1. Introduction

Unique features of semiconductor disk lasers (SDLs) include a successful combination of advanced mode control using extended cavity together with power scalability. SDLs accumulate the positive traits peculiar to the disk geometry concept, i.e. a low thermal lensing effect and a high capability for excess heat dissipation, enriched with a general merit of semiconductor lasers – the spectral versatility available through a big portfolio of semiconductor quantum-confined compounds. Regardless of the spectral flexibility natural to semiconductor materials, the operation of SDLs at certain wavelengths is hard to achieve, particularly in the range of ~1.2 – 1.5 µm [1,2]. The wavelength tailoring with quantum-well system suffers in this spectral range from critical growth conditions due to the strain effect.

A convenient and practical way to operate semiconductor gain medium in disk geometry is to monolithically integrate it with a high-reflective mirror. This arrangement employs the gain element as a cavity mirror, ensures advantageous alignment procedure, offers flexibility in using various intracavity components and allows for multiple-gain laser design. The thermal management, which is critical for high-power lasers, is also easier to implement when the gain medium operates in a reflection mode. However, this aspect of disk geometry prohibits the operation of SDLs at specific wavelengths due to the shortage of suitable semiconductor compounds and the obstacles associated with monolithic growth of high-reflection distributed Bragg reflectors and gain regions. The increased lattice mismatch results in critical growth conditions for quality materials due to high strain. Indeed, the operation wavelength of InGaAs quantum well based devices has been extended only up to ~1.18 µm [3]. Extending the operation wavelength to longer wavelengths using InGaAs is limited by the increased lattice mismatch to GaAs substrate. The addition of nitride into the InGaAs compound can further shift the operation of the device to longer wavelengths. However, the GaInNAs/GaAs material system suffers from enhanced non-radiative processes and requires critical optimization of the growth parameters [4].

The use of quantum-dot (QD) ensembles allows for extended wavelength tailoring of SDLs due to alleviated strain impact on the epitaxial structure. SDLs based on InAs/GaAs submonolayers and InGaAs Stranski–Krastanow grown QDs have both been successfully demonstrated [5]. An output power of 4 W has been achieved from an optically pumped InGaAs/GaAs QD SDL emitting at 1180 nm [6,7]. However, the maximum gain of a QD ensemble is typically limited to 4–8 cm⁻¹ per QD layer in the vertical configuration, which may cause problems for implementation of laser cavities with nonlinear frequency conversion. A possible solution to this problem utilizing multiple-gain cavity was demonstrated recently for frequency doubling [7]. The prospects of QD technology for operation at 1.3 µm and beyond are unclear at present.

In contrast to GaAs-based materials, InP-based gain materials are nearly ideal for long wavelength emitters. However, they are impaired in lasers with vertical cavities due to the low quality of distributed Bragg reflectors (DBRs) resulting from the low refractive index contrast of the compounds available in this material system. High reflective InP-based DBRs require a large number of layer pairs, thus seriously raising the thermal problem and making such a technique of low practical value [8].

The approach employing wafer fusion allows to combine disparate semiconductor materials, e.g. InP-based active materials with high quality GaAs/Al(Ga)As DBRs [9], which cannot be grown monolithically. Recently, we have demonstrated optically-pumped SDLs operating at wavelengths of 1.2 µm, 1.3 µm and 1.57 µm with AlGaInAs/InP gain materials and AlGaAs/GaAs distributed Bragg reflectors integrated by a wafer fusion process [10–14]. The results reveal an important finding: wafer bonding can be used in emitters with high power density. This approach promises substantial wavelength tailoring of SDLs as it allows for monolithic integration of non-lattice-matched compounds. In addition, it demonstrates a high potential for power scaling of long-wavelength SDLs, because the InP-based DBRs can be avoided.

The development of visible light sources has been growing at an unprecedented rate over the past decades [7,15–18]. However, genuine red 650 nm emission with multi-watt output power has remained a challenge for SDLs. These red lasers would be needed for television and laser projection technology where visible laser sources are expected to have a huge impact [19]. In the field of biophotonics, the development of small footprint, efficient red lasers could be conveniently used as replacements for the bulky excitation sources currently used in fluorescence imaging [20]. In medicine, the ability to produce reliable watt-level red lasers is of great significance, especially in photodynamic therapy, as fiber-coupled red lasers are needed for the treatment of malignant tumors [21,22].

