1. Introduction

Monolithic dual-λ semiconductor lasers are required for a number of applications that include dual-λ interferometry, spectroscopy, and especially for difference-frequency generation of terahertz radiation [1–4]. Different techniques have been developed to achieve dual-λ operation, such as laser sources utilizing two distributed Bragg reflectors (DBR) gratings [5], external cavity configurations [6], and Y-branch waveguide [7]. It is more realistic and significant for laser diodes using simple and compact structure, especially a traditional FP cavity, to accomplish simultaneous lasing at two well-defined wavelengths in the same optical path and along the same optical axis. However, simultaneous dual-mode lasing is usually forbidden in a common gain medium such like quantum wells (QWs), because stimulated recombination at a particular energy usually compete for carriers undergoing similar processes at other energies.

Simultaneous two-state lasing from QD based on FP cavity laser has been previously observed and discussed, which was attributed to the incomplete clamping of the excited state (ES) carriers population at the ground state (GS) threshold [8]. However, the quenching of the GS power with current increasing leads to an instable lasing spectrum and limited dual-λ operating range [9]. Some other reports demonstrated the two wavelength lasing phenomenon stems from carrier transition in different subbands of an inhomogeneously broadened QD material based on FP cavity, whereas the device characteristics such as the output power and linewidth are not attention-getting [10–12]. An alternative scheme of dual-λ lasing based on carrier transition in different dots has been presented by a numerical simulation using Monte Carlo markovian models [13]. It is predicted that coupled-cavity laser including QDs may operate in dual-mode far more easily than bulk and QW lasers because the carrier capture process in QD layers can be independent for each dot and the difference between homogenous broadening and inhomogeneous broadening helps decoupling the modes.

Bimodal-size distribution QDs provide subsets with obviously different sizes and therefore is a suitable model for dual-λ lasing. Previous research on growth condition, morphology, photoluminescence and growth kinetics have demonstrated that bimodal sized distribution of quantum dots could be achieved by deliberately varying the deposition conditions [14–17]. However, there are few reports on device, especially on laser diodes, using bimodal sized QDs as active material. Until recently, dual-λ lasing was first observed from different subsets of bimodal-sized InAs/GaAs QDs at temperature below 200 K, which demonstrated independent carrier transitions from the ground states of large and small dot ensembles [18]. In our previous work, bimodal sized InGaAs/GaAs QDs has been employed to make superluminescent diodes (SLDs) which exhibited a 3 dB bandwidth of 178.8 nm and power of 1.3 mW under CW conditions [19].

In this letter, we present high-power dual-λ FP cavity laser based on bimodal-sized InGaAs/GaAs QDs. The devices exhibit stable dual-λ operation in a large current range. A total power of 165.6 mW at 1015.2 nm and 1023.0 nm is obtained at current of 700mA under room-temperature CW mode. Through EL spectrum analysis, we attribute this dual-λ lasing to the independent carrier transitions from the LD ES1 and SD ES1, respectively. To our knowledge, this is the first report on high-power dual-λ lasing in bimodal-sized QDs. Our results demonstrate the possibility that stable high-power dual-λ lasing could be achieved based on carrier transition in different dot ensembles.

2. Experiment

The epitaxial structure of QD laser was grown by an AIXTRON 200-4 MOCVD system on an n-GaAs (001) substrate. The active region is composed of five repeated layers containing bimodal-sized In_0.5Ga_0.5As QDs, which are separated from each other by four 40-nm GaAs space layers. The QD layers were all deposited at 530 °C with the same amount of 5.26 monolayers (MLs) and high growth rate of 1.64 ML/s, and the V/III ratio is 45. The detailed epitaxial and device structure and the Atomic Force Microscope (AFM) picture of the QDs is described in Fig. 1. The chips were fabricated as tapered lasers that consist of a 6 µm × 750 µm ridge section and a tapered section with a full taper angle of 4° and a length of 750 µm which act as a power amplifier. Conventional lithography and dry etching was carried out before a 300 nm SiO₂ insulating layer was grown by Plasma-Enhanced Chemical Vapor Deposition (PECVD). Ti/Pt/Au and Ni/Ge/Au/Ni/Au ohmic contacts were evaporated on the top and back of the wafer, respectively. Different from our previously reported SLD with both facets anti-reflection coated to suppress lasing, both facets of the dual-λ laser are as cleaved and the chips were mounted p-side up on a copper sink for testing.

 

Fig. 1 (a) The epitaxial structure of the dual-λ QD LD. (b) AFM picture of uncapped QD sample. (c) The device structure of the dual-λ QD laser.

