Maximizing the mode instability threshold of a fiber amplifier ∗ ArleeV.Smith andJesseJ.Smith AS-Photonics,LLC,8500MenaulBlvd.NE,SuiteB335,Albuquerque,NMUSA87112 ∗[email protected] 3 1 0 Abstract: We show by detailed numerical modeling that stimulated 2 thermalRayleighscatteringcanaccountforthe modalinstabilityobserved n in high power fiber amplifiers. Our model illustrates how the instability a J threshold power can be maximized by eliminating amplitude and phase modulation of the signal seed and the pump, and by careful launch of the 5 1 signal seed. We also illustrate the influence of photodarkening and mode specificlossonthethreshold. ] s © 2013 OpticalSocietyofAmerica c OCIScodes:(060.2320)Fiberopticsamplifiersandoscillators;(060.4370)Nonlinearoptics, i t fibers;(140.6810)Thermaleffects;(190.2640)Stimulatedscattering,modulation,etc p o s. Referencesandlinks c 1. T. Eidam, C. Wirth, C. Jauregui, F. Stutzki, F. Jansen, H.-J. Otto, O. Schmidt, T. Schreiber, i s J. Limpert, and A. Tu¨nnermann, “Experimental observations of the threshold-like on- y set of mode instabilities in high power fiber amplifiers,” Opt. Express 19, 13218 (2011). h http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-19-14-13218. p 2. H.-J.Otto, F. Stutzki, F.Jansen, T.Eidam, C. Jauregui, J.Limpert, and A. Tu¨nnermann, “Temporal dynam- [ ics of mode instabilities in high-power fiber lasers and amplifiers,” Opt. Express 20, 15710–15722 (2012). http://www.opticsexpress.org/abstract.cfm?URI=oe-20-14-15710. 1 3. N.Haarlammert,O.deVries,A.Liem,A.Kliner,T.Peschel,T.Schreiber,R.Eberhardt,andA.Tu¨nnermann, v “Buildupanddecayofmodeinstabilityinahighpowerfiberamplifier,”Opt.Express20,13274–13283(2012). 9 http://www.opticsexpress.org/abstract.cfm?URI=oe-20-12-13274. 8 4. F. Stutzki, H.-J. Otto, F. Jansen, C. Gaida, C. Jauregui, J. Limpert, A. Tu¨nnermann. “High-speed modal 4 decomposition of mode instabilities in high-power fiber lasers”, Opt. Letters 36, 4572–4574 (2011). 3 http://www.opticsinfobase.org/ol/abstract.cfm?uri=ol-36-23-4572. 5. A.V.SmithandJ.J.Smith,“Modeinstabilityinhighpowerfiberamplifiers,”Opt.Express19,10180–10192 . 1 (2011).http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-19-11-10180. 0 6. A. V. Smith and J. J. Smith, “Influence of pump and seed modulation on the 3 mode instability thresholds of fiber amplifiers,” Opt. Express 20, 24545–24558 (2012). 1 http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-20-22-24545. : 7. K. R. Hansen, T. T. Alkeskjold, J. Broeng, and J. Lægsgaard, “Thermally induced v mode coupling in rare-earth doped fiber amplifiers,” Opt. Letters 37, 2382–2384 (2012). Xi http://www.opticsinfobase.org/ol/abstract.cfm?uri=ol-37-12-2382. 8. A.V.SmithandJ.J.Smith,“Asteady-periodicmethodformodelingmodeinstabilityinfiberamplifiers,”ArXiv r e-prints(2013).http://arxiv.org/abs/1301.1296. a 9. K.D.Cole,“Steady-periodic Green’sfunctions andthermal-measurementapplications inrectangular coordi- nates,”JournalofHeatTransfer128,DOI:10.1115/1.2194040(2006). 10. B. Ward, C. Robin, and I. Dajani, “Origin of thermal modal instabilities in large mode area fiber amplifiers,” Opt. Express 20, 11407–11422 (2012). http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-20-10-11407. 11. M.Laurila,T.T.Alkeskjold,M.M.Jorgensen,S.R.Petersen,J.Broeng,andJ.Lægsgaard,“Highlyefficienthigh powersingle-modefiberamplifierutilizingthedistributedmodefilteringbandgaprodfiber,”in“Proc.SPIE,”, vol.8237,E.C.HoneaandS.T.Hendow,eds.(2012),vol.8237,p.823710. 12. M.Laurila,M.M.Jorgensen,K.R.Hansen,T.T.Alkeskjold,J.Broeng,andJ.Lægsgaard,“Distributedmode filteringrodfiberamplifierdelivering292wwithimprovedmodestability,”Opt.Express20,5742–5753(2012). http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-20-5-5742. 13. H.Uranus,H.Hoekstra,andE.vanGroesen,“Modesofanendlesslysingle-modephotoniccrystalfiber:afinite elementinvestigation,”Proc.11thIEEESymp.Commun.andVehicularTechnol.(SCVT)(2004). 