We report on subpicosecond pulse generation using passively mode locked lasers (MLL) based on a low optical confinement single InGaAsP/InP quantum well active layer grown in one epitaxial step. Systematic investigation of the performances of two-section MLLs emitting at 1.54 µm evidenced pulse width of 860 fs at 21.31 GHz repetition rate, peak power of ~500 mW and a time-bandwith product of 0.57. A 30 kHz linewidth of the photodetected radio-frequency electrical spectrum is further demonstrated at 21 GHz which is, to our knowledge, the lowest value ever reported for a quantum well device.
©2008 Optical Society of America
Monolitic semiconductor mode-locked lasers (MLLs) are very attractive components for a wide range of applications. In optical telecommunication systems, they may cover such diverse applications as very high speed optical time division multiplexing (OTDM) sources (40 - 160 GHz), all-optical signal processing (high bit rate clock recovery, A/D conversion), 100 Gigabit Ethernet or low noise sampling for performance monitoring of optical networks. MLLs are also rising considerable interest in microwave photonics, with emphasis on radio-over-fibre for access networks, optical sampling for instrumentation or radar applications . They are finally particularly suited for optical interconnects in microprocessors, as pulsed sources and for clock distribution . In recent years, many investigations have been carried out on monolithic quantum-well (QW) based MLLs with the demonstration of very good performances in passive or hybrid schemes [3-6]. These investigations showed in particular that the pulse duration decreases with the number of wells in the active layer [3, 5] owing to the reduction of differential gain which results in a higher gain saturation energy, essential for efficient mode locking.
Recent work on low dimensional quantum dot and quantum dash material has shown that the linewidth of the radio frequency (RF) spectrum was found to decrease with the optical confinement factor  implying reduced coupled amplified spontaneous emission (ASE) to the optical modes. The narrow RF linewidth is a characteristic property indicating low timing-jitter in mode-locked lasers [8-9]. Some applications require ultra-short pulse generation, below 1 ps, combined with subpicosecond timing jitter, which has only recently been achieved using InAs/GaAs quantum dot MLLs .
In this work, we combine these important characteristics for MLLs by the use of an optimized QW layer structure allowing low propagation losses, low modal and differential gains (comparable to those of QD structures) to achieve high saturation energy. We report for the first time to our knowledge subpicosecond pulse generation without any external pulse compression scheme and the lowest ever measured RF spectrum linewidth  in a 2-section single QW MLL grown in a single epitaxial step. Basic optical gain measurements and dynamic performances are first reported for one-section Fabry-Perot devices. Pulsewidth and noise assessment is finally reported for 2-section MLL components.
2. Device fabrication and main characteristics
The waveguide structure is modelled and optimized for low optical propagation losses, low differential gain and a small optical confinement factor Γ of ~1%.
The quantum well structure was grown by gas source molecular beam epitaxy on a (100) InP substrate. It consists of a single InGaAsP quantum well enclosed within two 80-nm-thick separate confinement heterostructure (SCH) layers. The SCH layers are undoped lattice-matched quaternary Ga0.2In0.8As0.4P0.6 layers (λg=1.17 µm). Graded p-doping of the p-InP confinement layer is performed.
50 µm-wide broad area lasers were first processed from this structure. A modal gain of 11.6 cm-1, very low internal losses of ~7 cm-1 and internal differential quantum efficiency of ~20 % were extracted. Shallow ridge single mode waveguide lasers were then fabricated by a combination of dry and wet etching. After beam propagation modelling, a stripe width of ~3 µm is adopted for single transverse mode operation (Fig. 1). Electrode pads were formed by means of bisbenzocyclobutene (BCB), yielding low parasitic capacitance compatible with high frequency experiments.
Fabry-Pérot (FP) lasers and two-section devices consisting of a long gain and a short saturable absorber sections with as cleaved facets were investigated.
