Gain Inhomogeneity in L-band Phosphosilicate-based Erbium-Doped Fiber Amplifiers Li Qian 1, Davide Fortusini and S. D. Benjamin Corning Photonic Technologies, Corning Incorporated, SP-ZV-, Corning, New York, USA,1831 l.qian@utoronto.ca Abstract: We report large pump-mediated inhomogeneity in an L-band phosphosilicate-based EDFA, pumped in the 18 nm absorption band. We have investigated inhomogeneous effects as a function of average inversion level, input signal power, and pumping configuration. OCIS codes: (6.21) Fibers, erbium; (6.232) Fiber optics amplifiers and oscillators; Introduction The inhomogeneous behaviour of an Erbium-Doped Fiber Amplifier (EDFA) results from the Stark splitting levels associated with the I 13/2 and I 15/2 states being slightly different at different erbium sites within the glass host. It can have many adverse effects on the performance of the EDFA. For example, inhomogeneous gain saturation near the spectral region of a strong signal, also known as spectral hole burning, can cause havoc to amplifier gain control and degrade the transient performance of a wavelength-division-multiplexed network during a wavelength add/drop event [1-3]. Another manifestation of the erbium inhomogeneity is the pump-mediated inhomogeneous (PMI) effect, which refers to the non-homogeneous spectral gain deviation as a result of a small variation in pump wavelength. For long-haul optical systems or large optical networks that employ many EDFAs, PMI effects should be strictly controlled in order to meet the tight tolerance on EDFA gain deviation. It has been shown that PMI effects are very large (2-3 db of gain deviation) for C-band EDFAs [-6], but previous studies on L-band EDFAs have seen little PMI effect (not distinguishable from measurement uncertainty), whether the EDFA was pumped in the 98 nm or the 18 nm absorption band [7]. However, these studies were carried out on EDFAs made with germanosilicate or aluminosilicate erbium-doped fibers. Research has shown that modifying the erbium host material systems can further extend the L-band amplification bandwidth [8-1], and in particular, phosphosilicate-based erbium fiber has been successfully deployed in a 5 nm gain-flattened EDFA with full functionality operating in the extended L-band [11]. But a study of PMI effects in phosphosilicate-based erbium fiber has not been previously reported. In this paper, we report the first observation of large PMI effects in an L-band phosphosilicate-based EDFA pumped in the 18 nm absorption band. We have systematically investigated the PMI effect on phosphosilicate EDF gain spectrum as a function of average inversion, input signal power and pump configuration. 1 The corresponding author is currently with the Department of Electrical and Computer Engineering, University of Toronto, 1 King s College Road, Toronto, Ontario, Canada, M5S 3G. Tel: (16) 96-5517 Fax: (16) 971-32 Email: l.qian@utoronto.ca
Experimental Results and Discussion A single-coil amplifier operating at room temperature was used for this investigation. The active coil was a 2-m long phosphosilicate fiber doped with Er 3+ at a concentration of 37 ppm, fusion spliced to SMF28 pigtails with splice loss.1db at each end. Wavelength combiners (18/155 type) were placed both at the input and output of the coil to allow forward, backward, or bi-directional pumping. Narrow-band lasers of approximately equal power and uniformly covering the extended L-band (1568 to 1616 nm) at 1 GHz spacing were multiplexed and fed to the input of the EDFA. Gain was measured by taking the signal spectra at the input and output of the amplifier via an optical spectrum analyzer, and therefore includes the input- and output-stage component losses, which are constant for all the measurements. Two pairs of grating-stabilized pumps were used: the first pair at 165 nm, and the second pair at 187 nm. The wavelength variations for all pumps were within ±.5nm over their operating range. Pump power at the entrance of the erbium coil as a function of drive current was characterized for each pump. Table 1 EDFA Gain Measurement Conditions λ p = 165 nm λ p = 187 nm Forward pumping, average inversion.32 P in = 2dBm, 5dBm, 1dBm P in = 2dBm, 5dBm, 1dBm Forward pumping, average inversion.3 P in = 2dBm, 5dBm, 1dBm P in = 2dBm, 5dBm, 1dBm Forward pumping, average inversion.6 P in = 2dBm, 5dBm, 1dBm P in = 2dBm, 5dBm Bi-directional pumping, average inversion.9 P in = 2dBm, 5dBm, 1dBm P in = 2dBm, 5dBm Bi-directional pumping, average inversion.53 P in = 2dBm, 5dBm, 1dBm P in = 2dBm, 5dBm