joona€ koponen,€ marko€ laurila,€ and€ mircea€hotoleanu.€...

Joona Koponen, Marko Laurila, and Mircea Hotoleanu. 2008. Inversion behavior incore and claddingpumped Ybdoped fiber photodarkening measurements. AppliedOptics, volume 47, number 25, pages 45224528.

© 2008 Optical Society of America (OSA)

Reprinted with permission.

Inversion behavior in core- and cladding-pumpedYb-doped fiber photodarkening measurements

Joona Koponen,* Marko Laurila, and Mircea HotoleanunLIGHT Corporation, Sorronrinne 9, FIN-08500 Lohja, Finland

*Corresponding author: [email protected]

Received 17 June 2008; accepted 17 July 2008;posted 21 July 2008 (Doc. ID 97591); published 22 August 2008

Yb-doped fibers are widely used in applications requiring high average output powers and high powerpulse amplification. Photodarkening of the Yb-doped silicate glass core potentially limits the lifetime orefficiency of such fiber devices. In many studies of photodarkening, two principal methods of controllablyinducing an inversion are used, namely, cladding pumping and core pumping of the sample. We presentsimulation results describing the key differences in the inversion profiles of samples of different physicalparameters in these two cases, and we discuss the problems and possibilities that arise in benchmarkingfibers of various physical parameters. Based on the simulation and experimental work, we propose guide-lines for photodarkening benchmarking measurements and show examples of measurements made with-in and outside of the guidelines. © 2008 Optical Society of America

OCIS codes: 140.3380, 140.3510, 140.5680, 060.2270, 060.2290.

1. Introduction

Ytterbium (Yb)-doped fiber lasers and amplifiershave become an attractive option for many applica-tions ranging from materials processing to frequencyconversion to visible or ultraviolet wavelengths. Theinterest in silica based glass fibers doped with Ybarises from the low quantum defect that enables highefficiency, and a wide body of literature exists on theuse of Yb fibers in amplifiers and lasers. The highoptical efficiency in conjunction with double cladding(DC) fiber structure to convert large quantities ofpump radiation to high brightness signal havepushed Yb-doped sources from research demonstra-tions to commercially available laser devices [1–7].On the other hand, higher concentrations or pumpintensities have in some cases been known to inducedeleterious effects, such as photodarkening (PD),that manifests as a broad spectral attenuation inthe Yb-doped core of the fiber [8]. In the work towardfiber with increased PD resistance, the PD as a ma-terial property needs further study, and preferablyquantitative or qualitative measurement methods.

In our earlier work we have introduced two methodsof comparing, benchmarking, and studying of Yb-doped fibers, namely, the core pumping method [8]and the cladding pumping method [9]. This work de-scribes in detail the differences of the two methodsand discusses their limitations and possibilities incharacterizing fibers for PD.

Although good progress has been made in furtherunderstanding the PD process in silica glass, the me-chanism(s) involved in PD are under discussion, andseveral hypotheses have been suggested for the prin-cipal underlying process. In a recent experiment, Yooet al. [10] induced PD in both Yb-doped and Ge-dopedglass, and observed a strong absorption band at220nm that was attributed to oxygen deficiency cen-ters (for example, Yb—Yb or Yb—Si bonds) that couldact as PD precursors, leading to color centers. A clus-tering related process was also proposed by GuzmanChávez et al. [11], in which Yb3þ − Yb3þ pairs (or evenmore complicated clusters) would lead to the forma-tion of color centers or other structural defects. In adifferent set of experiments, Engholm et al. [12] ob-served an absorption band at 230nm. However, theabsorption was attributed to a charge transfer bandinvolving the Yb ion, implying the possibility of

0003-6935/08/254522-07$15.00/0© 2008 Optical Society of America

4522 APPLIED OPTICS / Vol. 47, No. 25 / 1 September 2008

inducing a change of Yb3þ to Yb2þ and vice versa,creating free electrons or holes that could act as pre-cursors for PD. Several different methods of fully orpartially reversing the PD have been demonstrated.The different methods included temperature anneal-ing [13], exposure to UVand visible light [11,14], andoxygen loading [10]. More recently, PD has also beenshown to bleach by the pump power itself [15].