We present first basic laser characteristics measured on single section Fabry Perot devices. A 1 mm-long sample was hence used to evaluate the net modal gain and linewidth enhancement factor (LEF). The device was mounted on a temperature stabilised holder. The threshold current was 14 mA at 20 °C and the slope efficiency was 0.15 W/A per facet. The differential gain (dg/dN) which is an important parameter in the mode locking process as it is related to gain saturation , was evaluated by the Hakki Paoli (HP) method  under threshold current. Fig. 2 shows the gain measurements performed below threshold current.
The material gain as a function of carrier density is approximated using a linear expression:
where dg/dN is the differential gain, N is the carrier density and N0 is the transparency carrier density. In order to extract the differential gain value, a relation between the injected current and the carrier density is required. We use the following formula:
where τ is the carrier lifetime, B is the bimolecular recombination coefficient, C is the Auger recombination coefficient, I is the injected current, q is the charge of the electron, S is the cross-sectional surface of the active layer and L is the length of the laser cavity. All the values of the parameters used for these calculations are listed in Table 1.
|τ||Carrier lifetime||~2×10-9 s|
|B||Bimolecular recombination coefficient||1×10-10 cm3.s-1 |
|C||Auger recombination coefficient||7.5×10-29 cm6.s-1 |
|S||Surface of the active region||1.7×10-10 cm2|
|L||Length of the cavity||1144×10-6 m|
The recombination coefficients B and C are typical values found in the literature, yielding an estimate of the order of magnitude of the differential gain (dg/dN) rather than an exact value. We estimate the differential gain at the lasing wavelength at ~1.3×10-17 cm2 by using the gain curves in Fig. 2. Typical differential gain of InGaAsP multiple quantum well (MQW) lasers amounts to about 10-16 cm2 , i.e. one order of magnitude higher than that of our single QW active medium. The differential gain in a MLL configuration might however be higher as carrier density is not pinned above threshold.
The linewidth enhancement factor (LEF) or Henry factor αH may have a non-negligible effect on the value of pulsewidth . The LEF quantifies the amount of phase-amplitude coupling generated in the device. In pulsed regime, it contributes to self-phase modulation (SPM) leading to inefficient spectral broadening . The LEF has been assessed by ASE measurements below threshold current and by the FM/AM method above threshold using the same device. The LEF is defined as follows :
where n is the refractive index, g is the gain per unit length, λ is the wavelength and N is the carrier density. As our structure has been optimized for small differential gain, we expect higher LEF values than those of standard QW lasers . This is indeed the case as shown in Fig. 3 where the LEF reaches a value of ~8.3 for an injection current of 100 mA. This relatively strong change with current might be related to incomplete carrier recombination in the quantum well leading to an increase of refractive index through a plasma effect. The role of the LEF in pulse broadening will be investigated in the next section.
3. Pulse characterization
Pulsewidth and repetition rate are characterized by means of a background-free autocorrelator based on second harmonic generation.
The laser used is a two-section device with a total length of 2130 µm with as cleaved facets. The corresponding repetition frequency is 21.31 GHz. Figure 4 shows the light-current (L-I) characteristics as a function of reverse voltage (VSA) and bias current (I) for a 145 µm long saturable absorber section (~7 % of total length) and a 1985 µm-long gain section. The threshold current with unbiased absorber section is 23 mA. The modulation of losses by the reverse bias is more important in the range between -5 V to -4 V as illustrated by the presence of pronounced hysteresis (Fig. 4 (b)).
Stable mode-locking is observed for injection currents ranging from 60 mA to 140 mA and reverse voltages from -4.5 V to -3.0 V. The use of longer absorber sections (up to 15 % of total length) resulted in increased absorption and much reduced voltage range for efficient mode locking. We investigated the pulse duration and time bandwith product (TBP) for different values of current in gain section and reverse bias applied to the absorber section.
We first focus on the deconvolution of autocorrelation traces. For this purpose, we compare gaussian and lorentzian shapes for the fit. From calculated standard deviations, the best fit is obtained for a lorentzian lineshape (Fig. 5).
