1. Introduction

Vertical-External Cavity Surface Emitting Lasers (VECSELs) have attracted a lot of attention lately because of attractive attributes including high-power, circularly-symmetric output beams and versatility in operating wavelength. The main element constituting a VECSEL is the semiconductor active structure: it is made by a quantum well gain region grown on top of a distributed Bragg reflector (DBR). The laser cavity is usually completed by an external mirror, or mirrors, separated from the semiconductor chip by an air-gap. Heat removal from the gain region is important in these devices in order to maximize the output power. One way to do this efficiently is to capillary-bond a transparent high-thermal conductivity crystal as heatspreader on to the VECSEL surface [1]. As well as facilitating power-scaling and broad wavelength coverage, this configuration has enabled a quasi-monolithic VECSEL format, the so-called microchip VECSEL (µ-VECSEL) [2, 3]. In this particular case, the heatspreader is directly mirror-coated on its outer surface, creating a plane-plane cavity. The µ-VECSEL is therefore an integrated version of the VECSEL with potential for volume production.

One drawback of the µ-VECSEL is that the laser cavity is mainly stabilized by a thermal lens appearing in the gain region when it is being pumped [4]. As the power is ramped up, the thermal lens effect increases and the fundamental mode spot size in the active region decreases. This means that mode-matching the pump to the fundamental laser mode is possible for one given power only and the output beam quality tends to degrade (due to the appearance of higher transverse modes) as the power is varied. To overcome this problem and ensure stable fundamental mode operation, a microchip laser having a plano-concave cavity is preferable [5-8], because the mode size is then mainly determined by the physical characteristics of the cold cavity.

In this paper we demonstrate the operation and investigate the properties of an array of micro-lensed µ-VECSELs whose plano-concave cavities are obtained by micro-patterning a diamond heatspreader. This array geometry, also suited for laser array operation, is here used to perform a parametric study of the influence of the cavity properties on the performance of micro-lensed µ-VECSELs. Section 2 of the present paper details the design and realization of the devices. In section 3 we report the experimental study of a single µ-VECSEL, i.e. when only one element in an array is probed at one time. Finally, Section 4 is devoted to the analysis of the experimental results. Attention is paid to the mode-matching conditions and aperture losses in the selection of the lasing modes.

2. Device design and fabrication

The 1055-nm-emitting micro-lensed µ-VECSELs under study are constituted of a semiconductor part liquid-capillarity bonded to a single-crystal diamond heatspreader, whose outer surface has been shaped into an array of spherical micro-lenses (µ-lenses) and subsequently mirror-coated. The semiconductor structure includes 10 InGaAs strain-compensated QWs (one per antinode) grown on top of a 30.5-pair AlAs/Al_0.2Ga_0.8As DBR. The micro-lenses are fabricated using the ‘resist-reflow’ technique followed by a inductively coupled plasma etching using Ar/O₂ [9]. The lenses obtained are accurately spherical with a rms roughness less than 3 nm [9]. A SiO₂/TiO₂ coating with nominal peak reflectivity of 95% at 1.055 µm, anti-reflection at 808 nm, was deposited onto the diamond outer surface in order to create the concave output couplers. A schematic of an individual plano-concave µ-VECSEL cavity is shown in Fig. 1. Fig. 2 shows a picture of the final µ-VECSEL array, where different µ-lenses (acting as concave micromirrors), having a diameter d ranging from 10 to 100 µm, are distributed over different sections of the array. The other devices parameters (as shown in Fig. 1) are t ~250 µm, h ~ 0.75 µm and the micro-lens radius of curvature R is given by R=(d ²+4h ²)/(8h).

 

Fig. 1. Schematic of an individual µ-VECSEL.

Download Full Size | PDF

 

Fig. 2. Picture of the µ-VECSEL array.

Download Full Size | PDF

3. Single µ-VECSEL experiment

In this section, we study the characteristics of individual µ-VECSELs having different lens diameters. A schematic of the experimental set-up is shown in Fig. 3. The µ-VECSEL array is directly mounted on a copper holder held at 10°C. The holder is fixed on a XYZ translation stage for alignment. The device is pumped, through the heatspreader, with polarization-coupled 810-nm diode lasers. The pump beam, with up to 210 mW of average power, is delivered using a 15-mm focal length lens (NA~0.25), yielding pump spot size in the 13-30-µm range. As can be seen in Fig. 3, the µ-VECSEL signal is separated from the pump by the use of a dichroic mirror followed by a high-pass filter.

