1. Introduction

The 2 µm waveband is beneficial for various photonic and optoelectronic applications in gas sensing, optical ranging, and next-generation telecommunication [1–8]. This wavelength regime is very useful for sensing environmental gases like CO₂, H₂O, and CH₄, which have fingerprint vibrational modes near 2 µm. As such, huge attention is being given to developing 2 µm avalanche photodiodes (APDs) and phototransistors as components for LiDAR systems, compact sensors, and spectrometers [1–4]. In data and telecommunication, the growing demand for higher data transmission has ushered the development of low-loss, low-nonlinearity hollow-core photonic bandgap (HC-PBG) fibers, which were found to have a minimum loss near 2 µm [6]. These fibers promise even higher capacities than current single mode fibers [7]. This waveband is also promising for silicon photonics as it exhibits a low two-photon absorption and a high Kerr nonlinear coefficient, which are both ideal for high-speed optical signal processing [9,10]. Although the development of APDs and novel optical fibers are burgeoning rapidly, the research on InP-based semiconductor laser gain materials emitting near 2 µm is proceeding at a slower pace.

In recent years, the research on 2 µm semiconductor laser diodes has been centered on quantum well-based (QW) lasers grown on GaSb substrates [11–13]. Although these lasers are well established and commercialized for certain applications, the manufacturing cost is still relatively high primarily due to the expensive GaSb wafer. Developing the lasers on cheaper InP substrates would be more advantageous as it would take advantages of the mature InP architecture built on the 1310 nm datacom and 1550 nm telecom technologies. Research on HC-PBG fibers frequently uses highly-strained InGaAs QW lasers on InP to study the viability of HC-PBG fibers [7,14,15]. More recently, innovations are being made in the fabrication of InGaAsSb QW lasers on InP. By using tapered waveguide, buried gratings, corrugated sidewalls, and heat dissipation strategies, 2 µm InP-based QW lasers have been shown to emit up to 81 mW [16,17].

However, decades of research have now established that lower dimensional gain mediums such as 0D quantum dots (QD) and quantum dashes (Qdash) can offer unique advantages over QWs, arising from their spatial and quantum confinement of carriers. QDs/Qdashes present improved thermal stability, defect tolerance, reflection insensitivity, and smaller linewidth enhancement factors [18–21]. These advantages have already been demonstrated for visible, O-band, and C-band lasers [19,22,23]. As such, developing a high-performance laser with QD or Qdash active gain materials is desirable in the long run.

The growth of InAs QDs or Qdash on GaAs or InP is very well-established for the 1310 nm, 1550 nm, and ∼1600 nm wavelength regimes [24–28]. However, extending the emission wavelength up to 2 µm wavelength proved to be difficult because the nanostructures need to be grown large and near their elastic strain limit, beyond which dislocations will inevitably form. Hence, one must carefully navigate the growth parameter space to obtain bright emission from InAs nanostructures. Previously we have successfully demonstrated the growth of InAs Qdashes on InP for 2 µm emission by optimizing the growth temperature, deposition thickness, As flux, ripening times, and punctuated growth interruptions [29,30]. Papatryfonos et al. also reported 2 µm Qdash distributed feedback (DFB) lasers [31,32]. However, the performance of their lasers was quite limited with relatively high threshold current density, probably due to an unoptimized Qdash active region design.

In this paper, we present room temperature continuous wave (CW) lasing from InAs/InP Qdash ridge waveguide lasers emitting at 1.97 µm enabled by punctuated growth. The punctuated growth increases Qdash photoluminescence intensity by 60%. We achieved a low threshold current density of 657 A/cm² under pulsed mode, 1183 A/cm² under QCW mode, and 1944 A/cm² under CW operation, respectively, for the 7.1 µm–wide laser. The lasers also exhibited good thermal stability, with T₀ value of 43 K. Due to their 0D nanostructure and 3D carrier confinement, the integration of Qdashes onto Si substrate should be interesting for on-chip light source for 2 µm silicon photonics.

