2050
OPTICS LETTERS / Vol. 36, No. 11 / June 1, 2011
Tunable synchronously-pumped fiber Raman laser in the visible and near-infrared exploiting MOPA-generated rectangular pump pulses Dejiao Lin,* Shaif-ul Alam, Peh Siong Teh, Kang Kang Chen, and David J. Richardson Optoelectronics Research Centre, University of Southampton, Southampton, SO17 1BJ, UK *Corresponding author:
[email protected] Received February 24, 2011; revised April 21, 2011; accepted April 21, 2011; posted April 26, 2011 (Doc. ID 142981); published May 26, 2011 We report a tunable synchronously pumped fiber Raman laser (SPFRL) in the near-infrared (NIR) and visible wavebands pumped by a pulsed, all-fiber PM 1060 nm master oscillator power amplifier (MOPA) and its frequencydoubled output, respectively. The seed was adaptively shaped to deliver rectangular output pulses, thereby enabling selective excitation of individual Raman Stokes lines. Using filtered synchronous feedback of the desired Raman Stokes line, the linewidth of the SPFRL was reduced by a factor of 4 and the extinction ratio of the desired Raman Stokes was improved by more than 3 dB relative to a simple single-pass conversion scheme. A continuous tuning range of 2:2 THz was obtained for each of the Raman Stokes orders in the visible (spanning from green to orange— first to fifth Stokes lines). A larger 5:0 THz tunable range was achieved in the NIR spectral region. © 2011 Optical Society of America OCIS codes: 140.3550, 060.2320, 320.5540.
Fiber Raman lasers have attracted a great deal of interest because they allow the ready generation of light at a very wide range of interesting wavelengths [1]. A substantial body of work has been conducted on CW fiber Raman lasers that allow for compact, robust fully fiberized cavities with pump powers typically in the range of hundreds of milliwatts to hundreds of watts [2,3]. Operation in the pulsed regime is, however, compromised by the power dependence of the gain: a pulse propagating through a length of fiber will experience differing amounts of Raman gain across its temporal profile and this can lead to complex temporal (and spectral) pulse-shaping effects [4]. This issue can be addressed by using rectangularshaped pulses that provide constant Raman gain across the pulse profile. This method has been exploited in single-pass Raman amplifier configurations to obtain the selective excitation of individual Raman Stokes lines of up to ninth order pumped at 530 nm, and fourth order pumped at 1060 nm [5]. The method has also been extended to the sequential generation of multiple Raman Stokes using multistep pulses [6]. One disadvantage of this technique is that it does not provide for wavelength tuning around the individual Raman order gain peaks and the output spectrum tends to broaden due to competing nonlinear effects at high Raman orders. Synchronously pumped fiber Raman lasers (SPFRLs) were studied both theoretically and experimentally in the 1970s and 1980s [7–9]. Compared to single-pass Raman generation schemes, SPFRLs have the potential for considerably lower Raman thresholds and wavelength tuning and narrower spectral linewidths. Operation around the first- or second-order Raman Stokes peaks is relatively simple in terms of system configuration [8,9], however, reliable operation at higher-order shifts (e.g., fourth) requires multiple cavity mirrors and complicated adjustment to ensure synchronous pumping for all Raman Stokes orders [7]. Moreover, the power residing in the residual pump and all of the lower order Stokes significantly reduces the conversion efficiency to the desired higher-order Raman Stokes line. 0146-9592/11/112050-03$15.00/0
Herein we demonstrate the use of pulse-shaping techniques and wavelength-selective feedback to realize fiber master oscillator power amplifier (MOPA) pumped, wavelength-tunable SPFRLs capable of operating at high Raman orders, and capable of providing both narrower linewidths and higher extinction ratios for neighboring Raman Stokes lines than single-pass pulsed Raman conversion schemes. A schematic of the tunable SPFRL configured to operate in the visible regime is illustrated in Fig. 1. The MOPAbased pump source was seeded by a 1060-nm-fiber Bragg grating stabilized semiconductor laser diode that passed through an inline electro-optic modulator (EOM), driven by an arbitrary waveform generator (AWG) with 4 ns resolution to allow active pulse shaping. The seed signal was amplified in an all-fiber, three-stage, Yb-doped PM amplifier chain to deliver single-mode, single-polarization output with the required rectangular pulse shape. A lithium triborate (LBO) crystal was used to frequencydouble the MOPA output. The second harmonic generation (SHG) at 530 nm was launched into a 10 dB tap coupler (Tap 1) whose 90% output port was spliced to a 250 m length of Pirelli Freelight fiber, which was used as the Raman gain medium. A length of 250 m was selected as it provided a reasonable compromise between the need to keep attenuation and competing Kerr nonlinearities (e.g., self phase modulation, SPM and four-wave mixing, FWM) at an acceptable level while providing adequate nonlinear Raman gain at reasonable peak powers. The other end of the Raman gain fiber was spliced to a second 10 dB tap coupler
