- Email: [email protected]
Optical properties of Yb+ 3-doped fibers and fiber lasers at high temperature
Optics Communications 284 (2011) 5774–5780
Contents lists available at SciVerse ScienceDirect
Optics Communications journal homepage: www.elsevier.com/locate/optcom
Optical properties of Yb + 3-doped fibers and fiber lasers at high temperature S.W. Moore a,⁎, T. Barnett a, T.A. Reichardt a, R.L. Farrow b a b
Sandia National Laboratories, P. O. Box 969, MS 9056, Livermore, CA 94551, USA JDSU,430 North McCarthy Boulevard, Milpitas, CA 95035, USA
a r t i c l e
i n f o
Article history: Received 12 July 2011 Received in revised form 24 August 2011 Accepted 26 August 2011 Available online 12 September 2011 Keywords: Spectroscopy Lasers Emission
a b s t r a c t Recent advances in power scaling of Yb+ 3-doped fiber lasers to the kilowatt level suggest a need to examine the performance of Yb+ 3-doped silica at temperatures well above ambient. We report experimental results for the absorption coefficient, emission cross-section, fluorescence lifetime, and slope efficiency of a Yb3+-doped large mode area (LMA) silica fiber for temperatures spanning 23 °C–977 °C. To the best of our knowledge these are the highest temperatures to date for which these optical properties have been measured. We find a sharp reduction in the energy storing capability and lasing performance of Yb+ 3:SiO2 above 500 °C that coincides with the onset of non-radiative transitions in the excited state manifold (thermal quenching). As the temperature increases from room temperature to 977 °C, absorption in the 1020–1120 nm operating band increases monotonically, concurrent with a reduction in absorption at the 920-nm and 977-nm pumping bands. Conversely, the spectral weight of the emission cross-section shifts from transitions above 1010 nm to those below, with the exception of the 977-nm emission band. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Advances in cladding-pumped, kilowatt-class, Yb-doped fiber laser systems have surged in recent years as these devices have proven especially suitable for material processing. The large surface area-to-volume ratio, small quantum defect (975-nm pumping vs. 1064- to 1100-nm lasing), and high slope efficiency of commercially available Yb-doped silica fiber reduces thermal management and operating costs while allowing for very high wall-plug efficiencies. Still, despite the intrinsic advantages in thermal management offered by fibers, a significant rise in core temperature is expected for fiber lasers and amplifiers operating at the 1–5 kW level. The operating temperature of standard Yb-doped silica fiber is limited by the protective acrylate/fluoroacrylate coating, which begins to degrade and eventually fails below 200 °C. Studies to date concerning the optical properties and performance of rare-earth doped glass oxide and crystal laser and amplifier systems have been conducted at or below this temperature [1–6]. However, to further reduce the cost of thermal management or, alternatively, to operate fiber lasers and amplifiers in high temperature environments by using cladding materials that can withstand higher temperatures such as polyimide, then it is important to evaluate the optical properties and performance of Yb+ 3-doped silica at temperatures exceeding 200 °C. In this paper we report the temperature dependence of the radiative lifetime, emission cross-section, absorption coefficient, and slope efficiency of Yb+ 3-doped (2.4× 1020ions/m3) large mode area (LMA) fiber
from 23 °C to 977 °C. We find that absorption in the 1020–1120 nm operating regime increases rapidly while absorption at the 920-nm and 977-nm pump bands decreases steadily with increasing temperature. For temperatures exceeding 500 °C, a sharp reduction in amplifier performance at 1064 nm is observed that coincides with an abrupt drop in fluorescence lifetime associated with thermally activated non-radiative losses (thermal quenching). As the temperature is increased from ambient to 977 °C, we observe a redistribution in the spectral weight of the emission cross-section from transitions in the 1020–1120 nm operating regime to transitions below 1010 nm. Interestingly, most of the spectral weight is transferred from transitions near the peak of the gain, 1020– 1040 nm, while the emission cross-section at longer wavelengths is nearly independent of temperature. The experimental set-up, data and subsequent analysis are divided into 5 sections. Section 2 gives a detailed description of the experimental set-up used to measure the absorption, emission, and slope efficiency as a function of temperature. In Section 3, the fluorescence lifetime is reported along with a description of the fitting routine used to determine the radiative lifetimes and energy separation of the individual Stark levels of the excited state manifold. Section 4 reports the temperature-dependent absorption coefficient and emission cross-section from 800 to 1200 nm. The saturated slope efficiency vs. temperature is presented in Section 5 followed by analysis and concluding remarks in Section 6. 2. Experimental set-up
⁎ Corresponding author. E-mail address: [email protected] (S.W. Moore). 0030-4018/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2011.08.064
Fig. 1 depicts the experimental set-up used to measure the emission and absorption cross-sections and slope efficiency of nLight
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
5775
Fig. 1. Experimental set-up for absorption and emission cross-section, fluorescence lifetime, and slope efficiency measurements.
