WO2001014918A1

WO2001014918A1 - Improvements in and relating to fibre optic devices

Info

Publication number: WO2001014918A1
Application number: PCT/GB2000/003206
Authority: WO
Inventors: George Kakarantzas; Timothy Eugene Dimmick; Timothy Adam Birks; Philip St. John Russell
Original assignee: The University Of Bath
Priority date: 1999-08-20
Filing date: 2000-08-17
Publication date: 2001-03-01
Also published as: AU6708300A; GB9919822D0

Abstract

A method of manufacturing an optical fibre coupler, comprising the steps of providing a first optical fibre and a second optical fibre, at least one of the fibres being a single-mode fibre; reducing the local diameter of a region of each of the first and the second fibres to produce a tapered waist (80) in each fibre; placing the waists (80) of the first and second fibres in contact with each other; and fusing the waists (80) together to form a fused region (70).

Description

Improvements in and relating to fibre optic devices

This invention relates to the field of single-mode optical fibres and, m particular, to the manufacture of optical fibre couplers and to the couplers themselves.

Glass optical fibres that support at most only a small number of modes ("single-mode" optical fibres) are widely used m applications such as telecommunications and sensing. There exist many devices for performing functions on light passing through such a fibre. Examples of elementary devices include: directional couplers, for splitting light m an input fibre between two output fibres; spectral filters, to block, pass or re-route light according to its wavelength; and reflectors, to reverse the direction of propagation of a light wave. More complex devices can be made by assembling a plurality of elementary devices, in the same way that an electronic circuit can be made from a plurality of electronic components .

Such fibre devices can be made from bulk optical components, or planar integrated optical devices, by "pigtailing" , wherein light leaving an input fibre or entering an output fibre is interfaced to a functional part of the device by a system of lenses. That usually results m a device that is large, environmentally unstable, mechanically weak and expensive to make. The interface can also introduce significant optical losses; so, it is often advantageous to employ all -fibre devices, which are made from fibres that have been processed m some way. In such a device, light never leaves the fibres and so there is no need for an interface and consequently far less loss. One basic all-fibre device is the fused coupler*_ which acts as a fibre beam-splitter. Light entering the device in an input fibre is divided, in some ratio, between two or more output fibres. Variants of this device can act as filters, where the splitting ratio depends on the wavelength of the light, and switches, where the ratio can be controlled electrically. Such devices are widely used m many technological applications.

A fused coupler is typically made from a pair of optical fibres. Their polymer coatings are removed over a suitable length, exposing the bare glass fibres. The fibres are held m substantially parallel contact. They are then heated using a heat source such as a gas flame, an electrical resistance heater or a laser and the two fibres are thereby fused together. At the same time, the fibres are stretched to reduce their diameter, a process called tapering. The result is a single fused structure with a cross-sectional area that is much smaller at the narrowest point (the "waist" of the coupler) than the cross-sectioned area of the original fibres. The four lengths of the two original fibres emerge from the waist via "taper transitions" . Two act as input fibres and the other two as output fibres.

Tapering of a single fibre causes light propagating in the core of a fibre to expand to fill the entire cross- section of the fibre. Therefore m a fused coupler, in broad terms, that allows the light in one fibre to leak across into the other fibre m the region where they are fused. More strictly, the light in an input fibre excites a number of electromagnetic modes of propagation that fill the waist of the coupler. The properties of these modes determines how the light is split between the output fibres. Modes like these, which fill the cladding of the coupler, are known as_; cladding modes .

In some conventional fused couplers, one or both fibres is/are heated and stretched individually (a process known as pre-tapenng) , before being held m contact with the other fibre or fibres, then heated and stretched further. However, it is important to note that m these conventional processes, even if pre-tapenng is involved, the fibres are then heated and stretched together m a second stage of the process, the diameter of the fibres being reduced by a greater ratio m that second stage than they were reduced during pre- tapering. Also, pre-tapenng does not proceed to the stage that the light originally m the core of the fibre becomes guided as a cladding mode at any point .

