Chirped pulse amplification of a femtosecond Er-doped fiber laser mode-locked by a graphene saturable absorber

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 203.106.166.178

This content was downloaded on 20/10/2013 at 10:58

Please note that terms and conditions apply.

Chirped pulse amplification of a femtosecond Er-doped fiber laser mode-locked by a

graphene saturable absorber

View the table of contents for this issue, or go to the journal homepage for more

2013 Laser Phys. Lett. 10 035104

(http://iopscience.iop.org/1612-202X/10/3/035104)

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING LASER PHYSICS LETTERS

Laser Phys. Lett. 10 (2013) 035104 (6pp) doi:10.1088/1612-2011/10/3/035104

LETTER

Chirped pulse amplification of afemtosecond Er-doped fiber lasermode-locked by a graphene saturableabsorberG Sobon1, J Sotor1, I Pasternak2, W Strupinski2, K Krzempek1,P Kaczmarek1 and K M Abramski1

1 Laser & Fiber Electronics Group, Wroclaw University of Technology, Wybrzeze Wyspianskiego 27,50-370 Wroclaw, Poland2 Institute of Electronic Materials Technology, Wolczynska 133, 01-919 Warsaw, Poland

E-mail: [email protected]

Received 24 November 2012Accepted for publication 4 December 2012Published 31 January 2013Online at stacks.iop.org/LPL/10/035104

AbstractIn this work we demonstrate for the first time, to our knowledge, a chirped pulse amplification(CPA) setup utilizing a graphene mode-locked femtosecond fiber laser as a seed source. Thesystem consists of a mode-locked Er-fiber oscillator operating at 1560 nm wavelength, agrating-based pulse stretcher, two-stage amplifier and a grating compressor. The presentedsetup allows the amplification of the seed up to 1 W of average power (1000 timesamplification) with linearly polarized 810 fs pulses and 20 nJ pulse energy at a 55 MHzrepetition rate. The whole design is based on single-mode fibers, which allows one to maintainexcellent beam quality, with M2 less than 1.17.

(Some figures may appear in colour only in the online journal)

1. Introduction

Ultra-fast fiber lasers emitting sub-picosecond pulses havealready found many applications in various branches ofscience and industry, e.g. in laser micromachining [1],micro-surgery [2], precise optical metrology [3], opticalimaging [4] or terahertz wave generation [5]. Typically,the optical power emitted directly from mode-locked fiberoscillators is at the level of several milliwatts, which isinsufficient for most practical applications. Therefore, itis necessary to amplify the signal, while maintaining theultra-short pulse duration and excellent beam quality. Thechirped pulse amplification (CPA) technique, first proposedby Mourou and Strickland almost 30 years ago [6] is stilla key tool enabling efficient amplification of ultra-short

laser pulses emitted from mode-locked lasers. In principle,the ultra-short pulses are temporally broadened in a pulsestretcher, then amplified, and afterwards recompressed totheir initial duration. So far, this approach has been appliedmostly to oscillators operating in the 1 µm wavelengthrange. The usage of ytterbium-doped fiber amplifiers allowedthe achievement of sub-picosecond pulses with extremelyhigh peak powers [7–10]. For example, Eidam et al [11]demonstrated a CPA system based on Yb-doped photoniccrystal fiber (PCF) emitting 680 fs pulses with 830 Waverage output power and a 78 MHz repetition rate. Thesecond popular wavelength exploited by the ultra-fast fiberlaser community is 1.55 µm. Lasers based on Er-dopedfibers are capable of generating sub-100 fs pulses withMHz-level repetition rates [12–15]. Constructions based

