Open Access
Add to BibSonomy
Abstract
We introduce a fully-integrated two-color sub-nanosecond fiber laser system that incorporates a backward-pumped polarization-maintaining (PM) Raman phosphosilicate fiber amplifier (RFA) followed by two fully-integrated fiber-coupled second harmonic generator (SHG) modules. The RFA is pumped by a continuous-wave (CW) Yb laser operating at 1116 nm. The pulsed signals are generated by gain-switched distributed feedback (DFB) laser diodes at 1178 nm and 1310 nm, respectively. The output pulsed DFB signals are independently or simultaneously amplified in the RFA. This amplification is achieved using both the broad SiO2 (∼13.2 THz) and relatively narrow P2O5 (39.9 THz) Stokes shifts. The laser system produces sub-nanosecond pulses at 589 and 655 nm, featuring repetition rates ranging from 40 to 100 MHz and an average power of up to 3 W (limited by the SHG crystal damage threshold) at each wavelength. The diffraction-limited output beams maintain an M2 value of < 1.05 across the entire range of output powers and repetition rates for both wavelengths.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Laser sources operating in both orange and red regions of the spectrum, featuring optimal pulse durations of a few hundred picoseconds and tens of MHz repetition rates find widespread applications in various biomedical fields, particularly in stimulated emission depletion (STED) microscopy [1–5]. This combination of parameters provides efficient depletion, mitigates photobleaching, and allows for high-speed acquisition rates [4].
Numerous studies have explored different combinations of Raman fiber converters and amplifiers, Yb lasers and amplifiers, and SHGs [6–13], as well as supercontinuum generation [5] for the development of pulsed visible sources. A thorough review and additional references of these studies are given in [3].
One method to generate visible beams involves unseeded Raman shifting of the pulsed green Yb laser second harmonic beam [1]. Although this approach allows for the generation of a few Stokes components in the yellow, orange and red spectral ranges, the stability, reliability and long-term operation of these systems have yet to be proven. Similar approaches where Raman shifted pulsed Yb laser output is frequency doubled suffer from the relatively low SHG efficiency due to the spectral broadening of the near infrared unseeded Stokes components and complexity of the SHG generators for different wavelengths.
Designs where a seed Stokes signal is used in the RFA [6–13] have been reported where a pulsed Yb fiber master oscillator power amplifier (MOPA) is serving as a Raman pump. A CW signal with a narrow linewidth at the desired Stokes wavelength is used as a seed. In these architectures, both the CW Stokes shifted signals and the pulsed pump propagate in the same direction in the RFA fiber. RFA-based pico- and nanosecond sources at 560 and 620 nm have been reported using this approach [8,9]. More recently, a watt-level nanosecond 743-nm pulsed source was demonstrated using a cascaded phosphosilicate RFA system with a 1240-nm pulsed RFA output used as a pump for a CW 1485-nm seed. [11].
In co-pumping RFAs, the pulsed pump peak power is significantly higher than the Raman amplified signal and, because the pump and amplified signal co-propagate, four-wave mixing (FWM) seeds a signal at the second Stokes wavelength, thus limiting the peak power of the first Stokes output, the conversion efficiency and the pulse-to-pulse stability [9–13]. It is worth pointing out that the repetition rate change of the pulsed pump in a co-pumping RFA results in the proportional change of the pump peak power and thus requires an additional control algorithm to keep the RFA gain at the desirable level. In this work, we use a counter-pumping RFA scheme with a CW pump and a pulsed signal. Noise transfer is substantially reduced in a counter-pumping RFA, where the pulsed signal averages the Raman gain over the RFA. As was shown in experiments [14–16], the noise transfer from the relatively broad pump spectrum to a counter-propagating single-frequency signal is limited to frequencies $\mathrm{\ \mathbin{\lower.3ex\hbox{$\buildrel< \over {\smash{\scriptstyle\sim}\vphantom{_x}}$}}\ }{\raise0.7ex\hbox{$c$} \!\mathord{/ {\vphantom {c L}}}\!\lower0.7ex\hbox{$L$}}$, where c is the speed of light and L is the length of the Raman amplifier fiber. Although it requires a longer RFA fiber, a counter-propagating RFA design eliminates FWM between pump and signal pulses. Moreover, it makes it possible to simultaneously amplify a pair of different wavelength pulses, separated by tuneable ns-scale time intervals, in one amplifier. Simple and efficient CW Raman pump lasers can provide CW RFA pump wavelengths well beyond the Yb amplifier range, thus enabling SHG pulsed output wavelengths up to 900 nm and with peak powers that can be significantly higher than the average power of the pump laser.
