CN1654981A - Method for amplifying optical input signal over widened optical bandwidth and optical amplifier - Google Patents

Abstract

A light input signal is input into a light wave-guide made of non-crystal material doped with rare earth elements in a method of amplifying light input signals on a wide bandwidth, said signal includes the wavelength sphere of at least 80nm and 160nm for the optimum. The pump light is provided to the light wave-guide to provide gain of light to the input signal to be amplified in the wave-guide and provides amplified light signals in the widened 80-160nm sphere, especially including light signals of one end and the other wavelengths.

Description

Method for amplifying optical input signal over widened optical bandwidth and optical amplifier

This application is a divisional application. The application number of the original application is 00819894.2. The original application is the international application PCT/US00/20578 that entered the national phase on March 14, 2003, and its international filing date is July 28, 2000. day.

technical field

The present invention relates to the field of optical amplifiers, and more particularly, the invention relates to the field of optical amplifiers comprising a length of optical fiber having an active dopant that fluoresces in response to pump light.

Background technique

Erbium-doped fiber amplifiers (EDFAs) are widely used in commercial optical communication systems, both as all-optical repeaters and as preamplifiers. The information-encoded signal is greatly weakened by fiber loss after passing through a long communication fiber (usually tens of kilometers), and the function of the erbium-doped fiber amplifier (EDFA) is to amplify these signals to a reasonable power level. The energy efficiency of EDFAs has now been optimized, reducing the (expensive) pumping power requirements. The noise performance of the EDFA has been further optimized so that the noise figure can now approach the theoretical limit by 3dB. The gain flatness of the best EDFAs is now tens of dB over a bandwidth of tens of nanometers. EDFAs are now designed so that their gain depends very little on the polarization of the input signal.

A still very active area of EDFA is gain bandwidth. This gain-bandwidth parameter is very important because it ultimately determines the number of signals at different wavelengths that can be amplified by a given EDFA. The wider the bandwidth, the greater the number of individual signals that can be amplified, and thus the greater the bandwidth (signal bits per unit time) that can be carried by a strictly single fiber. Since the host affects the spectrum of the erbium ion, several host materials, including silica, fluorophosphate glasses, and chalcogenides, have been and will continue to be investigated in an attempt to identify a ⁴ I _13/2 → ⁴ I for Er ³⁺ The _15/2 shift provides greater gain bandwidth to the matrix. In silica-based glasses, the bandwidth is usually divided into C-band and L-band. In close terms, the C-band is the portion of the spectrum below about 1565 nm, while the so-called L-band is the portion of the spectrum above about 1565 nm. The combined total bandwidth of the C-band and L-band is close to 80 nm, although this figure has so far only been obtained by cascading two EDFAs. (For example, see Y. Sun, et al., 80nm ultra-wideband erbium-doped silica fiber amplifier, Electronics Letters, Vol. 33, No. 23, November 1997, pp. 1965-1967). The same is true for fluorozirconate matrices. (See eg S. Kawai, et al., Wide bandwidth and long distance WDM transmission using highly gain-flattened hybrid amplifier, Proceedings of Optical Fiber Communication OFC'99, PaperFC3, February 1999, pp. 56-58). In tellurite fiber, the total bandwidth is also about 80 nm, but can be achieved with a single fiber. (see for example Y. Ohishi, et al., Optical fiber amplifiers for WDM transmission, NTT R & D, Vol.46, No.7, pp.693-698, 1997; Y. Ohishi, et al., Gain characteristics of tellurite -basederbium-doped fiber amplifiers for 1.5μm broadband amplification, Optics Letters, Vol.23, No.4, February 1998, pp.274-276.)

FIG. 1 shows an example of a standard EDFA configuration 100 having an erbium-doped fiber (EDFA) 110 . The optical signal is input into the erbium-doped optical fiber 110 through a first optical frequency isolator 120 and a wavelength division multiplexing (WDM) coupler 122 . An optical pump signal from the optical pump source 124 is also input into the erbium-doped fiber 110 through the WDM coupler. The amplified output signal from the erbium-doped fiber 110 is output through a second optical isolator 126 . The optical isolators 126, 120 are used to eliminate the backward reflection from the output port into the Erbium-doped fiber 110 and the backward reflection from the Erbium-doped fiber 110 to the input port, respectively. As shown in FIG. 1, the erbium-doped fiber 110 can be pumped forward, or backward, or both. Due to the extensive nature of the fiber gain medium, the configuration of Figure 1 yields gain over a large bandwidth. For example, erbium-doped tellurium fibers and erbium-doped chalcogenide fibers have been used in the configuration of FIG. Using tellurium fiber Gain bandwidths of approximately 80 nm have been produced.

Gain over the entire bandwidth (approximately 1525 nm to 1610 nm) cannot be provided with a single fiber when the fiber substrate is silica-based glass. Alternatively, a gain needs to be generated between two adjacent spectral regions, and the outputs from this attention then combined. A common way to obtain a wider gain bandwidth is to use a hybrid amplifier, where two or more amplifiers composed of different substrates are connected. These amplifiers are designed to provide complementary gain spectra, thus yielding a greater overall gain bandwidth than either alone. The method has been successfully demonstrated by following a silicon-based EDFA with a fluorine-based EDFA, which yielded a 0.5-dB-bandwidth of 17 nm. (See P.E. Wysocki, et al., Dual-stage erbium-doped, erbium/ytterbium-codoped fiber amplifier with up to+26-dBm output power and a 17-nm fiat spectrum, Optics Letters, Vol.21, November 1996, pp.1744 -1746.). Recently, the same concept was adopted for two fluorine-based EDFAs. (See Y. Sun, et al., 80nm ultrawidebanderbium-doped silica fiber amplifier, Electronics Letters, Vol.33, No.23, November 1997, pp.1965-1967.)

Figure 2 shows an example configuration 200 with two EDFAs 210 and 220. One EDFA is used to amplify the C bandwidth (approximately 1525-1565 nm), and the other EDFA (upper EDFA) is used to amplify the L bandwidth (approximately 1565-1620 nm). Both EDFAs 210, 220 include respective pump sources (not shown), which are coupled to Erbium-doped fibers using respective WDM couplers, as shown in FIG. 1 . The input signals of different wavelengths spaced apart by a certain amount are split into two parts with a WDM coupler 230 and the amplified output signals are combined in an output coupler 232 . An input opto-isolator 240 and an output opto-isolator 242 operate as described above. Due to the WDM coupler 230, signals with a wavelength shorter than 1565 nm are coupled into the lower branch and propagate to the C-band EDFA 210, and signals with a wavelength greater than 1565 nm are coupled into the upper branch and propagate to the L-band EDFA 22. (Actually, there is a narrow guard band between the C-band and the L-band to avoid overlap of signals in both sides). For example, a silicon-based EDFA can be designed to have an L-band with a flatness of the gain spectrum within 0.5 dB in the range of 1568-1602 nm. This gain flatness can be obtained in part by selecting the appropriate fiber length or employing filters. Both methods are well known in the art.

The C-band EDFA 210 and the L-band EDFA 220 can be formed from the same erbium-doped fiber, or from different fibers, or from different host materials. The C-band and L-band EDFAs 210, 220 can each differ in design, in particular pump wavelength, pump configuration and fiber length.

The upper limit for L-band EDFA 220 is about 1610 nm. Several research groups have made considerable efforts to increase this upper limit by tuning the host material. A difficulty in further increasing this upper limit is the presence of signal excited state absorption (ESA) around 1620 nm for tellurium glasses and around 1610 nm for silica glasses. The ESA constitutes an undesirable signal loss mechanism. According to these results, the bandwidth records of the present Si-based (Y. Sun et al. cited above) and fluorine-based EDFA (S. Kawai et al. cited above) are around 80-85 nm (i.e., the same as Y. Ohishi et al. cited above). The same bandwidth records for tellurium-based EDFAs published in the article).

Contents of the invention

An aspect of the present invention is to provide a method of amplifying an optical input signal over a widened optical bandwidth. The method involves rare earth doped amorphous Y ₂ SiO ₅ . The optical input signals include at least a first optical signal having a first wavelength and a second optical signal having a second wavelength, wherein the second wavelength is greater than the first wavelength. The method includes applying pump light to an optical waveguide, causing said waveguide to provide optical gain to an optical input signal to amplify at least said first and second optical signals.

In one embodiment of the method, the amorphous material is _Y2SiO5 doped with erbium, and the second wavelength is about 160 nanometers longer than the first _wavelength . Alternatively, the amorphous _Y2SiO5 material can also be doped with erbium and _ytterbium .

In a second embodiment of the method, the amorphous material is Lu ₃ Al ₅ O ₁₂ , and the second wavelength is approximately 160 nanometers longer than the first wavelength.

In a third embodiment of the method, the amorphous material is _Y3Ga5O12 , and the second _{wavelength is} about 140 nanometers longer than the first wavelength.

In a fourth embodiment of the method, the amorphous material is _Ca2Al2SiO7 , and the second wavelength _is about 130 nanometers longer than the first _wavelength .

In a fifth embodiment of the method, the _amorphous material is _Y3Sc2Ga3O12 , and the second wavelength is _about 130 nanometers longer than the first _wavelength . Alternatively, the amorphous Y ₃ Sc ₂ Ga ₃ O ₁₂ material can also be doped with erbium and ytterbium. Still alternatively, _{the amorphous Y3Sc2Ga3O12 material} can _also be doped with erbium and _chromium .

In a sixth embodiment of the method, the amorphous material is _Bi4Ge3O12 , and the second wavelength _{is about} 125 nanometers longer than the first wavelength.

In a seventh embodiment of the method, the amorphous material is _GdAlO3 , and the second wavelength is about 125 nanometers longer than the first wavelength.

In an eighth embodiment of the method, the amorphous material is _SrY4 ( _SiO4 ) _3O , and the second wavelength is about 125 nanometers longer than the first wavelength.

In a ninth embodiment of the method, the amorphous material is _LiYF4 , and the second wavelength is about 110 nanometers longer than the first wavelength.

In a tenth embodiment of the method, the amorphous material is _CaF2 - _YF3 , and the second wavelength is about 110 nanometers longer than the first wavelength.

In an eleventh embodiment of the method, the amorphous material is _YVO4 , and the second wavelength is about 90 nanometers longer than the first wavelength.

In a twelfth embodiment of the method, the amorphous material is LiErYP ₄ O ₁₂ , and the second wavelength is about 80 nanometers longer than the first wavelength.

Another aspect of the present invention is an optical amplifier capable of amplifying an optical input signal over an extended optical bandwidth. The optical amplifier includes an optical pump source for supplying optical pump light, and an optical waveguide including a rare earth-doped amorphous material. The optical waveguide is optically coupled to receive optical pump light from an optical pump source. The optical waveguide receives an optical input signal having a plurality of wavelengths. The optical input signals include at least one first optical signal having a first wavelength and at least one second optical signal having a second wavelength greater than the first wavelength. The pump light has a pump wavelength and intensity at the pump wavelength such that the optical waveguide provides optical gain to amplify at least said first and second optical signals.

In one embodiment of the device, the amorphous material is _Y2SiO5 doped with erbium, and the second wavelength is _about 160 nanometers longer than the first wavelength. Alternatively, the amorphous _Y2SiO5 material can also be doped with erbium and _ytterbium .

In a second embodiment of the device, the amorphous material is Lu ₃ Al ₅ O ₁₂ , and the second wavelength is about 160 nanometers longer than the first wavelength.

In a third embodiment of the device, the amorphous material is _Y3Ga5O12 , and the second wavelength is _about 140 nanometers longer than the first _wavelength .

In a fourth embodiment of the device, the amorphous material is _Ca2Al2SiO7 , and the second wavelength is about 130 nanometers longer than the first _wavelength .

