IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO.

2, MARCH/APRIL 2006

233

High-Power Ultrafast Fiber Laser Systems

Jens Limpert, Fabian R ser, Thomas Schreiber, and Andreas T nnermann, Associate Member, IEEE o u (Invited Paper)

AbstractThe recent demonstration of rare-earth-doped ber lasers with a continuous wave output power well above the kilowatt level with diffraction-limited beam quality has proven that ber lasers constitute a power-scalable solid-state laser concept. To generate intense pulses from a ber, several fundamental limitations have to be overcome. Nevertheless, novel experimental strategies and ber designs offer an enormous potential toward laser systems with high average powers and high pulse energies. This paper reviews the challenges, achievements, and perspectives of ultrashort pulse generation and amplication in bers. Index TermsChirped pulse amplication, ber laser and ampliers, microstructured bers, nonlinear ber optics, ultrafast optics.

I. INTRODUCTION

NUMBER of important practical as well as fundamental research applications of ultrafast lasers appeared over the last decades [1], a trend initiated by the step from old dye-laser technology toward solid-state lasers. These high-power ultrafast solid-state lasers use small rods as the amplier media for instance, Titanium-doped sapphire as the most widespread one [2]and have the potential to generate signicantly higher pulse energies, higher powers, and shorter pulse durations, in combination with greater reliability than dye-lasers. However, these systems are difcult to scale in average power and suffer from low efciencies because direct diode pumping is not possible. Furthermore, the complexity of short pulse high energy Ti:sapphire lasers still constrain the employment of ultrafast laser technology in industrial environments. Recently, the development of diode pumped solid-state lasers, such as Yb:YAG or Cr:LiSAF, has constituted a big step forward in terms of efciency. In order to overcome thermooptical effects, which limit the power scaling capability of these systems, several novel gain media designs, such as thin disk or slab, have been introduced [3], [4]. However, due to the low single pass gain of these amplier materials, very complex systems; e.g., regenerative amplication schemes, are required to obtain a reasonable output. Therefore, robustness, compactness, and long-term stability are restricted in short pulse bulk solid-state laser systems. Alternatively, forming the gain medium to be long and thin not only leads to outstanding thermooptical properties, but also

Manuscript received November 17, 2005. J. Limpert, F. R ser, and T. Schreiber are with the Institute of Applied o Physics, Friedrich Schiller University Jena, D-07745 Jena, Germany (e-mail: Jens.Limpert@uni-jena.de; roeser@iap.uni-jena.de; Thomas.Schreiber@unijena.de). A. T nnermann is with the Institute of Applied Physics, Friedrich Schiller u University Jena, D-07745 Jena, Germany, and also with the Fraunhofer Institute for Applied Optics and Precision Engineering, D-07745 Jena, Germany (e-mail: Andreas.Tuennermann@iof.fraunhofer.de). Digital Object Identier 10.1109/JSTQE.2006.872729

to a very high single pass gain. Fiber-based laser systems have the reputation of being immune to any thermooptical problems due to their special geometry. Their excellent heat dissipation is due to the large ratio of surface-to-active volume of such ber. The beam quality of the guided mode is determined by the ber core design, and is therefore power-independent. Due to the connement of both the laser and pump radiation, the intensity is maintained over the entire ber length and is not limited to the Rayleigh length, as is the case in longitudinally pumped bulk lasers. The gain of the laser medium is determined by the product of pump light intensity and interaction length with the laser radiation in the gain medium. Therefore, the decisive product can be orders of magnitude higher in bers than in other bulk solid-state lasers. This results in very efcient operation of ber laser systems exhibiting very high gain and low pump threshold values. Additionally, complete integration of the laser process in a waveguide allows for inherent compactness and long-term stability of ber lasers. In particular, Ytterbium-doped glass bers, which have a quantum defect of less than 10%, can provide optical-to-optical efciencies well above 80% and, therefore, low thermal load. These ber laser systems are especially interesting for highpower ultrashort pulse generation and amplication because of several unique properties [5]: Firstly, a broad emission spectrum allows for short pulse amplication. In ytterbium-doped glass bers, the amplication bandwidth of approximately 40 nm supports, in principle, pulses of durations as short as 30 fs. Furthermore, the absorption spectrum covers a wavelength range in which powerful diode lasers are commercially available. An additional point to note is that the long uorescence lifetime (1 ms) results in a high-energy storage capability. Exited-state absorption of pump or signal radiation, or concentration quenching by ion-ion energy transfer processes, does not occur with ytterbium because only two energy level manifolds are relevant for all optical wavelengths. High power ber lasers usually use the double-clad ber concept, invented in 1988 by Snitzer [6]. Such a double-clad ber is characterized by a second waveguide, which is highly multimode, surrounding the active core. Into this second waveguide, also called inner cladding or pump core, low brightness high power diode laser radiation can be launched. This pump light is gradually absorbed over the entire ber length and is converted into high brightness high power laser radiation. Thus, doubleclad rare-earth doped bers can provide a highly efcient brightness improvement by pump-to-laser radiation conversion by the laser process itself. The aforementioned properties make rare-earth-doped bers superior to other solid-state laser concepts in a variety of performance categories. This has become obvious following several

1077-260X/$20.00 2006 IEEE

234

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO. 2, MARCH/APRIL 2006

recent demonstrations of continuous wave ber laser systems exhibiting more than 1 kW of average power while maintaining an excellent beam quality [7][10]. II. BASIC LIMITATIONS OF SHORT PULSE AMPLIFICATION IN OPTICAL FIBERS The connement of the laser radiation and the long interaction length makes bers attractive for solid-state laser systems; however, nonlinear effects impose fundamental limitations on these systems, which may lead to severe pulse distortions and even to damage of the ber. The lowest-order nonlinear effect in optical bers originates from the third-order susceptibility (3) , and is responsible for an intensity-dependent refractive index in the form n = n0 + n2 I. Consequently, an optical eld propagating through a ber experiences a self-induced phase shift, a phenomenon referred to as self-phase modulation (SPM). A second important class of nonlinear effects results from stimulated inelastic scattering, whereby the radiation transfers a part of its energy to the glass host in form of excited vibrational modes. A large frequency shift (13 THz) is observed due to the excitation of optical phonons, a process called stimulated Raman scattering (SRS), whereas the excitation of acoustical phonons leads to a much smaller shift (17 GHz) called stimulated Brillouin scattering (SBS). Both manifest themselves as a signicant power loss mechanism in ber-based laser systems. In the context of ultrashort pulse amplication in bers, where the spectra are significantly broader than the Brillouin gain bandwidth, the effect of SBS can be neglected [11]. In general, the nonlinearity coefcients in glass bers are intrinsically small. Both the nonlinear index coefcient n2 and the gain coefcients of SRS and SBS are at least two orders of magnitude smaller than in other common nonlinear media [12]. Nevertheless, due to the large product of intensity and interaction length inside the ber core, nonlinear effects can be observed at very low power levels and basically limit the performance of pulsed rare-earth-doped ber systems. Pulse propagation in bers is described by the nonlinear Schr dinger equation (NLSE), shown in (1), which describes o the temporal and longitudinal dependency of the slowly varying pulse envelope A(z, T ) along the ber in the retarded time frame T A z