2. Photodarkening Measurement Methods

PD measurements can be realized either by measur-ing the properties of a small sample of the doped fiberof interest or by creating a full fiber laser or amplifiersystem, where the total efficiency and output powerare measured as a function of time. A fast, quantita-tive, and repeatable method is preferred for bench-marking or studying different Yb-doped fibers. Toachieve these objectives, the inversion of the investi-gated samples needs to be well controlled. Inducing acontrolled inversion is easier with short samplelengths, as shown below. The different PD bench-marking and studying methods that aim to induce arepeatable and controllable inversion to the fibersample can roughly be categorized to two dissimilarcases, namely, core-pumping [8,11,16] and cladding-pumping [9,15,17] methods. Both methods are basedon the observation that the excited state populationdensity is the main parameter affecting the rate ofPD [18].The measurement technique we proposed for

benchmarking single-mode Yb-doped fibers relieson saturating the inversion of the fiber to a flat andrepeatable level across the sample by core pumpingthe samplewith a high-irradiance, single-mode, fiber-coupled pump [8]. With this technique, PD can bemeasured with either a monochromatic or a white-light source coupled into the core of the fiber. Asthe area of the fiber core increases, the inversion be-comes harder to saturate using a single-mode pumpdiode, and othermeans of inducing the flat and repea-table inversion level in the fiber samplemust be intro-duced. Such means include cladding pumping of ashort fiber sample, but the typical pump irradianceis several orders of magnitude lower than with corepumping, and the inversion is thus difficult to satu-rate, making inversion dependent on the incidentpumppower [9]. Both of thesemethods are illustratedin Fig. 1. In the simplest form, the PD measurementconsists of measuring the transmission of the samplefiber before and after exposure to pump light, eitherby measuring a probe signal with a powermeter or awhite-light source with an optical spectrum analyzer.During PD of the sample, care should be taken to en-sure that the sample is not lasing, as any lasing sig-nificantly decreases the inversion level. Signal lightfeedback should be prevented by, for example, usingangled fiber ends. Care should also be taken not to in-fluence the measurement by using a powerful probelaser, as many reports indicate that the PD processcan be influenced by illumination by many differentwavelengths [11,14,15]. Potential probe laser influ-

ence can be decreased or eliminated, for example,by using very low intensities togetherwith a sensitivereceiver or by illuminating (and measuring) the sam-ple only at some time interval.

3. Simulated Inversion in Core- and Cladding-Pumping Measurements

To decrease the time required for each benchmarkingmeasurement, the inversion in the fiber sampleshould be maximized during the measurement in or-der to rapidly induce PD. On the other hand, the in-version level needs to be reproducible betweenmeasurements and samples. Inversion of a fiber sam-ple is defined by the pump brightness, the sampleproperties, and the pump wavelength, and the mostconvenient way of deriving a relatively accurate in-version profile of the fiber is by simulation. The re-sults presented here were calculated usingcommercially available fiber simulation softwareLiekki Application Designer v4.0. For simulating fi-ber types of different compositions, the absorptionand emission cross-section data need to be accuratefor each fiber type. The results shown here are basedon cross sectional data from aluminosilicate fibers(Fig. 2). The used cross sectional data were shown

Fig. 1. (a) Core-pumping method for benchmarking. Single-modepump laser diode is coupled to the sample fiber using a wavelengthdivision multiplexer (WDM). Either a probe laser, for example,He–Ne, or a white-light source (WLS) can be used to measurethe transmission of the sample before, after, or during PD usingthe powermeter or the spectrum analyzer, respectively. Samplecan be, for example, fusion spliced to the setup from each end.(b) Cladding-pumping method uses a multimode combiner(MMC, for example, a 6þ 1 to 1 tapered fiber device) to couple lightfrommultimode laser diodes to the sample fiber. The probe laser orother light source can be coupled to the core of the sample fiber byusing a multimode combiner equipped with a signal fiber. Pumppropagates in the cladding of the sample fiber.