2. Materials growth and device fabrication

All samples were grown on InP wafers using a solid-source molecular beam epitaxy system. The active region comprises 5 layers of 7 monolayer InAs Qdashes in a 10 nm In_0.53Ga_0.47As QW separated by 40 nm In_0.52Al_0.24Ga_0.24As (1.14 eV) barriers. The Qdash-in-a-well was grown at 510 °C using punctuated growth (PG), while the rest of the layers were grown at 500-510 °C as measured through optical pyrometry. In PG, the InAs deposition was subdivided into two 3.5 monolayer steps, with a 1-minute growth interruption in between. This was done to improve the photoluminescence (PL) intensity of the Qdashes. More details of the growth can be found elsewhere [29,30,33]. Figure 1(a) shows the room temperature PL spectra from calibration samples. These PL samples consist of buried Qdashes for optical property measurements and surface Qdashes for morphology studies. The PL from Qdashes grown with PG had 1.6× larger integrated PL emission compared to Qdashes that were continuously grown (CG). The inset shows atomic force microscope images of surface Qdashes. PG Qdashes are taller with slightly larger height distribution, giving rise to the slightly larger full-width at half max (FWHM) value in the PL spectrum (61.8 vs 66.8 meV). The blue-shifted emission (18 meV) is due to the change in quantum confinement effect as taller PG Qdashes are only partially buried by the InGaAs QW cap and become in close contact with the top InAlGaAs barrier. This blue-shift due to the change in potential barrier outweighs the red-shift due to the quantum size effect [29]. The laser structure shown in Fig. 1(b) follows a graded-index separate confinement heterostructure (GRINSCH) design to optimize both optical and carrier confinement [34]. The doping profile was also step-graded to minimize free carrier absorption near the active region. Figure 1(c) shows a bright-field transmission electron microscope (BF-TEM) image. The specimen was prepared by focused ion beam and oriented with [110] in the lateral direction, allowing us to see the cross-section of the Qdashes which are preferentially elongated along the [1–10]. No dislocations were observed under various two-beam axis conditions. Figure 1(d) shows a zoomed-in image of the active region. All the five Qdash layers were grown coherently with crisp QW/barrier interfaces. From the bottom to the top Qdash layer, there is an increase in size and a reduction in density, indicative of strain coupling between the Qdash layers. This can be improved by utilizing strain–compensating tensile layers in the spacer layers to cancel out the increasing compressive strain energy.

 

Fig. 1. (a) Effect of punctuated growth on the photoluminescence and morphology of quantum dashes. (b) Schematic of the entire laser structure. (c) Bright-field transmission electron microscope of the as-grown laser heterostructure, with [110] in the lateral direction. (d) Zoomed-in view of the 5 Qdash layers.

Download Full Size | PDF

The samples were then fabricated into narrow ridge waveguide lasers, shown schematically in Fig. 2(a). The laser cavity direction was oriented perpendicular to the Qdash elongation direction to maximize the gain and minimize the linewidth enhancement factor [35]. The mesas were defined using inductively-coupled reactive ion etching (RIE). A 300 nm thick SiO₂ layer was deposited by plasma-enhanced chemical vapor deposition to insulate the mesas from the probe pad. Probe pad vias were opened using RIE. Ti/Pt/Au and AuGe/Ni/Au were evaporated for the p- and n-metal contacts using an e-beam evaporator. The wafers were thinned down to 180–220 µm to facilitate cleaving. Rapid thermal annealing was performed at 400 °C for 1 min to improve the ohmic contacts. No high-reflection coating was deposited on the facets. Figure 2(b) shows a cross-sectional scanning electron microscope (SEM) image of a representative laser diode. The ridge widths were defined based on the average width of the active region as measured through SEM imaging instead of either the metal width or the mesa’s top width. For our targeted cavity widths of 4, 6, 8, and 10 µm, we estimated the actual active region to have average widths of 7.1, 8.7, 10, and 13 µm. Figure 2(c) displays an optical microscope of a cleaved bar with two finished laser diodes.

