2 Chip design and mounting

2.1 Chip design

As described in Ref. [Reference Pittroff, Eppich, Erbert, Platz, Tyralla and Traenkle10], the laser chip was optimized for QCW operation at ${\it\lambda}=940$ nm and both high output power and high efficiency. The vertical design incorporates an extreme double-asymmetric super large optical cavity (EDASLOC) epitaxial structure consisting of an InGaAs double quantum-well (DQW) embedded in a $3.6~{\rm\mu}\text{m}$ thick AlGaAs waveguide. The contact layer was structured into $5~{\rm\mu}\text{m}$ wide stripes with $10~{\rm\mu}\text{m}$ spacing (aperture width $W=1200~{\rm\mu}\text{m}$), which were defined by shallow proton implantation. The facets of the laser chip, with total width 2 mm and cavity length $L=6$ mm, were passivated and coated with a reflectivity of $R_{f}=0.9\%$ at the front side and $R_{r}=96\%$ at the rear side. The final laser chip was then mounted in ‘sandwich’ configuration between a 2 mm and 1 mm thick expansion matched CuW heat sink using AuSn solder (Figure 1), to produce a single stack element. These elements are characterized in QCW mode using a clamping test fixture, at a temperature of $25\,^{\circ }\text{C}$. Each element operates at $P_{\text{out}}=130$ W and 60% conversion efficiency (${\it\tau}=1$ ms, $f=100{-}200$ Hz). This corresponds to a power density of ${>}$1 kW per cm aperture, which is more than double that of conventional commercial QCW bars (300–400 W$/$cm aperture). The measured vertical and lateral divergence angles are ${\it\theta}_{V}\sim 43^{\circ }$ and ${\it\theta}_{L}\sim 12^{\circ }$ (95% power included), respectively at the nominal output power of 130 W. The lateral beam parameter product per stack element at the operation point is $\text{BPP}_{L}=0.25\times W\times {\it\theta}_{L}=63~\text{mm}~\text{mrad}$, accordingly the beam propagation ratio is $M^{2}\sim 220$.

Figure 1. Single stack element.

2.2 Stack design

The sandwich mounts described in Section 1 were subsequently used as building blocks (stack elements) for a diode laser stack consisting of 28 emitters with vertical spacing of 3.16 mm. Furthermore, for lateral heat dissipation, water-cooled DCB coolers were mounted to both sides of the stack using soft solder, which takes advantage of the long resonator length to cool the single emitters efficiently, leading to both low electrical and low thermal resistance^{[Reference Pittroff, Erbert, Eppich, Fiebig, Vogel and Tränkle7, Reference Pittroff, Eppich, Erbert, Platz, Tyralla and Traenkle10]}. The thermal resistance of a single stack element is 1.7 K$/$W, determined by tracking the change in centroid wavelength with varying bias in QCW mode (${\it\tau}=1$ ms, $f=100$ Hz). The use of expansion matched coolers in sandwich configuration and distributed cooling along the length of the single emitter minimize temperature gradients and mechanical stress, for enhanced performance, reliability and efficiency. Figure 2 shows the complete stack subassembly with fast-axis collimation (FAC). A summary of key performance characteristics data is given in Table 1.

Figure 2. 3.6 kW stack subassembly with fast-axis collimation.

Table 1. Typical stack data (measured).

The lateral BPP of the stack turned out to be significantly larger than the BPP of the single stack elements. The reason for this behavior is not clear and under current investigation. One possible reason is optical feedback (most likely from the FAC lenses) into the laser resonators causing the excitation of higher-order optical modes with higher divergence angles.

3 Fiber coupling

We developed an optical design based on simple geometrical coupling^{[Reference Pittroff, Erbert, Eppich, Fiebig, Vogel and Tränkle7, Reference Pittroff, Eppich, Erbert, Platz, Tyralla and Traenkle10]}, allowing for low-cost manufacturing and featuring high robustness. As noted in Section 2.2, each single stack element was vertically collimated using individual FAC lenses (with effective focal length 1.2 mm). Lateral focusing was achieved using a common acylindrical lens. The beams of two stacks are geometrically combined using a stack of thin right-angle prism plates, as illustrated in the optical simulation shown in Figure 3. The combined beam is subsequently coupled into a custom high-power optical fiber^[11] with core diameter $D=1.9$ mm and numerical aperture $\mathit{NA}=0.22$, by using one vertical acylindrical and one lateral cylindrical lens. All lenses as well as prism elements were fabricated with anti-reflective (AR) coatings, to minimize back-reflections. Geometrical coupling efficiency was estimated by forming an image of the beam into the expected fiber input plane into a high spatial resolution camera, as shown in Figure 4. An integration of the imaged beam profile at the fiber entry point implies a measured geometrical coupling efficiency of ${>}$98%, which is in good agreement with the measured total optical-to-optical coupling efficiency of ${\sim}$88%, taking into account 4% Fresnel losses at each fiber end, since the fibers were not AR coated, and 3% internal losses in the fiber, which is a characteristic of the custom fiber used (manufacturers data).