Fig. 1. (Color online) Layout of the tunable fiber Raman laser synchronously pumped by SHG from a PM MOPA. © 2011 Optical Society of America
June 1, 2011 / Vol. 36, No. 11 / OPTICS LETTERS
(Tap 2), the 90% port of which was used as the output of the Raman laser while the output of the 10% port was diffracted by a grating to select the wavelength for the feedback signal. The diffracted beam was coupled into another piece of Freelight fiber of 400 m length, providing sufficient delay to synchronize the ∼305 kHz pump and feedback pulses. The feedback signal was then spliced to the other input port of Tap 1 to combine it with the launched pump at 530 nm. The 10% output port of Tap 1 was monitored spectrally and temporally to ensure good synchronization between the pump and the feedback pulses. The diffraction angle of the grating could be rotated to allow wavelength tuning of the Raman laser. To configure the SPFRL for NIR operation, the SHG crystal was removed from the setup and the visible broadband taps were replaced with similar devices operating in the NIR. Because of the significant difference in group velocity between the visible and NIR, the repetition rate of the MOPA needed adjustment to maintain synchronization of the feedback signal and pump. We first experimentally investigated the Raman Stokes generation, starting with the first order, to the fifth order (covering from green, yellow to orange), for the visible SPFRL (note that due to group velocity dispersion, the repetition rate of the pump pulses had to be adjusted for different order Stokes pulses: for example, the pulse repetition rate was set at 304:6 kHz for the first-order Stokes and 306:0 kHz for the fifth-order Stokes). Figure 2 plots examples of spectra for the third-order Raman Stokes (yellow light) measured with an ANDO optical spectrum analyzer at a resolution setting of 0:1 nm. The fundamental pulse width was 20 ns and the repetition rate was set at 305:3 kHz to meet the need for synchronization. Even though the feedback signal power was less than 0.1% of the incident pump power, the linewidth of the desired third-order Raman Stokes output was decreased by a factor of 4 (0:5 nm versus 2:0 nm) and the extinction ratio was improved by >3 dB in the presence of the feedback signal (centered at 571:2 nm). The spectral bandwidth of the output signal broadened from that of the feedback signal due primarily to SPM-induced spectral broadening of the amplified signal in the long length of Raman gain medium. Four-wave mixing [10] and cross-phase modulation also contributed to this broadening; however, their impact was relatively small due to the large suppression (>10 dB) of the neighboring Raman lines.
Fig. 2. (Color online) Spectra of the third Raman Stokes output with and without feedback.
2051
Fig. 3. (Color online) Pulse shapes of SHG, total Raman output, the third Stokes (signal), second Stokes, and fourth Stokes (noise) (as spectrally resolved with a grating).
Figure 3 shows various output pulse shapes from the SPFRL measured at the 90% port of the Tap 2 output with the system configured to operate on the third Raman Stokes line. Compared with the SHG pulse in Fig. 3, the separated third Raman Stokes pulse is seen to be slightly shorter than the pump pulse because of walkoff among the Raman Stokes and pump wavelengths. The overall output pulse (pink curve in Fig. 3) exhibits a spike on its leading edge and multiple spikes on the trailing edge. These spike features result from the finite rise and fall times of the pump pulses (due to the 4 ns resolution of the AWG). The varying intensity in these regions leads to the generation of multiple Stokes components that increase in order with time on the leading edge and decrease in order with time on the trailing edge. During propagation, the group velocity mismatch (GVM) between pump and successive Raman orders causes these pulse components to coalesce on the leading edge of the output pulse, leading to the single spike feature. Conversely, at the trailing edge, successive Raman components move away from each other, resulting in distinct multiple pulses for the residual pump and the first and second Raman orders. The dynamic extinction ratio of the desired third Raman Stokes was estimated (from the temporal plots in Fig. 3) to be about 15 dB from the nearest neighbors (second and fourth Stokes), in agreement with estimates made from the time-averaged spectra recorded on an OSA (Fig. 2). Owing to the broad Raman gain bandwidth, the output of the Raman laser can be tuned by adjusting the wavelength of the feedback signal using the external bulk grating. As shown in Fig. 4, a continuous tuning range of 2:4 nm (equivalent to 2:2 THz in the frequency domain) was obtained for the third Stokes line. The peak powers of the frequency-doubled and the Raman laser outputs
Fig. 4. (Color online) Spectra of the tunable Raman laser of the third Stokes in the visible.
2052
OPTICS LETTERS / Vol. 36, No. 11 / June 1, 2011
Fig. 5. (Color online) Spectra of the visible tunable Raman laser at the (a) first Stokes line and (b) fifth Stokes line.
