Jump to ContentJump to Main Navigation
Show Summary Details
In This Section


Managing Editor: Sorger, Volker

6 Issues per year

IMPACT FACTOR 2016: 4.492

5-year IMPACT FACTOR: 5.723

In co-publication with Science Wise Publishing

Open Access
See all formats and pricing
In This Section

Photonic crystal fiber amplifiers for high power ultrafast fiber lasers

Thomas T. Alkeskjold
  • Corresponding author
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
  • Email:
/ Marko Laurila
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Johannes Weirich
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Mette M. Johansen
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Christina B. Olausson
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Ole Lumholt
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Danny Noordegraaf
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Martin D. Maack
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
/ Christian Jakobsen
  • NKT Photonics, Blokken 84, 3460 Birkerød, Denmark
Published Online: 2013-11-30 | DOI: https://doi.org/10.1515/nanoph-2013-0050


In recent years, ultrafast laser systems using large-mode-area fiber amplifiers delivering several hundreds of watts of average power has attracted significant academic and industrial interest. These amplifiers can generate hundreds of kilowatts to megawatts of peak power using direct amplification and multi-gigawatts of peak power using pulse stretching techniques. These amplifiers are enabled by advancements in Photonic Crystal Fiber (PCF) design and manufacturing technology. In this paper, we will give a short overview of state-of-the-art PCF amplifiers and describe the performance in ultrafast ps laser systems.

Reviewed publication: WillnerAlan

Keywords: photonic crystal fibers; fiber lasers; fiber design; ultrafast lasers

1 Introduction

The development of advanced ytterbium-doped Photonic Crystal Fiber (PCF) amplifiers has fuelled the rapid progress in high power ultrafast fiber lasers [1], and the upper limit of obtainable pulse energy, peak power and/or average power, combined with sufficient laser linewidth, beam quality and efficiency has continuously been pushed [2, 3]. The fundamental challenge for ultrafast fiber amplifiers is nonlinear effects, such as Self-Phase Modulation (SPM), Stimulated Brillouin Scattering (SBS), Stimulated Raman Scattering (SRS) and Transverse Mode Instability (TMI) [4].

The traditional nonlinear effects such as SPM, SBS and SRS are relatively well-known effects, which are related to peak power, pulse width and linewidth of the laser pulses. These nonlinear effects can be detrimental to an ultrafast system, but the impact can be significantly reduced with appropriate system design [5, 6] and/or optimizing the fiber design. The current approach towards mitigating nonlinear effects in fiber amplifiers is to increase the effective mode area of the fiber and/or increasing the pump absorption, thereby reducing the required fiber length. Mitigating SBS has also been pursued by acoustically tailoring the core of the fiber, thereby effectively broadening the Brillouin gain, which increases the SBS threshold particularly for narrow linewidth systems [7, 8].

TMI is a relatively new nonlinear effect, which is not dependent on peak power, but dependent on average power. The TMI manifests as a sudden onset of beam instabilities i.e., a (semi-)chaotic coupling between fundamental and Higher Order Modes (HOMs) on a millisecond timescale [4, 9–12]. This process is mainly thermally driven by the interference between different transversal modes in the amplifier, and particular Large Mode Area (LMA) fibers suffer from this, since the quantum defect driven thermo-optic effect in the core tend to cause a single-mode waveguide structure to support few modes at high average power due to a thermal lensing effect [13]. The relevant design parameters for increasing TMI threshold is to have highly core-delocalized HOMs (single-mode core) for as high thermal loads as possible, reduce the impact of the thermo-optic effect, which means reducing the core size and/or reducing the pump absorption and/or reducing the quantum defect by inband pumping schemes and/or reducing the thermo-optic effect of the material [11].

Therefore, the traditional design parameters for increasing thresholds of classical nonlinear effects (larger core, higher pump absorption etc) are in contrast to those needed for increasing TMI threshold. There exist therefore a design trade-off between obtainable peak power and average power in current LMA fiber technology.

Nonetheless, ultrafast fiber laser systems using LMA fiber amplifiers delivering hundreds of watts in average power has attracted significant academic and industrial interest in recent years. These amplifiers can generate megawatts of peak power [14, 15] and hundreds of watts in average power [16–18] using direct amplification and multi-gigawatts of peak power using chirped pulse amplification (CPA) [2, 19] or combined with divided pulse amplification [20]. Recently, more than 500 W and 1.6 mJ from a CPA femtosecond system was demonstrated by coherent beam combining 4 individual femtosecond fiber amplifiers [21].

Several types of fiber designs have been proposed to push the limits with respect to average power, peak power and/or pulse energy. Conventional ytterbium-doped air/silica hexagonal designed PCFs have been demonstrated with core diameters from 40 μm to 100 μm having pump absorption (at 976 nm) from 10 dB/m to more than 30 dB/m [22]. Air/silica so-called large-pitch fibers (LPFs) has pushed core diameters to 135 μm [2] with an impressive 105 μm MFD and hybrid approaches, combining TIR and PBG, has demonstrated ytterbium-doped single-mode cores up to 85 μm diameter [23] with more than 20 dB suppression of HOMs in an only 90 cm long straight ROD fiber [24]. Whereas de-localization of higher-order modes is adjusted by trimming the air hole diameter of the cladding structure in conventional PCF structures, LPFs obtain this by accurate refractive index tailoring of the core material combined with an appropriate structured cladding [25].

