We report what is to our knowledge the first Nd:BaY2F8 (Nd:BaYF) laser pumped with a multiwatt fiber-coupled diode array tuned at ≈804 nm. As much as 2.4 W were obtained with 6.2 W of absorbed pump power, showing efficient operation (51% slope efficiency), excellent beam quality (M 2=1.1), and weak thermal lensing. Small intracavity losses (<1%) were measured, indicating both reduced thermally induced aberrations and good optical quality of the laser crystal.
©2003 Optical Society of America
Neodymium fluoride laser crystals such as Nd:LiYF4 (Nd:YLF) are of interest for medium-high-power diode pumping because of their favorable thermo-optical properties. They allow for the generation of linearly polarized beams with excellent spatial quality, as well as significantly reduced thermal lensing with respect to Nd:YAG and Nd:YVO4 crystals .
BaY2F8 has been widely investigated and successfully employed as a crystal host with Er3+ dopant for generation of laser light near 3 µm . However, among 1-µm laser crystals Nd:BaYF was less investigated in the past [3,4], although its laser properties might be as attractive as those of Nd:YLF. Therefore, an accurate study of a multiwatt diode-pumped Nd:BaYF laser might assess the suitability of such a material for high power operation by measurement of the thermal lens and the thermal fracture parameter.
Owing to its relatively long fluorescence lifetime ≈470 µs and the high saturation fluence ≈1.35 J/cm2, Na:BaYF can generate efficiently high-energy Q-switching pulses.
A particularly interesting feature of Nd:BaYF among neodymium fluoride laser materials is the relatively broad fluorescence bandwidth of ≈2.6 nm (nearly twice that of Nd:YLF, ≈1.4 nm ), which can support the generation and the amplification of subpicosecond pulses.
A distinct advantage of Nd:BaYF over Nd:YLF is that it can be doped at relatively high levels (at least up to ≈3.75% ) without significant parasitic effects reducing laser efficiency, allowing for the fabrication of microchip lasers.
Another interesting feature of Nd:BaYF is the possibility to pump a secondary absorption peak near 804 nm . Since commercial high-power laser diodes are most often designed to pump at room temperature (≈25°C) the 808-nm line of other Nd-doped laser crystals, their operation at the lower temperature needed to generate this shorter wavelength is beneficial for extending their lifetime. Furthermore, pumping Nd:BaYF at the 804-nm peak is preferable to using high-power diode pumps appositely fabricated for operation at the main absorption peak, ≈799 nm. Indeed, the lower absorption allows for a better distribution of the pump power (and the associated heat) along the crystal length, facilitating the power upscaling.
Recently we reported a detailed investigation of a low-power diode-pumped Nd:BaYF , operated either at 1.05 µm or at 1.32 µm. Here we summarize the results of a power-scaled Nd:BaYF laser based on a crystal sample that was appositely grown at the University of Pisa for these experiments. As much as 6.2 W pumped the Nd:BaYF laser, nearly 10 times higher than the power used in our earlier experiments with a single-emitter pumping diode.
2. Experiments and modeling
The crystal was grown in a homemade Czochralski furnace with resistive heating and automatic diameter control system. Particular care was taken in order to avoid contamination in the sample: the starting materials were 5N fluoride powders (BaY2F8, NdF3, and BaF2), and the growth process took place in high purity Ar atmosphere. The growth parameters were a temperature of ≈976°C, 0.5-mm/hour pull rate, and a rotation rate of 5 revolutions per minute. As seed we used a piece of undoped crystal oriented along the a crystallographic axis.
The doping level in the melt was 1.8 Nd3+ at %. The boule was pencil-shaped with a maximum diameter of 16 mm and a length of about 53 mm. It was oriented and a 4×4×11 mm3 sample was cut with the b and c crystallographic axes lying in the input face (perpendicular to the crystal length). The crystal faces were ≈1° wedged. The sample was anti-reflection coated at 1.05 µm on one face, while the other face was coated for high reflectivity at 1.05 µm and high transmittivity at 0.8 µm. It may be worth noting that the monoclinic symmetry of the crystal makes this compound very difficult to characterize. In particular all its optical and thermo-mechanical features strongly depend on the orientation of the sample, for this reason its optimum choice is of crucial importance. It has been shown in the literature that, even if the c-direction is not one of the axes of the index ellipsoid, it is indeed the direction of maximum stimulated emission cross section of the main laser line of Nd3+(σc≈1.3·10-19 cm2) , and that the difference in the thermal expansion coefficients along the three crystallographic axes is minimized in the experimental configuration we chose .