1 September 2008 / Vol. 47, No. 25 / APPLIED OPTICS 4523

to fit experimental data well by Farrow et al. [19].The other typical simulation parameters used wereYb ion density of 1:2 × 1026 m−3 and an excited statelifetime of 0:85ms. The amplified spontaneous emis-sion (ASE) values were calculated for 101 wave-lengths between 1000 and 1100nm, and theinversion was calculated in 100 points along the fibersample, regardless of simulated fiber length.In the simulator the core propagating pump (and

ASE) transverse intensity profile is calculated basedon thegiven refractive indexprofile [20,21].Thepumpabsorption is derived through the rate equations ofthe active fiber (taking into account the concentrationof ions and cross section data), and the overlap of thecore propagating pumpmode and the doped region ofthe core is taken into account. The cladding propagat-ing pump is assumed to homogeneously fill the clad-ding and core. The inversion is calculated both in theradial direction and the longitudinal direction of thefiber, assuming cylindrical symmetry. The rate equa-tion model was described in more detail by Rebolledoet al. [22]. The inversion values shown later are typi-cally the average values of the total volume of eachfiber sample. The inversion is defined to be flat, whenthe standard deviation of the inversion data is <1%.This definition is based on what is readily attainableusing both the core-pumping and the cladding-pump-ing methods, and ensures that the sample volume re-presents well the behavior at the nominal (average)inversion fraction level. In the case of a flat inversion,also the radial dimension is typically flat, althoughpump and ASE distribution in the core affect the in-version profile.

A. Inversion in Core-Pumped Samples

The inversion profile of a core-pumped sample wassimulated as a function of pump power, samplelength, Yb ion concentration, and pump wavelengthfor a 4 μm core diameter fiber. The average inversionover the whole sample and the standard deviation ofthe inversion were calculated from the simulated in-version profiles. Each longitudinal inversion point isthe average of the radial inversion, and in caseswhere the inversion is defined to be flat (i.e., <1%standard deviation over the length of the sample)

also the radial inversion profile was confirmed tobe flat. Figure 3 illustrates how the inversion of a10 cm core-pumped sample saturates with increasingpump power at two different pump wavelengths. Theinversion level of a fiber changes very little with in-creased pump power when the inversion is flat. Dueto the short sample length and the high brightness ofthe pump, the ASE contribution to the inversion issmall. A general view of the influence of pump wave-length is shown in Fig. 4, where each point repre-sents the average inversion of a particular samplelength irradiated with 250mW of pump light. Asshown, the saturation level of inversion is mainly de-fined by the ratio of the absorption and emissioncross sections at the pump wavelength. The longersamples at 920nm show a decreased average inver-sion. The pump power used was not high enough tosaturate the inversion in those data points, and theinversion profile of the said points is not flat.

The scalability of the core-pumping method to lar-ger core diameters is shown in Fig. 5, where thesimulated pump power required for inversion satura-tion of a 10 cm sample is illustrated for two differentpump wavelengths. The simulations were done usingthe typical parameters listed above. The pump powerrequired to saturate a core-pumped sample increasesas the core diameter increases, but the required irra-diance is constant. This limits the core size of a prac-tical benchmarking measurement to approximately9 μm (920nm pump) and 18 μm (976nm pump) with500mW of coupled pump power. Practical limitationssuch as splice losses between dissimilar fibers andthe availability of high brightness pumps furtherlimit the core diameters, making the core-pumpingmethod suitable mostly to single-mode and few-modefibers up to 10 to 15 μm diameter when using the976nm pump wavelength.