Fig. 1. Impact of SPM in CPA system on pulse quality of recompressed pulses assuming a Gaussian pulse shape during amplication in comparison with transform-limited pulse (B = 0).

self-phase modulation as the lowest-order nonlinear effect. The phase imposed by SPM can lead to a nonlinear chirp, which cannot be compensated by standard dispersive elements. An important parameter is the accumulated nonlinear phase of the pulse traveling through the ber, the so-called B-integral, dened by (3) with I(z) as the pulse intensity varying over the ber length L [13] B= 2
L 0

n2 I(z) dz.

(3)

n
n =2

in 1 n n! T n

g A A = i|A|2 A 2

(1)

where g represents the gain or loss of the ber, n are the Taylor expansion coefcients of the dispersion parameter 2 (), and the pulse amplitude A(z, t) is assumed to be normalized such that |A|2 represents the optical power. The nonlinearity coefcient is dened by n2 0 = (2) cAe where Ae is the effective mode eld area, n2 is the nonlinear refractive index coefcient, c equals the speed of light, and 0 is the center frequency of the optical eld. With these parameters, (1) includes the linear effects of dispersion and gain/loss and

A B-integral of smaller than 1 is considered as a linear propagation, pulse quality degradation is expected above this value. To illustrate the consequence of SPM, a chirped pulse amplication (CPA) system is simulated by solving the NLSE using the split-step Fourier method and assuming a Gaussian pulse shape during amplication. Fig. 1 illustrates the recompressed pulse intensity proles after accumulating a nonlinear phase shift in comparison to the transform-limited pulse. As revealed, SPM leads to a pulse broadening and the evolution of sidelobes containing a considerable amount of pulse energy. Hence, even moderate self-phase modulation can signicantly distort a recompressed pulse after amplication, with the result of reduced peak power and degraded pulse contrast. In contrast to SPM, SRS is a process which is initiated at a certain threshold value. Above this threshold, energy is rapidly transferred to a frequency-downshifted Stokes wave. This process can be described quantum-mechanically as an inelastic scattering process, where an incident photon is annihilated to create a photon at a downshifted frequency, the Stokes wave, and a phonon with the right energy and momentum. Additionally, a frequency up-shifted photon, the so-called anti-Stokes wave, can be created if a phonon with the right energy and momentum is available. The threshold for the onset of SRS in a ber amplier is, to a good approximation, given by (4), where gR is the peak Raman gain and Le is the effective interaction length [11]
SRS Pthreshold

16 Ae . gR Le

(4)

LIMPERT et al.: HIGH-POWER ULTRAFAST FIBER LASER SYSTEMS

235

Fig. 4.

Coreless ber end cap to avoid facet damage.

Fig. 2.

Emission spectrum of t ber amplier as a function of pulse energy.

Fig. 5. Calculated bending losses in 30-m LMA ber (NA = 0.06) for rst four transverse modes subject to bending radius.

Fig. 3.

Damaged ber facet.

As an example, the threshold behavior of SRS and energy loss of the propagating optical pulse in a single-mode shortpulse ber amplier is shown in Fig. 2. The measured emitted spectrum reveals strong changes with increasing peak power. At lower pulse energies, only the signal at 1060 nm is observable. At peak powers higher than 2.3 kW, a signal at about 1120 nm increases rapidly, indicating that the threshold of stimulated Raman scattering is passed. At a peak power of 3.4 kW, even the frequency up-shifted anti-Stokes wave at about 1010 nm becomes detectable. At 4.6 kW, the pulse breaks up and most of the intense central part of the spectrum is transferred into the Stokes wave. Beside these basic nonlinear effects, ber damage has to be considered as a scaling limitation of high-energy ber laser systems. The surface damage uence threshold of fused silica, which is signicantly lower than the bulk damage uence threshold, at a wavelength of 1 m, is given by [14] Fdamage = 22 ( )0.4 J/cm2 (5)

This problem can be solved by special treatment of the ber end. One solution is to splice a coreless end cap on the output side of the ber amplier, as shown in Fig. 4. The expansion of the beam reduces the uence and avoids ber facet damage. Due to the aforementioned nonlinear distortions, the generation of high peak powers in pulsed ber laser systems is very challenging. Novel ber designs and experimental strategies have to be applied to scale up the output parameters. There are basically two approaches to amplify ultrashort laser pulses in bers: the parabolic pulse amplication, which actually uses nonlinearity, and the chirped pulse amplication, which relies on avoidance of nonlinearity. Before reviewing these two basic amplication concepts, ber designs with intrinsically reduced nonlinearity; i.e., an increased mode-eld diameter and a reduced ber length, will be discussed. III. LOW-NONLINEARITY FIBER CONCEPTS As discussed in Section III and shown in (3) and (4), the nonlinear effects are proportional to the ber length and the intensity in the ber core, and therefore are inversely proportional to the mode-eld area of the guided radiation in the ber. Thus, an enlargement of the mode-eld diameter and a reduction of ber length would help to avoid disturbing nonlinear effects. Using special techniques and ber designs the mode-eld-area of single-transverse mode ber devices could be signicantly increased in the past years. To investigate the modal properties, the normalized frequency V is introduced in (6). It is related

where is the pulse duration in nanoseconds. Therefore, the damage threshold is in general signicantly lower than the extractable energy. Fig. 3 shows an example of a damaged ber facet.

