Novel semiconductor membrane external-cavity surface-emitting laser

Optically pumped semiconductor vertical external-cavity surface-emitting lasers (VECSELs) exhibit many desirable properties^{1, 2} and have therefore become an important stand-alone class of solid-state lasers over the last 20 years. For example, VECSELs can be used nowadays to reach 100W-level continuous wave output.³ However, a large quantum defect (resulting from the energy difference between pump and laser photons) means that heat is incorporated into the active region of VECSELs. This gives rise to a strongly temperature-dependent performance⁴ caused by the interplay of gain and cavity resonance and the limited charge-carrier confinement. The limited charge-carrier confinement is a particular challenge in the aluminum gallium indium phosphide (AlGaInP) material system, i.e., in which the thermal conductivity^{5, 6} is low and the laser structure is based on a thick distributed Bragg reflector (DBR). Indeed, the thermal conductivity of this type of DBR is an order of magnitude lower than well-conducting metals (i.e., which are often used as backside heatsinks) and two orders of magnitude worse than diamond (commonly used for the backside or as an intracavity heat spreader).⁷ In addition, the semiconductor structure itself—with a thickness of several micrometers (for the active region and the DBR)—and the substrate (with a typical thickness of 350μm) impede the heat flow out of the active region.

To overcome the heat flow problems and to improve the performance of VECSELs, numerous thermal management strategies have been previously proposed. Such approaches include changes to the heat spreader arrangement,⁸ removing the substrate,¹ flip-chip processes,⁹ or the insertion of compound mirrors.¹⁰ According to the natural progression of these developments, all semiconductor components of a VECSEL that are not essential for building a whole laser could eventually be abandoned. This could be realized by growing the active region directly on the substrate (without the DBR), then removing the substrate, and finally embedding the released active region membrane between two transparent intracavity heat spreaders (made of diamond or silicon carbide). This concept for improving the cooling of a compact VECSEL has already been theoretically studied and simulated,¹¹ and a DBR-free VECSEL (with a released active region bonded to one side of an intracavity heat spreader) has also been realized recently.¹²

In another approach to improve the thermal properties of VECSELs, we have recently performed experiments to illustrate a proof-of-principle semiconductor membrane external-cavity surface-emitting laser (MECSEL).¹³ To fabricate our novel device (see Figures 1 and 2), we grow an active region of a VECSEL (containing multiple quantum wells) that is optimized for stand-alone operation on a substrate. We then use a wet chemistry technique to dissolve the substrate and finally sandwich the residual active region membrane between two antireflection-coated diamond heat spreaders.¹⁴

Figure 1. Picture of the semiconductor membrane external-cavity surface-emitting laser (MECSEL) in operation. From left to right, the out-coupling/resonator mirror, diamond-sandwiched semiconductor gain membrane (integrated into a brass mount), birefringent filter, and pump optics with a 532nm pump laser beam (behind the birefringent filter), and a highly reflective resonator can be seen (as illustrated schematically in Figure 2).

Figure 2. Schematic illustration of the MECSEL experimental setup.

We have also performed a fundamental characterization of our new laser and have thus demonstrated its superior properties. For example, for a barrier-pumped AlGaInP-based laser system at a heatsink temperature of 10°C emitting at a wavelength of about 660nm, we can obtain an output power of up to about 600mW (see Figure 3). Furthermore, the slope efficiency we achieve (22.3%) is particularly good for vertically emitting lasers in the AlGaInP material system under these ambient conditions. In our tests we also achieved a broad tuning range for the MECSEL of about 24nm (between 650 and 674nm), as illustrated in the inset to Figure 3.¹³

Figure 3. Experimentally measured output power of the MECSEL, plotted as a function of the incident pump power. Inset: Example emission spectra from the MECSEL as a function of wavelength. Two highly reflective mirrors were used for these measurements, as well as a 1mm birefringent filter for wavelength tuning. Int.: Intensity (in arbitrary units). T_hs: Heatsink temperature. φ_pumpspot: Pump spot diameter. R_outcoupler: Outcoupler reflectivity. η_diff: Differential efficiency. P_th: Threshold pump power. P_abs: Absorbed pump power.

With our MECSEL configuration (heat-spreader-sandwiched gain membrane) it should be possible to grow semiconductor structures that are not possible with conventional VECSELs (because of the limitations imposed by the need for lattice-matching of the DBR to the substrate or of the active region to the DBR). The absence of a DBR in our MECSEL thus makes the design and growth of semiconductor gain structures much simpler. It also reduces the growth time and, in turn, the growth costs. The choice of possible materials and compositions that can be used with our devices is also much larger than with VECSELs, which means that the accessible wavelength range is greatly extended. As an example, the gallium nitride material system (in which conventional VECSELs¹⁵ currently deliver relatively poor performance) could benefit strongly from our new laser design and could make efficient blue and green¹⁶ high-power laser emission possible. It should also be possible to produce AlGaInP-based active regions in the orange spectral range¹⁷ (currently impossible with classical VECSELs because of the absorption of the emitted light in the DBR). High-quality selective etching processes are also available for other material systems^{18, 19} and are scalable to the size of whole wafers.²⁰ It should thus be possible to use any material, which can be produced and processed to the same thickness as our thin gain membrane, as a laser medium.

In summary, we have developed and experimentally verified a novel semiconductor membrane external-cavity surface-emitting laser as an alternative to standard VECSELs. Our MECSEL includes an optically pumped semiconductor membrane sandwiched between two diamond heat spreaders. The results of our experiments indicate that we can achieve an output power of about 600mW and a broad tuning range at room temperature. We are now investigating a number of additional developments to our system. For instance, large-scale bonding processes²¹ can be applied to the device. Furthermore, it may be possible to achieve in-well and multipass pumping^{22, 23} in a transmission configuration (where the pump light is recycled and folded several times through the active region). We are also investigating how these processes can be adapted for classical solid-state thin-disk lasers (i.e., to better optimize their thermal management).^24–26 Lastly, we believe that stacking of gain materials and heat spreaders (previously investigated, theoretically, for solid-state lasers^{27, 28}) can be adapted to our MECSEL design.

The authors thank the Deutsche Forschungsgemeinschaft for funding (Mi 900/24-1 and Br 3606/4-1).