A key parameter regarding a benchmarking mea-surement is the length of the fiber sample. The irra-diance requirements presented above are based onthe absorption and emission cross sections shownin Fig. 2 and the typical Yb ion concentration of 1:2×1026 m−3. The same irradiance requirement can be

Fig. 2. Absorption and emission cross sections used in the simu-lations. Based onmeasurements made from aluminosilicate fibers.

Fig. 3. Average inversion and standard deviation of the inversionversus pump power for a core-pumped, Yb-doped fiber. The inver-sion of the sample saturates, i.e., additional pump power no longersignificantly increases the inversion. The saturated inversionlevel is mainly defined by the ratio of absorption and emissioncross sections.


used for fiber samples where the number of ions iskept constant, i.e., higher concentration samplesneed to be shorter and lower concentration samplescan (but do not necessarily have to) be made longer.Shorter sample lengths may also be required when,for example, studying spectral regions of high ab-sorption such as the shape of the visible photodar-kened spectra or the Yb ion absorption peak [11].

B. Inversion in Cladding-Pumped Samples

To achieve a flat inversion in >10 μm core diameterfibers, a different approach of inducing the inversionis required. Cladding pumping using multimodepumpdiodes, as shown inFig. 1(b), fulfills the require-ment. Figure 6 shows the calculated average level andstandard deviation of the inversion over fiber sampleswith a 20 μmcore diameter anda 400 μmcladdingdia-meter with varied pump power. The results indicatethat a flat (<1% standard deviation) inversion isachieved over the whole range of pump power, butthe average inversion achieved depends on the pumpirradiance. The operational regime is far from the ir-

radiance thresholds for inversion saturation;200kW=cm2 for a 400 μm diameter fiber translatesto approximately 250W of pump power, which is fea-sible but hardly practical for a benchmarking mea-surement. The pump wavelength dependence of alow irradiance cladding-pumped sample was studiedby simulating the inversion profile of a fiber sampleover a range of pump wavelengths and fiber lengths(Fig. 7). The inversion shown is the average inversionof the fiber sample of a certain length, i.e., the graphdoes not quantify the flatness of the inversionprofiles.However, the inversion in the samples was flat withinthe shown wavelengths and sample lengths. Thepump power was 5W, and the core and cladding dia-meters were 20 μm and 400 μm, respectively. Due tothe low irradiance of the pump light, the average in-version is defined mainly by the absorption cross sec-tion. In order to induce a comparable inversionbetween different fiber samples, one needs to havea repeatable pump wavelength and coupled pumppower to the samples.

4. Experimental Results

Some key differences of the two measurement meth-ods are experimentally demonstrated to show the

Fig. 4. Simulated average inversion of a 4 μm core diameter fibersamples core pumped with 250mW of pump power. Each datapoint represents a fiber sample of certain length. Inversion isthe average over the volume of the sample. The average inversionof a sample is mainly defined by the ratio of absorption/emissioncross sections. The inversion of each sample is flat excluding thepoints at the 920nm long sample’s corner, where the average in-version is seen to decrease.

Fig. 5. Pump powers required to saturate a 10 cm sample by corepumping to different core diameters. The error ranges representthe uncertainty due to the pump power step between each simu-lated data point. The dashed irradiance lines are fit to the simula-tion results.

Fig. 6. Average inversion and standard deviation of the inversionversus pump power for a cladding-pumped, Yb-doped fiber with a20 μm core diameter and 400 μm cladding diameter. The inversiondoes not have a practical threshold level after which the inversionis saturated, but rather the inversion is tunable and very flat overa range of pump power. The inversion level is mainly defined bythe absorption cross section and the pump power.