Direct red light generation from SDLs has been reported at wavelength 660 nm with 0.5 W of output power [23,24] and at wavelength 675 nm with 1 W output power [25]. This approach is practically limited at present by the lack of high-power short-wavelength diode sources. On the other hand, the development of frequency-doubled 650 nm sources is limited by the lack of multi-watt 1.3 µm SDLs [26]. Recently, we demonstrated a wafer-fused SDL producing an output power of 2.7 W at wavelength 1.3 µm, which represented the highest power achieved from SDLs at this wavelength to date. In this Letter we scale the power of the 1.3 µm wafer-fused semiconductor disk laser up to 6.6 W which allows ~3 W at 650 nm to be achieved by frequency doubling. The wafer fusion seems to be the most promising technique for generating high power radiation at a genuine red wavelength.

2. Semiconductor chip design

The active medium designed for a 980 nm pumping was grown on InP substrate by low pressure metallorganic vapor phase epitaxy (LP MOVPE). 5 pairs of compressively strained (1%) AlGaInAs quantum wells are positioned by lattice matched AlGaInAs spacers and InP window at optical field antinodes. A photoluminescence of quantum wells at room temperature is centered at 1263 nm, while the subcavity resonance located at 1315 nm sets a detuning of 52 nm. The DBR grown by solid-source molecular beam epitaxy (MBE) on GaAs substrate comprises 35 pairs of quarter-wave thick Al_0.9Ga_0.1As and GaAs layers [11]. The wafers were processed using a 2-inch wafer fusion technique, as described in [9,27]. After the fusion step, the InP-substrate and GaInAsP etch-stop layer were selectively etched from the top of the active region by wet etching and the monolithically integrated structure was cut into 2.5 × 2.5 mm² chips. The thermal management of the gain element was achieved by capillary bonding it to a type IIIa low bifringent diamond using de-ionized water [28]. The diamond had a 2° wedge in order to avoid the etalon effect. The assembly was then placed between two copper plates using a thin indium foil that ensures good thermal and mechanical contact. Finally, a dielectric antireflection coating at the signal wavelength was deposited on the surface of the diamond heat spreader. The optical pumping from a 980 nm fiber-coupled diode laser was focused onto a spot with a diameter of ~290 µm at the gain element. The schematic of the 1.3 µm laser cavity and the output power are given in Figs. 1 and 2 , respectively.

 

Fig. 1 The schematic of the 1.3 µm wafer-fused SDL. The spot diameter of the pump beam was ~290 μm at the gain element and the temperature of the gain element was kept at 7 °C. RoC – radius of curvature.

Download Full Size | PDF

 

Fig. 2 Output power of the 1.3 µm wafer-fused SDL pumped at 980 nm.

Download Full Size | PDF

3. Generation of 650 nm radiation by frequency conversion

The nonlinear crystal used in this study for frequency conversion is a critically phase-matched, Type I, 4 mm long BBO crystal. The crystal was AR coated for both fundamental and red wavelengths. The laser cavity was engineered to match closely the mode field diameter with the size of the pump spot at the gain mirror. The mode diameter on the gain element was 290 µm, whereas the beam diameter on the BBO crystal was varied in order to maximize the efficiency of frequency conversion [29]. The generic schematic of the cavity used for frequency conversion is given in Fig. 3 .

 

Fig. 3 Cavity schematic of the disk laser producing ~3 W at the wavelength of 650 nm by intracavity frequency conversion in the BBO crystal. The temperature of the gain element was kept at 7 °C. The beam diameter on the BBO crystal was varied by changing the distances between the given mirrors. RoC – radius of curvature.

Download Full Size | PDF

The frequency-converted 650 nm output power as a function of pump power is given in Fig. 4 for the optimal cavity configuration with beam diameter at the BBO crystal of 140 µm and Rayleigh range of 19 mm. The optical spectra of the frequency doubled emission and the fundamental frequency for the optimal cavity geometry are given for different pumping powers in Figs. 5(a) and 5(b), respectively.

 

Fig. 4 The power of red output versus pump power. The beam profile at output of ~3 W is shown in the inset. The beam diameter at BBO crystal was 140 µm and the Rayleigh range 19 mm.

Download Full Size | PDF

 

Fig. 5 The spectra of the frequency doubled output (a) and the fundamental frequency (b) for different pump powers.

Download Full Size | PDF

4. Conclusion

We demonstrated an optically pumped semiconductor disk laser producing ~3 W at 650 nm by intracavity frequency doubling. The optically-pumped 1.3 µm semiconductor disk laser with AlGaInAs/InP gain material and AlGaAs/GaAs distributed Bragg reflectors integrated by a wafer fusion process produces 6.6 W of output power representing the highest power reported to date from a semiconductor disk laser at this wavelength. The results demonstrate the high potential of wafer fusing technique for both wavelength tailoring and power scaling of semiconductor disk lasers.