14. Z.JiangandJ.R.Marciante,“Mode-areascalingofhelical-core,dual-cladfiberlasersandamplifiersusingan improvedbend-lossmodel,”J.Opt.Soc.Am.B23,2051–2058(2006). 15. B. G. Ward, “Bend performance-enhanced photonic crystal fibers with anisotropic numerical aperture,” Opt. Express 16, 8532–8548 (2008). http://www.opticsexpress.org/abstract.cfm?URI=oe-16-12-8532. 16. F. Stutzki, F. Jansen, C. Jauregui, J. Limpert, and A. Tu¨nnermann, “Non-hexagonal large- pitch fibers for enhanced mode discrimination,” Opt. Express 19, 12081–12086 (2011). http://www.opticsexpress.org/abstract.cfm?URI=oe-19-13-12081. 1. INTRODUCTION Aboveasharppowerthresholdtheoutputbeamfromamultimodefiberamplifierisradically altered[1,2,3,4].IfthefundamentalLP modeisinjectedatthefiberinput,slightlyabovethe 01 thresholdtheoutputislargelyinLP .Thismodalinstabilityiscausedbyastimulatedthermal 11 Rayleigh scattering (STRS) process[5, 6, 7]. Quantum defect heating creates a temperature grating,and consequentlya refractiveindexgrating,that is responsibleforcouplingbetween the two fibermodes.Numericalmodelsof the STRS processshow that,like otherstimulated Rayleigh scattering processes, the gain has a dispersive shape[6, 7] similar to that shown in Fig. 1. A frequencyoffsetbetweenthe two coupledmodesis necessaryto producethe phase shiftessentialformodecoupling[5, 7],andpowertransfer istowardthemodewith thelower frequency. 80 40 m] B/ d ain [ 0 G al ot T −40 −80 −100 −75 −50 −25 0 25 50 75 100 ∆ν [kHz] Fig.1.SamplegainofmodeLP versusitsfrequencyoffsetfrommodeLP .Thegain 11 01 includesbothlasergainandmodecouplinggainduetoSTRS.Lasergainaccountsforthe upwardshiftofapproximately15dB/m. Thepurposeofthispaperistoprovidesomeguidanceonhowthemodeinstabilitythreshold canbeheldatthehighestpossiblepower.Therearetwoapproachestothis:optimizingthefiber design,oroptimizingtheoperatingconditions.Thispaperisprimarilyconcernedwithfinding theoptimumoperatingconditions. 2. NUMERICALMODEL WereportedelsewhereourdetailednumericalmodeloftheSTRSprocess[5,8]basedontheas- sumptionthatthepowersofthetwocoupledmodesandthetemperaturevaryonlyperiodically. Thissteady-periodicassumptionallowstheuseofasteady-periodicGreen’sfunctiontocom- pute the time-dependenttemperature profiles responsible for mode coupling[9]. The Green’s functiontemperaturesolveriscombinedwithasplit-stepfast-Fouriertransformbeampropaga- tionmethodtomodelmodecoupling.Thisapproachhastheadvantagethatallfibermodesare simultaneouslyincluded,andthepopulationinversionprofilesarerealistic.Thehighlynumer- icalnatureofthe modelmakesit relativelyeasy to addother physicaleffects.Thismodelal- lowsustopredictthefrequencyshiftbetweenLP andLP whichproducesmaximummode 01 11 coupling gain. Our predicted frequenciesare in good agreement with observed signal output modulationfrequencies[2,10].Anotherkeyfeatureofthemodelistheinstabilitythresholdis well-defined,with a sharpchangeof LP contentwithincreasingpumppower,asillustrated 11 inFig.2. Inthefollowingsectionswewillexaminehowtheinstabilitythresholdisimpactedbypump andsignalamplitudemodulation,photodarkening,andmodespecificloss.Wealsosuggestthe possibilityofrestoringthethresholdbycounteringthepumpmodulationwithsignalmodula- tion.Asabaselinecase,withoutanyoftheseinfluences,we modeltheLPF45fiberamplifier described by Otto et al.[2], using the parameters listed in Table 1. This is a photonic crystal fiber,butwesimulateitusingaco-pumped,stepindexfiberwiththecoresizeandnumerical apertureadjusted to give an LP modeclose to the reportedsize. For the baseline amplifier, 01 weuseasignalseedpowerof10−16 WinthefrequencyshiftedLP modetosimulatequan- 11 tumnoiseseeding.Thisiscalculatedusinganoisespectralpowerdensityofhn multipliedby the amplified bandwidth of approximately 500 Hz. The baseline amplifier has a threshold at a normalizedpump input powerof 1.0. We will compareother thresholdsto this normalized value.Thresholdisdefinedhereasthepumpinputpoweratwhich5%oftheoutputsignalisin thehigherordermodeLP .Sincethispaperisonlyconcernedwithqualitativebehavior,and 11 weexpectsimilarbehaviorsforco-andcounter-pumpedfibers,wemodelonlytheco-pumped case. Theexperimentalmodecouplingperformanceof the LPF45 fiberhasbeenreported[2], permittingcomparisonsofsomeofourpredictionswithlaboratoryresults. Table1.LPF45fiberamplifierparametersusedinthenumericalmodel. d 81m m d 63m m core dope d 255m m P 10W clad 01 l 976nm l 1060nm pump signal s a 2.47×10−24m2 s e 2.44×10−24m2 pump pump s a 5.8×10−27m2 s e 2.71×10−25m2 sig sig n 1.45015 n 1.45 core clad N 3.0×1025m−3 L 1.2m Yb 3. MODULATEDSIGNALSEED While seeding the LP mode with quantum noise is unavoidable, additional sources of fre- 11 quencyshiftedLP lightcandramaticallyreducethethreshold.Forexample,iftheinputsig- 11 nallightis amplitudemodulated,frequencycomponentswithin the STRS amplificationband canbepopulated.Ifsomeofthisfrequencyshiftedseedlightisaccidentallyinjectedintomode LP , it will seed the amplified mode.Amongotherpossibilities, such amplitudemodulation 11 willbepresentiftheseedisanASEsource.Itisalsopresentatsomelevelinanyothersource oftheseedlight. 1 0.8 nt e nt o c 0.6 al d o m ge 0.4 a LP er 01 Av LP 11 0.2 LP 12 LP 02 0 0 0.5 1 1.5 Normalized Input Pump Power [arb.] Fig. 2. Fraction of the signal power in various modes versus the input pump power (in normalizedunits).Here,thefrequencyshiftedLP isseededbyquantumnoisesimulated 11 by10−16W.Thesharpmodeinstabilitythresholdoccursnearthenormalizedpumppower of1.0. We modelan amplitudemodulatedseed with 9.9W injectedintoLP and 0.1W injected 01 intoLP .Thefrequencyisadjustedformaximummodecouplinggain.Figure3demonstrates 11 theinfluenceofincreasingdepthofsignalmodulationonthethresholdwherethebaselinecase correspondstozeromodulation.Asthefigureshows,thethresholdisstronglyreducedevenfor smalllevelsofmodulation. 1 el.] 0.8 er [r w o 0.6 p p m u d p 0.4 ol h es hr 0.2 T 0 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01 Signal Modulation (Peak−to−Peak) Fig.3.Normalizedpumppoweratmodeinstabilitythresholdvs.magnitudeofsignalam- plitudemodulation.Themagnitudeisdefinedasthepeak-to-peakpowervariationnormal- ized totheaveragepower. Thresholdisdefined as5%of thesignal output inLP .The 11 leftmostpointcorrespondstothebaselinecase. The importantthing is the strength of the seed in the frequency-shiftedlight in LP . The 11 threshold reductionis similar if onlythe light in LP is amplitudemodulatedratherthen all 11 theseedlightaswasthecasefortheresultspresentedinFig.3.Thismeanssmallmechanical vibrations that affect the amount of light accidentally injected into LP could produce the 11 amplitude modulation in the amplified band. It is also unimportantwhether the seed light is amplitudemodulatedorphasemodulated.Theyproducesimilarthresholdreductions. 4. MODULATEDPUMP Evenif thesignalseedisunmodulated,pumpmodulationproducesalmostthesameeffectas signalmodulation.A modulatedpumpquicklyimpressesamplitudemodulationonthesignal light in LP , leading to population of the frequency shifted signal sideband. We model this 11 effectbyseedingLP with9.9WandLP with0.1W,bothunmodulated.We modulatethe 01 11 inputpumpbyvaryingamountsatthefrequencyofmaximummodecouplinggain.Theresults are shown in Fig. 4 with the baseline case correspondingto zero modulation. Again a small amountofmodulationleadstoastrongreductioninthreshold. 1 0.9 el.] 0.8 er [r 0.7 w o 0.6 p mp 0.5 u p d 0.4 ol esh 0.3 hr T 0.2 0.1 0 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01 Pump modulation (Peak−to−Peak) Fig.4.Normalizedpumppoweratmodeinstabilitythresholdvs.magnitudeofpumpampli- tudemodulation.Themagnitudeisdefinedasthepeak-to-peakpowervariationnormalized totheaveragepower.Thresholdisdefinedas5%ofthesignaloutputinLP . 11 5. COUNTERMODULATIONOFSEEDANDPUMP The similarity of Figs. 3 and 4 suggests that perhapsa small pump modulationcan be coun- teracted by an appropriately adjusted signal modulation. We tested this idea using a pump amplitudemodulationof 0.001,and show in Fig. 5 the results whenthe signalmodulationis optimizedinamplitudeandphase.Wecouldnotfullyrestorethethreshold,butdidimproveit by15%. 