1. INTRODUCTION

The early demonstration of lasing emission in a photonic-crystal microcavity (PCM) [1] opened new avenues toward very low threshold and highly efficient solid-state lasers [2,3] by taking advantage of two aspects. On one hand, PCMs allow a strong confinement of light through high quality factors ($Q$) and small mode volumes ($V_{\text{eff}}$). On the other hand, zero-dimensional nanostructures or quantum dots (QDs) were proposed as an optimal active region to reach low threshold and highly efficient laser sources [4,5]. Even though it has been argued that a truly thresholdless laser is not possible in practice [6], the term thresholdless is used in the literature [3] to identify lasers presenting two main features: a spontaneous emission coupling factor ($\beta$) close to 1 and low nonradiative losses. Nonradiative losses are reduced by several orders of magnitude at cryogenic temperatures, although they can never be completely suppressed. An ultimate thresholdless laser [3,7,8] operating at room temperature (RT) may have a strong impact in optical integrated circuits [9], which may require performance at temperatures as high as 85 °C and, for instance, in bio-organic sensing, where the 1.3 μm spectral window has been used for single-cell photonic nanocavity probes [10]. Ultralow threshold lasing was achieved using an ever-decreasing number of QDs within photonic-crystal cavities [11 –16]. That strategy was adopted by Strauf et al. [12] to demonstrate near-thresholdless lasing at low temperature (4.5 K) using a few QDs (two to four) as active emitters. The authors reported power threshold values as low as 124 nW (corresponding to an absorbed power of 4 nW) and a high $\beta =0.85$. Khajavikhan et al. recently demonstrated thresholdless lasing at low temperature (4 K) [17]. The authors also reported RT lasing, although without thresholdless characteristics. In this work we report a RT continuous wave (cw) laser with emission characteristics close to those of an ideal thresholdless laser.

2. DESIGN, FABRICATION, AND CHARACTERIZATION OF THE LASERS

A. Design

We have designed a PCM that consists of a hexagonal lattice of air holes with nine missing holes along the $\mathrm{\Gamma }\mathrm{K}$ direction (L9-PCM) fabricated on a GaAs suspended slab [18]. For that PCM, the best $Q/V_{\text{eff}}$ ratio is obtained for the fundamental mode. Therefore, the design of the L9-PCM was optimized to achieve spectral matching of the fundamental mode with the emission of the QDs. Figure 1(a) shows the mode profile of the L9-PCM fundamental mode, calculated by the finite-difference time-domain (FDTD) method (see Supplement 1). The calculated values for $Q$ and $V_{\text{eff}}$ are $4.4\times {10}^5$ and $1.43\times {(\lambda /n_{\text{eff}})}^3$ respectively, where $\lambda$ represents the spectral position and $n_{\text{eff}}$ is the effective refractive index of the fundamental mode.

 

Fig. 1. (a) Calculation of the electric field distribution ${|E|}^2$ of the L9-PCM fundamental mode. (b) Scanning electron microscopy image of a L9-PCM. (c) PL of the ensemble of the QDs outside of the PCM (black line) and PL of a L9-PCM (filled gray) showing the mode structure. The inset shows a schematic diagram of the epitaxial material.

Download Full Size | PDF

B. Fabrication

The fabrication of the L9-PCM involved a ${\mathrm{BCl}}_3/{\mathrm{N}}_2$-based plasma etching procedure, optimized for the enhancement of the $Q$ of the fundamental mode [19]. Figure 1(b) shows a scanning electron microscope image of the resulting L9-PCM.