In a fifth embodiment of the device, the _{amorphous material is Y3Sc2Ga3O12 ,} and the second wavelength is _about 130 nanometers longer than the first _wavelength . Alternatively, the amorphous Y ₃ Sc ₂ Ga ₃ O ₁₂ material can also be doped with erbium and ytterbium. Still alternatively, _{the amorphous Y3Sc2Ga3O12 material} can _also be doped with erbium and _chromium .

In a sixth embodiment of the device, the amorphous material is _Bi4Ge3O12 , and the second wavelength _is about 125 nanometers longer than the first _wavelength .

In a seventh embodiment of the device, the amorphous material is _GdAlO3 , and the second wavelength is about 125 nanometers longer than the first wavelength.

In an eighth embodiment of the device, the amorphous material is _SrY4 ( _SiO4 ) _3O , and the second wavelength is about 125 nanometers longer than the first wavelength.

In a ninth embodiment of the device, the amorphous material is _LiYF4 , and the second wavelength is approximately 110 nanometers longer than the first wavelength.

In a tenth embodiment of the device, the amorphous material is _CaF2 - _YF3 , and the second wavelength is about 110 nanometers longer than the first wavelength.

In an eleventh embodiment of the device, the amorphous material is _YVO4 , and the second wavelength is about 90 nanometers longer than the first wavelength.

In a twelfth embodiment of the device, the amorphous material is LiErYP ₄ O ₁₂ , and the second wavelength is about 80 nanometers longer than the first wavelength.

Description of drawings

These and other aspects of the invention are described below with reference to the accompanying drawings.

Figure 1 shows an example of a standard EDFA construction including an erbium-doped fiber;

Figure 2 shows an example of a configuration with two EDFAs, one EDFA amplifies the C band and the other EDFA amplifies the L band;

Figure 3 shows an amorphous YAG fiber and a replacement YAG fiber doped with various concentrations of erbium and with approximately 45 milli-pumped amorphous lanthanum at 980 nm;

Fig. 4 represents this amorphous YAG optical fiber and this amorphous lanthanum to replace the backward fluorescence spectrum measurement spectrum of YAG optical fiber;

Fig. 5 shows the influence that erbium molar concentration increases on the forward and backward bandwidth of erbium-doped amorphous YAG and erbium-doped amorphous lanthanum instead of YAG;

Fig. 6 A shows the temporary decay (relaxation) of the fluorescence of the amorphous YAG optical fiber radiation of doping erbium 1%;

Fig. 6 B represents the transient decay (relaxation) of the fluorescence of the amorphous YAG fiber radiation of doping erbium 5%;

Fig. 7 shows the influence of erbium concentration on the fast light (square mark in the lower curve) and the slow light (circle mark in the upper curve) of the erbium-doped amorphous YAG fiber fluorescence;

Fig. 8 represents the first construction of a kind of amplifier for realizing the first embodiment of the method of the present invention;

Fig. 9 represents the second construction of a kind of amplifier for realizing the second embodiment of the method of the present invention;

Figure 10 shows a pre-L-band amplifier comprising a single length of erbium-doped amorphous YAG optical fiber;

Figure 11 shows an amplifier configuration similar to that of Figure 8 but with a longer Erbium-doped amorphous YAG fiber used as an L-band amplifier;

Figure 12 shows the construction of an L-band amplifier in which an erbium-doped amorphous YAG fiber is placed between two WDM couplers that receive the individual pump signals and pump the erbium-doped amorphous YAG fiber forward and backward. YAG fiber;

Figure 13 shows a construction of an L-band amplifier comprising two erbium-doped amorphous YAG fibers connected in series by a WDM coupler, with pump light directed forward to pump the second fiber;

Figure 14 shows an amplifier configuration comprising a first Erbium-doped amorphous YAG fiber and a second Erbium-doped amorphous YAG fiber connected in series by a WDM multiplexer, with pump light;

Figure 15 represents the 1/e decay time constant of the fluorescence power measured from the same fluorescence decay curve shown in Figures 6A and 6B, where the 1/e time constant represents the average of the fast and slow time constants;

Figures 16-23 are reproductions of drawings in the international patent application PCT/US97/00466, the filing date of which is January 9, 1997, published on July 17, 1997, and the international publication number is WO97/25284, The drawings and description thereof are incorporated herein to illustrate a method of making an optical fiber for carrying out the invention wherein:

Figure 16 shows a schematic diagram of one embodiment of an optical fiber insertion tube drawing system for a method of making an optical fiber;

Figure 17 shows an embodiment of a method of drawing optical fiber from supercooled and floating melts in opposite directions according to the principles of WO97/25284;

Figure 18 shows an embodiment of a method of drawing optical fiber from a supercooled melt held in a conical nozzle floater according to the principles of WO97/25284;

Figure 19 shows a graph of typical cooling curves for a 0.3 cm diameter mullite sample floating and melting in an aeroacoustic floater;

Figure 20 represents a typical aluminum-silicon phase diagram;

Figure 21 shows an embodiment of a method of drawing optical fiber from a supercooled melt held in a non-isothermal vessel;

Figure 22 shows a portion of the anomalous aluminum-silicon phase diagram showing the behavior of mullite near its melting temperature; and

Figure 23 shows a typical tension test result diagram of a glass optical fiber containing a mullite composition;

Figure 24 shows the fluorescence spectrum of erbium-doped crystalline aluminum garnet (LuAG).

Detailed ways

The invention relates to a method of amplifying an optical signal over a wide gain bandwidth. The method relies in part on the use of an optical fiber with a broad fluorescent bandwidth. More specifically, Containerless Research, Inc. (CRI) has recently developed a method for fabricating optical fibers and small quantities of samples from a material that forms a nearly inviscid liquid at the equilibrium melting point of a solid. (See J.K.R.Weber, et al., Glass fibers of pure and erbium orneodymium-doped yttria-alumina compositions, Nature, Vol.393, 1998, PP.769-771; and Paul C. Nordine, er al., Fiber Drawing from Undercooled Molten Materials, its International Publication No. WO97/25284, published on July 17, 1997). This International Patent Application is hereby incorporated by reference. A detailed description of this International Patent Application is given as part of the specification of the present application.

The method involves completely melting a solid to form a homogeneous liquid at or slightly above the equilibrium melting point of the crystalline material. To obtain the viscosity required to support fiber drawing, the liquid is subcooled (ie cooled below its equilibrium melting point). The use of subcooling to increase viscosity is a major step in this method. Glass optical fibers are then drawn from this viscous subcooled liquid.

One embodiment of this fiber synthesis method uses a vessel-free processing technique that avoids complex phase nucleation from contact with solid vessel walls, allowing cryogenic cooling to temperatures below the melting point of the crystalline precursor material. Containerless handling technology eliminates contact with foreign objects such as liquid contamination from crucibles. They can also be used to synthesize glass spheres with a diameter of 2-3 mm from various oxidic materials. (see for example J.K.R. Weber, et al., Enhanced Formation of Calcia-Gallia Glass by Containerless Processing, J.Am.Geram.Soc.Vol.76, No.9, 1993, pp.2139-2141; and J.K.R.Weber, et al. ., Aeroacoustic lebitation-Amethod for containerless liquid-phase processing at high temperatures, Rev. Sci. Instrument., Vol.65, 1994, pp.456-465). The new fiber manufacturing process has been applied to a variety of "network" formations that contain neither silicon dioxide (silica) nor any other typical glass fiber components made by conventional fiber manufacturing techniques. This method allows the generation of optical fibers with diameters typically 10-30 micrometers up to 1 meter.

A prototype material _that can be fabricated by _this method is the _Y3Al5O12 composition. The materials discussed in this application are modifications of compositions formed by substituting lanthanum (La), ytterbium (Yb) and/or erbium (Er) atoms for yttrium (Y) atoms within the material. These materials are synthesized from pure oxides in a certain weight mixture, these oxides are Y ₂ O ₃ , Al ₂ O ₃ , La ₂ O ₃ , Yb ₂ O ₃ , Er ₂ O ₃ , the weighing error of each component About 0.5 mg of the total mass of 0.5 g. The pure oxide is 99.999% pure, smaller than 325 powder, and is available from Cerac, of Milwaukee, Wisconsin. Using a "master alloy" of _{Yb2O3 :Er2O3 equal} to 9: ₁ for _a composition containing _Er2O3 at 0.5 mol% enables easy weighing of dopants. The mixture of these oxide components was melted in a laser furnace melter by first melting and homogenizing 0.5 g of the sample, and then obtaining spheres of approximately 3 mm by melting the fragments inside the object again. (See eg JKR Weber, et al., Laserhearth melt processing of ceramic materials, Rev. Sci. Instrum., Vol. 67, 1996, pp. 522-524). These spheres are then lifted by a laser beam in an aeroacoustic levitator (e.g. as in JKR Weber, et al., Aero-acoustic levitation-A method for containerless liquid phase processing at high temperatures, Rev. Sci. Instrumen., Vol. 65, 1994, pp. 456-465) or in a conical nozzle suspender (such as JPCoutures, et al., Contactless Treatments of Liquids in a Large Temperature Range by an Aerodynamic Levitation Debice under LaserHeating, Proc. 6 ^th European Symposium on Materials under MicrogravityConditions , Bordeaux, France, December 2-5, 1986, pp.427-30; JKR Weber, et al., Containerless LiquidPhase Processing of Ceramic Materials, Microgravity Sci.Technol., Vol.7, 1995, pp.279-282; S .Krishnan, et al., Levitation apparatus for structural studies of high temperature liquids using synchrotron radiation, Rev. Sci.Instrum., Vol.68, 1997, pp.3512-3518) are heated to obtain suspension and melting.

Glass spheres can be prepared by rapidly cooling a liquid sample to room temperature without a container by cutting off the power to the heating laser beam. Glass fibers can be prepared by reducing the power of the heating laser until the temperature of the liquid is 1600°-1700°K, where the liquid viscosity is sufficient for fiber pulling. The tungsten insertion tube is then rapidly inserted into the melt and withdrawn at a constant speed of about 100 cm/sec to draw the fiber from the melt. The above-mentioned part of the process is described in Paul C.Nordine, et al., Fiber Drawing from Undercooled Molten Materials, which is an international patent application with a publication number of WO97/25284, published on July 17, 1997NA, and disclosed herein part.

The crystalline form of pure Y ₃ Al ₅ O ₁₂ is known as yttrium aluminum garnet or YAG for short. "YAG" as used herein refers to crystalline materials. Doping with neodymium ions (Nd ³⁺ ) was the earliest and most successful laser material to be demonstrated. Neodymium-YAG (Nd:YAG) lasers have been commercially available for over 25 years, and they are still one of the most efficient, highest powered and most widely used lasers. Conversely, amorphous fibers of Y ₃ Al ₅ O ₁₂ composition discussed here will be referred to as amorphous YAG, and Y ₃ Al ₅ O ₁₂ fibers substituted by lanthanum will be referred to as substituted by amorphous lanthanum (or substituted by La) YAG fiber.

There are two main differences between YAG and the materials discussed here. First, the materials described here are different because they are amorphous glass materials compared to YAG, which is a crystalline material. Second, the materials discussed herein contain lanthanum (La) atoms, ytterbium (Yb) atoms, and/or erbium (Er) atoms in place of some of the yttrium (Y) atoms in _{the Y3Al5O12} composition _. The materials discussed here will be referred to as erbium-doped amorphous YAG, ytterbium-doped amorphous YAG, amorphous lanthanum substitution YAG or amorphous YAG. The term originates from a different purpose and is used for erbium atoms, ytterbium atoms, and lanthanum atoms instead of yttrium atoms. The erbium and ytterbium atoms replace optically active dopants within the host glass material.