Other approaches include leaky-channel designs [26] and chirally-coupled core fibers [27], where an active step index fiber core is surrounded by a number of passive cores that spirals around the center core. The side cores are matched to the supported HOMs of the center core and the longitudinal spiraling couples the HOM content away from the core. Furthermore, significant work has also been done with HOM fibers [28], where operation is not performed in the fundamental mode of the fiber, but in a specific HOM and long period gratings are used to couple light between the FM and HOM of the fiber [29].

In this paper, we will describe the design and performance of state-of-the-art of ytterbium-doped flexible PCF amplifiers and PCF ROD fiber amplifiers. We will describe how the unique dispersion properties of PCFs are utilized to form a compact 40 μm single-mode single-polarization fiber amplifier, and also how mode dispersion and index uniformity is controlled in a single-mode 85 μm core ROD fiber amplifier, both amplifiers delivering >300 W of output power with high optical efficiency and diffraction limited beam quality.

2 PCF LMA vs. step-index LMA fiber technology for ultrafast systems

Reducing the impact of nonlinear effects is of high importance in ultrafast lasers, and they are traditionally met by increasing the core size, to reduce the intensity in the core, but also increasing the pump absorption and thereby reduce the nonlinear interaction length. When the core size is increased, the numerical aperture of the core needs to be reduced in order to keep the fiber single-mode. However, there is a practical lower limit to the NA and index precision of MCVD type cores, and scaling the core diameter beyond ~15 μm and maintaining single-mode is challenging due to requirements to reproducibility and uniformity. Instead, LMA step-index fibers are mostly produced with multimode cores, where the beam quality is controlled by careful mode-matching of the input seed and/or appropriate coiling which helps leak any excited HOMs [30]. This approach is feasible to a certain level, but often a penalty has to be paid such as reduced beam quality, beam instability, reduced polarization extinction ratio (PER) and coiling induced mode compression, which reduces the effective mode area in the fiber and thereby also increasing the impact of nonlinear effects [31].

PCF cores offer much higher precision and uniformity of the refractive index. This allows manufacturing of cores with very low NA and even “negative” NA (i.e., smaller refractive index compared to the cladding background index). Combined with careful microstructured cladding structures, manufacturing of single-mode fibers with core diameters from 10 μm to >100 μm is possible. Furthermore, by careful design of the dispersion of the cladding even single-polarization LMA fibers can be manufactured and also form-factor reducing designs are possible [32].

The pump absorption in double-clad fiber is mainly determined by the core composition and by the core to cladding area ratio, which means that high absorption fibers need a small pump cladding. This should have a relative large numerical aperture to accept low-brightness pump light from pump diodes or bars. In conventional step-index LMA fiber technology, the pump cladding is formed by a low-index polymer cladding deposited on the outside of the fiber.

This has the impact, that the pump absorption is reduced as the fiber is made thicker, and the nonlinear thresholds are therefore decreased as the fiber is made more rigid.

If the fiber core diameter is increased, the effective mode spacing in the fiber is reduced and the fiber becomes more susceptible to microbend loss i.e., microscopic variations along the length of the fiber, which can cause the fiber to be unstable and sensitive to small perturbations. This is usually solved by making the fiber more rigid, and stable large-core fibers requires a rather rigid fiber structure. In conventional step-index designs having low-index polymer claddings, this rigid requirement is in strong contrast to having a small pump cladding (and thereby a small outer diameter). However, in PCF designs, the pump cladding is often formed by a ring of airholes (an air cladding) surrounding the core and cladding of the fiber. This air cladding can be positioned almost arbitrarily within the fiber structure and most importantly, independent of the actual outer diameter of the fiber. The rigidity of the fiber is therefore no longer coupled to the pump absorption, whereas larger stable cores can more easily be realized.

3 PCF flexible fiber amplifier technology

Fiber amplifiers for ultrafast high power fiber lasers often require large core designs in order to suppress nonlinear effects. With increasing core diameters, LMA fibers often become multimoded and performance suffer from beam instability, reduced polarization purity, and/or reduced beam quality. However, PCF technology offer significant progress in many of those performance parameters, in the following, this is exemplified with theoretical and experimental results for the flexible fiber amplifier DC-200/40-Pz-Yb, – manufactured by NKT Photonics.

Figure 1 shows an optical micrograph of the crossection of the fiber. The 40 µm diameter Yb-doped index-neutral core is surrounded by airholes and stress applying parts. The core supports one fundamental mode with an effective area of ~750 μm2. The 200 µm double cladding (DC) structure is formed by a ring of larger air-holes, and is acting as a high NA pump-cladding (Typical NA >0.55). The outer diameter is ~450 µm, which effectively reduces microbend loss.

Figure 1

Optical micrograph of the DC-200/40-PZ-Yb fiber, consisting of a 40 µm diameter Yb-doped index-neutral core, 200 µm pump cladding, and large stress applying parts that create a polarizing waveguide structure i.e., only one transversal mode with one polarization can propagate. The outer diameter is ~450 µm, which effectively reduces microbend loss.

As discussed above, native single-mode properties are advantageous for fiber amplifier due to diffraction limited beam quality and high beam stability. Further, when the fiber is single- mode, there is no need for careful mode matching of the seed input. In order to achieve single mode behavior in large core fiber amplifiers, the core NA has to be reduced, which becomes increasingly cumbersome in step-index fibers with core diameter >15 µm, since the required small index differences between core- and cladding are difficult to manufacture. On the other hand, in PCFs, a low NA can be obtained by reducing the diameter of the cladding holes, which can be controlled very precisely.