The output of the 12-W fiber-coupled pump diode array used for the experiments had a diameter of 800 µm and a numerical aperture of 0.22, and was imaged into the crystal with a 480-µm spot diameter 2W p0. The laser crystal was mounted in a water-cooled copper holder, using an indium foil for optimum thermal contact. The fiber-coupled pump was tuned at λp≈803.6 nm and a fraction Aeff ≈95% of the pump power was absorbed in the Nd:BaYF sample. Therefore, the effective absorption coefficient was αeff=-1n(1-Aeff )/l≈0.27 mm-1 at λp≈803.6 nm (l=11 mm was the crystal length). Both the ≈2-nm wide diode spectrum and the pump beam depolarization contributed to an effective absorption which was smaller than the spectroscopic values measured for polarizations parallel to the axes b and c. To reduce the risk of optical damage, we limited initially the level of the absorbed pump power to 4.7 W for the measurements of laser gain and intracavity losses, as well as for the determination of the output beam quality and the thermal lensing. Later we found that the pump face of the crystal indeed experienced coating damage at ≈6–7 W of absorbed power.
We set up a 36-mm long flat-flat resonator. The output power characteristics with different output couplers are summarized in Fig. 1. As much as 2.4 W at 6.2 W absorbed pump power were obtained with a linear polarization parallel to the c-axis. The highest slope efficiency measured was 51% with a 90%-reflectivity output coupler. The beam quality at the maximum output power was M 2≈1.1, indicating a nearly diffraction-limited operation.
where Po is the output power, Psat is the saturation power, g 0 is the double-pass small-signal gain (proportional to the pump power absorbed Pabs ), and R is the output coupler reflectivity. L=L 0+Ld (Pabs ) is the total intracavity loss, where L 0 is the constant loss due to scattering, linear absorption and coating imperfections, whereas Ld (Pabs ) is the pump-induced loss. For each value of pump power in Fig. 1 there are several output power values Po correspondent to the output coupler reflectivity R used. One can perform a best-fit with Eq. (1) to extract pump-dependent parameters such as L and g 0. The characterization of the small signal gain was already the subject of a previous report . Here we will concentrate on the thermo-optical effects. With the nearly optimum output coupling of R=90%, and using the knife-edge technique we investigated the spatial quality of the output beam, obtaining information both on the M 2-parameter and on the intracavity mode size . This is related to the pump-induced thermal lens and therefore to the intrinsic thermo-optical properties of Nd:BaYF. The plane-plane Fabry-Perot geometry allows an accurate characterization of the thermal lensing, furthermore the high pump power makes the measure of the intracavity power-dependent loss in such a laser easier than in low-power devices . In order to make the analysis of the thermo-optical properties of Nd:BaYF more meaningful, we compared these results with those obtained with a 5-mm long Nd:YVO4 laser crystal pumped by the same fiber-coupled diode. Nd:YVO4 is a well-known laser crystal which has been extensively employed in high-power (few tens of Watts) diode-pumped lasers, and its thermo-optical properties are readily available . In this case the vanadate crystal (from Casix, Inc.) was 0.3%-doped, with flat parallel faces coated like those of the Nd:BaYF sample, with approximately the same effective absorption coefficient (αeff ≈0.26 mm-1 at λp≈808 nm), yielding nearly the same heat load per unit length in both crystals. The vanadate laser was a 70-mm long flat-flat resonator.
The thermal lensing was modeled according to the approximations discussed in Refs. [7,9]. In particular, we considered axially symmetric radial heat flow and a top-hat pump intensity distribution with beam radius
being np the refractive index at λp(np ≈1.5 for Nd:BaYF  and np ≈2.0 for Nd:YVO4 ) and z 0 the longitudinal position of the pump waist of radius W p0 . ≈250 is the beam-quality factor of the fiber-coupled array. The temperature perturbation with respect to the constant-temperature walls (equivalent radius rb ≈2.3 mm) is :
where ξ=1-λp/λ is the contribution of the pump quantum defect to the local heating, the thermal conductivity is Kc ≈0.06 W/cm/K for Nd:BaYF , Kc ≈0.05 W/cm/K for Nd:YVO4  and Θ is the Heavyside’s step function (u=Wp -r).