Yb1200 20/125 DCPM [Yb-doped, 20-μm core, 0.07NA (numerical aperture), 125-μm clad, double clad, polarizing maintaining] fiber at high temperatures. Based on the cladding absorption at 920 nm (7 dB/m), the Yb+ 3 ion concentration is estimated to be 2.4 × 1020ions/m3. Spectral measurements performed simultaneously on passive 20/125 DCPM fiber subject to the same thermal conditions served as a reference for determining the absorption coefficient of the gain fiber. The jacketing of three gain fibers and one passive fiber were removed and the fibers were placed inside a tube furnace as depicted in Fig. 1. The gain fibers and passive fiber were 2.0 cm, 0.47 m, 2.3 m and 0.94 m in length, respectively. With the exception of the 2.2 cm Yb-doped strand, the fibers were coiled on a high temperature 6.7-cm diameter ceramic spool mounted to a sliding platform and centered in the middle of the furnace to minimize thermal gradients. Two sets of baffles were inserted at both ends of the tube furnace to further minimize thermal gradients and stabilize the temperature in the vicinity of the coiled fiber. Matching passive 20/125 DCPM (20 μm core, 0.07NA, 125-μm clad, double clad, polarizing maintaining) fibers were spliced to both ends of all the fibers to guide light into and out of the tube furnace. APC/FC connectors were mounted to the bare fiber ends to insure reproducibility when swapping the fibers between fiber-coupled instruments. As with the gain fibers, the acrylate/fluoroacrylate coatings were removed along all sections of the passive fiber pigtails resting inside the tube furnace. Finally, one end of a 0.22-NA silica clad passive fiber with a 148-μm diameter core was mounted in a fixture attached to the ceramic platform and butted against the side of the 2.2-cm gain fiber, while the other end was fed out of the tube furnace to collect the fluorescence spectrum and measure the radiative lifetimes. Each fiber in Fig. 1 was used to measure different physical properties of the Yb-doped silica fiber as a function of temperature. The 2.2-cm Ybdoped silica fiber was pulse-pumped at 10 Hz with 5-ms, 3.5-W, 974-nm pulses from an Apollo F25-974 fiber-coupled diode bar to measure the radiative lifetimes. To measure the emission spectrum, the same gain fiber was pumped with 3.5 W of continuous wave (CW)
power from the same fiber-coupled source. For both measurements, a fraction of the power emitted from the side of the fiber was captured by the 148-μm diameter core, passive fiber side-coupled to the gain fiber, and routed to either an InGaAs detector (Thorlabs PDA10CS), to measure the radiative lifetime, or to an optical spectrum analyzer (Ando AQ 6317), to measure the emission spectrum from 600 to 1300 nm. Using a short, 2.2-cm piece of fiber and capturing the emission from the side of the fiber minimized radiation trapping and ASE depletion of the inversion, which can alter the measured fluorescence lifetime. One end of the 0.47-m YB1200 20/125 DCPM fiber was illuminated with an Oriel 77501 white-light source to measure the (core) absorption spectrum, with the transmission of the 0.94-m passive fiber serving as the reference. Both fibers were mode stripped on at least one side to remove light propagating in the cladding. Finally, the 2.3 m YB1200 20/125 DCPM fiber was seeded with 1064-nm light from an IPG fiber laser and end-pumped in a counter-propagating configuration with the aid of a 975 nm/1064 nm dichroic optic to measure the slope efficiency. 3. Radiative lifetimes The temperature dependence of the radiative lifetimes of YB1200 20/125DCPM is shown in Fig. 2. As observed by Newell et al. [1], the radiative lifetime decays approximately linearly with increasing temperature from room temperature to nearly 500 °C, at which point the decay rapidly accelerates due to thermal quenching of the excited state. The behavior below 500 °C can be best understood by examining the energy levels of Yb + 3 in a silica host as shown in Fig. 3 [1,9,10]. The single electron energy level diagram of Yb + 3: silica is relatively simple, consisting of the 2F5/2 (excited state) and 2F7/2 (ground state) manifolds. The degeneracy of the manifolds is partially lifted by the silica host, which splits the lower manifold into four sublevels and the upper manifold into three, labeled i, j, k, l, and a, b, c, respectively. The radiative lifetime is inversely
5776
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
and thermal quenching were ignored, and the lifetimes were fit to a simple kinetic two level model [1,7], 0
−ΔEb kT
1
1þe A τ ¼ τa @ −ΔEb 1 þ ττa e kT
ð1Þ
b
Fig. 2. Fluorescence lifetime vs. temperature.
proportional to the dipole strengths of transitions from the a, b, c sublevels of the upper-state manifold to the Stark levels of the 2F7/2 manifold, and so is weighted by the population densities of the 2F5/2 sublevels, which in turn are determined by Boltzman statistics. Fig. 4 displays the 2F5/2 and 2F7/2 sublevel populations as a function of temperature. As expected, the higher lying states in the upper and lower manifolds become more populated with increasing temperature as the 2F5/2 (a) and 2F7/2 (i) sublevels depopulate. As will be shown, the radiative lifetimes of the 2F5/2 (b) ⇒ 2F7/2 transitions are shorter than that associated with the 2F5/2 (a) ⇒ 2F7/2 transition. This accounts for the nearly linear decay of the radiative lifetime from room temperature to 500 °C. Above this temperature, the radiative lifetime is strongly influenced by thermal quenching, which has an effective dipole moment that is 2–3 orders of magnitude greater than that of the 2F5/2 ⇒ 2F7/2 dipole transitions [1]. The radiative lifetimes, along with the separation in energies between the a, b sublevels of the upper state manifold, were extracted by fitting the measured fluorescence lifetimes as a function of temperature to a Boltzman-like distribution from 20 °C to 400 °C. Over this temperature range contributions from the 2F5/2 (c) ⇒ 2F7/2 transition
where τa and τb are the radiative lifetimes of levels a and b, respectively, k is Boltzmann's constant, T is the temperature (K), and ΔEb is the energy separation between Stark levels Ea and Eb. Using values of ΔEb, τa, and τb from previous studies [1,4,8,9,10] as initial values for the fit described by Eq. (1) yields 879 μs, 451 μs, and 660 cm −1 for τa, τb, and ΔEb, respectively. The fitted values for τa and ΔEb agree well with those of reference [1]; however, τb is 17.5% smaller (451 μs vs. 547 μs). The reason for this unclear, but it is consistent with the more rapid decline in the fluorescence lifetime that we observe from 20 °C to 400 °C relative to ref. [1]. Above 400 °C the upper-state lifetime is strongly influenced by thermal quenching. Thermal quenching can be incorporated into fitting routines of the measured fluorescence lifetimes using theory based on multiphonon emission [11]. However, the multiphonon emission model described by Eq. (11) in ref. [11], coupled with the contribution of the 2F5/2 (c) Stark level to the fluorescence lifetimes at higher temperatures, introduces more variables than can be reliably fit from the temperature-dependent variation of the measured lifetime. Thus, the data was fit to Eq. (1) using a least-squares fitting routine between 20 °C and 400 °C, where thermal quenching and contributions from 2F5/2 (c) could be safely ignored, to extract τa, τb, and ΔEb as was done in [1]. 4. Absorption coefficient and emission cross-section The fiber-core transmission spectra versus temperature of YB1200 20/125DCPM from 600 to 1300 nm are shown in Fig. 5. Transmission through the 2.2-cm strand was measured to determine the absorption coefficient in the range of 875–1000 nm where strongly resonate 975-nm, 960-nm, and 920-nm absorption bands (N100 dB/m) preclude measurable transmission for longer lengths of fiber. For each temperature, the spectra of both samples were referenced to the transmission spectrum of the coiled 0.94-m long nLight 20/125 passive fiber with matching core NA to determine the absolute transmission and absorption coefficients. The absorption coefficient is determined from the simple relation, α¼
− ln½Tr L
Fig. 3. Energy level diagram of Yb+ 3:silica. From Reference [3].
ð2Þ
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
5777
Fig. 4. Boltzman distribution of the 2F5/2 (top) and 2F7/2 (bottom) manifolds vs. temperature of Yb+ 3:silica.
