We present a new geometry for edge-pumping of a solid-state microchip laser. This design, which consists of a thin-disk gain crystal that has on top a diffusion bonded undoped material that guides the pump light, allows a good thermal heat management by reducing the thickness of the gain media, whereas the pump optics is kept simple. Simulations show that more than 0.95 of the pump radiation with uniformity coefficient in excess of 0.95 can be absorbed in an Yb:YAG/YAG composite device that has a 200-μm thick, 15-at.% Yb:YAG of 3.6-mm diameter. First experiments with this configuration produced on-time 34 W output power for 220 W on-time pump power and 0.26 slope efficiency. Power scaling possibilities are discussed.
© 2006 Optical Society of America
Power scaling of microchip solid-state lasers is essential for various applications that require a high power, high-quality laser beam from a small dimensions device. The key point in designing a high-power microchip solid-state laser is related to the management of the inevitable generation of waste heat during the pumping process. Typically the diode pumped solid-state lasers have a non-uniform deposition of heat. This fact, combined with the propagation of that heat through the gain medium to the heat sink, results in temperature gradients that turns in gradients in refractive index, thermal lens effects, astigmatic and birefringence effects, and mechanical stress.
The quantum defect of Yb doped laser is typically almost three times lower than that associated with the Nd-doped crystals. However the performance of such a system is intrinsically sensitive to temperature since the thermal population of the lower laser level causes absorption at the laser wavelength.
One architecture that has been proposed to maximize the benefits of a material like Yb:YAG, whilst minimizing the associated difficulties, is the thin-disk [1, 2] or active mirror  laser. Over 1070 W continuous wave (CW) output power with 48% optical efficiency was reported . There are two outstanding advantages of this geometry: large ratio of the cooling surface to the pumped volume and the direction of heat flow is parallel to the laser cavity axis. This results in purely axial temperature gradients, which in turn implies that the intra-cavity laser field experiences no thermal lens .
Perhaps the main disadvantage of the thin-disk laser is the relative complexity of the optical system required to achieve multi-pass pumping and thus efficient absorption. An alternative is to use composite gain media to allow edge pumping of the disk. Over 90 W output power have been obtained from a CW diode edge-pumped 10-at.% Yb:YAG composite microchip of 400-μm thickness . By reducing the crystal thickness to 300 μm over 300 W output power have been reported . The decrease of the crystal thickness to values below 100 μm becomes difficult for edge pumped thin materials because of limited pumping spot size. One configuration that allows the decrease of gain crystal thickness preserving its strength is diffusion bonding of gain media to an index matched undoped cap . The complicated imaging pumping geometry used by thin-disk is replaced by non-imaging lens duct technology. By using re-imaging pumping system a similar arrangement for gain media reported 235 W output power , however, nonuniform distribution of the absorbed power was observed in this scheme.
In this work we propose a new pumping scheme for a thin-disk laser: it is a composite structure that consists of thin-disk laser crystal acting as active mirror that has on top of it a diffusion bonded undoped material of larger diameter and thickness. The pump beam is delivered through three windows and propagates inside the undoped cap and gain crystal by total internal reflection (TIR). This configuration allows the multi-pass of the pump beam by using only geometrical design of the composite gain media and keeps the pump focusing optics simple and easy to align. The gain media strength is enforced by the undoped cap which provides also the way to concentrate the pump beam uniformly into a small diameter gain media.
2. The composite microchip geometry
Figure 1 shows our new design for a thin-disk Yb:YAG/YAG composite microchip laser. The thin Yb:YAG gain media with diameter ϕ and thickness t2 is diffusion bonded to an undoped YAG cap of diameter Φ and thickness t1. Three pump windows spaced at 120° each one and 45° with respect to the optical axis were cut on the undoped YAG cap. The pump beam impinges perpendicular onto window surface and propagates by TIR between the top and bottom surfaces of the microchip as shown in Fig. 1(a). Let us consider a ray that propagates from the pump window and makes an angle γ with the corresponding radius, Fig. 1(b). The undoped YAG cap allows the transverse pump to multi-pass forth and back the gain medium through TIR analogous to the pumping method in a doubly clad fiber laser.