11612-2011/13/035104+06$33.00 c© 2013 Astro Ltd Printed in the UK & the USA

Laser Phys. Lett. 10 (2013) 035104 G Sobon et al

on SESAMs or NPR are already available as commercialproducts. However, CPA systems operating at 1.55 µm, basedon Er- and Er/Yb-doped fibers, have not experienced asmuch progress as those utilizing Yb-doped fibers. In 1997Minelly et al [16] demonstrated an eye-safe CPA system basedon multi-mode Er- and Er/Yb-doped fibers emitting 700 fspulses at a 5 kHz repetition rate and 100 µJ pulse energy.Another setup based on large-mode area (LMA) Er/Yb-dopedfiber was demonstrated by Morin et al [17], delivering605 fs pulses with 1.5 µJ pulse energy. Pavlov et al [18]reported amplification of high-repetition rate 450 fs pulses insingle-mode fibers up to 10 W, but the output power of thesystem was limited to 2.5 W due to the low efficiency of thecompressor. There are several applications where high-power1.55 µm sources may outperform lasers operating at 1 µm,e.g. corneal surgery, solar cell processing or nonlinearparametric generation. Additionally, 1.55 µm radiation maybe easily converted by second harmonic generation (SHG)in nonlinear crystals to the near-infrared (780 nm) and serveas a seed for efficient supercontinuum generation, instead ofTi:sapphire lasers. Furthermore, systems operating at 1.55µmmay be more cost-effective, because of the wide availability ofrelatively cheap telecom components (couplers, isolators, etc).

All CPA setups presented so far were based onoscillators mode-locked by semiconductor saturable ab-sorbers (SESAMs) [9–11] or nonlinear polarization rotationmechanism (NPR) [18]. Recently, a new saturable absorbermaterial—graphene—has attracted much interest from thelaser community. Graphene, due to its unique opticalproperties, such as high modulation depth, low non-saturablelosses and broadband absorption, is considered to be an idealsaturable absorber for mode-locking of fiber lasers. Sincethe first demonstration of a graphene mode-locked fiber laserin 2009 [19], a number of journal papers on various lasersetups have appeared. There are several methods of obtaininguniform graphene layers suitable for mode-locking. The mostcommon include chemical vapor deposition (CVD), chemicalfunctionalization, and mechanical exfoliation from graphite.The first method allows one to grow mono- and multilayergraphene on metallic substrates, such as copper or nickel,which are afterwards transferred onto optical substrates(mirrors, glass plates, connectors, etc) [20–23]. Grapheneflakes might also be obtained by direct exfoliation of naturalgraphite by ultrasonification in various solvents [24–27].Recently, for this purpose also derivatives of graphene,namely graphene oxide (GO) and reduced graphene oxide(rGO), have been used [28–31]. Mechanical exfoliation (theso-called ‘scotch-tape method’) is the easiest method ofobtaining graphene flakes. However, mechanical exfoliationdoes not allow one to control the layer thickness and therepeatability of the process is very poor. Nevertheless severalinteresting results with the use of ‘scotch-tape graphene’ havealready been presented [32–35].

The promising results presented so far lead to aconclusion that graphene might outperform other mode-locking techniques used in fiber lasers. Nevertheless, theoutput power of a graphene mode-locked oscillator is alsotypically very low, at the level of several milliwatts. To date,

Figure 1. Graphene mode-locked seed source for the CPA system.

there have been no reports in the literature on amplification ofa laser mode-locked by graphene. In our work we demonstratefor the first time, to our knowledge, a chirped pulseamplification setup, seeded by a fiber oscillator mode-lockedby a graphene saturable absorber. The developed systemallows amplification of the seed signal up to 1 W averagepower with 810 fs pulse duration at a 55 MHz repetitionrate. The whole design is based on single-mode fibers, whichallows one to maintain excellent beam quality, with M2 lessthan 1.17. With the achieved parameters, the presented systemmight be used as an effective source for nonlinear frequencyconversion or supercontinuum generation in nonlinear fibers.

2. Experimental setup

2.1. Graphene mode-locked oscillator

The experimental setup of the CPA system is shown infigure 2. The setup of the graphene mode-locked laser usedas a seed source for the CPA system is shown in figure 1. Itconsists of a 1 m long erbium-doped fiber (with 30 dB m−1

absorption), a fiber isolator, 980/1550 single-mode WDMcoupler, in-line fiber polarization controller, 10% outputcoupler and a fused silica plate with deposited bi-layergraphene, inserted between two GRIN-lens collimators(200 mm working distance). The total length of the fibers inthe resonator is around 3.7 m and all of them have anomalousgroup delay dispersion (GDD).

