GB2406918A

GB2406918A - Arrayed waveguide grating

Info

Publication number: GB2406918A
Application number: GB0323674A
Authority: GB
Inventors: Majd Zoorob; Martin Charlton; Greg Parker
Original assignee: Mesophotonics Ltd
Current assignee: Mesophotonics Ltd
Priority date: 2003-10-09
Filing date: 2003-10-09
Publication date: 2005-04-13
Anticipated expiration: 2023-10-09
Also published as: GB2406918B; GB0323674D0

Abstract

The present invention relates to arrayed waveguide gratings (AWGs). The present invention provides AWGs with reduced physical dimensions. This is achieved by introducing different materials (23) into the waveguides (22) of the AWG in order to provide an optical path difference between each of the waveguides. The overall length and shape of each of the waveguides can be chosen according to design requirements whilst the optical path length difference between each of the waveguides can be achieved by having a different length of material (23) having a particular effective refractive index or a different material in each of the waveguides (22).

Description

240691 8

ARRAYED WAVEGUIDE GRATING

Field of the Invention

The present invention relates to arrayed waveguide gratings (AWGs). In particular, the invention concerns AWGs of reduced physical dimensions.

Background to the Invention

AWGs are well known components in optical communications networks and integrated optical circuits. An AWG is a planar structure comprising an array of waveguides arranged side-by-side which together act like a diffraction grating. AWGs can be used as multiplexers or demultiplexers and a single design can be suitable for both.

An AWG is typically composed of inpuVoutputwaveguides, two focussing slab regions and a phased array of channel waveguides therebetween, with a constant path-length difference AL between adjacent waveguides, as shown in Figure 1.

In a demultiplexing regime, the first slab is designed to radiate the input power of the signal evenly in the slab region and hence allow it to be collected by the array of waveguides. To improve uniformity of coupling power into the array of waveguides additional peripheral input waveguides are used.

2 0 The number of waveguides in the array between the slab regions determines the spatial resolution of the Fourier transformed field at the focal position. This determines the level of crosstalk between the channels.

After travelling through the array of waveguides the light beams constructively interfere at one focal point in the second slab region. The location of this focal point is dependent on the signal wavelength, since the relative phase delay in each waveguide is given by QUA. The second slab region is therefore used to focus the entire field from the arrayed waveguide section onto a specific output channel.

The waveguides and slabs are typically formed using standard photolithographic techniques as a "core" layer on a silicon substrate. Cladding and buffer layers are also typically provided, the cladding provided by, for example, a Chemical Vapour Deposition (CVD) process.

In order to provide the path length difference in the array of waveguides they are curved. The focussing slabs are also aligned at an angle to each other to improve coupling into and out of the waveguides, as shown in Figure 1.

A typical 128 channel glass AWG operating at the 1.55,um region having a 0.2nm channel spacing (25GHz) requires a path length difference of AL=63, um between neighbouring arrayed waveguides. To provide a channel crosstalk of less than -20dB, 388 waveguides are required. The length of the input and output focussing slabs are 36.3mm each.

The overall size of the AWG device depends on two main factors: the minimum bend radius of the array, and the refractive index difference between the core and the cladding material, for minimal loss. If the bends are too tight or the refractive index contrast too small, unacceptable losses will result. However, the length of the outer waveguide in the array can be predicted to be of a minimum length of AL x the number of waveguides i.e. 63,um x388 = 24.5mm. This waveguide is usually longer due to the introduction of the lengths of the bends into and out of the focussing slabs.

As can be seen, AWGs are relatively large devices and this has been one of the main issues with AWG design, together with losses arising from the waveguide bends. They are conflicting problems as, increasing the bend radii decreases the losses of the AWG, but also increases the overall size of the device further. Larger bend radii also means that the material from which the waveguides are formed is required to be uniform over larger areas, which is difficult to achieve.

Furthermore, the size and shape of AWG structures impacts on manufacturing 2 o and in particular the shape affects the number of AWGs that can be packed onto a silicon substrate. This has a negative impact on the cost of AWGs as there is a great deal of wastage in production.

The present invention aims to provide a more compact and easily producible AWG device.