When the pump light intersects the gain media it is partially absorbed, then it is reflected and propagates towards cylindrical surface of the undoped YAG. After TIR on this surface the ray and the reflection plane make the same angle γ with respect to the radius. Therefore the ray hits again the gain media and the process continues until it is entirely absorbed, or eventually escapes through one of the three pump windows. The maximum pump beam divergence that still propagates through the YAG cap by TIR is 43.2°. The Yb:YAG chip has its bottom surface high-reflection (HR) coated at the lasing wavelength of 1030 nm and HR coated at 940 nm for 38° to 52° angles of incidence. A metallic coating can be finally used in order to solder the Yb:YAG/YAG to a cooling finger and to obtain an efficient heat transfer between them, as presented in Fig. 2.
3. Results and discussions
In experiments we considered a 200-μm thick, 15-at.% Yb:YAG gain crystal of 3.6-mm diameter that was attached by diffusion bonding to an undoped 1.0-mm thick YAG of 10-mm diameter [as shown in Fig. (1)]. Three windows, spaced at 120° each one, were cut at 45° on the YAG component. The windo’s dimensions are 3.6-mm length and 0.4-mm width. The upper side of YAG crystal was AR coated at the 1030-nm lasing wavelength. The bottom side of Yb:YAG crystal was HR coated at 1030 nm and at 940 nm in such a way that it assures protection against evanescent wave and TIR for pump beam. Because of laboratory conditions the Yb:YAG/YAG structure was attached to a micro-channel cooling system by using a thermo-conductive paste (thermal conductivity K=0.84 Wm-1K-1). We use for pump three fiber-coupled diode lasers JOLD-100-CAXF-15A (JenaOptik), each of it delivering 100 W at 940 nm (∆λ~4.0 nm, FWHM definition) through a fiber of 600-μm core diameter and 0.22 numerical aperture. A set of cylindrical and spherical lens was used to image the optical fiber’s end onto the pump window. Two different pump spot sizes were used in our experiments: one that has 0.4 mm × 0.93 mm and does not fill completely the gain media, the second has 0.39 mm × 3.6 mm and fill the gain media entirely. The pump beam distribution at each window was recorded with a CCD camera and used in simulations on the absorption efficiency and pump beam distribution into the Yb:YAG crystal.
We further comment that because of Yb:YAG thickness, t2, each time when the ray crosses the gain media a small displacement 2t2×tan[π/4-ArcSin(Sin β)] is added in the XY plane, β is the incident angle as shown in Fig. 1(b). Thus, the non uniformities reported in Ref  are eliminated because of a multi-pass absorption with different ray orientation instead of one pass absorption and due to the optical path added to the ray path each time when the ray strikes the gain media.
The pump-beam absorption efficiency and its distribution into the Yb-doped medium were numerically evaluated by OPTICA ray tracing software and the Monte Carlo method. The dependence of the absorption efficiency on geometrical and material factors was analyzed. Fig. 3(a) shows the numerical simulation of absorbed power distribution for three pumping directions and Fig. 3(b) shows the experimental fluorescence image of the pumped Yb YAG core. The agreement between simulations and experiment is very good taking into account the fabrication differences of diode lasers.
The pump absorption uniformity is important for keeping pure axial thermal gradients into Yb YAG gain media that makes possible the generation of laser beam with high quality. In order to characterize the distribution of the absorbed pump power into the gain medium, we defined a normalized uniformity as U=[〈I〉-σI]/〈I〉, where 〈I〉 is the average value of the irradiance absorption inside of the gain media and σI is the variance inside the gain media volume integrated over the optical axis. Following this definition, the optimum pump absorption distribution has U coefficient close to one. The uniformity coefficient varies periodically with cap diameter as Fig. 4(a) shows. Because the absorption efficiency, ηa depends on the product of absorption length and absorption coefficient in our case a good figure of merit for ηa is F= t2×αa, with αa the absorption coefficient. Figure of merit larger than 0.15 yields over 90% absorption efficiency, Fig. 4(b). Proper choice of geometrical parameters as well as doping level is important for obtaining, from the same microchip, good U and high ηa. For example a configuration that has the gain media of 25 at % Yb YAG of 75μm thickness gives more than 90 % absorption efficiency, Fig. 5(a). High absorption efficiency can be obtained by decreasing the cap thickness t1 and by increasing the doping level and Yb:YAG thickness t2, however this comes at the expense of a low homogeneity of the absorbed pump power distribution in the Yb:YAG crystal.