To mitigate the limiting effect of a variety of non-linear processes when attempting to scale to higher and higher pulse energies, the amplifier fiber length can be shortened while the pump power is increased. In the end, damage to the fiber is the ultimate limiting factor. In this regard, it is interesting to compare the pulse energy potential of co- and counter-pumped Raman amplifiers. In counter-propagating amplifiers, the output signal pulse peak power is several times higher than the average CW pump power (see paragraph 3.1 below) whereas, in co-pumped amplifiers, the peak power of the pulsed pump at the amplifier output is significantly larger than the amplified Stokes pulse peak power. If we assume a reasonable amplifier conversion efficiency of ∼ 30%, in the co-pumped scheme the maximum peak power (signal plus pulsed pump) travelling in the fiber is ∼ 3 times higher than in a counter-pumped amplifier providing the same Stokes output pulse power. This means that, respecting the same fiber damage limit, a counter-pumped amplifier is capable of producing amplified pulse powers several times higher than a co-pumped amplifier.
2. Laser description
Figure 1 shows the main optical components of the described laser source. A counter-propagating pump is provided by a polarization-maintaining Yb-doped double-clad fiber laser operating at 1116 nm with an output power of up to 25 W. The linewidth of the 1116-nm output gradually increased from 0.02 nm at threshold to 0.25 nm at the maximum output power. The Yb laser is pumped by two 30-W 915-nm multimode laser diodes. Pulsed signals at 1178 nm and 1310 nm are generated by two gain-switched DFB seed diodes. The electrical pump pulse for the seed diodes is deliberately shaped to have a long leading edge to avoid the typical gain-switched spike on the leading edge of the seed pulses. The pulse width, repetition rate, and temporal delay of both DFB seed diodes are electronically controlled by a clock generator board, allowing for a flexible range of adjustments. The repetition rates can vary from 40 to 100 MHz, and pulse durations from 700 to 1500 ps. The electronic control of the delay between the two seed pulses covers a range of 5 ns with an accuracy of approximately 10 ps. The pulsed seed signals are combined by an 1178/1310-nm PM fused wavelength division multiplexer (WDM). The pulses can be amplified individually or simultaneously using both the broad SiO2 (∼13.2 THz) and relatively narrow P2O5 (39.9 THz) Stokes shifts in the PM phosphosilicate fiber (IXF-PDF-5-125-PM from iXblue) which is about 200 m long. The length of the fiber was optimized experimentally to achieve a target of > 2 W output average power with reasonable RFA conversion efficiency and to keep the linewidth of the amplified signal less than the 100-pm acceptance bandwidth of the SHG. Following Raman amplification, the two pulse signals are separated from the counter-propagating CW pump using a single PM 1116/1178-nm WDM with losses at all three wavelengths below 0.3 dB. Residual Raman pump power travelling towards the seeds is removed by two PM WDMs in front of the two DFB seed diodes. The amplified signals are split by a high-power PM WDM and further converted into two second harmonic outputs at 589 nm and 655 nm using fiber-coupled modules based on bulk periodically-poled lithium niobate (PPLN) crystals 25 mm long. The SHG modules include focusing and collimating optics as well as a tap coupler and output monitoring photodetector. The focused input signal beam waist was ∼ 50 µm and was located in the middle of the crystal.
3. Experimental results
3.1 Single-wavelength operation
Figure 2(a) shows the amplified signal powers at 1178 nm and 1310 nm measured at a repetition rate of 80 MHz versus 1116-nm pump power, as well as the corresponding Raman gains.
Fig. 2. (a) Infrared 1178 and 1310 nm output signal powers and corresponding Raman gains vs. RFA pump power. (b) SHG powers of 589 nm and 655 nm vs corresponding signal powers at 1178 nm and 1310 nm. Repetition rate is 80 MHz.
The seed powers, measured prior to the RFA, are 3.2 mW and 3.7 mW with pulse widths of 782 ps and 820 ps, respectively. The maximum amplified average powers at 1178 nm and at 1310 nm reach 8 W and 5.6 W, for an 1116-nm pump power of 22 W. The maximum pulse peak power at 1178 nm was ≈ 128 W, which is six times higher than the CW 1116-nm pump power. The Raman power conversion efficiency is 37%. For the 1310-nm pulse power, the conversion efficiency reaches 25.5%. The maximum pulse peak power at 1310 nm is ≈ 86 W, representing a 3.9 times increase over the CW Raman pump power. The fact that the slope of the 1310-nm average power curve begins to level off above ∼18 W of 1116-nm pump power whereas the 1178-nm curve continues increasing with approximately the same slope is currently not understood and is being investigated. Peak powers were calculated based on the average power measurements, digital pulse shapes and the contrast ratio, i.e. the measured difference between pulse energy and the energy contained between pulses.
