This invention relates to optical glasses and to optical waveguide devices.

Recently, optical amplifiers using erbium-doped silica fibre have been used instead of electronic regenerators, to directly amplify optical signals in optical data transmission systems.

Erbium-doped fibre amplifiers are characterized by their high gain, low noise and high pump efficiency, which means that pumping can be achieved using a single stripe laser diode. These devices have been commercially available since about 1990 for long-haul telecommunication systems operating in the third low-loss window at a wavelength around 1.55 μm.

However, most of the world's installed land-based fibre-optic networks are operated at a wavelength close to 1.3 μm in the so-called second low-loss window of silica optical fibres used in the optical transmission systems. The erbium-doped amplifiers mentioned above are not suitable for amplifying optical signals in the 1.3 μm region.

There is therefore a need for fibre amplifiers to amplify optical signals in the 1.3 μm wavelength region.

At present, there are three major rare-earth candidates for use in making 1.3 μm fibre amplifiers namely; neodymium, praseodymium and dysprosium.

Neodymium has an optical transition between ⁴ F _{3 2} and ⁴ I _13/2 levels that shows fluorescence near 1.3 μm. Incidentally, however, there is excited state absorption (ESA) occurring around 1.3 μm as well, which is detrimental to the amplifier performance. Recent study shows that by choosing a low refractive index multi-component glass as host, the 1.3 μm ESA spectrum shifts to shorter wavelength region due mainly to the so-called "nephelauxetic" effect which subsequently allows gain to be observed near 1.3 μm.

Moreover, the neodymium ⁴ F _3/2 level has an advantage of exhibiting much less concentration quenching effect in host materials in general. This allows much shorter length of fibre to be used, thereby relaxes the need for the fibre to be ultra low-loss which is often difficult to attain in such multi-component glass fibres.

Nevertheless, because the branching ratio for emission in neodymium at 1.06 μm and 0.9 μm limits emission at 1.3 μm to about 10 %, there remains the challenge of how to limit the more efficient emission at the shorter wavelength, which tends to deplete the ⁴ F _3/2 level and subsequently decrease amplification at 1.3 μm. Trivalent praseodymium and dysprosium are the other two rare-earth ions that have been proposed for use in 1.3 μm fibre amplifiers.

Because the respective 1.3 μm fluorescent level in praseodymium or in dysprosium has a rather narrow energy-gap relative to the next adjacent energy level, it becomes crucial that a low phonon-energy host material is used in order to minimize the nonradiative decay via multiphonon relaxation process. This means that the commercially preferred silica fibre host cannot be used in this application.

The best performance reported so far has been from praseodymium doped fluoride fibre amplifiers (PDFFAs). Although a fluoride (ZBLAN) host glass exhibits only about a 4% quantum yield for the 1.3 μm emission, these devices have attained a small signal gain of around 23 dB with about 120 mW of pump power from a single stripe laser diode.

In practice however, the high pump power required severely shortens the lifetime of the pump diode and consequently reduces the commercial application of these devices. In order to construct a more pump efficient Pr ^3"1" - (or Dy ^{3 +} -) doped 1.3 μm fibre amplifier, it is therefore critical that the chosen host material must have a maximum phonon energy notably lower than that (580 cm ^"1 ) of fluoride ZBLAN glass. Accordingly, it has been proposed to use chalcogenide glass as the host, which largely comprises of group VI elements, i.e., sulphur and/ or selenium and/or tellurium, and metal and/ or semi-metal elements such as As, Ge, Ga, Sb, Al, In, Sn,

Pb, Bi, Zr, P etc.

Chalcogenide glass typically has a maximum phonon energy of less than 450 cm ^"1 , which sufficiently reduces the probability of nonradiative decay and substantially increases the quantum yield and, subsequently, the pump efficiency for the device. Modelling has shown that an optimized chalcogenide glass waveguide doped with praseodymium and pumped with about 50 mW at a suitable wavelength will provide a small signal gain in excess of 20 dB. A similar waveguide doped with

dysprosium will similarly produce in excess of 20 dB of small signal gain for 50 mW of pump power.

Among the known chalcogenide glass forming compositions, gallium lanthanum sulphide (GLS) glasses have been identified as one of the most promising candidates, mainly due to their good visible transmission and excellent rare-earth solubility.