Due to careful design of the core refractive index as well as the cladding structure, the DC-200/40-Pz-Yb fiber is single-mode for wavelengths around 1 μm. The single mode cutoff wavelength is defined by a crossing of the effective mode index of the LP11 mode and the first cladding mode (FCM). At this crossing, the LP11 couples from the core to the cladding, and the remaining core mode is the fundamental mode LP01. Figure 2 shows the simulated effective mode indices as function of wavelength for the fundamental mode, higher-order mode and first cladding mode in the DC-200/40-Pz-Yb fiber.

Figure 2

Effective mode indices of the first 3 modes in the DC-200/40-Pz-Yb fiber. The LP11 mode exhibits a short-wavelength cutoff at 1100 nm, and the fiber is single mode for shorter wavelengths.

The design exhibits a short wavelength cutoff, i.e., the fiber is single mode on the blue wavelength side of the cutoff. This mechanism is attributed to the core and cladding design, and this is different from step-index fibers, where a fiber is single-mode on the red wavelength side of a cutoff. The cutoff wavelength in the simulated fiber of Figure 2 is ~1100 nm, enabling single-mode operation at wavelengths below 1100 nm. Increasing or decreasing the hole size results in a blue- or red-shifting, respectively, of the cutoff wavelength.

Experimentally, the modal content of the fiber was investigated using Spatial and Spectral imaging (S2) [33]. Being an interferometric method, S2 is very sensitive and can characterize multiple HOMs simultaneously from LMA fibers where conventional M2 or cut-off measurements fail. The analysis of HOM content of the fiber with various launching and coiling configuration revealed a HOM suppression of more than 24 dB for a coil diameter of 28 cm [34]. Following the guidelines of the single-mode fiber standard [35], which requires a HOM suppression of >19.3 dB for a single-mode fiber, the DC-200/40-PZ-Yb fiber has higher HOM suppression than required and is therefore a single-mode fiber.

Figure 3 shows the HOM suppression for three different coupling conditions. The values were obtained as an average over 5 measurements for each coupling condition.

Figure 3

Relative power of the LP11 mode with respect to the fundamental mode (MPI) measured on a 2.2 m piece of the DC-200/40-Pz-Yb fiber.

Using the optimum coupling condition, the LP11 mode is relatively weak with an average HOM suppression of -32.7 dB. When offsetting the input beam, one lobe of the LP11 mode overlaps with the input beam and is excited. Depending on the off-setting direction, different mode symmetries of the LP11 mode can be excited.

Being a single-mode fiber, the fiber provides excellent beam quality. Figure 4 shows a typical M2 measurement. Fitting the formula for nearly Gaussian beams to the measured data reveals an M2 value of <1.15. This result was obtained by using a signal input splice to the DC-200/40-Pz-Yb fiber. The signal fiber was a step-index fiber having a mode field diameter of only 15 μm i.e., much smaller effective mode area than the DC-200/40-Pz-Yb fiber, and demonstrates that diffraction limited beam quality can easily be obtained even without careful mode-matching of the input signal.

Figure 4

Measured beam diameter vs. propagation distance for a collimated beam from a DC-200/40-Pz-Yb fiber. The beam diameter was measured both in x-direction and y-direction. The fits yield an M2 value better than 1.15.

The large stress applying parts as shown in Figure 1 creates a polarizing waveguide structure and only one transversal mode with one polarization can propagate. The induced birefringence in the core leads to a higher refractive index component in direction of the plane of the stress elements (slow axis/x-direction) than in the direction perpendicular to the stress element plane (fast axis/y-direction). The core NA for the slow axis mode is therefore higher than the NA for the fast axis mode, resulting in a spectral band where only the slow axis mode is guided.

A transmission scan for the two polarization directions is shown in Figure 5. While the slow-axis is well guided down to <1000 nm, the fast axis is highly attenuated for wavelengths below 1200 nm. This behavior significantly reduces polarization crosstalk and yields a very stable polarized output beam. Furthermore, the fiber can be bent to coil diameters below 25 cm without notable losses, as shown in Figure 6. This enables a compact amplifier design with small form factor.

Figure 5

(A) Transmission scan for the two polarization directions. The polarizing fiber structure yields a strong attenuation for the y-polarization for wavelengths <1200 nm, while the x-polarization is well guided down to <1000 nm.

Figure 6

Bendloss as function of coil diameter. The fiber can be bent to coil diameters below 25 cm without significant bending losses.

Due to the precision in the manufacturing process, DC-200/40-Pz-Yb based amplifiers deliver very uniform performance. To illustrate this, Figure 7 shows the output power of 8 amplifiers during a 24 h burn-in period. The amplifiers are operated in constant current mode and are counter-pumped using free-space optics through a high-power SMA-905 connector. A single-mode step-index fiber is spliced to the signal input end and the input splice is equipped with mode-stripper and packaged to facilitate thermal management. All amplifiers show very high uniformity on all measureable performance parameters such as beam quality, PER and, efficiency.

Figure 7

Output power vs. time duing the burn-in process.

In an experiment to test the high-power handling capabilities of the DC-200/40-Pz-Yb fiber, up to 700 W of pump was coupled into a 2.3 m fiber. Figure 8 shows the output power vs. pump power for 10 W seed coupled in a counter-pumped configuration and quasi continuous-wave operation regime. More than 500 W output was obtained for 700 W pump power. The optical to optical efficiency is between 65% and 75%, and the fiber showed no sign of TMI up to ~300 W output power, where small beam fluctuations started to appear.

Figure 8

Fiber output power vs. pump power for 10 W seed power.