Fig. 5. Core transmission through 2.2 cm (top) and 0.47 m (bottom) of YB1200 20/125 DCPM fiber.
where Tr is the absolute transmission referenced to the passive fiber and L is the fiber length. The absorption coefficients for the 47-cm and 2.2-cm samples in the ranges of 1010–1120 nm and 850–1100 nm are shown in Fig. 6. It is evident that the absorption coefficient increases with increasing temperature for all wavelengths spanning the 1000–1120 nm operating regime of Yb+ 3-doped silica. This trend is consistent with the increasing probability that levels j–l of the 2F7/2 manifold are occupied at higher temperatures (Fig. 4). Transitions from these levels to the lowest level of the upper state manifold account for the ~1020–1100 nm absorption band in Yb+ 3-doped silica. An overall decrease in absorption from 900 to 1000 nm is observed in Fig. 6 as the temperature is increased. This is particularly true for the 975-nm and 920-nm resonant absorption lines which undergo a ~50% reduction as the temperature increases from room temperature to 985 °C. As with the temperature-dependent absorption above 1000 nm, this is due to depopulation of the 2F7/2 (i) Stark level with increasing temperature. Transitions from this sublevel to the lowest two levels of the 2F5/2 manifold are responsible for the 977-nm and 920nm absorption bands, respectively. We note that the shoulder in the
absorption spectra centered at 960 nm grows with increasing temperature up to 473 °C before saturating and decreasing slightly at higher temperatures. Whether the observed saturation in absorption at this wavelength is real is obfuscated by uncertainty in the absolute absorption coefficient due to variability in absorption at the endpoints of the measured spectrum (where the absorption coefficients are expected to be the same for all temperatures). Nonetheless, the overall increase in absorption at 960 nm can be attributed to the increasing population density of the 2F7/2 (j) Stark level at higher temperatures. Lastly, above 800 °C a weak absorption band is observed at 900 nm associated with transitions from the 2F7/2 (k,l) sublevels. This is expected as the population densities of these Stark levels increases at higher temperatures. Reductions in absorption at the 920-nm and 977-nm pumping bands with increasing temperature, coupled with increases in absorption in the 1020–1120 nm signal band, could be very detrimental to the performance of three-level laser/amplifier systems such as Yb: SiO2 operating at the kilowatt level or in high-temperature environments. The effect of reduced absorption of the pump could be offset with either additional fiber or by double-passing the pump light
5778
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
Fig. 7. Emission spectra of 2.2 cm YB1200 20/125 DCPM fiber pumped at 977 nm.
As expected, opposed to the changes in the absorption coefficient, the emission increases below 980 nm and decreases above 1010 nm as the temperature is increased (see Fig. 7). The overall gain in emission below 980 nm is expected due to rising populations of the 2F5/2 (b,c) sublevels (Fig. 4) which are responsible for transitions in the 900–1000 nm region. Conversely, the population of the 2F5/2 (a) sublevel falls with increasing temperature, and so the associated transitions between 1010 and 1100 nm become weaker. The emission cross-section can be readily derived from the emission spectra using the Füchtbauer–Landenburg (F–L) equation [2,3,12–14], σe ðλÞ ¼
λ5 IðλÞ 8πcn5 τrad ∫ λ IðλÞdλ
ð3Þ
Δλ
where σe(λ) is the emission cross-section, λ is the wavelength, τrad is the radiative lifetime, I(λ) is the power or intensity spectrum of the
Fig. 6. Absorption of 2.2 cm (top) and 0.47 m (bottom) of YB1200 20/125 DCPM fiber.
through the fiber, while increases in the absorption coefficient across the signal band will reduce slope efficiency. From Fig. 6, the absorption coefficient in the signal band increases from ~ 2 to 5 dB/m as the temperature increases from 23 °C to 167 °C, a temperature regime seemingly within the realm of kilowatt class fiber lasers. The effects of this trend will be revisited in Section 5 when the temperature dependence of the slope efficiency is discussed. The emission spectra collected from the side of the 2.2-cm YB1200 20/125 DCPM fiber that is pumped with 974-nm light is shown in Fig. 7. The overall decrease in emission from 800 to 1200 nm with increasing temperature is consistent with the reduction in absorption at 974 nm observed in Fig. 6 and, for temperatures above 800 °C, may also be due to thermal damage to the 148-μm passive fiber used to capture the spontaneous emission. It should be noted that pump scatter into the 148-μm passive fiber compromised the recorded spectra from 972 to 978 nm. These spurious contributions from pump scatter were excised and “patched” with an interpolating routine to give a more accurate measure of the emission cross-section. Consequently, the experimental results presented here do not provide a direct measure of the emission at 975 nm.
Fig. 8. Emission cross-section calculated using Eq. (3) and the data from Fig. 7.
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
5779
measured fluorescence, Δλ is the bandwidth of the radiative transition, c is the speed of light in vacuum, and n is the index of refraction of the medium. Using the emission spectra from Fig. 7 and the radiative lifetimes from Fig. 2, the emission cross-sections are calculated and displayed in Fig. 8 from 22 °C to 478 °C. Consistent with the emission spectra, the emission cross-section increases markedly from 875 to 970 nm and 980–1010 nm, and decreases in the 970–980 nm and 1010–1045 nm spectral windows as the temperature increases. The emission cross-section from 970 to 980 nm is compromised by removal of the pump scatter and interpolating over the excised section prior to using Eq. (5). The peak of the emission cross-section (gain) associated with the 2F5/2 (a) ⇒ 2F7/2 transitions (1010–1100 nm) decreases by 17% and red-shifts from 1027 nm to 1030 nm from 22 °C to 478 °C. By contrast, the emission cross-section decreases only 10% from 1050 nm to 1100 nm and changes above 1100 nm are not observed at all, suggesting the performance of Yb+ 3-doped silica lasers and amplifiers operated above 1050 nm would be insensitive to temperature. The onset of thermal quenching (Fig. 2) precludes determination of the optical cross-section for temperatures exceeding 500 °C. Ideally, if the thermal quenching rate is known then the emission cross-section can be calculated by scaling Eq. (3) by the ratio of the non-radiative lifetime (thermal quenching) to the measured radiative lifetime [13]. However, in Section 3 it was determined that the uncertainty in the thermal quenching rate that would be obtained by a fit of the fluorescence lifetimes to a multiphonon emission model would be unacceptably high. Still, to the best of our knowledge we have reported the emission cross-section in Yb+ 3-doped silica at the highest temperature to date. 5. Slope efficiency Measurements of the slope efficiency were performed on a 2.3-m length of 20/125 YB1200 DCPM fiber from room temperature to 900 °C. The fiber was end-pumped in a counter-propagating configuration by an Apollo F25-974 25-W, 974-nm fiber-coupled diode laser and seeded with 575 mW of 1064-nm light from a diffraction-limited, 10-W IPG Photonics fiber laser. The free-space-to-fiber coupling efficiency of the 974-nm pump source into the 125-μm fiber cladding was measured to be 89%, and a dichroic optic (highly reflecting at 975 nm, anti-reflecting at 1064 nm) was placed between the seed laser and gain fiber to redirect and measure the unabsorbed pump power. The unabsorbed pump light was subtracted from the coupled pump power (0.89 × available power) to determine the absorbed power. The amplified seed power at 1064 nm (output power) was then plotted against the absorbed pump power to determine the intrinsic slope efficiency of the fiber. Spectral measurements of the amplified signal showed more than 40 dB of ASE suppression. The amplified signal power as a function absorbed pump power and the associated slope efficiencies for temperatures spanning 20 °C–985 °C in ~100 °C increments are shown in Fig. 9. Also included in this figure is the output power versus absorbed pump power for a fully clad (i.e. the fluoroacrylate jacketing is not removed), 2.0-m long YB1200 20/125 DCPM (Yb + 3-doped) reference fiber coiled to a 6.7cm diameter that was not placed in the furnace. The slope efficiency of the reference (clad) fiber is more than twice (76% vs. 31%) that of the unclad fiber coiled to the same diameter and placed inside the furnace. The difference owes to pump scatter out of the fiber cladding along regions of the passive and coiled gain fiber that are in contact with the high temperature ceramic platform and spool used to fixture the fiber in the oven. Consequently, more than 50% of the pump light is scattered out of the fiber channel before and during amplification of the 1064-nm seed. However, this loss of pump light only offsets the measured slope efficiencies and will not alter relative changes in slope efficiency with temperature and so, for purposes of this study, is of little consequence. From Fig. 9 it can be seen that the slope efficiency drops while the pump threshold for transparency, or the pump power at which the
Fig. 9. Saturated output power vs. absorbed 977 nm pump power (top) and corresponding slope efficiency vs. temperature (bottom). 2.3 m YB1200 20/125 DCPM fiber. Dashed lines are fit to the data.