This temperature perturbation produces a double-pass transverse phase perturbation of the resonant eigenmode (a TEM00 with radius wg in the laser crystal):
being -γr 2 the reference spherical wavefront regarding which the aberration amplitude is calculated. The accumulated thermal dephasing is defined as
In Eq. (5) the effective thermo-optical parameter at the laser wavelength and for the oscillating polarization, χ=∂n/∂T+CαT(1+ν)n includes both the refractive index thermal sensitivity ∂n/∂T and the longitudinal expansion contribution related to the thermal expansion coefficient αT (ν is the Poisson ratio). Although a more accurate model can be developed, for example using finite element methods, it is possible to include a phenomenological coefficient 0≤C≤1 taking into account local thermal gradients that reduce the longitudinal expansion with respect to the case of uniform heating . This helps to explain the slight overestimate of χ in Ref.  compared with the experimental result. It is important to note that for the laser designer the knowledge of χ is sufficient to model thermal lensing and diffractive losses for a given resonator geometry.
This result was obtained by using the perturbation theory, therefore its validity is restricted to pump regimes with small diffractive losses of few percents [7,10], typical of most diode-pumped cw lasers. The spherical component of the phase perturbation also determines the thermal focal length fth , since an ideal plane wave after a double pass through the aberrated crystal acquires a transverse spherical dephasing equivalent to that provided by a lens with
Both Ld and fth depend on the cavity mode radius wg inside the laser crystal and on W p0.
The power-dependent thermal focal length, measured for both Nd:BaYF and Nd:YVO4 crystals, is shown in Fig. 2 with the theoretical curves calculated from the numerical model, Eqs. (2–7). The Nd:BaYF sample shows a significantly reduced thermal lensing with respect to the Nd:YVO4 crystal, of nearly a factor 3 smaller. Accordingly, the diffractive loss contribution is reduced in the fluoride crystal (≈0.15% versus ≈0.7% in the vanadate), owing to the smaller thermal aberrations (Fig. 3). The background passive losses were smaller in the Nd:YVO4 (≈0.1%) likely because the flat-parallel faces minimize spurious intracavity reflections from the antireflection-coating, acting as an etalon. It is worth noting that in a flat-flat resonator the resonant spatial mode is determined by thermal lensing and, in part, by gain-guiding  (the latter being more important for extremely compact resonators such as microchip lasers). In particular, since in our setup the radius of the fundamental mode wg in the end-pumped laser crystal decreases when the pump power increases, the thermally induced diffractive losses tend to saturate at the maximum pump power levels employed. Conversely, an optical cavity designed to have a nearly constant resonant mode in the active medium exhibits an approximately quadratic increase of the thermal diffractive losses with the pump power [7,10]. From the best-fit procedure applied to the pump-dependent thermal focal length (Fig. 2) and to the pump-induced passive losses (Fig. 3), we obtained χNd:YVO 4≈7.7•10-6 K-1 and χNd: BaYF≈4•10-6 K-1. As we have remarked, the measure of χ without a detailed knowledge of the thermal stress distribution yields only a range of values for ∂n/∂T, corresponding to 0≤C≤1. The value of χNd:YVO 4 is in fair agreement with that measured in Ref. , and corresponds to C≈0.4 if one uses the parameters for Nd:YVO4 given in Ref. . The identification of the correct value of ∂nc /∂T for Nd:BaYF might be done only with a more sophisticated thermo-optical model , but this is beyond the scope of this work.
Eventually, we tried to perform a thermal-fracture experiment, but it was not clear whether the coating or the bulk started to damage first. There is the possibility that the process of coating deposition stressed and weakened the optical surface, making uncertain the origin of the pump-induced damage. Therefore, we rely on an earlier experiment conducted on a single uncoated sample 3.75%-doped, yielding RT =αeffξPabs, fracture /4π≈0.6 W/cm for the thermal fracture parameter . The maximum absorbed power was Pabs,fracture ≈2 W, the effective absorption coefficient αeff≈αc≈1.6 mm-1 (the pump was a narrow-line Ti:Al2O3 laser operating at ≈799 nm) and ξ=0.24. This preliminary test indicates a thermal fracture parameter lower than that of Nd:YLF, RT ≈2.0 W/cm [5,12], but more extensive experiments, especially on several samples with lower doping levels should be performed to assess this parameter more reliably.