(from the 90% output port of Tap 2) were 48:1 W and 9:5 W (30% of which was due to spontaneous Raman scattering), respectively. The estimated conversion efficiency with respect to the incident pump power was ∼15%, which is reasonable when the fiber launch efficiency and fiber attenuation in the visible region are taken into account. We also studied the tuning performance of the SPFRL for other Raman Stokes orders by varying the injected pump power while maintaining a rectangular pulse shape. Figure 5 shows examples for the first and fifth Stokes orders where tuning ranges for the first and fifth Stokes orders were 2:2 nm and 2:8 nm, corresponding to 2:2 THz and 2:3 THz in frequency, respectively. It is observed that the spectral bandwidth for the higherorder Stokes was broadened due to competing nonlinear processes. However, we could still achieve around a four-fold decrease in spectral linewidth by means of synchronous pumping relative to the single-pass case. The estimated peak powers of the Raman laser outputs for the first and fifth Stokes orders were 2:3 W and 18:7 W for incident pump peak powers of 11:6 W and 96:8 W. The corresponding conversion efficiencies were estimated to be approximately 18% and 12%. It is to be noted here that the average output powers of the first and fifth Stokes orders were increased by about 4 dB and 2 dB in the presence of feedback (the extinction ratios were 18 dB and 12 dB). Similar experiments were carried out in the NIR regime. It was possible to operate at wavelengths up to the fourth-order Raman Stokes order; beyond this, the competitive nonlinear Kerr effects became too strong to allow good performance due to the proximity of the zero dispersion wavelength (∼1300 nm). Equal lengths (250 m) of Raman gain and feedback fiber were used. A pulse duration of 16 ns and repetition rates around 400 kHz were chosen for these experiments. Figure 6 shows an example of a spectrum for the fourth Raman Stokes line. Again, a substantial improvement in terms of linewidth and extinction ratio relative to adjacent orders was observed [as shown by comparing the spectra with and without the feedback signal, see Fig. 6(a)]. As shown in Fig. 6(b), a wavelength tuning range of up to 28 nm was obtained for the fourth Stokes order, corresponding to 5 THz in the frequency domain. This is more than double that obtained in the visible. The primary reasons for this are (i) the substantially lower fiber attenuation, and (ii) the longer walk-off length between the pump and Stokes components. The peak powers of the MOPA and SPFRL outputs were estimated to be 281 W and 119 W, respectively. The estimated conversion efficiency in this instance was better than 50%, assuming
Fig. 6. (Color online) (a) Spectra of the fourth Raman Stokes in NIR with and without feedback; (b) spectra of the tunable Raman laser at 1:3 μm (fourth Stokes order).
80% pump coupling efficiency to the Raman gain medium. In conclusion, we have demonstrated synchronously pumped NIR and visible Raman laser sources pumped by a single-polarization all-fiber Yb-doped PM MOPA incorporating adaptive pulse shaping and its frequencydoubled output, respectively. A tuning range of 2:2 THz in the visible regime and 5 THz in the NIR were achieved within each Stokes order by means of an external bulk grating. Replacing Tap 1 with a suitable wavelengthdivision multiplexer would significantly improve the efficiency of the combination and accordingly enhance the tuning range and optical output power as well as the spectral performance of the SPFRL. Moreover, incorporation of a tunable fiber Bragg grating would enable an all-fiber tunable Raman laser in the visible and NIR regions. References 1. C. Lin, R. H. Stolen, and L. G. Cohen, Appl. Phys. Lett. 31, 97 (1977). 2. J. Au Yeung and A. Yariv, J. Opt. Soc. Am. 69, 803 (1979). 3. Y. Feng, L. R. Taylor, and D. B. Calia, Opt. Express 17, 23678 (2009). 4. R. H. Stolen, C. Lee, and R. K. Jain, J. Opt. Soc. Am. B 1, 652 (1984). 5. K. K. Chen, S. U. Alam, P. Horak, C. A. Codemard, A. Malinowski, and D. J. Richardson, Opt. Lett. 35, 2433 (2010). 6. D. Lin, S. U. Alam, P. S. Teh, K. K. Chen, and D. J. Richardson, Opt. Express 19, 2085 (2011). 7. C. Lin and W. G. French, Appl. Phys. Lett. 34, 666 (1979). 8. M. Nakazawa, M. Kuznetsov, and E. P. Ippen, IEEE J. Quantum Electron. 22, 1953 (1986). 9. K. Smith, P. N. Kean, D. W. Crust, and W. Sibbett, J. Mod. Opt. 34, 1227 (1987). 10. S. A. Babin, D. V. Churkin, A. E. Ismagulov, S. I. Kablukov, and E. V. Podivilov, J. Opt. Soc. Am. B 24, 1729 (2007).
