A new reduced mode overlap (RMO) single mode fiber design is proposed and demonstrated. For the first time saturated photo darkening operation is observed in a nominal 350 W Yb / Al co-doped silica fiber laser. After 1500 hours operation less than 7% slope efficiency degradation is found and further 500 hours operation show no degradation.
©2009 Optical Society of America
Ytterbium doped double-clad fiber sources have become increasingly popular because of their high output beam quality and high output powers. The single aperture output powers of Yb-doped double-clad fibers have steadily increased over the past years, rising to the multiple kW level [1,2]. As output power levels continue to increase, photo darkening (PD) has increasingly become an obstacle for further progress especially with regard to long term reliability. PD is observed as increase in propagation loss of rare earth doped glass material as a function of time. The underlying physical process for PD has been attributed to formation of color centers by photo ionization. For unseeded amplifiers the process is found to involve co-operative action of four to five excited ytterbium atoms . The color centers show absorption in the visible with long tails stretching into the near infrared leading to absorption both at pump and signal wavelength for ytterbium co-doped glass material . Direct scaling of the results from un-seeded amplifiers would for cw fiber lasers suggest that the low population inversion during operation should not give raise to problems with PD. This is, however, as shown in  not the case for Yb, Al co-doped cw fiber lasers operating at output power levels exceeding 200 W. On the other hand, report of PD degradation-free Yb, phosphate glass has been published . This result is obtained in un-seeded amplifiers under high population inversion where PD is known to be high in Yb, Al co-doped silica glasses. The slope efficiency under operation was, however, not reported in . For Yb, Al co-doped silica glasses launched pump power to output power “fiber slope efficiency” in excess of 70% is standard. In  this type of slope efficiency is found to be in excess of 60% for phosphate glasses co-doped with ytterbium. This suggests that for less than 10% reduction in slope efficiency due to photo darkening the Yb, Al co-doped silica glass material performs more efficiently than the phosphate glass. This fact becomes increasingly important as the output power enters the kW regime where a few percent extra loss represents considerable excess heat load on the laser system.
A ring design similar to the one proposed in this work is theoretically analyzed for a three-level fiber laser in . It is found to be efficient in suppressing amplified spontaneous emission and unwanted parasitic defects. The benefits in connection with a four-level fiber laser operation with output at wavelengths beyond 1050 nm is not considered in  and neither is the benefit in reducing the parasitic effect of PD developing over time on fiber lasers and amplifiers.
In this work a new fiber design with reduced mode overlap (RMO) is presented and tested for four-level fiber lasers and amplifiers. The RMO design is based on Yb, Al co-doped nanostructure glass material. A long term test with nominal 350 W output power and constant pump power show significant improvement compared with traditional fiber signal overlap designs made in similar glass material.
2. New RMO design
The RMO design is based on a single mode, passive, Ge co-doped silica core with Yb, Al co-doped nanostructure material index matched to silica placed in a ring around the core. The core and ring is placed in a second core structure surrounded by air-cladding to provide good guidance for the pump light. The air clad is made of 120 holes placed in a ring around the second core structure. The air clad is again surrounded by an outer silica ring which is protected by a thin polymeric coating. Cross section of the RMO design is shown in Fig. 1a ) and a schematic refractive index profile of the RMO fiber design is shown in Fig. 1b).
The nanostructure material is illustrated in the insert and is based on Yb, Al co-doped silica merged into F co-doped silica material. The nanostructure material is derived through a multiple stack and draw process . The individual Yb, Al co-doped areas become hereby less than ≈100 nm in diameter as indicated in Fig. 1. The overall area integration of the nanostructure material yield an average refractive index matched to silica which becomes a direct function of the Yb, Al co-doped area in F co-doped silica material dilution factor.