input signal power equals the output signal power, increases with rising temperature. Changes in the slope efficiency can be categorized into three temperature regimes. From 20 °C to 273 °C the slope efficiency decreases monotonically with increasing temperature, consistent with increases in the absorption coefficient at 1064-nm observed in Fig. 6. However, the slope efficiency remains unchanged to within experimental error for the next two temperature points up to 478 °C. This cannot be explained by the temperature dependence of the absorption coefficient and emission cross-section and is not understood at this time. However, the steady rise in threshold from 23 °C to 478 °C is expected from the decrease in population density of 2F5/2 (a) Stark sublevel with increasing temperature. From 480 °C to 680 °C the slope efficiency drops precipitously to less than 1%, reaching transparency (power in = power out) near 579 °C. From 680 °C to 995 °C the fiber becomes virtually opaque for all pump powers, transmitting only 20–30 mW for 575 mW of input seed power. The threshold for the sharp decline in slope efficiency coincides with the onset of thermal quenching observed near 500 °C in Fig. 2,
5780
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
implying that non-radiative losses associated with thermal quenching rapidly deplete the gain above 500 °C. Thus, we can conclude that the core temperature of Yb + 3:SiO2 should not exceed 500 °C for efficient lasing operation. Under conditions of pulsed amplification in the kHz-regime where the time between the onset of pumping and stimulated emission from the seed pulse ranges from 10 μs to 1 ms, this threshold could be lower still, as the net inversion along the fiber will be less than that of a fiber under these conditions continuously pumped for several upper-state lifetimes. While it is unlikely that a Yb-doped gain fiber would be operated under conditions where the core temperature approached or exceeded 500 °C, as this would cause the protective acrylate/fluoroacrylate coating to melt or burn, it is clear from Fig. 9 that degradation of the laser performance should be expected for temperatures as modest as 80 °C–160 °C. Even as the core temperature is raised from 20 °C to 85 °C, a likely operating regime for kilowatt class fiber lasers, there is an 18% drop in slope efficiency. For temperatures approaching 165 °C the slope efficiency drops further to a net reduction of 21%. Thus, proper thermal management of any high power fiber laser/amplifier system is essential to ensure the highest achievable efficiency. 6. Conclusions We have reported the fluorescence lifetime, absorption coefficient, and emission cross-section of Yb + 3 over the pump absorption and signal bands, and the slope efficiency of a Yb + 3-doped LMA fiber as a function of temperature from 23 °C to nearly 1000 °C. We found elevated temperatures, particularly above 500 °C, severely degraded the lasing performance of Yb + 3-doped fibers. This was due to sharp increases in the absorption coefficient coupled with modest decreases in the emission cross-section in the signal band and a sharp degradation of the fluorescence lifetime, particularly above 500 °C. The rapid reduction in fluorescence lifetime above 500 °C due to the onset of thermal quenching limits the energy storing capacity of Yb + 3:SiO2 and effectively prohibits its use at higher temperatures. This was confirmed in measurements of the slope efficiency where we found that 977 nm-pumped Yb + 3-doped silica becomes opaque for temperatures exceeding 600 °C. Even for core temperatures as low as 85 °C−160 °C,
a potential operating regime for kW-class fiber laser systems, we observed a 21% reduction in both the slope efficiency at 1064 nm and the absorption coefficient at 975 nm. This suggests that thermal issues should be considered and managed even at the modest temperatures found in typical kW-class fibers lasers. Acknowledgments Sandia is a multi-program laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000. Funding for this work was provided by the Defense Advanced Research Project Agency (DARPA) under grant DOE-NNSA. The views expressed are those of the author and do not reflect the official policy or position of the Department of Defense or the U. S. Government. References [1] T.C. Newell, P. Peterson, A. Gavrielides, M.P. Sharma, Optics Communication 273 (2007) 256. [2] H. Kiriyama, N. Srinivasan, M. Yamanaka, Y. Izawa, T. Yamanaka, S. Nakai, Japan Journal of Applied Physics 36 (1997) L1165. [3] D. Sumida, T.Y. Fan, OSA Proceedings of the Advanced Solid State Lasers 20 (1994) 100. [4] D.A. Grukh, A.S. Kurkov, V.M. Paramonov, E.M. Dianov, IEEE Journal of Quantum Electronics 34 (6) (2004) 579. [5] Z. Burshtein, Y. Kalisky, S.Z. Levy, P. LeBoulanger, S. Rotman, IEEE Journal of Quantum Electronics 36 (8) (2000) 1000. [6] J. Dong, M. Bass, Y. Mai, P. Deng, F. Gan, Journal of Optical Society American B 20 (9) (2003) 1975. [7] Z. Zhang, K.T.V. Gratten, A.W. Palmer, Journal of Applied Physics 73 (7) (1993) 3493. [8] P.L. Pernas, E. Cantelar, Physica Scripta T118 (2005) 93. [9] H.M. Pask, R.J. Carman, D.C. Hanna, A.C. Tropper, C.J. Mackechnie, P.R. Barber, J.M. Dawes, IEEE Selected Topics Quantum Electronics 1 (1995) 2. [10] J.Y. Allain, M. Monerie, H. Poignant, Electronics Letters 28 (1992) 988. [11] C.B. Layne, W.H. Lowdermilk, M.J. Weber, Physics Review B 16 (1) (1977) 10. [12] D.E. McCumber, Physics Review 134 (No. 2A) (1964) A299. [13] P.F. Moulton, Journal of Optical Society American B 3 (1) (1986) 125. [14] W.L. Barnes, R.I. Laming, E.J. Tarbox, P.R. Morkel, IEEE Journal of Quantum Electronics 27 (1991) 1004.