 

Fig. 2. (a) Three-dimensional illustration of the narrow ridge waveguide laser along with an AFM image. The cavity was oriented perpendicular to the direction of Qdash elongation. (b) Cross-sectional scanning electron microscope of a fabricated laser. (c) Optical microscope image showing the two fabricated ridge waveguide lasers. The laser presented has a cavity length of 1.3 mm.

Download Full Size | PDF

3. Results and discussion

Figure 3(a) shows room temperature light-current-voltage (LIV) curve from a 7.1 µm × 3 mm laser diode measured under various electrical pumping conditions. In SPW mode, a pulse width of 1 µs with 1% duty cycle was chosen to minimize the effect of heating under high current injection. In QCW mode, the pulse width was increased to 50 µs, long enough to approach steady-state conditions while minimizing the effect of heating. Increasing the pulse width to 100 µs yielded the same lasing behavior. The threshold current, i_th (threshold current density, j_th) for this device is 140 mA (657 A/cm²), 252 mA (1183 A/cm²), and 414 mA (1944 A/cm²) under SPW, QCW, and CW operation. These j_th values correspond to 131, 237, and 389 A/cm² per Qdash layer. Our SPW j_th values are the lowest j_th values reported to date for MBE-grown, Sb-free InAs QD or Qdash lasers on InP emitting from 1.85–2.1 µm [31,32,36–39], even lower than the j_th for a broad area laser measured under pulsed mode [36]. We believe that the Qdash morphology and optical properties granted by punctuated growth, the GRINSCH design, and doping profile all contribute to the low j_th observed. Our CW j_th value is comparable with previous ridge waveguide lasers with improved fabrication technology, including DFB lasers, DBR lasers, and lasers with high-reflection coating and optimal heat dissipation strategies. We expect that incorporating these strategies can further lower our i_th and j_th, approaching QCW values which we believe represent the ideal CW lasing performance for our material when processed under an optimal fabrication procedure [16].

 

Fig. 3. (a) Light-current-voltage curves of a 7.1 µm × 3 mm laser measured under SPW, QCW, and CW injection at 20 °C. The voltage was measured under CW injection. (b) Threshold current density at various cavity dimensions measured under SPW (solid circles) and QCW (hollow circles).

Download Full Size | PDF

Figure 3(b) summarizes the j_th of lasers with different cavity widths and lengths. In general, j_th drops as the laser cavity become longer. The lower j_th from longer cavities is due to lower mirror loss (α_m) contributions, which is inversely proportional to the cavity length, α_m = ln(1/R) L^-1. Therefore, a cavity half as long will suffer from mirror loss that twice as large. Based on these data, the transparency threshold current densities are 278 A/cm² (SPW) and 301 A/cm² (QCW) for the 7.1 µm × 3 mm device. The effect of sidewall recombination appears to be independent with cavity size. Our lowest j_th values were obtained from our narrowest mesas. Previous research done by Wan et al. has shown that Qdash lasers could suppress sidewall surface recombination due to lower carrier diffusion length in Qdashes as opposed to QWs [40]. They have also shown large variabilities in threshold current density which was attributed to the dry etching and metallization steps [40]. This width-independence could also be due to the less ideal current spreading through the trapezoidal mesas with non-vertical sidewalls. For the 7.1 µm wide laser, we observed a peak power of 15 mW under SPW (Fig. 3(a), inset). The largest peak power we observed was 32 mW for a 13 µm × 3 mm laser under SPW at 15 °C.