The amplified Stokes pulse peak power (PS) can be significantly higher than the pump CW average power (PP) since the Stokes pulses are travelling in the opposite direction to the pump and therefore, they can collect a substantial part of the energy “stored” in the length of the Raman amplifier fiber; PS can be estimated using the following expression:
where η is the average power conversion efficiency, F is the repetition rate and τ is the amplified pulse duration. This process is similar to the well-known Raman or Brillouin pulse compression technique, see for example [17,18].Notably, the pulse duration and spectral width of the amplified pulsed seeds at both wavelengths remain almost constant across varying output powers. The spectral width of the 1178-nm pulse was at 46 pm, and 33 pm for the 1310-nm pulse. These results emphasize the negligible impact of non-linear spectral broadening in the counter-pumping RFA, even when considering the relatively long length of the Raman fiber and the high peak power. The low nonlinear impact observed in the backward-pumped Raman amplifier further distinguishes it from co-pumping configurations.
Figure 2(a) illustrates the Raman gain of the 1178-nm and 1310-nm signals. The maximum gains reached more than 32 dB and showed saturation at both wavelengths beyond a Raman pump power of 10 W due to pump depletion. At 22 W of 1116-nm Raman pump power, a Raman gain of 34 dB was achieved at 8 W of the Stokes output power at 1178 nm, while at 1310 nm, it was 32 dB at 5.6 W output power. In Fig. 2(b), the SHG powers at 589 nm and 655 nm are presented as functions of the amplified power of the 1178-nm and 1310-nm signals. To ensure the safety of the periodically-poled lithium niobate (PPLN) crystals, the maximum output power from both SHGs was maintained at 3 W. We have found that, in our SHG module configuration, the PPLN crystals experience catastrophic damage at an average power of ∼ 4 W while hundreds of similar devices of ours in the field operating at 3 W have shown long-term stable operation. With this setup, the measured SHG conversion efficiencies were 50% for 589 nm and 55.7% for 655 nm. No signs of SHG saturation or a decline in conversion efficiency were observed since the linewidths at both wavelengths remained well below the spectral acceptance bandwidth of 0.1 nm for our SHG converters even at the highest powers. This distinct feature contrasts with co-pumping Raman amplifier configurations, where spectral broadening of the amplified seed signal limits SHG conversion efficiency [11,13].
Figure 3 presents the optical pulse characteristics of the frequency-doubled pulses at both wavelengths when the visible average output power is fixed at 2.5 W. Measurements were done using a sampling scope, Agilent Infinium DCA-J 86100 C, and fast photodetectors providing a temporal resolution better than 10 ps. Values of the pulse duration and corresponding spectral width for visible output powers from 1 to 2.5 W are shown in Fig. 4. The peak power for the 589-nm and 655-nm pulses is 43 W, and 48.8 W, respectively, and the pulse energies were > 31 nJ. Pulse energy increase is almost inversely proportional to the repetition rate in the range of 40-100 MHz.
Fig. 3. Sampling scope measurement of optical pulse duration at (a) 589 nm (b) 655 nm. Output power is 2.5 W and the repetition rate is 80 MHz.
Fig. 4. Pulse duration and spectral width measured at different output powers at (a) 589 nm (b) 655 nm. The repetition rate is 80 MHz.
In the case of the 589-nm pulse, both duration and spectral width exhibit slow, nearly linear broadening with increasing output power. However, for the 655-nm pulse, both remain practically constant across output powers ranging from 1 W to 2.5 W. These differences might be caused by a distinctive chirp of the corresponding pulses and different Raman gain shapes of the P2O5 and SiO2 Stokes components, which requires additional investigation. Also, it is worth mentioning that the linewidth of the 1310-nm second harmonic pulses was bigger than the linewidth of the corresponding near IR pulse. We do not have a reasonable explanation for this phenomenon. Most probably, the visible light broadening occurs inside the PPLN crystal.
When considering the application of STED microscopy, where resolution relies on the depletion pulse energy, and repetition rate defines the data acquisition speed [1,2], pulse and beam characteristics at different repetition rates are very important. The power of the 589-nm and 655-nm pulses was measured at repetition rates ranging from 40 to 100 MHz, and the results are presented in Fig. 5 as a function of 1116-nm Raman pump power. Measurements were carried out up to 1.5 W average power for the repetition rates of 40 and 50 MHz to ensure that the peak pulse power does not exceed the safe operating range for the PPLN (3 W at 80 MHz) and a maximum average power of 2 W was chosen for the higher repetition rates up to 100 MHz.