The gain spectrum was measured at three different input powers and five different average inversion levels, as given in Table 1. The five inversion levels (estimated values given in Table 1) cover the typical operating range of erbium coil average inversion in an EDFA. Note that, for a multi-coil EDFA, the average inversion of the individual coils can be quite different from each other, and therefore our investigation covers a large range of average inversions. For each inversion level, the condition where P in = 2 dbm and λ p =165 nm was used as a reference, and for all other conditions in the same row as the reference in Table 1, the pump powers were adjusted so that the gain shape has the best match to the reference gain shape. This was to ensure that any dynamic gain tilt was minimized, and the average inversion level was kept to be as close to the reference level as possible. When bi-directional pumping was used, in order to ensure the same pump distribution was used for 165 and 187 nm pumping, the forward and backward pump powers were set to be equal. 2-2 - -6-8 -1 165nm forward pump 187nm forward pump -12 156 158 16 162 8 7.5 7 6.5 6 5.5 5.5 165nm forward pump 187nm forward pump 156 157 158 159 16 161 162 13 12 11 1 9 8 7 6 5 165nm bidirectional pump 187nm bidirectional pump 156 158 16 162 Gain Deviation (db) Forward (.32) Forward (.3) 1.2 Forward (.6) Bi-dir (.9) Bi-dir (.53) 1.8.6..2 -.2 -. -.6 156 158 16 162 (a) (b) (c) (d) Figure 1. Comparison of gain spectra under 165nm and 187nm pumping at P in = 2dBm. (a) Forward pumping at average inversion of.32. ; (b) Forward pumping at average inversion of.3; (c) Bi-directional pumping at average inversion of.9. (d) Gain difference between 165nm and 187nm pumping for different inversion levels (indicated in parentheses) and pumping configuration.
As seen from Figure 1, significant PMI effects, which resulted in a peak-to-peak gain variation as much as 1.2 db, were observed in the extended L-band. These PMI effects are qualitatively similar to those reported for C-band in germanosilicate or aluminosilicate EDFAs [-5]. Moreover, by comparing the spectral gain deviation obtained between 165 and 187 nm pumping for various inversion levels and pump configurations (Figure 1d), we can clearly see that the PMI effect are much more affected by pump configuration than by the change in average inversion level. The significant difference made by pump configuration can be qualitatively explained as follows: in a forward pumping configuration, the pump first generates C-band ASE, and then the ASE amplifies the L-band signal as they propagate together along the fiber; while in a bi-directional pumping configuration, the L-band signals are directly amplified by the pump near the output of the erbium coil, and hence a more significant PMI effect. Our observations at other input power levels (5 dbm and 1 dbm) are consistent with those made at 2 dbm input level. The pump wavelength difference used here is 22 nm, which is generally considered large compared to the typical wavelength variation found among a batch of pumps. However the central wavelength of a typical 1xx nm pump without grating stabilization can drift as much as 2 nm over the operating range of drive current. Consequently, when an EDFA is operating under different power conditions, considerable gain deviation can result from pump wavelength drift. Therefore, low-ripple amplifier designs, especially those operating in the 1568 162 nm range, would benefit from wavelength-stabilization of the pump lasers, for example, by means of a fiber grating..3.2.1 -.1 -.2 forward pump avg. inv =.3 Pin=1dBm and 2dBm -.3 156 158 16 162 16.3.2.1 -.1 -.2 forward pump avg. inv =.3 Pin=1dBm and 2dBm -.3 156 158 16 162 16.3.2.1 -.1 -.2 bi-dir pump avg. inv =.9 Pin=1dBm and 2dBm 156 157 158 159 16 161 162 163 (a) (b) (c) Figure 2 Inhomogeneous gain deviation observed for different input levels (2dBm, 5dBm and 1dBm) under fixed λ p =165 nm. (a) forward pumping, at average inversion of.3. (b) bi-directional pumping at medium average inversion of.9. (c) bi-directional pumping at high average inversion of.53..2.1 -.1 -.2 -.3 Pin=5dBm Pin=1dBm 156 157 158 159 16 161 162 163 Figure 3. Inhomogeneous gain deviation between forward and bi-directional pumping, observed for input power of 2, 5, and 1 dbm. Pump wavelength was fixed at 187nm, average inversion for all cases was held at ~.9. Since gain inhomogeneity originates from the fact that Er 3+ ions occupy slightly different sites in the glass matrix and therefore have different spectral broadening as well as slightly different Stark split levels, we should expect inhomogeneous behaviour even without pump wavelength variation. If different portions of these inhomogeneous sites were excited due to different power levels or different