The elements yttrium, lanthanum, erbium and ytterbium are elements of the "rare earth" group, ie these elements have atomic numbers 21, 39, 57 to 71. These rare earth elements have a valence of +3, although some rare earth elements, notably ytterbium, also have a valence of +2. The general chemical formulas of all materials synthesized in this specification can be written as RE ₃ Al ₅ O ₁₂ , where the subscript "3" of "RE" represents the total amount of Y, La, Er, and Yb. These glass spheres and fibers contain lanthanum and or erbium and are fabricated in air, pure argon or pure oxygen. Although the foregoing rare earth elements have been described herein, it is understood that other rare earth materials may also be used.

The chemical composition of the material of the synthetic amorphous YAG composition with lanthanum atom, erbium atom and ytterbium atom replacing yttrium atom is given by table I: Chemical composition, mole percent Formed glass material Al ₂ O ₃ Y ₂ O ₃ La ₂ O ₃ Er ₂ O ₃ Yb ₂ O ₃ optical fiber sphere 62.5 37.5 0 0 0 x x 62.5 37.49 0 0.01 0 x 62.5 37.45 0 0.05 0 x x 62.5 37.4 0 0.1 0 x x 62.5 37.2 0 0.3 0 x x 62.5 37.0 0 0.5 0 x x 62.5 36.5 0 1.0 0 x x 62.5 35.5 0 2.0 0 x x 62.5 32.5 0 5.0 0 x x 62.5 29.5 0 8.0 0 x x 62.5 12.5 12.5 12.5 0 x x 62.5 17.75 17.75 2.0 0 x x 62.5 17.75 11.75 8.0 0 x x 62.5 32.0 0 0.5 0.5 x

One of the most attractive features of doped and substituted amorphous YAG materials is that they can be doped with very high concentrations of Er ³⁺ , Yb ³⁺ or other rare earth ions. The maximum concentration that can be obtained in the glass depends on the dissolution rate of Er ₂ O ₃ in the glass. If the concentration of rare earth oxides in the glass melt is too high, crystalline oxide particles precipitate within the matrix. These crystalline particles are generally detrimental to the performance of lasers and amplifiers incorporating the material. A second factor that limits dopant concentration in some materials, especially silicon-based glasses, is the formation of clusters of dissolved dopant ions in the host material, which occur at concentrations less than the prescribed concentration of harmful oxide particles.

Instead of being uniformly distributed in the matrix like the erbium ions, the clustered erbium ions are subjected to a deleterious process known as cross-relaxation. (See eg JL Wagener, et al., modeling ofion pairs in erbium-doped fiber amplifiers, Optics Letters, Vol. 19, March 1994, pp. 347-349). Through this process, when two ions are photoexcited to a metastable (laser) state, one of the ions rapidly loses its energy and gives it to the other ion, resulting in the unwanted loss of population inversion and optical Huge reduction in buff. In silica-based glasses, the maximum concentration _of erbium that can be used is approximately 100 parts per million (ppm) molar _Er2O3 before this effect becomes significant. To avoid this effect, most silicon-based fiber lasers and amplifiers use less than 500 ppm molar Er ₂ O ₃ . This number is relatively high in fluorine-based fibers, where concentrations of several thousand ppm are acceptable. (See eg JSSanghera, et al., Rare earth doped heavy-metal fluoride glass fiber, in Rare Earth Doped Fiber lasers and Amplifiers, MJF Digonnet, Ed., Marcel Dekker, Inc., NY, 1993). In tellurium fibers, this number is a few thousand ppm, or a few tenths of a mole percent. (See Y. Ohishi, et al., Gain characteristics of tellurite-based erbium-doped fiber amplifiers for 1.5-μm broadband amplification, Optics Letters, Vol. 23, No. 4, Febreary 1198, pp. 274-276). The main consequence of these concentration limitations is that so many ions per unit length can be combined in a single fiber, and long fibers are expensive and bulky for making useful devices. For example, a typical silicon-based EDFA requires tens of meters of doped fiber length.

There have been early experimental studies of the spectroscopic properties of new glass materials doped with Er ³⁺ . One aim of these studies is to add several new optical properties to these materials, either in the form of spheres or fibers. In particular, the study is directed to the absorption and radiation cross-section of a material towards the presence or absence of clusters (cross-relaxation) and, at such concentrations, the bandwidth of the fluorescence spectrum (which reflects the gain bandwidth of the material).

For optical characterization purposes, short lengths of erbium-doped amorphous YAG fiber were inserted into capillaries, then bonded with UV-cured epoxy, and then polished. Typically four to five fibers with the same matrix and erbium concentration are installed in the same capillary. Several capillaries were prepared, each with several concentrations (0.5, 1, 2, 5, 8 molar _{percent Er2O3} ). These fibers are exposed. These fibers have an outer diameter of about 30-15 microns or less and a fiber length of about 5 mm or about 10 mm after polishing. Experiments have been carried out with two new oxide glass matrix compositions, namely amorphous YAG (Y ₃ Al ₅ O ₁₂ , with erbium replacing Y) and amorphous lanthanum replacing YAG fibers in which 50% of the yttrium is replaced by lanthanum ( La _1.5 Y _1.5 Al ₅ O ₁₂ , replacing La and Y with Erbium). For all measurements, the fibers were diode pumped from a commercially available braided fiber laser with an optical power of 50 mW at 980 nm.

Fiber absorption is measured by measuring the ratio of output to input pump power, then assuming 100% coupling of the pump into the fiber, and correcting for Fresnel reflections (approximately 8.3%) at each fiber end. Fresnel reflections were calculated assuming that the material has the same refractive index as crystalline YAG, which is approximately 1.8 at these wavelengths. From these data it is possible to deduce the absorption coefficient α _a , and the absorption crossover σ _a = α _a /N ₀ , where N ₀ is the erbium concentration in the fiber. The average value observed in several samples with two different concentrations was σ _a =2.4*10 ⁻²¹ cm ² . It is comparable to Erbium-doped silica fiber with a standard value of 2.2*10 ^-21 cm ² .

Figure 3 shows the fluorescence spectra measured in amorphous YAG fibers and in amorphous lanthanum-substituted YAG fibers, where the fibers are 0.5-1.0 cm in length, 30 microns in diameter, doped with various concentrations of erbium, at 980 nm at approximately 45 milliwatts pumped out. These spectra are measured from the front, ie the spectra represent the fluorescence radiating in the same direction as the fiber is pumped out. All spectra are broad and unaffected by the transition lines of erbium-doped crystals. This observation corroborates the results of X-ray diffraction analysis, which showed that fibers drawn from supercooled liquids of YAG composition are amorphous. (See eg JKR Weber, et al., Glass fibers of pure and erbium orneodymium-doped yttria-alumina compositions, Nature, Vol. 393, 1998, pp. 769-771). In fibers with low erbium concentration, the spectrum exhibits the characteristics of the low power Er ³⁺ fluorescence spectrum typically found in other oxide glasses, i.e. a peak around 1530 nm with a flat flank around 1550 nm. At higher concentrations, these properties become less pronounced and the spectrum shifts toward longer wavelengths because the pump power is insufficient to excite all erbium ions, and the ground state absorption screens out the 1530-1550 nm bandwidth (C-band). In these fibers, the L-band (wavelength greater than 1565 nm) is more prominent. The L-band of this material is extremely wide. The L-band broadens to 1653 nm in 1 cm and 8% doped fiber. It has been observed that the 3dB fluorescence bandwidth reaches 116 nm in 1 cm and 5% YAG doped fiber. This is much higher than any previously reported bandwidth of erbium transitions within a matrix. As a point of comparison, in a standard erbium-doped silica-based fiber at 1650 nm, the fluorescent power typically drops by 50 dB from 1530 nm. In contrast, in the amorphous YAG fiber of the present invention, the fluorescent power is only reduced by about 12 dB (see FIG. 3 ). This can be interpreted as a fluorescence spectrum that further broadens from 20-40 nm to longer wavelengths than erbium-doped silica-based fibers.

The post-fluorescence spectrum is represented in Figure 4, which characterizes the C-band, as expected, but the spectrum here is also very broad. In 0.5 cm doped 8% YAG fiber, the highest 3-dB bandwidth observed for the back spectrum was 121 nm.

Both pre- and post-fluorescence bandwidths generally increase with increasing concentration, as shown in Figure 5. Diamonds represent erbium-doped amorphous YAG, while circles represent erbium-doped amorphous lanthanum instead of YAG. Front and back fluorescence are represented by solid and open symbols, respectively. Within experimental error, both materials have substantially identical properties.

Fluorescence temporal is measured by placing an 80 MHz acousto-optic modulator in the path of the input pump beam and modulating the pump intensity into a square pulse about 20 ms long and short decay time (about 5 ms). relax. The temporal decay of the fluorescence emitted by the fiber after each pump pulse is then recorded with a fast photodetector. For low concentrations, the relaxation curve exhibits mono-exponential decay, as shown for 1% Erbium-doped amorphous YAG fiber in FIG. 6A. For higher erbium concentrations, the relaxation curve is no longer mono-exponential, but decays exponentially with time constant with increasing time. For a 5% erbium-doped amorphous YAG fiber, this is shown in Figure 6B. The curve begins to decay with a time constant _τ1 . Over time, this constant increases (decay slope) until the time constant reaches an asymptotic value _τ2 (greater than _τ1 ) at the lower end of the decay curve.

Another strikingly important result in Figures 6A and 6B is that these fibers are largely unaffected by erbium clusters, even at the highest concentration. If these clusters are present, the relaxation curve exhibits first a very fast relaxation component due to the cluster ions, and then a slow component due to the non-clustered ions. (See for example P.F. Wysocki, et al., Evidence and modeling of paired ions and other loss mechanisms in erbiim-doped silica fibers, in SPIE Proceedings on Fiber Laser Sources and Amplifiers IV, Vii. 1789, 1993, pp. 66-79). In other materials, the fast component is usually 2-3 orders of magnitude smaller than the radiation lifetime (ie expect it to be under 100 microseconds). The fact that this composition has been observed each suggests the presence of negligible clusters of erbium ions in the two new matrices, reaching a concentration of 80% in amorphous YAG, which is approximately 600% higher than the cluster-free concentration in silica-based glasses. times.

The lifetimes _τ1 and _τ2 deduced from the decay curves measured for all concentrations are plotted in Figure 7. The squares represent the lifetime of the fast component (τ ₁ ), the circles represent the lifetime of the slow component (τ ₂ ), and the solid and hollow symbols represent amorphous YAG matrix fiber and amorphous lanthanum substituted YAG matrix fiber, respectively. Both lifetimes decrease monotonically with increasing erbium concentration, as is commonly observed in other substrates. The slow component _τ2 is the radiation lifetime. The lowest radiant lifetime measured in 5% and 8% YAG fibers was about 4 milliseconds, while the shortest lifetime was 0.6-1.0 milliseconds (also in these materials). In contrast, the lifetime of erbium in tellurium glasses, which are now commercial EDFAs, is a few milliseconds and typically 8-10 milliseconds in silica-based glasses.

To better estimate the relative performance of the two materials, Figure 15 shows the 1/e decay time constant of the fluorescence power measured from the same fluorescence decay curves as in Figures 6A and 6B. The 1/e time constant represents the average of the fast and slow time constants, in the sense that it provides the measured average lifetime of the excited state of the erbium ion (this average lifetime controls, inter alia, the dynamics and saturation performance of the amplifier). Figure 15 shows that the 1/e lifetime decreases monotonically with increasing erbium concentration, and that erbium in the lanthanum replacement material has a shorter lifetime, although the difference between the two materials is within experimental error. It was concluded that erbium in both materials exhibited approximately the same bandwidth and the same lifetime, with a slight advantage over amorphous YAG.