Figure 3. Optical simulation of the 6 kW fiber coupling system.

Figure 4. Measured beam profile near the focus plane, which is identical with the coupling plane. The plotted circle corresponds to the fiber core diameter.

Following standard practice, in order to evaluate the achieved beam quality, we determined the beam propagation factor $M^{2}$ of the geometrically combined stacks by a caustic measurement. This has been done by measuring the beam diameter at different longitudinal positions around the beam waist, with results shown in Figure 5 for the full complete system, containing all elements except for the fiber. The $M^{2}$ values in lateral and vertical direction were obtained by a fit to the measurement data. The values achieved are $M^{2}\sim 330$ (lateral) and $M^{2}\sim 360$ (vertical), which is close to the measured $M^{2}$ of the single stacks before beam combination. The results indicate that the tolerances of the fast-axis collimation, regarding beam direction and beam divergence, as well as the vertical stack adjustment are sufficient for the collimated beams to pass the prism stack without significant degradation of beam quality. In addition, no fine alignment of the position and orientation of the large coupling lenses was necessary, showing the relaxed tolerances of the complete beam forming system. The measured $M^{2}$ is only half of the $M^{2}$ of the fiber ($M^{2}\sim 700$) allowing efficient fiber coupling. It should also be noted that the system should also be compatible (when appropriate beam forming is used) with a standard 1.5 mm core diameter fiber ($M^{2}\sim 550$), although this was not attempted, as it is not necessary for the current application.

Figure 5. Measured caustic of the focused laser beam.

4 Output characteristics

Figure 6 shows the completely assembled 6 kW pump module. The two stacks have separate water-cooling and are separately driven each with a high-current pulse driver with maximum duty cycle $dc=10\%$. The fiber connector and the fiber mounting flange are also water-cooled. The power characteristics are shown in Figure 7. The measurement was conducted with ${\it\tau}=1$ ms, $f=100$ Hz ($dc=10\%$) and cooling water temperature $T=17\,^{\circ }\text{C}$. We measured the average optical power $\langle P_{\text{out}}\rangle$ using a Gentec high-power thermo-detector and calculated peak power $P_{\text{out}}=\langle P_{\text{out}}\rangle /dc$. At operating current $I=140$ A, the pump module reaches $P_{\text{out}}=6$ kW, corresponding to a pulse energy $E_{\text{p}}=6$ J, and electro-optical efficiency ${\it\eta}=50\%$. The performance was subsequently remeasured on a separate test station, in dependence on duty cycle, as shown in Figure 8(a), with higher duty cycle not significantly lowering $P_{\text{out}}$. (Small differences in total power between Figures 7 and 8(a) are due to an offset in calibration of the two test stations.) The temperature measured at the top of the stack (with in-line Pt-100 sensor) did not exceed $30\,^{\circ }\text{C}$ ($I=140$ A, $dc=20\%$). The optical spectrum is shown in Figure 8(b). It can be seen that the center wavelength shifts from ${\it\lambda}=937.5$ nm at $dc=10\%$ to ${\it\lambda}=939.5$ nm at $dc=20\%$. This is consistent with a temperature rise ${\rm\Delta}T\approx {\rm\Delta}{\it\lambda}/(0.3~\text{nm}/\text{K})=10~\text{K}$ in the average temperature of the diode lasers. Spectral width with 95% power content is 7 nm. This is wider than the typical 5 nm for a single stack element, potentially due to chip to chip variation, but consistent with the requirements for pumping the broad Yb:YAG absorption band at 940 nm. All modules showed stable operation over more than 500 h of operation. Life tests with single stack elements over 1000 h of operation were also performed without failure^{[Reference Pittroff, Eppich, Erbert, Platz, Tyralla and Traenkle10]}, as were initialization tests of the pump module units over 12 h of operation (50–100 Hz, 130 A). In continuous wave mode, diode lasers assembled using comparable epitaxial designs and process technology were shown to operate for many 1000 h in continuous wave mode at much higher optical intensities (7 W per $30~{\rm\mu}\text{m}$ output aperture^{[Reference Knigge, Crump, Wenzel, Erbert and Tränkle12]}, more than double optical intensity to that used here). Therefore, the chip technology is anticipated to enable a lifetime of many 1000 h in QCW mode, but special aging studies are necessary to confirm this for the complete systems.

Figure 6. 6 kW pump module with 1.9 mm high-power fiber. Dimensions without fiber: 68 cm $\times$ 32 cm $\times$ 20 cm.

Figure 7. Power characteristics of the pump module (out of fiber).

Figure 8. Peak output power (a) and spectral distribution (b) of the pump module depending on duty cycle.