pumping configurations, inhomogeneous gain deviation would result, even when average inversion is kept constant. It is important for EDFA designers to learn the magnitude of such inhomogeneous effects under different excitation conditions. We therefore measured the inhomogeneous gain deviation with fixed pump wavelength. As seen in Figure 2, the inhomogeneous effects observed for the different input levels are small but noticeable (peak to peak ~.2.dB). The inhomogeneous effect also tends to be larger for wavelength region below 158 nm, which, as expected, coincides with the region where gain slope is the steepest. Even for the same input level, we observed inhomogeneous gain deviation between forward pumping and bi-directional pumping configuration (Figure 3), but its magnitude is also relatively small (.3 db peak-to-peak). Conclusion We have investigated the various manifestations of erbium gain inhomogeneity in an L-band phosphosilicate-based EDFA. We have observed large pump-mediated inhomogeneous gain deviation in the extended L-band (1568 to 1616 nm) when the EDFA was pumped in the 18 nm absorption band. This behaviour is distinctly different from L-band germanosilicate- or aluminosilicate-based EDFAs, and is qualitatively similar to those reported for C-band conventional EDFAs. The PMI effect was much more pronounced for bi-directional pumping configurations. Although pump wavelength variation has the largest effect on spectral gain deviation, small inhomogeneities were also observed for different input power levels and different pumping configurations under a fixed pumping wavelength. References 1. F. A. Flood, Impact of Pump and Signal Wavelength on Inhomogeneous Characteristics of L-band EDFA s, in Technical Digest of Optical Fiber Communications Conference, paper WG6-1 (Baltimore, USA, 2) 2. D Bayart, Y. Robert, P. Bousselet, J-Y Boniort and L. Gasca, Impact of spectral hole-burning for EDFAs operated in the long-wavelength band, in Technical Digest of Optical Amplifiers and Their Applications Conference, paper WD5-1 (Nara, Japan, 1999) 3. M. Kakui and S. Ishikawa, Long-wavelength-band optical amplifiers employing silica-based erbium doped fibers designed for wavelength division multiplexing systems and networks, IEICE trans. Electron., E83-C, 799-815 (2).. K. W. Bennett, F. Davis, P. A. Jakobson, N. Jolley, R. Keys, M. A. Newhouse, S-J Sheih and M. J. Yadlowsky, 98 nm band pump wavelength tuning of the gain spectrum of EDFAs, in Technical Digest of Optical Amplifiers and Their Applications Conference, paper PD (Victoria, Canada, 1997) 5. M. J. Yadlowsky and L. J. Button, Pump-mediated inhomogeneous effects in EDFAs and their impact on gain spectral modeling, in Technical Digest of Optical Fiber Communications Conference, paper TuG5 (San Jose, USA, 1998) 6. N. E. Jolley, F. Davis, Measurement of pump mediated inhomogeneity in an EDFA at liquid nitrogen temperatures, in Technical Digest of Optical Amplifiers and Their Applications Conference, paper TuD6 (Vail, USA, 1998) 7. P. N. Kean, S. J. Wilson, M. Healy, R. Di Muro, N. E. Jolley and F. Davis, Pump induced inhomogeneity of gain spectra in conventional and extended-band EDFAs, in Technical Digest of Optical Fiber Communications Conference, paper WA-1 (San Diego, USA, 1999) 8. Y. Ohishi, A. Mori, M. Yamada, H. Ono, Y. Nishida, and K. Oikawa, Gain characteristics of telluritebased erbium-doped fiber amplifiers for 1.5-um broadband amplification, Opt. Lett., 23, 97-9 (1998) 9. A.J.G. Ellison, D.E. Goforth, B.N. Samson, J.D. Minelly, J.P. Trentelman, D. I. McEnroe, B.P. Tyndell, Extending the L-band to 162nm using MCS fiber, in Technical Digest of Optical Fiber Communications Conference, paper TuA2 (Anaheim, USA, 21)
1. M. Kakui, T. Kashiwada, M. Onishi, M. Shigematsu, and M. Nishimura, Optical amplification characteristics around 1.58um of silica-based erbium-doped fibers containing phosphorous/alumina as codopants in Technical Digest of Optical Fiber Communications Conference, paper TuC3 (Vail, USA, 1998) 11. L. Qian, D. Fortusini, S. D. Benjamin, G. Qi, P.V. Kelkar, and V. L. da Silva, Gain-flattened, extended L-band (157-162 nm), high power, low noise erbium-doped fiber amplifiers, in Technical Digest of Optical Fiber Communications Conference, paper ThJ, (Anaheim, USA, 22)