Homogeneous and Inhomogeneous Line Broadening in EDFA

Transcription

1 Homogeneous and Inhomogeneous Line Broadening in EDFA Laurent Bigot, Stephan Guy, Bernard Jacquier, Anne Marie Jurdyc Laboratoire de Physico-Chimie des Matériaux Luminescents, UMR du CNRS n 5620 Université Lyon 1, Domaine Scientifique de La Doua, 10 rue A.M. Ampère, Villeurbanne FR Dominique Bayart, Stéphanie Blanchandin, Laurent Gasca Alcatel, Research & Innovation Dpt, Route De Nozay Marcoussis FR Abstract This paper is addressed to determination of the spectroscopic key parameters which are relevant to explain broadening and flatness of gain profile of Erbium Doped Fiber Amplifier (EDFA) versus glass composition. Absorption and fluorescence of bulk glass material give a classical identification of erbium energy levels in the near infrared at 1,54µm. Low temperature measurements allow to assign Stark splittings of both the ground and the excited states. Using site selective spectroscopy, homogeneous and inhomogneous linewidths are determined by resonant and non resonant fluorescence line narrowing measurements (FLN and RFLN respectively). Knowing these paremeters, it is shown a reconstrution of the spectral fluorescence profiles as a function of temperature within the K range. A comparison of all these three parameters determined for several glass compositions gives a good agreement with gain characteristics. Finally room temperature homogeneous halfheight linewidth is extrapolated from the low temperature data in agreement with Gain Spectral Hole Burning determination (GSHB). corresponding author: B. Jacquier, address: jacquier@pcml.univ-lyon1.fr key words: EDFA, luminescence, line broadening, rare-earth-doped glasses Introduction Information carrying capacity of optical fiber has been increased by several orders of magnitude in the last decade [1]. Thanks to the outstanding properties of the Erbium Doped Fiber Amplifier (EDFA), the volume of data transported has doubled during the 30

2 last two years, for instance. However a constant effort is necessary to increase the performances of EDFA used for the 1.5µm amplificationso as to answer to the strong demand of Internet traffic. During the nineties, many progresses have been done with the use of Wavelength Division Multiplexing (WDM) technology. Today, several solutions are studied in order to improve the cost performance of WDM systems: - The further reduction of the spacing between the channels which transport information (Dense Wavelength Division Multiplexing: DWDM) - The broadening of the present C ( nm) and L ( nm) amplification bands. - The use of distributed Raman amplification over the C- or L-band These two solutions bring the benefit of complying with the existing EDFA-based network technologies. Also, they imply to improve the gain characteristics (width, flatness) of the present EDFA. Therefore, a better understanding of the key parameters that can explain those characteristics namely, Stark splitting, homogeneous and inhomogeneous broadening of the transitions involved is then required. Using high resolution spectroscopy (i.e., Resonant Fluorescence Line Narrowing (RFLN, [3]), site selective spectroscopy [4] and Spectral Hole Burning [5], we report evaluation of those parameters in various glassy hosts currently presented as favorable media for broad band amplification (silica, fluoride and tellurite, see Figure 1). 1,0 0,9 Gain (a.u.) 0,8 0,7 ZBLA Si-Ge-Al TeO2 0, Wavelength (nm) Figure 1: relative spectral gain profiles of three main glass compositions for amplifiers The results obtained are very helpful to understand: - the modifications of gain profile with the glass compositions - the behavior of an EDFA working at a saturated regime of the signal. 31

3 First part of the paper reports experimental procedure and the results and second part analyses these results with comparison of extraploted homonegeneous linewidth with previous Gain Spectral Hole burning determination. Then reconstruction curves of the emission profile are shown. Experimental Experimental set-up and description of all the bulk samples used in that study have been reported in a previous paper [3]. Low temperature absorption and broadband excitation of the 1,54µm fluorescence gives an assignment of all the Stark splittings of the ground and first excited states assuming that inhomogeneous broadening is small compare with the average crystal field splitting. It is to be noted such determination of the 4 I 13/2 multiplet is not unambiguous for alumino-silica composition due to a large broadening. A comparison of all these Stark splittings have been reported in a previous paper [see tables I and II in ref. 3]. Furthermore, in all the fluoride glass compositions we have studied, two well defined sites have been identified through modulation spectroscopy [4]. However, if they differ by their emission probabilities i.e different lifetimes, the Stark splittings of both sites are very similar. We also reported low temperature RFLN experiments [3] which allow to determine both inhomogneous and homogenous broadenings. Following the temperature dependences of the homogeneous linewidths recorded in the range of K, a model including Raman and one-phonon relaxation processes shows a good agreement with experiment and permits to extrapolate the room temperature parameters. In the table I, we have compiled all the broadenings measurements which can be used in a reconstruction of the fluorescence spectra and then to build gain profile. Another determination of the homogeneous linewidths have been reported using a saturation technique of the gain of the amplifier, namely Gain Spectral Hole Burning [6]. It consists to work the amplifier in a saturating regime and then by polarization discrimination and spectral substraction to reach hole spectrum at the signal wavelength. In this technique it is not clear what is really measured in terms of width (half homogeneous linewidth or full homogeneous linewdth), this will introduce further developments. On the other hand, low temperature spectral hole burning was recorded for several glass compositions in order to get information on glass structure and more specifically on the role of the interaction between low excitation Two Levels System (TLS) which are usually involved to explain 32