Contents lists available at SciVerse ScienceDirect
Optics Communications journal homepage: www.elsevier.com/locate/optcom
Optical properties of Yb + 3-doped fibers and fiber lasers at high temperature S.W. Moore a,⁎, T. Barnett a, T.A. Reichardt a, R.L. Farrow b a b
Sandia National Laboratories, P. O. Box 969, MS 9056, Livermore, CA 94551, USA JDSU,430 North McCarthy Boulevard, Milpitas, CA 95035, USA
a r t i c l e
i n f o
Article history: Received 12 July 2011 Received in revised form 24 August 2011 Accepted 26 August 2011 Available online 12 September 2011 Keywords: Spectroscopy Lasers Emission
a b s t r a c t Recent advances in power scaling of Yb+ 3-doped fiber lasers to the kilowatt level suggest a need to examine the performance of Yb+ 3-doped silica at temperatures well above ambient. We report experimental results for the absorption coefficient, emission cross-section, fluorescence lifetime, and slope efficiency of a Yb3+-doped large mode area (LMA) silica fiber for temperatures spanning 23 °C–977 °C. To the best of our knowledge these are the highest temperatures to date for which these optical properties have been measured. We find a sharp reduction in the energy storing capability and lasing performance of Yb+ 3:SiO2 above 500 °C that coincides with the onset of non-radiative transitions in the excited state manifold (thermal quenching). As the temperature increases from room temperature to 977 °C, absorption in the 1020–1120 nm operating band increases monotonically, concurrent with a reduction in absorption at the 920-nm and 977-nm pumping bands. Conversely, the spectral weight of the emission cross-section shifts from transitions above 1010 nm to those below, with the exception of the 977-nm emission band. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Advances in cladding-pumped, kilowatt-class, Yb-doped fiber laser systems have surged in recent years as these devices have proven especially suitable for material processing. The large surface area-to-volume ratio, small quantum defect (975-nm pumping vs. 1064- to 1100-nm lasing), and high slope efficiency of commercially available Yb-doped silica fiber reduces thermal management and operating costs while allowing for very high wall-plug efficiencies. Still, despite the intrinsic advantages in thermal management offered by fibers, a significant rise in core temperature is expected for fiber lasers and amplifiers operating at the 1–5 kW level. The operating temperature of standard Yb-doped silica fiber is limited by the protective acrylate/fluoroacrylate coating, which begins to degrade and eventually fails below 200 °C. Studies to date concerning the optical properties and performance of rare-earth doped glass oxide and crystal laser and amplifier systems have been conducted at or below this temperature [1–6]. However, to further reduce the cost of thermal management or, alternatively, to operate fiber lasers and amplifiers in high temperature environments by using cladding materials that can withstand higher temperatures such as polyimide, then it is important to evaluate the optical properties and performance of Yb+ 3-doped silica at temperatures exceeding 200 °C. In this paper we report the temperature dependence of the radiative lifetime, emission cross-section, absorption coefficient, and slope efficiency of Yb+ 3-doped (2.4× 1020ions/m3) large mode area (LMA) fiber
from 23 °C to 977 °C. We find that absorption in the 1020–1120 nm operating regime increases rapidly while absorption at the 920-nm and 977-nm pump bands decreases steadily with increasing temperature. For temperatures exceeding 500 °C, a sharp reduction in amplifier performance at 1064 nm is observed that coincides with an abrupt drop in fluorescence lifetime associated with thermally activated non-radiative losses (thermal quenching). As the temperature is increased from ambient to 977 °C, we observe a redistribution in the spectral weight of the emission cross-section from transitions in the 1020–1120 nm operating regime to transitions below 1010 nm. Interestingly, most of the spectral weight is transferred from transitions near the peak of the gain, 1020– 1040 nm, while the emission cross-section at longer wavelengths is nearly independent of temperature. The experimental set-up, data and subsequent analysis are divided into 5 sections. Section 2 gives a detailed description of the experimental set-up used to measure the absorption, emission, and slope efficiency as a function of temperature. In Section 3, the fluorescence lifetime is reported along with a description of the fitting routine used to determine the radiative lifetimes and energy separation of the individual Stark levels of the excited state manifold. Section 4 reports the temperature-dependent absorption coefficient and emission cross-section from 800 to 1200 nm. The saturated slope efficiency vs. temperature is presented in Section 5 followed by analysis and concluding remarks in Section 6. 2. Experimental set-up
⁎ Corresponding author. E-mail address: [email protected] (S.W. Moore). 0030-4018/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2011.08.064
Fig. 1 depicts the experimental set-up used to measure the emission and absorption cross-sections and slope efficiency of nLight
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
5775
Fig. 1. Experimental set-up for absorption and emission cross-section, fluorescence lifetime, and slope efficiency measurements.
Yb1200 20/125 DCPM [Yb-doped, 20-μm core, 0.07NA (numerical aperture), 125-μm clad, double clad, polarizing maintaining] fiber at high temperatures. Based on the cladding absorption at 920 nm (7 dB/m), the Yb+ 3 ion concentration is estimated to be 2.4 × 1020ions/m3. Spectral measurements performed simultaneously on passive 20/125 DCPM fiber subject to the same thermal conditions served as a reference for determining the absorption coefficient of the gain fiber. The jacketing of three gain fibers and one passive fiber were removed and the fibers were placed inside a tube furnace as depicted in Fig. 1. The gain fibers and passive fiber were 2.0 cm, 0.47 m, 2.3 m and 0.94 m in length, respectively. With the exception of the 2.2 cm Yb-doped strand, the fibers were coiled on a high temperature 6.7-cm diameter ceramic spool mounted to a sliding platform and centered in the middle of the furnace to minimize thermal gradients. Two sets of baffles were inserted at both ends of the tube furnace to further minimize thermal gradients and stabilize the temperature in the vicinity of the coiled fiber. Matching passive 20/125 DCPM (20 μm core, 0.07NA, 125-μm clad, double clad, polarizing maintaining) fibers were spliced to both ends of all the fibers to guide light into and out of the tube furnace. APC/FC connectors were mounted to the bare fiber ends to insure reproducibility when swapping the fibers between fiber-coupled instruments. As with the gain fibers, the acrylate/fluoroacrylate coatings were removed along all sections of the passive fiber pigtails resting inside the tube furnace. Finally, one end of a 0.22-NA silica clad passive fiber with a 148-μm diameter core was mounted in a fixture attached to the ceramic platform and butted against the side of the 2.2-cm gain fiber, while the other end was fed out of the tube furnace to collect the fluorescence spectrum and measure the radiative lifetimes. Each fiber in Fig. 1 was used to measure different physical properties of the Yb-doped silica fiber as a function of temperature. The 2.2-cm Ybdoped silica fiber was pulse-pumped at 10 Hz with 5-ms, 3.5-W, 974-nm pulses from an Apollo F25-974 fiber-coupled diode bar to measure the radiative lifetimes. To measure the emission spectrum, the same gain fiber was pumped with 3.5 W of continuous wave (CW)