It can be seen from Fig. 5(a) that there is an optimum value for cap diameter. Small cap diameter gives low absorption efficiency because the ratio between pump window surface and cap surface is large therefore there is high probability for ray to escape. Very large cap diameter implies many TIRs therefore the probability for scattering by surface defects increases and the losses increase too. Figure 5(b) shows that the absorption efficiency increases slightly when cap thickness decreases. From Fig. 4(a) and Fig. 5(a) it can be observed that it is possible to obtain over 95% absorption efficiency with uniformity factor of 0.95.
The temperature distribution into Yb:YAG gain media can be calculated following the method proposed by Cousins . Whilst the analytical modeling is very good for identifying the important parameters, the complexity of the structure shown in Fig. 2 requires the use of finite element analysis (FEA). The temperature distribution in Yb:YAG/YAG structure generated by the absorbed power was calculated by using the finite element software, ANSYS 7.0. The FEA analysis was performed taking into account several parameters like: crystal’s thickness, diameter and its thermo-mechanical properties, the thermal conductivity of the heat sink’s material, input power, crystal to heat sink thermal impedance.
Figure 6(a) shows the temperature of the upper YAG’s cap surface versus pump power when thermal conductivity of the heat sink’s material is copper-tungsten (Cu30-W70, K=180 Wm-1K-1) and diamond like material (DMCH60, K=550 Wm-1K-1). We neglect in this case the thermal impedance introduced by HR dielectric layer and suppose very thin Au-Sn die bonding. Figure 6(b) shows the temperature of the YAG upper surface versus pump power when Yb:YAG thickness varies from 200 μm to 50 μm. Comparing these two figures it comes out that the Yb:YAG thickness reduction has a much more influence on temperature than the increased thermal conductivity of the heat sink. Increasing the heat sink thermal conductivity from 180 Wm-1K-1 to 550 Wm-1K-1 makes the temperature decrease by 20% while the decrease of gain media thickness from 200 μm to 50 μm makes temperature decrease from 140°C to 50°C even if the absorbed pump density increases four times. We found that the design of a thin Yb:YAG gain media assembled with undoped YAG develops a thermal gradient very close to that of a cap-less thin-disk. Thus, this composite structure maintains the thermal advantages of a thin-disk geometry, while enabling high pump power to be easily injected through the pump windows, trapping it within and hence multi-passing forth and back the gain media. The cap simply rises to a uniform, constant temperature, a harmless effect that does not affect the thermal gradients in the thin gain media.
Similar computations were made for different crystal to heat sink thermal impedances, resulting that good heat transfer is crucial for high average power operation. The thermal impedance problem has two aspects. First, it is the design of the HR coating that should be as thin as possible in order to avoid a thermal barrier between crystal and heat sink. Secondary, contacting technology should be based on metallic type, for instance Au-Sn technology , instead of indium foil or thermo-conductive paste. This analysis concluded that the increase in the heat dissipation capacity of a disk varies inversely with the Yb:YAG thickness, t2 and thus the average laser output intensity of a thin composite microchip scales as l/t2. In order to maximize the output intensity, for a given Yb:YAG diameter, the thinnest possible disk that is consistent with the pump geometry should be used.
The output power versus input pump power performances of this new geometry was investigated in a short plane-concave resonator, as shown in Fig. 2. The thermo-conductive paste, that we used for heat sink contacting, do not allow high-power CW operation, however it is possible to test the laser oscillation under quasi-CW operation. Figure 7 presents the output power under quasi-CW pumping with 10 ms pulse duration and 5% duty cycle. The best results were obtained in a 50-mm long resonator with a 100-mm radius output mirror of 0.05 transmission at 1030 nm: the on-time output power was 34 W for an on-time pumping power of 220 W with 0.26 slope efficiency. Under pumping with extended pump spot size the fluorescence image of the pumped gain media shows uniform fluorescence emission while under pumping with small pump spot size the fluorescence image shows higher intensity at the center, as shown in the inset of Fig. 7. Because three pump beams with 0.4 mm 1 mm pump spot size give higher absorption at the gain media center the pump intensity at center overcomes in higher proportion the laser oscillation threshold giving better slope efficiency.