The epitaxial material was grown by solid-source molecular beam epitaxy (MBE) on a semi-insulating GaAs substrate. The inset in Fig. 1(c) shows the epitaxial structure of the wafer, which consisted of the active region embedded in a 190 nm thick GaAs layer and a 1 μm thick ${\mathrm{Al}}_{0.75}{\mathrm{Ga}}_{0.25}\mathrm{As}$ sacrificial layer underneath. The active region includes a single layer of InAsSb QDs with luminescence at 1.3 μm. The microcavities were patterned by electron beam lithography on a 365 nm thick ZEP-520A resist homogeneously spun onto a 90 nm thick ${\mathrm{SiO}}_{\mathrm{x}}$ hard mask deposited by plasma enhanced vapor deposition (PECVD). Reactive ion beam etching (RIBE) was performed to drill the ${\mathrm{SiO}}_{\mathrm{x}}$ layer previously deposited on the epitaxial material and inductively coupled plasma-reactive ion etching (ICP-RIE) was used to transfer the pattern to the active GaAs slab. The final step consisted of the removal of the sacrificial layer underneath the L9-PCMs by means of $\mathrm{HF}\text{:}{\mathrm{H}}_2\mathrm{O}$ wet etching. Four large rectangular holes surrounding every structure were fabricated to facilitate the renewal of the HF solution and the removal of the remaining material underneath every structure during the wet etching [19].

C. Optical Characterization

A 785 nm laser diode (LD) operating in cw mode was used for the optical pumping of the photonic structures. The excitation beam was focused down to a 1.5 μm diameter spot by means of a microscope objective ($20\times$, $NA=0.4$). The same objective was used to direct the light emitted from the sample to a 0.85 m long double spectrometer through an optical fiber. An InGaAs-cooled photodiode array was used as a detector.

3. RESULTS AND DISCUSSION

Figure 1(c) shows the photoluminescence (PL) spectra at RT of the QD ensemble (black) and the L9-PCM (filled gray). The spectrum of the QD ensemble shows the fundamental transition that corresponds to the excitonic emission centered at 1280 nm with an inhomogeneous broadening of 23 meV. The incorporation of Sb in the InAs QDs induces an upward shift of the conduction and valence band edges; it results in a deeper hole confinement and more efficient emission at RT [20 –24], thus overcoming the thermal ambipolar escape of carriers that reduces the RT efficiency of conventional InAs QDs [25,26]. At low temperature, the performance of the InAsSb QDs is similar to that of regular InAs QDs. However, the emission of the InAsSb QDs clearly outperforms that of regular InAs QDs at RT [21,22]. Therefore, we take advantage of the good optical properties of our InAsSb QDs at RT: the integrated emission of InAsSb QDs at 28 K is 1.8 times that of InAs QDs, but at RT it is 10.2 times larger [24]. The study of the emission of the devices as a function of the temperature is not straightforward due to the thermal drift of the QD emission, which is much larger than that of the cavity mode. Therefore, the cavity mode will be highly detuned as the temperature is changed. The only way to make this study is to use a different cavity for each temperature, but that will not be conclusive because the devices cannot be made completely identical. The observed resonances from the PL spectrum of the L9-PCM correspond to the cavity mode structure, with the fundamental mode emitting at 1286 nm.

Figure 2(a) presents the integrated intensity of light emitted by the fundamental mode of the L9-PCM laser as a function of cw optical pump power (i.e., ${\mathrm{L}}_{\text{in}}$ versus ${\mathrm{L}}_{\text{out}}$ or LL-curve) at RT. We have estimated an effective power lasing threshold of 860 nW (see Supplement 1). To determine $\beta$, we used a coupled rate equation model where $\beta =0.85$ optimizes the fit to the experimental data (see Supplement 1). Figure 2(a) also shows the calculated LL-curves for different $\beta$ values (gray curves), including the fit for $\beta =0.85$ (red curve). This value is comparable to those reported in state-of-the-art lasers designed for their operation at temperatures few degrees above absolute zero [12,17]. The assertion that the device indeed reaches lasing is further substantiated by the linewidth dependence on pump power [Fig. 2(b)]. Three regions corresponding to different emission regimes are identified [17] from the linewidth behavior: the PL region, below threshold, where the light is predominantly spontaneously emitted [27]; the lasing region, above threshold, where light is mostly influenced by the stimulated emission; and the amplified spontaneous emission (ASE) region around the threshold [gray region in Fig. 2(a)] as a transition between the two regimes. In the ASE region, a small plateau in the evolution of the linewidth ($\mathrm{\Delta }\lambda$) with the power is observed [12,13,17,28], whereas a linewidth-narrowing behavior can be measured both within the PL and the lasing regions [17]. Such behavior is observed in our device at RT.