The graphene layers used for mode-locking wereproduced by chemical vapor deposition on the surface of thincopper foils and transferred onto a fused silica plate in thesame way as presented by the authors previously [36]. Theglass window with deposited bi-layer graphene was placedinto the resonator on a three-axis stage, near to the Brewsterangle, in order to avoid parasitic back-reflections.

2.2. Chirped pulse amplification system

The pulses from the graphene oscillator are temporallybroadened in a Martinez-type stretcher, based on tworeflection gratings with 900 l mm−1 line density (SpectrogonPC0900), optimized for 45◦ angle of incidence. The telescope

2

Laser Phys. Lett. 10 (2013) 035104 G Sobon et al

Figure 2. Experimental setup of the CPA system seeded by a graphene mode-locked laser.

placed between the gratings consists of two bi-convex lenseswith 250 mm focal length. The applied separation of thegratings yields GDD of the stretcher at the level of 7.4 ps2.We estimate the stretched pulse duration to be approx. 50 ps.The beam is afterwards coupled to a single-mode fiberwith an aspheric lens. Next, the signal is amplified in atwo-stage amplifier. The first stage is based on a 50 cm long,highly Er-doped fiber (Liekki Er110) pumped by a 980 nmlaser diode (120 mW). The high-power stage is based on aEr/Yb-doped double-clad fiber (CorActive DCF-EY-7/128).It is backward pumped by a 975 nm laser diode. Both stagescontain fiber isolators to avoid parasitic back-reflections. Afteramplification, the pulses are compressed in a Treacy-typecompressor based on two gold-coated reflection gratingswith 950 l mm−1 line density (developed by ShimadzuCorporation). The grating separation is optimized to matchthe stretcher and compensate the anomalous dispersion of thefibers. The two in-line polarization controllers allow one toadjust the polarization to the maximum diffraction efficiencyof the gratings.

3. Experimental results

The performance of the system was observed using anoptical spectrum analyzer (Yokogawa AQ6370B), 7 GHzradio frequency (RF) spectrum analyzer (Agilent EXAN9010A), 13 GHz digital oscilloscope (Agilent Infiniium

DSO91304A) coupled with a 30 GHz photodetector (OptiLabPD-30), Thorlabs BP beam profiler and an interferometricautocorrelator.

3.1. Graphene-based oscillator

First, we characterized the seed laser before amplification.The graphene mode-locked oscillator provides 1 mW ofaverage output power at around 30 mW of pumping. Theoptical spectrum, shown in figure 3, is centered at 1560 nmand has a soliton-like shape with typical Kelly’s sidebands,which is a consequence of the all-anomalous cavity design.The full width at half maximum (FWHM) of the spectrumis 8 nm. No signs of parasitic continuous wave (CW) lasingwere observed. Figure 4 illustrates the RF spectrum recordedwith 520 Hz resolution bandwidth (RBW) and 1 MHz span.The laser operates at a 54.9 MHz repetition rate, whichcorresponds to a ∼3.7 m long resonator. The signal to noiseratio (SNR) in the RF spectrum is higher than 60 dB, whichis consistent with previous reports on graphene mode-lockedlasers. The autocorrelation trace of the output pulses is shownin figure 5. Assuming a sech2 shape, the pulse width is equalto 430 fs. No signs of pulse pair generation were observed.With 8 nm FWHM bandwidth (985 GHz), the time-bandwidthproduct (TBP) is equal to 0.423, which is close to thetransform limit for sech2 pulses (0.315). The calculated pulseenergy is lower than 20 pJ.

3

Laser Phys. Lett. 10 (2013) 035104 G Sobon et al

Figure 3. Optical spectrum of the seed laser recorded with 0.02 nmresolution.

Figure 4. RF spectrum of the seed laser recorded with 520 HzRBW. Inset: full RF spectrum with 7 GHz span.

Figure 5. Autocorrelation trace of the 430 fs pulse from the seedoscillator.

3.2. Chirped pulse amplification of the seed laser

As mentioned before, the average output power of the seed is1 mW. Due to the 35% coupling efficiency from the stretcher

Figure 6. Average output power of the system measured before(blue trace) and after compression (red trace).

Figure 7. Optical spectrum of the CPA output beam.

to single-mode fiber, the average power after stretching is atthe level of 350 µW, so the signal needs to be pre-amplifiedbefore entering the double-clad stage. The average power atthe erbium-doped fiber amplifier output is 5 mW.