The nanostructure islands are “invisible” to the propagating field due to their small dimensions and the nanostructure holds high flexibility with regard to both refractive index choices and uniformity. The ring structure yields a reduced signal mode overlap to the active material which lends it the name RMO design. The pump overlap to the active material is unchanged with unchanged pump absorption as result. In the present version the core is made in Ge co-doped silica such that gratings can be written directly in the core. It will, however, be advantageous to apply pure silica cores surrounded by an index depressed Yb, Al co-doped ring and F co-doped pump cladding for high power booster amplifiers.
3. Experimental results from traditional fiber designs and the RMO fiber design
This section reports 3 different rare earth material compositions which placed in nanostructure core designs results in 0.7 – 0.9 dB/m pump absorption (at 915 nm) for a 250 µm air clad pump core as given in Table 1 . For fiber 1 and 2 the rare earth nanostructure is part of the index guiding single mode core (with a dilution factor that produces NA of 0.06), whereas fiber 3 is the new single mode design described in section 2 with the nanostructure as part of the cladding (and a dilution factor that yield refractive index match to silica). The numeric aperture of all pump cores is ≈0.6 and signal mode to active material area overlap ΓSignal is calculated in Table 1.
PD absorption per length in unseeded amplifiers operated with uniform population inversion through 40 cm un-seeded amplifier fiber sections are shown in Fig. 2 a ) for the three fibers measured at various pump power levels. The absorption spectrum is measured through the pump core for 1 min while the pump power is shut off for every 15 min during 46 hours total measurement time. Further details for the accelerated PD test are referred to . The measured data is fitted with the stretched exponential function α(t) = αeq.PD [1 – exp(-(t/τ)β)], for which the applied parameters αeq.PD, τ and β are given for the individual fits in Fig. 2 a). The equilibrium loss αeq.PD is shown in Fig. 2 b) as function of pump power. Here the characterized values are plotted along with their PD model predictions .
It is from Fig. 2 b) to be observed that the saturated PD level at 605 nm scales with the local ytterbium concentration [Yb] (counting oxygen) and a considerable improvement in PD performance between fiber 1 and fiber 2 is obtained. The decrease in equilibrium PD level scales linearly with the decrease in local ytterbium concentration. Due to the increase in Yb co-doped area fraction of the total core area the pump absorption is unchanged between fiber 1 and fiber 2. The fiber 3 material is from Fig. 2 b) found to perform with slightly improved PD performance over the fiber 2 material in good agreement with the findings of . Here an increase in [Al] to [Yb] ratio is found to improve PD performance.
In Fig. 3 long term test of master oscillator power amplifier (MOPA) devices established in the 3 fibers are shown. The MOPA device lengths are chosen to hold ≈13 dB pump absorption leading to an almost identical number of ytterbium ions in the three devices.
The MOPA devices 13 dB pump absorption is divided in master oscillator and power amplifier lengths as indicated in Table 1. The output coupler of the master oscillator is ≈1 dB reflectivity fiber Bragg grating written directly in the Ge co-doped core for all systems. The pump stack is in the test coupled to the MOPA through a 1:1 power combiner in a co-propagating single side pump configuration. The output is coupled directly to a power meter. The test is performed with constant pump power such that the measured output is a direct measure for the device degradation during the test. The “fiber slope efficiency” is determined for the various MOPA devices as the ratio of output power to coupled in pump power – which yields >70% for all systems.
The long term high power test with initial 350 W output power for constant pump power of Fig. 3 show that both fiber 1 and fiber 2 performs with a reduction in “fiber slope efficiency” from about 73% to 50% during the first ≈1000 hour operation. This is remarkable because the material of fiber 2 performs considerable better in the accelerated PD test compared with fiber 1 as shown in Fig. 2 b). From the high power PD model predictions  superimposed the measured data the reason for the small realized difference is found to be the 43% signal to active material overlap of fiber 2 compared with 30% for fiber 1. The higher signal to active material overlap of fiber 2 gives a considerable higher effective contribution of PD. This even though the amount of PD centers generated in fiber 2 is less than in fiber 1 as they both works at equal population inversion levels. Based on this observation – which is found to hold true for all previously realized MOPA fiber designs – the new RMO fiber core design is proposed and tested in fiber 3. This RMO fiber design (fiber 3) show less than 7% slope efficiency degradation when the equilibrium state (saturated PD) is reached after 1500 hours operation. The saturation is to be observed in the final 500 hours operation with no change in output power. The pump power was maintained constant during all 2000 hours operation. This is to the best of our knowledge the first time PD is observed to saturate for a 350 W output power fiber laser based on Yb, Al co-doped silica.