The Nd:BaYF crystal that has been grown for these experiments showed high optical quality and excellent laser performance also with pumping in the multiwatt regime, allowing for a significant upscaling of our earlier low-power results. Nd:BaYF seems a promising material for efficient, high-power diode-pumped lasers, especially for applications requiring thermally insensitive resonator performance, such as cw intracavity doubling, in which thermal diffractive losses must be carefully controlled. Indeed, thermal lensing was significantly weaker (≈1:3) than in a 0.3%-doped Nd:YVO4 crystal under the same pumping conditions. The pump-induced losses were also smaller (≈1:5) than for the end-pumped low-doped vanadate laser.
We have shown that Nd:BaYF matches the performance of Nd:YLF (its most natural competitor) at the main laser transition at 1 µm also with multiwatt pumping, as is expected owing to the similarity of the laser parameters, i.e., cross sections (σc=1.3•10-19 cm2 versus σc=1.9•10-19 cm2 for Nd:YLF) and fluorescence lifetimes (τf=470 µs versus τf=460 µs for Nd:YLF) [5,6]. The significantly broader fluorescence line (2.6 nm versus 1.4 nm) favors Nd:BaYF for short-pulse generation. To date, the major drawback seems to be a significantly smaller thermal fracture parameter with respect to that of Nd:YLF (≈1:3). However, we note that our numerical models suggest that Nd:BaYF crystals with lower doping levels should allow for the design of high-beam-quality diode end-pumped lasers with 10-W output.
References and links
1. W. A. Clarkson, P. J. Hardman, and D. C. Hanna, “High-power diode-bar end-pumped Nd:YLF laser at 1.053 µm,” Opt. Lett. 23, 1363–1365 (1998). [CrossRef]
2. H. J. Eichler, J. Findeisen, B. Liu, A. A. Kaminskii, A. V. Butachin, and P. Peuser, “Highly efficient diode-pumped 3-µm Er3+:BaY2F8 laser” IEEE J. Quantum Electron. 3, 90–94 (1997). [CrossRef]
3. A. A. Kaminskii, “New room-temperature diode-laser-pumped efficient quasi-cw and cw single-mode laser based on monoclinic BaY2F8:Nd3+ crystal,” Phys. Stat. Sol. A 137, 61–63 (1993). [CrossRef]
4. N. P. Barnes, K. E. Murray, A. Cassanho, K. M. Dinndorf, and H. P. Jenssen, “Flashlamp pumped Nd:BaY2F8,” in OSA Proceedings on Advanced Solid-State Lasers, A. A. Pinto and T. Y. Fan, eds., Vol. 15 (Optical Society of America, Washington, D.C., 1993), pp. 24–27.
5. J. J. Zayhowski and J. Harrison, in Handbook of Photonics, M. C. Gupta, ed. (CRC Press, Boca Raton, Fla., 1997), pp. 326–392.
6. A. Agnesi, A. Guandalini, G. Reali, E. Sani, A. Toncelli, and M. Tonelli, “Spectroscopic analysis and diode pumped laser results of Nd: BaY2F8,” IEEE J. of Quantum Electron. (to be published).
7. A. Agnesi and P. Uggetti, “Measurement of thermal diffractive losses in end-pumped solid-state lasers,” Opt. Commun. 212, 371–376 (2002). [CrossRef]
8. A. E. Siegman, Lasers (University Science, Mill Valley, Calif., 1986).
9. Y. F. Chen, T. M. Huang, C. F. Kao, C. L. Wang, and S. C. Wang, “Optimization in scaling fiber-coupled laser-diode end-pumped lasers to higher power: influence of thermal effect,” IEEE J. Quantum Electron. 33, 1424–1429 (1997). [CrossRef]
10. S. C. Tidwell, J. F. Seamans, M. S. Bowers, and A. K. Cousins, “Scaling cw diode-end-pumped Nd:YAG lasers to high average powers,” IEEE J. Quantum Electron. 28, 997–1009 (1992). [CrossRef]
11. E. J. Grace, G. H. C. New, and P. M. W. French, “Simple ABCD matrix treatment for transversely varying saturable gain,” Opt. Lett. 26, 1776–1778 (2001). [CrossRef]
12. C. Pfistner, R. Weber, H. P. Weber, S. Merazzi, and R. Gruber, “Thermal beam distortions in end-pumped Nd:YAG, Nd:GSGG, and Nd:YLF rods,” IEEE J. Quantum Electron. 30, 1605–1615 (1994). [CrossRef]
13. Y.-F. Chen, “Design criteria for concentration optimization in scaling diode end-pumped lasers to high powers: influence of thermal fracture,” IEEE J. Quantum Electron. 35, 234–239 (1999). [CrossRef]