1. Introduction

The cubic sesquioxides Sc₂O₃, Y₂O₃, and Lu₂O₃ are very attractive host materials for high power solid-state lasers due to their outstanding thermo-mechanical properties [1]. In comparison with other oxide laser host materials, they exhibit comparably low phonon energies [2]. Consequently, the multi-phonon relaxation rates in sesquioxide hosts are low, which is highly relevant for mid-infrared laser operation. Efficient continuous-wave (cw) laser operation has been demonstrated using sesquioxides [1], e.g., when doped with the heavier rare earth ions Ho³⁺ at 2.1 µm [3], Er³⁺ at 1.5 µm [4] and 3 µm [5–7], Tm³⁺ at 2 µm [8] and in particular Yb³⁺ [9,10] as the active ion, but also when doped with Nd³⁺ ions lasing at wavelengths between 0.9 µm and 1.5 µm [11]. The strong crystal field of sesquioxides furthermore supports a large splitting of the Stark levels, resulting in broad spectra which are well suited for the generation of ultrafast fs-pulses in mode-locked operation [12–17].

To date, high-quality sesquioxide laser crystals are not commercially available. This is owed to their high melting temperatures of more than 2400 °C making their growth very challenging. Up to now, high-quality single crystals of Lu₂O₃, Sc₂O₃, or Y₂O₃ with cm³-scale volume were only obtained by the heat-exchanger method using rhenium as the crucible material [18]. However, this approach is not suitable for growth on a commercial level due to the high costs and difficulties of rhenium crucible fabrication and insulation materials. Moreover, the use of rhenium demands a very precise control of the oxygen partial pressure in the growth atmosphere to avoid damage of the crucible.

To avoid the costs and risks associated with the growth from a crucible, the crucible-free optical floating zone method was applied for the growth of sesquioxides, too [19–23]. In this method, a thin rod of pressed starting material of the desired composition is slowly moved into the focus of a high-power lamp, where it melts. The small molten zone is held together by surface tension, and after translation through the hot zone, the melt crystallizes either by spontaneous nucleation or initiated by a seed crystal, while at the same time new material is fed into the hot zone. Up to now, the high melting temperatures and the strong thermal gradient intrinsic to this method hindered the growth of large sesquioxide single crystals and resulted in strong tensions [20].

The cubic structure of sesquioxides makes them also ideally suited for ceramic fabrication techniques [24–27]. In this approach, the starting powders are sintered at temperatures below the melting temperature under high pressure. Thus the transparent ceramic formation is a very slow solid-state reaction. However, up to now the laser results obtained did not reach the performance obtained with single crystals [28], which may be due to scattering at grain boundaries and/or detrimental effects due to sintering aids often used in the fabrication process of laser ceramics [29].

It is well known that the sesquioxides Lu₂O₃, Sc₂O₃, and Y₂O₃ form mostly complete solution series, i.e. their solid solutions of the form (Lu_x,Sc_y,Y_z)₂O₃ with x + y + z = 1 also crystallize in the cubic bixbyite structure [30–33]. These solid-state solutions are often referred to as mixed sesquioxides. To overcome the issues associated with the synthesis of high-quality, large-size sesquioxide single crystals, we recently revisited the ternary phase diagram of Y₂O₃ – Sc₂O₃ – Lu₂O₃ with respect to compositions with reduced melting temperatures [34]. We successfully identified mixed sesquioxides with melting temperatures reduced to values below 2200 °C. This is well in range for the growth by the Czochralski method [35] from conventional iridium crucibles [36], but also other crucible based growth techniques such as the Bridgman method [37], the Bagdasarov method [38] or the micro pulling down (µ-PD) method [39] are feasible.

In a first experimental verification, we were able to successfully grow mixed sesquioxide crystals of compositions close to (Y_0.5Sc_0.5)₂O₃, summarized as YScO₃, by the Czochralski method [34]. The large difference in the ionic radii of Y (0.9 Å) and Sc (0.75 Å) [40] leads to a strong inhomogeneous broadening of the emission peaks of laser active rare-earth ions in this matrix [28,41], making mixed sesquioxides and in particular YScO₃ even more attractive for the generation of ultrashort pulses [28,42]. Consequently, using YScO₃ doped with Tm³⁺, we recently obtained continuous-wave [43] and mode-locked [44] laser operation and initial results with Yb³⁺-doped samples presented at conferences [45,46].

