1548 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 10, NO. 11, NOVEMBER 1998 Modeling of Dy -Doped GeAsSe Glass 1.3- m Optical Fiber Amplifiers D. T. Schaafsma, L. B. Shaw, B. Cole, J. S. Sanghera, and I. D. Aggarwal Abstract— We present a model for optical amplification at 1.3 m using Dy in fibers made from a low phonon energy glass, based on GeAsSe. This model uses in-band pumping at 1.28 m, takes into account the spectral distribution of amplified spontaneous emission, and allows for bottlenecking of excited ions into the intermediate states in Dy as well as the excited state absorption (ESA) from those levels. Using data obtained from spectroscopic measurements and Judd–Ofelt calculations, our model shows that very high gain ( 30 dB) is possible in short lengths (40–100 cm) of fiber. Given the very high quantum efficiency of the radiative transition in this glass, we show that bottlenecking and ESA should not have a significant impact on device performance. We also predict that devices made from this fiber should have a very high tolerance to the passive loss of the fiber. Index Terms— Communication system performance, infrared spectroscopy, optical fiber amplifiers, optical fiber communica- tion, optical fiber materials, optical materials/devices. I. INTRODUCTION D Y -DOPING for 1.3- m optical fiber amplifiers has recently been investigated in a variety of chalcogenide glasses [1]–[3] but little attention has been paid to the effects of ESA, bottlenecking, and amplified spontaneous emission (ASE). Dy is of interest for 1.3- m optical fiber amplifiers because of its high product in chalcogenides, but due to the presence of intermediate states in Dy, a low phonon energy host is needed to minimize the effects of multiphonon quenching of the excited state. We present here a model for Dy in such a glass which takes into account the spectral distribution of amplified spontaneous emission (ASE), the bottlenecking of populations in the intermediate states, and excited state absorption (ESA) from those states. Our model is based on spectroscopic data and Judd–Ofelt calculations for these glasses [4]. For Dy -doped amplifiers, selenide-based glasses are a natural choice over sulfides, fluorides, or oxides, due to their lower phonon energy (highest peak at 350 cm ), good chemomechanical durability, amenability to fiberization, high rare-earth solubility (up to 1500 ppm), and high radiative quantum efficiency. Though the absorption edge of the Se glass is slightly red-shifted from a comparable S-based glass ( 200 nm from As S ), the reduction in multiphonon quenching should more than compensate for the increase in passive loss. Manuscript received April 27, 1998; revised June 17, 1998. D. T. Schaafsma is with the U.S. Naval Research Laboratory, Code 5606, Washington, DC 20375 USA. L. B. Shaw, B. Cole, J. S. Sanghera, and I. D. Aggarwal are with the U.S. Naval Research Laboratory, Code 5606, Washington, DC 20375 USA. Publisher Item Identifier S 1041-1135(98)07940-3. Fig. 1. Excitation and decay mechanisms in Dy pumped near 1.3 m. MP indicates multiphonon decay. We have made Dy -doped fibers from this composition with low losses ( 1–3 dB/m) in the 1–2 m region [5]. Spec- troscopy [4] has shown that the radiative quantum efficiency of the 1.3- m( F ) level is very high ( 90% with no clustering or quenching observed up to 1100 ppm). It has been suggested [1] that codoping with Tb is needed to overcome the bottlenecking problem in Dy and that fiber loss could be critical due to the long length required for amplifiers made from sulfide fibers. In the selenide-based fibers, the H level is somewhat longer-lived, and thus codoping would not work the same way that it does in sulfides, but our model and the spectroscopic data suggest that a low phonon energy selenide glass fiber doped solely with Dy should be relatively insensitive to bottlenecking and ESA problems, as well as to passive loss. We have considered three basic models for this system. (For reference, a schematic representation of emission and absorption processes in Dy pumped near 1.3 m can be found in Fig. 1.) The simplest model is that of a quasi-four-level (Q4L) laser (or a simple two-level laser, depending on your preference for nomenclature), with ASE and ground state absorption (GSA) in the signal band, as has been described in many texts [6]. In order to simplify the analysis slightly, we assume that the two excited levels ( H 2.97 ms, and F 0.3 ms) act as one state, with the decay dominated by the short-lived state ( F ). We have computed 1041–1135/98$10.00  1998 IEEE