Thus, the pump intensity is only 1.3 times higher than pump intensity at threshold: one could expect a significantly improved slope efficiency if the pump intensity exceeds threshold intensity by more than eight times. In addition, these results are severely impeded by optical fiber delivery diode laser used. The rectangular shape required for efficient coupling the pump beam to gain media is much easier to obtain from a stack diode laser. Moreover, for these preliminary experiments we used thermo-conductive paste therefore the heat transfer was not good enough for high power operation: a metallic coating would be the most appropriate and considered in further work. The laser results are bellow expectations, but encouraging for the experimental conditions we used. Thus, we consider that by solving these problems, CW operation of 300 W output power with over 40% slope efficiency is possible.
In conclusion, a new design for microchip laser that allows a reduced thickness of the gain media while keeping the pump optics simple and preserve high absorption efficiency, good thermal management, and uniform pump-beam absorption was demonstrated. The configuration consists of a thin-disk Yb:YAG gain medium that is diffusion bonded to an undoped YAG of bigger diameter, the last acting as a guide for the pumping light. The cap allows pump beam multi-pass the gain media and gives the flexibility to choose the proper gain media diameter for specific applications. Numerical model shows the dependencies between cap thickness and diameter, dopant concentration and gain media thickness. It is shown that there is an optimum cap diameter and the uniformity parameter varies periodically with the cap diameter. Also, it is concluded that for the absorption efficiency over 90% the product of gain media thickness and absorption coefficient should be over 0.15. First experiments yielded on-time 34 W output power for an on-time pump power of 220 W with 0.26 slope efficiency and shows a very homogeneous pump beam. Further work plans using high-power stack diode laser and die-bonding Au-Sn technology for high-power cw operation. The device has large capability for power scaling since available input power is not dependant by gain media thickness like edge-pumping scheme and the undoped YAG cap provides multi-pass efficient absorption.
This research work was funded by CREATE - JST Fukui Association for Industry and Technology program.
References and links
1. A. Giesen, H. Hugel, A. Voss, K. Wittig, U. Brauch, and H. Opower, “Scalable concept for diode-pumped high-power solid-state lasers,” Applied Physics B: Lasers and Optics 58, 365–372 (1994). [CrossRef]
2. C. Stewen, K. Contag, M. Larionov, A. Giesen, and H. Hugel, “A 1-kW CW thin disc laser,” IEEE J. Sel. Top. Quantum Electron. 6, 650–657 (2000). [CrossRef]
4. T. Dascalu, T. Taira, and N. Pavel, “Thermo-optical effects in high-power diode edge-pumped microchip composite Yb:YAG laser,” Conference Digest of CLEO/QELS Europe 2003, CA8-3, Munchen, Germany, June 22–27, 2003.
5. T. Dascalu, N. Pavel, and T. Taira, “90 W continuous-wave diode edge-pumped microchip composite Yb:Y3Al5O12 laser,” Appl. Phys. Lett. 83, 4086–4088 (2003). [CrossRef]
6. M. Tsunekane, T. Dascalu, and T. Taira, “High-power diode-edge pumped single-crystal Yb:YAG/ceramic YAG composite microchip Yb:YAG laser for material processing,” Conference Digest of CLEO/QELS 2005, CTuZ3, Baltimore, Maryland, USA, May 22–27, 2005.
7. L. E. Zapata, S. M. Massey, R. J. Beach, and S. A. Payne, “High average power Yb:YAG laser,” presented at Solid State and Diode Laser Technology Review, Albuquerque, New Mexico, May 21–25, 2001.
8. S. Yamamoto, T. Yanagisawa, and Y. Hirano, “High power continuous-wave operation of side-pumped Yb:YAG thin disk laser,” Conference Digest of CLEO/QELS 2004, CWO3, San Francisco, CA USA May 16–21, 2004.
9. A. Kemp, G. I. Valentine, and D. Burns, “Progress towards high-power, high-brightness neodymium-based thin-disk lasers,” Progress in Quantum Electron. 28, 305–344 (2004). [CrossRef]
10. A. Cousins, “Temperature and thermal stress scaling in finite-length end-pumped laser rods,” IEEE J. Quantum Electron. 28, 1057–1069 (1992). [CrossRef]