Summary of the Invention

According to a first aspect of the present invention, an arrayed waveguide grating (AWG) structure comprises: an input waveguide; an input focussing slab region coupled to the input waveguide; an output focussing slab region; a phased 3 0 array of at least three channel waveguides coupled between the input focussing slab region and the output focussing slab region; and an output waveguide coupled to the output focussing slab region, wherein each of the channel waveguides includes a region having a first effective refractive index and a region having a second effective refractive index, and wherein the optical path length of the region having a second effective refractive index is different in each of the channel waveguides.

The introduction into the waveguides of regions having a second effective refractive index provides an additional or alternative means to provide an optical path length difference between waveguides to waveguide bends. It allows the shape and the size of the waveguides to be chosen whilst ensuring a particular optical path length difference between the waveguides. Preferably, the channel waveguides are linear. Preferably, the second effective refractive index is higher than the first effective refractive index.

In order to give rise to a different optical path length in each waveguide, the regions of second effective refractive index can be of different absolute length in each waveguide and/or can have different material properties in each waveguide.

Preferably, each channel waveguide comprises a waveguide core and the region of a first effective refractive index includes a core formed from a first material and the region having the second effective refractive index includes a core ore portion of the core formed from a second material. Two alternative arrangements can be set up. The second material can have a lower or higher refractive index compared to the first material. To reduce the divergence losses from the two interfaces (first material/second material and second material/first material) and maximise the mode 2 o overlap between the two regions, a higher index for the second material is preferred.

In order to minimise losses associated with divergence and scattering from the interfaces the depth of the second material can be altered. If the second material has a higher index than the first material then it is desirable that the second material is of lesser depth than the first material.

2 5 Waveguides typically include a core layer and a buffer layer and may have a cladding layer as well. The region having a second effective refractive index can be formed by including a different material in the core layer, the cladding layer, the buffer layer or any combination of the three.

The mode profile of an optical signal is affected by the effective refractive 3 0 index that it experiences. Preferably, the cross-section of the region of the core having the second effective refractive index is formed such that the mode profile of an optical signal in that region is matched to the mode profile of the optical signal in the region having the first effective refractive index. Matching the mode profile minimises scattering and reflections at the interfaces between the two regions. If the second effective refractive index is higher than the first effective refractive index then the cross-section of the region of second effective refractive index should be smallerthan that of the region having a first effective refractive index.

Each of the regions having a second effective refractive index in the array of waveguides may be formed, at least in part, by a photonic crystal extending longitudinally along each of the waveguides. The photonic cystal sections may be 1 dimensional,2-dimensional or 3- dimensional photonic crystals. The photonic crystal sections can provide a delay to optical signals passing through them, with each waveguide providing a different delay. Detail on how photonic crystals can be used to provide a time delay to optical signals can be found in W098/53351 and PCT/GB03/02143. The photonic crystal can be formed by patterning the waveguide by providing a plurality of sub-regions of a different material in a core layer, a buffer layer, or a cladding layer of the waveguide, or any combination of the three. Detail of the properties of photonic crystal structures in the various layers of waveguides can be found in co- pending US patent application no. 10/421949.

Alternatively, each of the regions having a second effective refractive index may include two or more regions of a different effective refractive index to the first effective refractive index. The two or more regions may have the same or different effective refractive index as each other.

Each channel waveguide may alternatively include two core layers and an intermediate cladding layer, wherein a photonic crystal section is formed in the intermediate cladding layer. Detail of this kind of waveguiding structure can be found in co-pending US patent application no.10/465559.

Preferably, the photonic crystal section includes defects. Defects can include missing sub-regions, different size or shaped sub-regions or differently filled sub regions. The introduction of arrays of defects into the photonic crystal can lead to a greater bandwidth of operation and simultaneous filtering or cross talk improvement.

Similarly, grading the pitch of the regions forming the photonic crystal section index or grading the refractive index or size or filling fraction of those regions, can broaden the bandwidth of operation and can provide an engineered, linear time delay across all of the waveguides for a broad bandwidth of wavelengths.

Materials with tunable refractive indices such as electro-optic or magneto-optic materials may be used in the region of second effective refractive index. This provides tunability for the AWG by altering the second effective refractive index and hence the optical path length in each waveguide.