In known devices, the waist of the coupler is typically between 5 and 10 mm in length, but the complete device is considerably longer because of the taper transitions, which generally contribute at least another 20 mm or so to the length of the device, both because of the constraints of the fabrication technique and also m order to keep the optical losses of the coupler to an acceptable level . Hence a typical practical coupler, complete with a rigid housing for strength and stability, occupies a cylinder of about 60 mm length and 3 mm diameter.

Consequently, each fused coupler that is added to an assembly usually adds another length of 60 mm. The assembly can be shortened by holding individual couplers in parallel and connecting them with fibre loops, but loops with a radius of curvature of less than about 10 mm tend to introduce bend losses of their own. There is, therefore, a limit to the miniaturisation that is possible with conventional fused couplers .

It is an object of the invention to provide a method of manuf cturing an optical fibre coupler that makes it possible to make a coupler with an especially small fused region. It is a further object of the invention to provide miniaturised optical devices which incorporate such couplers.

According to the invention there is provided a method of manufacturing an optical fibre coupler comprising the following steps: providing a first optical fibre and a second optical fibre, at least one of the fibres being a single-mode fibre; reducing the local diameter of a region of each of the first and the second fibre to produce a tapered waist in each fibre; placing the waists of the first and second fibres m contact with each other; and, fusing the waists together.

The term "single-mode fibre" is a well-known term of the art and refers to a fibre that supports one or at most only a few transverse modes m a wavelength range of interest; for example at least one of the fibres may support βetween one and ten modes at at least one wavelength. In contrast, a "multi-mode fibre" is a fibre that supports many (perhaps hundreds of) transverse modes.

By reducing the local diameter of the first and second fibres individually before fusing the fibres together, so that their waists are of similar cross-sectional area to that which they will have in the final coupler, it becomes possible to limit the length of the fused region (those parts of the waists that have been fused) , preferably to less than 2 mm. Indeed, the length is advantageously limited even more: to less than 1 mm or, more preferably, to less than 500 microns. More preferably, the length of the fused region is less than 200 microns and it may even be less than 100 microns. Such a short fused region m a prior art fibre coupler has not been previously demonstrated. The reduction m the local diameter of each of the fibres should be sucn that light (of whatever wavelength is to be passed through the fibre) is guided m the waist of each fibre as a cladding mode .

The invention therefore permits the manufacture of a coupler that is on a miniature scale (a "raicrocoupier"¹ ; that makes it possible to house a coupler m a very small volume and, also, to house complex devices m much smaller volumes than is customary.

Preferably, the first and second optical fibres are single-mode fibres. Miniaturisation is of particular advantage m respect of couplers formed of single mode fibres; couplers embodying the invention and described below with reference to the accompanying drawings are formed of single mode fibres.

In US patent No. 5,138,676 (Stowe et al) various optical fibre devices are described and m Fig. 13 an optical coupler is described m which it is proposed that, m order to provide coupling, two narrowed fibres, each bent around 180°, are fused together at a point of tangency and it is said that coupling would be achieved because the optical field travelling around a bent fibre of reduced cross-sectional area is distributed with most of the optical energy concentrated near the outer circumference of the bend, so that when two bends are brought into contact, evanescent coupling occurs. A problem that we believe arises in trying to follow such a proposal, however, is that the stresses m the bent fibres would obstruct joining together of the bent portions by fusing, since the resistance to those stresses would be substantially reduced during the fusing operation and would be likely to cause the spring-like fibres to straighten where not fused and deform (causing loss) or even break where fused as soon as an attempt at fusion was made. In the present invention, no reliance is placed upon having the fibres bent and the mechanism relied upon m the present invention to achieve coupling is not reliant on such a bend so that m the present invention straight fibres can be joined together to form a coupling.