Following our report of the growth of the mixed sesquioxide crystal YScO₃ from iridium crucibles [34], the aim of this work is to investigate the potential of the Yb³⁺-doped mixed sesquioxide crystal Yb³⁺:(Y_1-zSc_z)₂O₃ grown by two different iridium-crucible based crystal growth techniques for laser applications. To this end, we grew high-quality crystals by the Czochralski as well as the µ-PD method for the first time and analyzed their structural, compositional, and spectroscopic properties. Finally, we summarize the first laser results obtained with Yb³⁺:YScO₃ by presenting efficient laser operation based on crystals grown by both methods.

2. Crystal growth

2.1 Growth by the Czochralski method

The setup for the Czochralski growth was similar to the setup described in [34]. We utilized a 41 mm high induction-heated iridium crucible with an inner diameter of 42 mm. The crucible was embedded in ZrO₂ granulate surrounded by Al₂O₃ ceramics. In order to reduce the thermal gradients, an iridium afterheater was used. The growth of crystals with a composition according to Yb^3 +(0.4 at.%):(Y_0.5Sc_0.5)₂O₃ using 5N-purity Y₂O₃, Sc₂O₃, and Yb₂O₃ starting materials (Lehmann & Voss) was conducted under pure argon atmosphere at a growth rate of 0.7 mm/h and a rotation rate of 10 rpm utilizing automated diameter control.

The progress of the growth results is shown in Fig. 1. In the initial growth experiments no seed crystal was available and the growth was initiated by a 5 mm diameter iridium rod. One can clearly see in Fig. 1 (a) that the top of this first crystal is polycrystalline, while in the bottom highly transparent, apparently single crystalline regions are found. In the second growth attempt with the result shown in Fig. 1 (b), the crystal growth was successfully initiated by a (211)-oriented seed crystal of 5 mm diameter and 20 mm length cut from the clear region of the first crystal. This crystal is transparent and shows facets corresponding to {111} planes; however, the length of the crystal was limited by the onset of a foot-formation seen in the bottom left part of Fig. 1 (b). This effect was also observed for Er:YScO₃ [34] and occurs frequently in high-temperature Czochralski-growth experiments [47]. In the third growth attempt using an identical setup and seed crystal, we obtained a longer crystal. After a ∼2 cm long neck region we were able to grow a nearly 5 cm long faceted boule with a cross-sectional area of ∼2 × 1.5 cm² and a weight of 47 g. At about 2 cm from the bottom, this crystal shows a slight deviation in its growth direction, which was corrected by manual adjustment of the growth parameters. Figure 1 (d) shows the same boule placed between crossed polarizers. The crystal clearly shows stress birefringence and striations are visible which we attribute to the regulation cycles of the automatic diameter control.

 

Fig. 1. (a) – (c) Photographs of the progress in the results of Yb^3 +(0.4 at.%):YScO₃ by the Czochralski method. (d) Boule from the 3^rd run under crossed polarizer inspection.

Download Full Size | PDF

2.2 Growth by the micro-pulling down method

In a second growth approach, Yb³⁺:YScO₃ crystals were grown by the µ-PD method. This method allows the quick fabrication of small samples of different compositions, which is very useful e.g., for spectroscopic tests.

The crystals were grown in a vacuum-tight induction-heated stainless steel chamber utilizing different iridium crucibles. The best results were obtained with a capillary diameter of 4 mm. In all cases, the crucible was closed by an iridium lid and placed on an iridium afterheater. The setup was surrounded by zirconia felts and alumina ceramic tubes for thermal isolation. Afterheater and isolation each had a small window to view the growth process.

The 5N-purity starting materials of Y₂O₃, Sc₂O₃, and Yb₂O₃ (Lehmann & Voss) were mixed to the desired composition and about 3 g of the starting materials were placed in the crucible. After evacuation to ∼3 × 10⁻⁴ mbar, the chamber was flushed with 5N nitrogen (N₂) and the growth was conducted under floating N₂ atmosphere at a flow of 500 ml/min. After melting the starting materials using a 10-kW inductive RF-heater, the melt meniscus hanging from the bottom of the capillary was brought into contact with a randomly oriented piece of a 1-mm diameter undoped YScO₃ rod from an earlier growth test. Figure 2 shows the seeding process with photos; Fig. 2 (a) before the seeding, Fig. 2 (b) just after the seeding, Fig. 2 (c) during increasing the diameter and Fig. 2 (d) during the full diameter pulling process.

