A laser based on a ribbon multicore ytterbium doped fiber where different cores amplified different spectral bands, has been mode-locked with a single saturable absorber mirror. Tunable dual wavelength synchronized picosecond pulses were obtained. Compensation of differential cavity roundtrip times was achieved in the fiber.
© 2015 Optical Society of America
Passively mode-locked solid-state and fiber lasers are commercial products which are widely used as a source of ultrashort pulses (from tens of picoseconds down to few tens of femtoseconds) for scientific, medical and industrial applications. However, lasers delivering ultrashort pulses simultaneously at multiple wavelengths are still a research topic in view of their various applications in fields such as difference frequency generation, pump-probe measurements, remote sensing, coherent anti-stokes Raman scattering (CARS), single cycle optical wave synthesis, THz sources, etc. Considering fiber lasers only, a number of approaches have been reported showing successful dual wavelength mode-locking [1–6] (and on up to seven wavelengths ). These investigations have concerned different rare earth doped fiber lasers featuring both active and passive mode-locking using different techniques including amplitude and phase modulators [1,3], nonlinear polarization evolution [2,4,6] and real saturable absorbers [7,8]. In the most encountered configurations, the lasers include a single amplifying fiber to provide gain in combination with a spectral filter to ensure multi-wavelength operation. With a single rare earth doped fiber it is almost impossible to get stable mode-locking on two channels which are too close in wavelength, because of gain competition in the homogeneously broadened gain linewidth, or which are too far in wavelength, because of the limits of the gain bandwidth. For that reason, a few contributions have reported the use of separate gain fibers to reach a larger frequency separation. In one case, two separate erbium doped fibers were used to provide gain in C-band (~1540 nm) and L-band (~1580 nm) in order to make a source suited for broadband optical communications . In a second demonstration, M. Zhang et al. had coupled an Er doped fiber ring with an Yb doped fiber ring using a single wall carbon nanotube saturable absorber . More recently, passive synchronization of Er- and Tm-doped mode-locked fiber lasers have been achieved through a common graphene saturable absorber [9,10]. Timing jitter as low as 67 fs has been measured between the two pulse trains .
In this paper, we present a dual wavelength all-normal dispersion mode-locked laser based on an Yb doped multicore fiber (Yb-MCF) for the first time to our knowledge. The different cores serve to amplify different wavelengths which are combined in a common section of the cavity comprising a semiconductor saturable absorber mirror that leads to mode-locked operation. The multicore fiber serves also for filtering out different spectral bands in the laser. In addition to offer a large flexibility in terms of operation wavelengths, this fiber laser platform constitutes a solution for energy scaling through the addition of power from the different cores and it avoids gain competition due to homogeneous broadening with closely spaced laser frequencies.
Getting simultaneous mode-locked operation of a laser on two carrier wavelengths does not always ensure that the two pulse trains are at the same repetition frequency and does not guarantee they perfectly overlap in time in case they exhibit the same repetition rate. In the following, we report experiments in which cavity round trip times were adjusted and balanced in the multicore fiber itself thanks to bending and twist in a loop arrangement. Perfect pulse synchrony was ensured by the common saturable absorber.
The proposed cavity based on a multicore fiber amplifier offers new opportunities for the realization of laser sources of synchronized multiwavelength pulses: a large versatility in the choice of operating wavelengths, pump controlled switching, in-fiber delay tuning, power combining, compactness, stability, etc.
2. Experimental set-up
Experiments have been carried out with a multicore ribbon fiber made up of a linear array of 15 Ytterbium doped single mode waveguides [inset in Fig. 1]. The fiber was previously used for the amplification and synthesis of femtosecond pulses by coherent spectral combining . The guided mode diameter is 5 μm. The 30 μm separation distance between neighboring cores is sufficient to avoid couplings in the array. The different cores could be pumped separately by 976 nm fiber coupled laser diodes with a maximum power of 300 mW. The heavily Yb-doped MCF was ~1 m long so that the pump was completely absorbed. A schematic drawing of the laser is given in Fig. 1.
