Radio-frequency analysis of self-mode-locked quantum dot laser

Radio-frequency analysis of self-mode-locked quantum dot laser

Available online at www.sciencedirect.com ScienceDirect Materials Today: Proceedings 7 (2019) 908–911 www.materialstoday.com/proceedings NanoFIS 20...

730KB Sizes 0 Downloads 11 Views

Available online at www.sciencedirect.com

ScienceDirect Materials Today: Proceedings 7 (2019) 908–911

www.materialstoday.com/proceedings

NanoFIS 2017

Radio-frequency analysis of self-mode-locked quantum dot laser Christoph Webera,*, Paolo Bardellab, Lorenzo Columbob, Mariangela Gioanninib, and Stefan Breuera b

a Institute of Applied Physics, Technische Universität Darmstadt, Schlossgartenstraße 7, 64289 Darmstadt, Germany. Dipartimento di Elettronica e Telecomunicazioni, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy.

Abstract We numerically study and confirm experimentally a transition from unlocked multi-modal emission to self-mode-locked emission of a single-section InAs/InGaAs quantum dot laser in dependence on the laser injection current. At this transition, the radio-frequency beat note linewidth as well as the modal power fluctuations quantified by an integrated relative intensity noise of all modes are found to be significantly reduced. A linewidth reduction from several hundreds of MHz to below 1 MHz is observed, with the lowest value observed being as small as 20 kHz. Experiment and simulation results are in good qualitative and quantitative agreement. © 2019 Elsevier Ltd. All rights reserved. Selection and/or peer-review under responsibility of NanoFIS 2017 - Functional Integrated nano Systems. Keywords: Self mode-locking; semiconductor laser; quantum-dot laser; relative intensity noise; radio-frequency linewidth

1. Introduction Semiconductor diode lasers (SCLs) generating optical frequency combs (OFCs) consisting of phase locked, frequency equidistant lasing modes with nearly equally distributed spectral power and low phase noise are compact sources for high-data rate optical interconnections based on silicon photonics optical modulators [1], sub-THz signal generation by comb line mixing on fast photo detectors [2,3] or dual-comb spectroscopy applications with high speed and good precision [4]. Self mode-locking (SML) in single-section SCLs with active regions based on quantum-well [5], quantum-dash [6] or quantum-dot (QD) [7] material or in quantum-cascade lasers [4] without any saturable absorber has been proven to be one way to produce OFCs where phase locking of the individual optical

* Corresponding author. Tel.: +49 6151 16 20160. E-mail address: [email protected] 2214-7853 © 2019 Elsevier Ltd. All rights reserved. Selection and/or peer-review under responsibility of NanoFIS 2017 - Functional Integrated nano Systems.

Weber et al. / Materials Today: Proceedings 7 (2019) 908–911

909

modes occurs by internal physical nonlinear processes in the semiconductor active material. Short pulses are observed either directly at the laser output facet [7] or after dispersion compensation [8]. By modelling a singlesection SML QD SCL using a time-domain traveling-wave (TDTW) model the underlying physical progress has been investigated and a transition from unlocked multi-mode emission to SML has been proposed [9]. In this work, we study experimentally and theoretically, applying the model in [9], the radio-frequency (RF) beat note linewidth and the relative intensity noise (RIN) dependencies on laser injection current. In particular, we find a correlation between SML and narrowing of the RF beat note as well as a reduction of the integrated RIN. 2. Laser and experimental details In our simulations we consider a 250 µm long single-section QD laser with a corresponding mode difference frequency of 178 GHz. The laser is a ridge waveguide laser with a ridge width of 5 µm and the number of QD layers in the active region amounts to 15. The reflectivity of both facets amounts to 60 %. For insight into the TDTW model and further simulation parameters see Bardella et al. [9]. The laser under experimental investigation is a 1 mm long single-section ridge waveguide device with an active region consisting of 10 layers of InAs/InGaAs QDs. The ridge width amounts to 4 µm and the facets are as-cleaved. The laser is mounted on a copper cooling block, whose temperature is held constant at 10° C. The length of the laser corresponds to an optical mode spacing of approximately 40 GHz. A gain current up to 400 mA is injected into the laser. By analyzing the emitted light using a fast photo diode and an electrical spectrum analyzer the RF beat note frequency corresponding to the frequency difference between the optical modes and the RF beat note linewidth by applying a Lorentzian fit to the measured RF beat note is investigated. With an InGaAs photo-detector in combination with an electrical spectrum analyzer the integrated RIN from 40 MHz to 50 MHz as a measurement for amplitude fluctuations of the optical power is accessed as described in [10] and [11]. 3. Results First, the RF beat note spectra are measured for increasing gain injection current. Despite the offset in the beat note frequency, stemming from the different device lengths, simulation and experiment exhibit similar RF beat note spectra. With increasing gain current, the RF beat note frequency increases. Fig.1 and Fig. 2 depict the resulting RF

Fig. 1. Simulated RF beat note linewidth as a function of gain current. A transition from unlocked optical modes (broad RF beat note) to phase-locked optical modes (small RF beat note) is observed.