The evolution of lorentzian FWHM pulsewidth as a function of both reverse bias and current is illustrated in Fig. 6.
Pulsewidths range from 0.860 to 6.6 ps. The shortest pulses are obtained at a large reverse bias, less than -4.1 V, when the absorber recovery time is the shortest, and for injection currents below 100 mA, where SPM impact is the smallest. We note typical trends, i.e., pulses broaden with increasing injection current and shorten with increasing reverse bias.
Below a current of 70 mA, the laser wavelength does not allow pulse characterization. Above a current of 160 mA, and out of voltage range from −2.8 V, incomplete mode-locking occurs: the pulses become broader and we observe an increase of continuous wave (CW) background. Previous work showed 1.2 ps pulse generation at 40 GHz using bandgap shifted growth of the saturable absorber section  and 2.8 ps pulse generation at 40 GHz using a single epitaxial growth step . In the latter, the investigation shows that pulse duration decreases with the number of wells in the active layer owing to the reduction of differential gain and, thus, a higher gain saturation energy essential for efficient mode locking.
In classical mode-locking theory, the pulsewidth is proportional to the ratio between the saturation energies in the gain and absorber sections. The saturation properties of a semiconductor gain medium is often quantified by the pulse saturation energy above which the gain is significantly reduced. The saturation energy is defined as
where hv is the photon energy, A is the mode area and dg/dN is the differential gain.
The optimization of the 1-QW active medium with a reduced optical confinement factor and low differential gain (dg/dN ~1.3×10-17 cm2), should yield high saturation energy of the gain section  implying very good pulsewidth performances. Moreover, pulsewidth reduction is amplified by the reduced carrier lifetime in the absorber section for high reverse bias as shown in Fig. 6.
This optimized laser structure allows to achieve the shortest pulsewidth value for the bias conditions shown in Fig. 7. This figure illustrates the autocorrelation trace for a current of 74 mA and a reverse bias of −3.8 V. After deconvolution of the lorentzian line-shape, we obtain a subpicosecond pulse of ~860 fs duration at the output of the laser, without any pulse compression scheme. As the full width at half maximum (FWHM) of the optical spectrum is 5 nm (Fig. 7(b)), the TBP is 0.57, indicating that the pulses are not transform-limited. This implies chirped pulses as the TBP limit for a lorentzian shape amounts to ~0.22. The extinction ratio of the pulse correlation trace is estimated to be higher than 25 dB. An average power of 10 mW and a pulse peak power of ~500 mW are achieved at the laser facet.
Time-bandwith products (TBPs) for different values of reverse bias VSA applied to the absorber section and LEF are shown in Fig. 8.
Simulations  showed that for QW MLLs the time-bandwidth product scales linearly with the LEF through SPM, provided this parameter remains the same for gain and absorber sections. Figure 8(a) shows the evolution of the measured TBP as a function of the LEF for a fixed VSA. One can notice that for the lowest LEF values, the TBP remains constant at around 0.6. The amount of chirp in the pulse is therefore not SPM driven, but dispersion driven. For higher values of LEF (>8.5), the amount of SPM generated in the laser cavity becomes large enough to generate an aditionnal chirp. This correlation between the TBP and LEF values shows this latter parameter plays a certain role, although other effects may contribute to pulse broadening . Figure 8(b) depicts the effect of applied bias on the TBP for a injection current of 120 mA. As expected, the TBP decreases with high reverse bias in absorber section owing to absorber recovery-time shortening.
4. Relative Intensity Noise (RIN) and electrical spectrum
Both RIN and electrical spectrum have been measured on a passively mode-locked laser. The RIN was measured in the 0 - 20 GHz range (Fig. 9), using a battery-powered low-noise driving source. A low RIN level of ~-153 dB/Hz was measured in the 6 - 20 GHz range. The broad peak at the frequency of ~500 MHz corresponds to the relaxation frequency of the 1 mm-long laser under test. This low RIN value is due to the very low differential gain of the single quantum well structure. This measurement shows that reducing the relaxation frequency with a low differential gain could be a way to reduce the RIN of lasers .
