1. Introduction

Optical communication systems predominantly function in the 1.3 µm and 1.55 µm bands, chosen for their minimal signal loss in optical fibers. Within this framework, the incorporation of self-assembled InAs quantum dots (QDs) as active regions, developed through the Stranski-Krastanov mode, has been a sustained area of focus for laser technologies operating at these specific wavelengths [1–3]. This interest is largely attributed to the unique three-dimensional carrier confinement in QDs, which leads to a δ-function density of states. This unique characteristic is anticipated to bestow QD lasers with superior performance traits compared to quantum well lasers. These advantages, as documented in a multitude of studies [4–7], encompass lower threshold current, diminished power consumption, higher gain, and enhanced temperature stability. Collectively, these attributes establish self-assembled InAs QDs as an efficient and robust light source for optical communications.

However, challenges arise in the development of 1.55 µm InAs/InP QD lasers, primarily due to a moderate lattice mismatch of 3.2% between InAs and InP. This relatively small mismatch, along with the complex strain distribution tends to cause the InAs to elongate, forming elongated nanostructures, i.e. quantum dash (Qdash), rather than the desired round-like dots, resulting in substantial size fluctuations and inhomogeneously broadened transition energy levels, thereby adding complexities to the process of epitaxial growth [2]. Typically, the InAs QDs (Qdashes) grown on InP substrates have a larger broadening photoluminescence (PL) spectrum, which is several times higher than those for QDs grown on GaAs substrates [8]. The broadening of the PL spectrum in InAs/InP QDs leads to a decrease in the saturated gain and an increase in the threshold current. Specifically, the broad spectrum causes the spectral gain associated with the ground state transition to saturate more rapidly. This early onset of saturation is a limiting factor in achieving high modal gain in lasers [9–13]. To address these issues, developing high-density InAs QDs with uniform size distribution is crucial.

Recent advancements in molecular beam epitaxy (MBE) have enhanced the fabrication of 1.55 µm InAs QDs, particularly addressing the morphological challenges associated with QD elongation. A key development involves the substitution of As₄ with As₂ during the InAs QDs growth process, facilitating the formation of InAs QDs with a more uniformly circular morphology. This refinement has been instrumental in fulfilling PL linewidths as narrow as 17 meV at 10 K [14]. Additionally, an “alternate growth” method of QDs via MBE has led to the five stacked InAs QDs, characterized by a full width at half maximum (FWHM) of 35 meV at room temperature (RT) [15]. The introduction of an ultra-thin GaAs prelayer before the growth QDs decreased the surface roughness of the InAlGaAs buffer, consequently enhancing dot uniformity with a high QD density of 5 × 10¹⁰cm⁻². The resultant five-stack QD PL exhibited a peak at 1.577 µm with a FWHM of 32 meV [16].

On the other hand, metal-organic chemical vapor deposition (MOCVD) remains an attractive deposition technique of choice for 1.55 µm QD lasers due to its ability to grow high-quality compound semiconductors with few defects at high temperatures and scalability to high volume production. For these reasons, the preparation of 1.55 µm InAs QDs by MOCVD has been widely carried out [8,17]. The InAs QD emission wavelength has been proven to be thermally stable during high-temperature growth of top cladding layers in laser structure by MOCVD, which suggests that MOCVD-fabricated lasers could potentially match or surpass those produced by MBE in performance [18,19]. Similar to MBE, the role of the ultra-thin GaAs layer between the InAs QD and the buffer layer has been studied in MOCVD. It is found that the GaAs sublayer plays a crucial role in suppressing the As/P exchange between InAs QDs and InGaAsP barriers, hence mitigating issues caused by the exchange reaction and aiding in tuning the QD emission wavelength towards 1.55 µm [18,20,21]. The fivefold stacked InAs QDs PL FWHM of 110 meV at 1.58 µm was obtained [22].

The double-cap (DC) technique, another growth method, effectively reduces QD height dispersion and tunes emission wavelengths [23]. This method employs a two-step capping process for QD fabrication. The first cap layer (FCL), thinner than the height of the QDs, is deposited at a temperature similar to that used for QD growth. This step is followed by depositing a second cap layer (SCL), which acts as a spacer. The process of As/P exchange or thermal annealing during growth interruption enables the homogeneous QD height distribution and planar growth fronts [24,25]. Luo et al. reported a reduction in PL linewidth to 87 meV for five-stack InAs/InGaAsP QDs using this method [26]. Shi et al. achieved an even narrower PL linewidth of 71 meV at RT for five-stack InAs/InAlGaAs QDs [27]. The DC technique in MOCVD, while effective in certain aspects, faces limitations in producing uniformly high InAs QDs on InP substrates possibly due to complex reaction processes [26]. Consequently, the 1.55 µm InAs QDs produced by MOCVD using the DC technique exhibit relatively broader PL spectra compared to those produced by MBE. For instance, the FWHM for five-stack InAs QD layers has been successfully reduced to 50.9 meV using MBE on (001) InP substrates [28]. Therefore, optimizing MOCVD parameters via the DC procedure is essential for reducing the optical linewidth of 1.55 µm emission InAs QDs and enhancing QD laser performance.