 

Fig. 2. Seeding process during the µ-PD-growth of mixed sesquioxides: a) shortly before seeding (scale holds for (a) – (d)), b) shortly after seeding, c) during diameter increase, d) during constant diameter growth at full diameter. e) 6.4 cm long µ-PD-grown mixed sesquioxide crystal with a composition of Yb^3 +(1 at.%):(Y_0.52Sc_0.48)₂O₃.

Download Full Size | PDF

During the seeding we set a growth velocity of 0.25 mm/min, which was reduced to 0.10 mm/min for the actual growth. If required the heating power was manually adjusted to maintain a constant fiber diameter. As we found the cleaning process of the iridium crucibles in hot acids to be very time consuming for mixed sesquioxides, we aimed to completely empty the crucibles in each growth run. The as-grown crystals had diameters of ∼4 mm and lengths between 3.5 and 6.4 cm, an example is shown in Fig. 2 (e). The lengths was mainly limited by the load of the crucibles, yielding total growth durations below 11 hours in all cases.

3. Structural and compositional characterization

Assuming rare-earth ions like Sc³⁺, Y³⁺ and Yb³⁺ to be randomly distributed on the two different cation sites in the cubic bixbyite structure of sesquioxides [48,49], the composition of Yb³⁺:YScO₃ is thus most correctly described by (Sc_1-y-x,Y_y,Yb_x)₂O₃ with 0 $\le $ x, y $\le $ 1. However, for the sake of simplicity, pronunciation, and to account for notations common for laser crystals, in most cases we use the notation Yb^3 +(x·100 at.%):(Y_zSc_1-z)₂O₃, where z = y/(1-x), thus normalizing the total amount of Y- and Sc-atoms with x*100 describing, which at.% fraction of the host ions is replaced by the dopant. This notation is advantageous in particular when describing different amounts of Yb³⁺-doping in a host with the same Y:Sc-ratio. As an example, Yb^3 +(1 at.%)- and Yb^3 +(5 at.%)-doped (Y_0.6Sc_0.4)₂O₃ would become (Sc_0.396,Y_0.594,Yb_0.01)₂O₃ and (Sc_0.38,Y_0.57,Yb_0.05)₂O₃ in the correct notation, respectively, where it is difficult to see that it is the same host material, i.e., Y:Sc-ratio. Moreover, due to the average compositions of the grown crystals remaining close to the starting materials we stick to the starting materials compositions when describing the samples.

To confirm the cubic structure of the as-grown crystals, we ground pieces of an µ-PD-grown Yb^3 +(1 at.%):(Y_0.5Sc_0.5)₂O₃ crystal and performed X-ray powder diffraction measurements. The results are shown in Fig. 3 in comparison to spectra calculated [50] for Sc₂O₃ (a = 9.846 Å [51]) and Y₂O₃ (a = 10.604 Å [52]). All expected lines and no additional peak are found at angles between those for the parent compounds. Similar patterns were found for all other samples under investigation.

 

Fig. 3. Powder diffraction pattern and Miller indices of (Yb_0.01Sc_0.495Y_0.495)₂O₃ (bottom) compared to patterns calculated by the software Fullprof based on data for Sc₂O₃ [53] and Y₂O₃ [54].

Download Full Size | PDF

We thus conclude that Yb³⁺:YScO₃ crystallizes in the cubic bixbyite structure (Ia$\bar{3}$) without foreign phase contributions such as the high-temperature hexagonal phase occurring in yttria [31]. Using the software Fullprof, the cubic lattice parameter was calculated from this diffraction pattern to be 10.233 Å, in excellent agreement with the interpolated value of 10.231 Å obtained by Vegard’s law [51,52,55,56].