In another experiment, the DC-200/40-Pz-Yb fiber was tested for long term stability at an output power of ~55 W. Figure 9 shows the output power during a time span of 860 days, where output power level was stable and only little power loss was observed due to photo darkening. Several component failures (diodes, isolators, computer etc) were observed during the period. Typical polarization extinction ratio is 15–20 dB at high power operation.

Figure 9

Long-term test of the DC-200/40-Pz-Yb. Only little degradation was observed over a time span of 850 days.

4 Monolithic PCF amplifiers

An all-fiber turn-key laser solution is critical for reliable and compact high power systems in many industrial applications. A fused fiber combiner capable of handling kilowatts of pumps power is a critical component for achieving monolithic systems for high-power ultrafast lasers. A typical combiner contains a single-mode signal fiber surrounded by multimode pump fibers, all bundled and tapered down to match the diameter and NA of the active fiber as shown in Figure 10. This configuration has the disadvantage that the core diameter of the signal fiber is significantly reduced in the tapering process, resulting in a signal loss. In order to reduce the signal loss, an improved pump/signal combiner for large mode area PCFs has been developed [36]. The component combines six 400/440 µm 0.22 NA pump fibers and a 25/440 µm PM single-mode signal fiber into the active fiber. The combiner preserves the core diameter through the taper, resulting in a great reduction of signal loss.

Figure 10

Combiner setup with one signal and six pump ports fused together in fiber bundle, tapered down and spliced to a PCF.

The amplifier in [36] has been constructed by integrating 20 pump bars and coupling these into three of the pump fibers with up to 80% coupling efficiency.

The seed source was constructed by pumping 20 m of 20/400 µm Yb-doped LMA fiber at 915 nm and feeding the 75 W of output power at 1085 nm into the signal port of the combiner. The two remaining 915 nm ports were also coupled to the pump combiner. A schematic of the amplifier is shown in Figure 11. Up to 976 W was achieved with a slope efficiency of 75% using this configuration.

Figure 11

Illustration of the forward pumped amplifier with the high power PCF combiner as the key element.

5 Power scaling in narrow linewidth high power amplifiers

Power scaling of narrow linewidth system is limited by nonlinear effects – primarily by the onset of SBS. Several measures can be taken to increase the SBS threshold of the amplifier. The nonlinearity is proportional to the length of the fiber and inversely proportional to the core size, thus increasing both the core size and the pump absorption will improve the fiber with respect to SBS. Furthermore, the SBS threshold can be increased by reducing the Brillouin gain in the fiber.

Typically, a large core PCF consists of seven elements with uniform doping causing both the optical and the acoustic index of the refraction in the core to be preserved. Whereas the optical index determines the mode properties, the acoustic index determines the SBS gain of the core, and with seven uniformly doped elements the SBS wave will be amplified to a maximum. In order to circumvent this amplification, the core can be built up of elements with different acoustic velocities resulting in a reduction in SBS gain. The acoustic indices can be manipulated by co-doping each element with different concentrations of e.g., Ge or Al while maintaining the same Yb-concentration [7, 8].

In addition to the power limitations caused by SBS, thermal effects are seen to impose a limit to the beam quality and pointing stability at very high power levels.

In [8] a fiber has been designed which employs both acoustic and preferential gain tailoring of the core. The fiber core is shown in Figure 12. Preferential gain tailoring is achieved by replacing three of the core elements with silica such that the overall pump absorption in the outer region of the core is reduced. An additional acoustic region is achieved by ensuring that the central Yb-doped element has a different co-doping composition compared to the outer elements. It is, however, essential for the mode quality that the index of refraction is kept uniform across the whole core. Figure 13 shows how a Brillouin shift of more than 200 MHz can be achieved by varying the concentration of the co-dopants. This fiber was used to amplify a single-frequency signal with a linewidth of 300 MHz to approximately 990 W with a slope efficiency of 73% with no sign of modal instabilities and with M2 <1.3.

Figure 12

Fiber core design with both preferential gain tailoring and segmented acoustic tailoring. Yb-1 and Yb-2 have different levels of co-doping, and all three elements have the same optical refractive index.

Figure 13

Illustration of Brillouin shift in a fiber with segmented acoustic tailoring using three different material compositions.

6 PCF ROD fiber technology

Rod type fibers are so-called very large mode area (VLMA) fibers, with core diameter from 50 μm all up to 135 μm. These fibers are sensitive to bending, which compresses the mode area and can give loss. Therefore, ROD type fibers are manufactured with an outer diameter from around 1 mm to 2 mm to provide rigidity, hence ROD fibers are operated in straight configurations of typical lengths between 30 cm and 100 cm.

Several types of fiber designs have been suggested to push the effective mode area and the obtainable average power, while maintaining diffraction limited beam quality and good optical efficiency. ROD fibers are often designed with small pump claddings of about 200–300 μm diameter to ensure high pump absorption and high conversion efficiency.

High power extraction of 100s of W/m from rod fibers has recently been demonstrated using a 1.2 m large-pitch fiber (LPF) and a 1.0 m distributed mode filtering fiber (DMF) [16, 17] having similar mode field diameters. As mentioned previously, the LPF design achieves delocalization of HOMs using precise index tailoring of the core, where the DMF design is based on a hollow resonant cladding structure that enables filtering or delocalization of HOMs from the core within a well-defined wavelength region.