6. PHOTODARKENING Photodarkeningcanalsostronglyreducetheinstabilitythreshold.Wemodelthisusingavery simplephotodarkeningmodel.Weassumeauniformlinearabsorptionofthesignallightacross the full doped region of the fiber. A more realistic model would account for the transverse shape of the absorptionat each z location,butourmodelshouldgive a reasonableindication oftheinfluenceofphotodarkening.Italsoshowsthe effectothersourcesoflinearabsorption mighthave.We assumetheabsorbedpowerisfullyandinstantaneouslyconvertedtoheat,so theheatingprofilematchestheirradianceprofileoverthedopedregion. Figure6showsinthebluetrace(circles)howthethresholdfallswithincreasingsignalab- sorption.Itisimportanttorealizethatattheabsorptions consideredheretheefficiencyofthe amplifierisonlyslightlyreduced,asshowninthegreencurve(squares).Areductionof60% 1 el.] 0.8 er [r w o 0.6 p p m u d p 0.4 ol h es hr 0.2 T 0 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01 Pump Modulation (Peak−to−Peak) Fig.5.ThisisthesameasFig.4withtheadditionofthestarindicatingtheimprovement byoptimizedcountermodulationoftheinputsignal. in threshold correspondsto a reduction in efficiency of only 7%. One importantsignature of actual photodarkeningis that it turns on graduallyin the presence of pumping,and it can be reversedbyopticalbleachingorthermalbleaching[11,12]. 1 95 0.9 94 el.] 0.8 93 mp power [r 000...567 999012ciency [%] eshold pu 00..34 8889Slope effi Thr 0.2 87 0.1 86 0 85 00 00..0055 00..11 00..1155 00..22 00..2255 00..33 00..3355 00..44 00..4455 00..55 Linear absorption [dB/m] Fig.6.Threshold(circles)andefficiency(squares)dependenceonlinearabsorption.Here, LP isseededwith10WandLP isseededwith10−16Watthefrequencyofmaximum 01 11 gain.Efficiencyisdefinedastheincreaseinsignalpowerdividedbythereductioninpump power. 7. MODESPECIFICLOSS IfthelossofLP canbemadelargewithoutsignificantlyaffectingLP ,itshouldbepossible 11 01 to increase the instability threshold while maintaining overall efficiency. Such loss for LP 11 mightbe created by bend loss orby cleverfiber engineering.We introducean LP loss that 11 isconstantalongthefulllengthofthefiber,withzerolossto LP , withtheresultsshownin 01 Fig. 7. We seed LP with 10 W and the frequency shifted LP with 10−16 W so the zero 01 11 losspointisagainthebaselinecase.Itisclearfromthefigurethatlargemodespecificlosses arerequiredtosignificantlyincreasethethreshold.Thisisnotsurprisingbecausethebaseline modecouplinggainislargerthan170dBatthreshold,andthe gainisapproximatelylinearin pumppower. Severalstudies ofmodediscriminationby fiberdesignorbendinghave suggestedthe pos- siblityoflargemodespecificloss[13,14,15,16].Unfortunately,itiscommonthatreportsof experimentalobservationsof modeinstability thresholds neglectto includemention of mode specificlosses. 1.5 el.] er [r 1 w o p p m u p d ol 0.5 h es hr T 0 0 10 20 30 40 50 60 70 LP −specific loss (dB/m) 11 Fig. 7. Influence of mode LP losson therelative threshold pump power ina quantum 11 seededamplifier. 8. CONCLUSION Careful characterization of fiber amplifiers will be required to maximize the mode instabil- ity threshold. All pump lasers and seed sources have some degree of spectral and amplitude modulation,andweshowedthatevensmallmodulationscanstronglyreducethethreshold.It shouldbeclearthatthepumpandseedinputsmustbeextremelywellcontrolledtomaximize thresholds,andthattheircharacterizationis an essentialpartof meaningfulreportson modal instability.Additionally,preciseaccountingofthesignalandpumppowersmaybenecessary inordertodetectanysmalllinearlossessincetheyalsostronglyreducethresholds.Further,it isclearthatattemptstoraisethemodeinstabilitythresholdsignificantlybyengineeringlosses forthehigherordermodeswillrequirelargelosses.Doublingthethresholdfromthebaseline requiresabout170dBoflossforLP . 11