1. Introduction

Photonic-crystal (PhC) structures incorporating III–V active layers open new perspectives for performing all-optical functionalities in compact integrated photonic circuits. A lot of effort has been devoted towards efficient PhC laser sources where the laser operation is based either on defect-nanocavity [1–2] or on band-edge light confinement effects [3]. The vast majority of PhC laser demonstrations were obtained under pulsed excitation, whether optical or electrical. However, for future integrated photonic applications, continuous-wave (CW) laser operation is of paramount importance. Recently, two-dimensional (2D) distributed feedback surface emitting lasers, operating under continuous-wave current injection at -20°C [4] and room temperature [5], were reported. In these structures the active layer is separate from the grating layer. Therefore, the distributed feedback from the PhC to the lasing mode is weak, which is especially suited to sustaining laser operation over a broad area. In order to realize more compact devices, it is necessary to exploit a stronger feedback from the corrugated structure, by etching completely through the active medium with the photonic lattice. However, patterning air-holes across the quantum wells entails a significant reduction in the carrier lifetime, leading to higher operating threshold. The required increase in pumping to achieve population inversion makes the heat sinking under CW operation an even more challenging task. Recently, careful choice of either the active layer or the heat sink materials has allowed room temperature CW operation of PhC lasers in defect-cavity configuration. For instance, the superior carrier confinement properties of an InAs/GaAs quantum dot active medium were exploited to provide PhC laser emission at 1.3 μm in a suspended 2D PhC slab [6]. Alternatively, a suitable choice of bonding materials with efficient heat dissipation, such as sapphire and AlOx, allowed CW PhC lasing near 1.55 μm [7–8].

Here we report on the first continuous-wave operation of a PhC band edge laser at telecom wavelengths at room temperature, to the best of our knowledge. The active InP/InGaAs PhC slab is transferred on to Silicon by Au/In metallic dry bonding [9]. The choice of a high-reflectivity metallic layer provides us simultaneously with two advantages, an improved photon confinement inside the PhC resonator by an interference phenomenon [10], and an efficient thermal dissipation through the substrate. The present work is timely given the rapidly developing domain of hybrid structure devices. Such systems have recently received a lot of attention due to the exciting possibilities that they offer for monolithically integrating passive Silicon and active III–V devices into a unique and compact architecture [11–13].

2. Design and fabrication of the PhC structure

The epitaxial III–V material layers were grown by MOCVD. The full structure consists of a 265 nm thick InP slab incorporating four lattice-matched In_0.53Ga_0.47As QWs, emitting near 1.53 μm. The scheme of the hetero-structure is shown in Fig. 1(a).

 

Fig. 1. (a). Schematic hetero-structure. (b) SEM top image of the PhC structure.

Download Full Size | PDF

The 2D PhC structure was designed to provide a slow group velocity Bloch mode with high Q factor and strong light confinement near 1.55 μm. The in-plane wave vector was positioned near to the Γ point of the Brillouin zone to enable vertical device operation. The Γ point operation is also characterized by a small angular cone for the radiative modes. Further control over the optical losses is obtained by placing a high reflectivity layer below the PhC structure, consequently increasing the photon lifetime in the resonator. The distance between the bottom mirror and the active layer was judiciously chosen with the view to inducing destructive interference of vertical radiative losses [10], thereby enhancing the quality factor of the Bloch mode resonator. We chose SiO₂ as the material spacer, to benefit from the high index contrast between InP and SiO₂, leading to a strong optical confinement in the InP-based PhC slab. The thickness of SiO₂ layer is ≈785 nm. The metallic mirror, made of an Au/In alloy, bonds the III–V PhC layer on top of a Silicon substrate. The InP slab was fully drilled to obtain a 2D graphite PhC structure of air holes. The graphite lattice constant and the hole diameter were chosen to be 745 nm and 290 nm, respectively. This positions the mode indicated by the arrow on the photonic band diagram [Fig. 2(a)] within the gain bandwidth of the QWs. A Scanning Electron Microscope image is shown in Fig. 1(b). The patterning of air holes into the semiconductor was made by inductive plasma etching through a silicon nitride mask, itself obtained by e-beam lithography and reactive ion etching.

 

Fig. 2. (a). 2D photonic band calculation. The arrow indicates the mode of interest. (b) H_z field component associated with the mode at the Γ point for a/λ = 0.263 (a = a_graphite/√3). The circles represent the air holes of the graphite lattice. Red and blue colors corresponds to positive and negative values, respectively. Light blue indicates the field antinodes.

Download Full Size | PDF

We characterized the spectral response associated with the lasing photonic resonance by pump and probe spectroscopy, in reflection at normal incidence [14]. In particular, we investigated the reflectivity spectrum of the PhC resonance under near-transparency condition for the gain material [15], in order to evaluate the Q factor of the unloaded resonator, which is about 8000.

3. Laser characterization

The experimental study was conducted by pumping the structure using a CW diode laser emitting at 830 nm. The pump beam is focused at normal incidence on the sample surface with a 10X microscope objective to a spot size of about 15 μm. The PhC laser emission in the vertical direction is collected using a fiber-coupled optical spectrum analyzer, with a 20 pm resolution. Figure 3(a) shows the PhC laser emission centered at 1586 nm, which is single-mode and linearly polarized, with a side mode suppression ratio of 28 dB. Figure 3(b) shows the same laser spectrum but zoomed in to reveal a Full Width Half Maximum (FWHM) value as small as 70 pm. Figure 3(c) shows the far field radiation pattern as recorded on a CCD infrared camera. The lasing photonic mode displays a doughnut-shaped emission pattern.

