An arrayed-waveguide grating
This invention relates to the field of telecommunications and in particular to arrayed-waveguide gratings.
Arrayed-waveguide gratings (A G) are well-known devices used for signal-processing in optical telecommunications systems based on wave-division multiplexing (WDM) . Prior-art AWGs are planar devices; that is, they are essentially two- dimensional devices manufactured, in a similar way to electronic semiconductor devices, on a silicon substrate. A typical prior-art AWG (Fig. 1) comprises a silicon substrate 10, on which are provided one or more input waveguides 20, free-propagation regions 30, 40, one or more output waveguides 60 and intermediate waveguides 50. The waveguides 20, 50, 60 may be formed from silicon themselves or from silica. Because of the planar nature of the device, the waveguides 20, 50, 60 are of rectangular cross-section. Light is confined to them in use by total internal reflection at each of the four sides of the rectangle.
Free-propagation regions 30, 40 are so-called because light is not constrained in use by the structure of the regions 30, 40 to any particular area in the plane of the device. Rather, the light is able to diffract freely. However, the light is constrained by the small dimensions of the device in the direction perpendicular to the plane of the device; the light is guided even in the free-propagation regions in that axis.
Free-propagation regions 30, 40 each include a lens 70. The lens in region 30 is arranged so that the entrance ports of intermediate waveguides 60 are Fourier planes for light leaving the input waveguides 20 (i.e. the exit ports of the input waveguide 20 are situated at the focus of the lens) . Similarly, the lens in region 40 is arranged so that the
entrance ports of output waveguides 60 are Fourier planes for light leaving the intermediate waveguides 50.
The lengths of adjacent ones of intermediate waveguides 50 increase across the waveguide array. Thus, shortest waveguide 51 has a length L and adjacent waveguide 52 has a length L + ΔL . There are p+1 waveguides, each ΔL longer than its neighbour, so that the (p+l)th waveguide 53 is of length L + pΔL. As each waveguide is of a different optical length, light travelling in different ones of the waveguides 50 is subjected to a different phase shift. The operation of the AWG is as follows.
Light enters input waveguides 20 and is guided to free- propagation region 30 where it is no longer guided but diffracts freely, although it is focused by lens 70 so that it is in the far field of the input waveguide 20 exit ports when it reaches intermediate waveguides 50. The AWG is arranged such that light entering on any of the input waveguides 20 is evenly distributed across intermediate waveguides 50. However, the phase shift, and hence the delay, to which the light is subjected as it is guided in waveguides 50 varies from waveguide to waveguide, as discussed above. Thus light that propagates in the longest waveguide 53 is delayed relative to light propagating in the shortest waveguide 51 (and all of the other waveguides 50) . On leaving the intermediate waveguides 50, the light propagates through free-propagation region 40, in the same way as it did through free-propagation region 30. However, the phase differences (in this case a linear phase 'tilt') between the light travelling in each of the waveguides 50 is such that the light interferes destructively at the entrance ports to the output waveguides 60 such that different wavelength components of the light are selectively output along different ones of the output waveguides 60.
That behaviour can be understood as follows. If waveguides 50 were of equal length, the AWG would be symmetric and optically reciprocal. Consequently, if light of no matter what wavelength were to enter the AWG on a particular one of waveguides 20, it would be output from the device on the corresponding one of waveguides 60. However, waveguides 50 are not of equal length, but rather introduce a phase shift that increases linearly across waveguides 50. This breaks the reciprocity of the device, causing a wavelength-dependent tilting in the phase front at the output of the waveguide. This linear phase shift has no effect on the shape of a spot of light at those ports. However, it does shift the position of such a spot, because it changes the average spatial frequency of the spectrum. Consequently, monochromatic light entering the AWG on a particular one of waveguides 20 will exit on a different one of waveguides 60, because of that shift in spot position.
Moreover, the phase slope introduced by waveguides 50 vary with wavelength because path lengths in waveguides 50 vary with wavelength.
Thus light entering the AWG on a particular one of waveguides 20 will be split into spectral components that will exit the AWG on different ones of waveguides 60.
The AWG may thus operate as a wavelength demultiplexing device. If a light signal comprising a plurality of wavelength channels is input to one of the input waveguides 20, each wavelength channel will be output to a different one of the output waveguides 60.