4 the low temperature physical properties of glass. These new results will be reported in a forthcoming paper. inhomogeneous broadening (cm -1 ) Homogeneous 1 broadening (cm -1 ) Homogeneous broadening 2 (cm -1 ) Temperature dependence Alumino-silicate Fluoride Tellurite T 1.39±0.04 T 1.53±0.12 T 1.7±0.13 Table I : comparison of inhomogeneous and inhomogeneous linebroadenings of the three glass compositions: 1 homogeneous width determined by RFLN, 2 homogeneous width determined by GSHB ( half holewidth ) and the temperature dependences were measured in the K range in RFLN experiment. Analysis With the knowledge of the three main spectroscopic parameters determined above it is proposed to reconstruct the fluorescence spectra in the range of temperature investigated (15 to 150 K). The formulation we used was proposed earlier in terms of temperature dependence of emission cross section as follows: 8 σ ( λ) = ( ). σ 2 k1k 2 k = 1 k = p k T where p k2 (T) is the population of the k 2 sublevel of the 4 I 13/2 multiplet and σ k1,k2 (λ) is the cross section of individual transition between the sublevels k 1 and k 2 of the 4 I 15/2 and 4 I 13/2 manifolds. It is a Voigt profile which includes the measured homogeneous and inhomogeneous widths. No further parameters were included but some contributions were estimated from the spectral areas. The two following rebuilt curves for which only temperature is varied give examples of comparison with experiment (Figure 2). This example was developed for the ZBLA glass because we have a better confidence in the knowledge of all the spectroscopic parameters. We may note the ( λ) 33

5 reasonably good agreement with experiment. With such procedure, it is possible to derive room temperature spectrum if one can get more assignements of the transitions between Stark components of high lying crystal field levels in both the ground and excited multiplets. Figure 2: Reconstruction of the fluorescence spectra recorded at 15 and 150 K. The 15 K fluorescence spectrum was normalized to the experimental curve and the most intense curve at 150 K is the experimental one. Note that the vertical scale of the 150 K fluorescence spectrum is comparable with the 15 K spectrum. Good agreement of the reproduced spectra (without no parameter other than temperature T) gives us some confidence to extrapolate the procedure to room temperature. However to reach quantitative description at room temperature, it is 34

6 necessary to include the transitions involving the high lying levels of both multiplets which may contribute to emission profile. Then relative gain calculation can be performed knowing absorption spectrum of the 4 I 15/2 4 I 13/2 transition. Conclusion Along with this report together with other references [3,4] we have shown that the three spectroscopic parameters: Stark splitting, inhomogeneous and homogeneous broadenings may describe unambiguously the emission profiles of the 1,54µm fluorescence. This procedure could be applied to the other glass compositions in order to get some insight on the emission profile when conventional glass composition is modified by codopants in a predictive way. Furthermore, it is a first step in the understanding of gain spectral hole burning which affects notably the long distance optical transmissions. In a undergoing next step, low temperature spectral hole burning is carried out to understand the different behaviors of the studied glass compositions. Acknowledgements: This research program was supported by a joint agreement between the Laboratoire de Physico-Chimie des Matériaux Luminescents associated with CNRS CNRS at the University of Lyon 1 and Alcatel Research and Innovation. References [1] Alcatel Research and Innovation Department website [2] EDFAs, Device and System Development, E. Desurvire, D. Bayart, S. Bigo, B. Desthieux, Wiley (2002) [3] L. Bigot, A.M. Jurdyc, B. Jacquier, L. Gasca, and D. Bayart, Phys. Rev. B, 66, (2002) [4] S. Guy, L. Bigot, I. Vasilief, B. Jacquier, B. Boulard, Y. Gao, J. Non-Cryst. Solids 336, , 2004 [5] L. Bigot, S. Choblet, A.M. Jurdyc, B. Jacquier and J.L. Adam, JOSA, B n , 2004 [6] E. Desurvire, J.L. Zyskind, IEEE Photon. Tech. Lett. 2(4) , 1990 ; J.W. Sulhoff, A.K. Sivrastava, IEEE. Photon. Tech. Lett. 9(12) , 1997; E. Rudkevich, D.M. Baney, IEEE. Photon. Tech. Lett. 11(5) ,