Each of the array of waveguides may include a Mach-Zehnder interferometer.

This provides tunablity for the AWG and could form a switch, a tunable lens, a signal s modulator or tunable channel shapes.

According to a second aspect of the present invention, a method of fabricating an Arrayed Waveguide Grating (AWG) comprises the steps of: forming an input waveguide; forming an input focussing slab coupled to the input waveguide; forming an output focussing slab; forming an array of at least three channel waveguides between the input focussing slab and the output focussing slab; and forming a plurality of output waveguides coupled to the output focussing slab, wherein each of the channel waveguides includes a region having a first effective refractive index and region having a second effective refractive index, and the optical path length of the region having the second effective refractive index is different in each of the channel waveguides.

Brief description of the Drawings

Examples of the present invention will now be described in detail with reference to the accompanying drawings, in which: Figure 1 shows a conventional Arrayed Waveguide Grating (AWG); Figure 2 is a generic schematic illustration of an AWG in accordance with the present invention; Figure 3 shows a waveguide in an AWG in accordance with the present invention in cross section; Figure 4 shows an AWG according to a first embodiment of the present invention; Figure 5 illustrates sections of the waveguides having different cross sectional 3 0 area in dependence on their refractive index for optimal mode matching; Figure 6 shows an AWG in accordance with the present invention including materials with tunable refractive index; Figure 7 shows an AWG according to a second embodiment of the present invention; Figure 8 shows detail of the waveguides of Figure 7; Figure 9 shows an AWG according to a third embodiment of the present invention; Figures 10a-10c show detail of possible waveguide configurations; Figures 11a-11c show waveguides for an AWG according to a fourth embodiment of the present invention; Figures 12-13 illustrate the introduction of defects into photonic crystals in the waveguides.

Figure 14-15 illustrate graded structures in the photonic crystal sections; and, Figure 16 shows the use of liquid crystal in the photonic crystal sections.

Detailed Description

Figure 1 shows a conventional arrayed waveguide grating (AWG). The AWG comprises an input waveguide 10, an input focussing slab 11, an array of channel waveguides 12, an output focussing slab 13 and a plurality of output waveguides 14.

In order to improve the uniformity of the coupling power into the array of waveguides 12 from the input focussing slab 11 additional peripheral input waveguides may be used. The mode coupling from the excited input waveguide to the neighbouring dummy waveguides guarantees that the coupling conditions for the external 2 o waveguides is uniform.

The input light is radiated into the input focussing slab which excites the array of waveguides. The number of arrayed waveguides defines the spatial resolution of the Fourier transformed field at the focal position. This determines the higher spatial components of the Gaussian or side lobes, which in turn determines the cross talk 2 5 level between the channels.

The arrayed waveguides have an effective path length difference between adjacent waveguides of AL. Typically, AL is a constant across the array but may be graded or vary according to some other predetermined function in order to engineer the output waveform.

3 0 Aftertravelling through the arrayed waveguides, the light beams constructively interfere into one focal point in the output focussing slab. The location of this focal point depends on the signal wavelength, since the relative phase delay in each waveguide is given by ALIA.

The second slab region is used to focus the entire field from the arrayed waveguide section onto a specific output channel which is achieved by the summation of the Fourier transforms of the individual fields in every arrayed waveguide.

The path length difference (I\L) of adjacent waveguides in the arrayed waveguide section of the AWG is defined as m)O aL= no where I0 is the central wavelength of operation, no is the effective index of the array waveguide and m is the diffraction order of the AWG.

The waveguide spacing for the input or output side (if the AWG is assumed to be symmetrical and the waveguide separation is constant) of the AWG is defined as n demo n dDn AL = N f or alternatively aL = N f ' where An iS the wavelength spacing between adjacent channels.

'to is the central wavelength of operation.

nS effective index of the array waveguides.

d arrayed waveguide separation at output of first slab.

D waveguide separation at the input of first slab.

NC group index of the array waveguides (Nc=nc - ldnc/dI).

2 o f radius of curvature of the focal slabs.

AL path length difference between adjacent arrayed waveguides.