We further investigated the compositional and structural properties of the Czochralski-grown Yb^3 +(0.4 at.%):(Y_0.5Sc_0.5)O₃ crystal using a 1 mm thick sample prepared from the transparent part of the first Czochralski-boule. A photograph of this sample is seen in Fig. 4 (a). Figure 4 (b) shows that the crystal is subject to significant internal stress resulting in birefringence as evidenced by the crossed polarizer image. By micro-X-ray fluorescence (µ-XRF) using a Bruker M4 TORNADO spectrometer we investigated the spatial distribution of Yb³⁺ and retrieved the Yb³⁺-density by type calibrated quantification using a sintered Yb:YScO₃ pellet reference sample with known composition [57]. As seen in Fig. 4 (c) the crystal shows a mostly homogeneous Yb³⁺-doping ion distribution with variations between 0.50 at.% and 0.55 at.% indicating a segregation coefficient slightly higher than 1 for Yb³⁺ ions in YScO₃. This is in agreement with the data presented in [34]. Using the µ-XRF spectrometer, an energy-dispersive Laue mapping (EDLM) was performed [58]. The different colors in Fig. 4 (d) indicate grains with different crystallographic orientations. As expected, this first crystal grown without seed is polycrystalline with several millimeter-sized single crystalline regions.

 

Fig. 4. Sample from the first Czochralski-grown Yb^3 +(0.4 at.%):(Y_0.5Sc_0.5)O₃ boule. (a) Photo on mm-grid. (b) Crossed polarizer image. (c) Relative Yb³⁺-distribution measured by µ-XRF. (d) Energy-dispersive Laue mapping image. Regions of one color indicate single crystalline parts.

Download Full Size | PDF

In contrast, Fig. 5 shows the same measurements for a sample from the 3^rd Czochralski-grown boule shown in Fig. 1 (c). It should be noted that the diameter of the sample in Fig. 5 (a) is much larger than that shown in Fig. 4. The sample shows a much more regular stress pattern in Fig. 5 (b) and the Yb³⁺-distribution is perfectly homogeneous (Fig. 5 (c)). The main difference is found in Fig. 5 (d), in which the EDLM-measurement indicates a perfectly single crystalline growth. However, a closer look reveals ring-like patterns, which indicate minor variations of the lattice constant. A closer investigation of these variations is in progress and may help to reveal further details of the mechanisms of stress in Czochralski-grown mixed sesquioxide crystals.

 

Fig. 5. Sample from the third Czochralski-grown Yb^3 +(0.4 at.%):(Y_0.5Sc_0.5)O₃ boule. (a) Photo on mm-grid. (b) Crossed polarizer image. (c) Relative Yb³⁺-distribution measured by µ-XRF. (d) Energy-dispersive Laue mapping image indicating single crystalline growth.

Download Full Size | PDF

Figure 6 shows the same data for typical samples from µ-PD grown rods. Due to the thin diameter of the rod of 4 mm in combination with internal stress, it was difficult to avoid surface cracks as seen in Fig. 6 (a). When placed between crossed polarizers a rather irregular stress pattern is visible (Fig. 6 (b)). This pattern is attributed to the strong radial temperature gradients of hundreds of degrees per cm emerging during the µ-PD growth process. We assume that these can be reduced by reducing the growth velocity and/or improving the isolation setup. Figure 6 (c) and (d) show the spatial Yb³⁺ distribution and the EDLM image recorded with the µ-XRF spectrometer. Figure 6 (c) reveals a strong radial gradient of the Yb³⁺-doping concentration decreasing from 1.2 at.% in the center to 0.7 at.% at the outer face. Such a gradient was observed for all samples under investigation. It points towards Yb₂O₃ increasing the melting temperature thus being favorably incorporated in the center of the crystals at the non-equilibrium conditions featuring strong thermal gradients during the µ-PD growth. In contrast, we found a depletion of yttrium in the crystal from 55.6 at.% at the outside to 51.3 at.% in the center. However, the average composition of this section of the crystal of Yb^3 +(1 at.%):(Y_0.53Sc_0.47)₂O₃ deviates only slightly from the composition of the starting powders of Yb^3 +(1 at.%):(Y_0.5Sc_0.5)₂O₃ and the segregation coefficient for Yb³⁺ remains close to unity. We grew Yb^3 +(1 at.%):YScO₃ rods with different Y:Sc-ratios of the starting materials between 54:46 and 46:54 by the µ-PD method.

