The operation of a self-starting, passively harmonic modelocked, figure-eight laser is experimentally demonstrated. A stable pulse train with near half duty-cycle is produced at a repetition rate of 1.7 GHz at 1536 nm wavelength without any starting/triggering mechanism and stays modelocked as long as it is being pumped.
© 2009 Optical Society of America
Compact fiber lasers have emerged as the practical alternative to conventional bulk solid-state lasers in many applications [1, 2]. The main advantages relative to conventional bulk solid-state lasers include high beam quality, high efficiency, compactness, robustness, and excellent heat dissipation. Since Duling  reported the first passive mode-locking of a figure-eight fiber laser (F8L), the conventional F8L generates very short pulses (mostly sub-picosecond) directly from all-fiber lasers [1–4]. However, for many applications, including most optical communication systems, conventional F8Ls have several serious drawbacks. Typically their pulses are short and their repetition rate is slow. This gives a very small duty cycle and many applications including communications typically require duty-cycles closer to a half . Furthermore, the primary limitation of conventional F8Ls is the low pulse repetition rate, which is determined by the cavity length, typically more than a few meters in an F8L configuration. Whilst there are reports of much higher repetition rates, up to 71 GHz, they are not of stable pulse trains but disordered sequences of optical bursts [3, 6]. A harmonically mode-locked erbium fiber ring laser has been reported to achieve 7th order harmonic oscillation with repetition rate of 220 MHz . Optical clock-division operation at 2.5GHz has also been demonstrated using a mode-locked F8L but it used an external driving signal of 2.5 GHz . In Ref. , a 10 GHz external signal was used to achieve 20 GHz repetition rate. Y. Ueno, et al, used a complex symmetric Mach-Zehnder-type (SMZ) delayed-interference signal-wavelength convertor (DISC) switch and an energy-distributed Mach-Zehnder interferometer (ED-ZMI) to achieve 10 GHz repetition rate . Clearly, to date, no one has demonstrated a simple system for generating high repetition rate stable periodic pulse trains directly from the F8L configuration. A third drawback is that they are generally not self-starting so that the F8Ls require an extra starting mechanism to be realized. Reported self-starting F8Ls [3, 11] were not repeatable (they do not guarantee self-starting) nor were true self-starting, where the laser always generates a stable pulse train (not burst pulses) without any starting mechanism or technique.
In this paper, for the first time to our knowledge, we report an F8L that solves all three of these problems with only five components; two couplers, a polarization controller (PC), an isolator, and a semiconductor optical amplifier (SOA). Frisken et al reported a similar F8L  but it requires many more components; an Erbium-doped fiber (EDF), a pump laser-diode (LD), two PCs, a variable optical attenuator, and an isolator. Frisken’s does not support half duty-cycle either. The experimental demonstration is given of a near half duty-cycle, self-starting fiber laser that employs an SOA as the gain and switching medium. This laser automatically generates a stable pulse train with near half duty-cycle at a repetition rate over 1 GHz; this pulse train commences soon after the pump power is supplied and remains stable for as long as it is being pumped. The simplicity, high repetition rate, high duty-cycle, and self-starting ability of this all-optical modelocked laser make it highly practical for a number of applications including optical logics and communications.
A schematic diagram of the experimental setup is shown in Fig. 1. The figure-eight laser consisted of a nonlinear amplifying loop mirror (NALM) and a linear loop connected by a coupler to recirculate the switched pulse through a polarization-insensitive (PI) optical isolator. Since the optical isolator is polarization-insensitive, there is no polarization selection in the linear cavity. The nonlinear loop consisted of a 90:10 center coupler, a PC, an SOA, and a 70:30 output coupler. The asymmetric center coupling ratio (90:10) reduces the switching energy of NALM  and this makes the self-starting easier. The output coupling ratio (30:70) was selected to get as high output power as possible in the given structure whilst maintaining good stability. Finally, a 4.53 m long linear loop and a 6.45 m long nonlinear loop were constructed. The SOA, QSOA-1550 of QPhotonics, LLC, with 30 nm spectral width at 1527 nm center wavelength provides the nonlinear phase shift and gain that effect nonlinear switching and enable self-starting of the optical pulse. The PC in the nonlinear loop compensates for the stress birefringence of the fiber and controls the operation modes of the lasers.