Additionally, the number of available wavelength channels NCh is given by, 2 5 Oh n,dD To facilitate the introduction of the path difference in the arrayed waveguides, the focussing slabs are aligned at an angle to each other. This provides easier coupling into the array waveguide bends which then steer the light back to the output focussing slab.

As will be appreciated, crucial to the operation of an AWG is a path length difference between each of the waveguides in the array of waveguides between the input and output focussing slabs. This path length difference is achieved by curving the array of waveguides, with each waveguide in the array having a different radius of curvature. However, the curvature of the array of waveguides leads to a number of problems, not least of which is the size of the arrayed waveguide grating. If the waveguides in the array of waveguides could be made straighter then the size of the arrayed waveguide grating could be substantially reduced, as could the losses associated with the curved waveguides. Furthermore, if the AWG could be linearised, many more of them could be packed into the same region on a silicon chip. This would have a great impact on the cost of the device, as at least twice as many devices could be packed in the same wafer space as for a conventional design. Additionally, a linear AWG would be far simpler to define using e-beam lithography and is easier to design.

In present invention, the path length difference between adjacentwaveguides in the AWG is achieved by altering the optical path length rather than the absolute path length of each waveguide. Figure 2 shows schematically an arrayed waveguide grating (AWG) in accordance with the present invention. As with a conventional design, the device comprises an input waveguide or waveguides 20, an input focussing slab region 21, an array of waveguides coupled to the input focussing slab region 22, an output focussing slab region 24 and a plurality of output waveguides 25.

However, included in the array of waveguides 22 in Figure 2 is a section 23 which is comprised of a region having a different effective refractive index to the rest of the waveguide. The section 23 may be of different length in each waveguide or may have different material properties in each waveguide. The section 23 in each of the waveguides allows the optical path length of each waveguide to be chosen, whilst keeping the waveguide straight or giving it a particular desired configuration.

The AWG shown in Figure 2 may be formed on a silicon substrate. The waveguides 22 include a buffer layer 31 formed on the substrate 30, a core layer 32 formed on the buffer layer, and a cladding layer 33 formed on the core layer. This is shown in cross section Figure 3. The core layer is typically formed of silicon oxynitride or silicon nitride and the buffer and cladding layers formed of silicon dioxide.

Figure 4 shows a first embodiment of an arrayed waveguide grating according to the present invention. The device includes a input waveguide or plurality of input waveguides 40 and an input focussing slab region 41. Coupled to the input focussing slab region are an array of channel waveguides 42 which in turn are coupled to an output focussing slab 44 and a plurality of output waveguides 45. Each of the waveguides 42 includes a section 43 of material having a refractive index which is different to the refractive index of the core layer in the rest of the waveguides 42. In this example, the section of material extends through the cladding layer and into the core layer. The sections of material could exist in the buffer layer, the core layer or the cladding layer, or any combination of the three, as long as the effective refractive index experienced by an optical signal propagating through the waveguide is different in the region corresponding to section 43 to that experienced in the rest of the waveguide. In this embodiment each of the sections 43 are made of the same material but the length of each section is different in each waveguide. However, different material could be used in different waveguides to provide a relative phase shift between channels. As shown in Figure 4, the difference in the length of each section 43 in adjacent waveguides is AL. This introduces an optical path length difference between adjacent waveguides of AL (n2-n,) where n2 is the effective refractive index of section 43 and n, is the effective refractive index of the rest of the 2 0 waveguide. This is equivalent to a path length difference between adjacent waveguides of NIL (n2-n,)/n, in a conventional AWG with curved waveguides.

From a fabrication point of view, the simplest arrangement for the sections 43, is to cut slots corresponding to the sections in the waveguides and to fill the slots with a suitable material. An AWG of this type can be fabricated in two stages. Initially, the 2 5 AWG is fabricated as a conventional AWG, e.g. using flame hydrolysis, Silicon-on lnsulators etc. Detail of AWG fabrication techniques can be found in Fundamentals of optical waveguides, Chapter 9, Katsunari Okamoto, Academic press 2000. An additional patterning stage is required which defines the slots, possibly using for example e-beam lithography. These slots are then etched to the desired depth and filled with the desired substance.