236

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO. 2, MARCH/APRIL 2006

to the effective refractive indices of the core nC and cladding nClad of the ber with a core radius ae . One can show that a ber becomes single-mode for V < 2.405 [15] V = 2 2 ae NA = ae n2 n2 . C Clad (6)

Thus, one approach to increase the core diameter is to decrease the numerical aperture (NA) at a certain value of V . However, in conventional step-index bers, a reduction in the numerical aperture below a certain limit is not tolerable in terms of propagation losses and precision of core fabrication method. More sophisticated ber designs are based on modied index proles, which increase the single-mode area by using an outer ring structure. Such large-mode area bers with up to few 10-m core diameter and diffraction-limited output could be demonstrated [16]. Preferential gain to the fundamental mode is created by an optimally overlapping rare-earth dopant distribution [17]. This concept can be extended to gain and loss managed multimode bers to discriminate higher order modes [18]. Stable fundamental mode propagation over more than 20 m at 1.5-m wavelength is obtained in a conventional step-index core double-clad ber with a core diameter of 45 m with a numerical aperture of 0.13 [19]. In [19], it is stated that the mode-coupling between transversal modes is affected by the cladding diameter and the ber fabrication process. Thus, for maintaining the fundamental mode, it is preferable to use large cladding diameters relative to the core diameters and the high-quality modied chemical-vapor deposition (MCVD) process. Further discrimination of higher order modes is achieved by a careful optimization of the seed launching conditions and incorporated tapered sections [20] inside the ber laser or amplier. Using these techniques, the single-mode operation of a 50-m core (MFD 30 m) ber amplier at 1.06 m was reported [21]. In recent years, the most commonly employed lownonlinearity ber concept is the low-numerical aperture largemode-area (LMA) ber. Such an LMA ber has V-parameters in the range of 510, and can therefore guide several higher order transverse modes. However, bending losses can be applied to achieve stable fundamental mode operation of a ber laser or amplier [22]. Fig. 5 shows the calculated bending losses for the fundamental and three higher-order modes as a function of bending radius in a 30-m LMA ber (NA = 0.06) [23]. This calculation reveals a signicant discrimination of higher order transverse modes. As an example, at a bending radius of 50 mm, the induced bending loss for the LP01 mode is 0.01 dB/m and for the rst higher order mode LP11 is 52 dB/m. This difference of approximately ve orders of magnitude enforces fundamental mode operation. Robust and environmentally stable fundamental mode operation in even larger cores is only possible in a truly single-mode ber. Therefore, a strictly single-mode ber with large mode areas would be a signicant achievement. Microstructuring the ber by including air-holes adds several attractive properties to conventional bers. These so called photonic crystal bers (or holey bers) are currently a subject of intense research [24]. Solid core photonic crystal bers consist of a regular array of air-lled holes characterized by the hole diameter d and the

Fig. 6. (a) Structural parameters, and (b) modal characteristics of a one holemissing photonic crystal ber.

pitch [Fig. 6(a)]. To describe the modal properties, (6) can also be used for such one-hole-missing PCFs where the param eter ae corresponds to / 3. The main difference between a standard step index ber and a PCF is that for PCF, the effective cladding index is strongly dependent on /. Also, in the limit of / 0, the effective cladding index reaches the core index. This property becomes more interesting by plotting the single mode boundary V = 2.405 of the normalized wavelength over the relative hole diameter, as done in Fig. 6(b). For d/ > 0.4, the ber turns from multimode operation into single mode operation if the relative wavelength is large enough. For d/ < 0.4 the ber becomes single mode for all wavelengths / due to the fact that the wavelength dependence of nClad keeps the NA low enough to stay single mode. This regime is called endlessly single mode. It was observed in the rst photonic crystal bers and was investigated theoretically soon afterward [25], [26]. Such an endlessly single mode operation is not known from step index bers, and leads to a concept of scaling the core diameter of a photonic crystal ber. If the modal properties do not depend on the normalized wavelength /, the mode diameter of a PCF, which is proportional to , can theoretically be scaled to innity at a given wavelength. Of course, the scaling of the core size is limited by increasing propagation losses [27]. If the V -parameter of the photonic crystal ber is smaller than one, connement of the mode is too weak and leakage occurs through a PCFs nite cladding. On the other hand, if the value / becomes too small (<0.1), scattering losses due to longitudinal nonuniformities increase; e.g., losses due to microbending, macrobending, and dielectric imperfections play an important

LIMPERT et al.: HIGH-POWER ULTRAFAST FIBER LASER SYSTEMS

237

role. If further scaling of the mode-eld diameter is necessary, the core can be extended to consist of more than one missing hole; i.e., three missing holes [28]. Even seven missing hole PCFs are demonstrated with mode eld diameters of 50 m and fundamental mode guidance [29]. The gain medium of a ber laser can be fabricated by replacing the pure silica core by a rare-earth-doped rod. In general, the core is codoped with uorine to compensate for the refractive index increase due to the rare-earth ion. This ensures that the refractive index of the rod closely matches that of the silica cladding. The refractive index step can be reduced to 105 , even at relatively high ytterbium doping levels, so that the guiding properties are determined by the photonic crystal structure surrounding the core, and not by any index step due to the dopants. A further advantage of microstructuring a ber is the possibility of forming an air-cladding region to create double-clad bers [30]. Double-clad PCFs can be achieved by surrounding the inner cladding with a web of silica bridges, which are substantially narrower than the wavelength of the guided radiation. The result is a very high numerical aperture of the inner cladding up to 0.8. This allows for reducing the diameter of the inner cladding, while maintaining sufcient brightness acceptance for efcient pumping. The advantage of shrinking the inner cladding is that the overlap ratio of the core to the inner cladding increases, leading to shorter absorption lengths, and therefore to reduced nonlinearity [31], [32]. Recently, a ber design has been developed which exhibits extremely reduced nonlinearity. This type is called rod-type photonic crystal ber [33]. The basic idea of this ber design is to have outer dimensions of a rod laser, meaning a diameter in the range of a few millimeters and a length of just a few tens of centimeters, but to include two important waveguide structures, one for pump radiation and one for laser radiation. The pump light is conned by an air cladding and the laser light is guided in an array of air holes. Finally, such a ber has an extremely reduced nonlinearity, and therefore allows for signicant power and energy scaling. The cross section of this ber is shown in Fig. 7. The inner cladding has a diameter of 180 m. The numerical aperture is as high as 0.6. The ytterbium-doped core has a diameter of 60 m; to our knowledge, this represents the largest single-mode core ever reported for doped bers. This ber possesses a pump light absorption as high as 30 dB/m at 976 nm. Usually, the extraction of high-power levels from short ber lengths is limited by thermooptical problems. A detailed analysis of the thermooptical behavior of high-power ber lasers (including photonic crystal bers) has revealed that power scaling is restricted by damage to the polymer coating, which occurs at ber surface temperatures between 100 C and 200 C [34]. These temperatures are easily reached if power levels in the 100W/m range are extracted. In a conventional double-clad ber, the coating has an optical function. The coating has to have a lower refractive index than fused silica, and therefore forms the waveguide for pump radiation. In contrast, in a microstructured air-clad ber, it just serves to protect the ber from mechanical damage and chemical attack. The most straightforward way to avoid damage of the coating is to remove it. This can be done if