Figure 6 shows the average output power of the CPAsystem versus launched pump power measured before (bluetrace) and after compression (red trace). The maximum outputpower before compression is at the level of 1.24 W. Themeasured efficiency of the compressor is at the level of 81%,which yields the total optical to optical efficiency of thesystem at the level of 25.6%. The compression efficiencyis over three times higher in comparison to the resultspresented by Pavlov et al [18]. The maximum output powerafter compression is 1.02 W, which means that the signalis amplified over 1000 times (total net gain factor equal to30 dB). The calculated pulse energy is 20 nJ.

The optical spectrum of the compressed beam is depictedin figure 7. It can be seen that after amplification the spectrumnarrowed and the FWHM bandwidth is at the level of 7 nm.

The autocorrelation of the pulse after recompressionis plotted in figure 8. Assuming a sech2 pulse shape, thepulse duration is at the level of 810 fs. With 7 nm opticalbandwidth (862 GHz), the time-bandwidth product (TBP) is

4

Laser Phys. Lett. 10 (2013) 035104 G Sobon et al

Figure 8. Autocorrelation of the compressed pulses (810 fs pulseduration).

Figure 9. Recorded output pulse train.

equal to 0.698 (assuming sech2 pulse shape). It means that thepulses are slightly chirped, probably due to uncompensatedhigh-order dispersion. With 20 nJ pulse energy, the pulse peakpower reaches 25 kW.

The oscilloscope pulse train from the CPA output isshown in figure 9. The pulses are equally separated by around18 ns, corresponding to a 55 MHz repetition frequency. Nosigns of double-pulsing or pulse breaking were observed.

In order to verify the quality of the amplified andrecompressed beam, we have performed a M2 measurementusing a beam profiler. The results are plotted in figure 10.The output beam is slightly elliptical, which might becaused by slight compressor misalignment. Nevertheless, themeasurement confirmed the excellent spatial properties of thebeam, with M2 value lower than 1.17.

4. Summary

In conclusion, we have demonstrated a high-repetition rate,high-power and eye-safe chirped pulse amplification schemeusing a graphene mode-locked fiber laser as a seed source.The graphene-based CPA system delivered 810 fs solitonpulses with 55 MHz repetition frequency and 1 W of output

Figure 10. M2 measurement results.

power. The obtained pulse energy was at the level of 20 nJand almost 25 kW peak power. The designed compressor,based on reflection gratings, allowed the achievement of81% compression efficiency. In addition, the whole system isdesigned using only single-mode fibers, which allows one tomaintain excellent beam quality at the output. The presentedsystem might be used as an effective source for nonlinearfrequency conversion or supercontinuum generation, e.g. inphotonic crystal fibers.

Acknowledgments

The work presented in this paper was supported by theNational Centre for Research and Development (NCBiR,Poland) under a research and development project entitled‘Ultra-fast graphene-based fiber lasers’. The authors wouldlike to acknowledge the kind support of ShimadzuCorporation (Japan), in particular Tatsuyoshi Fujiwara forproviding the pulse compression gratings. The researchfellowship of one of the authors (GS) is co-financed by theEuropean Union as part of the European Social Fund.

References

[1] Gattass R R and Mazur E 2008 Nature Photon. 2 219–25[2] Mamalis N 2011 J. Cataract Refract. Surg. 37 1177–8[3] Balling P, Kren P, Masika P and van den Berg S A 2009 Opt.

Express 17 9300–13[4] Tang S, Liu J, Krasieva T B, Chen Z and Tromberg B J 2009

J. Biomed. Opt. 14 030508[5] Stehr D, Morris C M, Schmidt C and Sherwin M S 2010 Opt.

Lett. 35 3799–801[6] Strickland D and Mourou G 1985 Opt. Commun. 56 219–21[7] Sumimura K, Yoshida H, Fujita H and Nakatsuka M 2007

Laser Phys. 17 339[8] Shah L, Liu Z, Hartl I, Imeshev G, Cho G and

Fermann M 2005 Opt. Express 13 4717[9] Roser F, Eidam T, Rothhardt J, Schmidt O, Schimpf D N,