Fig. 7. Simulated average inversion of a 20 μm core diametersample cladding pumped to 400 μm cladding with 5W of pumppower. Each data point represents a fiber sample of certain length.Inversion is the average over the volume of the sample. The inver-sion of cladding pumped samples is mainly defined by the absorp-tion cross section, as the irradiance of the pump is more than anorder of magnitude lower than in typical core-pumping case.


behavior of samples under different conditions. Core-pumped measurements using a setup as shown inFig. 1(a) were done to long fiber samples using a corepropagating probe laser operating at 633nm. Thealuminosilicate fiber samples with an Yb concentra-tion of approximately 2 × 1026 m−3 were exposed todifferent pump powers at 974nm wavelength. Themeasurement results are shown in Fig. 8, and the si-mulated inversion profiles are as inset. During therelatively short exposure to pump light, all samplesexhibit a different PD rate and degree of PD. The rea-son can be seen from the simulated inversion pro-files: as pump power is increased, a longer piece ofYb fiber is exposed to high inversion, causing an in-crease in the measured total PD signal.The simulation results suggest using a short fiber

sample length in order to induce a comparable PDrate and degree of PD between samples. Measure-ment results from shorter fiber lengths with approxi-mately 1:2 × 1026 m−3 Yb concentration are shown inFig. 9 for several pump powers at 974nm wave-length. The inversion profiles of the samples areshown as inset; however, due to overlapping datapoints, only one profile can clearly be seen. The probetransmission of samples exposed to different pumpintensities are the same within experimental errorbetween the different samples. This is attributedto the identical inversion profiles shown in the inset,regardless of the coupled pump irradiance.PD measurements with the cladding-pumping

method were done using a setup as described inFig. 1(b). The core propagating probe diode operatedat 795� 10nm (coupled power 68 μW), and the clad-ding coupled pump wavelength was 976� 2nm(coupled power 8W). The results are shown in Fig. 10,where the normalized transmission of the probe lightper unit length is shown as a function of time. Thesimulated inversion profiles of the different samplelengths are illustrated in the inset. The fiber corediameter was 30 μm, the octagonal cladding flat-to-

flat diameter was 250 μm, and the Yb concentrationin the core was approximately 1:2 × 1026 m−3. Sam-ples with longer lengths have lower overall averageinversion, and particularly the simulated inversionof the longest sample shows inversion depletion atthe input and output ends of the fiber due to ASE.The results show different PD rates for each samplelength. The results highlight that, in cladding pump-ing, the repeatability of the sample length also is ofconcern, and, to control the inversion in the samples,a short enough length should be chosen to not depletepart of the inversion by ASE.

5. Discussion

The principal differentiator of the core- and cladding-pumped methods is the pump brightness. The differ-ence in the achieved inversion profiles is illustratedin Figs. 4 and 7, where the core-pumped saturated

Fig. 8. Measured normalized probe transmission in a corepumped setup with varying pump powers. The pump wavelengthwas 974nm, and the probe wavelength was 633nm. The samplelength is too long to achieve inversion saturation, and the rateand degree of PD after the observed time period are dependenton the pump power. Inset: simulated inversion profiles for eachpump power at 974nm.

Fig. 9. Measured normalized probe transmission in a core-pumped setup with varying pump powers. The pump wavelengthwas 974nm, and the probe wavelength was 633nm. The inversionis saturated, and the degree of PD and the rate of PD are not de-pendent on the applied pump power after the observed time per-iod. Inset: simulated inversion profiles for each pump power at974nm (data points overlap).

Fig. 10. Measured normalized probe transmission in claddingpumped samples with varying lengths. The pump wavelengthwas 976� 2nm, and the probe wavelength was 795� 10nm. Tomake signals comparable, they were normalized with the samplelength. Two measurements were done for each length (shown asline, dotted line). Inset: simulated inversion profiles for each fiberlength. The simulation data shows the depletion of inversion dueto ASE at the input and output of the long sample fibers.