Conversely, the AWG may operate as a wavelength multiplexing device, If a plurality of light signals each consisting of a single wavelength channel are each input at a different one of the 'output' waveguides 60, a single light signal consisting of multiple wavelength channels will be output on one of the 'input' waveguides 20.
AWGs can thus multiplex/demultiplex many wavelength channels simultaneously, in contrast with fibre Bragg gratings, for example, which can operate only on one wavelength at a time. It is known that AWGs can be designed to provide more complex optical signal processing. For example, an AWG can be arranged to separate or combine wavelengths in groups rather than singly or it may be arranged to add or drop channels to or from a signal. The different signal- processing functions may be provided by altering the relationships between the phase shifts provided by the intermediate waveguides 50.
Prior-art AWGs have many advantages. They are manufactured by a similar process to electronic semiconductor devices and so their manufacture can take advantage of many of the advanced techniques developed in the semiconductor industry. Manufacture can be automated to a large extent. All of the components of the AWG may be manufactured on a single chip, making possible a tight density of components and hence a very compact device. The chip may be packaged as a whole. In contrast, large numbers of devices such as thin- film filters (TFFs) and fibre Bragg gratings (FBGs) are required to perform the function of a single AWG.
Moreover, there is the possibility of incorporating further optical devices onto the chip carrying the AWG, to provide optical 'integrated circuits' analogous to electronic devices .
Finally, insertion losses for AWGs are essentially independent of the number of channels, whereas losses for TFFs and FBGs, although low per channel, increase in high- density systems because many TFFs or FBGs are required for same function as a single AWG.
However, prior-art AWGs also suffer from a number of problems .
Coupling of light into a planar waveguide is very difficult. Many telecommunications systems are based on optical-fibre technology and a fibre must be very accurately aligned for light to be coupled into a planar waveguide. A fundamental problem is that optical fibres generally have a core of circular cross-section, whereas planar waveguides have a guiding region of rectangular cross-section; significant coupling losses can result from that mis-match. Planar devices are also prone to high insertion loss resulting from the material from which they are made, particularly in the case of silicon-on-silica waveguides.
Planar waveguide devices are very sensitive to environmental fluctuations such as temperature fluctuations. Temperature changes alter the phase properties of the device and can seriously degrade its operation. In order to avoid such problems, prior art AWGs require active temperature control, which adds manufacturing and operating cost and complexity.
The problems with prior-art AWGs are sufficiently severe to hamper their widespread uptake by the telecommunications industry, which often looks instead to devices based on fibre Bragg grating or other standard fibre-based technology, which cannot offer many of the advantages of AWGs
An object of the invention is to provide an AWG that does not suffer, or suffers less, from the problems afflicting prior art devices.
According to one aspect of the invention there is provided an arrayed-waveguide grating (AWG) comprising an input waveguide, a first free-propagation region to which the input waveguide is connected, an output waveguide, a second free-propagation region to which the output waveguide is connected, and a plurality of intermediate waveguides, of different optical lengths, connecting the first and the second free-propagation regions, characterised in that the AWG comprises an optical fibre having a plurality of cores
and the plurality of cores form the plurality of intermediate waveguides .
Use of a multicore fibre to provide the intermediate waveguides of the AWG can provide a device that is as compact as a planar-waveguide device, because the cores can be arranged to be as close together as cross-talk limits permit. Moreover, use of a multicore fibre overcomes many of the problems associated with planar technology. For example, insertion losses may be reduced, because the AWG of the invention may be made using optical materials in general use in fibre optics which cause lower losses than, for example, silicon on silica waveguides. Polarisation-mode-dispersion and polarisation-dependent loss are significantly reduced by moving away from planar structures . Another advantage of the AWG of the invention is that it is less sensitive to environmental fluctuations than prior- art AWGs, because the plurality of intermediate cores are all encased within the same fibre. Changes in temperature and the like are likely to result in similar changes to all of the waveguides, which is preferable to each waveguide changing in a different way.
Moreover, many telecommunications systems are based on optical fibres and the AWG may be incorporated into fibre- based telecommunications systems more easily than is possible with prior-art AWGs. Light can be more easily coupled from a fibre into a fibre than from a fibre into a planar waveguide. One reason for that is that all relevant fibre cores may readily be formed having a circular cross-section, whereas a planar waveguide has a rectangular cross-section, which leads to lossy coupling, as discussed above. Preferably, the input waveguide is an optical fibre. Preferably, the output waveguide is an optical fibre.
