However, a new problem arises when phosphides are not used. The problem lies in the ease of migration of indium adatoms due to the high indium content, making it challenging to form InAs QDs. In such cases, quantum dashes that are elongated in the\(\left[1\overset{\overline{}}{1}0\right]\) direction, which promotes indium adatom migration, can be formed. In particular, quantum dash structures [11] have been reported to exhibit luminescent thermalization at high temperatures due to their reduced carrier confinement dimensions compared to QDs. This has a significant impact on laser performance [12, 13]. Several C/L-band lasers employing only As-based growth materials have been reported, however they are either not grown on on-axis InP (001) substrates [14, 15] or utilize additional structures like tunnelling junctions [16]. These factors limit their practicality and applicability.
In this study, we aim to overcome these limitations and explore the growth of a QD laser using only III-arsenide layers on an on-axis InP(001) substrate with a standard QD structure. By focusing on on-axis growth and utilizing only arsenide materials, we can achieve a more practical and efficient laser design. The findings of this study demonstrate successful lasing in the L-band with a threshold current density of 633 A/cm2. This achievement represents a significant milestone as it sets a new record for the lowest threshold current density in the 1.6 μm-wavelength region for a laser grown on an exact InP(001) substrate.
Experimental method: We utilized a RIBER Compact21 DZ solid source molecular beam epitaxy (MBE) system for sample growth. The Group-III materials Gallium (Ga), Indium (In), and Aluminum (Al) were selected for the experiment, with standard dual filament sources providing these necessary materials. Arsenic dimer (As2) served as the Group V material throughout the entire growth process. We heated the substrate to a temperature of 530 °C under an Arsenic atmosphere, a step necessary to eliminate the oxide layer. Silicon (Si) and Beryllium (Be) were chosen as n-type and p-type doping materials, respectively.
We grew the epitaxial layer structure on a quarter of a 3-inch n-type InP (001) substrate. Directly on the InP substrate, we grew the first layer, an In0.52Al0.48As lower buffer layer, which notably features a low refractive index (3.20 at 1550 nm). Following the lower clad layer, we then grew a 100-nm thick In0.52Al0.24Ga0.24As waveguide layer. Both layers were lattice-matched to the InP substrate. Subsequently, we grow a 5-layer stacked QD structure, beginning with a 1-nm GaAs layer to promote QD formation, followed by the deposition of 3 monolayers (MLs) of InAs.The InAs QDs were subsequently capped with a 21-nm In0.52Al0.24Ga0.24As spacer layer. After completing the QD layer growth, another 100-nm thick In0.52Al0.24Ga0.24As As layer was grown, followed by a 1700-nm thick In0.52Al0.48As upper clad layer. As the final step in the growth process, we grew a 200-nm thick In0.53Ga0.47As p-type contact layer. The complete epitaxial layer structure can be shown in Fig. 1. We subsequently fabricated the prepared sample into a traditional Fabry-Perot (FP) laser. Both p-type and n-type electrodes utilized AuGeNi/Au as the electrode material. Utilizing a metal mask, we deposited the p-type electrode in a stripe pattern with a width of 100 μm. The n-type electrode was deposited on the backside of the InP substrate. Following these steps, the sample was then cleaved into multiple lengths, ranging from 500 to 3500 μm. We chose not to apply high reflection coating on the laser facets.
Results and discussion: The laser bar samples were characterized through pulsed current injection, with pulse conditions set at a 1 μs signal, 999 μs delay, and 0.1% duty. A Peltier device actively controlled the sample stage temperature. The growth process was evaluated using atomic force microscopy (AFM) to assess surface flatness, QD size, and areal densities. Photoluminescence (PL) was used to characterize the emission spectra.
Figure 2 displays an AFM image of the grown QDs. Due to the limited migration length of indium adatoms on the GaAs thin layer, QD formation exhibited minimal elongation in the\(\left[1\overset{\overline{}}{1}0\right]\) direction. The resulting QDs demonstrated an areal density of 1.61 × 1010 cm-2, an average height of 10 nm, and an average diameter of 56 nm. Figure 3 displays the light-current (L-I) characteristics of a Fabry-Perot (FP) laser at 25 °C, having a cavity length of 3049 μm and a width of 100 μm. The laser’s threshold current (Ith) and threshold current density (Jth) were measured at 1.93 A and 633 A/cm² respectively. The sample displayed multi-mode lasing with a fundamental lasing wavelength of approximately 1620 nm. The device exhibited a series resistance of 3.4 Ω, and a slope efficiency of 1 mW/mA was determined. Figure 4 shows the correlation between the reciprocal of the average threshold current density and the cavity length, demonstrating a reduction in the threshold current density as a result of mirror loss, which increases with the cavity length. The projected threshold current density is 48 A/cm² for an infinite cavity length. Remarkably, both the threshold current density and the projected threshold current density for an infinite cavity length, represent the lowest values ever reported for a 1.6 μm QD laser.
Conclusion: All III-arsenide InAs QD lasers on InP with low threshold current density have been successfully demonstrated. Each epitaxial layer of the QD laser was grown through MBE, and no phosphorus sources were employed. The fabricated laser demonstrates the threshold current density of 633 A/cm2, which is the minimum value for QD lasers in the 1.6 μm-wavelength range. This result suggests a high cost-effectiveness and paved the way toward a large-scale production technology for high-performing C/L/U-band QD lasers.