In this study, we optimized the growth of InAs QDs on (001) InP substrates via MOCVD, leading to a QD density of 4.13 × 10¹⁰cm⁻². This was attained at a V/III ratio of 9, with a deposition thickness of 4.5 monolayers (ML), and a growth temperature of 480 °C. Under these optimized conditions, a five-stack InAs QD layer exhibited a narrowed PL FWHM of 40.1 meV, utilizing the DC technique. This technique enhanced the homogeneity of QD heights and fine-tuned the emission wavelength to 1577 nm. Broad-area (BA) lasers, incorporating the five-stack InAs QD layer, were fabricated using standard optical lithography and wet etching techniques. The laser with a 50 µm ridge width and a 1 mm cavity length demonstrated a low threshold current density of 80 A/cm² per QD layer and a saturation power of 163 mW in continuous-wave (CW) mode at RT. These lasers demonstrated lasing spectra around 1.55 µm, which aligns with the target operational wavelength for 1.55 µm laser applications.

2. Methods

The InAs/InAlGaAs/InP QD structures were deposited on sulfur-doped (001) InP substrates at a low pressure of 100 mbar using a horizontal flow MOCVD system (Aixtron 200/4). The Group III reactive precursors used were trimethylgallium (TMGa), trimethylindium (TMIn), and trimethylaluminum (TMAl), while the Group V precursors comprised high-purity arsine (99.9999% AsH₃), and phosphine (99.9998% PH₃). Pure hydrogen (H₂) served as the carrier gas.

As depicted in Fig. 1(a), the initial epitaxial structures consisted of one buried QD layer intended for PL studies, along with an uncapped QD layer on the surface for atomic force microscopy (AFM) analysis. The substrate underwent deoxidation in a PH₃ atmosphere, followed by the growth of a 200 nm InP buffer layer and a lattice-matched In_0.521Al_0.148Ga_0.331As layer at 660 °C, which exhibited PL emission at 1265 nm. Subsequently, the substrate was cooled for QD deposition in the presence of AsH₃ to prevent surface decomposition. The growth rate was consistently held at 0.04 ML/s. The effects of V/III ratio ranging from 9 to 91 on the morphology of InAs QDs were investigated. A 30 nm In_0.521Al_0.148Ga_0.331As, covering the QDs, was grown at the same temperature as the QDs. Investigations focused on varying the thickness of QD depositions, ranging from 3.5 ML to 4.5 ML, and on adjusting the growth temperatures between 460 °C and 490 °C while maintaining a constant V/III ratio of 9. In these experiments, the thicknesses of the In_0.521Al_0.148Ga_0.331As FCL were set at either 1.2 nm or 0.8 nm. Then, the temperature was raised to 660 °C for the growth of upper In_0.521Al_0.148Ga_0.331As SCL and InP layers. The total thickness of FCL, and SCL situated beneath the upper 200 nm InP layer was 100 nm as shown in Fig. 1(b). In the case of the five-stack QD layers as in Fig. 1(c), a similar growth protocol was employed. The total spacer thickness between the multilayer QDs was established at 40 nm. To ensure homogeneity across uncapped QDs, a further 100 nm of In_0.521Al_0.148Ga_0.331As was deposited atop the upper InP layer. Finally, an uncapped surface QD layer was deposited using identical parameters to those of the buried QD layer for morphological studies.

 

Fig. 1. (a) Schematic cross-section diagram of initial InAs/InAlGaAs QD structures, where the 6.5 ML InAs QDs were grown at 490 °C at different V/III ratios. (b) The epitaxial structures with introduced upper InP layer containing single InAs/InAlGaAs QD layer. (c) Five-stack InAs /InAlGaAs QD layers with InAlGaAs FCL.

Download Full Size | PDF

The optical characterizations of the QD samples were conducted using the RPMBlue (Onto Innovation Inc.), equipped with an InGaAs photodiode detector. A 532 nm laser with an output power of 40 mW and power density of 128 W/cm², was used as the excitation source. The QD morphological studies were performed using the Multi-Mode 8 AFM (Bruker). The density and morphology of QDs were assessed by employing high-frequency non-contact mode for optimal imaging. All measurements were conducted at RT.