The glass optical fiber is doped with more than one kind of rare earth ions, which takes advantage of the interaction between two or more kinds of doped organic sample materials. For example, erbium-doped fibers are doped with Yb ³⁺ , which can pump erbium ions close to 1.06 microns. (See, for example, JE Townsend, et al., Yb ³⁺ sensitised Er ³⁺ doped silica optical fiber with ultrahigh transfer efficiency and gain, Electronics Letters, Vol. 10, No. 21, 1991, pp. 1958-1959). The reason is that in some respects it is useful to pump out about 1.06 microns of EDFA (other erbium-doped fiber devices), a low-cost laser wavelength that is available at higher powers from commercial Nd:YAG lasers or cladding Nd-doped fiber lasers are more favorable than the 980nm and 1480nm laser diodes currently used as pump sources. Er ³⁺ cannot exhibit an absorption band at 1.06 microns, but Yb ³⁺ can. In fibers doped with Er ³⁺ and Yb ³⁺ and pumped out near 1.06 μm, the pump energy is absorbed by the Yb ³⁺ ions. Since the excited state of Yb ³⁺ ( ² F _5/2 ) has energy close to ^{the 4} I _11/2 energy level of Er ³⁺ , the energy of the excited Yb ³⁺ ion is quickly transferred to the adjacent Er ^{3 The 4} I _11/2 energy level of the ⁺ ion. The next relaxation from this ⁴ I _11/2 level brings the erbium ion to a potentially excited state ( ⁴ I _13/2 level). In effect, the ytterbium ions act as mediators by absorbing the pump and transferring the required energy to the erbium ions. This principle is used for the same purpose in optical fibers as for the ytterbium/erbium-doped amorphous YAG optical fibers indicated in the last entry in Table I above.

Glass fibers doped with other rare earth ions besides erbium have also demonstrated potential commercial importance for other optical fiber and waveguide devices. In particular, continuous-wave (CW), Q-switched, and mode-locked fiber lasers have been demonstrated operating from the ultraviolet to the far-infrared. (See, eg, Rare Earth Doped Fiber Lasers and Amplifiers, MJF Digonnet, Ed., Marcel Dekker, Inc., New York, 1993). These above experimental results have been collected together with the ⁴ I _13/2 → ⁴ I _15/2 transition of Er ³⁺ , but their generality should not be ignored.

The erbium-doped amorphous YAG fiber described herein is characterized by its ability to provide a means of amplifying an input signal over a wider bandwidth than that provided by prior devices and methods. By substituting erbium-doped amorphous YAG fiber for silicon-based erbium-doped fiber, the same amplifier configuration provides amplification over a bandwidth extending to about 1650 nm, compared to the upper limit of about 1620 nm for prior known amplifiers.

A first configuration 800 employing the method of the present invention is shown in FIG. 8 . FIG. 8 is similar to FIG. 1 , but in FIG. 8 , the conventional erbium-doped fiber 110 is replaced by erbium-doped amorphous YAG fiber 810 . In association with FIG. 1 above, a first optical isolator 820 , a wavelength division multiplexing (WDM) coupler 822 , a second optical isolator 824 , and an optical pumping source 830 are also included.

The operation of the method of the present invention in FIG. 8 is similar to that of the EDFA 100 in FIG. 1 . However, in contrast to the method of operation of the configuration in FIG. 1 , the optical signal applied to the embodiment in FIG. 8 includes wavelengths in the upper range of the L-band of approximately 1610 nanometers to 1653 nanometers. Since the bandwidth of the erbium-doped amorphous YAG fiber 810 is widened, the optical signal with a wavelength of 1610-1653 is amplified in the erbium-doped amorphous YAG fiber 10 . Thus, the constructed method of operation of FIG. 8 provides a widened gain bandwidth compared to previously constructed methods of operation. In other words, except that the pump is input toward the rear or toward the front and rear directions, the same scheme as that in FIG. 8 can be adopted.

Figure 3 shows that both the amorphous YAG material and the amorphous lanthanum-substituted YAG material have extremely broad C-bands, which broaden to about 1481 nm (3-dB point). For example, in 0.5 cm and erbium-doped 8% amorphous YAG fiber, ^{the 4} I _13/2 → ⁴ I _15/2 conversion spectrum mainly contains the post-fluorescence spectrum of the shorter wavelength part from about 1482 nm to 1603 nm (3dB point). Another aspect of the invention is a method of operating a C-band amplifier employing the configuration of FIG. 8 to amplify a plurality of signals having wavelengths of about 1480 nm or above to 1565 nm or below 1565 nm. The lower wavelengths in this range, approximately 1480 nm, are substantially lower than reported values for the same conversion Er3+ in other known matrices.

Figure 9 shows yet another configuration 900 using the method of the present invention. The configuration 900 of FIG. 9 is the same as the configuration 200 of FIG. 2 and includes a lower C-band EDFA 910 and an upper L-band EDFA 920. These input optical signals are coupled to EDFAs 910 and 920 through a WDM coupler 930 . The outputs of EDFA 910 and 920 are combined in coupler WDM 932 and these combined outputs are provided to one output port. Each EDFA 910 and 920 can have one or more respective pumping sources (not shown). In FIG. 9, the upper (L-band) EDFA 920 includes an erbium-doped amorphous YAG fiber 940 as described above. The lower (C-band) EDFA 910 also includes an erbium-doped amorphous YAG fiber 950, however, the lower EDFA 910 and upper EDFA 920 could also be a conventional EDFA using conventional silica-based or other glass-based optical fibers. The first opto-isolator 60 prevents back reflections from the EDFA 910, 920 into the input port. The second opto-isolator 962 prevents back reflections from the output ports into the EDFAs 910 and 920 .

The configuration 900 in FIG. 9 is operated by applying an input optical signal having a wavelength of about 1481 nanometers or below to about 1653 nanometers or above 1653 nanometers. The WDM coupler 930 splits the input optical signals such that input optical signals having wavelengths less than about 1565 nm are coupled to the lower EDFA 910 and input optical signals having wavelengths greater than about 1565 nm are coupled to the upper EDFA 920. The lower EDFA 910 amplifies the signal in the range of about 1481 nm-1565 nm, and provides the amplified output signal to the output port through the coupler 932 . The taller EDFA 920 amplifies the signal in the range of about 1565 nm-1653 nm, and provides the amplified signal to the output port through the coupler 932 . Those of ordinary skill in the art will appreciate that a deadband within the optical input signal may be provided around 1565 nanometers, precluding amplification of the signal at this wavelength by the EDFAs 910 and 920. As previously mentioned, the use of Erbium-doped amorphous YAG in either the upper EDFA 920 or the lower EDFA 920 or both EDFAs provides amplification over a wider bandwidth than before.

The taller (L-band) EDFA 920 in FIG. 9 can also be constructed like one of the embodiments described in FIGS. 19-14.

Figure 10 shows a pre-L-band amplifier 1000 having a single length 1010 of Erbium-doped amorphous YAG fiber as described above. The fiber 1010 is pumped by an optical pump source 1020 through an input of a first WDM coupler 1022 . The input optical signal of the input port 1030 is provided to the output port 1034 through a first optical isolator 1032 and a second input of a first WDM coupler 1022 . The output of the optical fiber 1010 is provided to the output port 1034 through a second optical isolator 1036 . The input port 1030 of the amplifier 1000 in FIG. 10 is coupled to the upper output port of the WDM coupler 930 in FIG. 9 , and the output port 1034 is coupled to the above-mentioned inlet of the coupler 932 in FIG. 9 . In FIG. 10, the optical pump source 1020 includes a first optical pump 1050 providing pump light at a first optical pump wavelength λ _p1 and a second optical pump 1050 providing pump light at a second optical pump wavelength λ _p2 . Optically pumped 1052. The outputs of the two pumps 1050 , 1052 are coupled in a second WDM coupler 1054 . The output of the second WDM coupler 1054 is provided to the first input of the first WDM coupler 1022 as a composite pump signal. The two pump signals can be multiplexed in other known ways. In a preferred method of operation, the first pump wavelength λ _p1 is 980 nm or 1480 nm (or another Er ³⁺ pump wavelength which is preferably not absorbed by the pump excited state), and the second pump wavelength λ _p2 is about 1555 nm. These pump signals can be generated with semiconductor lasers or with other solid state lasers. The two wavelengths λ _p1 , λ _p2 enter the optical fiber 1010 from the first WDM coupler 1022 . The first pump wavelength λ _p1 plays the same role as the pump wavelength in the C-band EDFA. In particular, the first pump wavelength _λp1 produces a population inversion within the fiber 1010, thereby providing gain. However, the gain spectrum is largest in the C band and weaker in the L band. As a result, the simultaneous radiation in the C-band is greatly amplified, producing a strong C-band amplified spontaneous emission (ASE), which attenuates the gain over the entire gain bandwidth of fiber 1010, including in the L-band gain. Consequently, the gain in the L-band is reduced, and generally the gain in the L-band is too low to be of practical use. In order to increase the gain of the L-band, the ASE of the C-band is reduced. In the method of operation employing the configuration of Fig. 10, the C-band ASE is reduced by the second pump at wavelength λ _p2 . The wavelength of the second pump was chosen to be around 1555 nm, coinciding with the high-gain region within the C-band. The pump at wavelength [lambda] _p2 entering fiber 1010 is strongly amplified, resulting in attenuation of gain in the C-band and L-band. At that time, the λ _P2 pumping with a wavelength of λ also significantly reduced the power in the C-band ASE (for example, see JFMassicott, et al., Low noise operation of Er3+dopedsilica fiber amplifier around 1.6μm, Electronics Letters, Vol.28, No. 20, Sept. 1992, pp. 1924-1925). The reduced power in the C-band ASE results in an increase in both in-band gains. Of these two opposing effects, the latter predominates, so that the L-band gain increases even more than the C-band gain. This type of L-band amplifier has good (low) noise characteristics, but its gain efficiency is low (it needs a high pump power to generate a certain gain).

FIG. 11 shows a configuration 1100 similar to that of FIG. 8 . The configuration 1100 includes a pre-pumped Erbium-doped amorphous YAG fiber 1110 . The fiber 1110 can be pumped from the optical pump source 1120 through a WDM coupler 1130 at 980 nm or 1480 nm (the other erbium wavelength without ESA pumping). The first and second optical isolators 1140 and 142 operate as described above.

The configuration represented in Figure 11 is a pre-pumped configuration. Post-pumping and bi-directional pumping are also possible. Erbium exhibits strong ground state absorption (GSA) in the C-band, but little or no GSA in the L-band. GSA decreases with increasing wavelength in the L-band. Then, as the fiber 1110 increases, at some point the GSA at the C-band wavelength becomes so large that there is no gain for a given pump power, but rather a loss within the C-band. On the other hand, there is very little GSA in the L-band, where gain can be obtained. Therefore, one conventional approach to amplifier design with considerable gain in the L-band is to choose a fiber that is sufficiently long, eg, twice as long as a C-band amplifier requires for the same type of fiber. This method destroys the gain of the C-band and greatly reduces the ASE in the C-band. Thus, the configuration of FIG. 11 becomes an L-band amplifier that can be combined with the upper side of the configuration of FIG. 9 .

FIG. 12 shows yet another configuration of an L-band amplifier, which is similar to that of the amplifier shown in FIG. 11. The configuration 1200 includes an Erbium-doped amorphous YAG fiber 1210 positioned between WDM couplers 1220,1222. The first WDM coupler 1222 is coupled to the output port 1240 through a first optical isolator 1232 . The second WDM coupler 1222 is coupled to the output port 1240 through a second optical isolator 1242 . A first optical pumping source 1250 is coupled to the optical fiber 1210 through a first WDM coupler 1220 , and the pumping light propagates in the optical fiber 1210 along the direction from the first WDM coupler 1220 to the second WDM coupler 1222 . A second optical pump edge 1252 is coupled to the optical fiber 1210 through the second WDM coupler 1222 , and pump light propagates in the optical fiber 1210 in a direction from the second WDM coupler 1222 toward the first WDM coupler 1220 . Thus, the pump light from these two sources propagates bidirectionally within the fiber 1210 . A similar bidirectional configuration using conventional erbium-doped fibers is disclosed for example in H. Ono, et al., Gain-flattened Er3+-doped fiber amplifier for a WDM signal in the 1.57-1.60-m wavelength region, IEEE Photonics Technology Letters, Vol.9, No.5, May 1997, pp.596-598, which shows a similar embodiment using a conventional erbium-doped silica fiber. The configuration of FIG. 12 operates in the same manner as the configuration of FIG. 11 . The second pump source 1252 provides pump light backwards for adjusting the gain spectrum. In particular, the gain spectrum can be smoothed over a large L-band bandwidth by properly adjusting the post-pump power.