Limpert J and Tunnermann A 2007 Opt. Lett. 32 3495[10] Eidam T, Rothhardt J, Stutzki F, Jansen F, Hadrich S,

Carstens H, Jauregui C, Limpert J and Tunnermann A 2011Opt. Express 19 255

5

Laser Phys. Lett. 10 (2013) 035104 G Sobon et al

[11] Eidam T, Hanf S, Seise E, Andersen T V, Gabler T, Wirth C,Schreiber T, Limpert J and Tunnermann A 2010 Opt. Lett.35 94

[12] Ilday F, Chen J and Kartner F 2005 Opt. Express 13 2716–21[13] Ma D, Cai Y, Zhou C, Zong W, Chen L and Zhang Z 2010

Opt. Lett. 35 2858–60[14] Nikodem M and Abramski K 2010 Opt. Commun. 283 109–12[15] Deng D, Zhan L, Gu Z, Gu Y and Xia Y 2009 Opt. Express

17 4284–8[16] Minelly J D, Galvanauskas A, Harter D, Caplen J E and

Dong L 1997 CLEO’97: Conf. on Lasers andElectro-Optics (Baltimore, USA, 18–23 May); IEEE Conf.Proc. 11 475–6

[17] Morin F, Druon F, Hanna M and Georges P 2009 Opt. Lett.34 1991

[18] Pavlov I, Ilbey E, Dulgergil E, Bayri A and Ilday F O 2012Opt. Express 20 9471

[19] Bao Q L, Zhang H, Wang Y, Ni Z H, Shen Z X, Loh K P andTang D Y 2009 Adv. Funct. Mater. 19 3077–83

[20] Zhang H, Tang D Y, Zhao L M, Bao Q L and Loh K P 2009Opt. Express 17 17630–5

[21] Zhang H, Tang D Y, Zhao L M, Bao Q L, Loh K P, Lin B andTjin S C 2010 Laser Phys. Lett. 7 591–6

[22] Zhang H, Tang D, Knize R J, Zhao L, Bao Q andLoh K P 2010 Appl. Phys. Lett. 96 111112

[23] Huang P L, Lin S, Yeh C, Kuo H, Huang S, Lin G, Li L,Su C and Cheng W 2012 Opt. Express 20 2460–5

[24] Popa D, Sun Z, Torrisi F, Hasan T, Wang F andFerrari A C 2010 Appl. Phys. Lett. 97 203106

[25] Sun Z, Hasan T, Torrisi F, Popa D, Privitera G, Wang F,Bonaccorso F, Basko D M and Ferrari A C 2010 ACS Nano4 803–10

[26] Sun Z, Popa D, Hasan T, Torrisi F, Wang F, Kelleher E J R,Travers J C, Nicolosi V and Ferrari A C 2010 Nano Res.3 653–60

[27] Cunning B V, Brown C L and Kielpinski D 2011 Appl. Phys.Lett. 99 261109

[28] Xu J, Liu J, Wu S, Yang Q and Wang P 2012 Opt. Express20 15474–80

[29] Xu J, Wu S, Li H, Liu J, Sun R, Tan F, Yang Q-H andWang P 2012 Opt. Express 20 23653–8

[30] Sobon G, Sotor J, Jagiello J, Kozinski R, Zdrojek M,Holdynski M, Paletko P, Boguslawski J, Lipinska L andAbramski K M 2012 Opt. Express 20 19463–73

[31] Choi S Y, Cho D K, Song Y-W, Oh K, Kim K,Rotermund F and Yeom D-I 2012 Opt. Express 20 5652–7

[32] Sobon G, Sotor J and Abramski K M 2012 Laser Phys. Lett.9 581–6

[33] Sotor J, Sobon G and Abramski K M 2012 Opt. Lett.37 2166–8

[34] Martinez A, Fuse K and Yamashita S 2011 Appl. Phys. Lett.99 121107

[35] Sotor J, Sobon G, Krzempek K and Abramski K M 2012 Opt.Commun. 285 3174–8

[36] Sobon G, Sotor J, Pasternak I, Grodecki K, Paletko P,Strupinski W, Jankiewicz Z and Abramski K M 2012J. Light. Technol. 30 2770–5

6