3. Results and discussions

3.1 Effects of V/III ratio in InAs QD

Figure 2(a-d) present AFM images depicting the impact of varying V/III ratios in both 5 × 5 µm² and 1.5 × 1.5 µm² areas. QD densities were quantified from the 1.5 × 1.5 µm² images, whereas the densities of the larger, coalesced islands were determined from the 5 × 5 µm² images. Figure 2(e) elucidates the relationship between the V/III ratio and the densities of both the QDs and the large coalesced islands. A trend was observed wherein a decrease in the V/III ratio of InAs led to an increase in QD density and a concurrent decrease in the density of large coalesced islands. Specifically, at V/III ratios of 91, 60, 30, and 9, the QD densities were recorded as 6.9 × 10⁹cm⁻², 1.26 × 10¹⁰cm⁻², 2.05 × 10¹⁰cm⁻², and 3.76 × 10¹⁰cm⁻², respectively. The corresponding densities of coalesced islands were 3.2 × 10⁸cm⁻², 2.3 × 10⁸cm⁻², 2 × 10⁸cm⁻², and 2.2 × 10⁸cm⁻². Additionally, at higher V/III ratios (e.g., 91), the islands exhibited irregular geometries, whereas, at lower V/III ratios, the enhanced diffusion capacity of indium adatoms led to more nucleation sites and QDs tended to be more circular, suggesting that lower V/III ratios favor the formation of round-like QDs, consistent with previous studies [29,30].

 

Fig. 2. (a-d) The 5 × 5 µm² AFM images of 6.5 ML InAs QDs grown at 490 °C with V/III of 91, 60, 30, and 9 are shown above from left to right, and the corresponding 1.5 × 1.5 µm² images are shown below. (e) The plot of QD density and large coalesced island density with V/III ratio. (f) The PL spectra of different V/III ratios, the inset is the variation of PL intensity and FWHM with V/III ratio. The FWHM values of various V/III ratios are marked in the inset.

Download Full Size | PDF

In Fig. 2(f), the PL spectra of QDs are presented across a range of V/III ratios, accompanied by an analysis of PL intensity, and FWHM. It is observed a discernible variation in PL intensity corresponding to different V/III ratios. A general trend of decreasing PL intensity was noted with increasing V/III ratios. This decreased intensity was attributed to the increase of coalesced dots, leading to the degraded intensity. The broadened PL spectrum was found at a high V/III ratio, due to the strong size fluctuation of QDs. In contrast, a reduction in the V/III ratio was found to be associated with a narrowing of the PL spectra, indicative of a more homogeneous distribution of QD sizes. Remarkably, the minimum FWHM, measuring 41.6 meV, was attained at the lowest V/III ratio of 9.

3.2 Influences of InAs deposition thickness, growth temperature, and the DC technique

An upper InP confinement layer was introduced to enhance the PL intensity, and the DC technique was used to tune the emission wavelength and improve the homogeneity of QDs. Figure 3(a-b) present the 5 × 5 µm² and 1.5 × 1.5 µm² AFM images of different QD depositions of varying thickness, respectively. Figure 3(c) shows the height histograms of QDs based on AFM results from 1.5 × 1.5 µm² AFM images, and Fig. 3(d) describes the corresponding PL spectra of single buried InAs/InAlGaAs QD layers with thicknesses of 3.5 ML, 4 ML, and 4.5 ML, all grown at 470 °C. The InAlGaAs FCL in these samples maintained a consistent thickness of 1.2 nm. Significantly, the AFM images revealed an absence of large coalesced islands in Fig. 3(b). The predominant formation of QDs, as opposed to Qdashes, can be attributed to the smooth growth front of InAlGaAs buffer grown at 660 °C [31].

 

Fig. 3. (a) The 5 × 5 µm² AFM surface images for 3.5 ML, 4 ML, and 4.5 ML InAs QD grown at 470 °C. (b) The corresponding 1.5 × 1.5 µm² AFM images for various deposition thicknesses. (c) The height histogram of the surface QDs, the dashed and solid lines are Gaussian component and fitting results. (d) The PL spectra for 3.5 ML, 4 ML, and 4.5 ML InAs QD. The V/III ratio was maintained at 9 and the FCL thickness was 1.2 nm.