Fibre fabrication from the gallium-lanthanum sulphide (GLS) glasses has been demonstrated recently, but not without difficulties. The thermal stability of the GLS glasses at the temperature where the fibre drawing can be performed is relatively low. This is chiefly because the temperature of suitable viscosity for fibre drawing is just slightly below that of the onset of crystallization in these glasses. As a result, successful fibre drawing requires a great deal of dedication in keeping the fibre drawing furnace with a highly reliable temperature stability and accuracy. This in reality can be extremely difficult to fulfil. Therefore, in order to allow a fibre amplifier to be fabricated easily at 1.3μm, there is a need for an optical glass with a relatively low phonon energy (such as the phonon energy of a GLS glass) but which has an improved thermal stability with respect to GLS glass so as to avoid devitrification in fibre drawing and a good optical transmission at the ultraviolet(UV)/ Visible region for efficient pumping of active ions. Further previous arrangements are disclosed in US-A-5 392 376 and US-A-5

378 664.

This invention provides optical glass having a composition: 50 to 80 mole-percent NS _X ; 5 to 50 mole-percent MS _X ; and 1 to 40 mole-percent RH _X ; where

S is sulphur; H is a halide selected from the group consisting of Iodine, Bromine and Chlorine; N is gallium, or gallium with at least one cation selected from the group consisting of indium, aluminum, arsenic, antimony, bismuth, germanium, boron, silicon and phosphor; M is lanthanum, or lanthanum with at least one cation selected from the group consisting of the rest of the lanthanide series, yttrium, sodium, potassium, calcium, barium, zinc, cadmium, tin, lead, lithium, mercury, silver, thallium and strontium; and R is at least one cation selected from the group

consisting of caesium, rubidium, potassium, sodium, barium, strontium, calcium, zinc, lead, lanthanum, lutetium, yttrium, scandium, lithium, beryllium and magnesium.

Further aspects of the invention are defined in the appended claims. Embodiments of the present invention disclose halide-containing gallium- lanthanum sulphide glasses suitable for optical waveguide device applications such as waveguide (either fibre or planar) lasers, optical amplifiers and superfluorescent sources. While preserving the essential characteristics of low phonon energy and good rare-earth solubility, these modified GLS glasses in addition exhibit enhanced optical transmission further into the UV/ Visible region of the electromagnetic radiation spectrum, and more importantly, some also exhibit improved glass thermal stability, compared to the related non-halide containing gallium-lanthanum sulphide glasses. Both factors are beneficial to their optical waveguide device applications, providing respectively ideal conditions for efficient pumping of active ions and for fibre drawing (when the waveguide of concern is a fibre).

In particular, when doped with rare-earth ions such as Pr ³⁺ or Dy ^{3 +} , these glasses form the core of a waveguide as a pump-efficient optical amplifier operating in the second telecommunication window at a wavelength close to 1.3 μm.

Glasses according to embodiments of the invention have a general composition formula NS _X - MS _X - RH _X , expressed in the compound as per atomic metal basis, wherein S is sulphur, H is I, Br or Cl, and N, M, R are the associated metal elements. Embodiments of the glasses comprise, in mole percent, of 50 - 80% NS _X , where N is gallium or further at least one other network forming cation selected from the group consisting of indium, aluminum, arsenic, antimony, bismuth and germanium, 10 - 50% MS _X , where M is Lanthanum or further at least one other network modifying cation selected from the group consisting of the rest of the lanthanide series, yttrium, sodium, potassium, calcium, barium, zinc, cadmium, tin and lead, and 1 - 40% RH _X , where R is at least one cation selected from the group consisting of caesium, rubidium, potassium, sodium, barium, strontium, calcium, zinc, lead, lanthanum, lutetium, yttrium and scandium. The suitable amount of RH _X substituted into the primary GaSj ₅ - LaS _{j 5} system further promotes the glass formation, where the substituting halide is structurally theorized as intermediate in the

glass compositions.

The invention will now be described by way of example with reference to the accompanying drawings, throughout which like parts are referred to by like references, and in which: Figures la and lb are schematic graphs illustrating thermal characteristics of glasses in the system of όSGaS ₅ (35-X)LaS _{j 5} XCsI, where X = 2, 5, 10, 15, 20, 25, 30;

Figure 2 is a schematic graph illustrating a differential thermal analysis (DTA) curve of a Csl modified glass showing definitions for Tg, Tx and Tm; Figure 3 is a schematic graph illustrating UV/ Visible transmission spectra of

Csl modified and pure GLS glasses;

Figure 4 is a schematic graph illustrating the 1.3 μm fluorescent decay of a Pr ^{3 +} doped Csl modified GLS glass;