power from the same fiber-coupled source. For both measurements, a fraction of the power emitted from the side of the fiber was captured by the 148-μm diameter core, passive fiber side-coupled to the gain fiber, and routed to either an InGaAs detector (Thorlabs PDA10CS), to measure the radiative lifetime, or to an optical spectrum analyzer (Ando AQ 6317), to measure the emission spectrum from 600 to 1300 nm. Using a short, 2.2-cm piece of fiber and capturing the emission from the side of the fiber minimized radiation trapping and ASE depletion of the inversion, which can alter the measured fluorescence lifetime. One end of the 0.47-m YB1200 20/125 DCPM fiber was illuminated with an Oriel 77501 white-light source to measure the (core) absorption spectrum, with the transmission of the 0.94-m passive fiber serving as the reference. Both fibers were mode stripped on at least one side to remove light propagating in the cladding. Finally, the 2.3 m YB1200 20/125 DCPM fiber was seeded with 1064-nm light from an IPG fiber laser and end-pumped in a counter-propagating configuration with the aid of a 975 nm/1064 nm dichroic optic to measure the slope efficiency. 3. Radiative lifetimes The temperature dependence of the radiative lifetimes of YB1200 20/125DCPM is shown in Fig. 2. As observed by Newell et al. [1], the radiative lifetime decays approximately linearly with increasing temperature from room temperature to nearly 500 °C, at which point the decay rapidly accelerates due to thermal quenching of the excited state. The behavior below 500 °C can be best understood by examining the energy levels of Yb + 3 in a silica host as shown in Fig. 3 [1,9,10]. The single electron energy level diagram of Yb + 3: silica is relatively simple, consisting of the 2F5/2 (excited state) and 2F7/2 (ground state) manifolds. The degeneracy of the manifolds is partially lifted by the silica host, which splits the lower manifold into four sublevels and the upper manifold into three, labeled i, j, k, l, and a, b, c, respectively. The radiative lifetime is inversely
5776
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
and thermal quenching were ignored, and the lifetimes were fit to a simple kinetic two level model [1,7], 0
−ΔEb kT
1
1þe A τ ¼ τa @ −ΔEb 1 þ ττa e kT
ð1Þ
b
Fig. 2. Fluorescence lifetime vs. temperature.
proportional to the dipole strengths of transitions from the a, b, c sublevels of the upper-state manifold to the Stark levels of the 2F7/2 manifold, and so is weighted by the population densities of the 2F5/2 sublevels, which in turn are determined by Boltzman statistics. Fig. 4 displays the 2F5/2 and 2F7/2 sublevel populations as a function of temperature. As expected, the higher lying states in the upper and lower manifolds become more populated with increasing temperature as the 2F5/2 (a) and 2F7/2 (i) sublevels depopulate. As will be shown, the radiative lifetimes of the 2F5/2 (b) ⇒ 2F7/2 transitions are shorter than that associated with the 2F5/2 (a) ⇒ 2F7/2 transition. This accounts for the nearly linear decay of the radiative lifetime from room temperature to 500 °C. Above this temperature, the radiative lifetime is strongly influenced by thermal quenching, which has an effective dipole moment that is 2–3 orders of magnitude greater than that of the 2F5/2 ⇒ 2F7/2 dipole transitions [1]. The radiative lifetimes, along with the separation in energies between the a, b sublevels of the upper state manifold, were extracted by fitting the measured fluorescence lifetimes as a function of temperature to a Boltzman-like distribution from 20 °C to 400 °C. Over this temperature range contributions from the 2F5/2 (c) ⇒ 2F7/2 transition
where τa and τb are the radiative lifetimes of levels a and b, respectively, k is Boltzmann's constant, T is the temperature (K), and ΔEb is the energy separation between Stark levels Ea and Eb. Using values of ΔEb, τa, and τb from previous studies [1,4,8,9,10] as initial values for the fit described by Eq. (1) yields 879 μs, 451 μs, and 660 cm −1 for τa, τb, and ΔEb, respectively. The fitted values for τa and ΔEb agree well with those of reference [1]; however, τb is 17.5% smaller (451 μs vs. 547 μs). The reason for this unclear, but it is consistent with the more rapid decline in the fluorescence lifetime that we observe from 20 °C to 400 °C relative to ref. [1]. Above 400 °C the upper-state lifetime is strongly influenced by thermal quenching. Thermal quenching can be incorporated into fitting routines of the measured fluorescence lifetimes using theory based on multiphonon emission [11]. However, the multiphonon emission model described by Eq. (11) in ref. [11], coupled with the contribution of the 2F5/2 (c) Stark level to the fluorescence lifetimes at higher temperatures, introduces more variables than can be reliably fit from the temperature-dependent variation of the measured lifetime. Thus, the data was fit to Eq. (1) using a least-squares fitting routine between 20 °C and 400 °C, where thermal quenching and contributions from 2F5/2 (c) could be safely ignored, to extract τa, τb, and ΔEb as was done in [1]. 4. Absorption coefficient and emission cross-section The fiber-core transmission spectra versus temperature of YB1200 20/125DCPM from 600 to 1300 nm are shown in Fig. 5. Transmission through the 2.2-cm strand was measured to determine the absorption coefficient in the range of 875–1000 nm where strongly resonate 975-nm, 960-nm, and 920-nm absorption bands (N100 dB/m) preclude measurable transmission for longer lengths of fiber. For each temperature, the spectra of both samples were referenced to the transmission spectrum of the coiled 0.94-m long nLight 20/125 passive fiber with matching core NA to determine the absolute transmission and absorption coefficients. The absorption coefficient is determined from the simple relation, α¼
− ln½Tr L
Fig. 3. Energy level diagram of Yb+ 3:silica. From Reference [3].
ð2Þ
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
5777
Fig. 4. Boltzman distribution of the 2F5/2 (top) and 2F7/2 (bottom) manifolds vs. temperature of Yb+ 3:silica.