FIG. 13 shows another configuration 1300 of an L-band amplifier. The configuration 1300 includes two Erbium-doped amorphous YAG fibers 1310, 1312 connected in series by a WDM coupler 1320. In particular, an input port 1330 is coupled to one end of the first optical fiber 1310 through a first optical isolator 1332 . The other end of the first optical fiber 1310 is connected to one input of the WDM coupler 1320 . A pump source 1340 is connected to the second input of the WDM coupler 1320 . The output of the WDM coupler 1320 is connected to one end of the second optical fiber 1312 . The other end of the second optical fiber 1312 is connected to the output port 1350 through a second optical isolator 1352 . The wavelength of the pump light provided by the pump source 1340 is preferably about 980 nanometers or about 1480 nanometers. The pump source 1340 pumps the second optical fiber 1312 to generate gain in the second optical fiber 1312 . The second optical fiber 1312 produces amplified simultaneous emission (ASE) in the backward direction, ie, toward the WDM coupler 1320 . ASE from the second fiber 1312 is coupled into the first fiber 1310 through the WDM coupler 1320 . The ASE centered at approximately 1550 nm (C-band), the ASE is absorbed by the first optical fiber 1310 . The first fiber 1310 is thus optically pumped out and produces gain in the L-band region. See for example J. Lee, et al., Enhancement of the powerconversion efficiency for an L-band EDFA with a secondary pumping effect in the unpumped EDF section, IEEE Photonics Technology Letters, Vol.11, No.1, Jan.1999, pp. 42-44, which show a similar embodiment using conventional erbium-doped silica fibers.

FIG. 14 shows a configuration 1400 similar to that of FIG. 13 . The configuration 1400 includes a first Erbium-doped amorphous YAG fiber 1410 and a second Erbium-doped amorphous YAG fiber 14112 connected in series by a WDM multiplexer 1420 . In particular, an input port 1430 is coupled to one end of the first optical fiber 1410 through a first optical isolator 1432 . The other end of the first optical fiber 1410 is connected to the input of the WDM coupler 1420 . Pump source 1440 is connected to the second input of WDM coupler 1420; however, unlike FIG. The output of the WDM coupler 1420 is connected to one end of the second optical fiber 1412 . The other end of the second optical fiber 1412 is connected to an output port 1450 through a second optical isolator 1452 . In configuration 1400, pump light from a pump source 1440 is directed onto a first fiber 1410 whose ASE around 1555 nanometers provides pump light to a second fiber 1412. The backward pumped dual EDFA in Figure 14 is slightly less effective than the forward pumped dual EDMA in Figure 13. See for example J. Lee, et al., Enhancement of the power conversion efficiency for an L-band EDFA with a secondary pumping effect in the unpumped EDF section, IEEE Photonics Technology Letters, Vol.11, No.1, Jan.1999, pp.42 - 44, which shows a similar embodiment using a conventional erbium-doped silica fiber.

Obviously, in the art, these different configurations offer certain advantages, such as efficiency and noise characteristics, but also size, complexity, cost, repeatability of production, etc. Furthermore, while one particular configuration may be preferred in a certain type of application, another configuration may be preferred in another application.

Compared with known methods of using other optical fibers or waveguides, including silicon, fluorozirconate, tellurite, etc. to form amplifiers, the method of the present invention has many advantages. For example, one advantage is that the bandwidth of the C-band is wider than that of erbium in other known materials. Fluorescence as small as about 1481 nm (3-dB point) has been observed in some material compositions, indicating increased bandwidth on the short wavelength tail of the C-band. Therefore, in these materials of amorphous YAG composition, the gain bandwidth is generally broadened in the short wavelength tail of the C-band.

A second feature of this approach is that the bandwidth of the L-band is wider than that of erbium in other known broadband materials. Fluorescence up to about 1653 nm (3dB point) has been observed within some material compositions, indicating an increase in bandwidth on the long wavelength tail of the L-band. The L-band bandwidth is approximately 60% wider than that of erbium in any other known fiber material.

The third feature of this method is that by combining the above-mentioned L-band amplifier and the above-mentioned C-band amplifier, using the general circuit shown in Figure 9, or using the circuit shown in Figure 8, the resulting amplifier adds L to the C-band over a bandwidth. The band gain is wider than that of Erbium in any known broadband material. For example, using the circuit of Figure 9, one can observe fluorescence from about 1481 nm to 1653 nm, or a 3dB bandwidth of about 172 nm, which is about twice the broadest fluorescence bandwidth observed with other known broadband materials, inner erbium .

A fourth feature of this approach is that it can be realized using very short L-band and C-band fiber amplifiers, a feature that has several benefits such as more compactness and reduced cost. Another technically important benefit is that the approach reduces unwanted nonlinear effects, which are a concern with known L-band amplifiers. In any optical fiber, when the result of the transmission signal strength is too large over the transmission length of the transmission signal, some nonlinear effects start to occur, such as stimulated Brillouin scattering (stimulated Brillouin scattered, SBS) and cross-phase modulation (cross-phase modulation, XPM). Both effects are detrimental to the operation of EDFAs. The SBS converts forward-traveling light into backward-traveling light with an associated frequency shift. Both the backward transmission of light and the generation of frequency shift are detrimental. The XPM effect can cause information encoded on one signal to be partially transmitted to another, resulting in unwanted crosstalk. The SBS effect and the XPM effect occur within the communication part of the fiber optic connection and within the EDFA itself. The communication fiber is very long (tens of kilometers), but the mode field of the communication fiber is relatively large (ie, low intensity for a given pump power), which reduces the XPM and SBS effects in the communication fiber. An EDFA, on the other hand, involves a shorter length of fiber, but its mode field is usually quite small (ie, erbium-doped fiber carries higher signal strength than communication fiber). Therefore, the XPM effect may be stronger in an EDFA, especially in conventional L-band EDFAs, which involve a longer Erbium-doped fiber than conventional C-band EDFAs. The erbium-doped amorphous YAG material used in the method of the present invention avoids these problems by reducing the fiber length substantially within the ratio of the erbium concentration, which is usually a factor of 100 or more. Therefore, the efficiency of the XPM effect and the SBS effect in these fibers is also reduced due to the reduced fiber length.

For each of the configurations described here, the lengths of the various erbium-doped fibers, the power within the various pumps, and other parameters can also be evaluated theoretically and experimentally using theoretical models and experimental techniques described in the literature and known techniques , to optimize certain EDFA parameters, such as gain, gain curve, noise characteristics, pump efficiency, etc.

Although the method of the present invention has been described above in connection with erbium-doped amorphous YAG fibers, it should be understood that the method can also be implemented using planar waveguide technology. In particular, the method can also be implemented with planar channel waveguides when such waveguides are constructed using erbium-doped amorphous YAG material.

The method of the present invention stems in part from the discovery that the ⁴ I _13/2 → ⁴ I _15/2 transition of Er ³⁺ in amorphous YAG provides an extremely broad fluorescence bandwidth, i.e. about is 121 nm, partly due to the discovery that higher concentrations of Er ³⁺ (and perhaps other rare earth element oxide ions) lead to wider fluorescence bandwidths and wider optical gain spectra.

Applicants have described an amplification method for an L-band erbium-doped fiber amplifier in which a new family of materials is used to produce a ⁴ I _13/2 → ⁴ I _15/2 transition of Er ³⁺ that is more efficient than any other known material Broad L-band spectrum. Applicants have described an amplification method for C-band Erbium-doped fiber amplifiers in which a new family of materials is used to produce a ⁴ I _13/2 → ⁴ I _15/2 transition of Er ³⁺ that is stronger than any other known material Broad C-band spectrum.

Description of International Publication WO97/25284

Paul C.Nordine, et al., in Fiber Drawing from Undercooled MoltenMaterials, WIPO International Publication No.WO97/25284 was published on July 17, 1997, which discloses a method for manufacturing optical fibers, which can be well used to complete this paper invention. The specification thereof is hereby incorporated by reference in its entirety. A detailed description thereof will be given below in connection with Figures 16-23, which correspond to Figures 1-8 of WO97/25284.

According to the international patent application WO97/25284, Paul C. Nordine et al surprisingly discovered that supercooling a certain liquid melt under control can form a melt with sufficient viscosity to draw optical fibers without a large amount of liquid recrystallization , including melts whose melting point viscosity is too low to allow such manipulation. Described here are several examples of optical fibers drawn from supercooled melts of several oxidic materials, whereby optical fibers can be drawn from melts at or above the melting point. Using the method of WO 97/25284 it is possible to rapidly draw optical fiber from this melt at supercooled temperatures and more than 20% below the equilibrium melting temperature. It is also surprising that glass optical fibers can also be drawn from supercooled melts of chemical compositions containing higher concentrations of additives than prior art optical fibers. Furthermore, it can be noted that the fibers drawn according to the method of WO97/25284 have extremely high tensile strength, presumably due to the relatively flawless surface of the fibers of WO97/25284.

The method of WO97/25284 utilizes an "insert tube" to initiate the drawing. The ability to grow optical fiber is significantly influenced by the physical properties of the insert, several conditions under which the insert is used such as material, size, surface finish and insertion depth of the insert tip of the insert, and the residence time of the liquid inside the insert before drawing begins. Control of these properties and processes affects surface wetting and adhesion, allowing meaningful control during fiber drawing.

Briefly, the method in WO97/25284 comprises the steps of: (1) melting a sample of the selected material without a container to produce suspended droplets, or with a container such as in a crucible, (2) The liquid is cooled below the melting temperature, even if the liquid is supercooled, and (3) the supercooled liquid is brought into contact with the insertion tube probe, the probe is pulled out in the desired state, and the optical fiber is drawn from the liquid.

In all cases, control of the fiber diameter can be achieved by controlling (1) the viscosity of the liquid (by varying the melting temperature and/or gas environment), and (2) the fiber draw rate. Gases used in the tests described below include, for example, oxygen, air and argon, although other gases such as nitrogen, helium, carbon monoxide, carbon dioxide, hydrogen and water vapor may also be used. Fast draw speeds and/or slow draw speeds generally favor small diameter fibers. The upper fiber diameter limit is determined by the minimum draw speed that can be used without causing crystallization in the bulk supercooled melt. The lower limit on fiber diameter is determined by the maximum draw speed that can be achieved without breaking the fiber or drawing the fiber from the melt.

Furthermore, optical fibers drawn according to the method of WO97/25284 can be converted into crystalline optical fibers by heating the glass optical fibers to a temperature at which specific glass crystallization occurs.

The method described in WO97/25284 enables the production of high-purity optical fibers from several materials known to have components of these composites that are strong, stiff, slow in creep, resistant to oxidation, or chemically compatible at elevated temperatures Good sex. It is permissible for the optical fiber to be constructed of materials with low electromagnetic radiation absorption properties, such as, but not limited to, materials used in the field of telecommunications. The method disclosed in WO97/25284 also enables the synthesis of homogeneous glass fibers including, for example, the use of high concentrations of dominant components in the field of fiber lasers and fiber laser amplifiers, but not limited to these fields. According to WO97/25284, these fibers can be drawn rapidly, can reduce production costs, can be crystallized to form stable materials, and can be used in oxidation-resistant compounds in areas with very high temperature structures, such as turbo combustion chamber liner and jet deflector. The method of WO97/25284 allows the synthesis of optical fibers with increased tensile strength and stiffness for use in the field of polymeric matrix materials. Furthermore, the method of WO97/25284 allows the construction of biocompatible materials used in the field of organic living organisms. Thus, the method of WO97/25284 greatly expands the range of materials from which optical fibers can be drawn from liquid melts.