inversion is shown to follow the ratio of absorption/emission cross sections, whereas the cladding-pumped inversion mainly follows the absorptioncross section shape. Therefore the highest corepumping inversion can be achieved by pumping atshorter wavelengths (e.g., 920nm). In the typicalcladding-pumped case, the pump brightness is lower,and higher inversion is attained at the 976nm wave-length pump.The core-pumping method using a single-mode,

fiber-coupled, high-brightness diode relies on inver-sion saturation to achieve inversion flatness and re-peatability between different samples. In the caseof core pumping too long a fiber, the inversion profileis very flat in the beginning of the fiber, where the in-version is saturated.When the coupled pumppower isinsufficient to saturate the full length of the active fi-ber, the high core absorption leads to a sharp drop inthe longitudinal inversion profile as the propagatingpump irradiance is reduced. By increasing thecoupled pump power, the longitudinal inversion pro-filemay retain a similar shape, but the saturated por-tion of the fiber becomes increasingly longer. Theeffect becomesmore evidentwith increasing core size,as is the case with large mode area (LMA) fibers. Atsome point the fiber sample can no longer be satu-rated using reasonable pump power, and the inver-sion over the fiber sample is no longer flat andrepeatable.Fromthe simulation results the thresholdlevels for core-pumping saturation using 920nm and976nm pumping were approximately 700kW=cm2

and 200kW=cm2, respectively. For a single-mode,fiber-coupled pump with 500mW output power, themaximum core diameters feasible for inversion sa-turation using core pumping are approximately9 μm and 17 μm for 920nm and 976nm pumping, re-spectively, suggesting that the preferred wavelengthfor core pumping experiments is near the 976nm ab-sorption band. In practical measurement setups oneshould operate further away from the calculated lim-its due to coupling losses to the sample fiber and un-certainties in the simulation results. Given the above,the sample lengths need to be limited from a few cen-timeters to a few tens of centimeters (depending onthe Yb concentration) to ensure that the inversionwithin the sample is saturated.In the case of cladding pumping the longitudinal in-

version profile as a function of input pump power isquite different from the core-pumping method. Evenwith small input pump powers the inversion tends tobevery flat throughout the fiber, andby increasing thepump power the inversion increases evenly through-out the short sample, as shown inFig. 6. The inversionacross the length (and in the radial direction) is rela-tively uniform. The cladding-pumped sample lengthis limited by the onset of ASE in the fiber, as theASE suppresses the inversion in both the input andoutput ends of the fiber. The onset of ASE limitsthe practical length of cladding-pumped samples toa few tens of centimeters. In the cladding pumpingcase the inversion at the end part of the sample de-

creases due to reduced pump irradiance. If onewishesto compare samples at similar inversion levels, thepump absorption of different Yb fiber samples shouldbe kept constant in order to induce a similar inversionprofile between the fibers, or one should further flat-ten the inversion profile by using a two-directionalpumping scheme. Another approach is to keep the ex-cited Yb ion density constant, as suggested before [9].To keep the excited state density constant betweensamples of different core/claddingdimensions, onepo-tentially needs to tune both the coupled pump powerand the wavelength.

To reliably determine the inversion level of acladding-pumped sample, both the pumpwavelengthand the coupled pump power must be known. In ad-dition to knowing well the experimental parameters,the material parameters used in the simulation needto be accurate in order to obtain reliable inversion re-sults. If the glass composition is changed consider-ably, for example, by introducing or changing theamounts of Al, P, B, or Ge doping in the glass, the ab-sorption and emission cross section data need to beupdated. Accurate determination of both absorptionand emission cross sections and their error rangesto ensure the accuracy of the simulated inversion isunder way and will be the subject of a later report.

6. Conclusions

The benchmarking of PD from different types of fibersamples is feasible by inducing a flat, known, and re-peatable inversion to the sample volume. In order toinduce the PD to the sample in a short amount oftime, the inversion level within the sample shouldpreferably be high. The simulation results illu-strated that, depending on fiber type, this can beachieved both by core pumping (single-mode fibers,cores typically smaller than 10 μm) and by claddingpumping (LMA fibers). The four principal conclu-sions are the following:

– The critical parameter defining the inversion ofa sample in the core-pumping method is the pumpwavelength. Pump irradiance and sample lengthhave lesser effects when working in the inversion sa-turation regime. The pump wavelength can easily bestabilized, for example, by using grating stabilizedsources.