Two alternative arrangements can be set up. A lower or higher index material compared to the rest of the waveguide can be used for the sections 43. To reduce the divergence losses from the two interfaces, i.e. the waveguide/section and section/waveguide interface and in order to maximise the Gaussian overlap between the two waveguiding structures, a high index material is preferred for the sections.

Additionally, this eliminates the out of plane losses associated with leakage into the buffer layer which might otherwise be of a higher index than the sections.

The etch depth is optimised depending on the waveguide material and geometry and additionally the filling material. To provide minimal loss due to divergence and scattering from the interface between the waveguides and the sections, and additionally out of the sections into thewaveguide, and optical matching, it is desirable that the etch depth is not deeper than the core for a high index material section, while it is desirable that the etch depth is more than the core depth for a lower index material section.

The higher index material could be silicon nitride or silicon for fused silica AWGs, while a lower index material could be silicon dioxide or silicon nitride for silicon on insulator (SOI) AWG devices.

In order to minimise the mix-match between the waveguide section and the sections and hence reduce the coupling losses between the two regions, the sections can be designed so that the field profile in the region of the section is matched to the field profile in the region of the waveguide. In this arrangement the effective index in the section remains different to that of the waveguide but the field profile is matched as closely as possible. Figure 5 shows the two possible scenarios. Figure 5a shows high index material sections 50 which would have smaller dimensions than the waveguides 51 in order to give rise to a beder matched mode profile in both the waveguide and the section. Figure 5B shows the converse, in which the sections 52 are of a material having a lower refractive index that the waveguides 53, leading to the cross-section of the sections being made larger than that of the waveguides.

2 5 Additionally, tunable or non-linear materials such as liquid crystal can be used to fill the slots to form the sections. This is shown in Figure 6. This can provide great benefits in eliminating fabrication tolerances. Figure 6 shows an AWG with input waveguides 60, input focussing slab 61, an array of channel waveguides 62, an output focussing slab 63 and output waveguides 64. Each waveguide 62 includes a slot 65 3 o cut into it. Each slot 65 is of a different length. The device also includes a tunable voltage source 67 which is applied across the slots 65 via electrodes 66. The refractive index of the material filling the slot, in this case liquid crystal, can be varied by varying the applied voltage to provide a different effective path length giving rise to a greater degree of tunability in the selection of the desired wavelengths for the same number of channels. The effective path length difference between each channel AL can be tuned according to the electrical signal applied to each of the sections.

Figure 7 shows a second embodiment of the present invention, in which photonic crystal sections are used in each of the waveguides to introduce an optical delay between each waveguide. The photonic crystals in effect introduce a time delay to the optical signals passing through them. This is due to the fact that photonic crystals can provide a large change in the group velocity in an optical signal. Detail on the mechanisms and design considerations associated with using photonic crystals to provide optical delay can be found in W098/53351 and co-pending US patent application no. 10/438316. A path length AL can be converted into a time delay at using the expression: ilt= ncoreAL c Figure 7 shows an array of input waveguides 70, an input focussing slab 71, an array of channel waveguides 72, an output focussing slab 73 and an array of output waveguides 74. Each of the channel waveguides includes a number of sub regions of a different effective refractive index to the rest of the waveguide. As shown in Figure 7, each of the waveguides has a different number of the sub-regions, which 2 o are formed by placing rods 75 of material across the waveguides 72. This is shown in more detail in Figure 8. The rods 75 form, in effect, a 1-dimensional photonic crystal along the length of each waveguide 72 and the photonic crystals give rise to a photonic band structure. The photonic band structure includes a number of allowable modes which are highly dispersive. The group velocity of the optical signals 2 5 in these modes is dependent upon the gradient of the mode on a band diagram. In a photonic crystal having bandgaps using a mode operating close to the band edge allows large reductions in group velocity and hence large time delays to be achieved.

However, operating close to the band edge does result in greater losses and so there is a trade off between the maximum allowable loss and the achievable delay. With larger losses tolerable, longer time delays are possible and hence a smaller photonic crystal section can be used. Additionally, the bit length of the optical signal scales linearly with the time delay for a given loss. Therefore, for a 40 Gb/s system with a maximum of 5 dB loss a photonic crystal of four times longer will be required than for a 10 Gb/s system with a 5 dB loss.