Example 1 - Optical fiber drawing device

Figure 16 shows a preferred scheme for drawing optical fiber from a supercooled melt using the principles disclosed in WO 97/25284 under melt conditions with or without a vessel. It should be noted that a new and important feature of the insertion tube in WO 97/25284 is that the optical fiber is not drawn through a die or similar shaping device. In this example, a containerless state is depicted, although the principles and fiber drawing methods can use any melt to draw the desired fiber, including containerized melts.

The container-free state described in FIG. 16 can also be obtained using an air acoustic insertion tube (AAL) to insert a droplet from the drawn fiber. The method utilizes aerodynamic forces from gas injectors 32 , stabilized by using acoustic influences from a three-axis acoustic positioning system 33 . According to WO97/25284, this and other devices for inserting samples are devices described in the prior art, the use of any method for inserting supercooled samples is within the scope of WO97/25284. These methods include, for example, electromagnetic levitation and electrostatic levitation. These methods involve the suspension and retention of the melt under high vacuum, which enables the easy application of the optical fiber drawing method of WO97/25284 to metals, alloys, and materials sensitive to reactions with gases present in air and gaseous environments.

A suspended droplet is formed by heating and melting the sample with a beam from a carbon dioxide laser, although any method of heating is considered to be within the scope of WO97/25284, such as incandescent or arc lamps, microwave heating, induction heating, oven or Suspended in thermal air. Furthermore, any laser beam capable of providing sufficient heat to the sample can be used in the method of WO97/25284. In this particular example, the carbon dioxide laser beam is split into two beams 34 which are focused on opposite sides of the suspended sample, causing the sample to melt. The melt is kept at high temperature until it is completely melted, then the molten droplets are subcooled and kept subcooled by switching off or reducing the heating power supplied.

The stinger and fiber drawing device 31 consist of a 0.01 cm diameter tungsten wire stinger connected to a rod 3, manipulated by a solenoid actuator 4, positioned so that the tungsten Insert the tip of the wire insertion tube into the droplet 1 of the suspension. The contact between the tungsten wire insert and the supercooled melt must be carefully controlled to avoid heterogeneous crystal nuclei in the supercooled melt due to contact with the insert. Although such nuclei are not normally caused by the fiber drawing operation of WO97/25284, the problem of out-of-phase nuclei can be mitigated if previously formed glass optical fibers are used as the insertion tube material. Although tungsten wire is used as the insert in this embodiment, it is contemplated that inserts of various materials and sizes will be utilized depending on the viscosity and desired fiber properties, and such alternative inserts are also within the scope of WO 97/25284.

In this example, a spring operated drawing mechanism 5 provides the pulling force for drawing a defined length of fiber, although any method of drawing fiber is within the scope of WO97/25284. The pull force of the spring can be adjusted so that its force constant is in the range of k+0.1-0.25 lb/in. The fiber drawing speed is further controlled by the frictional damper 6. The electronic control circuit 7 is used to initialize the solenoid actuator and maintain the insertion tube within the droplet for a predetermined time when the solenoid actuator is released for the fiber drawing operation. A high-speed pyrometer 35 is used to monitor the temperature of the suspended sample, which can be displayed in real time on a computer screen as a graph of temperature versus time. The temperature of the melted droplet is maintained at the desired subcooling temperature by increasing or decreasing the intensity and time of exposure to the laser beam.

Certainly, in order to obtain the optical fiber of the required size, the force constant of the spring and the drawing speed of the optical fiber should be adjusted as required. In addition, the fiber drawing device can be any suitable device, such as a motor and wheel assembly, and the pulling force can be adjusted according to the desired physical properties of the fiber and the fiber drawing method.

Fiber drawing is initiated when the laser beam is first blocked to heat and the temperature of the cooled droplet is detected (displayed as a temperature versus time graph on the computer screen). The solenoid actuator is manually activated when the temperature reaches a preselected value, preferably within the optimum drawing temperature range. In this particular embodiment, the solenoid is designed to automatically shut off after sample insertion. The interposer is then withdrawn by the action of the drawing mechanism, drawing the fiber from the droplet. Droplet temperature control is a key part of the method in WO97/25284. At temperatures above the optimum temperature range for drawing fiber, the insertion tube can be pulled from the droplet without drawing fiber. At temperatures below the optimum temperature range for drawing optical fiber, the viscosity of the liquid is so high that the force exerted by the insertion tube on the droplet must be increased beyond the recovery force of the levitation device. Instead of pulling an optical fiber from a droplet, the droplet is drawn from the position. Furthermore, if the melting temperature is too low, the resulting fiber will be shorter than necessary. At intermediate subcooling temperatures, optical fibers of various lengths, from less than 1 millimeter to over 60 millimeters in diameter, can be formed.

Although a range of fiber sizes are reported in this example, it is also understood that a wide range of fiber sizes can be produced, depending on the drawing conditions. The diameter of the fiber is relatively large when drawn at low speed. The diameter of the optical fiber is relatively small when drawn at high speed. The window of an optical fiber is limited by two effects. First, at lower temperatures, the forces on the droplet will eventually pull the droplet from its suspended position. Second, at higher temperatures, the diameter of the fiber increases with the pulling speed, so that the tension can no longer overcome the surface tension and the pulling of the fiber from the liquid is interrupted. In the appropriate drawing temperature range, extremely long optical fibers can be drawn. For example, drawing a 10 mm diameter fiber until a 0.35 cm diameter droplet is reduced to a 0.25 cm diameter can produce a fiber longer than 18,000 cm.

Figure 17 shows a preferred method according to the principles of WO 97/25284, in which optical fiber can be drawn in more than one direction from a suspension droplet under container-free conditions. In this method, suspension droplets 15 are initially formed in suspension nozzles 16 or using another suspension melting technique. Optical fibers 17 are simultaneously drawn in opposite directions from opposite sides of the drop by means of motors and wheels 18 or other drawing means capable of controlling insertion tube operation and fiber drawing speed. The pulling forces are opposite and can be controlled so that they are nearly equal or opposite so that the force generated on the droplet is reduced and the droplet cannot be pulled from its initial position. The figure also shows that heating by a laser beam 19 or other radiant heat source is focused by a lens 20 on the top surface of the suspended liquid droplet. The temperature within the heated region remains above the melting point, while this temperature will drop in other regions of the droplet and can be sufficiently subcooled on the sides of the droplet to allow fiber to be drawn. The fiber optic material pulled from the droplet can be supplemented by adding and melting solid material 21 in the area heated by radiation, which can be added to the suspending liquid at a controlled rate with a powder stream of solid material and one or more small rods . By controlling the fiber drawing speed, adding and melting the material in the area heated by radiation, and maintaining the supercooled area of the drawn fiber, a continuous process for producing long and continuous optical fibers can be obtained. The method disclosed in WO97/25284 envisages The orientation draws multiple fibers from a position without significant displacement relative to the levitated position.

Insertion tube status and operation includes activating the insertion tube by contacting the melt at a temperature above the melting point prior to drawing fiber from the supercooled melt, and allowing the insertion tube to contact the molten droplet for a period of time (typically 1-50 milliseconds ), although priming time may vary depending on the insertion tube, the composition of the containing melt, and the viscosity of the melt. , the distance the insertion tube is inserted into the melt and the insertion and withdrawal speed of the insertion tube from the melt. If the temperature is too high, nucleation through the insertion tube can be avoided by rapidly inserting and withdrawing the insertion tube, but the withdrawal speed does not have to be so high as not to draw the fiber from the melt. At the appropriate supercooling temperature of the melt, the viscosity of the melt increases and the optical fiber can be drawn with a reduced rate of crystallization below that observed close to the melting point of the material.

Example 2 Drawing Optical Fiber Using Tapered Nozzle Suspension

Figure 18 shows a scheme for drawing fiber using a motor and wheel assembly and a Cone Nozzle Levitator (CNL). Fibers with a diameter of 0.25-0.40 cm can be drawn and levitated, although levitating large samples depends on their surface tension and density. . A suspended air flow 8 passes through a plenum 9 , through the nozzle 10 and over the suspended sample 11 . The suspended sample is heated and melted with a carbon dioxide laser beam focused by ZnSe lens 13 on the top surface of the sample. The temperature of the sample is controlled by blocking the laser heating beam with any signal blocking device and varying the laser power. Fiber is drawn from the bottom surface of the subcooled melt using a tungsten insertion tube 77 fed through the nozzle and driven by a reversible stepper motor and wheel assembly 14. The insertion tube consists of a long tungsten wire attached to the wheel, which wraps around the wheel as the fiber is pulled. Of course, it is also conceivable to use other lasers for heating the sample, for example a continuous wave Nd-yttrium-aluminum-garnet (Nd-YAG) laser. Of course, any heating method that can effectively melt these materials and not interfere with the drawing operation can be used instead of a laser. In addition to the insertion tube method and stepper motor and wheel assembly of Example 1 above, it is conceivable to utilize any means capable of powering the drawing process.

A computer can be used to control the direction and acceleration of the motor and wheel assembly 14, manipulate the insertion tube, vary the acceleration of the fiber pulling speed, and obtain a constant fiber pulling speed. A high-speed pyrometer is used to monitor sample temperature and observe cooling behavior. The insertion tube and resulting fiber are wound on a wheel connected to the motor 14 without further mechanical processing, such as drawing through a die. In this example, the fiber drawing speed was up to 120 cm/sec, although the fiber drawing speed was dependent on the individual devices (here motors and wheels) used to power the drawing. The acceleration of the insertion tube was computer controlled and one acceleration equal to 1200 cm/ ^s2 was used, although WO97/25284 contemplates that other accelerations could be used depending on the particular drawn material and desired fiber properties. Fibers up to 60 cm long and 5-20 mm in diameter can be drawn with this device, although other lengths and diameters can be obtained using different drawing conditions. The insertion and fiber drawing operations are typically completed in a cycle time of less than 0.6 seconds, although this time can vary depending on the viscosity of the melt, the speed at which the fiber is pulled, and the crystallization speed. Fiber pulling must be started at a certain pulling fiber speed.

The speed at which fiber is initially and continuously drawn with the CNL apparatus can be slower than that with the AAL apparatus of Example 1 above because the liquid sample is not drawn away from the fiber at high draw forces. At low temperature, when the viscosity of the melt is high, and when the drawing force of the optical fiber is high and the drawing speed is fast, the drawing force is sufficient to displace the melt so that the melt touches the side of the suspension nozzle. The melt is crystallized by contact with the nozzle, but fiber drawing continues until the melt crystallizes to the point of fiber drawing. At temperatures high enough for the draw force to allow the melt to contact the nozzle, the vigilant growth rate is typically low enough to keep the center of the sample in a liquid state for a period of time sufficient to continuously draw fibers with lengths greater than 60 cm.

For example, when contacting a nozzle at a draw speed of 120 cm/s, crystal growth rates are typically less than 1 cm/ Second. The point of contact between the liquid and the nozzle is approximately 0.2 cm from the fiber drawing point. Thus, fiber drawing lasted greater than 0.2 seconds after contact with the nozzle, drawing fibers with lengths of 24-60 cm. A method that allows contact with a supercooled liquid with a mechanical confinement device as a result of displacement of the fiber drawing force can thus be used to draw useful lengths of fiber. Surprisingly, crystal nucleation through the contact mechanical confinement device propagates at a defined velocity, without disturbing the continuous drawing of the optical fiber, until these crystals reach the point at which the CCP is drawn from the liquid.