Figure 5 is a schematic graph illustrating a 1.3 μm fluorescent spectrum of a Pr ^{3 +} doped Csl modified GLS glass;

Figure 6 is a schematic graph illustrating the quantum efficiency of Pr ³⁺ 1.3 μm ^ ₄ - ³ H ₅ transition in a Csl modified GLS glass;

Figure 7 is a schematic graph illustrating the quantity Tx-Tg against halide concentration for chlorine, bromine and iodine containing glass; Figure 8 is a schematic graph illustrating the decay lifetime of Pr ³⁺ 1.3um emission against CsCl concentration in CsCl modified GLS glass;

Figure 9 is a schematic diagram of an optical fibre waveguide fabricated using the above glasses;

Figure 10 is a schematic diagram of a planar waveguide fabricated using the above glasses;

Figure 11 is a schematic diagram of an optical amplifier; and

Figure 12 is a schematic diagram of an optical fibre laser.

In order to improve not only the thermal property but also the UV/Visible transmission property of the glass, the embodiments relate to glass formation with the addition of halide compound(s) into the binary GaS _{j 5} - LaS _{t 5} (GLS) glass system.

Fabrication of the halide-containing GLS glasses can be achieved by melting, then quenching, the designated batch mixture of a suitable amount of high purity (3N

to 5N) sulphides and halide(s) through a multi-step temperature process in a vitreous carbon crucible inside a silica tube.

The well-mixed powder batch is prepared in a nitrogen-purged glovebox before moving into the crucible placed in a silica tube furnace purged with argon gas. The multi-step temperature process is aimed at fully compounding the batch constituents to facilitate in assuring a structurally homogeneous glass. One typical temperature process, for example, is to increase the temperature slowly at l°C/min first to 300°C for 5hrs, and at 2°C/min to 500°C for another 5hrs, then at 4°C/min to 650°C again for 5hrs and finally at 5°C/min to 1150°C for 3hrs for a batch of about 50 grams in weight.

After quenching, the glass is transferred into an annealing furnace operating at about 500°C - 550°C, depending on the glass composition, for 6-12 hours.

Figures la and lb show the thermal characteristic temperatures of a series of caesium iodide (Csl) modified GLS glasses identified by a standard differential thermal analysis technique using a Perkin-Elmer DTA-7 thermal analyzer operating at a heating rate of 20°C/min. The Csl is introduced by substituting an equivalent amount, in mole percent, of LaS _{j 5} following a formula όSGaS _{! 5} (35-X)LaS _{1 5} XCsI, where X = 2, 5, 10, 15, 20, 25, 30 and 35.

The thermal stability of a glass is commonly expressed either by the temperature difference between the onset of crystallization, Tx, and the glass transition, Tg, ie, (Tx - Tg), or by the so-called Hruby factor, Hr = (Tx - Tg)/(Tm -

Tx), where Tm is the melting temperature. The greater is the (Tx - Tg) or the Hr, the more stable the glass is.

A typical differential thermal analysis curve for a Csl modified GLS glass is shown schematically in Figure 2, together with the definitions for Tg, Tx and Tm.

As can be seen from Figure 1 , the thermal stability of Csl modified GLS glasses initially decreases, then increases at the Csl concentration of about 10 mol% and gradually reaches a peak plateau at a Csl concentration around 25 mol% .

Structurally, the results indicate that initially the iodine primarily acts as a network terminator in the form of -S -Ga-I, splits the covalent glass-forming network consisting of [GaS ₄ ] tetrahedra and reduces the glass stability. However, as the Csl concentration increases to about 10 mol%, a major fraction of the iodine

starts to form a bridging structure, -S-Ga-I → Ga-S-, entering the glass-forming network through the formation of an additional coordinate (dative) bond, thus further enhancing the glass stability. This is very similar to the case of the formation of an Cs:Ga:S:Cl glass.

The ratio of the bridging iodine to the non-bridging (terminating) iodine, (Ga-I→Ga)/(Ga-I), undergoes a maximal plateau around the Csl concentration of 25 mol% in the system.

Table 1 lists the data of thermal properties of the glasses in the system presented in Figure 1.

Table 1: Thermal properties of 65GaS (35-X)LaS _S XCsI, X = 0, 2, 5, 10, 15. 20,

25, 30 .