 

Fig. 3. (a). Lasing spectrum under CW optical excitation at room temperature. (b) Zoom in of the laser spectrum. (c) Far field emission pattern.

Download Full Size | PDF

As shown in Fig. 2(b), the calculated H_z vertical field component of the mode of interest exhibits an even parity, forbidding the coupling to the radiative modes in the vertical direction [16]. This explains the observed node at the center of the emission profile.

The output characteristic of the CW PhC laser is plotted in Fig. 4(a) as a function of the excitation power. Fitting the experimental points gives a laser threshold of about 4.5 mW. The corresponding pump fluence is estimated to be about 0.6 μJ/cm², given the measured value of carrier recombination time of 215 ps [17]. Light output power in the vertical direction of ≈7 μW was measured at room temperature for an excitation power of about 5.2 mW. This value of output power is close to other results recently reported for a PhC laser [18]. The overall efficiency of the present laser device may be explained as follows. On one hand, the radiative recombination rate of electron-hole pairs is decreased because of the creation of non-radiative recombination centers at the interfaces of the etched holes. On the other hand, as in second order distributed feedback lasers, light is not only emitted in the vertical direction, but also in the plane of periodicity. This represents an additional limitation for the overall efficiency of the device as a surface-emitting laser. Figure 4(b) shows the FWHM of the emission spectra as a function of the excitation power. Above laser threshold, the linewidth reaches a minimum value of less than 60 pm. All the measurements in Fig. 4 were obtained without any modulation of the pump intensity. Despite this, as is obvious from Fig. 4(c) no significant spectral red-shift of laser mode is observed.

 

Fig. 4. (a). Laser output power at ≈1586 nm as a function of the CW excitation power. (b) Linewidth of emission for different values of excitation power. (c) Peak wavelength versus the excitation power.

Download Full Size | PDF

4. Comparison of different bonding methods

Figure 5(a) shows the laser intensity spectra measured under CW and pulsed excitation. This comparison allows an initial estimate of the thermal load affecting the device during continuous lasing operation. The laser characterization in pulsed regime made use of an ultrafast Ti:Sa pump source, delivering 130 fs pulses, at a repetition rate of 82 MHz and wavelength of 810 nm. The spot size is about 30 μm and the average (external) pump power is about 0.75 mW. The corresponding pump fluence is estimated ≈1 μJ/cm², which is comparable to the CW case. This very low value of threshold is related to the high Q photonic resonator. Comparing the wavelength position of pulsed and CW laser emission spectra we observe that the CW lasing takes place at a red-shifted wavelength, meaning that a substantial heating occurred in the PhC active layer. We also observed a remarkable difference in laser line width distinguishing the CW emission from pulsed one, the former being ≈9 times narrower. This is due to the spectral broadening of emission in pulsed regime. Figure 5(b) shows a CCD image of the far field emission pattern of pulsed laser. The same doughnut-like shape than in CW regime is observed, confirming that the same photonic mode is involved.

 

Fig. 5. (a). Laser intensity spectra under CW (red line) and pulsed (black line) excitation. (b) Far field emission pattern of pulsed PhC laser.

Download Full Size | PDF

The choice of appropriate materials, i.e., both the Au/In bonding and SiO₂ cladding layer appears decisive in enabling CW laser operation. In order to verify this statement, we compare the performance obtained by the present CW laser and another PhC laser device, both with the same InP/InGaAs active layer, but having different bonding and cladding layers. The reference device is an InP-based 2D PhC bonded on silicon by means of benzocyclobutene (BCB) layer. In that structure BCB acts as the cladding layer instead of SiO₂, and a gold-bottom mirror layer was also positioned between the BCB and the silicon substrate. Due to the rear gold mirror and the refractive index of BCB (n = 1.53) being very close to that of silica, the optical confinement in both devices is comparable in the vertical direction. The bulk thermal conductivity of BCB, K_th = 0.24 W/(m*K), is, however, ≈5 times lower than that of SiO₂, K_th = 1.26 W/(m*K) [19]. As reported elsewhere [20], the Au/BCB-based PhC structure provided laser emission peaked near 1590 nm under pulsed excitation, with threshold of about 3.4 μJ/cm². The corresponding photonic band-edge resonator displayed a Q factor comparable to that of the present device. Nevertheless, no CW laser operation was observed in the case of the Au/BCB sample.

5. Conclusion

In summary, we have demonstrated for the first time, to our knowledge, a CW surface-emitting PhC laser, integrated on a silicon chip and operating in the telecom range at room temperature. In order to reach CW PhC laser operation at room temperature several conditions must be fulfilled. A high Q photonic mode is required to provide the strong optical feedback to sustain laser oscillation. We used a band edge slow Bloch mode with Q ≈8000. The InGaAs QW active layer was placed in the middle of the InP slab and fully drilled by the 2D PhC, to optimize the superposition and interaction between the photonic mode and active material. Lastly, the most important advance of this structure towards obtaining CW lasing has consisted of using cladding and bonding materials which achieve better management of photon losses, allowing high-Q resonators, as well as better thermal sinking to remove heat from the active layer more efficiently.