As in prior-art AWGs, the lengths of the intermediate waveguides may be chosen to perform a desired signal- processing operation on an input signal. Preferably, there
are a plurality of input waveguides connected to the first free-propagation region. The lengths of the intermediate waveguides may then be such that optical signals of different wavelengths that enter on different ones of the plurality of input waveguides are output on a single output waveguide (that is, the AWG may act as a wavelength-division multiplexer) . Preferably, there are a plurality of output waveguides connected to the second free-propagation region. The lengths of the intermediate waveguides may then be such that an optical signal that enters on a single input waveguide is split into wavelength components and the wavelength components are output on different ones of the plurality of output waveguides (that is, the AWG may act as a wavelength-division demultiplexer) . Preferably, the plurality of input waveguides are a plurality of cores of a multicore optical fibre. Preferably, the plurality of output waveguides are a plurality of cores of a multicore optical fibre.
Although advantageous over coupling to a multiple- planar-waveguide device, coupling to a multicore fibre can present problems of its own. Preferably, the multicore optical fibre forming the input and/or output waveguides comprises a first, multicore portion, in which all of the cores are surrounded by the same cladding region and a second portion in which each of the cores is surrounded by its own, discrete cladding region (so that in the second portion a plurality of separate optical fibres may be distinguished) , the cores being continuous between the first and the second portions . Splitting the multicore fibre into separate fibres in that way greatly facilitates coupling into the cores, since coupling into each separate fibre is readily achievable.
Preferably, the first and/or second free-propagation region comprises a lens. More preferably, the lens is a microlens. Preferably, the first and/or second free-
propagation region comprises a length of glass-fibre having no guiding core. The AWG may thus be an integrated device comprising, for example, three multicore optical fibres separated by two microlenses and such core-less glass fibre as is necessary to ensure that light entering the free- propagation region from the input cores is in its far-field at the intermediate-waveguide cores.
The plurality of cores may be arranged in a straight line in the cross-section of the optical fibre. Such a one- dimensional (in cross-section) arrangement corresponds to the arrangement of prior art waveguides based on planar technology. However, the AWG of the invention is not restricted by planar technology. The plurality of cores may be arranged in a two-dimensional arrangement in the cross- section of the optical fibre. Preferably, the plurality of cores is arranged in a two-dimensional array in the cross- section of the optical fibre. Preferably, the array is a periodic grid.
Arrangement in such a two-dimensional array enables still further miniaturisation of the AWG in comparison with prior art devices. The AWG of the invention may be arranged so that its cores form a plurality of 'stacked' planar-type devices; that is, each row of a rectangular grid of cores may form a distinct AWG operating in a plane of the device. More complex arrangements are also envisaged, in which the lengths of the intermediate waveguides are chosen so that signal processing is carried out through interactions of cores that do not form a straight line with each other. Thus, more integrated and more complex signal processing becomes possible. Preferably, the AWG in which the intermediate waveguides are comprised is arranged to raster with wavelength light across the two-dimensional array.
At least one optical fibre comprised in the AWG, for example the fibre in which the intermediate waveguides are comprised and/or the fibre in which the input and/or the
output waveguides are comprised, may be an optical fibre that guides light in a core by total internal reflection. The fibre may then comprise a cladding region that is made substantially of one material. Alternatively, at least one fibre comprised in the AWG may be an optical fibre that guides light in a core by photonic-band-gap guidance.
For either guidance mechanism, the cladding region may include microstructure . The microstructure may be a plurality of elongate regions comprising a dielectric material different from that of matrix regions in which the elongate regions are embedded. The dielectric material may be different from the matrix regions because it is differently doped. The elongate regions may be arranged in a periodic array. The elongate regions may be holes. The different lengths of the plurality of intermediate waveguides may be different physical lengths. Alternatively, the different lengths may be different optical lengths resulting from different effective refractive indices. The different refractive indices may for example result because the cores are doped with different types or concentrations of dopants or because the cores are of different cross-sectional areas. Preferably, the different lengths result because the optical fibre in which the intermediate waveguides are comprised is bent. Bending can result in refractive index changes due to material stress and hence the different lengths may result from differences between the refractive indices of the waveguides resulting from the bending. If the intermediate waveguides form a two-dimensional pattern, the optical fibre in which they are comprised may be bent in two orthogonal planes and/or twisted to give the desired optical path length differences. Preferably the optical fibre in which the intermediate waveguides are comprised forms a helix.