Fig. 7. (a) Microscope image of rod-type photonic crystal ber. (b) Close-up view of inner cladding and core regions.

the ber itself has enough mechanical stability; i.e., if the ber is thick enough. The ber shown in Fig. 7 has an outer cladding diameter as large as 2 mm and possesses no coating. In addition, the larger outer diameter improves the heat dissipation capabilities of this ber [34], and also reduces the propagation loss of weakly guided radiation because of the increased rigidity. A comparison of nonlinearity between a double-clad ytterbium-doped step-index single-mode ber, a low-numerical large-mode-area ber, and the previously described singletransverse mode rod-type photonic crystal ber, shows the achievement of this novel ber design (Table I). The nonlinearity, basically given by the effective mode area and the absorption length of the ber, is normalized to the value of the standard single mode step-index ber in the 1 m wavelength region. This comparison reveals a reduction of nonlinearity by a factor of about 2000 in the rod-type PCF. As stated previously, whenever power or energy scaling is limited by nonlinearity, such a ber offers a signicant potential. IV. SHORT PULSE AMPLIFICATION IN THE PRESENCE OF NONLINEARITY If short optical pulses are amplied in a ber, nonlinearity cannot be neglected as long as there are no special techniques

238

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO. 2, MARCH/APRIL 2006

TABLE I COMPARISON OF NONLINEARITY IN DIFFERENT FIBER DESIGNS

Fig. 8. Evolution of Gaussian input pulse toward parabolic output pulse in 10-m normal-dispersion ber amplier. Intensity in 2-m increments in: (a) time domain (linear scale) and (b) spectral domain (logarithmic scale).

applied to avoid it. Dispersion, gain, and nonlinearity affect the pulse evolution during the amplication in a ber. This section concentrates on the interplay of these effects and how they can be used to generate high power ultrashort pulses. As discussed in Section II, SPM leads to a phase shift across the pulse that is directly proportional to the pulse intensity itself. Thus, as the instantaneous frequency is the negative rst derivative of the phase, new frequencies are generated across the pulse, where a red shift occurs on the leading edge and a blue shift on the trail of the pulse. In addition to the temporal frequency chirp, a second consequence of SPM is, of course, a spectral broadening due to this generation of new frequencies. The interplay of SPM and dispersion can lead to dramatic temporal and spectral changes with the result that the generated frequency chirp cannot be compensated to obtain transform-limited pulses by standard dispersive elements such as a pair of gratings or prisms [35], which require a linear chirp. Thus, the evolution of a pulse, which maintains a linear chirp during amplication, would be favorable. The pulse shape with a perfect linear chirp is a parabolic shape. If this shape is maintained, the limits of the accumulated phase, the B -integral, can even be overcome. The next two sections discuss the experimental realization of this idea. Firstly, parabolic pulses are generated during direct amplication of femtosecond pulses by a balanced interplay of dispersion, gain, and nonlinearity. Secondly, the amplication of parabolic pulses generated directly by a ber oscillator is demonstrated.

A. The Generation of Parabolic Pulses in a Fiber Amplier The numerical investigation of the NLSE with gain and second order dispersion 2 (1) reveals that the interplay of normal dispersion, nonlinearity (self-phase modulation), and gain produces a linearly chirped pulse with a parabolic shape, which resists optical wave breaking [36], [37]. The linear chirp can be efciently removed, allowing for high quality pulse compression. For an amplier with constant distributed gain, an exact asymptotic solution has been found corresponding to a parabolic pulse that propagates self-similarly [38]. The asymptotic pulse characteristics are not inuenced by the shape or width of the input pulse. Only the initial pulse energy determines the nal pulse amplitude and width. As an example, the spectral and temporal evolution of a Gaussian input pulse at 1060 nm center wavelength with realistic parameters of a 10 m long ytterbium-doped ber amplier is plotted in Fig. 8. The initial pulses have a duration of TFWHM = 500 fs and energy Ei = 400 pJ. The ber amplier provides a gain of 30 dB (g = 0.7 m1 ), 2 = 0.02 ps2 m1 , and = 5.0 104 W1 m1 . The nal parabolic pulses have a pulse duration (FWHM) of 6.7 ps and a spectral width of 49.3 nm, corresponding to a time-bandwidth product of 84.6. It can be seen that the temporal shape is parabolic (Fig. 8(a)), which is also indicated by steep edges in the logarithmic spectrum (Fig. 8(b)). The setup to experimentally demonstrate this evolution into a linearly chirped parabolic pulse is shown in Fig. 9. The ber

LIMPERT et al.: HIGH-POWER ULTRAFAST FIBER LASER SYSTEMS

239

Fig. 9.

Experimental setup of parabolic pulse ber amplier.

amplier system consists of a passively mode-locked, diodepumped solid-state laser system, a diode-pumped ytterbiumdoped ber amplier, and a diffraction-grating compressor based on transmission gratings. As a femtosecond seed source, a Nd:glass laser system is applied, which is based on a semiconductor saturable absorber mirror. The laser is running at a 75 MHz repetition rate, producing pulses as short as 180 fs at 1060 nm center wavelength and an average power of 100 mW. The high power amplier is constructed using 9 m length of low-NA large-mode-area ber with a 30-m diameter, 0.06-NA step-index ytterbium-doped core, and a 400-m D-shaped inner cladding with NA = 0.38. Bending losses are used to extract fundamental mode beam quality from this ber. The combined action of gain, nonlinearity, and normal dispersion ensures the generation of linearly chirped parabolic pulses which experience large temporal and spectral broadening. The spectral broadening in the parabolic amplier is illustrated in Fig. 10, showing the emitted spectrum at different output powers in a logarithmic scale. The conversion of the spectrum of the initial sech2 pulse to the spectrum of a linearly chirped parabolic pulse can be identied and compared to the simulation in Fig. 8. At an average output power of 17 W, the spectral width has increased to 40.5 nm. Transmission gratings in fused silica are employed to remove the chirp of the high power parabolic pulses [39]. Fig. 11 shows the measured autocorrelation trace of the recompressed pulses at an output power of 17 W of the ber amplier. We determined the FWHM pulse duration p = 80 fs corresponding to a time-bandwidth product of 0.86. The low-intensity wings in the autocorrelation trace have their origin in uncompensated higher-order phase contributions of the ber amplier grating compressor setup. A compressor efciency of 60 % is reached, resulting in an average power of 10.2 W after compression, corresponding to a pulse energy of 140 nJ and a pulse peak power of 1.7 MW. B. Amplication of Parabolic Pulses in a Fiber Amplier As previously described, special pulse shapes, herein called parabolic pulses, can create themselves during amplication in normal dispersive amplier bers and overcome limitations due to nonlinear effects. Of course, these pulse shapes can also be externally generated and injected into the amplier. The most elegant way of doing this is to create parabolic pulses directly

Fig. 10. Experimentally obtained output spectra of ber amplier.