Nonetheless, the length of the photonic crystal required in orderto produce the required time delays is fairly small. In fact the length of photonic crystal is orders of magnitude less than the AL achieved. This allows the arrayed waveguides to be made even shorter than in the first embodiment. For example, for a 100 Ghz channel spacing a desired AL of 63, um can be achieved with less than 5 dB loss in a length of photonic crystal of 1.5,um.

Figure 9 shows another embodiment of the present invention in which a photonic crystal is used to provide time delay in each of the channel waveguides. The AWG again comprises an input waveguide orwaveguides 90, an input focussing slab 91, an array of channel waveguides 92, an output focussing slab 93 and a plurality of output waveguides 94. The photonic crystal sections are formed from a monorail of rods along each waveguide. The photonic crystal can be composed of rods of a different material in each of the waveguide cores.

Figure 10 shows arrangements whereby the photonic crystal section can alternatively be formed only in the cladding layer rather than in the core layer. Figure 10a shows an array of slots 101, as shown in Figures 7 and 8 but formed only in a cladding layer above the waveguide cores 100. Figure 1 Ob shows an array of rods 102 2 o as shown in Figure 9 but formed only in a cladding layer above the waveguide cores 100. Figure 10c shows either Figure 10a or Figure 10b in cross section. The waveguides comprise buffer layer 103, core layer 104 and cladding layer 105. The photonic crystal slots or rods 106 are also shown. This type of arrangement provides for easy fabrication and reduces out-of- plane losses. However, arrangements of this sort may require double the amount of slots or rods than if the slots or rods are provided through the core. In order to reduce the number of slots or rods required it is possible to push the mode profile up into the cladding layer so as to increase the interaction between the mode and the photonic crystal section. This technique is described in more detail in co-pending US patent application no.10/421949.

Figures 11 a to 11 c show a further embodiment of the invention, similar to that shown in Figure 10, but the embodiment in Figure 11 shows twocores fommed on either side of a photonic crystal section. Details of this kind of waveguiding structure are provided in more detail in co- pending application no. 10/465,559. Figure 11a shows an array of twin waveguides 110, each pair of waveguides separated by a buffer layer including a number of slots 111 forming a photonic crystal section. Optical signals propagating through the waveguides interact with the photonic crystal. Figure 11 b shows a similar arrangement to Figure 11 a but with an array of rods 112 located in the buffer layer between waveguides 110. The same principles apply. Figure 11 c shows Figure 11a or Figure 11b in cross section. The rods or slots 112,111 are located in a buffer layer 114 between the core layers 110. The whole structure is supported on another buffer layer 113.

The introduction of defects into the photonic crystals can provide a larger bandwidth of operation. Figure 12 shows an example of a number of defects 121 introduced into the 2-dimensional photonic crystal 120 formed in the waveguides. The AWG includes an input waveguide or waveguides 122, an input focussing slab 123, an array of channel waveguides 124, an output focussing slab 125 and a plurality of output waveguides 126. The photonic crystal 120 spans the array of waveguides 124.

In order to give rise to a large change in group velocity the photonic crystal is designed so that a band edge occurs close to the frequency of operation. This is shown for a photonic crystal without defects in Figure 13a. As can be seen, the photonic crystal as a delaying structure has a relatively narrow band of operation, and it is desirable to increase the bandwidth for many AWG applications. The introduction of defects into the photonic crystal can lead to an increased bandwidth of operation by their 2 o effect on the photonic band structure. This is shown in Figure 13b. The three sets of defects 121 shown in Figure 12 lead to three passbands in the band gap. Defects can take the form of different size, shape or filled regions or missing regions in the photonic crystal. Altematively, defect states may be introduced by the superposition of two different photonic crystal structures to form a Moire pattern.

2 5 Another way to increase the bandwidth of operation is to use graded photonic crystal structures as shown in Figure 14. Figure 14 shows an array of channel waveguides 140 for an AWG with a graded pitch for the sub-regions 141 forming the photonic crystal. Also shown in Figure 14 is the corresponding transmission characteristic, from which it can be seen that the band edge is spread over a larger range of frequencies as a result of the graded structure.