The DMF rod fiber design is shown on Figure 14. This is a hybrid fiber design utilizing both total internal reflection (TIR) and the photonic band gap (PBG) principles [23]. The core is formed by 19 ytterbium doped cells, and the hybrid cladding is formed by air holes arranged in a triangular lattice with a number of hollow high-index Germanium-doped silica rings (DMF elements) placed in a kagome-type structure. The air holes confine light by TIR and the DMF elements form a band of coupled cladding states, which are in-resonance with the FM at a first wavelength region and in-resonance with the HOMs at a second wavelength region, whereas the FM is well guided and the rod is single-mode in the second wavelength region. An air cladding surrounds the inner cladding, enabling high pump absorption of 20–30 dB/m. The resonance wavelength is determined by the width of the Germanium rings, which can be controlled during fiber draw process by inflating or deflating the air holes. This allows fine spectral control of the SM region, and enables positioning of the HOM resonance, and thereby HOM suppression, around the gain peak of ytterbium-doped rod fibers (~1030–1040 nm) where the rod only supports one guided spatial mode. Figure 15 shows simulated modal overlaps of the FM (A) and LP11 type HOMs (B-E) from 1020 to 1100 nm wavelength. From 1035 nm to 1055 nm, the FM has high overlap (80–93%) with the core, which allows for efficient FM amplification, and the HOMs has very low core overlap of ~35%. This low overlap allows for efficient mode scrambling of the HOMs and thereby efficient HOM suppression [37, 38]. The HOM suppression in passive operation has been measured with a technique called cross-correlated (C2) imaging [39], which is especially feasible for this type of fibers because the HOM suppression can be resolved with a relatively narrow light source of only 5–10 nm. This allows resolving HOM suppression by correlating the weighted power of the guided modes by a reference mode in a reference fiber. Using C2 imaging, the HOM suppression within the HOM resonance band was measured to >20 dB on a straight 90 cm long 85 μm core ROD fiber [24]. Comparing this to standard SMF28 fibers at the cutoff wavelength, where HOMs are only suppressed by ~19 dB on a 2 m coiled sample [35], the DMF structure is very efficient in suppressing HOMs.

Figure 14

Optical micrograph of the DC-250/85-Yb-ROD fiber based on a distributed mode filtering design.

Figure 15

Core overlap as a function of wavelength. The FM has high core overlap in the SM region, while the HOMs are efficiently suppressed due to resonant coupling with the DMF elements in the cladding structure.

At high average power operation, the core of the rod fiber heats up due to the quantum defect from the active core. Quantum defect heating yields a small refractive index increment via the thermo-optic effect in silica, and handling this effect becomes important for extremely low NA rod fibers because it modifies the guiding mechanisms [38, 40]. A heat load Q is generated in the core resulting in a temperature profile that is parabolic across the core and decays logarithmic outside the core [41]. The temperature profile adds a refractive index profile on top of the transverse fiber structure from the thermo-optic effect, see illustration in Figure 16.

Figure 16

Quantum defect heating results in a refractive index perturbation that changes the guiding mechanisms as the rod fiber is operated at high average output power.

In order to assess the effects of quantum defect heat generation on the SM regime of ROD fibers (in general), different design approaches have been compared through simulations [38, 42]. The DMF rod fiber is compared to a standard PCF with a 19 cell core, and a Step-Index-Fiber (SIF) with equal core area and aircladding as the DMF fiber. The SIF has a V-parameter of 2.40 and requires an unrealistic index step between core and cladding to be produced, thus it is a purely theoretical SM SIF. The cross sections of the DMF, PCF and SIF are illustrated in Figure 17. Essentially, the DMF rod fiber is a PCF with added DMF elements, and therefore is expected to have properties similar to the PCF outside spectral resonance regions. Within the spectrally resonance regions the DMF rod fiber has SM properties and thus resembles a theoretical SM SIF. The spectral location of the SM region depends on relative small index steps in the structure, and reproducibility demands high precision on allowable index variations as well as feature structures. Small perturbations in the core refractive index will red- or blue-shift the spectral location of the SM region [17, 37]. Rod fiber are typically very low NA fibers and considerations of performance at high output power must be engineered into the design. The DMF rod fiber has proven robust to thermal load perturbations [38], since quantum defect heating is a nonlocal effect, which also affects the cladding structure and thereby the DMF elements. As the refractive index of the core increases the refractive index of the DMF elements also increases allowing the resonant mechanism to follow the core heating, which means that the DMF rod fiber experience superior HOM suppression at high average power compare to standard PCF rod fibers, and only a small blue shift of the SM region is observed even at high average output power [17]. The SM properties of the DMF rod fiber is compared to a theoretical SM SIF for increasing thermal loads in W/m in Figure 17(A), by considering the core overlap of the FM and first HOM. The DMF rod fiber and SIF have equal active core area and aircladding dimensions, and the SIF’s refractive indices are chosen to yield a V-parameter of 2.40, thus it is strictly SM in the cold case. In Figure 17(B) the DMF rod fiber is simulated to be operated in the SM region at 1040 nm. The DMF rod fiber has superior properties compared to the SIF with a slightly larger FM core overlap and a significant higher suppression of HOM due to the resonant cladding structure until a thermal load of ~40 W/m, where the HOM becomes confined to the core, and is no longer in resonance with the DMF elements. A thermal load of 40 W/m corresponds approximately to an output power of 300–350 W.

Figure 17

(A) FM and HOM core overlap as a function of thermal load in the active core of a DMF rod fiber and a SM SIF with V=2.40. (B) MFD for increasing thermal load simulated for the DMF rod fiber, a standard PCF with similar structure as the DMF rod fiber but without the DMF elements in the cladding, and a SM SIF.