Fig. 2. Experimentally estimated RF beat note linewidth as a function of gain current. As in the simulation, a transition to phase-locked optical modes is evident. The inset depicts the linewidth in the phaselocked case in a logarithmic scale. The lowest measured linewidth amounts to 20 kHz.

910

Weber et al. / Materials Today: Proceedings 7 (2019) 908–911

beat note linewidths from simulations and experiment, respectively, as a function of the gain injection current. In both Figures, two regimes can be clearly differentiated depending on the gain current. For low currents the RF beat note is very broad whereas at a sharp transition the RF beat note is narrowed significantly, from two-digit MHz to below 100 kHz in the simulations. In the experiment the lowest measured linewidth amounts to 20 kHz at a gain current of 230 mA. The simulations suggest that in the first regime, at low gain currents, the emitted light consists of several longitudinal modes without a fixed phase relation and thus varying frequency difference between the modes. At the threshold to the second regime, these multimode emission changes to an OFC generated by SML. The phasedifference between adjacent modes is fixed as is the frequency difference. Therefore, the beat note linewidth decreases significantly. In Fig.3 and Fig.4 the simulated and the experimentally estimated integrated RIN are depicted as a function of the gain injection current. The simulated RIN is integrated from 10 MHz to 20 GHz. In the experiment the RIN can be accessed up to 75 MHz and has been measured between 40 MHz and 50 MHz where the spectral noise density is observed to be flat. In the simulation a significant RIN decrease at the threshold to the second regime, obtained in the linewidth measurements, is evident. A RIN decrease for currents above this threshold is also confirmed by experiment. 4. Conclusion We investigated experimentally and by simulations the onset of SML and the generation of an OFC in a singlesection QD SCL. Both simulation and experiments indicated a current threshold where the optical modes appear to phase-locked and the laser emission changes from multi-mode to SML. This threshold is characterized by a reduction of the RF beat note linewidth, with the lowest experimentally measured linewidth of 20 kHz, and a reduction of the integrated RIN. Our experiments are in good quantitative and qualitative agreement. Acknowledgements The authors thank I. Krestnikov and his team from Innolume GmbH, Dortmund, Germany, for growing the excellent QD wafer.

Fig. 3. Simulated integrated RIN (10 MHz to 20 GHz) as a function of gain current. A and B denote the locking regimes observed in the RF beat note linewidth analysis.

Fig. 4. Experimentally measured integrated RIN (40 MHz to 50 MHz) as a function of gain current. A and B denote the locking regimes observed in the RF beat note linewidth analysis.

Weber et al. / Materials Today: Proceedings 7 (2019) 908–911

911

References [1] P.J. Delfyett, S. Gee, M.-T. Choi, H. Izadpanah, W. Lee, S. Ozharar, F. Quinlan, and T. Yilmaz, IEEE. J. Lightwave Technol. 24 (2006) 2701. [2] S. Koenig, D. Lopez-Diaz, J. Antes, F. Boes, R. Henneberger, A. Leuther, A. Tessmann, R. Schmogrow, D. Hillerkuss, R. Palmer, T. Zwick, C. Koos, W. Freude, O. Ambacher, J. Leuthold, and I. Kallfass, Nat. Photonics 7 (2013) 977-981. [3] S. Latkowski, F. Surre, and P. Landais, Appl. Phys. Lett. 92 (2008) 081109. [4] J. Faist, G. Villares, G. Scalari, M. Rösch, C. Bonzon, A. Hugi, and M. Beck, Nanophotonics 5 (2016) 272-291. [5] K. Sato, Electron. Letts. 37 (2001) 763-764. [6] V. Panapakkam, A. P. Anthur, V. Vujicic, R. Zhou, Q. Gaimard, K. Merghem, G. Aubin, F. Lelarge, E. A. Viktorov, L. P. Barry, and A. Ramdane, IEEE J. Quantum Electron. 52 (2016) 1-7. [7] J. Liu, Z. Lu, S. Raymond, P. J. Poole, P. J. Barrios, and D. Poitras, Opt. Lett. 33 (2008), 1702-1704. [8] K. Merghem, C. Calò, R. Rosales, X. Lafosse, G. Aubin, A. Martinez, F. Lelarge, and A. Ramdane, IEEE J. Quantum Electron. 50 (2014) 275-280. [9] P. Bardella, L. L. Columbo, and M. Gioannini, Opt. Express. 25 (2017) 26234-26252. [11] T. Gensty, W. Elsäßer, and C. Mann, Opt. Express 13 (2005) 2032-2039. [10] R. Pawlus, S. Breuer, and M. Virte, Opt. Lett. 42 (2017) 4259-4262.