According to another aspect of the invention there is provided a method of processing an optical telecommunication
signal, the method comprising: guiding, in an input waveguide, light carrying the optical signal, propagating the light from the input waveguide to an optical fibre having a plurality of cores, guiding the light in the plurality of cores, propagating the light from the plurality of cores to an output waveguide and guiding the light in the output waveguide; wherein signal processing results from interference effects manifest after propagation of the light to the output waveguide . According to another aspect of the invention there is provided a method of making an arrayed-waveguide grating (AWG) , the method comprising: providing a bundle of canes, including at least some canes that are to form a plurality of core regions; drawing the bundle into an optical fibre comprising the plurality of core regions; providing an input waveguide and an output waveguide; arranging the optical fibre, the input waveguide and the output waveguide so that free-propagation regions are provided between the input waveguide and the optical fibre and the optical fibre and the output waveguide .
Preferably, the input and/or output waveguides is an optical fibre.
Preferably, a plurality of input and/or output waveguides are provided. Preferably, the method includes the step of providing a further bundle of canes, including at least some canes that are to form a plurality of core regions; drawing the further bundle into an optical fibre in which the plurality of core regions form the plurality of input or output waveguides. Preferably, both the input and output waveguides are formed in this way.
Preferably, the method includes the step of arranging a lens in the free-propagation regions. Preferably, the lens is a microlens.
Preferably, the method includes the step of forming at least one of the free-propagation regions from a length of optical fibre not having a guiding core.
Preferably, the method includes the step of including in the bundle and/or the further bundle an optical fibre that is partially drawn into a core when the bundle is drawn into a fibre. Light may, in that case, be relatively easily coupled into the undrawn portion of the fibre.
Preferably, the method includes the step of including in the bundle and/or the further bundle canes that form microstructure in the fibre drawn from the bundle. Preferably, the bundle and/or the further bundle includes tubes that form microstructure in the form of elongate holes in the fibre drawn from the bundle. In either case, the microstructure may form a cladding region that guides light in a fibre core by total internal reflection or by photonic-band-gap guiding.
Preferably, the method includes the step of bending the fibre to give the cores in the optical fibre different optical path lengths. Preferably, the method includes the step of bending the fibre in two planes. Preferably, the method includes the step of twisting the fibre.
Preferably, the method includes the step of arranging the cores in a two-dimensional arrangement, as discussed above .
An embodiment of the invention will now be described, by way of example only, with reference to the accompanying drawings, of which:
Fig. 1 is a perspective view of a prior-art AWG based on planar waveguide technology;
Fig. 2 is a schematic view of an AWG according to the invention, including a perspective view of a fibre (right hand side of Fig. 2) and magnified plan views of two portions of the fibre (left hand side of Fig. 2) ;
Fig. 3 is a schematic cross-sectional view of the AWG of Fig. 2, taken in the plane A-A' ;
Fig. 4 is a plan view of an arrangement for coupling optical fibres into the AWG of Fig. 2. Operation of the prior-art AWG of Fig. 1 is described above .
The AWG of Fig. 2 comprises input waveguides 120, free- propagation regions 130 and 140, intermediate waveguides 150, output waveguides 160 and lenses 170; the function of each of those elements is that of corresponding elements 20, 30, 40, 50, 60 and 70 in the prior-art AWG of Fig. 1.
However, the AWG of Fig. 2 is formed using optical fibres rather than planar waveguides. Intermediate waveguides 150 are cores in an optical fibre 110 (Fig. 3) . The cores 150 are arranged in a square-grid pattern in the cross-section of the fibre 110. In this embodiment, optical fibre 110 is a photonic-crystal fibre (such fibres are also known as holey fibres or microstructured fibres) . Waveguiding cores 150 are elongate regions of solid silica. Light is guided in the cores 150 by total internal reflection from a cladding of silica containing an array of elongate holes 180. The holes lower the effective refractive index of the cladding to a value between that of air and silica (the exact value depends on the shape of the guided mode) ; hence total internal reflection can occur at the boundary between the higher-index, silica regions and the lower-index cladding regions. The guidance mechanism of the fibre is thus essentially the same as that of a standard fibre, but it is not necessary to use dopants to provide the refractive-index step between the core and the cladding. Note that although in this embodiment the array of holes 180 is periodic, that is not necessary for total-internal reflection effects to occur and it does not in this case result in photonic-band- gap guidance (although such guidance could be relied upon in other embodiments of the invention) .