 

Fig. 2. (a) Light-in versus light-out (LL-curve) characteristics of the L9-PCM laser for different $\beta$ values (gray lines) in logarithmic scale; the best fit (red line) is for $\beta =0.85$. (b) Evolution of the linewidth of the L9-PCM resonant mode with the excitation power. (c) Analysis of the differential efficiency calculated from the LL data (dots) and from the fit for $\beta =0.85$ (line). Error bars in (b) are referred to the statistical deviation analysis. Gray region in (a) and (b) marks the ASE region.

Download Full Size | PDF

The $Q$-factor, given by $\lambda /\mathrm{\Delta }\lambda$ at the lowest excitation power ($P_{\mathrm{exc}}=370\mathrm{nW}$), is $Q=7400$ [27]. As the excitation power increases, the optical mode becomes narrower due to gain mechanisms in lasing systems [29]. At high excitation power above threshold ($P_{\mathrm{exc}}=10.14\mathrm{\mu }\mathrm{W}$), $\lambda /\mathrm{\Delta }\lambda =12100$ (see Supplement 1). Spontaneous emission enhancement [i.e., Purcell enhancement (PE)] has been widely measured in nonlasing microcavities containing self-assembled QDs [30]. Nevertheless, in a laser microcavity one has to be especially careful because the contribution of stimulated emission begins at very small excitation powers below the threshold. We have shown previously that the PE in a lasing L7-InP microcavity embedding InAs self-assembled quantum wires and exhibiting a high-$\beta$ value of 0.1 can only be determined from an extrapolation of the experimental data to zero pump power [27,31]. Reducing the excitation power to such small values makes impossible the measurement of the PE in the present devices. Using the approach of [32] and assuming no cavity–emitter detuning, an homogeneous broadening of 1 meV results in a PE $\sim 20$. As we are operating at RT, the homogeneous linewidth increase (up to around 10 meV [33]) results in a maximum PE of $\sim 2$. This is, however, compatible with a high-$\beta$ value, since the main parameter determining $\beta$ is the spontaneous rate into nonlasing modes [34]. This point has been also reported for nanobeams [14] and subwavelength lasers [35].

The evolution of the linewidth of the fundamental mode with the excitation power at RT follows a trend similar to that observed in near-thresholdless lasers at 4.5 K [3,12]. Finally, an important feature of a laser system is the differential efficiency (DE) given by the derivative ${\mathrm{dL}}_{\text{out}}/{\mathrm{dL}}_{\text{in}}$ [28]. For a lasing device, the DE increases with the power, tending to a constant value. The DE at RT is represented in Fig. 2(c) (dots) with the calculated DE curve (line) for $\beta =0.85$. The calculated DE follows the experimental data and deviates at excitation powers well above the threshold. This effect is due to the saturation of the QDs, which was not considered in the model.

A further confirmation of lasing could be provided from the analysis of the second-order correlation function, $g^{(2)}$ [36]. We have measured before the $g^{(2)}$ function of single QDs emitting at 980 nm in a Hanbury–Brown and Twiss interferometer [37]. Nevertheless, measurements at the emission wavelength of the InAsSb QDs (around 1300 nm at RT) are too noisy to extract reliable information. The use of pulsed excitation can enhance the signal-to-noise ratio [38,39], but in that case the induced excitonic dynamics are difficult to be transferred from a pulsed regime to cw operation.

In summary, the present experimental results are consistent with near-thresholdless laser operation in the 1.3 μm telecom window at RT for a system combining a single layer of InAsSb QDs with a PCM. We showed that thermally activated processes, such as nonradiative recombination, are not an insurmountable obstacle to the realization of laser sources with characteristics similar to those observed in near-thresholdless lasers at temperatures a few degrees above absolute zero. Such highly efficient systems are very promising candidates for applications in optical integrated circuits, potentially enabling low consumption electrically injected devices based on PCMs [40] at RT.