We have found that it is especially advantageous for at least one, and preferably both, of the fibres to be annealed by heating prior to fusion, m order to remove any stresses m the fibres, such as those due to the way the fibres are held. For example, even clamps intended to hold a fibre straight may m fact impose some bend, twist or stretch deformations. Preferably, regions that will be m a stressed state immediately prior to fusion are annealed. Such an annealing step is not customary when manufacturing optical fibre couplers but we have found it to be especially advantageous m the present invention. The annealing step is preferaoly carried out prior to fusion of the fibres but may be carried out instead, or m addition, after fusing of the fibres. Preferably, at least the waist of at least one of the fibres is annealed by heating prior to fusion. Preferably, at least the fused portions of the fibres are annealed by heating after fusion. The annealing step preferably involves applying heat for a period of more than 2 minutes, that being a longer period than is generally used simply for fusion; the annealing step would usually be carried out by heating fibres to a temperature less than the temperature at which they fuse. During annealing, heating is not strong enough_^ to deform the bulk structure of the fibres but does allow stresses m the fibre to relax, for example by the microscopic internal rearrangement of interatomic bonds.

As already indicated, the local diameter of a region of each of the first and the second fibres should be reduced such that light is guided m the waist of each fibre as a cladding mode. In general, that means that the local diameter of a region of each of the first and second fibres is reduced to a diameter of 30 μm oi less The narrowed regions may nave a αiameter of, for example, 20 μm, 15 μm, 10 μm, 5 μm or less.

The waists may be fused together over a substantially shorter length than the lengths of the waists of the fibres.

Whilst the present invention can be used to join fibres of any reasonable curvature, it does not rely, as already explained, on the fibres having any particular curvature.

Thus, the radius of curvature of the waist of at least one of the fibres, preferably each of tne fiores, may be greater than 100 microns when the waists are fused together and may be greater than 600 microns, 1 mm, 2 mm, 1 cm or 5 cm. Similarly the first and/or second fibre (s) may be bent by less than, for example, 180°, 90 °, 45 °, or 10 °. Furthermore the waist of at least one of the fibres, preferably each of the fibres, may be substantially straight when the waists are fused together.

The waists of the fibres will usually be fused together with the waists substantially parallel to one another, but it may also be advantageous m some cases for the waists of the fibres to be fused together with the waists inclined to one another. In either case, the waists may be twisted around each other during their fusing together; for example, to help maintain contact between the fibres .

Although reference is made above to fusing first and second optical fibres, it should he understood that to make certain particular devices, it will be desired that the waists of the first and second fiores that are fused together are separate portions of the same optical fibre; m such a case it will be understood that the first optical fibre and the second optical fibre are separate portions of the same optical fibre, the portions being spaceα apart from one another longitudinally by an intermediate portion

It is also possible to manufacture optical fibre couplers m which more than two fibres are coupled together. Thus, the method may further include the steps of providing a further optical fibre, reducing the local diameter of a region of the further optical fibre to produce a tapered waist, placing the waist of the further fibre m contact with a waist of the first and/or second fibres and fusing the waists together. The further fiores may be fused to tne first or second fibre at a location spaced from the fused joining of the first and second fibres. The further fibre may be fused to the same waist of the first or second fibre as forms part of the fused coupling of the first and second fibres, or it may be fused to another waist of one of the first and second fibres.

As well as reducing the local diameter of a region of each of the first and second fibres before fusing the fibres together, the method may further comprise the step of stretching the fibres during the fusing of the waists so that the cross-sectional area of the waisted fibres is further reduced. In such a case, the further reduction would be of an amount which was small compared with the reduction prior to fusion.

Preferably the optical response of the coupler is monitored as the coupler is being manufactured. That can be achieved by launching light into one fibre and detecting the outputs at the far ends of the first and second fibres.