The passive mode-locking of conventional F8Ls typically starts from noise fluctuations in the laser due to perturbation. The fundamental principle of the self-starting of this novel laser is however based on self amplitude modulation due to the fast saturation (femtosecond order) and relatively slow recovery (initial recovery in a picosecond and full recovery around 500ps) properties of the SOA gain. First, the SOA generates the amplified spontaneous emission (ASE) noise, which will be the initial input signal to the SOA. The SOA amplifies the noise signal with high gain (23.5 dB at 10 μW input signal at 300 mA pump) but the gain is rapidly saturated (the saturation energy of the SOA is below 1 pJ) and then recovers slowly compared to saturation. As the pulse duration is very long (initially continuous wave ASE) compared to the recovery time of the SOA gain (a few tens of picoseconds), the signal is affected by the recovering gain of the SOA resulting in the slow decrease of the pulse amplitude in the top part of the pulse shown in Fig. 4. With the nonlinear phase shift of NALM, this will establish initial pulse-like signals that are seeds for the laser. As the input signal power to the NALM increases, the NALM periodically transmits and reflects the input signal due to the nonlinear phase shift so that there are quantized optical power levels that satisfy the transmission condition [6, 13]. Once a pulse is initiated and the power satisfies the transmission condition of NALM, the pulse keeps circulating in the F8L cavity. On the other hand, pulses not satisfying the transmission condition are reflected by the NALM and suppressed by the isolator. Once the mode-locking is started, the circulating pulse saturates the laser gain to a level that is just sufficient to compensate for the losses to the pulse itself, whereas any other circulating low-intensity light experiences more loss than gain and thus dies out during the following cavity round-trips. As the pulse signals circulate, the pulse shape and the repetition rate are thus stabilized. The PC serves to select a preferential polarization eigenstate of the light in the cavity.
When the appropriate SOA current was supplied, a stable pulse train was automatically generated and the output characteristics of the self-starting fiber laser were measured at port Pout with an oscilloscope, an RF spectrum analyzer and an optical spectrum analyzer.
The optical pulse was generated within a current range for each repetition rate as shown in Fig. 2. For example, on increasing the SOA current, the fiber laser generated a stable pulse train with the repetition rate of 1.2 GHz when the SOA current exceeded 170 mA until it reached 220mA. At around 220 mA the repetition rate of the generated pulse changed to 860 MHz and the laser kept generating the stable pulse until the SOA current reached 450 mA. Over 450 mA the laser could no longer generate a stable pulse train. On decreasing the SOA current, at around 220 mA the repetition rate changed to 1.2 GHz again and below 170 mA the pulses vanished. At the other PC setting, the laser started to generate pulses with the repetition rate of 1.7 GHz from near 300 mA SOA current and kept generating pulses until the current reached 385 mA. This phenomenon that the pulse generation occurs within the current range can be explained by a fast saturable absorber theory where the relaxation time of the absorber is short compared with the pulse width . Within the lasing current range, the overall gain of the laser cavity exceeds the overall loss of the cavity resulting in pulsed lasing. The transition from one repetition rate to another is initiated by alteration of the PC’s position, which affects the polarization of the signal inside the cavity, the gain of the laser and consequently the phase between the pulsed signal and non-pulsed signal. That is combined with the nonlinear phase shift in the SOA to change the switching properties of the NALM, which results in the change of the repetition rate. The detailed phenomena inside the laser cavity can be understood through theoretical analysis similar to . The measured amplitudes of the pulses as a function of the launched SOA current are shown in Fig. 2 for various repetition rates; 860 MHz, 1.2 GHz, and 1.7 GHz. The generated pulses had the maximum amplitude and were most stable near the center of each current range. Since the cavity energy is quantized, the pulse repetition rate changes in steps, not in a continuous fashion.