 

Fig. 3. Left: Experimental set-up. Right: schematic of the single µ-VECSEL probing.

Download Full Size | PDF

The equivalent plane-plane µ-VECSEL is characterized first by probing a region with no lenses (such a region can be seen at the bottom centre of Fig. 2). Figs. 4 (a) and 4 (b) show the results. A 25-mW threshold and a pump limited output power of 60 mW are found. As expected [4], the M² varies with the power indicating that the mode-matching condition changes with the power. At 50 mW output power the M² is 1.9.

 

Fig. 4. Plane-plane µ VECSEL (a): Output power transfer function (b): M2 measurements.

Download Full Size | PDF

Micro-lensed µ-VECSELs with lens diameter d of 90, 75, 50 and 42 µm, respectively, were then measured. For each device the output power transfer function, the spectrum and the M² of the output beam were recorded. The results are summarized in Fig. 5. There is no large variation in the threshold value between devices. We can nevertheless observe that the 90-µm device threshold is at 25 mW, while it is around 15 mW for the others. The slope efficiency of the 90, 75 and 50-µm VECSELs is 35%. The maximum output power for all devices is limited by the available pump power and is around 70 mW for both the 50 and 75-µm devices. It is slightly lower, 60 mW, for the 90-µm device due to its higher threshold. The slope efficiency drops dramatically to ~15% for the 42-µm VECSEL and consequently the output power is limited to 25 mW. This decrease in efficiency can be attributed to both an increase in the mode-mismatch and in the aperture losses (see next section). We can further note that µ-VECSELs with lower lens diameters (down to 10 µm) did not lase. These achieved power levels are consistent with previous micro-cavity laser reports [2, 3, 5]. The absence of roll-over and the low device thermal resistance, measured to be 140 K/mW, demonstrate the effective heat sinking provided by the diamond heatspreader and suggest that single emitter operation with output powers greater than 100 mW would be achievable. A detailed report and analysis of the latter effects will be presented elsewhere.

An example of M² measurements for the 90-µm µ-VECSEL is shown in Fig 5 (c), while results for all devices are plotted in Fig. 5 (b). The µ-VECSELs operate in the TEM₀₀ mode with M²<1.3 for lens diameters above 75 µm (M²~1.1 for the 90-µm device). The 50-µm µ-VECSEL has a device M²~2.8. The 42-µm has a lower M² of 1.6. This can also be explained by the increasing aperture losses filtering out the highest order modes. Unlike the plane-plane case, M² values are found not to vary significantly with the power.

In Fig. 5 (d), spectra for the 90, 50 and 42-µm µ-VECSELs at their respective full powers are plotted. Depending on the operation conditions, the spectrum consists of one or several peaks (separated by 0.9 nm) representing the longitudinal modes. The spectrum is similar for other devices, except that satellite peaks, due to the presence of transverse modes [6], are present in the 50-µm µ-VECSEL case.

 

Fig. 5. (a): Devices’ power transfer functions, (b): summary of M²values, (c): Examples of M² measurements for the 90 µm device. Inset: beam image at full power, (d): Spectrum at full power of the 90-µm, 50-µm and 42-µm devices.

Download Full Size | PDF

These results show that plano-concave cavities can be a solution to ensure stable single-transverse mode lasing in µ-VECSELs at power levels of several tens of mW.

4. Micro-lensed µ-chip analysis

The previous results concerning the mode selection and evolution with d can be explained as the combination of two phenomena: the variation of the mode-matching condition and the aperture losses. Given the micro-lensed µ-VECSEL parameters, the fundamental mode size in the gain region, i.e. at the semiconductor plane, can be inferred by:

In Eq. (1), λ is the wavelength in vacuum, n is the heatspreader refractive index (and for diamond at λ~1 µm, n~2.42) and R is the radius of curvature of the micro-lens.

 

Fig. 6. Fundamental mode (2ω₀) and pump (2ω_p) diameter evolution as a function of the lens diameter d (bottom axis) and of the radius of curvature R (top axis).

Download Full Size | PDF

From Fig. 6 we can see that the fundamental mode ω₀ size increases with the lens diameter d (resp. R) and that a cut-off exists at d=25 µm (R~105 µm). The cavity is therefore stable only for d>25 µm. For d varying from 40 to 100 µm (R~267 to 1667 µm) the fundamental mode diameter ranges from 12 to 22.5 µm. The plano-concave cold cavity supports a wide range of transverse modes, which can be written in a basis of Hermite-Gaussian or Laguerre-Gaussian modes (noted TEM_mn where the fundamental mode is TEM₀₀). The number of modes that can lase depends first on the overall gain seen by the different modes. This means that it depends on the mode-matching conditions. If the pump spot size matches the fundamental mode size, the laser operates mainly in the TEM₀₀ mode because the higher order modes have a bigger spot size. On the other hand, if the pump spot is big enough to excite higher order modes, then the laser tends to operate in a superposition of modes.