Download Full Size | PDF

In 1.5 × 1.5 µm² AFM images of Fig. 3(a), the QD densities for the 3.5 ML, 4 ML, and 4.5 ML depositions were measured as 2.7 × 10¹⁰cm⁻², 3.2 × 10¹⁰cm⁻² and 4 × 10¹⁰cm⁻², respectively. An observed increase in QD density correlated with the increase in deposition thickness is attributed to more nucleation sites from increased indium adatoms deposition on the growth surface. Gaussian fitting of the QD height histograms revealed that the peaks of Gaussian curves rose with increasing deposition in Fig. 3(c). These heightened QDs resulted in lowered ground energy levels, thereby causing a redshift in emission wavelength. As shown in Fig. 3(d), the emission wavelengths observed were 1529.1 nm, 1585.8 nm, and 1624.9 nm for 3.5 ML, 4 ML, and 4.5 ML depositions, respectively. When the deposition thickness increased, an enlargement in the size of all QDs, particularly in their height, was expected, which was correlated with a redshift in the PL spectra. A decreasing trend in PL intensity with increasing deposition was also noted, likely due to the formation of more strain-induced defects on the QDs surface. The strain-induced defects worsened recombination intensity, rendering the dots optically inactive, which may be responsible for PL intensity deterioration with increased QD deposition thickness. Instead, the slight variations of PL FWHM in these QDs indicated that the homogeneous QDs were well kept during the change of deposition thickness process. Although the 3.5 ML and 4 ML QDs outperformed the 4.5 ML in terms of emission wavelength and PL intensity, the higher QD density of the 4.5 ML QD made it more favorable for high modal gain per QD layers for applications of lasers.

Figure 4(a-c) showcase AFM images of InAs QDs, revealing how varying growth temperatures affect QD density. At temperatures of 460, 470, and 490 °C, the QD densities were 2.74 × 10¹⁰, 4.01 × 10¹⁰, and 3.35 × 10¹⁰cm⁻², respectively. The increase in growth temperature led to more pronounced indium adatom diffusion, hence a higher nucleation density. However, a notable decrease in QD density at 490 °C was observed, attributed to the increased evaporation of QDs due to elevated In and As adatoms desorption rates [26]. Figure 4(d) displays the PL spectra for these QDs, while Fig. 4(e) compares their emission wavelengths and FWHMs. A blueshift in the PL emission peak was noted as the temperature increased, with the highest PL intensity at 460 °C. This trend can be explained by the enhanced desorption rates of In and As adatoms at higher temperatures, affecting both QD size and PL intensity. Additionally, the varied adatom diffusion at elevated temperatures brought inconsistent dot size distributions, broadening the PL FWHM with increasing temperature.

 

Fig. 4. (a-c) 1.5 × 1.5 µm² AFM surface images of V/III = 9, 4.5 ML InAs QDs grown at 460 °C, 470 °C, and 490 °C. (d) The corresponding PL spectra, (e) the extracted wavelength and FHWM of InAs QDs. (f) the PL spectra of 4.5 ML QDs grown at 470 °C with different FCL thicknesses of 1.2 nm and 0.8 nm.

Download Full Size | PDF

For optimal modal gain in QD lasers, QDs developed at 470 °C were considered ideal for such applications due to their high density. However, the emission wavelength of these QDs, positioned at 1624.9 nm, did not align with the preferred 1550 nm band. Implementing the DC technique and reducing the FCL thickness to 0.8 nm facilitated a blueshift in emission to 1574.7 nm, as evidenced in Fig. 4(f). Yet, this adjustment led to a broader FWHM of 55.4 meV and slightly reduced intensity; the latter is likely due to less effective confinement of photo-generated carriers within the smaller-sized dots.

3.3 Optical properties of five-stack QDs

Five-stack QDs structures were grown using previously optimized parameters, which included a 4.5 ML deposition thickness, a temperature of 470 °C, and 0.8 nm FCL. The PL peak wavelength and FWHM of the five-stack QDs were in good agreement with those of single-layer QDs, as shown in Fig. 5(a) by the dashed line. The PL intensity was improved compared to the single-layer QDs due to the multilayer gain medium. The foregoing discussion on QD growth temperature revealed that the maximum density was reached at 470 °C, whereas QD density decreased above 490 °C due to increased adatom desorption at high temperatures. Speculating that 480 °C might further increase the density of QDs, five-stack QDs were prepared at this temperature. The 1.5 × 1.5 µm² AFM image presents a density of 4.13 × 10¹⁰cm⁻², slightly increased compared to the 470 °C counterpart. Unexpectedly, a narrower PL linewidth of 46.5 meV was observed at 480 °C, as depicted with a dotted line in Fig. 5(a). Furthermore, the previous discussion on single-layer QDs with FCL thicknesses of 1.2 nm and 0.8 nm suggested that a thickness within this range could reach a wavelength of around 1.55 µm, high PL intensity, and low PL FWHM. Consequently, a five-stack QDs structure with a 1 nm FCL, grown at 480 °C, demonstrated a PL linewidth of 40.1 meV and an emission wavelength of 1577 nm, maintaining consistent PL intensity, as shown by the solid line in Fig. 5(a). Figure 5(b) summarizes the PL FWHM values of InAs QDs reported so far using MOCVD [8,22,26,27,32–35], highlighting the enhancement in QD size distribution homogeneity achieved in this work.