Download Full Size | PDF

The morphology and size distribution of the QD sample were evaluated by AFM. The EL and lasing spectrum were measured at room-temperature CW mode. The output of the device was coupled to a multimode fiber with an NA of 0.35 and then acquired by an optical spectrum analyzer. To avoid signal saturation, the lasing spectrum was measured after attenuated by an integrating sphere.

3. Result and discussion

The histogram of size distribution of the uncapped QD sample is presented in Fig. 2(a). Two groups of QDs with different size can be distinguished obviously. The “small dots (SDs)” have an average height of 3.5 nm, whereas the “large dots (LDs)” are 4.8 nm in height. The base area of QDs mainly distribute in the range from 400 to 2000 square nanometers. Besides the two groups of major QDs, there is a small amount of marginal dots with height below 3nm and above 6nm, respectively. The total density of the QDs is ~9.1x10⁹ cm⁻², which accord with the density interval in which the size distribution tends to be bimodal according to theoretical calculation [17].

 

Fig. 2 Bimodal-sized distribution and Electroluminescence of the QDs. (a) Height distribution of uncapped QD sample. The insert is the base area distribution. (b) and (c) EL spectrum with resolution of 3nm at 20 mA and 160 mA (white circles) under CW mode, with the fitted Gaussian components for emission from the large dot ensemble (blue lines) and small dot ensemble (red lines). The black line is the peak sum of Gaussian fitting. (d) Variation of the peak wavelengths obtained from Gaussian fitting as a function of injection current.

Download Full Size | PDF

Gaussian fitting analyses for the EL spectrum of the device with current increasing from 20 mA to 160 mA are carefully carried out. The representative curves at 20 mA and 160 mA are shown in Figs. 2(b) and 2(c) and the fitted peaks are summarized in Table 1. The variation of peak wavelengths obtained from Gaussian fitting as a function of injection current is depicted in Fig. 2(d). Although the accurate QD size needed to determine transition energy cannot be obtained only with AFM, the origin of the fitted peaks could be deduced according to the peak wavelengths, full width at half maximum (FWHM), wavelength shifting and relative intensity variation of fitting peaks. The output from the low energy peaks at around 1128 nm and 1081 nm dominate the spectrum at low filling condition, and are surpassed by high energy peaks at high filling condition. The intensity ratio of these two peaks keeps almost unchanged from 20 mA to 160 mA as described in the following Fig. 4(a). So they probably originate from the carrier transitions in the ground states of the two groups of quantum dots, which can be defined as ground state of lager dot (LD GS) and small dot (SD GS) ensemble, respectively. The FWHM of LD GS (69.7 nm) is large than SD GS (36.6 nm) at 20 mA, which indicate that the size fluctuate of LD ensemble is much larger than that of SD ensemble. As depicted in Fig. 2(d), due to the band-filling effect, the peak at 1048.9 nm shifts to 1025.4 nm with the current increasing from 20 mA to 160 mA. Based on the above analysis, we deduce that the peak at 1048.9nm shifting to 1025.4 nm at 160 mA originates from LD ES1. Similarly, the peak shifting from 1019.4 nm (at 20 mA) to 1013.3 nm (at 160 mA) should stem from SD ES1. The output at ~1000 nm keeps relatively weak all through the current range and probably stem from the ground state of some ultra small dots (USD GS).

Table 1. Fitted peaks of Gaussian components

View Table

The room-temperature output of the dual wavelength QD laser as a function of injection current under CW mode is shown in Fig. 3. Simultaneous dual-λ lasing is illustrated at current of 340 mA and 700 mA (Inset (a) and (b)). Two peaks with frequency difference of 2.8 THz and total power of 48 mW is achieved at 340 mA. The light power increases linearly with current at a slope of 0.3W/A. The maximum power of 165.6 mW with simultaneously dual wavelengths at 1015.2 nm and 1023 nm, corresponding to a frequency difference of 2.2 THz, with both the FWHM of ~1 nm is achieved at current of 700 mA. The intensity of the two lasing peaks is almost equivalent. These results, to our knowledge, are the best record of dual wavelength QD lasers based on FP cavity and are comparable to the output of high-power monolithic two-mode distributed feedback QW lasers which exhibited a maximum output power of 435 mW [20]. Our technical approach described in this paper addresses the need for compact size and low fabrication cost, which is significant for the development of future THz sources.

 

Fig. 3 Room-temperature CW output of the dual wavelength QD laser versus injection current. Inset (a) and (b) are the lasing spectrum of the device with resolution of 1nm at 340 mA and 700 mA, respectively.