– As discussed above, the core-pumping methodrequired inversion saturation. To decrease the pumppower required for the inversion saturation, the useof 976nm pump wavelength region is preferred. Thecoupled pump power should be well above the simu-lated minimum irradiance due to simulation uncer-tainty, possible coupling losses, and PD inducedpump losses during the measurement.

– The cladding-pumpingmethod hasmultiple cri-tical parameters: pumpwavelength, pumppower, andsample length all have a significant effect on the leveland profile of the induced inversion. The pump wave-length induced inversion uncertainty can be mini-mized by using a pump wavelength away from


976nm. The pump power induced inversion uncer-tainty can be decreased by increasing the pumppower. The sample length induced inversion uncer-tainty can be decreased by decreasing the samplelength.– The main sample length limitation in the core

pumped case is the available pump power. In the caseof cladding pumped setup the main sample lengthlimitation is the onset of ASE, which in turn reducesthe inversion at the ends of the sample.

The harmful measurement artifacts of using toolong fiber samples were demonstrated with bothcore- and cladding-pumping methods. When properlyapplied as a benchmarking measurement, the core-pumping method is more repeatable, as the pumpwavelength is the principal parameter in definingthe inversion level of the sample. The cladding-pumping method, however, is more suitable forLMA fibers and advanced studies of the PD phenom-enon, such as the rate of PD at different levels of in-version. In such work, care needs to be taken tovalidate the simulated inversion levels. The simula-tion parameters need to be quantified through ex-periments, and the simulation model needs to bedemonstrated to fit the experimental data. Docu-mentation of the methodology and results of suchwork are under way and will be the subject of a laterpublication.

References1. S. Maryashin, A. Unt, and V. Gapontsev, “10mJ pulse energy

and 200W average power Yb-doped fiber laser,” Proc. SPIE6102, 61020O (2006).

2. S. Norman, M. Zervas, A. Appleyard, P. Skull, D. Walker,P. Turner, and I. Crowe, “Power scaling of high power fiberlasers for micromachining and materials processing applica-tions,” Proc. SPIE 6102, 61021 (2006).

3. A. N. Starodoumov; D. Dudley, S. McLean, A. Steinmetz, andN. Hodgson, “Hybrid fiber MOPA-bulk amplifier system forfrequency conversion,” Proc. SPIE 6871, 68710V (2008).

4. M. Kauf, R. Patel, J. Bovatsek, and W. Gries, “High power UVQ-switched and mode-locked laser comparisons for industrialprocessing applications,” Proc. SPIE 6871, 687123 (2008).

5. D. A. V. Kliner, F. Di Teodoro, J. P. Koplow, S. W. Moore, andA. V. Smith, “Efficient second, third, fourth, and fifth harmo-nic generation of a Yb-doped fiber amplifier,” Opt. Commun.210, 393–398 (2002).

6. F. Röser, T. Eidam, J. Rothhardt, O. Schmidt, D. N. Schimpf,J. Limpert, and A. Tünnermann, “Millijoule pulse energy highrepetition rate femtosecond fiber chirped-pulse amplificationsystem,” Opt. Lett. 24, 3495–3497 (2007).

7. K. C. Hou, M. Y. Cheng, D. Engin, R. Changkakoti,P. Mamidipudi, and A. Galvanuaskas, “Multi MW peak powerscaling of single-mode pulses using 80 μm core Yb-doped LMAfibers,” presented at the Directed Energy Professional Society19th Solid State and Diode Laser Technology Review Confer-ence, Albuquerque, New Mexico, 13–15 June 2006.

8. J. J. Koponen, M. J. Söderlund, H. J. Hoffman, andS. K. T. Tammela, “Measuring photodarkening from single-mode ytterbium doped silica fibers,” Opt. Express 1411539–11544 (2006).