Fig. 11. Intensity autocorrelation trace of recompressed 80-fs pulses (dashed curve: t).

in a short pulse ber oscillator. Indeed, the similariton regime has been realized in ber oscillators [40]. Such a laser emits linearly chirped parabolic pulses which, due to their pulse shape, constitute an ideal seed source for high power ber ampliers. For this purpose, we developed a mode-locked allpolarization maintaining ber oscillator comprising a saturable

240

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO. 2, MARCH/APRIL 2006

Fig. 12.

Experimental setup of nonlinear regime parabolic pulse amplier.

Fig. 14. Measured autocorrelation trace of the recompressed pulses out of the nonlinear regime amplier. In comparison, simulation of recompressed amplied pulse is shown in the inset using chirped parabolic pulse (black) and chirped sech2 pulse (gray). Fig. 13. SEM image of Yb-doped air-clad photonic crystal ber with six index-matched stress applying parts.

absorber mirror (SAM) [41]. The SAM makes the oscillator self-starting, and the exclusive use of polarization maintaining bers provides environmental stability. Linearly chirped parabolic pulses with a spectral bandwidth of 10 nm around 1035 nm, about 0.2 nJ pulse energy, and a pulse repetition rate of 17 MHz are emitted. These pulses can be externally compressed to a pulse duration of 260 fs. The setup of the oscillator and the subsequent amplication stages are shown in Fig. 12. The output of the oscillator is preamplied in a ytterbiumdoped single-clad PM ber. The average power is boosted to 50 mW and, due to the additional ber length, the pulses are stretched to 10 ps. The power amplier consists of a ytterbium-doped single-polarization single-mode double-clad photonic crystal ber [42]. This special ber comprises indexmatched stress-applying elements (SAP) as part of the photonic cladding. A cross section of this ber is shown in Fig. 13.

The proximity of the SAP to the core ensures high differential stress, and thus high birefringence. Additionally, due to the index matching of the SAP array to the air-hole cladding, the light is conned by both parts of the photonic cladding: the air holes and the index matched regular arrays of SAPs. Furthermore, the amount of birefringence is enough to split two polarization states of the weakly guided fundamental mode in that way that the effective index of one polarization is below the cladding index. This results in a single-polarization large-mode-area ber in a certain wavelength range, which is determined by the design and, in our case, lies within the amplication bandwidth of Ytterbium. The mode eld area of 700 m2 is larger than that of any step-index single-mode ber. The pump cladding is formed by an air-cladding outside the inner photonic cladding, and has a diameter of 170 m and an NA of 0.6. This structure has a pump light absorption of 14 dB/m. A ber coupled diode laser emitting at 976 nm pumps the main amplier, which is built using 1.2 m of the Yb-doped

LIMPERT et al.: HIGH-POWER ULTRAFAST FIBER LASER SYSTEMS

241

Fig. 15.

Schematic setup of high average power ber CPA system. OI: optical isolator.

single-polarization ber. Up to 29 W average power with a slope efciency of 72% is obtained at a launched average seed power of 30 mW. The polarization extinction ratio is as high as 24 dB (1:250) even at the highest power level. Again, a pair of transmission gratings is applied to recompress the pulses. The grating separation is just 2 cm, and due to the perfect linear polarization of the amplied output the throughput efciency is 73%, resulting in 21 W of recompressed average power. The measured autocorrelation is shown in Fig. 14. The autocorrelation width is 320 fs, corresponding to a 240-fs pulse duration. The recompressed pulse quality is almost unaffected by the imposed nonlinear phase, even if the B-integral is in the range of 6. The slight reduction in pulse duration compared to the oscillator is made possible by the bandwidth enhancement in the amplication stages. At the highest power levels, wings in the autocorrelation trace are developing; however, they are containing only a minor part of energy. Consequently, the pulse peak power is as high as 5 MW. In comparison, a simulation of the system is done by solving the NLSE with constant gain. The resulting autocorrelation after recompression is shown in the inset of Fig. 14 for a chirped parabolic pulse and a comparable chirped sech2 pulse coming out of the laser. As shown, the quality of the output pulses is signicantly better when using a parabolic pulse shape. The autocorrelation of the amplied sech2 pulse has a considerable wing structure containing a substantial part of the pulse energy. To our knowledge, this is the rst experimental conrmation that parabolic pulses can lead to a signicant improve of pulse quality in ber amplier systems. Further power and energy scaling will be possible in the future by stretching the parabolic pulses to nanosecond duration, as described in the following experiments.

Fig. 16. Output characteristics of ytterbium-doped ber power amplier.

scaling potential of ber based chirped pulse amplication is discussed. A. High Average Power Fiber CPA System In a ber-based CPA system, sufcient pulse stretching and the enlargement of the mode-eld diameter of the ber to reduce the peak power, and therefore nonlinear effects such as the previously discussed effects of SRS and SPM, are the key points to scale the output parameters. The setup of the state-of-theart high average power femtosecond ber laser system [44] is shown in Fig. 15. It consists of a passively mode-locked solidstate laser system, a grating stretcher, a two-stage ytterbiumdoped single-mode photonic crystal amplier, and a fused-silica transmission grating compressor. The femtosecond seed source is a passively mode-locked Yb:KGW laser system producing 150 fs pulses at a 73 MHz repetition rate, and a 1040 nm center wavelength. These pulses are stretched to 120 ps using a conventional gold-coated 1200 lines/mm diffraction grating. The preamplier and the power amplier are constructed with air-clad photonic crystal bers possessing an active core diameter of 40 m (NA = 0.03) and an inner cladding diameter of 170 m (NA = 0.62). The used ber length is just 1.2 m. In the rst stage, the signal is amplied to 5 W, which is the seed of the power amplier. The output

V. SHORT PULSE AMPLIFICATION IN THE ABSENCE OF NONLINEARITY To signicantly scale up the average power or pulse energy of ber laser systems, the well-known technique of chirped pulse amplication (CPA) [43] must be employed. Stretching the pulses in the time domain reduces the pulse peak power and, consequently, nonlinear pulse distortion can be avoided to a certain extent. In the following sections, the achievements and