The quantum defect driven heating of the core yields higher mode confinement as the fiber is operated at higher average power. Figure 18 shows simulated mode profiles for the DMF rod fiber with average output power from 0 W and up to 350 W [38]. The HOM becomes guided as the rod fiber is operated at higher output power. The FM is well guided for all average output power levels with a slight decrease in MFD with increasing output power. The MFD vs. thermal load for the DMF rod fiber is compared to a standard PCF with similar structure, but without the DMF elements in the cladding and to the SM SIF in Figure 17(B). The cold DMF rod fiber has a FM MFD comparable to the SM SIF. For high thermal loads the FM MFD approaches the one of a standard PCF. Experimentally a significant decrease in MFD with increasing output power has been observed for LPF- and PCF-type rod fibers [40], which makes it important to engineer rod fiber designs to be operated at high average output powers.

Figure 18

FM and HOM profiles at average output power from 0 W to 350 W.

7 High power amplifiers using ROD fibers

To demonstrate the amplifier performance of the DMF ROD fibers, we used a 25 ps mode-locked linearly polarized seed source with 4–40 MHz repetition rate and center wavelength of 1032 nm. The seed was amplified to between 5 and 15 W with ~20 dB PER in a 2 m long counter-pumped DC-200/40-Pz-Yb fiber. The 10 dB spectral width of the seed was 700 pm. The seed was used as input signal to the ROD fibers, which were all between 80 cm and 100 cm long and equipped with 5 mm AR coated endcaps.

The ROD fiber was counter-pumped pump with a high-power pump module delivering up to 600 W at 976 nm and both the signal and the pump light were single-passed through the rod fiber. A schematic of the setup is shown on Figure 19.

Figure 19

Schematic of setup for testing ROD fiber amplifiers in ultrafast laser system.

Figure 20(A) shows the average output power as function of pump power for an 80 cm long rod, and shows that about 100 W of output power is achieved for 150 W of coupled pump light. The pump used in this experiment was not wavelength locked to 976 nm at low power levels, but locked at 976 nm at higher power levels. This impacted the efficiency at low power, but at 150 W pump, the efficiency increased to ~67%, as shown on Figure 20(B). The efficiency is not measured as function of absorbed power but as function of coupled pump power, and demonstrates high conversion efficiency at high average power operation. In this experiment the repetition rate was 40 MHz and at 100 W output power, the peak power was 100 kW.

Figure 20

(A) Signal output power as function of coupled pump power in the ROD fiber. (B) Optical conversion efficiency as function of signal power. The saturation is caused by the pumps which wavelength-lock at higher power levels.

The laser system was running continuously for >100 h and the beam quality and PER was monitored to exclude any photodarkening effect on the optical mode and the polarization.

Figure 21 shows the M2 and the PER monitored over 100 h, and show no sign of degradation. Figure 22 shows the output power over a 250 h period, and shows an initial power decay during the first 40–50 h, which is attributed to photo darkening, but after 50 h of operation the output power saturates and shows only very small sign of power degradation, which can easily be compensated in modern power-locked laser systems. This experiment demonstrates that the ROD fibers deliver very high and stable performance at both high average power and high peak power.

Figure 21

(A) Beam quality (M2) measured at 100 W output power of a ROD fiber over 100 h of continous operation. (B) PER measured at 100 W output power over same 100 h period.

Figure 22

Output power from a ROD fiber amplifier measured over 250 h of continuous operation.

In a second experiment, a pulse picker was used to reduce the pulse repetition rate to 4 MHz and the system was operated at an average output power of 75 W, corresponding to a peak power of 750 kW. The system was running continuously for >500 h and showed no signs of beam quality or PER degradation. This experiment also validated the performance and reliability of ROD fibers in high-peak power operation.

In a third experiment, we used a 1.00 m long rod and operated the rod up to ~310 W of output power with a peak power of about 300 kW. Above 310 W average power, we observed the onset of TMI. Above 250 W output power, the optical conversion efficiency was as high as 69%.

Overall, VLMA ROD fibers show excellent conversion efficiency, diffraction limited beam quality and high PER of typical >15 dB. These fibers can deliver high average power of ~300 W with an optical gain of >13 dB, having diffraction limited beam quality, and simultaneously delivering hundreds of kW in peak power, potentially several MW of peak power. Continuous operation for many hundreds of hours have been performed and neither the beam quality nor the PER shows any signs of degradation. The rods exhibit initial power degradation due to photodarkening, but the output power saturates quickly where after negligible degradation is observed

8 Conclusion

Photonic crystal fiber technology for large-mode area fiber amplifiers delivers state-of-the-art performance in ultrafast laser systems (fs, ps, and ns systems). Both flexible fiber designs and ROD fiber designs can be manufactured with single-mode properties, large effective mode area, high conversion efficiency and deliver unprecedented spatial and spectral brightness with high polarization purity. Various laser system tests show very high degree of uniformity, performance and reliability, both with respect to power, beam quality and polarization purity. The fibers/rod are manufactured with features such as coil control, which enables ease of use, and the single-mode and/or single-polarization property enables much less critical requirements to input signal matching, which increases performance uniformity.


  • [1]

    Tünnermann A, Schreiber T, Limpert J. Fiber lasers and amplifiers: an ultrafast performance evolution. Appl Opt 2010;49:F71–8. [Crossref]

  • [2]

    Eidam T, Rothhardt J, Stutzki F, Jansen F, Hädrich S, Carstens H, Jauregui C, Limpert J, Tünnermann A. Fiber chirped-pulse amplification system emitting 3.8 GW peak power. Opt Exp 2011;19:255–60.