9. J. Koponen, M. Söderlund, H. J. Hoffman, D. A. V. Kliner,J. P. Koplow, and M. Hotoleanu, “Photodarkening rate inYb-doped silica fibers,” Appl. Opt. 47, 1247–1256 (2008).

10. S. Yoo, C. Basu, A. J. Boyland, C. Sones, J. Nilsson, J. K. Sahu,and D. Payne, “Photodarkening in Yb-doped aluminosilicatefibers induced by 488nm irradiation,” Opt. Lett. 32, 1626–1628 (2007).

11. A. D. Guzman Chávez, A. V. Kir’yanov, Yu. O. Barmenkov, andN. N. Il’ichev, “Reversible photo-darkening and resonantphotobleaching of ytterbium-doped silica fiber at in-core977nm and 543nm irradiation,” Laser Phys. Lett. 4(10),734–739 (2007).

12. M. Engholm, L. Norin, and D. Åberg, “Strong UV absorptionand visible luminescence in ytterbium-doped aluminosili-cate glass under UV excitation,” Opt. Lett. 32, 3352–3354(2007).

13. J. Jasapara, M. Andrejco, D. DiGiovanni, and R.Windeler, “Ef-fect of heat andH2 gas on the photo-darkening of Ybþ3 fibers,”in Conference of Lasers and Electro-Optics CLEO TechnicalDigest (Optical Society of America, 2006), paper CTuQ5.

14. I. Manek-Hönninger, J. Boullet, T. Cardinal, F. Guillen,S. Ermeneux, M. Podgorski, R. Bello Doua, and F. Salin,“Photodarkening and photobleaching of an ytterbium-doped silica double-clad LMA fiber,” Opt. Express 15, 1606(2007).

15. S. Jetschke, S. Unger, U. Röpke, and J. Kirchhof, “Photodar-kening in Yb doped fibers: experimental evidence of equili-brium states depending on the pump power,” Opt. Express15, 14838–14843 (2007).

16. T. Kitabayashi, M. Ikeda, M. Nakai, T. Sakai, K. Himeno, andK. Ohashi, “Population inversion factor dependence of photo-darkening of Yb-doped fibers and its suppression by highlyaluminum doping,” in Conference of Lasers and Electro-OpticsCLEO Technical Digest (Optical Society of America, 2006), pa-per OThC5.

17. A. V. Shubin, M. V. Yashkov, M. A. Melkumov, S. A. Smirnov,I. A. Bufetov, and E. M. Dianov, “Photodarkening of alumosi-licate and phosphosilicate Yb-doped fibers,” in Conference ofLasers and Electro-Optics / Europe CLEO/Europe TechnicalDigest (Optical Society of America, 2007), paper CJ3-1-THU.

18. J. Koponen, M. Söderlund, H. J. Hoffman, D. A. V. Kliner, andJ. P. Koplow, “Photodarkening measurements in large modearea fibers,” Proc. SPIE 6453, 6453E-1-11 (2007).

19. R. L. Farrow, S.W.Moore, P. E. Schrader, and A. V. Smith, “Nu-merical simulations of Yb3þ-doped, pulsed fiber amplifiers:comparison with experiment,” in Conference of Lasers andElectro-Optics CLEO Technical Digest (Optical Society ofAmerica, 2008), paper CThL6.

20. D. Marcuse, “Field deformation and loss caused by curvatureof optical fibers,” J. Opt. Soc. Am. 66, 311–320 (1976).

21. D. Marcuse, Theory of Dielectric Optical Waveguides, 2nd ed.(Academic, 1991), pp. 328–329.

22. M. A. Rebolledo, S. Jarabo, M. Hotoleanu, M. Karasek,E. Grolmus, and E. Juanart, “Analysis of a technique to deter-mine absolute values of the stimulated emission cross sectionin erbium-doped silica fibres from gain measurements,” PureAppl. Opt. 6, 425–433 (1997).


joona€ koponen,€ marko€ laurila,€ and€ mircea€hotoleanu.€...

Documents