The reduction m diameter of at least one fibre to produce a tapered waist may be achieved by various means including neatmg and stretching, chemical etcning, for example, using hydrofluoric acid, or a combination of heating and stretching and chemical etching. Other processes which could be used include: plasma etching, ion milling, solvent processing, grinding, and polishing, or any appropriate comhination of those processes.

The fibres may be fused together using, for example, light from a carbon dioxide laser, a flame, and electrical resistance heater or an electric arc discharge.

The present invention also provides a fibre coupler comprising a first optical fibre and a second optical fibre, at least one of the fibres being a smge-mode fibre, each having an elongate narrowed region, the first and second fibres being fused together along at least one portion of their narrowed regions, the length of the fused region being less than 2 mm.

Preferred features of a fibre coupler according to the invention will be apparent from the description above relating to a method of making a fibre coupler, but some of the more significant features are nonetheless set out below: The first and second optical fibres may be single-mode fibres. The or each fused portion of the coupler may be of substantially shorter length than the lengths of the narrowed regions. The radius of curvature of the narrowed region of the first and/or second fibre may be greater than 100 microns. The first and/or second fibre may be bent by less than 180° m the narrowed region. At least one of the fibres may be substantially straight m the narrowed region. The fibres may be substantially parallel or inclined to one another m the narrowed region. At least one fibre may have been annealed. The coupler may support light m a single transverse mode at at least one wavelength. At least two portions of the same fibre may be fused together. The narrowed region may have a diameter of less than 30 microns. More than two fibres may be fused together. One fibre may have a plurality of other fibres fused to it at respective locations spaced along the fibre. A plurality of other fibres may be fused to the same narrowed portion of said one fibre at respective locations spaced along the fibre.

Many optical devices may include at least one fibre coupler according to the invention. Optical devices may include more than one fibre coupler according to the invention. Such complex devices may, for example, be formed by repeating the contact and fusion steps at additional sites within the tapered region. Little extra space is taken up by each additional coupler; because the raw material is fibre that has already been narrowed, each component microcoupler m the complex device does not bring with it the overhead of its own taper transitions. Furthermore, since the fibres adjoining the coupler are themselves narrowed, they can be bent very tightly without loss, thus permitting further miniaturisation. Thus, complex devices can be housed m a similar volume to that required for a single conventuonal component. Indeed, three-dimensional interconnections are possible. Being made entirely from fibre, there is generally no need for interfacing and so the loss is low. Hence an advantage of the coupler is that it can lead to miniature low-loss complex fibre devices.

An optical device made from the coupler might be, for example, a Sagnac mirror, a ring resonator, a Mach-Zehnder interferometer, a coupler array (comprising a plurality of fibre couplers according to the invention formed m a network of narrowed fibres) , or a four-port filter m which at least one fibre has been narrowed by chemical etching and at least one fibre grating has been inscribed m that fibre.

Embodiments of the invention will now be described in greater detail, by way of example only, with reference to the accompanying drawings, of which:

Fig. 1 shows a conventional fused coupler.

Fig. 2 shows the evolution of light through a low- loss tapered single-mode fibre.

Fig. 3 shows a microcoupler according to the invention.

Fig. 4 shows a Sagnac mirror including (a) a conventional coupler and (b) a microcoupler according to the invention.

Fig. 5 shows a ring resonator including a microcoupler according to the invention. Fig. 6 shows a Mach-Zehnder filter including a microcoupler according to the invention.

Fig. 7 shows a coupler array including a microcoupler according to the invention.

Fig. 8 shows a four-port filter including a microcoupler according to the invention.