Figure 4(a)-(b) shows L-I curves obtained at different stage temperatures (setup limited) for the 7.1 µm × 3 mm laser under SPW (Fig. 4(a)) and QCW (Fig. 4(b)) mode. The characteristic temperature (T₀) was estimated by fitting the j_th values at different temperatures (Fig. 4(c)). The T₀ values under QCW and SPW modes are 34 K and 43 K, respectively. For wider laser cavities, the T₀ values ranged from 34-40 K under QCW and 42-45 K under SPW. These T₀ values are comparable to those from 2 µm InGaAsSb/InP QW lasers [41]. Unfortunately, our lasers lased CW up to 25 °C only, signifying device heating problems. Various heat management strategies are discussed later in this paper.

 

Fig. 4. (a-b) Temperature-dependent L-I curves for a 7.1 µm × 3 mm laser measured under (a) SPW and (b) QCW mode. (c) Temperature-dependent threshold current density versus stage temperature.

Download Full Size | PDF

The lasing spectra at various injection currents were measured using a monochromator and extended InGaAs photodetector that can detect up to a wavelength of 2.4 µm. The laser diode was wire-bonded onto a custom laser holder for ease of measurement. Figure 5(a) shows the photoluminescence spectrum of the as-grown laser sample prior to fabrication and the lasing spectra of the fabricated laser diodes measured 0.3 A (near threshold) and at 3 A under SPW at room temperature. The PL emission has a peak wavelength at 1.94 µm with a broad FWHM value of 302 nm (105 meV). The laser has a peak emission at 1.97 µm coinciding with the ground state energy of the Qdash ensemble. Figure 5(b) shows two light–current density (LI) curves from the same laser diode. The solid line was measured on a laser testing setup where the bar was mounted on a large copper heat sink, while the dotted line was obtained by integrating the electroluminescence spectra at various injection currents. There is good correspondence between the two measurements, with the discrepancy due to poorer heat dissipation of the custom laser holder used. Figure 5(c) shows the electroluminescence spectra at different injection currents. The laser exhibits multimode lasing centered at around 1.97 µm, but the modes could not be resolved due to the resolution limitations in our spectrometer setup. There is a clear broadening of the emission due to multimode lasing and device heating. At 0.3 A, the FWHM value is 4.5 nm, while for 3 A, it is 22 nm. We did not observe excited emission from our lasers.

 

Fig. 5. (a) Photoluminescence spectrum of the as-grown epitaxial structure and lasing spectra of the fabricated device at injection currents of 0.3 A (red) and 3 A (blue). (b) LI curves of an 8.7 × 2.5 mm laser measured on a copper heat sink (solid line) and integrated from current-dependent spectra (circles). (c) Emission spectra at various injection currents.

Download Full Size | PDF

The threshold current densities reported here could still be improved by applying high-reflective coating on the facets. Depositing 5 nm of Al₂O₃ via atomic layer deposition is also expected to improve the performance of these narrow mesas by suppressing sidewall carrier recombinations. Also, increasing the p-InGaAs contact layer thickness from 100 nm to 300 nm and improving the vertical aspect ratio of the mesas would reduce the hole spreading problem. Lastly, the large heating in our devices can be improved by electroplating In or AuSn on the back and bonding the laser onto an AlN submount. Alternatively, the laser can be flipped and bonded with the mesa down to improve heat dissipation in the active region.

4. Conclusion

In conclusion, we have demonstrated room-temperature, low-threshold InAs/InP Qdash lasers emitting at 1.97 µm. The ridge waveguide lasers lase at threshold current densities of 657, 1183, and 1944 A/cm² under SPW, QCW, and CW operation, respectively, for a 7.1 µm × 3 mm laser. We observed good thermal stability, with T₀ values of 34 K (QCW) and 43 K (SPW). In O-band communication, the adoption of QDs as laser gain materials is poised to offer advantages for silicon photonics and photonic-integrated circuits. We believe that the adoption of QDs or Qdashes as the laser gain material will also be a key to unlocking the 2 µm window for next-generation photonics and optoelectronics.