The electrical spectrum measurement of the RF tone yields information on an idea of the frequency stability of the generated pulses. Measurement is made with a 50 GHz-bandwidth photodiode coupled to a 50 GHz-bandwidth electrical spectrum analyzer with a 10 kHz resolution bandwidth. Depending on bias conditions (current, absorber voltage), the -3dB RF linewidth decreases from 150 MHz to a lowest value of 30 kHz as illustrated in Fig. 10 for the 21.31 GHz pulse train. This extremely small RF linewidth is attributed to the small coupled amplified spontaneous emission noise owing to the low optical confinement factor. These results are similar to those obtained on quantum dash based self pulsating lasers at 1.55 µm [8, 9], emphasing the importance of optical confinement for achieving low phase noise.
This result highlights the potential of ultra-low “high frequency” timing jitter in active or hybrid scheme. As already demonstrated, in hybrid or active schemes, direct electrical modulation of saturable absorber or gain section may be used to reduce drastically the low-frequency timing jitter .
It is important to note that it is not possible to achieve both a very short pulsewidth and a very low RF spectrum linewidth with this device. The bias condition for a very low RF linewidth corresponds to incomplete mode locking with high CW background and, in an inverse manner, the shortest pulsewidth conditions correspond to a large RF spectrum linewidth which amounts to 150 MHz. However, we can find a trade-off between the optimisation of pulsewidth and RF linewidth.
Further, investigations are in progress to fully understand the mode-locking behaviour, in particular by the mode-phase characterization and noise measurements to extract the timing jitter.
In this work, single InGaAsP/InP quantum-well based passive mode-locked lasers emitting at 1.54 µm have been investigated. An optimized low differential gain structure enables a high saturation energy of the gain section and hence allows a subpicosecond pulse-train emission at 21.31 GHz repetition frequency without any pulse compression scheme. Pulse measurements show an extinction ratio in excess of 25 dB and pulse peak power of ~500 mW. The evaluation of TBP indicates non-transform-limited pulses that could further be compressed down to less than 500 fs provided the chirp is linear.
Furthermore, these low differential gain structures allow a very low RF spectrum linewidth thanks to low coupling of the ASE noise on the mode-locked laser modes and in particular low propagation losses. These RF performances highlight the potential of quantum well MLLs for extremely low timing jitter applications in active or hybrid schemes.
References and links
1. E. A. Avrutin, J. H. Marsh, and E. L. Portnoi, “Monolitic and multi-Gigahertz mode locked semiconductor lasers: Constructions, experiments, models and applications,” IEE Proc.-Optoelectron. 147, No 4 (2000).
2. G. A. Keeler, B. E. Nelson, D. Agarwal, C. Debaes, N. C. Helman, A. Bhatnagar, and D. A. B Miller, “The benefits of ultrashort optical pulses in optically interconnected systems,” IEEE J. Quantum Electron. 9, 477–485 (2003). [CrossRef]
3. K. A. Williams, M. J. Thompson, and I. H. White, “Long-wavelength monolitic mode-locked diode lasers,” New J. Phys. 6, 179 (2004). [CrossRef]
4. K. Sato, “Optical pulse generation using Fabry-Pérot lasers under continuous-wave operation,” IEEE J. Sel. Top. Quantum Electron. 9, 5 (2003).
5. D. Kunimatsu, S. Arahira, Y. Kato, and Y. Ogawa, “Passively mode-locked laser diodes with bandgap-wavelength detuned saturable absorbers,” IEEE Photon. Technol. Lett. 11, 1363–1365 (1999). [CrossRef]
6. K. Yvind, D. Larsson, L.J. Christiansen, C. Angelo, L. K. Oxenlwe, J. Mrk, D. Birkedal, J. M. Hvam, and J. Hanberg, “Low-jitter and high-power 40-GHz all-active mode-locked lasers,” IEEE Photon. Technol. Lett. 16, 975–977 (2004). [CrossRef]
7. F. Lelarge, B. Dagens, J. Renaudier, R. Brenot, A. Accard, F. Van Dijk, D. Make, O. Le Gouezigou, J.-G. Provost, F. Poingt, J. Landreau, O. Drisse, E. Derouin, B. Rousseau, F. Pommereau, and G.-H. Duan, “Recent advances on InAs/InP quantum dash based semiconductor lasers and optical amplifiers operating at 1.55 µm,” IEEE J. Sel. Top. Quantum. Electron. 13, 11 (2007).