Download Full Size | PDF

For preliminary explanation of the mechanism under the progressive dual-λ lasing, a complete spectrum characteristic of the device as a function of current is described in Fig. 4(a) and 4(b). We analyzed the EL spectrum from 20 mA to 160 mA by fitting Gaussians as depicted in Figs. 2(b) and 2(c). A total of five characteristic peaks were obtained by fitting each of the spectral curves at different current. The position of the peaks was marked on the spectrum in Fig. 4(a) using a red rhombus to show the peak shift. Previous analyzes have revealed that the peaks at 1048.9 nm and 1019.4 nm at 20 mA originate from LD ES1 and SD ES1, respectively. It can be obviously found in Fig. 4(a) that both the LD ES1 and the SD ES1 blueshift toward the lasing peaks at high currents. The lasing spectrum and the corresponding wavelengths shifting with the injection current are shown in Figs. 4(b) and 4(c), respectively. Dual-λ lasing starts to appear at about 280 mA and continues until 700 mA. A simple linear fitting to the two lasing wavelengths in Fig. 4(c) shows redshift of 0.015 nm/mA and 0.02 nm/mA due to the heating-effect and state-filling effect and wavelength difference between 7 nm and 8 nm in a relative large current range.

 

Fig. 4 Measured (a) electroluminescence with resolution of 3nm and (b) Lasing spectrum of the dual-λ QD laser with resolution of 1nm at different injection currents under CW mode. (c) Dependence of the lasing wavelengths on the injection current

Download Full Size | PDF

There are two primary theory models proposed to explain the dual-λ lasing in QD lasers in previous research. One attributed the dual-λ lasing to the simultaneous carrier transitions in ground and excited states from single QD subset where the normal wavelength difference of the two lasing peaks was about tens of nanometers [8,9], which is quite different from our results. The other explanation is based on the model of Rabi oscillation [10,21], in which the single emission peak from an exciton trapped by a QD may split into double peaks when the exciton state interacts with a strong resonant electromagnetic field. The key feature of Rabi oscillation is that the frequency difference of the two peaks linearly depends on the square root of the power density of the electromagnetic field. However, as shown in Fig. 4(c), the frequency difference of our device remains almost unchanged with the increasing of light power, which qualitatively violates the model of Rabi oscillation. According to above analyses and the tendency of peak shifting, we phenomenologically attribute the dual-λ lasing of our devices to independent emissions from LD ES1 and SD ES1 as marked by the blue arrows in Fig. 4(a). This is a new concept for dual-λ lasing from FP laser in spite the necessary and sufficient realization condition is still unclear. One obvious advantage of this formation is that the process of carrier transition in high energy states provides the possibility of high-power dual-λ operation as the high energy states has higher degeneracy than low energy states and can accommodate more carriers [22].

Finally, we focus on the disordered region in Fig. 4(c) and the temperature dependence of lasing spectrum. In current range from 380 mA to 480 mA, an unexpected peak around the long lasing wavelength is excited and disturbs the whole lasing spectrum. This may stem from the small amount of the marginal sized dots, which are populated independently and reach saturation soon at relatively high carrier concentration. At current range exceed 720 mA, the long lasing wavelength exhibits a decline and even complete quenching light power. This may be result from the onset and aggravation of mode competition between the two lasing wavelengths, due to the self-heating and homogeneous broadening at high carrier concentration. Temperature dependence of the lasing spectrum at 300 mA and 700 mA under CW mode is depicted in Fig. 5. The temperature was changed by a Thermo Electric Cooler (TEC). Dual-λ lasing could maintain at least from 15°C to 25°C under 300mA and 700mA. Under higher temperature, the spectrum degeneration probably results from that the elevated temperature accelerates the carriers transfer between dots and leads to mode competition between the two lasing wavelengths. Thermal effects also induce a redshift of about 0.34nm/°C for both of the lasing peaks. This spectrum degeneration could be avoided by improving the bimodal-sized distribution and increasing the size distinction of QD ensembles.

 

Fig. 5 Temperature dependence of lasing spectrum at (a) 300 mA and (b) 700 mA with resolution of 1 nm under CW mode.

Download Full Size | PDF

4. Conclusion

In summary, we demonstrated a dual-λ FP laser utilizing bimodal-sized QDs as active material. The device exhibits simultaneous dual-λ lasing at 1015.2 nm and 1023 nm with total power of 165.6 mW at a CW current bias of 700 mA. The dual-λ lasing is attributed to independent carrier transitions from LD ES1 and SD ES1 according to the Gaussian fitting analysis for EL spectrum. This is a new concept of dual-λ lasing in QD laser and opens a new possibility to achieve high-power dual-λ operation only utilizing FP cavity, which is significant for dual-λ source with compact size and low fabrication cost.