  • [3]

    Stutzki F, Jansen F, Liem A, Jauregui C, Limpert J, Tünnermann A. 26 mJ, 130 W Q-switched fiber-laser system with near-diffraction-limited beam quality. Opt Lett 2012;37:1073.

  • [4]

    Eidam T, Wirth C, Jauregui C, Stutzki F, Jansen F, Otto H, Schmidt O, Schreiber T, Limpert J, Tünnermann A. Experimental observations of the threshold-like onset of mode instabilities in high power fiber amplifiers. Opt Exp 2011;19:13218–24. [Crossref]

  • [5]

    Zeringue C, Dajani I, Naderi S, Moore G, Robin C. A theoretical study of transient stimulated Brillouin scattering in optical fibers seeded with phase-modulated light. Opt Exp 2012;20:21196–213. [Crossref]

  • [6]

    Henry L, Shay T, Hult D, Rowland K. Enhancement of output power from narrow linewidth amplifiers via two-tone effect – high power experimental results. Opt Exp 2010;18:23939–47. [Crossref]

  • [7]

    Li M, Chen X, Wang J, Gray S, Liu A, Demeritt J, Ruffin A, Crowley A, Walton D, Zenteno L. Al/Ge co-doped large mode area fiber with high SBS threshold. Opt Exp 2007;15:8290–9. [Crossref]

  • [8]

    Robin C, Dajani I, Zeringue C, Ward B, Lanari A. Gain-tailored SBS suppressing photonic crystal fibers for high power applications. Proc. SPIE 8237, Fiber Lasers IX: Technology, Systems, and Applications, 82371D, 2012.

  • [9]

    Jauregui C, Eidam T, Limpert J, Tünnermann A. Impact of modal interference on the beam quality of high-power fiber amplifiers. Opt Exp 2011;19:3258–71. [Crossref]

  • [10]

    Smith AV, Smith JJ. Mode instability in high power fiber amplifiers. Opt Exp 2011;19:10180–92. [Crossref]

  • [11]

    Hansen KR, Alkeskjold TT, Broeng J, Lægsgaard J. Thermally induced mode coupling in rare-earth doped fiber amplifiers. Opt Lett 2012;37:2382–4. [PubMed] [Crossref]

  • [12]

    Ward B, Robin C, Dajani I. Origin of thermal modal instabilities in large mode area fiber amplifiers. Opt Exp 2012;20:11407–22. [Crossref]

  • [13]

    Hansen KR, Alkeskjold TT, Broeng J, Lægsgaard J. Thermo-optical effects in high-power Ytterbium-doped fiber amplifiers. Opt Exp 2011;19:23965–80. [Crossref]

  • [14]

    Brooks CD, Teodoro FD. Multimegawatt peak-power, single-transverse-mode operation of a 100 μm core diameter, Yb-doped rodlike photonic crystal fiber amplifier. Appl Phys Lett 2006;89:111119. [Crossref]

  • [15]

    Di Teodoro F, Morais J, McComb T, Hemmat M, Cheung E, Weber M, Moyer R. SBS-managed high-peak-power nanosecond-pulse fiber-based master oscillator power amplifier. Opt Lett 2013;38:2162–4. [Web of Science] [PubMed] [Crossref]

  • [16]

    Stutzki F, Jansen F, Eidam T, Steinmetz A, Jauregui C, Limpert J, Tünnermann A. High average power large-pitch fiber amplifier with robust single-mode operation. Opt Lett 2011;36:689–91. [PubMed] [Web of Science] [Crossref]

  • [17]

    Laurila M, Jørgensen MM, Hansen KR, Alkeskjold TT, Broeng J, Lægsgaard J. Distributed mode filtering rod fiber amplifier delivering 292 W with improved mode stability. Opt Exp 2012;20:5742–53.

  • [18]

    Sangla D, Saby J, Cocquelin B, Salin F. High power picosecond fiber laser emitting 50 W at 343 nm at 80 MHz. Proc. SPIE 8237, Fiber Lasers IX: Technology, Systems, and Applications, 82371D, 2012.

  • [19]

    Röser F, Eidam T, Rothhardt J, Schmidt O, Schimpf DN, Limpert J, Tünnermann A. Millijoule pulse energy high repetition rate femtosecond fiber chirped-pulse amplification system. Opt Lett 2007;32:3495–7. [Web of Science] [PubMed] [Crossref]

  • [20]

    Zaouter Y, Guichard F, Daniault L, Hanna M, Morrin F, Hönninger C, Mottay E, Druon F, Georges P. Femtosecond fiber chirped- and divided-pulse amplification system. Opt Lett 2013;38:106–8. [PubMed] [Crossref] [Web of Science]

  • [21]

    Klenke A, Breitkopf S, Kienel M, Gottschall T, Eidam T, Hädrich S, Rothhardt J, Limpert J, Tünnermann A. 530 W, 1.3 mJ, four-channel coherently combined femtosecond fiber chirped-pulse amplification system. Opt Lett 2013;38:2283. [Web of Science]

  • [22]


  • [23]

    Alkeskjold TT, Laurila M, Scolari L, Broeng J. Single-mode ytterbium-doped large-mode-area photonic bandgap rod fiber amplifier. Opt Exp 2011;19:7398–409. [Crossref]

  • [24]

    Laurila M, Barankov R, Jørgensen M, Alkeskjold T, Broeng J, Lægsgaard J, Ramachandran S. Cross-correlated imaging of single-mode photonic crystal rod fiber with distributed mode filtering. Opt Exp 2013;21:9215–29. [Crossref]