242

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO. 2, MARCH/APRIL 2006

Fig. 17. Measured autocorrelation trace of high average power femtosecond pulses. Fig. 18. Summary of all limiting effects in ber CPA system, assuming MFD of nal amplier of 50 m, and stretched pulse duration of 1 ns.

characteristic of the main amplier is shown in Fig. 16. Up to 175 W of average power is obtained with a slope efciency of 75% with respect to the launched pump power. The emitted beam quality is nearly diffraction limited (M2 < 1.2). The stretched and amplied pulses are recompressed using the transmission gratings in fused silica, and are almost matched to the grating stretcher by having a period of 1250 lines/mm. Due to the higher average power damage threshold compared to conventional gold-coated diffraction gratings, these gratings are again one of the key elements of the high average power ber CPA system. Fig. 17 shows the measured autocorrelation trace at the highest output power of the ber amplier. The best compression is achieved at a grating separation of 0.90 m. This optimal grating distance stayed unchanged with increasing power. This means that the inuence of nonlinearity can be neglected, which was also conrmed by numerical simulations. The autocorrelation width is determined to be 340 fs, corresponding to 220-fs pulse duration (FWHM), assuming a sech2 pulse shape. The pulse duration is constant over the whole power range, no limitations or pulse quality degradation are observed. The difference in pulse duration between the oscillator and the recompressed pulses is due to slightly mismatched gratings in the stretcher and compressor. The measured compressor throughput efciency is 75%, resulting in 131 W average power of femtosecond pulses. The peak power is as high as 8.2 MW. To our knowledge, this is the highest average power ever reported for ultrashort pulse solid-state laser systems. The achieved results were only limited by the available pump power and not by nonlinear pulse distortions; therefore, further power scaling is possible even with the current system. The recent demonstration of >1 kW output of a continuous-wave ber laser with nearly diffraction-limited beam quality endorses this statement. Furthermore, the spectral width of the oscillator is completely supported by the amplier system and stayed unchanged at about 8 nm at all power levels; therefore, even sub-100 fs pulses are obtainable just by employing an oscillator which delivers shorter pulses.

B. High-Energy Fiber CPA System The generation of high-energy pulses from a ber based laser system is certainly the most challenging task. Even signicant stretching of pulses cannot avoid the formation of enormous peak power levels in the ber amplication stages. As discussed in Section II, ber facet damage can be prevented by proper ber preparation. Thus, nonlinear pulse distortion is the most challenging scaling limitation. The experimental setup of a high-energy ber CPA system is quite similar to that presented in the previous section. Pulses from a femtosecond oscillator are stretched, amplied by the high single-pass gain in a certain number of ber ampliers, and nally recompressed with a pair of gratings or other suitable dispersive elements. The main difference is the larger necessary stretching factor and the employment of an acoustooptical modulator to reduce the pulse repetition rate. Consequently, the pulse energy is increased if the average power is kept high. Pursuing this approach, we realized a ber CPA system producing 100 J at a 200-kHz repetition rate corresponding to an average power of 20 W. The pulse duration of the recompressed pulses is 800 fs. The system employs identical bers as used in the high average power ber CPA system; i.e., a 40 m core single-transverse-mode ytterbium-doped photonic crystal ber. The pulse duration during amplication is 700 ps. It is noteworthy that the average power is only limited by the use of gold-coated gratings in the compressor; thus, an upscaling is possible by employment of dielectric gratings, as previously described. To show the pulse energy scalability, Fig. 18 summarizes all of the important restricting effects in a ber CPA system using a ber with 50 m mode-eld diameter in a nal amplication stage and a stretched pulse duration of 1 ns. As can be seen, the surface damage limits the pulse energy to few 100 J if no mode expansion is applied. By preparation of the ber with a coreless endcap, the damage threshold can be increased to well above 10 mJ. A fundamental limitation is self-focusing; i.e., the spatial Kerr effect. The critical peak power for self-focusing in fused silica is about 3.8 MW [45]. The assumed pulse duration

LIMPERT et al.: HIGH-POWER ULTRAFAST FIBER LASER SYSTEMS

243

of 1 ns translates this to 3.8 mJ of energy. As shown in Fig. 18, the nonlinear effects of SPM and SRS are occurring before any other limitations. However, the mJ-level can be reached by using bers with extremely low nonlinearity. If a modeeld diameter of 50 m is assumed, a ber length as short as 0.5 m is necessary to avoid a B-integral larger than 1. Such a ber is exactly the design referred to as a rod-type photonic crystal ber in Section III. As previously discussed, pure linear amplication is not mandatory. By choosing a suitable pulse shape; e.g., parabolic pulses, an accumulated nonlinear phase of several can be handled without pulse quality degradation. Thus, even multi-mJ femtosecond pulses are feasible. As a result, we are convinced that we can demonstrate a ber-based CPA system delivering 100 W average power at 100 kHz repetition rate, corresponding to 1 mJ pulse energy and subpicosecond pulse duration, in the near future. Such a laser system would be an ideal source for a number of interesting applications, such as high-speed high precision micromachining of various solid materials. VI. CONCLUSION We reviewed the main advantages of ultrafast ber laser systems compared to bulk solid-state lasers. Their heat dissipation capability, very high single-pass gain, broad gain bandwidth, compactness, robustness, and simplicity of operation make ber lasers attractive for a host of applications. In particular, rareearth-doped photonic crystal bers offer several unique properties, which allow an upward scaling of performance compared to conventional ber lasers. Their extended optical parameter range and the transfer of additional functionality to the ber by microstructuring indicate that such systems have an enormous potential to scale the performance of next-generation laser systems. To prove this statement, we demonstrated high power and high energy photonic crystal ber laser and amplier systems using the advantages of an air-clad PCF with very high pump core NA, and an extended mode eld area of an intrinsically single-mode core. Furthermore, we summarized the limitations of ultrafast ber laser systems. Besides the innovative ber designs, experimental strategies were discussed to push these restrictions as far as possible. In conclusion, in the near future, such short pulse ber laser systems will deliver very high average powers, and therefore high repetition rates, in combination with high pulse energies, so that applications demanding high processing speed such as high precision micromachining will be in reach.