Csl (mol%) Tg (°C) Tx (°C) Tm (°C) Tx-Tg CC) Hr

0 576 700 812 124 1.11

2 579 701 815 122 1.07

5 574 694 795 120 1.18

10 566 693 775 127 1.54

15 558 691 762 133 1.87

20 547 690 760 143 2.04

25 536 692 757 156 2.40

30 515 638 734 123 1.28

The colour of the glasses in the system changes from pale red to light yellow as the Csl concentration increases. This is in agreement with the expected characteristic additivity of band-gap for a glass upon its constituent compositions. Figure 3 shows the UV/Visible transmission property of a 15 mol% Csl doped GLS glass.

In comparison, the transmission property of a pure GLS glass is also presented in the figure. This much enhanced glass transparency in the UV/Visible region of the electromagnetic radiation spectrum is of great importance for their applications in optical waveguide devices. It allows the operation of efficient pump of active ions and also a wider accessible wavelength range for such a pump.

The embodiment of the subject glasses has a general composition formula NS _X - MS _X - RH _X expressed in the compound as per atomic metal basis, wherein S is sulphur, H is I, Br or Cl, and N, M, R are the associated metal elements. The glasses comprise, in mole percent, of 50 - 80% NS _X , wherein N is gallium or further at least one other network forming cation selected from the group consisting of indium, aluminum, arsenic, antimony, bismuth and germanium; 5 - 50% MS„, wherein M is Lanthanum or further at least one other network modifying cation

selected from the group consisting of the rest of the lanthanide series, yttrium, sodium, potassium, calcium, barium, zinc, cadmium, tin and lead; and 1 - 40% RH _X , wherein R is at least one cation selected from the group consisting of caesium, rubidium, potassium, sodium, barium, strontium, calcium, zinc, lead, lanthanum, lutetium, yttrium and scandium. The suitable amount of RH _X substituted into the primary GaS _{j 5} - LaS _{j 5} system further promotes the glass formation, where the substituting halide is structurally theorized as an intermediate in the glass compositions.

The thermally more stable glasses, which warrant fibre production when the waveguide in concern is a fibre, can be located in each system experimentally by a simple differential thermal analysis (DTA) technique, as the example shown above at the previous paragraph. Whereas, such a process may not be necessary if the glass is intended for use in a planar structure.

All the compositions possess improved UV/Visible transmission properties compared to the corresponding non-halide containing pure GLS glass.

As an example for optical waveguide device applications for the glasses when doped with active rare-earth ions such as Pr ^{3 "1"} , Figures 4 and 5 present the 1.3 μm fluorescent spectrum and decay-curve, respectively, for Pr ³⁺ from the *G ₄ level in a 500 ppm (in weight) Pr ₂ S ₃ doped Csl modified GLS glass. The fluorescence is centred at 1.345 μm showing a somewhat stretched exponential decay (stretch factor

0.94) with a fitted lifetime of 329 μsec. Whereas, the commonly defined first , second and third e-folding lifetimes are 327 μsec, 350 μsec and 364 μsec respectively.

It is believed that this non-exponential decay characteristic for the Pr ^{3 +} 1.3 μm fluorescent emission has two possible causes. The first is owing to the structurally different sites in which the Pr ^{3 +} ions may reside. The second is probably associated with the competing non-radiative multiphonon relaxation process.

The 329 μsec lifetime observed for the 1.3 μm transition in the glass is longer than the 280 μsec shown in a typical pure GLS host, presumably owing to the reduced refractive index of the glass through the addition of Csl. Quantum efficiency for the Pr ^{3 +} 1.3 μm G ₄ - ³ H ₅ transition in the glass, measured using a self- calibrating technique, has a value 60% as shown in Figure 6. This is comparable to the 63 % measured in a pure GLS host, but considerably higher than the 43 % shown

in a Ba:Ga:Ge:S host glass both measured by the same method (see Quimby et al, Optics Lett, 20, (19), (1995), 2021).

The good quantum efficiency, long lifetime and superior UV/Visible transmission make the glass an extremely attractive host for Pr ^{3 +} in applications as efficient 1.3 μm optical amplifiers.

Technical Background Discussion

Two arguments behind the choice regarding enhancement for the properties concerned are discussed in detail below. Firstly the thermal property, it has become widely accepted that in general halide is structurally compatible with chalcogenide in glass formation and greatly extends the glass-forming region from the initial chalcogenide to the so-called chalcohalide glasses with homogeneous structure.