Fig. 5. (a) Power of 589-nm pulses at different rep. rates 40-100 MHz vs power of Yb laser at 1116 nm (b) Power of 655-nm pulses at different rep. rates 40-100 MHz vs power of Yb laser at 1116 nm
As is clear from Fig. 5, the average output power for a given 1116-nm pump power is approximately the same for all repetition rates, demonstrating that the energy per pulse is approximately inversely proportional to the repetition frequency. For the same RFA pump power, with a decrease in repetition rate, not only does the signal peak power go up, but so does the SHG efficiency, which makes the behavior more complex.
In the full range of powers and repetition rates, the M2 of the SHG output beams have been measured and found to be < 1.05 in both axes, confirming diffraction-limited beam quality.
3.2 Dual-wavelength and dual-pulse operation
A regime where both seed pulses are present in the RFA is illustrated in Fig. 6. Pulses shown in this figure were measured at an RFA pump power of 16 W and a repetition rate of 80 MHz. The amplitudes of the input signals have been controlled electronically using embedded firmware which makes it possible to control the relative powers of the output signals. Due to the considerable length of the RFA fiber, multiple pulse pairs (∼100) coexist simultaneously within the Raman amplifier. Consequently, the relative amplitudes of pulses at both wavelengths remain unaffected across the entire range of output powers, even when the time delay between pulses is changed. It was possible to change relative powers in the range of 1% to 99%.
Fig. 6. Relative powers of the two output pulses for different input signals settings (a) P(589)/P(655) = 99, (b) P(589)/P(655) = 1, (c) P(655)/P(589) = 99.
Figure 7(a) shows the amplified signal powers at 1178 and 1310 nm when both seed pulses with an average power of ∼ 3 mW each were launched into the amplifier. The delay between pulses was set at 4 ns. The individual infrared powers were measured after a splitting WDM and normalized with respect to the WDM losses. Figure 7(b) shows the corresponding second harmonic powers.
Fig. 7. Dual pulse operation. (a) Powers of the signals at 1178 nm and 1310 nm as well as the total power before splitting WDM when both are present in the amplifier. (b) Corresponding SHG powers. Repetition rate is 80 MHz
Comparing Figs. 7(a) and 2(a) one can see that the total output power in the case when both seed pulses are launched (with 6.9 mW total launched signal power at the two wavelengths) into the amplifier is slightly smaller than the power at 1178 nm when only a single 3.3-mW 1178-nm seed was launched because the effective Raman gain at 1310 nm is smaller than at 1178 nm. This smaller gain requires that the seed diode power at 1178 nm would have to be attenuated to set its launch power at ∼ 0.5 mW to achieve equal pulse amplitudes at both 655 and 589 nm, keeping in mind that, as the 1178-nm input power to the RFA is reduced, not only is the 1178-nm output power reduced but the 1310-nm output is increased since they both compete for the same pump power.
4. Conclusion
We have successfully demonstrated a two-color, dual-pulse compact laser system capable of producing sub-nanosecond visible outputs. This laser is based on a counter-pumped single-stage polarization-maintaining phosphosilicate fiber RFA, incorporating pulsed seed signals and a CW Raman pump. The RFA is a single-stage backward-pumped PM phosphosilicate-fiber Raman amplifier which features both SiO2 and P2O5 Stokes shifts. The pulsed seed signals are generated by gain-switched DFB lasers operating at 1178 nm and 1310 nm and are independently or simultaneously amplified in the RFA and frequency-doubled by two fiber-coupled PPLN-based SHG modules, providing wavelengths of 589 nm and 655 nm. The clock generator board firmware provides control of the repetition rate, pulse width, as well as the time delay between pulses at the two wavelengths. The maximum measured Raman gain was 34.0 dB for the 1178-nm and 31.5 dB for the 1310-nm pulses. Pulses were measured at a repetition rate of 80 MHz with the Raman amplifier operating in the saturation regime at both wavelengths. The spectral width of both output signals remains narrow (< 60 pm) in the full range of visible output powers up to 3 W, limited by the potential damage of the PPLN crystals. The pulse shape and duration exhibit very small changes in the full range of measured output powers and repetition rates, and the output beams are diffraction-limited with M2 < 1.05, over the entire range of output powers for both wavelengths. The presented set of parameters makes this developed laser a useful tool for various applications.
