Abstract

Quantum Dot lasers (QD) have been in the spotlight of attention for the last decades due to their unique ability to allow simultaneous lasing form different sub-bands. Especially in the case passive mode locking techniques are employed, the multi-energy emission capabilities of QD lasers allowed the exploration of multi-band modelocking [1], and optical feedback/injection stability [2, 3]. Typical, higher state excitation in QD lasers has been achieved by varying the laser’s bias conditions and thus modify the carrier populations in each band. In this paper we investigate an all optical switching scheme that is based only on the unidirectional injection of two QD Fabry-Perot passively mode locked lasers. By varying the injection strength, effects like clock recovery and wavelength switching were observed even in the case that the lasers exhibited significant optical and electrical detuning. The master (ML) and slave laser (SL) utilized were 2mm long, multi-section InAs/InGaAs QD laser grown on a GaAs (100) substrate. The lasers have 5 layers of QDs, waveguide ridge width of 6μm, whereas the facets exhibit a reflectivity of 95%–15% at 1280nm. In order to achieve passive mode locking, 15% of the total length was reverse biased. The experimental setup utilized is presented in Fig.1a. The optical spectrum analyzer had a resolution of 0.1nm, the RF analyzer an optical bandwidth of 26GHz, whereas the injection strength was regulated through a variable optical attenuator and a polarization controller. The SL was biased at 180mA (I_th=130mA) and V_abs=−4V and only ground state (GS) emission could be recorded with repetition frequency R_f=20.47GHz, whereas minor Q-switching side-peaks (Δf≈600MHz) were observed due to the proximity to excited state (ES) threshold. The bias of the ML was varied in order to achieve different operational regimes like single GS/ES and dual state lasing. Two cases are presented. The first consists of injecting GS emission from the ML (Fig.1b-i), the two emission peaks have an optical detuning of 4.4nm, while a partial overlap is present due to the inhomogeneous broadening. In Fig1b-ii the RF peaks exhibit GS mode locking from both lasers with a difference of ΔR_f=324MHz, due to the different bias conditions, while the optical power of the ML before SL injection was set to 1.16mW. By reducing the injection strength (Fig.1b iii-iv) the SL is mostly unaffected apart from a minor peak that corresponds to ML-GS pulses. On the other hand by increasing injection strength the GS emission at the SL is saturated and ES emission is achieved (ΔP_ES≈40dB), whereas a clear RF peak that corresponds to SL ES mode locking is evident (Fig1b v-vi). Moreover, a secondary peak can be observed that coincides with the ML-GS R_f and implies dual state locking is achieved following the dynamics (R_f) of the ML (Fig.1b-vi). In the second case, ML is set to I=380mA, and only ES lasing is evident, alongside a clear ML-ES RF peak and Q-switching related sidebands, the ML power was set to 0.06mW (Fig.1c i-ii). In the case that injection strength is kept minimum (Fig.1c iii-iv) only a slight increase in the RF noise is observed, while state switching did not occur. On the other hand, the increase of injection strength induces a reduction by 32dB for the SL-GS and SL-ES is enhanced by 40dB. In the RF domain GS mode locking is switched off and clear ES mode locking is evident with minor Q-switching instabilities (Fig.1c v-vi).

© 2015 IEEE