 

Fig. 6. (a) Photograph of a sample prepared from an µ-PD-grown Yb^3 +(1 at.%):(Y_0.48Sc_0.52)₂O₃ rod on mm-grid. (b) Crossed polarizer image of the same sample. (c) Relative Yb³⁺-distribution in an Yb^3 +(1 at.%):(Y_0.5Sc_0.5)₂O₃ measured by µ-XRF. (d) Energy-dispersive Laue mapping images with regions of one color indicate single crystalline parts.

Download Full Size | PDF

As the influence of the doping ion on the thermal conductivity is low for disordered materials [28], we did not perform measurements of the thermal conductivity for Yb:YScO₃ and consider it to be similar to the previously reported room-temperature value of 4.1 Wm⁻¹K⁻¹ for Er:YScO₃ [34].

4. Spectroscopic characterization

We performed spectroscopic investigations on the laser-relevant properties to reveal possible differences between the crystals fabricated by different methods, as well as the influence of the host composition on the spectroscopic features.

To determine the ground state absorption cross-sections, near-infrared room-temperature transmission spectra were recorded with a resolution of 0.3 nm using a dual-beam UV/Vis/NIR spectrophotometer (Lambda 1050, Perkin Elmer). For the determination of the stimulated emission cross-sections, we excited the samples at a wavelength of 901 nm using a wavelength tunable Ti:sapphire laser (SolsTiS, MSquared Lasers) and recorded the fluorescence with a resolution of 0.3 nm using a grating monochromator (M1000, HORIBA) and a near-infrared photomultiplier tube (R5108, Hamamatsu). The same detection setup was also utilized for the determination of the fluorescence lifetimes, while in this case a 5-ns optical parametric oscillator (OPO, versaScan, GWU-Lasertechnik) with a repetition rate of 10 Hz and tuned to a wavelength of 976 nm was utilized as the excitation source. To minimize reabsorption effects, all fluorescence signals were measured using the pinhole method [59,60].

The absorption cross-sections were calculated from the transmission spectra by the Beer-Lambert law utilizing the doping concentrations determined by µ-XRF as detailed above, while the emission cross-sections were determined by a combination of the Füchtbauer-Ladenburg method [61] and the McCumber relation [62,63].

The resulting absorption cross-sections σ_abs and stimulated emission cross-sections σ_em are shown in Fig. 7 (a) and (b), respectively. As the impact of compositional tuning is low, for better visibility, we only show the data for the scandia- and yttria-richest µ-PD-grown Yb³⁺:YScO₃ crystals with Y:Sc-ratios of 46:54 and 54:46, respectively, in comparison to the cross-sections of Yb³⁺ in the parent compounds Sc₂O₃ and Y₂O₃. The data for the 50:50 composition are e.g., found in [28,42].

 

Fig. 7. Spectroscopic characteristics of the mixed sesquioxides Yb³⁺:(Y_0.54Sc_0.46)₂O₃ and Yb³⁺:(Y_0.46Sc_0.54)₂O₃ in comparison to the parent compounds Yb³⁺:Y₂O₃ and Yb³⁺:Sc₂O₃: (a) Ground state absorption cross-sections σ_abs. (b) Stimulated emission cross-sections σ_em. The data for Yb³⁺:Y₂O₃ and Yb³⁺:Sc₂O₃ are from [1]. The insets enlarge the main absorption and emission peaks to highlight the impact of compositional tuning.

Download Full Size | PDF

All absorption and emission peaks are found at wavelengths between those of the parent compounds, while the zero-phonon line remains at 976 nm for all compositions. The absorption and emission features of the mixed crystals are inhomogeneously broadened and significantly broader than the spectra of the parent compounds. This is a result of disorder, i.e., the statistical distribution of the cations on the lattice sites leading to different ligand fields for different doping ions, with the broadened spectra being the average over all these differently absorbing and emitting atoms. As a consequence, the peak cross-sections are reduced to ∼2.05 × 10⁻²⁰ cm² for the zero-phonon line absorption around 976 nm and ∼0.86 × 10⁻²⁰ cm² and ∼0.21 × 10⁻²⁰ cm² for the emission peaks around 1037 nm and 1083 nm, respectively. As seen in the insets of Fig. 7 (a) and (b), the influence of compositional tuning is low in the investigated range and amounts to a shift of the peak wavelength of about 1.6 nm in all cases.