Acknowledgment
Portions of this work were presented at the Laser Congress in 2023, AM5A.6.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. B. R. Rankin and S. W. Hell, “STED microscopy with a MHz pulsed stimulated-Raman-scattering source,” Opt. Express 17(18), 15679–15684 (2009). [CrossRef]
2. S. W. Hell and J. Wichmann, “Breaking the diffraction resolution limit by stimulated-emission: stimulated-emission-depletion fluorescence microscopy,” Opt. Lett. 19(11), 780–782 (1994). [CrossRef]
3. T. H. Runcorn, F. G. Görlitz, R. T. Murray, et al., “Visible Raman-shifted fiber lasers for biophotonic applications,” IEEE J. Sel. Top. Quantum Electron. 24(3), 1–8 (2018). [CrossRef]
4. M. Castello, G. Tortarolo, I. C. Hernández, et al., “Gated-sted microscopy with subnanosecond pulsed fiber laser for reducing photobleaching,” Microsc. Res. Tech. 79(9), 785–791 (2016). [CrossRef]
5. D. Wildanger, E. Rittweger, L. Kastrup, et al., “STED microscopy with a supercontinuum laser source,” Opt. Express 16(13), 9614–9621 (2008). [CrossRef]
6. P. Dupriez, C. Farrell, M. Ibsen, et al., “1W average power at 589 nm from a frequency doubled pulsed Raman fiber MOPA system,” Proc. SPIE 6102, 61021G (2006). [CrossRef]
7. E. E. Rowen, G. Vashdi, J. Lasri, et al., “A scalable high-power yellow laser source based on frequency doubling of a combined Yb-Raman fiber amplifier,” Proc. SPIE 8961, 89611P (2014). [CrossRef]
8. T. H. Runcorn, R. T. Murray, E. J. R. Kelleher, et al., “Duration-tunable picosecond source at 560 nm with watt-level average power,” Opt. Lett. 40(13), 3085–3088 (2015). [CrossRef]
9. T. H. Runcorn, T. Legg, R. T. Murray, et al., “Fiber-integrated frequency-doubling of a picosecond Raman laser to 560 nm,” Opt. Express 23(12), 15728–15733 (2015). [CrossRef]
10. A. M. Chandran, T. H. Runcorn, R. T. Murray, et al., “Nanosecond pulsed 620 nm source by frequency-doubling a phosphosilicate Raman fiber amplifier,” Opt. Lett. 44(24), 6025–6028 (2019). [CrossRef]
11. A. M. Chandran, R. A. Battle, R. T. Murray, et al., “Watt-level 743 nm source by second-harmonic generation of a cascaded phosphosilicate Raman fiber amplifier,” Opt. Express 29(25), 41467–41474 (2021). [CrossRef]
12. M. Eibl, S. Karpf, H. Hakert, et al., “Pulse-to-pulse wavelength switching of a nanosecond fiber laser by four-wave mixing seeded stimulated Raman amplification,” Opt. Lett. 42(21), 4406–4409 (2017). [CrossRef]
13. T. H. Runcorn, R. T. Murray, and J. R. Taylor, “Highly efficient nanosecond 560 nm source by SHG of a combined Yb-Raman fiber amplifier,” Opt. Express 26(4), 4440–4447 (2018). [CrossRef]
14. D. Wei, V. Karpov, N. Guo, et al., “A 100-W 1178-nm continuous-wave single-frequency linearly-polarized Raman fiber amplifier,” Proc. SPIE 10703, 29 (2018). [CrossRef]
15. D. Wei, N. Guo, and W. R. L. Clements, “A mature 100-W 1178-nm single-frequency linearly polarized Raman fiber amplifier for laser guide star assisted optical ground station adaptive optics systems,” Proc. SPIE 12777, 45 (2023). [CrossRef]
16. V. Karpov, F. Eory, J. K. Kosenko, et al., “10-Watt 640-nm Single-Frequency Source,” Laser Congress (ASSL, LAC), AM5A.5, Optical Society of America (2023). [CrossRef]
17. J. R. Murray, J. Goldhar, D. Eimerl, et al., “Raman pulse compression of excimer lasers for application to laser fusion,” IEEE J. Quantum Electron. 15(5), 342–368 (1979). [CrossRef]
18. V. A. Gorbunov, S. B. Papernyĭ, V. F. Petrov, et al., “Time compression of pulses in the course of stimulated Brillouin scattering in gases,” Sov. J. Quantum Electron. 13(7), 900–905 (1983). [CrossRef]