8. J. Renaudier, R. Brenot, B. Dagens, F. Lelarge, B. Rousseau, F. Poingt, O. Legouezigou, F. Pommereau, A. Accard, P. Gallion, and G.-H. Duan, “45 GHz self-pulsation with narrow linewidth in quantum dot Fabry-Perot semiconductor lasers at 1.5 µm,” IEE Electron. Lett. 41, 1007 (2005). [CrossRef]
9. C. Gosset, K. Merghem, A. Martinez, G. Moreau, G. Patriarche, G. Aubin, A. Ramdane, J. Landreau, and F. Lelarge, “Subpicosecond pulse generation at 134 GHz using a quantum-dash-based Fabry-Perot laser emitting at 1.56 µm,” Appl. Phys. Lett. 88, 241105 (2006). [CrossRef]
10. M. G. Thompson, A. R. Rae, R. V. Penty, I. H. White, A. R. Kovsh, S. S. Mikhrin, D. A. Livshits, and I. L. Krestnikov, “Absorber Length Optimisation for Sub-Picosecond Pulse Generation and Ultra-Low Jitter performance in Passively Mode-Locked 1.3µm Quantum-Dot Laser Diodes,” Anaheim, USA, OFC OThG3 (IEEE, Anaheim, 2006).
11. B. W. Hakki and T. Paoli, “Gain spectra in GaAs double heterostructure injection lasers,” J. Appl. Phys. 46, 1299–1305 (1975). [CrossRef]
12. Y. Barbarin, E. A. J. M. Bente, G. Servanton, L. Mussard, Y. S. Oei, R. Nötzel, and M. K. Smit, “Gain measurements of Fabry-Perot InP/InGaAsP lasers using an ultrahigh-resolution spectrometer,” Appl. Opt. 45, 35, 9007–9012 (2006). [CrossRef] [PubMed]
13. G. P. Agrawal and N. K Dutta, Long wavelength semiconductor lasers, (Van Nostrand Reinhold Co. Inc. New York1994).
14. L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated Circuits, (Wiley, New York, 1995). Chap. 5.
15. G. P. Agrawal and N. A. Olsson, “Self-phase modulation and spectral broadening of optical pulses in semiconductor laser amplifiers,” IEEE J. Quantum. Electron. 25, 2297–2306 (1989). [CrossRef]
16. C. H. Henry, “Theory of the linewidth of semiconductor lasers,” IEEE J. Quantum Electron. 18, 259–264 (1982). [CrossRef]
17. U. Bandelow, M. Radziunas, A. Vladimirov, B. Hüttl, and R. Kaiser, “40 GHz mode-locked semiconductor lasers: theory, simulations and experiment,” Opt. Quantum Electron. 38, 495–512 (2006). [CrossRef]
18. G. Baili, M. Alouini, D. Dolfi, F. Bretenaker, I. Sagnes, T. Merlet, and J. Chazelas, “Novel architectures of very low RIN semiconductor lasers in extended cavities for high performance microwave links” International Topical Meeting on Microwave Photonics, 2006 Page(s): 1–4
19. D. J. Derickson, R. J. Helkey, A. Mar, J. R. Karin, J. G. Wasserbauer, and J. E. Bowers, “Short Pulse Generation using Multisegment Mode- Locked Semiconductor Lasers,” IEEE J. Quantum Electron. 28, 2186–2202 (1992). [CrossRef]