We also performed absorption and emission measurements at 10 K by placing the Yb³⁺:YScO₃ samples in a cryostat. However, even at such low temperature the spectra remained very broad, as seen in [64]. Thus, it is not useful to assign individual Stark energy level positions. The reason is the disordered nature of the mixed sesquioxide lattice with many different environments for the doping ion, each showing a slightly modified Stark splitting.

Nevertheless, the plot of the peak position of the absorption and emission lines vs. the cubic lattice parameter in Fig. 8 shows that, in good approximation, the peak positions (and thus the “average” Stark level positions) depend linearly on the lattice constant, which allows to tailor the composition to match a particular peak emission. A similar trend was previously found for Nd³⁺-doped LuScO₃ [65].

 

Fig. 8. Spectral shift of the absorption (a) and emission (b) peaks vs. cubic lattice parameter in the binary solid-solution series of the mixed sesquioxide Yb³⁺:YScO₃.

Download Full Size | PDF

The absorption and stimulated emission cross-sections allow to calculate gain cross-sections as shown in Fig. 9 by using the equation

 

Fig. 9. Gain cross sections of Yb^3 +(1.0 at. %):(Y_0.46 Sc_0.54)₂O₃ for different inversion levels β. With data from [42].

Download Full Size | PDF

Finally, we measured the fluorescence lifetimes for various pinhole diameters. The results are shown in Fig. 10 (a). Within the range of compositions under investigation, the lifetime values τ₀ resulting from the linear extrapolation to zero pinhole diameter do not show any significant variation, we found values of 770 µs, 790 µs, and 785 µs for Y:Sc-ratios of 54:46, 50:50, and 46:54, respectively. These values are in good agreement with the data for Yb^3 +(1.3 at.%):Sc₂O₃ reported in [66]. No data are available for Y₂O₃ doped with ∼1 at.% of Yb³⁺ and the lifetime of Yb³⁺ in sesquioxides is known to depend significantly on the doping concentration [18]. However, the general literature trend indicates slightly longer lifetimes for Yb³⁺:Y₂O₃ compared to Yb³⁺:Sc₂O₃, which is expected due to the larger lattice constant and the corresponding weaker crystal field. This also causes the lower cross-sections of Yb³⁺:Y₂O₃ (cf. Figure 7). In view of this, the lifetimes of Yb³⁺:YScO₃ seem to be indeed slightly reduced compared to its constituents. Pirri et al. attribute this to quenching by lattice defects [67], an effect which may indeed be more prominent in laser ceramics investigated there. However, it is noted that in [67] high doping concentrations of more than 5 at.% were investigated, which can result in concentration quenching. We thus regard it more likely that the slightly reduced lifetime is caused by the reduced symmetry caused by the disordered lattice, further relaxing the Laporte selection rule.

 

Fig. 10. (a) Fluorescence lifetimes for different pinhole diameters of two Yb³⁺-doped mixed sesquioxide crystals. The linear fit extrapolates the data to the zero pinhole diameter lifetime τ₀. (b) Fluorescence decay curves for 0.4 at.% and 1.0 at.% Yb³⁺:YScO₃ at 1.2 mm pinhole diameter with corresponding fit functions.

Download Full Size | PDF

Figure 10 (b) shows examples of fluorescence decay curves for 0.4 at.% and 1.0 at.% Yb³⁺:YScO₃ both recorded at a pinhole diameter of 1.2 mm. The curve for the lower doped sample shown in blue is not single exponential. Instead, the data can be fitted with very high confidence by a bi-exponential function. This is tentatively attributed due to the two different cation sites in the bixbyite-structure of sesquioxides. At low doping the energy migration is impeded and the individual decay times of the sites become apparent. In contrast, the decay curve for the higher doped sample is perfectly fitted by a single-exponential decay with a decay time of 876 µs (note that this value is not the extrapolated pinhole lifetime) due to strong migration between the two sites. A similar behavior was found in Er³⁺-doped sesquioxides [68].