REFERENCES
[1] M. E. Fermann, A. Galvanauskas, and G. Sucha, Ultrafast Lasers. New York: Marcel Dekker, 2002. [2] P. F. Moulton, Tunable solid-state lasers, Proc. IEEE, vol. 80, no. 3, pp. 348364, Mar. 1992. [3] J. M. Eggleston, T. J. Kane, K. Kuhn, R. L. Byer, and J. Unternahrer, The slab geometry laser ITheory, IEEE J. Quantum Electron., vol. 20, no. 3, pp. 289301, Mar. 1984. [4] A. Giesen, H. H gel, A. Voss, K. Wittig, U. Brauch, and H. Opower, u Scalable concept for diode pumped high-power solid state lasers, Appl. Phys., vol. B58, pp. 365372, 1994.

[5] R. Paschotta, J. Nilsson, A. C. Tropper, and D. C. Hanna, Ytterbiumdoped ber ampliers, IEEE J. Quantum Electron., vol. 33, no. 7, p. 1049, Jul. 1997. [6] E. Snitzer, H. Po, F. Hakimi, R. Tumminelli, and B. C. McCollum, Double-clad, offset core Nd ber laser, in Opt. Fiber Sens., Opt. Soc. Amer. Tech. Dig. Ser., vol. 2, Postdeadline Paper PD5, 1988. [7] J. Limpert, A. Liem, H. Zellmer, and A. T nnermann, 500 W continuousu wave bre laser with excellent beam quality, Electron. Lett., vol. 39, p. 645, 2003. [8] Y. Jeong, J. K. Sahu, D. N. Payne, and J. Nilsson, Ytterbium-doped large-core ber laser with 1.36 kW continuous-wave output power, Opt. Express, vol. 12, pp. 60886092, 2004. [9] G. Bonati, H. Voelckel, T. Gabler, U. Krause, A. T nnermann, J. Limpert, u A. Liem, T. Schreiber, S. Nolte, and H. Zellmer, 1.53 kW from a single Yb-doped photonic crystal ber laser, presented at the Photonics West, Late Breaking Developments, Session 5709-2a, San Jose, CA, 2005. [10] A. T nnermann, T. Schreiber, F. R ser, A. Liem, S. H fer, H. Zellmer, u o o S. Nolte, and J. Limpert, The renaissance and bright future of bre lasers, J. Phys. B, At. Mol. Opt. Phys,., vol. 38, pp. 681693, 2005. [11] G. P. Agrawal, Nonlinear Fiber Optics. San Diego, CA: Academic, 1995. [12] R. R. Alfano, The Supercontinuum Laser Source. New York: SpringerVerlag, 1989. [13] M. Perry, T. Ditmire, and B. Stuart, Self-phase modulation in chirpedpulse amplication, Opt. Lett., vol. 19, pp. 21492151, 1994. [14] W. K chner, Solid-State Laser Engineering. Berlin, Germany: Springero Verlag, 1999. [15] D. Gloge, Weakly guiding bers, Appl. Opt., vol. 10, pp. 22522258, 1971. [16] J. A. Alvarez-Chavez, H. L. Offerhaus, J. Nilsson, P. W. Turner, W. A. Clarkson, and D. J. Richardson, High-energy, high-power ytterbiumdoped Q-switched ber laser, Opt. Lett., vol. 25, p. 37, 2000. [17] J. M. Sousa and O. G. Okhotnikov, Multimode Er-doped ber for singletransverse-mode amplication, Appl. Phys. Lett., vol. 74, pp. 15281530, 1999. [18] J. Limpert, H. Zellmer, A. T nnermann, T. Pertsch, and F. Lederer, Supu pression of higher order modes in a multimode ber amplier using efcient gain-loss-managment (GLM), presented at the Advanced Solid State Lasers 2002, Queb c City, QC, Canada, Paper MB20. e [19] M. Fermann, Single-mode excitation of multimode bers with ultrashort pulses, Opt. Lett., vol. 23, pp. 5254, 1998. [20] J. A. Alvarez-Chavez, A. B. Grudinin, J. Nilsson, P. W. Turner, and W. A. Clarkson, Mode selection in high power cladding pumped bre lasers with tapered section, in Conf. Lasers and Electro-Optics, OSA Tech. Dig., 1999, p. 247. [21] A. Galvanauskas, Mode-scalable ber-based chirped pulse amplication systems, IEEE J. Sel. Topics Quantum Electron., vol. 7, no. 4, pp. 504 517, Jul.Sep. 2001. [22] P. Koplow, D. Kliner, and L. Goldberg, Single-mode operation of a coiled multimode ber amplier, Opt. Lett., vol. 25, pp. 442444, 2000. [23] J. I. Sakai and T. Kimura, Bending loss of propagation modes in arbitraryindex prole optical bers, Appl. Opt., vol. 17, pp. 14991506, 1978. [24] J. P. Russell, Photonic crystal bers, Science, vol. 299, pp. 358362, 2003. [25] J. Knight, T. Birks, P. Russell, and D. Atkin, All-silica single-mode optical ber with photonic crystal cladding, Opt. Lett., vol. 21, p. 1547, 1996. [26] T. Birks, J. Knight, and P. Russell, Endlessly single-mode photonic crystal ber, Opt. Lett., vol. 22, pp. 961963, 1997. [27] N. A. Mortensen and J. R. Folkenberg, Low-loss criterion and effective area considerations for photonic crystal bers, J. Opt. A, Pure Appl. Opt., vol. 5, p. 163, 2003. [28] N. A. Mortensen, M. D. Nielsen, J. R. Folkenberg, A. Petersson, and H. R. Simonsen, Improved large-mode-area endlessly singlemode photonic crystal bers, Opt. Lett., vol. 28, p. 393, 2003. [29] J. Limpert, A. Liem, T. Schreiber, S. Nolte, H. Zellmer, A. T nnermann, u J. Broeng, A. Petersson, C. Jacobsen, H. Simonsen, and N. A. Mortensen, Extended large-mode-area single mode microstructured ber laser, presented at the Conf. Lasers and Electro-Optics 2004, San Francisco, CA, Session CMS, Paper CMS6. [30] W. J. Wadsworth, R. M. Percival, G. Bouwmans, J. C. Knight, and P. S. J. Russell, High power air-clad photonic crystal bre laser, Opt. Express, vol. 11, pp. 4853, 2003. [31] J. Limpert, T. Schreiber, S. Nolte, H. Zellmer, T. T nnermann, R. Iliew, u F. Lederer, J. Broeng, G. Vienne, A. Petersson, and C. Jakobsen,

244

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 12, NO. 2, MARCH/APRIL 2006

[32]

[33]

[34]

[35] [36] [37] [38] [39]

[40] [41] [42]

[43] [44] [45]