A dichroic beam splitter combined the pump radiation with the laser fields which shared the same free space optics for coupling into the waveguides. A bulk mirror was butt coupled to the rear end of the Yb-MCF and closed the linear cavity so that the laser radiations make a round trip in the amplifier. The cavity included a diffraction grating (600 gr/mm) which performed spectral beam combining, in the forward direction, so as to form a single beam in the front part of the resonator. In the backward direction, the grating achieved spectral beam filtering and directed two spectral bands toward two different cores of the Yb-MCF. The common part of the cavity was closed by a semi-conductor saturable absorber mirror (SESAM). The parameters of the SESAM were as follows: non saturable absorption ANS = 20%, saturable reflectivity ΔR = 30%, relaxation time of 500 fs. A pinhole was introduced close to the SESAM to better ensure selection of a common direction (wave vector) for the combined beams. The laser output was provided by the reflection of the intracavity beams on the grating (zero order) leading to ~10% out-coupling. The focal length of lenses between the Yb-MCF and the grating, the beams incident angles on the grating together with the fiber geometry fixed the separation between the central wavelengths (and the widths) of the selected spectral bands. The configuration thus offers a great spectral flexibility. The output beam was analyzed (i) by a spectroscopic device, to get the optical spectrum, (ii) by a fast photodiode (1 GHz bandwidth), for display of the pulse trains and RF spectrum, and (iii) by a background free SHG autocorrelator for pulse duration measurements.
3. Single and dual wavelength mode locking
We chose to pump two cores separated by 90 µm and set the grating to address two beams to these channels with a difference in wavelength of ~2.6 nm. We started by switching on a single laser channel, corresponding to a center wavelength of 1028.37 nm, and raised the pump level. After some adjustments, in particular the focusing on the SESAM to reach the saturation criteria, the laser turned to mode-locked operation as indicated by the pulse train on the scope that displayed the photodiode signal. At the same time, the laser spectrum broadened from 0.1 nm to ~0.9 nm full width half maximum in intensity (FWHMI). The spectrum bandwidth was limited by the intracavity grating dispersion. From the autocorrelation trace we measured the pulse duration to be 2.7 ps with a profile close to a Gaussian [Fig. 2]. The pulses were not Fourier transform limited with a time bandwidth product of ~2.1.
The generation of positively chirped-pulses was expected because of the all-normal dispersion of the cavity (~4.105 fs2) . However, the low pulse peak power hindered the formation of highly chirped dissipative solitons with larger spectral widths . Although the grating was a polarization sensitive device that obliged to add wave-plates to compensate for the fiber birefringence, the pulse peak power was too low to ensure nonlinear polarization evolution in the Yb-MCF. So, mode-locking initiation and stabilization is mainly attributed to the combined actions of the saturable absorber mirror and the effective spectral filter. The impact of the spectral filter is highlighted by numerical simulations based on the round trip laser model . We use the standard split-step method with the arrangement of the laser cavity elements as shown in Fig. 1, in a single channel configuration. Pulse propagation along the gain fiber is described by the extended nonlinear Schrödinger equation (NLSE), including second order dispersion, Kerr nonlinearity and saturated gain with a finite bandwidth of 40 nm . Numerical results obtained for a 2 nm bandwidth spectral filter, assuming a Gaussian spectral shape, are shown in Fig. 3. Pulse evolution along the cavity shows that passive spectral filtering plays a key role in pulse stabilization [Fig. 3(a)]. Its combination with the amplitude modulation provided by the saturable absorber allows to satisfy pulse self-consistency in the spectral and temporal domains. Moreover, the calculated solution presents a quasi-Gaussian temporal profile with 2.6 ps FWHMI and a narrow spectral width of 0.85 nm which are very close to experiments [Fig. 3(b-c)].
Then the channel previously used was switched off and the pump of the second channel was switched on. The pump power was increased until mode-locking start up similarly to the previous case. The center wavelength was here at 1031.05 nm. The pulse features were close to those measured for the previous channel. We further switched the first channel back on. The two wavelength bands operated simultaneously and the mode-locked regime was preserved for both channels. We measured the optical spectrum of the laser output and a typical recording is given on Fig. 4.
Tunability of the selected wavelengths can be achieved by mechanical rotation of the grating G over the gain bandwidth of the Ytterbium ions. Choice of the laser diode pumps to be switched on is another option. For example it was possible to change in that way the separation in wavelength between the two mode-locked channels.
Figure 5 reports some recordings showing on one hand a variation from 1.6 nm to 5.6 nm of the separation between the dual wavelengths and on the other hand a change of the center wavelength from 1029 nm to 1031 nm. We got also mode-locked operating regime simultaneously on three spectral bands. It was not possible to get mode-locked operation on more than three channels because of the inhomogeneity of the Yb-MCF in terms of birefringence. Polarization eigen-axes vary from core to core in the array as well as the magnitude of the birefringence. Therefore the settings of the wave-plates in the free space part of the cavity to get low losses and efficient mode-locking could not be identical for all waveguides. That could be solved only by use of a Faraday mirror or by the fabrication of a polarization maintaining multicore fiber.