In one embodiment of the invention (not illustrated) realised m the laboratory, two lengths of standard single-mode glass fibre were reduced m diameter from 125μm to 15μm by stretching m a travelling flame. The two narrowed fibres were laid across one another so that they were then held m substantially parallel contact over a short length. The fibres were annealed m this position by further exposure to the flame, though at a lower temperature, for 10 minutes. They were then fused together using a high power beam from a 20W carbon dioxide laser (giving a 1 mm spot size) . Whilst being fused, the fibres were further stretched by an amount which resulted m a further reduction of their cross- sectional area of less than 20%. The microcoupler was the fused region that acts as a beam-splitter. In this embodiment, the fused region was 200μm long, the excess loss of the coupler was 0.2 dB, and the splitting ratio was 100% for light of a wavelength of 1550 nm. The coupler was so short because the required interaction length varies approximately as the square of the coupler's width (F.P. Payne, CD. Hussey, M.S. Yataki , Electron. Letters 2JL (1985)461), and the fibre from which it was made was already very narrow. However, it should be appreciated that by varying the details of the fabrication process, other splitting ratios, wavelength responses, losses and/or dimensions can be obtained. It should also be appreciated that the coupler can be formed from more than two portions of fibre .

In a conventional fibre coupler (Fig. 1) , fibres 10 are brought into approximately parallel contact, stretched and simultaneously fused to produce a common waist 20. A large part of the length of the waist 20 is fused. Taper transition regions 30 result between the waist 20 and the remaining untapered fibre lengths 40 to 43. Light propagating along input fibres lengths 40, 41 exits through output fibres lengths 42, 43 m a ratio determined by the evolution of light as it propagates.

Fig. 2 shows a single tapered fibre: m the untapered regions 45, 46 the light is confined to the core region 50 by the cladding 60; m the waist 20, light spills out into the cladding 60 and is confined therein by total internal reflection at the interface between the cladding 60 and the environment surrounding the waist 20 of the fibre.

In a fibre coupler according to the invention, as shown m Fig. 3, the fused region 70 occurs along only a short length of the fibre waist 80; that is, the region where the fibre is narrow and of a substantially constant diameter.

Figs. 4 to 8 show various examples of the devices that can be made incorporating a microcoupler.

One important device that can be constructed from the microcoupler is a Sagnac mirror (Fig. 4) . Such a mirror is made by bending a fibre 90 by about 180° so that two portions of the fibre are brought into parallel contact. Conventionally, a coupler 100 is used at the point of contact (Fig. 4a) ; however, m a narrowed fibre 91 a microcoupler 110 may instead be formed at the point of contact (Fig. 4b) .,If the splitting ratio of the coupler is arranged to be 50% at the wavelength of interest (by control of the heating time, temperature and length) , the resulting structure acts as a mirror for that wavelength. If the fibre diameter is about 15μm or less, the bend can be of diameter 2 mm or less, making the device very small.

Another important device is a ring resonator (Fig. 5) . In that device, a fibre 120 is bent by about 360° so that two portions of the fibre lie m parallel contact. A microcoupler 130 is formed with a splitting ratio that is close to 100%. The resulting resonator can have a spectral response with a free spectral range of the order of 0.8 nm (the channel spacing m dense wavelength division multiplexed telecommunication systems) if the loop diameter is about 0.6 m .

A further important device is the Mach Zehnder interferometer (Fig. 6), m which two successive microcouplers 140, 141, with an unfused region 160 m between, are formed on the same pair of parallel narrowed fibres, 150, 151. By varying the path lengths of the two portions of fibre 161, 162 forming the intermediate unfused region 160, a number of useful filtering responses can be obtained.

Another important device is the coupler array (Fig. 7), in which microcouplers are formed m a network 170 of narrowed fibres, m such a way that the light split by the first coupler encountered 180 is itself split by a further coupler 190. This permits division of light from one input fibre between a number of output fibres. A further important device is a four port filter (Fig. 8) . That device can be made by forming a microcoupler where a narrowed fibre 200 has been narrowed by chemical etching, for example, m hydrofluoric acid, and a fibre grating 210 has been inscribed m that fibre. If the fibre has a photosensitive core (for example, if it is dopeo with germanium) , a Bragg grating can be written in that core; for example, by exposure to ultraviolet light. The grating can be written within the fused region 220 of the microcoupler (as shown m Fig. 8) , and/or it can be written outside it.