High-power air-clad large-mode-area photonic crystal ber laser, Opt. Express, vol. 11, pp. 818823, 2003. J. Limpert, A. Liem, M. Reich, T. Schreiber, S. Nolte, H. Zellmer, A. T nnermann, J. Broeng, A. Petersson, and C. Jakobsen, Lowu nonlinearity single-transverse-mode ytterbium-doped photonic crystal ber amplier, Opt. Express, vol. 12, pp. 13131319, 2004. J. Limpert, N. Deguil-Robin, I. Manek-H nninger, F. Salin, F. R ser, o o A. Liem, T. Schreiber, S. Nolte, H. Zellmer, A. T nnermann, J. Broeng, u A. Petersson, and C. Jakobsen, High-power rod-type photonic crystal ber laser, Opt. Express, vol. 13, pp. 10551058, 2005. J. Limpert, T. Schreiber, A. Liem, S. Nolte, H. Zellmer, T. Peschel, V. Guyenot, and A. T nnermann, Thermo-optical properties of air-clad u photonic crystal ber lasers in high power operation, Opt. Express, vol. 11, pp. 29822990, 2003. E. B. Treacy, Optical pulse compression with diffraction gratings, IEEE J. Quantum Electron., vol. 5, no. 9, p. 454, Sep. 1969. D. Anderson, M. Desaix, M. Karlson, M. Lisak, and M. L. QuirogaTeixeiro, Wave-breaking-free pulses in nonlinear-optical bers, J. Opt. Soc. Amer. B, Opt. Phys., vol. 10, p. 1185, 1993. K. Tamura and M. Nakazawa, Pulse compression by nonlinear pulse evolution with reduced optical wave breaking in erbium-doped ber ampliers, Opt. Lett., vol. 21, no. 1, p. 68, 1996. V. I. Kruglov, A. C. Peacock, J. D. Harvey, and J. M. Dudley, Self-similar propagation of parabolic pulses in normal-dispersion ber ampliers, J. Opt. Soc. Amer. B, Opt. Phys., vol. 19, p. 461, 2002. T. Clausnitzer, J. Limpert, K. Z llner, H. Zellmer, H. J. Fuchs, E. B. Kley, o A. T nnermann, M. Jupe, and D. Ristau, Highly efcient transmission u gratings in fused silica for chirped-pulse amplication systems, Appl. Opt., vol. 42, pp. 69346938, 2003. F. O. Ilday, J. R. Buckley, W. G. Clark, and F. W. Wise, Self-similar evolution of parabolic pulses in a laser, Phys. Rev. Lett., vol. 92, p. 213902, 2004. C. K. Nielsen, B. Ortac, T. Schreiber, J. Limpert, R. Hohmuth, W. Richter, and A. T nnermann, Self-starting self-similar all-polarization maintainu ing Yb-doped ber laser, Opt. Express, vol. 13, pp. 93469351, 2005. T. Schreiber, F. R ser, O. Schmidt, J. Limpert, R. Iliew, F. Lederer, A. Peo tersson, C. Jacobsen, K. P. Hansen, J. Broeng, and A. T nnermann, Stressu induced single-polarization single-transverse mode photonic crystal ber with low nonlinearity, Opt. Express, vol. 13, pp. 76217630, 2005. D. Strickland and G. Mourou, Compression of amplied chirped optical pulses, Opt. Commun., vol. 55, pp. 447449, 1985. F. R ser, J. Rothhard, B. Ortac, A. Liem, O. Schmidt, T. Schreiber, o J. Limpert, and A. T nnermann, 131 W 220 fs ber laser system, Opt. u Lett., vol. 30, pp. 27542756, 2005. G. Fibich and A. Gaeta, Critical power for self-focusing in bulk media and in hollow waveguides, Opt. Lett., vol. 25, pp. 335337, 2000.

Fabian R ser was born in Saalfeld, Germany, on o September 17, 1977. He received the Diploma degree in general physics from the Friedrich Schiller University Jena, Jena, Germany, in 2003. Currently, he is working toward the Ph.D. degree in the Laser Development Group at the Institute of Applied Physics, Friedrich Schiller University Jena. His research works include high-power ber lasers and ampliers, focusing on ultrashort pulse generation and amplication. Mr. R ser is a member of the German Physical o Society and the Optical Society of America.

Thomas Schreiber was born in Gera, Germany, on September 24, 1976. He received the Diploma degree in general physics from the Friedrich Schiller University Jena, Jena, Germany, in 2001. Currently, he is working toward the Ph.D. at the Laser Development Group, Institute of Applied Physics, Friedrich Schiller University Jena. He worked with uorescence lifetime imaging as a biology application of laser physics with Prof. P. French at Imperial College, London, U.K., in 2003. His research work includes ber lasers and ampliers, especially applications of photonic crystal bers in the eld of ultrafast optics, and modeling ultrashort pulse propagation. Mr. Schreiber is a member of the German Physical Society and the Optical Society of America.

Jens Limpert was born in Jena, Germany, on December 10, 1975. He received the Diploma degree and the Ph.D. degree in general physics from the Friedrich Schiller University Jena, Jena, Germany in 1999 and 2003, respectively. After a one-year Postdoctoral position at the University of Bordeaux, France, he is currently leading the Laser Development Group at the Institute of Applied Physics, Friedrich Schiller University Jena. He has published over 50 conference and journal papers in the eld of ber lasers. His research interests include high-power ber lasers in the pulsed and continuous-wave regime, in the near-infrared and visible spectral range. Dr. Limpert is member of the German Physical Society and the Optical Society of America.

Andreas Tunnermann (M95A96) was born in Ahnsen, Germany on June 10, 1963. He received the Diploma and Ph.D. degrees in physics from the University of Hannover, Hannover, Germany, in 1988 and 1992, respectively. In 1997, he received the habilitation for his work on ultrastable light sources for interferometric gravitational wave detectors. He was the Head of the Department of Development at the Laser Zentrum Hannover, Hannover, Germany, from 1992 to 1997. In 1994, he became the National Scientic Coordinator for the LASER 2000 program. In 1998, he joined the Friedrich Schiller University Jena, Jena, Germany, as a Professor. Since 2003, he has been the Director of the Institute of Applied Physics and also the Director of Fraunhofer Institute for Applied Optics and Precision Engineering, Jena, Germany. His main research interests include scientic and technical aspects associated with the tailoring of light, and the design and manufacturing of novel micro- and nanooptical photonic devices using high-end microlithography. Prof. T nnermann is a member of the German Physical Society, the Optical u Society of America, and the Institute of Electrical and Electronics Engineers. He received the Roentgen Award in 1997, the WLT Award in 1998, the Otto Schott Award in 2003, the Leibinger Innovation Award in 2004, and the Leibniz Award in 2005.