In the above mentioned results, the two pulse trains were not synchronized leading to interweaved pulse trains in the time trace of the photodiode signal. This resulted from the difference in cavity length from the two laser arms. Despite the fact that the two wavelengths were amplified in two waveguides of the Yb-MCF, so that they shared a common physical length, several mechanisms might explain a difference in optical path. First of all, chromatic dispersion of the fiber, measured to be −34 ps/km.nm, contributed to a cavity length difference of 55 μm (in air). In addition, propagation constant of the different guides of the array are extremely dependent on the refractive index value and on the core diameter. In practice, we measured, by spectral interferometry, that group delay varied by nearly 270 fs between the two cores involved here for the 1 m long Yb-MCF. This contribution to optical path length difference on a roundtrip was therefore of 162 μm (in air). The fiber was not polarization maintaining and extra contributions from the fiber birefringence may be expected. Usually, synchronization of two mode-locked lasers requires the use of a tunable delay line to get cavity lengths sufficiently close to each other. Here we used a trick previously exploited in a multicore fiber amplifier . The Yb-MCF was arranged so that it makes a loop between two straight and fixed sections. When the loop stands in a plane which is perpendicular to the linear core array of the straight fiber sections, there is no impact on the differential delay. On the contrary when the loop is tilted, phase and group delay change across the array. The variation is linear from side to side in the array, the slope being proportional to the sine of the tilt angle. So, in the laser, we compensated for the initial difference in cavity length between the channels by tilting the Yb-MCF loop. Once the proper orientation was reached the loop was left fixed. The use of an Yb-MCF is a compact and robust method to obtain and maintain pulse trains synchronisation.
The RF spectrum of the laser single output beam detected by the photodiode served for the adjustments. The peaks characteristic of the two wavelength channels were initially separated and they merged in a single narrow peak after proper setting of the loop. Once the cavity lengths were sufficiently close to each other, the SESAM ensured the synchronization of the two channels as reported in . A typical recording is shown on Fig. 6 together with the dual wavelength pulse train on a short time scale which attests of the two pulse trains overlap. The RF cavity frequency peak was close to 53.75 MHz and had a width limited by the analyzer resolution (<10 kHz) with very low pedestal (−70 dB contrast).
The autocorrelation trace of the dual wavelength pulses, characterized by a profile with strong periodic modulations, is shown on Fig. 7 for the case of Δλ = 2.56 nm. It is consistent with the autocorrelation expected from a pulse resulting from the coherent superposition of a couple of pulses with different carrier frequencies. The theoretical trace in red open circles on Fig. 7 is in good agreement with the experimental data. The corresponding pulse shape with sharp intensity beatings at 735 GHz is given in inset. The synthesized pulse consisted in a train of ~700 fs (FWHMI) pulses separated by ~1.4 ps within an envelope of 5 ps duration (FWHMI).
It is worth mentioning that the intensity autocorrelation profile is not dependent on the phase relationship between the carrier wavelengths. Therefore the channeled pattern in the recorded traces does not provide any information on the relative round-trip phase slip between the two pulses of different color. The fact that the two laser arms almost shared the same free space path and the fact that they shared the same amplifying optical fiber (Yb-MCF) would suggest that their optical path length difference is highly stable. In the case of amplification with such a ribbon Yb-MCF, interferometric phase stability between neighboring cores in the array was previously observed during several hours in an unprotected environment . For that reason and also because it has been already observed that two mode locked lasers synchronized through a common saturable absorber can be occasionally phase-locked, we expected to get phase-locking of the two carriers. Phase-locking means here that the difference Δfceo in carrier envelope offset (CEO) frequency between the two wavelength channels could be almost cancelled.
We have used an ytterbium doped multicore fiber to build up a multiline laser with intracavity spectral combining. Mode locked operation was achieved simultaneously on two wavelengths thanks to a single saturable absorber mirror implemented on the common arm. To the best of our knowledge this is the first time a multicore rare earth doped fiber laser has been mode-locked with spatially dispersed amplification in the multicore fiber. Synchronized ~5 ps pulse trains were obtained after adjustment of the cavity lengths through twist and bending of the gain fiber to get a common repetition frequency. Final synchronization of the pulses resulted from the sharing of the saturable absorber. Separation in wavelength has been varied through a change in the active channels of the amplifier array and through rotation of the grating. Autocorrelation traces of the two-tones pulses were recorded showing strongly modulated THz beatings. The mode locked Yb-MCF laser synthesized trains of ultrashort pulses with duration (FWHMI) in the range 1.1 ps (Δλ = 1.6 nm) to 315 fs (Δλ = 5.6 nm) in the current design. Shorter pulses are easily achievable by changing the intracavity spatial dispersion. The laser was of low power here but the concept can be scaled to high power with the use of a cladding pumped large core multicore fiber such as the ones already exploited in . Polarization maintaining would help to get mode-locking on a larger number of channels. The used multiple core fiber consisted of a ribbon of identical Yb-doped waveguides. It would be straightforward to make a similar fiber with a set of cores with different gain bands using different rare earth ions for doping (Yb, Er, Tm).