 

Fig. 5. (a) The five-stack QD structures PL spectra with different growth temperatures and FCL thickness, the inset was the 1.5 × 1.5 µm² AFM images of 4.5 ML QD grown at 480 °C, (b) the PL FWHM of InAs QD grown by MOCVD in this work compared with other works.

Download Full Size | PDF

3.4 Device fabrication and characteristics

To characterize the five-stack InAs QDs performance for laser applications, BA lasers with a cavity length of 1 mm were fabricated using standard optical lithography and wet etching technology. The back facets of lasers were coated with a 97% high-reflection (HR) coating at 1550 nm while the front facets were as-cleaved. As shown in Fig. 6(a), the laser structure consists of five-stack QD layers grown at 480 °C with 4.5 ML InAs, separated by 40 nm thick In_0.521Al_0.148Ga_0.331As layers. The lower n-doped cladding layers were 200 nm InAlGaAs and 600 nm InP layers, while the upper p-doped cladding layers were 300 nm InAlGaAs and 1700nm InP. As the p-contact layer, a heavily p-doped InGaAs layer was deposited.

 

Fig. 6. (a) The schematic of epitaxial structure for laser fabrication. (b) The P-I curves of BA QD lasers with three ridge widths and a 1 mm cavity length under CW mode at RT. (c) The V-I curves of lasers corresponding to Fig. 6(b). (d) An enlarged view of the threshold current of BA lasers. (e) The normalized lasing spectrum of BA QD lasers at 1.2 I_th under CW mode at RT. (f) The lasing spectra of 100-µm-width laser in CW mode for different currents.

Download Full Size | PDF

Figure 6(b) and 6(c) show the output power versus injection current (P-I) and voltage versus injection current (V-I) curves for BA lasers in CW mode at RT. The BA lasers, with ridge widths of 25, 50, and 100 µm, reached saturated output powers of 95, 163, and 189 mW, respectively. Figure 6(d) presents the P-I curves of BA lasers operating near threshold current. The corresponding threshold current densities were 560 A/cm², 400 A/cm², and 500 A/cm², equating to per-layer threshold densities of 112 A/cm², 80 A/cm², and 100 A/cm². These values are notably lower than those reported in previous studies, such as those by Li et al. (220 A/cm²), Luo et al. (447 A/cm²), and Semenova et al. (1000 A/cm²), who demonstrated higher threshold current densities [35–37]. This lower threshold is attributed to a narrower QD size distribution, indicated by the reduced PL FWHM from prior optimization.

Figure 6(e) presents the normalized lasing spectrum at 1.2 times the threshold current under CW conditions. In devices with ridge widths varying from 25 µm to 100 µm, multi-mode lasing emissions were observed, with wavelengths ranging from 1545 nm to 1552 nm. The observed redshift in the lasing wavelength as the ridge width increased can be attributed to thermal effects during CW mode operation. Figure 6(f) illustrates the lasing spectra of a 100 µm wide laser at varying injection currents: at the threshold current, and at 1.1 and 1.5 times the threshold current. The lasing predominantly occurs around 1550 nm, but with increasing injection current, there is a noticeable shift to longer wavelengths. At higher injection currents, the increased current leads to a broadening of the emission spectrum, reflecting the laser spectrum showing a multi-mode envelope peak.

4. Conclusions

In summary, the growth conditions, including V/III ratio, deposition thickness, and growth temperature in MOCVD, have been optimized. Optimal conditions achieved a QD density of 4.13 × 10¹⁰cm⁻² at the deposition thickness of 4.5 ML, growth temperature of 480 °C, and a V/III ratio of 9. This led to a narrowed PL FWHM of 40.1 meV and an emission wavelength shortened to 1577 nm for five-stack InAs QDs, utilizing a lattice-matched In_0.521Al_0.148Ga_0.331As FCL through the DC technique in MOCVD. BA lasers fabricated with these parameters demonstrated a low threshold current of 80 A/cm² per QD layer and a saturation power of 163 mW in CW mode at RT, with the lasing spectrum around 1550 nm. Our results offer valuable insights for the development of high-performance QD lasers operating at the 1.55 µm band on (001) InP substrates.