5. Laser experiments

To evaluate the laser properties of the Yb³⁺:YScO₃ crystals grown by different methods, we performed laser experiments using a simple hemispheric cavity with a length of 48 mm as sketched in Fig. 11. The plane input coupling (IC) mirror was transparent for the pump wavelength and highly reflective (HR) for the laser wavelength. The output coupling (OC) mirrors had a radius of curvature of 50 mm and transmission values (T_OC) between HR (< 0.2%) and 10.2% at the laser wavelength. As a pump source we utilized an optically pumped semiconductor laser (OPSL, provided by Coherent, Inc.) tuned to a wavelength of 975 nm. The pump light was focused into the laser crystal by a 75 mm lens resulting in a pump waist diameter of about 90 × 80 µm². The uncoated and surface-polished laser crystals were mounted on a water-cooled copper heat sink kept at a temperature of 11°C.

 

Fig. 11. Schematic of the resonator setup of the Yb³⁺:YScO₃ lasers. IC: Input coupling mirror, OC: Output coupling mirror.

Download Full Size | PDF

In this configuration we successfully obtained cw laser operation with crystals grown by both methods. Figure 12 (a) shows the selected laser characteristics of a sample extracted from the 3^rd Czochralski grown Yb^3 +(0.4 at.%):(Y_0.50Sc_0.50)₂O₃ boule. The laser sample (shown in Fig. 5 (a)) had a length of 3.0 mm, resulting in a single-pass absorption efficiency of 52% at the pump wavelength. Despite the evident stress in the sample (cf. Figure 5 (b)), we did not observe a significant dependence of the laser efficiency on the laser mode position on the crystal. The highest slope efficiency ${\eta _{Slope}}$ of 83% was obtained at the highest available output coupler transmission of 10.2%. In this case the laser threshold amounted to 155 mW. However, nearly the same slope efficiency of 82% was found for T_OC = 2.9% at a significantly lower threshold of 55 mW. Consequently, the highest output power of 0.89 W under 1.17 W of absorbed power was obtained using this output coupler. The corresponding optical-to-optical efficiency was as high as 76%. The laser wavelength (λ_las) was around 1040 nm with a slight blue-shift towards higher output coupler transmission. For T_OC = 0.5% and below the laser emitted at a wavelength of about 1086 nm which suggests corresponding inversion levels below 0.05 (cf. Figure 9) and is an indication for comparably low losses. A rough estimation of the maximum losses ${L_{max}}$ according to

 

Fig. 12. (a) Laser characteristics of a 3-mm long sample from the Czochralski-grown Yb^3 +(0.4 at.%):(Y_0.50Sc_0.50)₂O₃ boule. (b) Laser characteristics of a 2-mm long

Yb^3 +(1.0 at.%):(Y_0.46 Sc_0.54)₂O₃ crystal grown by the µ-PD method. The inset shows a photo and a crossed polarizer image of the laser sample.

Download Full Size | PDF

Table 1. Laser parameters of all tested samples

View Table

We also performed laser experiments with three different µ-PD grown crystals. The best results were obtained with Yb^3 +(1.0 at.%):(Y_0.46 Sc_0.54)₂O₃ and are shown in Fig. 12 (b), all other results are listed in Table 1. The 2-mm thick laser sample had a diameter of 4 mm, but cracks at the edges reduced the free aperture to a diameter of about 2 mm. The µ-PD-grown sample enabling the best laser results is seen in the inset of Fig. 12 (b).