Fig. 5. Core transmission through 2.2 cm (top) and 0.47 m (bottom) of YB1200 20/125 DCPM fiber.
where Tr is the absolute transmission referenced to the passive fiber and L is the fiber length. The absorption coefficients for the 47-cm and 2.2-cm samples in the ranges of 1010–1120 nm and 850–1100 nm are shown in Fig. 6. It is evident that the absorption coefficient increases with increasing temperature for all wavelengths spanning the 1000–1120 nm operating regime of Yb+ 3-doped silica. This trend is consistent with the increasing probability that levels j–l of the 2F7/2 manifold are occupied at higher temperatures (Fig. 4). Transitions from these levels to the lowest level of the upper state manifold account for the ~1020–1100 nm absorption band in Yb+ 3-doped silica. An overall decrease in absorption from 900 to 1000 nm is observed in Fig. 6 as the temperature is increased. This is particularly true for the 975-nm and 920-nm resonant absorption lines which undergo a ~50% reduction as the temperature increases from room temperature to 985 °C. As with the temperature-dependent absorption above 1000 nm, this is due to depopulation of the 2F7/2 (i) Stark level with increasing temperature. Transitions from this sublevel to the lowest two levels of the 2F5/2 manifold are responsible for the 977-nm and 920nm absorption bands, respectively. We note that the shoulder in the
absorption spectra centered at 960 nm grows with increasing temperature up to 473 °C before saturating and decreasing slightly at higher temperatures. Whether the observed saturation in absorption at this wavelength is real is obfuscated by uncertainty in the absolute absorption coefficient due to variability in absorption at the endpoints of the measured spectrum (where the absorption coefficients are expected to be the same for all temperatures). Nonetheless, the overall increase in absorption at 960 nm can be attributed to the increasing population density of the 2F7/2 (j) Stark level at higher temperatures. Lastly, above 800 °C a weak absorption band is observed at 900 nm associated with transitions from the 2F7/2 (k,l) sublevels. This is expected as the population densities of these Stark levels increases at higher temperatures. Reductions in absorption at the 920-nm and 977-nm pumping bands with increasing temperature, coupled with increases in absorption in the 1020–1120 nm signal band, could be very detrimental to the performance of three-level laser/amplifier systems such as Yb: SiO2 operating at the kilowatt level or in high-temperature environments. The effect of reduced absorption of the pump could be offset with either additional fiber or by double-passing the pump light
5778
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
Fig. 7. Emission spectra of 2.2 cm YB1200 20/125 DCPM fiber pumped at 977 nm.
As expected, opposed to the changes in the absorption coefficient, the emission increases below 980 nm and decreases above 1010 nm as the temperature is increased (see Fig. 7). The overall gain in emission below 980 nm is expected due to rising populations of the 2F5/2 (b,c) sublevels (Fig. 4) which are responsible for transitions in the 900–1000 nm region. Conversely, the population of the 2F5/2 (a) sublevel falls with increasing temperature, and so the associated transitions between 1010 and 1100 nm become weaker. The emission cross-section can be readily derived from the emission spectra using the Füchtbauer–Landenburg (F–L) equation [2,3,12–14], σe ðλÞ ¼
λ5 IðλÞ 8πcn5 τrad ∫ λ IðλÞdλ
ð3Þ
Δλ
where σe(λ) is the emission cross-section, λ is the wavelength, τrad is the radiative lifetime, I(λ) is the power or intensity spectrum of the
Fig. 6. Absorption of 2.2 cm (top) and 0.47 m (bottom) of YB1200 20/125 DCPM fiber.
through the fiber, while increases in the absorption coefficient across the signal band will reduce slope efficiency. From Fig. 6, the absorption coefficient in the signal band increases from ~ 2 to 5 dB/m as the temperature increases from 23 °C to 167 °C, a temperature regime seemingly within the realm of kilowatt class fiber lasers. The effects of this trend will be revisited in Section 5 when the temperature dependence of the slope efficiency is discussed. The emission spectra collected from the side of the 2.2-cm YB1200 20/125 DCPM fiber that is pumped with 974-nm light is shown in Fig. 7. The overall decrease in emission from 800 to 1200 nm with increasing temperature is consistent with the reduction in absorption at 974 nm observed in Fig. 6 and, for temperatures above 800 °C, may also be due to thermal damage to the 148-μm passive fiber used to capture the spontaneous emission. It should be noted that pump scatter into the 148-μm passive fiber compromised the recorded spectra from 972 to 978 nm. These spurious contributions from pump scatter were excised and “patched” with an interpolating routine to give a more accurate measure of the emission cross-section. Consequently, the experimental results presented here do not provide a direct measure of the emission at 975 nm.
Fig. 8. Emission cross-section calculated using Eq. (3) and the data from Fig. 7.
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
5779
measured fluorescence, Δλ is the bandwidth of the radiative transition, c is the speed of light in vacuum, and n is the index of refraction of the medium. Using the emission spectra from Fig. 7 and the radiative lifetimes from Fig. 2, the emission cross-sections are calculated and displayed in Fig. 8 from 22 °C to 478 °C. Consistent with the emission spectra, the emission cross-section increases markedly from 875 to 970 nm and 980–1010 nm, and decreases in the 970–980 nm and 1010–1045 nm spectral windows as the temperature increases. The emission cross-section from 970 to 980 nm is compromised by removal of the pump scatter and interpolating over the excised section prior to using Eq. (5). The peak of the emission cross-section (gain) associated with the 2F5/2 (a) ⇒ 2F7/2 transitions (1010–1100 nm) decreases by 17% and red-shifts from 1027 nm to 1030 nm from 22 °C to 478 °C. By contrast, the emission cross-section decreases only 10% from 1050 nm to 1100 nm and changes above 1100 nm are not observed at all, suggesting the performance of Yb+ 3-doped silica lasers and amplifiers operated above 1050 nm would be insensitive to temperature. The onset of thermal quenching (Fig. 2) precludes determination of the optical cross-section for temperatures exceeding 500 °C. Ideally, if the thermal quenching rate is known then the emission cross-section can be calculated by scaling Eq. (3) by the ratio of the non-radiative lifetime (thermal quenching) to the measured radiative lifetime [13]. However, in Section 3 it was determined that the uncertainty in the thermal quenching rate that would be obtained by a fit of the fluorescence lifetimes to a multiphonon emission model would be unacceptably high. Still, to the best of our knowledge we have reported the emission cross-section in Yb+ 3-doped silica at the highest temperature to date. 5. Slope efficiency Measurements of the slope efficiency were performed on a 2.3-m length of 20/125 YB1200 DCPM fiber from room temperature to 900 °C. The fiber was end-pumped in a counter-propagating configuration by an Apollo F25-974 25-W, 974-nm fiber-coupled diode laser and seeded with 575 mW of 1064-nm light from a diffraction-limited, 10-W IPG Photonics fiber laser. The free-space-to-fiber coupling efficiency of the 974-nm pump source into the 125-μm fiber cladding was measured to be 89%, and a dichroic optic (highly reflecting at 975 nm, anti-reflecting at 1064 nm) was placed between the seed laser and gain fiber to redirect and measure the unabsorbed pump power. The unabsorbed pump light was subtracted from the coupled pump power (0.89 × available power) to determine the absorbed power. The amplified seed power at 1064 nm (output power) was then plotted against the absorbed pump power to determine the intrinsic slope efficiency of the fiber. Spectral measurements of the amplified signal showed more than 40 dB of ASE suppression. The amplified signal power as a function absorbed pump power and the associated slope efficiencies for temperatures spanning 20 °C–985 °C in ~100 °C increments are shown in Fig. 9. Also included in this figure is the output power versus absorbed pump power for a fully clad (i.e. the fluoroacrylate jacketing is not removed), 2.0-m long YB1200 20/125 DCPM (Yb + 3-doped) reference fiber coiled to a 6.7cm diameter that was not placed in the furnace. The slope efficiency of the reference (clad) fiber is more than twice (76% vs. 31%) that of the unclad fiber coiled to the same diameter and placed inside the furnace. The difference owes to pump scatter out of the fiber cladding along regions of the passive and coiled gain fiber that are in contact with the high temperature ceramic platform and spool used to fixture the fiber in the oven. Consequently, more than 50% of the pump light is scattered out of the fiber channel before and during amplification of the 1064-nm seed. However, this loss of pump light only offsets the measured slope efficiencies and will not alter relative changes in slope efficiency with temperature and so, for purposes of this study, is of little consequence. From Fig. 9 it can be seen that the slope efficiency drops while the pump threshold for transparency, or the pump power at which the
Fig. 9. Saturated output power vs. absorbed 977 nm pump power (top) and corresponding slope efficiency vs. temperature (bottom). 2.3 m YB1200 20/125 DCPM fiber. Dashed lines are fit to the data.