Several kinds of stable pulse trains were obtained with various repetition rates by adjusting the polarization controller and the SOA currents. The repetition rate of 1.7 GHz shown in Fig. 3 corresponds to 92 times the fundamental repetition rate generated from a conventional fiber laser with the same cavity length. This indicates that the stabilizing scheme is frequency selective. The typical fundamental repetition rate by so-called fundamental mode-locking, which is determined by the cavity length, is ~18.59 MHz, corresponding to ~10.98 m long cavity. Once a pulse train was generated by starting pumping the SOA or adjusting the PC, it stayed in a stable state for days without readjustment. Fig. 4 shows the pulse shapes with 1.2 GHz (a) and 1.7 GHz (b) repetition rate. Digital oscilloscopes were used to record the pulses. The measured full-width half-maximums (FWHM) of the pulses were ~0.42 ns for 1.2 GHz and ~0.28 ns for 1.7 GHz. The estimated timing jitter was below 20 ps for 1.7 GHz pulses. The pulse duty-cycles of most of the generated pulses were around a half.
The average optical output power was ~0.82 mW at 180 mA SOA current, at which current the SOA generated 1.41 mW ASE noise. The output optical spectrum corresponding to Fig. 3 (a) and Fig. 4 (a) is shown in Fig. 5, which was measured with an optical spectrum analyzer. The laser output had a 0.92 nm bandwidth with a center wavelength of 1535.7 nm, which is slightly longer than the center wavelength of the SOA. This mainly came from red shift in the laser cavity.
The RF spectrums of the generated pulses with 1.2 GHz and 1.7 GHz repetition rates were recorded using RF spectrum analyzer and are shown in Fig. 6. The supermode suppression ratios degraded from ~ 33 dB to ~ 28 dB when the repetition rate changed from 1.2 GHz to 1.7 GHz.
The proposed and demonstrated laser operates in harmonic mode-locking mode. Harmonic mode-locked pulse trains are generally randomly spaced. Under certain conditions, self-organized pulse trains with equal spacing can be formed through pulse-to-pulse interactions. The SOA provides the nonlinear phase shift and gain that effect nonlinear switching and enable self-starting of the optical pulse by self amplitude modulation. The slight asymmetric gain experienced across a pulse imparts a small frequency drift to that pulse. Qualitatively, the active ion population inversion of the gain medium depletes while transferring energy to a traversing pulse. Before the next pulse arrives, the population inversion will have a given amount of time to recover before being depleted once again. In the process of depleting the inversion, the pulse experiences a time-dependent gain, i.e. the leading edge of the pulse receives more gain than the trailing edge. This time-dependent depletion generates a group-velocity drift of the pulse toward the region of higher gain. The magnitude of this drift is related to the amount of gain given to the pulse and thus to the degree to which the inversion has recovered since the previous pulse. The result is an effective repulsion force between adjacent pulses and a steady-state condition consisting of equally spaced pulses. The time dependence of the free-carrier density reflects the phase modulation effect in the SOA. Therefore, the SOA in the laser cavity provides not only self-starting and cleaner pulses but also a passive phase modulation that stabilizes the laser repetition rate. Since the interplay between the self amplitude modulation and phase modulation is intricate, to fully elucidate the behavior of this laser will require a numerical model, which is beyond the scope of this paper [15, 16].
In conclusion, we have experimentally demonstrated, for the first time to our knowledge, a passively mode-locked figure-eight fiber laser with self-starting capability and near half duty-cycle based on the fast saturation and slow recovery mechanism of a semiconductor optical amplifier. A stable pulse train with near half duty-cycle, at the center wavelength of 1535.7 nm was obtained at repetition rates up to 1.7 GHz. The simple fiber laser started mode-locking without any triggering methods and kept generating a stable pulse train until the laser was turned off. The repetition rate of the generated pulses is 92 times the fundamental repetition rate. This scheme provides a new method to achieve high duty-cycle, high repetition rate pulse trains with the F8L configuration. The simplicity, stability, and self-starting capability of this mode-locked laser configuration make it highly practical for a number of applications including optical logic and communications applications.