4. Discussion of the RMO fiber design
The RMO design appears to present by concept a contradiction to the general belief that the effect of PD scales with population inversion. The small signal gain of active fiber material is known to scale directly with the signal to active material overlap. A decrease in signal overlap will force the fiber material to operate with proportional higher population inversion to produce equal output as found in Table 1 for fiber 1and 2 versus 3 for 350 W output power in pristine systems. This should result in increased PD for reduced signal to active material overlap – which it does. However, the effective PD scales with population inversion as well as power intensity of pump and signal. For high power systems it has been suggested that PD saturates dependent on the system output power . Absorption of the color centers is here given by Eq. (1):Eq. (1) isolated this again point in the opposite direction of the RMO design in that placing the active material in areas with low effective power intensity will drive the color centers more out of saturation, which should lead to higher effective absorption. However, the lifetime of the color center is in the model predicted to follow an exponential function of phonon density as given by Eq. (2):Eq. (3):
The postulate of the PD model [5,8] is that color centers reside on non-binding oxygen atoms found on cluster surfaces of Yb, Al complexes included in the silica network. Color centers are created by the action of two pump photons in collaboration with phonons produced by excited Yb. An unstable color center comprises a complex of three oxygen atoms where an electron is transferred between to neighbor oxygen atoms. The complex –O: O – O+ holds an electron trap on –O and a color center on the O – O+ complex. The lifetime of the color center is determined by the relaxation time for the O(1D) – O(3P) forbidden transition that assisted by a coulomb field gives away a continuum of states up to ≈2 eV. The relaxation time is determined by the local phonon density on the cluster surface because the transition from the O(1D) excited state to the O(3P) ground state requires impulse change to become allowed. In this process the hole of the color center either remains where it is or shifts back to the neighbor oxygen. Steady state is reached with a given number of color centers once the creation rate for electron traps and color centers matches the bleaching rate by the phonons produced in the color center.
For low power intensity the phonon density on cluster surfaces is dominated by excited Yb ions relaxing part of their energy to phonons hereby influencing the lifetime of color centers. For high intensity the phonon production of color centers by far dominates the local phonon density and hereby the lifetime of excited color centers. The model works with two cluster types: 1) with one Yb atom and 2) with several Yb atoms. The distribution of Yb among these types of clusters is highly influenced by the addition of aluminum as is the average cluster size.
It could be speculated that the RMO design gives away fewer co-operative emitted photons which could be an alternative explanation for the reduced PD . This effect is, however, not included in the model predictions of Fig. 2 and Fig. 3 and is considered to play a minor role for PD for output signals in the near infrared at wavelengths < 1150 nm.
To convert the achieved results in this work to real system applications the “fiber slope efficiency” has to be corrected for diode laser to pump fiber coupling efficiency, pump fiber combiner coupling efficiency and splice loss to yield the optical “system slope efficiency”. This gives with present available diodes and (61:1) combiners a pristine “system slope efficiency” of ≈65% for the RMO fiber system (fiber 3) that is expected to saturate at ≈58% after 1500 hours operation. Often numbers exceeding 80% is quoted for the slope efficiency of ytterbium co-doped fibers. This holds also true for “deposited pump” to output power slope efficiency of a pristine RMO fiber system. The deposited pump is coupled in pump subtracted residual pump at the output. This is, however, misleading as the residual pump usually is of no practical use as also is the case for lost pump power and signal power in various connections.
In this work it is shown that PD of ytterbium co-doped silica glass material in cw fiber lasers can be alleviated by fiber design. By reducing the signal mode to active material overlap the parasitic losses by PD can be reduced at limited expense for the saturated gain of cw fiber laser and amplifiers.