Photonic-crystal fibres generally exhibit strong waveguiding and can be designed to exhibit low polarisation- mode dispersion.
Input waveguides 120 and output waveguides 160 are cores in multicore photonic-crystal fibres 190, 200, respectively. Free-propagation regions 130, 140 are lengths of glass fibre that do not contain cores. However, microlenses are embedded in the fibres in these regions 130, 140. The microlens 170 in region 130 is arranged such that the intermediate waveguides 150 are at the Fourier plane of the input waveguides 120; that is, the lens is arranged such that the far-field of the exit ports of the input waveguides 120 is at the entrance ports of the intermediate waveguides 150. The lens in region 140 is similarly arranged in respect of the intermediate waveguides and the output waveguides 160. Each of fibres 110, 190, 200 has a numerical aperture that is sufficiently large to accept all light passing through lenses 170.
Optical fibre 110 is wrapped around a cylinder 155 to form a helix, such that it is bent in two planes (Fig. 2) so that stresses are induced in the waveguides 150, the stresses being different in different ones of the waveguides 150, so that the optical path lengths of the waveguides 150 are different. The degree and direction of bending is chosen to give a selected phase relationship between light travelling in the different waveguides, so that the AWG has a selected optical signal processing function. The phase differences are readily trimmed by slight changes in the bending and/or twisting. Optical path lengths in the cores of a bent multicore fibre may readily be calculated. Consider a multicore fibre bent in one plane to a radius of curvature R. Cores on the inside of the bend will have a shorter optical path length than those on the outside. Radius R is measured to the arc at the centre of the fibre where the length Lf does not change.
The change ΔI, due to the bend, in optical path length of a core positioned at a radial distance x from that arc is given by
A/ = .πn(l - χ)xlf λR where n is the refractive index, λ is wavelength and χ is related to a change in polarisability due to the bending (and equals 0.22 for silica).
The bending of fibre 110 in the horizontal plane of the cylinder about which it is wrapped is relatively fast, whereas the bending in the longitudinal direction is relatively slow. The bending causes a change in optical path length between different cores, which causes a phase shift, which in turn causes different wavelengths to be output to different ones of cores 160. The core to which a wavelength travels is selected by a raster-like effect. As the phase shift caused by bending in the horizontal plane of the cylinder changes with wavelengths, the output spot corresponding to the input channel of an input signal is displaced further and further horizontally across the face of fibre 200. Eventually, the phase shift exceeds 2π and so the spot returns to its original position and begins to sweep across again. However, the bending of fibre 110 in the vertical direction causes that second row of spots to be displaced vertically relative to the first row.
Thus the full array of cores 160 is utilised; different wavelength channels of a signal input on one of cores 120 are each output on a different one of cores 160.
An advantage of this embodiment of the invention is that standard optical fibres can readily be coupled into it. For example, multicore photonic-crystal fibre 190 is drawn from a
bundle 210 of rods 220 and tubes (which form holes 180) . The bundle 210 is drawn into fibre 190 on a standard fibre- drawing tower in a manner well known in the art (Fig. 4 shows the bundle 210 partially drawn) . Prior to drawing, standard optical fibres 230 are incorporated in the bundle 210, in place of certain of the rods, at the sites in the bundle 210 corresponding to cores 120 in the drawn fibre 190. Thus, when the fibre 190 is drawn, the material of the fibres 230 is incorporated in the core regions 120. One end of the bundle 210 remains undrawn, so that part of the length of fibres 230 remains unchanged. The transition from the undrawn parts of the fibres 230 to the cores 120 in the drawn fibre is sufficiently long such that light can propagate along the fibres 230 and into the cores 120 substantially without loss. Undrawn parts of the fibres 230 remain protruding from the bundle 210 and light can readily be coupled into them using standard techniques such as splicing. The phase-shifts provided by cores 150 of fibre 110 are trimmed, after the fibre has been drawn, to provide desired magnitude. Permanent trimming is carried out by heat-treating the fibre to partially collapse some of holes 180 (Expansion of holes would be achievable if the holes 180 were to be pressurised during heating) .