We have further contemplated the published information on the particular structural aspects of the GLS glasses in the literature. The structure of GLS glasses reported consists of a covalent network of [GaS ₄ ] tetrahedra, intercalated by the relatively ionic La - S channels, characterised by using an EXAFS (extended X-ray absorption fine structure) technique. This description is a close resemblance to the modified-random-network structural model proposed for the alkali silicate glasses. However, it is in deep contrast to the structural rules set for the conventional chalcogenide glasses.

In conventional chalcogenide glasses, all chemical bonds are essentially covalent and the coordination number of each participating element complies with the so-called 8-N rule, wherein N is the group number of the element concerned in the Periodic Table.

We have evaluated the spectral properties of the trivalent praseodymium doped into the GLS glasses using the well established Judd-Ofelt analysis procedure. Among the three Judd-Ofelt intensity parameters Ω ₂ , Ω ₄ and Ω ₆ , it is well known that the parameter Ω ₂ is directly proportional to the covalency of the chemical bond relating the rare-earth in the glass. It is found that the Ω ₂ has a value of 7.3xl0" ²⁰ cm ² for the trivalent praseodymium in a typical 70GaS _{j 5} 30LaS _{! 5} glass. This compares with 13.2xl0 ^"20 cm ² and 4.95xl0 ^"20 cm ² for the same ion in a conventional

chalcogenide glass based on GeS ₂ and in a typical oxide glass respectively.

The results suggest that the La - S bonds in GLS glasses do appear to be rather ionic, assuming the praseodymium has similar strucmral sites as the lanthanum does in the glasses, thus agree with the conclusions made by Benazeth et al on the structure of GLS glasses.

Since the structure of GLS glasses follows essentially the modified random- network structural model, one approach to further stabilize the glasses therefore is to look for a further component which has a structural function of intermediate in the glasses. This is similar to the case for alkali silicate glasses, where alumina is frequently used as the intermediate to further enhance the stability.

Consequently, we have postulated that the halide (except fluoride) would play the strucmral role of intermediate in the resulting modified GLS glasses. This is chiefly because of the very similar electronegativity and covalent radius between sulphur and chlorine, bromine and iodine. The relevant parameters are shown in table 2.

Table 2: Some elemental properties of S, Cl, Br and I

Property Sulphur Chlorine bromine Iodine

Electronegativity 2.4 2.8 2.7 2.2

Covalent radius 104 99 114 133

Because of the unusual structure of the GLS glasses when compared to the conventional chalcogenide glasses, the addition of halide enhances the glass formation in such a way that the halogen enters the covalent network by forming a structural tetrahedra unit, for example [GaS ₃ H], where H is I, Br or Cl.

In the glass structure, apart from the natural covalent bond, an additional halogen covalent dative bond (Ga-H Ga) is formed to account for the sulphur inadequacy of Ga ₂ S ₃ itself forming a tetrahedra so that an extended three dimensional covalent network is constructed. Whereas, the charge is compensated by the cation coming into the system as the halide. Such a mechanism has been shown in a

Cs:Ga:S:Cl glass.

However, this process is not possible for the conventional chalcogenide glasses. Because the 8-N rule requires that the halogen is bonded to one cation only, it functionally decomposes the structure by terminating the essentially continuous glass-forming network as evidenced, for example, in the Ge(As,Sb)-S(Se,Te)-I(Br, Cl) systems.

Thus, the halogen reduces the glass stability in the conventional chalcogenide glasses.

It is unique that halogen takes the role of intermediates in GLS glasses.

However, intermediates only stabilize the respective glasses when they begin to enter the glass-forming network. At what concentration and at what fraction do they proceed such a process is complex in theory and depends on the properties of all the constituents involved in the system.

However in practice, the thermally more stable glasses can easily be uncovered experimentally, for example, using a standard thermal analysis technique for the glass system of interest.

Secondly, in terms of improving the glass UV/Visible transmission edge, it has been recognised that the band-gap of a glass, which defines the UV/Visible transmission edge, shows an additive characteristic with regard to its constiment composition. This means the property can be represented by regarding the glass as a simple mixture of components each of which contributes independently to the overall effect.

Since the halide in general possesses a much wider band-gap and therefore transmits much further into the UV/Visible region than the sulphide, the addition of halide into a sulphide based glass is expected to enhance the UV/Visible transmission of the resulting glass, due to the characteristic additivity of band-gap upon the glass composition.

