Quantum dashes promise higher speeds for tomorrow's networks

Lasers based on self-assembled nanostructures are attracting considerable attention owing to a range of favorable characteristics, such as low threshold current, high gain and good thermal stability. These advantages over quantum-well (QW) and bulk lasers have even led to the commercial launch by Innolume, formerly NL Semiconductor, of InAs-on-GaAs quantum-dot lasers emitting at 1.1–1.3 μm.

While Innolume s technology is based on quantum dots, there is also a growing interest in "quantum dashes", which are essentially elongated dots. These structures share many of the advantages of a dot, but offer an even higher gain that can improve laser efficiency. Like quantum-dots lasers, they can be mode-locked to produce short, high-bit-rate pulses with very low jitter. These characteristics make the dash-based lasers promising candidates for providing high-repetition-rate sources at bit-rates of 40 and 160 Gbit/s, the transmission speeds expected in tomorrow s networks. These properties also lead to an intrinsic "beating" emission at microwave frequencies. Thus quantum-dash lasers are also potential sources of microwaves in the 10–300 GHz range with very high spectral purity.

Quantum-dot and quantum-dash lasers based on the InAs/InP material system are of particular interest because they can be used for 1.5 μm transmission in the fiber-optic spectral window. However, fabricating dot-based lasers that emit at this wavelength is not easy. MOCVD is currently unproven for producing InAs quantum dots with high-quality cladding layers, while MBE on InP substrates with standard orientations tends to produce dashes rather than dots. This has led researchers either to investigate dash-based structures on (100) InP or produce dots on the (311) B surface.

At the Alcatel-Thales III-V Lab in Palaiseau, France, we are using MBE to produce dashes on (100) InP. This has the significant advantage of being compatible with existing standard fabrication processes for QW and bulk devices, including regrowth processes. This has enabled us to produce CW room-temperature lasers with quantum dashes in InGaAsP QWs and barriers. A group from Korea s Electronics and Telecommunication Research Institute and a research team at the University of Würzburg, Germany, have had similar success with InGaAlAs barriers.

Our lasers are based on two designs that feature either dashes-in-a-barrier or dashes-in-a-well, and are produced by gas-source MBE on sulfur-doped substrates using the Stransky–Krastanow growth mode. Strain relaxation drives a 1 nm-thick InAs layer that has a 4% lattice mismatch with the underlying InGaAsP to form quantum-sized structures. In our case, this "self-organization" is highly sensitive to the surface anisotropy of InGaAsP, and dashes are formed along the [1–10] direction with a surface density of 1 − 4 × 1010 cm–2. They are 15–20 nm wide and 40–300 nm long, depending on the growth conditions (see figure 1). The dimensions of the dashes influence the laser s carrier-confinement properties and, ultimately, device performance.

One downside of making lasers from quantum-sized structures is their relatively small interaction with the optical modes of the device. Typically, just 0.15% of the power in the optical modes of the laser is actually confined in the quantum-dash layer – six times less than that for a QW layer. Consequently, to improve the modal gain, layers of quantum dashes are stacked close together between spacer layers. Unfortunately, this affects the dash density and dimensions, reduces modal gain at the lasing wavelength and hampers device performance (see figure 2). This shift is accentuated when the spacer layers are thinner than 60 nm.

We have overcome this problem by tweaking the growth conditions for each dash layer to compensate for this wavelength shift. This has enabled us to stack up to 12 layers. The reproducibility of these layers can be judged by examining the full-width half-maximum (FWHM) of the photoluminescence emitted by this structure (see figure 2). This shows that uniformity between the quantum-dash layers is good enough for structures up to nine layers thick.

With this approach we have grown dashes-in-a-barrier and dashes-in-a-well structures targeting 1.5–1.6 μm emission (see figure 3). For initial analysis, we processed broad-area (BA) lasers that emit over a wide spectral range of 1250–1650 nm. Gain saturation, a problem that has plagued the InAs/GaAs system, is absent from these spectra. Threshold-current measurements reveal that the dashes-in-a-well structure provides better carrier injection – threshold-current density was 110 A/cm2 per layer for this design, compared with 190 A/cm2 per layer for the dashes-in-a-barrier laser. Internal quantum efficiency for a six-layer quantum-well-based design is 80%.

We have assessed temperature performance by the conventional approach, determining the value of the "characteristic temperature" T0, which is a well-known figure of merit. This figure is influenced by threshold-current stability and carrier confinement, and a higher value indicates that the laser can operate at higher temperatures. For a dashes-in-a-well system, T0 can be increased by narrowing the width of the well, an approach that raises the energy of the well, and suppresses the electron flow from the dashes.

With this scheme, we have built devices with a T0 of 100K for operating temperatures of 20–80 °C. This approach has also produced single-transverse-mode Fabry–Pérot lasers with a threshold current of just 12 mA at 25 °C (see figure 4) and a T0 of 80 K between 25 and 85 °C. This is a significant improvement in temperature performance over standard 1.55 μm aluminum-free multiple-quantum-well (multiple-QW) lasers.

We have developed these Fabry–Pérot lasers with buried-ridge-stripe (BRS) lasers. These were fabricated using standard processing steps established for InGaAsP multiple-QW lasers: plasma etching to form 1–2 μm-wide ridge waveguides; MOCVD regrowth to add the p-doped InP cladding layer and GaInAs contact layer; and proton implantation to define the current flow through the device. A distributed feedback (DFB) structure defined by electron-beam lithography was added for singlemode emission.

The devices that we have made include a 205 μm-long BRS laser that has a high-reflection-coated rear facet and DFB structure with six dashes-in-a-well layers. This device produces CW operation between 15 and 85 °C (see figure 5), and has a dominant lasing mode at 1512 nm with a side-mode suppression ratio (SMSR) of 45.5 dB. It is the first quantum-dash or quantum-dot laser to produce a floor-free bit-error-rate measurement when directly modulated for 10 Gbit/s transmission (see figure 6). The low error rates, which can be less than one part in 1011 for transmission down a 16 km length of fiber at a received power of –6 dBm, show the promise of quantum-dash lasers for telecom applications.

These lasers also have the potential to be used in microwave applications. Eighteen months ago we reported the demonstrated low-jitter, high-bit-rate short pulses using mode-locked 1.5 μm quantum-dot lasers and all-optical clock recovery. The intrinsic beating spectrum of a Fabry–Pérot quantum-dot laser has a typical FWHM of 15 kHz, compared with 1.5 MHz for a bulk DBR laser (see figure 7). Our quantum-dash lasers offer significant improvements over their bulk and QW-based predecessors, and we expect that they could deliver a superior performance at microwave frequencies.

We will direct future efforts at refining our control over the shape of these nanostructures. This will optimize the quantum-dash heterostructure design and boost the dynamic properties of the laser. We will also continue to develop quantum-dash-based mode-locked lasers for fiber-optic communications and millimeter-wave generation, and explore their excellent phase-noise and time-jitter characteristics.