The authors thank Karen Delplace and Damien Labat for their contributions to the multicore fiber fabrication. G. Bouwmans and L. Bigot acknowledge the financial support of Labex CEMPI (ANR-11-LABX-0007), Equipex FLUX (ANR-11-EQPX-0017), and the CPER Campus Intelligence Ambiante (CPER-CIA 2007-2013).
References and links
1. L. R. Chen, G. E. Town, P.-Y. Cortes, S. LaRochelle, and P. W. E. Smith, “Dual-wavelength actively modelocked fibre laser with 0.7 nm wavelength spacing,” Electron. Lett. 36(23), 1921–1923 (2000). [CrossRef]
2. Y. D. Gong, X. L. Tian, M. Tang, P. Shum, M. Y. W. Chia, V. Paulose, J. Wu, and K. Xu, “Generation of dual wavelength ultrashort pulse outputs from a passive mode-locked fiber ring laser,” Opt. Commun. 265(2), 628–631 (2006). [CrossRef]
3. S. Pan and C. Lou, “Stable multiwavelength dispersion tuned actively mode-locked erbium doped fiber ring laser using nonlinear polarization rotation,” IEEE Photonics Technol. Lett. 18(13), 1451–1453 (2006). [CrossRef]
4. Z. W. Xu and Z. X. Zhang, “All-normal dispersion multiwavelength dissipative soliton Yb-doped fiber laser,” Laser Phys. Lett. 10(8), 085105 (2013). [CrossRef]
5. D. Pudo and L. R. Chen, “Actively modelocked, quadruple wavelength fibre laser with pump controlled wavelength switching,” Electron. Lett. 39(3), 272–274 (2003). [CrossRef]
6. Z. Yan, X. Li, Y. Tang, P. P. Shum, X. Yu, Y. Zhang, and Q. J. Wang, “Tunable and switchable dual-wavelength Tm-doped mode-locked fiber laser by nonlinear polarization evolution,” Opt. Express 23(4), 4369–4376 (2015). [CrossRef] [PubMed]
7. Z.-C. Luo, A.-P. Luo, and W.-C. Xu, “Tunable and switchable multiwavelength passively mode-locked fiber laser based on SESAM and inline birefringence comb filter,” IEEE Photonics J. 3(1), 64–70 (2011). [CrossRef]
8. M. Zhang, E. J. R. Kelleher, A. S. Pozharov, E. D. Obraztsova, S. V. Popov, and J. R. Taylor, “Passive synchronization of all-fiber lasers through a common saturable absorber,” Opt. Lett. 36(20), 3984–3986 (2011). [CrossRef] [PubMed]
9. J. Sotor, G. Sobon, I. Pasternak, A. Krajewska, W. Strupinski, and K. M. Abramski, “Simultaneous mode-locking at 1565 nm and 1944 nm in fiber laser based on common graphene saturable absorber,” Opt. Express 21(16), 18994–19002 (2013). [CrossRef] [PubMed]
10. J. Sotor, G. Sobon, J. Tarka, I. Pasternak, A. Krajewska, W. Strupinski, and K. M. Abramski, “Passive synchronization of erbium and thulium doped fiber mode-locked lasers enhanced by common graphene saturable absorber,” Opt. Express 22(5), 5536–5543 (2014). [CrossRef] [PubMed]
11. P. Rigaud, V. Kermene, G. Bouwmans, L. Bigot, A. Desfarges-Berthelemot, D. Labat, A. Le Rouge, T. Mansuryan, and A. Barthélémy, “Spatially dispersive amplification in a 12-core fiber and femtosecond pulse synthesis by coherent spectral combining,” Opt. Express 21(11), 13555–13563 (2013). [CrossRef] [PubMed]
12. W. H. Renninger, A. Chong, and F. W. Wise, “Dissipative solitons in normal-dispersion fiber lasers,” Phys. Rev. A 77(2), 023814 (2008). [CrossRef]
14. T. Walbaum, M. Löser, P. Gross, and C. Fallnich, “Mechanisms in passive synchronization of erbium fiber lasers,” Appl. Phys. B 102(4), 743–750 (2011). [CrossRef]
15. L. J. Cooper, P. Wang, R. B. Williams, J. K. Sahu, W. A. Clarkson, A. M. Scott, and D. Jones, “High-power Yb-doped multicore ribbon fiber laser,” Opt. Lett. 30(21), 2906–2908 (2005). [CrossRef] [PubMed]