The cooling rate of the drawn wire can also be evaluated. For example, for an optical fiber with a diameter of 10 mm and drawn at a rate of 100 cm/s in air using a CNL setup, the cooling rate can be calculated as follows

For example, assume that the temperature of the liquid oxide droplet is 1500 degrees Celsius. The thickness of the thermal boundary layer of the droplet is much smaller than the sample diameter at the stagnation point of the suspended gas flow, which is the same point at which fiber is drawn from the liquid. For a typical diameter of 0.3 cm, and a draw speed of 100 cm/sec, the fiber material is drawn through the boundary layer in less than .003 sec. Assuming that the fiber material is in thermal equilibrium with the gas, the cooling rate is on the order of 500,000 degrees Celsius/second. This cooling rate would occur if the heat flux at the fiber surface was about 700 W/cm2 as calculated from the thermal properties of alumina and from the rate of enthalpy change per unit surface area of a 10 mm diameter fiber.

Now assume that the fiber remains hot. At ambient air temperatures as cold as 1500°C, the convective heat flux q" of the fiber is given by the following formula:

${\mathrm{<span\; class="notranslate">\; q</span>}}^{<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; Nu</span>}}_{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; k</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; a</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; d</span>}}$

where T _f and T _a are the fiber and ambient temperature,

k _f is the thermal conductivity of the gas at the average gas "film" temperature (T _f +T _a )/2,

d is the fiber diameter, and

Nu _h is the Nusselt number for heat transfer.

For the assumed conditions, kf is approximately 4*10 ^-4 W/(cm°C), Nu _h is approximately 1, and q" is approximately 600 W/cm. Therefore, the assumption that the fiber is not cooled results in a heat flux that can be maintained as required Compared with the heat flux of the ambient gas heat balance. It can therefore be inferred that the CNL fiber drawing method obtains a cooling rate of hundreds of thousands of degrees Celsius per second for the drawn fiber for a fiber with a diameter of 10 cm. For a fiber with a larger diameter, the cooling rate is smaller , is roughly proportional to the square of the fiber diameter. Therefore, a cooling rate of more than 4000 degrees Celsius per second will be generated for a fiber diameter of 50 cm.

Example 3 - Fiber Drawing from Mullite Melt

Figure 19 shows the time and temperature conditions for drawing optical fiber from a supercooled melt of mullite composition, wherein the molar ratio of _Al2O3 : _SiO2 is 60:40, _using the optical fiber drawing method of WO97/25284.

Figure 19 shows a typical temperature-time course of a suspended sample in an optical fiber drawing test, as a function of temperature measured with an optical pyrometer as a function of time. Prior to the stated time period, the sample was melted with a carbon dioxide laser beam while suspended in the AAL apparatus in a flow of argon and kept at a constant temperature. The temperature range over which the fibers were drawn was determined by drawing the fibers at various temperatures using the insertion and drawing apparatus described in Figure 16 and Example 1. The decrease in temperature over the recording time interval of 0-2.0 seconds indicates that the liquid cools once the laser heating beam is blocked. The temperature range in which fiber was successfully drawn from the subcooled liquid during the cooling cycle is shown on the graph. Over a time period of approximately 2.0-2.2 seconds, the temperature is indicated to rise to the melting point of the sample. This temperature rise occurs when a supercooled liquid spontaneously crystallizes. The energy released by crystallization is sufficient to heat the sample to the melting point, where the temperature remains constant while crystallization continues. Finally, after all the gas has been consumed, the temperature drops due to heat loss from the solid sample.

As shown in Figure 22, the composition of crystalline mullite is in equilibrium with the high temperature liquid and is not contained within the low temperature mullite phase domain. The figure thus shows that mullite, which forms in equilibrium with a liquid at the highest temperature, is thermodynamically unstable at low temperatures. In contrast, the composition of glass optical fibers formed according to the principles of WO 97/25284 can be independently selected from the mullite phase domain at the temperature of the intended application. These glass fibers can be heated to convert them into pure crystalline mullite fibers that are stable with respect to the precipitation of the second phase at the application temperature.

Glass optical fibers can also be drawn in many cases where recalescence (the heat released by crystallization causing the temperature to rise to the melting point) is observed. For example, mullite composition glass fibers as described in this example can be obtained even when reglow is observed.

EXAMPLE 4 - DRAWING OF FIBER FROM SUPERCOOLED MELTS USING A CONTAINED SYSTEM

The description in WO97/25284 also envisages drawing the fiber in a containerized system. Figure 21 shows a preferred embodiment indicating a method using a liquid container which facilitates recrystallization-free drawing of optical fiber from a supercooled melt. An important feature of the method involves establishing and maintaining a temperature gradient within the vessel so that part of the melt is subcooled. In this method, the material of interest 30 is placed in an open container 2, such as a crucible, which is maintained at a temperature above the melting point. The lid 25 is also maintained at a temperature above the melting point, and the lid can initially be placed on the container, achieving a thermal equilibrium inside the container and effecting the melting of the material. The cover can then be lifted or removed, allowing heat loss and cooling to melt the surface. The heat transfer state of the melting surface 23 can be controlled so that the exposed central region of the melting surface is subcooled, thereby drawing an optical fiber from the supercooled liquid. The inner walls of the container 28 and the small portion of liquid 27 adjacent to these inner walls of the container can be maintained at a temperature above the melting temperature so that heterogeneous nucleation cannot occur on these walls.

Fig. 21 shows heating crucible 22 and molten material 30, can draw optical fiber 24 through the opening larger than optical fiber in the cover 25 of lifting, also can utilize the effect such as motor and wheel 26 or other drawing device through the opening of independent guide Drawn optical fiber. Fiber material removed from the melt can be replenished by adding and melting solid material in regions where the melting temperature exceeds the melting point. Fiber drawing is initiated using one or more insertion tubes (not shown) as described in Example 1, or by action of a motor and wheel assembly or other drawing device.

The temperature of the liquid and top surface of the crucible is graphically represented in the bottom of Figure 21 as a function of lateral position on the top surface of the crucible and contained liquid. The equilibrium melting temperature is represented by T _m on the coordinates of this portion of FIG. 21 . The temperature of the part of the liquid and the crucible near the wall of the crucible is above _Tm , while the temperature of the surface of the liquid away from the wall of the vessel drops below _Tm . The core temperature of the exposed liquid surface can be estimated as follows, assuming that the diameter of the container is much greater than the depth of the liquid, allowing heat to transfer from the bottom to the surface. In order to estimate the magnitude of the temperature gradient, it is also assumed that convective heat loss is negligible, that heat is only lost from the liquid surface by radiation, and that radiative heat is no longer reflected back to the liquid surface. The temperature drop within the liquid approximates the following equation, where the right hand side gives the heat flow from the bottom surface to the top surface of the liquid, maintained at the crucible temperature, where the left side gives the radiative heat loss from the liquid surface:

$\mathrm{<span\; class="notranslate">\; \&sigma;\&epsiv;</span>}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="4"\; label="T\; S"\; filenames="A20051000234800441.png,A20051000234800451.png"\; state="\{\{state\}\}">T</figure-callout></span><figure-callout\; id="4"\; label="T\; S"\; filenames="A20051000234800441.png,A20051000234800451.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}^{\mathrm{<span\; class="notranslate">S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k</span>}\frac{{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; C</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}}}{\mathrm{<span\; class="notranslate">\; h</span>}}$

Where σ=5.67*10 ^-12 watts/(cm ² degrees K ⁴ ), the Stephen Boltzmann constant ε for liquid oxides is 0.8, the emissivity of the liquid surface Ts is the temperature of the liquid surface, and To is the temperature of the crucible , h is the depth of the liquid layer.

The average thermal conductivity k of oxides at high temperatures is 0.02-0.2 W/(cm degrees Celsius).

Using mullite as an example, Tc = 1900° _C (slightly above melting point), _Ts = 1670°C (approximately 200°C subcooling), Nordine et al. obtained h = 0.045 cm based on the actual k value.

The above calculations show that the estimated liquid depth is less than 0.5 centimeters, sufficient to obtain cryogenic cooling on the surface of the liquid within the container maintained above the melting point. This depth is small enough to quickly satisfy the assumption that the depth of the liquid is much smaller than the diameter of the container.

Example 5 - Gas Environment and Reglow Effect

It has been found that undercooling, gob formation and fiber drawing conditions depend on the gas environment. In this example, fiber optics have been reported drawn under three different gas environments: air, pure oxygen, and pure argon. However, it is also conceivable to use other gases, such as nitrogen, helium, carbon monoxide, carbon dioxide, hydrogen and water vapour.

For example, a glass block of Y ₃ Al ₅ O ₁₂ composition can be formed without crystallization in argon gas. In air or oxygen, the liquid _Y3Al5O12 composition spontaneously crystallizes upon _{supercooling .} A second example is provided using pure alumina, for which the liquid can be cooled to below its melting point in argon at 450°C before spontaneous crystallization. Optical fibers can be drawn in all cases forming bulk glass samples, crystallization does not occur as the melt cools. Glass fiber can usually be drawn under many conditions where reglow is observed. For example, in an oxygen atmosphere, glass fibers of the mullite composition can be obtained, in which reglow can also be observed. These fibers can be drawn at temperatures above the temperature at which crystal nucleation and spontaneous crystallization of the melt occur.

Under the condition of spontaneous crystallization of the supercooled melt with re-glow, the cooling rate of a large amount of liquid is usually 100-500 degrees Celsius per second. It is known that glass fibers will be formed from the melt if the cooling temperature exceeds the critical cooling rate for glass formation; thus, observations by reflashing indicate that the critical cooling rate cannot be obtained in bulk liquids. However, glass fibers can also be obtained by drawing fibers at liquid temperatures above the spontaneous crystallization temperature. These results indicate that the fiber drawing process results in a cooling rate within the fiber that exceeds the droplet free cooling rate.

Example 6 - New Fiber Composition

Table II lists some new fiber compositions that can be obtained using the method of WO97/25284. Table II lists that these fibers can be drawn using a variety of methods, including the insertion tube, and the drawing apparatus described in Example 1, and the insertion tube and motor wheel assembly shown in Figure 18 and described in the Example. The thaw may be suspended using any suspending device, including the AAL and CNL devices described above, or the thaw may also be included as in Example 4 above. Solid samples are formed from essentially pure oxides melted by a laser furnace. Neodymium or erbium and other materials are used as additives, wherein the ratio of Al ₂ O ₃ :SiO ₂ is 50:50, and the ratio of Al ₂ O ₃ : _{Y 2} O ₃ is 63:37.

For temperatures at or above the melting point, the oscillations and fluid flow observed within the suspended melt indicate a low viscosity of the melt, comparable to that of liquid alumina and substantially less than that of typical glass-forming materials such as pure silica or rich Viscosity of silicon melts. The low viscosity of these melts is also indicative of the fact that optical fibers cannot be drawn from the melts at temperatures above the melting point. However, in all cases described here, glass fibers can be drawn from supercooled melts in the manner of WO97/25284. Glass optical fibers drawn from these melts were uniform in appearance in all cases. Visual inspection under a microscope showed no evidence of second phase precipitation within the fiber.

Glass fiber with high concentration _{of optically} active dopant can _{be synthesized} by _{adding Nd 2} O ₃ and _{Er 2} O ₃ . The additive concentration used was about 1% higher than typical for prior art optical fibers. The method of WO97/25284 obtains optical fibers with high concentrations of these additives by first heating the material to a temperature at which all components form a completely molten liquid. Since the supercooled melt does not crystallize, it remains homogeneous, enabling the formation of glass optical fibers with high additive concentrations.