Figure 7 is a schematic graph illustrating the quantity Tx-Tg against halide concentration for chlorine, bromine and iodine containing glass; and Figure 8 is a schematic graph illustrating the decay lifetime of Pr ^{3 +} 1.3um emission against CsCl concentration in CsCl modified GLS glass. These two Figures illustrate the advantageous results that can be achieved by selecting CsCl in the composition of the glass (with respect to the results obtained for CsBr and Csl).

For use as an amplifier, a fibre formed at least in part (e.g. the core, the cladding or both) from the glass described herein is connected in an otherwise standard fibre amplifier configuration. A suitable pump wavelength for use with the dopant praseodymium is 1.1 μm (withe the praseodymium concentration being preferably from about 200 to about 2000 ppm. Suitable pump wavelengths for use with the dopant dysprosium are 0.8, 0.9, 1.1 and 1.24 μm, with the dysprosium concentration being preferably from about 200 to about 20000 ppm. This higher preferred upper limit makes dysprosium a preferred dopant, as it allows the amplifier to be a physically shorter device. In summary, therefore, embodiments of the present invention disclose the halide-containing gallium-lanthanum sulphide glasses suitable for optical waveguide device applications such as waveguide (either fibre or planar) lasers, optical amplifiers and superfluorescent sources. While reserving the essential characteristics of low phonon energy and good rare-earth solubility of the pure GLS, these modified GLS glasses in addition exhibit enhanced optical transmission further into the

UV/Visible region of the electromagnetic radiation spectrum and, some also exhibit improved glass thermal stability when compared to the related non-halide containing pure gallium- lanthanum sulphide glasses. Both are beneficial to their optical waveguide device applications, providing ideal conditions for efficient pumping of active ions and for fibre drawing (when the waveguide of concern is a fibre), respectively. When doped with active rare-earth ions, the glasses show efficient emission from levels having an energy gap, relative to the next adjacent level, of not less than 1600 cm ^"1 . In particular, when doped with Pr ³⁺ or Dy ^{3 +} , these glasses form the core of a waveguide as pump-efficient optical amplifier operating in the second telecommunication window at a wavelength close to 1.3 μm.

In embodiments of the invention the glasses can be used as the core of optical waveguides such as optical fibres, and in further embodiments such waveguides can be employed in waveguide devices such as optical amplifiers, for example for use with signals in the 1.3 μm region. In other embodiments of the invention, various different elements can be substituted for those described above. In general, the composition of the glasses can be defined as:

50 to 80 mole-percent NS _X ; 5 to 50 mole-percent MS _X ; and 1 to 40 mole-percent RH _X ; where

S is sulphur; H is a halide selected from the group consisting of Iodine, Bromine and Chlorine; N is gallium, or gallium with at least one cation selected from the group consisting of indium, aluminum, arsenic, antimony, bismuth, germanium, boron, silicon and phosphor; M is lanthanum, or lanthanum with at least one cation selected from the group consisting of the rest of the lanthanide series, yttrium, sodium, potassium, calcium, barium, zinc, cadmium, tin, lead, lithium, mercury, silver, thallium and strontium; and R is at least one cation selected from the group consisting of caesium, rubidium, potassium, sodium, barium, strontium, calcium, zinc, lead, lanthanum, lutetium, yttrium, scandium, lithium, beryllium and magnesium.

Figure 9 is a schematic diagram of an optical fibre waveguide 10 fabricated using the above glasses, comprising a glass cladding 20 surrounding a glass core 30.

One or both of the cladding and core (or parts thereof) can be fabricated using one or more of the above glasses.

Figure 10 is a schematic diagram of a planar waveguide fabricated using the above glasses, comprising a light guiding core 40 surrounded by a substrate glass 50. Again, one or both of the substrate and core (or parts thereof) can be fabricated using one or more of the above glasses.

Figure 11 is a schematic diagram of an optical amplifier. Signal light to be amplified is received at a port 60 of a coupler 70, and pump light from a pump source 80 is received at aport 90 of the coupler 70. A length of doped optical fibre 100 acts as the amplifying medium, with the amplified signal emerging from the fibre

100.

Figure 12 is a schematic diagram of an optical fibre laser. Light form a pump source 110 is focused onto a length of amplifying optical fibre 120 formed using one of the above glasses. At one end of the fibre a reflector 130 has a substantially 100% reflectivity at the lasing wavelength, but only about a 97 % transmission at the pump wavelength. At the other end of the cavity a fibre grating 140 provides a 40% reflectivity at the lasing wavelength. Laser light emerges from the end 150 of the

cavity.