input signal power equals the output signal power, increases with rising temperature. Changes in the slope efficiency can be categorized into three temperature regimes. From 20 °C to 273 °C the slope efficiency decreases monotonically with increasing temperature, consistent with increases in the absorption coefficient at 1064-nm observed in Fig. 6. However, the slope efficiency remains unchanged to within experimental error for the next two temperature points up to 478 °C. This cannot be explained by the temperature dependence of the absorption coefficient and emission cross-section and is not understood at this time. However, the steady rise in threshold from 23 °C to 478 °C is expected from the decrease in population density of 2F5/2 (a) Stark sublevel with increasing temperature. From 480 °C to 680 °C the slope efficiency drops precipitously to less than 1%, reaching transparency (power in = power out) near 579 °C. From 680 °C to 995 °C the fiber becomes virtually opaque for all pump powers, transmitting only 20–30 mW for 575 mW of input seed power. The threshold for the sharp decline in slope efficiency coincides with the onset of thermal quenching observed near 500 °C in Fig. 2,
5780
S.W. Moore et al. / Optics Communications 284 (2011) 5774–5780
implying that non-radiative losses associated with thermal quenching rapidly deplete the gain above 500 °C. Thus, we can conclude that the core temperature of Yb + 3:SiO2 should not exceed 500 °C for efficient lasing operation. Under conditions of pulsed amplification in the kHz-regime where the time between the onset of pumping and stimulated emission from the seed pulse ranges from 10 μs to 1 ms, this threshold could be lower still, as the net inversion along the fiber will be less than that of a fiber under these conditions continuously pumped for several upper-state lifetimes. While it is unlikely that a Yb-doped gain fiber would be operated under conditions where the core temperature approached or exceeded 500 °C, as this would cause the protective acrylate/fluoroacrylate coating to melt or burn, it is clear from Fig. 9 that degradation of the laser performance should be expected for temperatures as modest as 80 °C–160 °C. Even as the core temperature is raised from 20 °C to 85 °C, a likely operating regime for kilowatt class fiber lasers, there is an 18% drop in slope efficiency. For temperatures approaching 165 °C the slope efficiency drops further to a net reduction of 21%. Thus, proper thermal management of any high power fiber laser/amplifier system is essential to ensure the highest achievable efficiency. 6. Conclusions We have reported the fluorescence lifetime, absorption coefficient, and emission cross-section of Yb + 3 over the pump absorption and signal bands, and the slope efficiency of a Yb + 3-doped LMA fiber as a function of temperature from 23 °C to nearly 1000 °C. We found elevated temperatures, particularly above 500 °C, severely degraded the lasing performance of Yb + 3-doped fibers. This was due to sharp increases in the absorption coefficient coupled with modest decreases in the emission cross-section in the signal band and a sharp degradation of the fluorescence lifetime, particularly above 500 °C. The rapid reduction in fluorescence lifetime above 500 °C due to the onset of thermal quenching limits the energy storing capacity of Yb + 3:SiO2 and effectively prohibits its use at higher temperatures. This was confirmed in measurements of the slope efficiency where we found that 977 nm-pumped Yb + 3-doped silica becomes opaque for temperatures exceeding 600 °C. Even for core temperatures as low as 85 °C−160 °C,
a potential operating regime for kW-class fiber laser systems, we observed a 21% reduction in both the slope efficiency at 1064 nm and the absorption coefficient at 975 nm. This suggests that thermal issues should be considered and managed even at the modest temperatures found in typical kW-class fibers lasers. Acknowledgments Sandia is a multi-program laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000. Funding for this work was provided by the Defense Advanced Research Project Agency (DARPA) under grant DOE-NNSA. The views expressed are those of the author and do not reflect the official policy or position of the Department of Defense or the U. S. Government. References [1] T.C. Newell, P. Peterson, A. Gavrielides, M.P. Sharma, Optics Communication 273 (2007) 256. [2] H. Kiriyama, N. Srinivasan, M. Yamanaka, Y. Izawa, T. Yamanaka, S. Nakai, Japan Journal of Applied Physics 36 (1997) L1165. [3] D. Sumida, T.Y. Fan, OSA Proceedings of the Advanced Solid State Lasers 20 (1994) 100. [4] D.A. Grukh, A.S. Kurkov, V.M. Paramonov, E.M. Dianov, IEEE Journal of Quantum Electronics 34 (6) (2004) 579. [5] Z. Burshtein, Y. Kalisky, S.Z. Levy, P. LeBoulanger, S. Rotman, IEEE Journal of Quantum Electronics 36 (8) (2000) 1000. [6] J. Dong, M. Bass, Y. Mai, P. Deng, F. Gan, Journal of Optical Society American B 20 (9) (2003) 1975. [7] Z. Zhang, K.T.V. Gratten, A.W. Palmer, Journal of Applied Physics 73 (7) (1993) 3493. [8] P.L. Pernas, E. Cantelar, Physica Scripta T118 (2005) 93. [9] H.M. Pask, R.J. Carman, D.C. Hanna, A.C. Tropper, C.J. Mackechnie, P.R. Barber, J.M. Dawes, IEEE Selected Topics Quantum Electronics 1 (1995) 2. [10] J.Y. Allain, M. Monerie, H. Poignant, Electronics Letters 28 (1992) 988. [11] C.B. Layne, W.H. Lowdermilk, M.J. Weber, Physics Review B 16 (1) (1977) 10. [12] D.E. McCumber, Physics Review 134 (No. 2A) (1964) A299. [13] P.F. Moulton, Journal of Optical Society American B 3 (1) (1986) 125. [14] W.L. Barnes, R.I. Laming, E.J. Tarbox, P.R. Morkel, IEEE Journal of Quantum Electronics 27 (1991) 1004.