For the first time saturated PD operation is observed after 1500 hours operation with less than 7% slope efficiency degradation from nominal 350 W output power in Yb, Al co-doped material operated under constant pump power in a reduced mode overlap (RMO) fiber design. Beyond 1500 hours (up to 2000 hours) operation no further degradation is observed.
The obtained result with the RMO design is surprising in that by reducing the signal to active material overlap the population inversion for the device under operation is increased proportional with the decrease, which is expected to give raise to increased PD.
The benefit of the RMO design is to be explained in that the generated power of the saturated fiber laser is primarily carried in the passive core while the active part amplifies the tails of the field. The effect of PD from the generated signal field is reduced with the signal overlap. This leads to a reduced phonon production by color centers. The lifetime of the color center excited state is modeled as an exponential function of the local phonon density which gives a considerable increase in excited state lifetime through reduced phonon density. As there is a limited number of color centers in the material a longer excited state lifetime is found to lead to a reduced effective absorption by the color centers as they saturates with beneficial effect on PD.
References and links
1. A. Tünnermann, T. Schreiber, F. Röser, A. Liem, S. Höfer, H. Zellmer, S. Nolte, and J. Limpert, “The renaissance and bright future of fibre lasers,” J. Phys. At. Mol. Opt. Phys. 38(9), S681–693 (2005). [CrossRef]
2. D. Gapontsev, “6 kW CW single mode ytterbium fiber laser in all-fiber format”, Conf. on Solid State and Diode Laser Tech. Review, Directed Energy professional society, Albuquerque, New Mexico (2008)
4. J. Koponen, M. Söderlund, H. J. Hoffman, D. Kliner, and J. Koplow, “Photodarkening Measurements in large mode area Fibers”, Proc. SPIE, Vol. 6453, 645350 (2007)
5. K. E. Mattsson and J. Broeng, “Photo darkening of ytterbium cw fiber lasers,” Proc. SPIE 7195, 71950V (2009). [CrossRef]
6. Y. W. Lee, S. Sinha, M. J. F. Digonnet, R. L. Byer, and S. Jiang, “Measurement of high photodarkening resistance in heavily Yb3+-doped phosphate fibres,” Electron. Lett. 44(1), 14–16 (2008). [CrossRef]
7. W. J. Wadsworth, J. C. Knight, and P. St. J. Russell, “Large mode area photonic crystal fibre laser”, Proc. of CLEO, CWC1 (2001)
8. K. E. Mattsson, S. N. Knudsen, B. Cadier, and T. Robin, “Photo darkening in ytterbium co-doped silica material”, Proc. SPIE, Vol. 6873, 68731C (2008)
9. S. Yoo, C. Basu, A. J. Boyland, C. Sones, J. Nilsson, J. K. Sahu, and D. Payne, “Photodarkening in Yb-doped aluminosilicate fibers induced by 488 nm irradiation,” Opt. Lett. 32(12), 1626–1628 (2007). [CrossRef] [PubMed]
10. I. A. Bufetov, S. L. Semenov, A. F. Kosolapov, M. A. Mel’kumov, V. V. Dudin, B. I. Galagan, B. I. Denker, V. V. Osiko, S. E. Sverchkov, and E. M. Dianov, “Ytterbium fibre laser with a heavily Yb3+ -doped glass fibre core,” Quantum Electron. 36(3), 189–191 (2006). [CrossRef]
11. J. Nilsson, J. D. Minelly, R. Paschotta, A. C. Tropper, and D. C. Hanna, “Ring-doped cladding-pumped single-mode three-level fiber laser,” Opt. Lett. 23(5), 355–357 (1998). [CrossRef]
12. T. Kitabayashi, M. Ikeda, M. Nakai, T. Sakai, K. Himeno and K. Ohashi, “Population Inversion Factor Dependence of Photodarkening of yb-doped Fibers and its Suppression by highly Aluminum Doping”, Optics Letters, OFC/NFOEC 2006, OThC5, (2006)