The 2-mm long Yb^3 +(1.0 at.%):(Y_0.46Sc_0.54)₂O₃ µ-PD sample possesses a single-pass absorption efficiency of 67% at the pump wavelength. It enabled a slope efficiency of almost 89% at T_OC = 2.9% with a maximum output power of 1.21 W at 1.43 W of absorbed power, amounting to a very high optical-to-optical efficiency of 85%. However, in contrast to the Czochralski-grown sample, in all µ-PD-grown samples efficient lasing was only obtained in the center part of the samples, which we attribute to the lower doping concentration at the edges (cf. Figure 6 (c)) and strong stress (see inset of Fig. 12 (b)). In particular, only the sample shown in Fig. 12 (b) showed a somewhat regular stress pattern in the center which is similar to the one observed for the Czochralski-grown laser sample (cf. Figure 5(b)). This finding might explain why the highest laser efficiencies were achieved only with this particular µ-PD-grown sample (see Table 1). The emission wavelength of this laser was around 1040 nm for all output coupler transmission values. However, according to Eq. (2) the maximum losses of this sample are found to be below 0.2%. The threshold values are comparable or even below the values obtained for the Czochralski-grown sample, but it should be noted that at the same inversion level the µ-PD-grown samples exhibit higher gain as their higher doping concentration overcompensates the shorter length. For the remaining two samples we calculated higher maximum losses of about 1% and 2%. Consequently, none of these samples achieved slope efficiencies of more than 70%. The slope efficiencies obtained here using the µ-PD-grown sample are the highest values that have been obtained with Yb³⁺-doped mixed sesquioxides to the best of our knowledge [28] and are comparable to the results obtained with Yb³⁺:Lu₂O₃ [18] and Yb³⁺:Sc₂O₃ [69] in a similar laser setup. Up to 82% of slope efficiency were also reported for ceramic Yb³⁺:Y₂O₃ [70] enabled by its lower synthesis temperature. In contrast, using Yb³⁺:Y₂O₃ crystals grown from the melt not more than 73% of slope efficiency were realized [71], which is attributed to scattering losses arising from the above mentioned foreign phase contributions due to the high-temperature phase transition in yttria [31]. Our results indicate that this phase transition is circumvented in Yb³⁺:YScO₃, which is a significant advantage when it comes to the growth of large, scattering free single crystals from the melt e.g. by the Czochralski or the µ-PD method.

6. Conclusion

In conclusion we have shown for the first time that the mixed sesquioxide Yb³⁺:YScO₃ with compositions close to (Y_0.5Sc_0.5)₂O₃ can be grown in laser quality from iridium crucibles by the µ-PD method and the Czochralski method. We have demonstrated this by growing Yb^3 +(0.4 at.%):(Y_0.5Sc_0.5)₂O₃ boules by the Czochralski method and Yb^3 +(1 at.%):YScO₃ rods with different Y:Sc-ratios of the starting materials between 54:46 and 46:54 by the µ-PD method.

The Czochralski method allows for the growth of several cm long mixed-sesquioxide boules with large single-crystalline areas. We regard it possible to reduce the tensions in the crystals by a further improvement of the thermal isolation in future.In contrast, the strong tensions and non-uniform incorporation of Yb³⁺ in the grown crystals seem to be an intrinsic issue of µ-PD-grown crystals. Nevertheless, this technique is extremely useful as it allows for a fast synthesis of crystals with different compositions and/or doping ions which are useful for spectroscopic investigations and even laser experiments.

In our spectroscopic investigations we determined the absorption and emission cross-sections of Yb³⁺:YScO₃ for the µ-PD-grown samples with the highest and lowest yttria content. For the first time, we quantified the peak wavelengths to shift by ∼1.6 nm for Y:Sc-ratios 54:46 and 46:54 and found a linear relation with the lattice constant of the host composition that allows to tailor the composition to achieve a particular peak wavelength. In the time-dependent fluorescence spectroscopy, we found that the decay kinetics of Yb^3 +(0.4 at.%):YScO₃ can be well described by a bi-exponential function, which we attribute to a decreased energy migration between the two different cation sites in sesquioxides at such low doping concentrations. In contrast, the decay kinetics of an Yb^3 +(1 at.%):YScO₃ sample are well described by a single exponential decay with lifetimes slightly below those of the parent compounds Yb³⁺:Y₂O₃ and Yb³⁺:Sc₂O₃ at the same doping concentration. This is explained by a reduced symmetry caused by owing to the disordered lattice.

Finally, our laser experiments unveiled the groundbreaking potential of Yb³⁺:YScO₃, showcasing its capability for highly efficient continuous-wave laser operation. Achieving slope efficiencies nearing 90% and optical-to-optical efficiencies surpassing 80% in samples marked by substantial stress-induced birefringence is a remarkable outcome of these initial laser experiments. These results are comparable to the results obtained with pure sesquioxides like Yb:Lu₂O₃ or Yb:Sc₂O₃. In combination with the broad emission spectra caused by the inhomogeneous broadening, remaining broad even at cryogenic temperatures, this gives rise to a plethora of applications for Yb:YScO₃ single crystals in mode-locked lasers and amplifiers at room- or cryo-temperatures.