Ultrafast Gain Recovery in Quantum Dot based Semiconductor Optical Amplifiers

摘要

Semiconductor optical amplifiers (SOAs) are key to GHz operation in the new optical telecommunication networks. Quantum dot (QD) based SOAs are outperforming their bulk and quantum well based competitors. The limiting factor in ultrahigh bit rate amplification is the ultrafast population recovery in the resonant level, which is mainly limited by carrier capture and relaxation processes in the QD. We use pump-probe measurements resonant to the QDs confined states energies (ground and excited state) to investigate the response to a four fs-pulse train of 1 THz repetition rate. A deep insight about the capture process implied is then obtained, and direct capture from the wetting layer is identified as the dominant mechanism in the high current regime. The device active medium is formed by 15 layers of MBE-grown InGaAs QD-in-a-well nanostructures and lays in the center of an AlGaAs waveguide, contacts are made as in the sketch of fig.1. Seeking a faster overall performance, p-modulation doping is used in the QDs layers. The deeply etched ridge waveguide (with a width of 2 μm and a length of 1 mm, as show in Fig 1) allows us to work at higher effective currents than normal in this kind of devices, filling the complete ensemble of QDs due to a better current distribution. Lasing is suppressed using a broadband antireflection coating in both ends of the amplifier, allowing single pass transmission experiments. The gain dynamics in the amplifier are measured by femtosecond pump-probe technique with heterodyne detection. Using two Michelson interferometers, up to four 150 fs long Fourier-limited pump-pulses are generated with a controllable delay in a picosecond time range. The gain dynamics is measured at room temperature as a function of injection currents and excitation power, either resonant to the ground state (GS) at 1.3 μm or resonant to the excited state (ES) at 1.2 μm. Fig 2. shows the gain evolution after one, two and four pulses trains when incident beam power is enough to momentary deplete the ground state population (complete depletion of the gain can be observed (Gain below 1)).