Devices incorporating Bragg gratings can function as compact wavelength division multiplexing filters.

It should be appreciated that a wide range of other interesting devices can be formed m similar ways.

Claims

1. A method of manufacturing an optical fibre coupler, comprising the following steps: providing a first optical fibre and a second optical fibre, at least one of the fibres being a single-mode fibre ; reducing the local diameter of a region of each of the first and the second fibres to produce a tapered waist in each fibre; placing the waists of the first and second fibres m contact with each other; and, fusing the waists together.

2. A method as claimed m claim 1, m which the length of the fused region is less than 2 mm.

3. A method as claimed m claim 1 or 2 , m which the first and second optical fibres are single-mode fibres.

4. A method as claimed in any preceding claim, m which at least the waist of at least one of the fibres is annealed by heating prior to fusion.

5. A method as claimed m any preceding claim, m which at least the fused portions of the fibres are annealed by heating after fusion.

6. A method as claimed m any preceding claim, m which the local diameter of a region of each of the first and the second fibres is reduced to a diameter of less than 30 μm.

7. A method as claimed m any preceding claim, m which the waists are fused together over a substantially shorter length than the lengths of the waists of the fibres.

8. A method as claimed m any preceding claim, in which the radius of curvature of the waist of at least one of the fibres is greater than 100 microns when the waists are fused together.

9. A method as claimed m claim 8, in which t e waist of at least one of the fibres is substantially straight when the waists are fused together.

10. A method as claimed m claim 8 or 9, m which the radius of curvature of the waist of each of the fibres is greater than 100 microns when the waists are fused together

11. A method as claimed m claim 10, in which the waists of each of the fibres is substantially straight when the waists are fused together.

12. A method as claimed m any preceding claim, m which the waists of the fibres are fused together with the waists substantially parallel to one another.

13. A method as claimed m any of claims 1 to 11, m which the waists of the fibres are fused together with the waists inclined to one another.

14. A method as claimed m any preceding claim, m which the waists are twisted around each other during their fusing together.

15. A method as claimed m any preceding claim, m which the waists of the first and second fibres that are fused together are separate portions of the same optical fibre .

16. A method as claimed m any preceding claim, further including the steps of providing a further optical fibre, reducing the local diameter of a region of the further optical fibre to produce a tapered waist, placing the waist of the further fibre m contact with a waist of the first and/or second fibres and fusing the waists together.

17. A method as claimed m claim 16, m which the further fibre is fused to the first or second fibre at a location spaced from the fused joining of the first and second fibres.

18. A method as claimed m claim 16 or 17, m which the further fibre is fused to the same waist of the first or second fibre as forms part of the fused coupling of the first and second fibres.

19. A method as claimed m any preceding claim, which further comprises the step of stretching the fibres during the fusing of the waists so that the cross- sectional area of the waisted fibres is further reduced by an amount which is a smaller proportionate reduction than the proportionate reduction prior to fusion.

20. A method as claimed in any preceding claim, m which the optical response of the device is monitored as the coupler is being manufactured.

21. A method as claimed m any preceding claim, m which the reduction m diameter of at least one fibre to produce a tapered waist is achieved by heating and stretching.

22. A method as claimed in any of claims 1 to 20, m which the reduction m the diameter of at least one fibre to produce a tapered waist is achieved by chemical etching.

23. A method as claimed m any of claims 1 to 20, m which the reduction m diameter of at least one fibre to produce a tapered waist is achieved by a combination of heating and stretching and chemical etching.

24. A method as claimed m claims 1 to 20, m which the reduction m diameter of at least one fibre to produce a tapered waist is achieved by plasma etching.

25. A method as claimed m claims 1 to 20, m which the reduction m diameter of at least one fibre to produce a tapered waist is achieved by ion milling.