This work was supported in part by the Post-doctoral Fellowship Program of the University of Sydney and ARC Discovery Grant DP0666484.
References and links
2. S. Min, Y. Zhao, and S. Fleming, “Repetition rate multiplication in figure-eight fibre laser with 3 dB couplers,” Opt. Commun. 277, 411–413 (2007). [CrossRef]
3. I. N. Duling III, “All-fiber ring soliton laser mode-locked with a nonlinear mirror,” Opt. Lett. 16, 539–541(1991). [CrossRef]
4. T. O. Tsun, M. K. Islam, and P. L. Chu, “High-energy femtosecond figure-eight fiber laser,” Opt. Commun. 141, 65–68 (1997). [CrossRef]
5. M. Singh, A. K. Sharma, and R. S. Kaler, “Optimizing 10 Gbps optical communication system with duty cycle selection of return to zero pulse,” Optik - Int. J. Light Electron. Opt. 119, 359–364 (2008). [CrossRef]
6. A. B. Grudinin, D. J. Richardson, and D. N. Payne, “Energy quantization in figure eight fibre laser,” Electron. Lett. 28, 67–68 (1992). [CrossRef]
7. K. S. Abedin, J. T. Gopinath, L. A. Jiang, M. E. Grein, H. A. Haus, and E. P. Ippen, “Self-stabilized passive, harmonically mode-locked stretched-pulse erbium fiber ring laser,” Opt. Lett. 27, 1758–1760 (2002). [CrossRef]
8. L. Xu, M. Yao, B. C. Wang, I. Glesk, and P. R. Prucnal, “All-optical clock division with mode-locked figure-eight laser based on the slow carrier recovery rate in semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 14, 402–404 (2002). [CrossRef]
9. T. Papakyriakopoulos, K. Vlachos, A. Hatziefremidis, and H. Avramopoulos, “20-GHz broadly tunable and stable mode-locked semiconductor amplifier fiber ring laser,” Opt. Lett. 24, 1209–1211 (1999). [CrossRef]
10. Y. Ueno, S. Nakamura, and K. Tajima, “5-ps, 10-GHz pulse generation from an all-optical semiconductor switch embedded in a ring cavity,” Appl. Phys. Lett. 79, 2520–2522 (2001). [CrossRef]
11. V. Tzelepis, S. Markatos, S. Kalpogiannis, Th. Sphicopoulos, and C. Caroubalos, “Analysis of a passively mode-locked self-starting all-fiber soliton laser,” J. Lightwave Technol. 11, 1729–1736 (1993). [CrossRef]
12. S. J. Frisken, C. A. Telford, R. A. Betts, and P. S. Atherton, “Passively mode-locked Erbium-doped fibre laser with nonlinear fibre mirror,” Electron. Lett. 27, 887–889 (1991). [CrossRef]
13. N. H. Seong, D. Y. Kim, and S. P. Veetil, “Mode-locked fiber laser based on an attenuation-imbalanced nonlinear optical loop mirror,” Opt. Commun. 280, 438–442 (2007). [CrossRef]
14. H. A. Haus, “Theory of mode locking with a fast saturable absorber,” J. Appl. Phys. 46, 3049–3058 (1975). [CrossRef]
15. A. B. Grudinin and S. Gray, “Passive harmonic mode locking in soliton fiber lasers,” J. Opt. Soc. Am. B 14, 144–154 (1997). [CrossRef]
16. J. Nathan Kutz, B. C. Collings, K. Bergman, and W. H. Knox, “Stabilized pulse spacing in soliton lasers due to gain depletion and recovery,” IEEE J. Quantum Electron. 34, 1749–1757 (1998). [CrossRef]