In addition, the synthesis of optical fibers with very high purity and optical fibers with small concentrations of additives is also possible. The use of vessel-free conditions to maintain the melt enables purification of the melt by (1) evaporation of impurities and (2) reactive vaporization of impurities. For example, alumina, which initially contains about 0.0005 mole percent chromium (5 parts per million chromium), can be purified by melting and heating the liquid to 2400 degrees Celsius without a vessel. The analyzed chromium concentration is reduced to parts per million within a few minutes of processing time. Likewise, many oxides can be purified by evaporation by means known in the art. When the material is handled in the container at very high temperatures, dissolution of the container material in the melt will prevent purification of the liquid. Thus, by purifying the liquid under vessel-free conditions, an optical fiber containing less than 0.0001 mole percent (parts per million) of impurities can be formed. Also, by first purifying the liquid, the additive can be used to control the concentration of the additive within the optical fiber drawn from the liquid in the range of less than 0.0001 mole percent to 50 mole percent.

The synthesis method of glass optical fiber with chemical composition CaAl ₂ O ₄ can use the method of WO97/25284 under the condition of no container, so that the melt is sufficiently supercooled to draw the optical fiber. The fiber pulling method is depicted in Figure 18, the suspension is in oxygen. These fibers are drawn from a melt that is supercooled to a temperature about 200 degrees Celsius below the melting temperature of the material. Once further supercooled, crystallization does not occur, and a large number of CaAl ₂ O ₄ glass fiber samples can be obtained. It is contemplated that other methods of fiber pulling can be used, such as the stinger pulling device of Figure 16, any method capable of pulling fiber is also within the scope of WO97/25284.

Using the method of _WO97 /25284, glass optical fibers can be synthesized from CaO- _Al2O3 melts, these fibers can be used as biocompatible structural materials, and will not cause silicosis if they are inhaled, as described in US patent 5552213 As disclosed, the entire specification thereof is hereby incorporated by reference.

Using _the method of WO97/25284, it is also possible to form glass optical fibers with the chemical composition mineral forsterite _Mg2SiO4 . This mineral is thermodynamically compatible with the mineral enstatite _Mg2Si2O6 , which is known in the prior art, and _{Mg2SiO4 ,} which is known in _the prior art, is an interphase weakening coating ₍ interphase weakening coating), for toughening composite materials. Forsterite optical fibers were formed using the fiber insertion tube assembly shown in Figure 16 and from the melt suspended and supercooled within the conical nozzle suspension assembly.

Figure 20 shows a phase equilibrium diagram of the aluminum-silicon system showing the total composition between pure silica and alumina. As can be seen in Table II the work disclosed in WO 97/25284 obtains glass optical fibers in a wide range of compositions including pure mullite compositions which are stable at low temperatures.

Example 7 - Optical fiber crystallization

Optical fibers obtained according to the method of WO97/25284 can also be crystallized. Table III reports mullite composition glass fibers such as 60:40 _Al2O3 : _SiO2 at temperatures of 1100 _and 1200 degrees Celsius. These results suggest that drawing glass fibers from supercooled melts and then heating to intermediate temperatures can produce crystalline fibers with controlled chemical compositions that are stable at intermediate temperatures.

The fiber draw rate is controlled, usually exceeding the crystallization rate of the supercooled melt, and the resulting cooling rate within the fiber is usually greater than the critical cooling rate for glass formation in the material from which the fiber is being drawn. The crystallization rate or critical cooling rate of glass formation cannot be strictly a function of temperature. For mullite, a drawing speed of 30 cm/s was sufficient to avoid crystallization of the melt. The crystallization rate of mullite is about 3 cm/s at a temperature of 200K below the melting point.

The liquid yttrium oxide-aluminum crystallization rate is greater than the crystallization rate of the liquid mullite composition. Using the motor and wheel assembly described in Figure 18, a yttria-alumina glass fiber was drawn at a rate of 30 cm/s in a flow of pure argon, with a length of several millimeters and a length of 60 cm at 100 cm/s . The liquid is cooled to about 200 degrees Celsius below its melting point.

As shown in Example 2, the cooling rate in the fiber decreases as the fiber diameter increases. The draw speed for obtaining an optical fiber of a given diameter also decreases as the diameter of the optical fiber increases. Therefore, in the drawing of a large-diameter optical fiber, a state may occur in which the cooling rate obtained in the optical fiber is less than the critical cooling rate for glass forming. The optical fiber obtained in this state contains at least some crystalline material. In addition, if the crystallization speed exceeds the fiber drawing speed in the fiber drawing state, the crystallization formed in the fiber will propagate in the fiber, causing crystallization of the supercooled liquid forming the fiber, thus ending the fiber drawing process.

Table III provides experimental tension data for glass fibers drawn from supercooled melts and for crystalline fibers formed by heating the drawn fibers in air. It should be noted the high tensile strength of the fiber thus drawn. Compared to the mullite composition fiber internal tensile strength values up to 6.4 GPa obtained according to the principles of WO97/25284, the tensile strength achievable in prior art fibers with the same composition is limited to less than 3 GPa.

Table II Chemical composition of glass optical fibers drawn from supercooled melts

Chemical composition, mole fraction Additives

Alumina-Silicon Oxide Materials:

0.50Al ₂ O ₃ +0.50SiO ₂

0.50Al ₂ O ₃ +0.50SiO ₂ Nd ₂ O ₃ , 1% to 20% by weight

0.50Al ₂ O ₃ +0.50SiO ₂ Er ₂ O ₃ , 1% to 20% by weight

0.60Al ₂ O ₃ +0.40SiO ₂

0.67Al ₂ O ₃ +0.33SiO ₂

0.69Al ₂ O ₃ +0.31SiO ₂

0.70Al ₂ O ₃ +0.30SiO ₂

0.71Al ₂ O ₃ +0.29SiO ₂

Alumina-Yttrium Oxide Material

0.63Al ₂ O ₃ +0.37Y ₂ O ₃

0.63Al ₂ O ₃ +0.37Y ₂ O ₃ Nd ₂ O ₃ , 5mol% substitute Y ₂ O ₃

other materials

0.50Al ₂ O ₃ +0.50CaO

0.30Al ₂ O ₃ +0.70CaO

0.67MgO+0.33SiO ₂ (forsterite)

0.50Al ₂ O ₃ +0.50La ₂ O ₃

0.35Al ₂ O ₃ +0.35LiO+SiO ₂

Table III Optical fiber characteristics of mullite composition

Fiber status Fiber diameter, _m Tensile strength, Gpa

Such as drawing 32.0 6.45

Such as drawing 20.5 4.68

Such as drawing 32.7 5.21

Such as drawing 30.5 6.14

Such as drawing 33.0 5.55

Crystallized at 1100°C 19.0 0.78

Crystallization at 1200°C 8.0 1.00

Crystallization at 1200°C 28.0 0.66

Description of other matrix fiber materials

Most crystalline oxide laser substrates do not yield amorphous products, so the optical properties of an amorphous form have not been tested. There is therefore no information on the optical properties of this matrix in the current literature. In particular, the document does not recommend the fluorescence bandwidth of the amorphous form. The applicant has studied the fluorescence spectra of many reported erbium-doped laser crystals, and the applicant has used the fluorescence spectrum in a given crystal as information to begin to calculate the corresponding amorphous fluorescence bandwidth of a material. It is hoped that amorphous crystalline materials can also have a wide bandwidth. fluorescence bandwidth.

For example, the fluorescence spectrum of erbium-doped crystalline lutetium aluminum garnet (LuAG) is shown in Fig. 24, as in AA Kaminsky, et al., Investigation of stimulated emission from _{Lu 3} Al ₅ O ₁₂ crystals with Ho ³⁺ , Er ³⁺ and Tm ^{3 +} ions, phys.stat, sol.(a), Vol.18, K3l, 1973 described. The LuAG spectrum exhibits a series of discrete peaks. Because in any crystal, each peak corresponds to one or more of the Stark sub-levels between the ⁴ I _13/2 energy level of Er ³⁺ and the Stark sub-level of ^{the 4} I _15/2 energy level. Different radiative relaxation transitions. The total fluorescence bandwidth Δλ of Er ³⁺ within the material was determined as the distance between the longest and shortest fluorescence wavelengths in the fluorescence spectrum. As can be seen in Figure 24, for Erbium-doped LuAG, these wavelengths are 1520 nm and 1670 nm, and the bandwidth is 150 nm. Next, the erbium fluorescence bandwidth in amorphous LuAG is estimated to be close to the measured crystalline bandwidth Δλ. In other words, it is presumed that within the LuAG crystal the discrete fluorescence spectrum of erbium at about 1550 nm transforms into the overall continuous spectrum of amorphous LuAG, as found by comparing the properties of erbium-doped crystalline YAG with those of erbium-doped amorphous YAG The transitions are similar.

Previous studies have used a large number of erbium-doped crystals to identify a variety of materials with potentially large fluorescence bandwidths. In particular, these studies are for YAG improvements, as it is expected that small chemical modifications of the YAG matrix (such as replacing lutetium with yttrium) will not substantially alter the favorable bandwidth fluorescence and gain properties of erbium-doped amorphous YAG.

Table IV lists the maximum bandwidths thus determined. The first column of Table IV lists the chemical formula of the host fiber material, and the second column of Table IV lists the dopants and co-dopants contained in the host fiber material. The third column in Table IV lists the fluorescence bandwidths Δλ extrapolated from published fluorescence spectra using the method described above. The fourth column of Table IV indicates whether the ESA was observed in the crystal at 980 nm. Table IV lists amorphous materials that have broad fluorescence bandwidths around 1550 nm.

Table IV Potential Broadband Oxides Er3+ doped crystal dopant fluorescence bandwidth YAG Er 160nm Y ₂ SiO ₅ Er 160nm Y ₂ SiO ₅ Er, Yb Lu ₃ Al ₅ O ₁₂ Er 160nm Y ₃ Ga ₅ O ₁₂ Er 140nm Ca ₂ Al ₂ SiO ₇ Er 130nm Y ₃ Sc ₂ Ga ₃ O ₁₂ Er, (Yb, Cr) 130nm Bi ₄ Ge ₃ O ₁₂ Er 125nm GdAlO ₃ Er 125nm SrY ₄ (SiO ₄ ) ₃ O Er 125nm LiYF ₄ Er 110nm CaF ₂ -YF ₃ Er 110nm YVO ₄ Er 90nm LiErYP ₄ O ₁₂ Er 90nm

Although Table IV now provides little information on the performance of the investigated pumped ESAs, Table IV provides a starting point for the study of EDFA broadband amorphous substrates.

Although described here in connection with fiber amplifiers, the method of amplifying optical signals of the present invention can be implemented using integrated optical amplifiers fabricated from bulk samples of amorphous YAG using one of several techniques known in the art of.

Claims (4)

1. A method of amplifying an optical input signal over a widened optical bandwidth, said method comprising: inputting an optical signal into an optical waveguide comprising rare earth-doped amorphous SrY ₄ (SiO ₄ ) ₃ O material, the optical signal comprising at least one first optical signal having a first wavelength and one second optical signal having a second wavelength an optical signal, wherein the second wavelength is about 125 nanometers greater than the first wavelength; and Pump light is applied to the optical waveguide such that the waveguide provides optical gain to the optical input signal to amplify at least the first and second optical signals. 2. The method of claim 1, wherein the amorphous _SrY4 ( _SiO4 ) _3O material is doped with erbium. 3. An optical amplifier for amplifying an optical input signal over a widened optical bandwidth, said optical amplifier comprising: an optical pump source providing optical pump light; and An optical waveguide comprising a rare earth-doped amorphous SrY ₄ (SiO ₄ ) ₃ O material, the optical waveguide optically coupled to receive optical pump light from an optical pump source, the optical waveguide receiving light having a plurality of wavelengths an optical input signal comprising at least one first optical signal having a first wavelength and at least one second optical signal having a second wavelength that is about 125 nanometers greater than the first wavelength, The pump light has a pump wavelength and intensity at a pump wavelength such that the optical waveguide provides optical gain to amplify at least the first and second optical signals. 4. The optical amplifier of claim 3, wherein said amorphous _SrY4 ( _SiO4 ) _3O material is doped with erbium.

Images