26. A method as claimed m claims 1 to 20, m which the reduction m diameter of at least one fibre to produce a tapered waist is achieved by solvent processing.

27. A method as claimed m claims 1 to 20, m which the reduction m diameter of at least one fibre to produce a tapered waist is achieved by grinding.

28. A method as claimed m claims 1 to 20, m hich the reduction in diameter of at least one fibre to produce a tapered waist is achieved by polishing.

29. A method as claimed m claims 1 to 20, m which the reduction in diameter of at least one fibre is achieved by a combination of any of the processes set out m claims 21 to 28.

30. A method as claimed m any of claims 1 to 29, m which the fibres are fused together using light from a carbon dioxide laser.

31. A method as claimed m any of claims 1 to 29, in which the fibres are fused together using a flame.

32. A method as claimed m any of claims 1 to 29, m which the fibres are fused together using an electrical resistance heater.

33. A method as claimed m any of claims 1 to 29, m which the fibres are fused together using an electric arc discharge .

34. A fibre coupler comprising a first optical fibre and a second optical fibre, at least one of the fibres being a single mode fibre, each having an elongate narrowed region, the first and second fibres being fused together along at least one portion of their narrowed regions.

35. A fibre coupler as claimed m claim 34, m which the length of the fused regions is less than 2 mm.

36. A fibre coupler as claimed m claim 34 or claim 35, in which the first and second optical fibres are single mode fibres .

37. A fibre coupler as claimed m any of claims 34 to 36, m which the or each fused portion is of substantially shorter length than the lengths of the narrowed regions.

38. A fibre coupler as of claims 34 to 37, m which the radius of curvature of the narrowed region of the first and/or second fibre is greater than 100 microns.

39. A fibre coupler as claimed m any of claims 34 to 38, in which the first and/or second fibre is bent by less than 180° m the narrowed region.

40. A fibre coupler as claimed m any of claims 34 to 39, m which at least one of the fibres is substantially straight in the narrowed region.

41. A fibre coupler as claimed m any of claims 34 to 40, m which the fibres are substantially parallel m the narrowed region.

42. A fibre coupler as claimed m claim 41, m which the fibres are inclined to one another m the narrowed region.

43. A fibre coupler as claimed m any of claims 34 to 42, m which at least one fibre has been annealed.

44. A fibre coupler as claimed m any of claims 34 to 43 which supports light m a single transverse mode at at least one wavelength.

45. A fibre coupler as claimed m any of claims 34 to 44, m which at least two portions of the same fibre are fused together.

46. A fibre coupler as claimed m any of claims 34 to 45, m which the narrowed region has a diameter of less than 30 microns.

47. A fibre coupler as claimed m any of claims 34 to 46, m which more than two fibres are fused together.

48. A fibre coupler as claimed m claim 47, m which one fibre has a plurality of other fibres fused to it at respective locations spaced along the fibre.

49. A fibre coupler according to claim 48, m which a plurality of other fibres are fused to the same narrowed portion of said one fibre at respective locations spaced along the fibre.

50. An optical device including at least one fibre coupler according to any of claims 34 to 49.

51. An optical device including more than one J:ιbre coupler according to any of claims 34 to 49.

52. An optical device as claimed m claim 50, which is a Sagnac mirror.

53. An optical device as claimed m claim 50, which is a ring resonator.

54. An optical device as claimed m claim 50, which is a Mach-Zehnder interferometer.

55. An optical device as claimed in claim 50, which is a coupler array, comprising a plurality of fibre couplers according to any of claims 34 to 49 formed m a network of narrowed fibres.

56. An optical device as claimed m claim 50, which is a four-port filter, m which at least one fibre has been narrowed by chemical etching and at least one fibre grating has been inscribed m that fibre.

57. A method as claimed m any of claims 1 to 33, for making a fibre coupler as claimed m any of claims 34 to 56.