity for || to the c-axis inscribed waveguides inscribed with ∼1 µJ pulse energy, ∼25 µm track distance using σ-polarized light. The dots on the grey background were obtained using kHz-inscribed waveguides [4], all others rely on MHz-inscribed waveguides. The solid splines are guides to the eye with no physical meaning.

Download Full Size | PDF

For higher inscription velocities the slope efficiencies (black dots in Fig. 5) decreased to values between 52% and 57%. The lowest threshold of 0.76 W was observed for the sample inscribed with the highest velocity of 100 mm/s which also showed the best slope of 57% among the faster-inscribed waveguides. It is rather counterintuitive that the observed laser thresholds decrease with the writing velocity whereas the waveguide losses increase. However, the measurement of the threshold contains a higher uncertainty than e.g. the slope efficiency. Furthermore, the incoupling-conditions are different for the loss measurements and the laser experiments. As threshold and slope efficiency are stated vs. incident power and thus depend on the incoupling-efficiency, too, waveguides with high losses in the transmission measurements could easily show a better laser performance due to a different incoupling-efficiency and vice versa. It should also be noted that the cross section of the guided mode is not only determined by the track distance. The induced refractive index change, which depends e. g. on pulse energy and writing velocity could also influence the dimensions of the waveguide cross section significantly. We are confident that further parameter optimization may still improve the performance of waveguide lasers inscribed at such high velocities. Again, more statistic is required here, too.

Figure 6 shows the input-output curves of the best waveguide lasers inscribed at 0.1 and 100 mm/s at MHz-repetition rates. The slope efficiencies are 67% and 57%, and the maximum output power amounts to 3.9 W and 3.4 W, respectively. In both cases, the laser wavelength was centered around 1039 nm. Depending on power level and alignment conditions, laser lines between 1019 nm and 1043 nm were observed as well, highlighting the potential of these waveguides for ultrashort pulse generation. The remaining tiny airgap between HR-mirror and crystal showed a strong influence during alignment as described before in [4]. Inset (a) in Fig. 5 shows the circular laser mode, indicating TEM₀₀ operation.

 

Fig. 6. Input-output laser curve of the most efficient MHz-inscribed Yb:CALGO waveguide laser. Inset (a) shows the laser mode generated in the waveguide at an output power of 350 mW.

Download Full Size | PDF

4. Summary

In conclusion, we inscribed waveguides with a 1 MHz-repetition rate fs-laser into Yb:CALGO crystals. Waveguides inscribed parallel to the optical axis showed efficient waveguide laser operation. This is – to the best of our knowledge – the first demonstration of waveguide laser operation in MHz-repetition rate fs-laser-inscribed crystalline waveguides. Waveguides inscribed at a translation velocity of 0.1 mm/s enabled a slope efficiency of 67% while those inscribed at 100 mm/s still reached 57% slope efficiency at a maximum output power of 3.4 W. The inscription velocity of 100 mm/s represents the highest processing speed for any fs-laser-inscribed waveguide laser up to now. For the first time, such structures were inscribed within milliseconds, making the mass production of such monolithic waveguide lasers feasible. Higher efficiencies seem to be possible even for the highest inscription velocities by further optimization of the parameters. The broad gain spectrum of Yb:CALGO triggers further interest in ultrafast modelocked operation of these waveguide lasers. Thus, our results open the path to various kinds of integrated and ultrafast applications of waveguides fabricated by a fast production process.