In our experiment the pump is focused down onto a single array element through the top of the device. For a given focuser, the pump spot size that can be achieved on the gain region depends on the µ-lens focal length, which itself depends on the µ-lens diameter. The expected pump spot can be calculated if one knows the pump beam characteristics before the focuser and by considering the µ-lens as a lens of focal given by: f (d)=R(d)/(n-1). We measured the pump beam to have a 1/e² radius of 1.25 mm and M²~1.6 right before the focuser. Results of the pump spot calculations for a 15-mm focal length focusing lens are shown in Fig. 6 along with the fundamental spot size. A minimum pump spot size smaller than the fundamental mode size indicates that mode-matching is possible by simply changing the position of the focus with respect to the µ-VECSEL. We can see that in our case, mode-matching is possible for devices with µ-lens diameter above ~65 µm (i.e. R>705 µm). Below this value the pump spot is wider than the fundamental mode, which favors the excitation of higher transverse modes. This effect explains why the M² shoots up to almost 3 for d=50 µm, while it is below 1.3 for d equal or above 75 µm (see previous section). However, another factor is responsible for the improvement of the M² recorded at d=42 µm.

Due to its finite transverse size, the concave mirror (i.e. the µ-lenses) also acts as an aperture and introduces a certain amount of loss for the transverse modes. To illustrate this point, we plot in Fig. 7, for a few Hermite-Gaussian modes, the losses L due to the reflection on the concave mirror. This was done by taking the overlap integral of the mode power over a circular aperture and dividing it by the total power carried by the mode.

In Eq. (2), H_m and H_n are the Hermite polynomial of order m and n respectively, x and y are the spatial Cartesian coordinates. Here, ω is the fundamental mode radius at the aperture plane and can be deduced from ₀ ω using the standard Gaussian propagation formulas. The denominator of Eq. (2) is equal to [10]: (2^mm! √π) (2ⁿ n! √π)ω ²/2. The results are shown in Fig. 7.

 

Fig. 7. Aperture losses of a few Hermite-Gaussian modes reflecting on concave micro-mirrors.

Download Full Size | PDF

The losses increase for decreasing mirror diameter, i.e. aperture. Losses are higher for higher order modes because of their bigger size. The aperture of the concave mirror therefore acts as a filter for the higher order modes. This feature can be interesting in order to reduce the number of µ-VECSEL transverse modes when the condition of mode-matching is not fulfilled. It explains the improvement in the M² and the cleaner optical spectrum (see Fig. 5 (d)) at d=42 µm (when compared to d=50 µm) even though the mode-mismatch is worse for this device. However, TEM₀₀ losses need to be kept at a minimum and consequently the µ-lenses aperture needs to be wide enough to accommodate the fundamental mode. Please note that Fig. 7 represents the losses incurred by the mode for one reflection. However, the lasing modes experience many round-trips in the cavity and the aperture effect is exacerbated. In consequence, such µ-VECSELs with µ-lens diameter below 30 µm are too lossy to reach lasing threshold.

The combination of the two phenomena discussed above is believed to be the cause for the dramatic decrease in the slope efficiency at low d recorded in section 3.

5. Conclusion

In this paper, we have reported the demonstration of monolithic micro-lensed µ-VECSELs. The devices were fabricated by micro-patterning a diamond heatspreader in a µ-lens array format. The heatspreader was then high-reflection coated and capillary-bonded to a semiconductor DBR-active region structure to create the µ-VECSEL array. We have studied the importance of mode-matching the pump beam and of the µ-lens aperture in the selection of the lasing modes. Mode-matched µ-lensed µ-VECSELs were found to address the main issue of plane-plane µ-VECSELs, by ensuring stable fundamental mode emission. Pump limited output power up to 70 mW and slope efficiency of 35% were reported. Higher pump power should readily enable these devices to reach powers above 100 mW. Increasing the micro-lens radius of curvature in order to increase the mode size in the active region should allow the use of higher power pump laser. This, in conjunction with the mode diameter increase, should yield microchip devices emitting several hundreds of mW in the fundamental mode.