  • [25]

    Jansen F, Stutzki F, Otto H, Baumgartl M, Jauregui C, Limpert J, Tünnermann A. The influence of index-depressions in core-pumped Yb-doped large pitch fibers. Opt Exp 2010;18:26834–42. [Crossref]

  • [26]

    Dong L, McKay H, Fu L, Ohta M, Marcinkevicius A, Suzuki S, Fermann M. Ytterbium-doped all glass leakage channel fibers with highly fluorine-doped silica pump cladding. Opt Exp 2009;17:8962–9. [Crossref]

  • [27]

    Galvanauskas A, Cheng MY, Hou KC, Liao KH. High peak power pulse amplification in large core Yb-doped fiber amplifiers. IEEE J Sel Top Quantum Elect 2007;13:559–66. [Crossref] [Web of Science]

  • [28]

    Ramachandran S, Fini JM, Mermelstein M, Nicholson JW, Ghalmi S, Yan ME. Ultra-large effective-area, higher-order mode fibers: a new strategy for high-power lasers. Laser Photonics Rev 2008;2:429–48. [Web of Science] [Crossref]

  • [29]

    Nicholson JW, Fini JM, DeSantolo AM, Monberg E, DiMarcello F, Fleming J, Headley C, DiGiovanni DJ, Ghalmi S, Ramachandran S. A higher-order-mode erbium-doped-fiber amplifier. Opt Exp 2010;18:17651–7. [Crossref]

  • [30]

    Koplow JP, Kliner DAV, Goldberg L. Single-mode operation of a coiled multimode fiber amplifier. Opt Lett 2000;25:442–4. [PubMed] [Crossref]

  • [31]

    Fini J. Bend-resistant design of conventional and microstructure fibers with very large mode area. Opt Exp 2006;14:69–81. [Crossref]

  • [32]

    Petersen S, Alkeskjold T, Poli F, Coscelli E, Jørgensen M, Laurila M, Lægsgaard J, Broeng J. Hybrid Ytterbium-doped large-mode-area photonic crystal fiber amplifier for long wavelengths. Opt Exp 2012;20:6010–20. [Crossref]

  • [33]

    Nicholson J, Yablon A, Ramachandran S, Ghalmi S. Spatially and spectrally resolved imaging of modal content in large-mode-area fibers. Opt Exp 2008;16:7233–43. [Crossref]

  • [34]

    Laurila M, Alkeskjold TT, Lægsgaard J, Broeng J. Modal analysis of a large-mode area photonic crystal fiber amplifier using spectral-resolved imaging. Opt Eng 2011;50:111604.

  • [35]

    TIA-455-80-C: FOTP-80 IEC-60793-1-44 Optical Fibres – Part 1-44: Measurement Methods and Test Procedures – Cut-off Wavelength.

  • [36]

    Sipes DL, Tafoya JD, Schulz DS, Ward BG, Carlson CG. A 967 W single mode all-fiber PM Yb PCF fiber amplifier. Proc. SPIE 7914, Fiber Lasers IX: Technology, Systems, and Applications, PD, 2011.

  • [37]

    Jørgensen MM, Petersen SR, Laurila M, Lægsgaard J, Alkeskjold TT. Optimizing single mode robustness of the distributed modal filtering rod fiber amplifier. Opt Exp 2012;20:7263–73. [Crossref]

  • [38]

    Johansen MM, Hansen KR, Laurila M, Alkeskjold TT, Lægsgaard J. Estimating modal instability threshold for photonic crystal rod fiber amplifiers. Opt Exp 2013;21:15409–17. [Crossref]

  • [39]

    Schimpf D, Barankov R, Ramachandran S. Cross-correlated (C2) imaging of fiber and waveguide modes. Opt Exp 2011;19:13008–19. [Crossref]

  • [40]

    Jansen F, Stutzki F, Otto H-J, Eidam T, Liem A, Jauregui C, Limpert J, Tünnermann A. Thermally induced waveguide changes in active fibers. Opt Exp 2012;20:3997–4008. [Crossref]

  • [41]

    Brown DC, Hoffman HJ. Thermal, stress and thermo-optic effects in high average power double-clad silica fiber lasers. IEEE J Quantum Electron 2001;37:207–17. [Crossref]

  • [42]

    Coscelli E, Poli F, Alkeskjold TT, Jørgensen MM, Leick L, Broeng J, Cucinotta A, Selleri S. Thermal effects on the single mode regime of distributed modal filtering rod fiber. IEEE J Lightwave Technol 2012;30:3494–9. [Web of Science] [Crossref]

About the article

Corresponding author: Thomas T. Alkeskjold, NKT Photonics, Blokken 84, 3460 Birkerød, Denmark, e-mail:

Received: 2013-09-23

Accepted: 2013-10-30

Published Online: 2013-11-30

Published in Print: 2013-12-16

Citation Information: Nanophotonics, ISSN (Online) 2192-8614, ISSN (Print) 2192-8606, DOI: https://doi.org/10.1515/nanoph-2013-0050. Export Citation

Citing Articles

Here you can find all Crossref-listed publications in which this article is cited. If you would like to receive automatic email messages as soon as this article is cited in other publications, simply activate the “Citation Alert” on the top of this page.

Fabian Stutzki, Florian Jansen, Hans-Jürgen Otto, Cesar Jauregui, Jens Limpert, and Andreas Tünnermann
Optica, 2014, Volume 1, Number 4, Page 233

Comments (0)

Please log in or register to comment.
Log in