A similar effect can be achieved by grading the size of the regions forming the photonic crystal, as shown in Figure 15, or by grading the refractive index of the material filling the regions along each waveguide. Figure 15 shows an array of waveguides 150 for an AWG with a photonic crystal section in each waveguide formed by a plurality of rods 151, where the size of the rods is graded along each waveguide. The corresponding transmission diagram shows that the band edge is spread over a larger range of frequencies than for a regular photonic crystal.

The operating frequency of an AWG incorporating photonic crystal sections in the channel waveguides can be made tunable by the introduction of electro-optic or magneto-optic material into the regions forming the photonic crystal, and applying a control signal which controls the refractive index of the material. Figure 16 is a schematic illustration of such an arrangement, and shows an arrangement similar to that shown in Figures 7 and 8. However, it should be clear that the tunability concept can be applied to any of the photonic crystal arrangements described herein, including photonic crystals with defects, graded structures and photonic crystals not in the waveguide core. Three waveguides 160 are shown 1,2,3, each of which are intersected by slots 161 forming a photonic crystal section in each waveguide 160.

The slots 161 are filled with a liquid crystal which has a refractive index which alters in the presence of an electric potential. Each waveguide is positioned between a pair of electrodes 162 which are connected to a voltage source 163. A change in applied voltage gives rise to a change in refractive index of the liquid crystal and hence a change in the photonic band structure of the photonic crystal sections and in particular the band edge frequencies.

2 o If the effective refractive index experienced by the modes guided by the array is designed to reside in the tunable range of the refractive index of the liquid crystal or other tunable material, the structure can be designed to flip from the slots having a higher refractive index than the waveguide to having a lower refractive index than the waveguide. This provides great operational flexibility as shown in Figure 16.

2 5 Some of the waveguides can be designed to operate in a positive dispersion regime, e.g. waveguide 3 as shown, while others operate in a negative dispersion regime, e.g. waveguides 1 and 2 as shown. In such an arrangement the bandwidth of each channel can be shaped as desired and the signal can be automatically dispersion compensated as well. Each waveguide will ordinarily introduce a different amount of dispersion owing to their different lengths but the dispersion of each waveguide can be tuned using the tunable material to cancel the ordinary waveguide dispersion or to ensure that each waveguide gives rise to the same dispersion.

Other material choices can also be advantageous. For example, as explained above photonic crystal structures inevitably lead to some loss, so that the use of active material, such as erbium doped glass, for the waveguides can provide a means to amplify the signal while it propagates along the waveguide and hence provide lossless devices.

Claims

Claims 1. An arrayed waveguide grating structure comprising an input

waveguide; an input focussing slab region; an output focussing slab region; an array of at least three channel waveguides coupled between the input focussing slab region and the output focussing slab region; and an output waveguide coupled to the output focussing slab region; wherein each of the channel waveguides includes a region having a first effective refractive index and a region having a second effective refractive index; and, wherein the optical path length of the region having a second effective refractive index is different in each of the channel waveguides.
2. An arrayed waveguide grating structure according to claim 1, wherein the channel waveguides are linear.
3. An arrayed waveguide grating structure according to claim 1 or 2, wherein the second effective refractive index is higher than the first effective refractive index.
4. An arrayed waveguide grating structure according to any one of the preceding 2 0 claims, wherein each channel waveguide comprises a waveguide core and the region of a first effective refractive index includes a core formed from a first material and the region having the second effective refractive index includes a core or a portion of the core formed from a second material.
5. An arrayed waveguide grating structure according to claim 4, wherein the second material has a higher refractive index than the first material.
6. An arrayed waveguide grating structure according to any preceding claim, wherein the cross-section of the region of the core having a second effective refractive index is formed such that the mode profile of an optical signal in that region is matched to the mode profile of the optical signal in the region having the first effective refractive index.
7. An arrayed waveguide grating structure according to any one of claims 1 to 3, wherein each of the multiple channel waveguides includes a core layer and a cladding layer and wherein the region having a second effective refractive index is formed by including a different material in the cladding layer.
8. An arrayed waveguide grating structure according to any preceding claim, wherein the region having the second effective refractive index includes a photonic crystal section in the waveguide.
9. An arrayed waveguide grating structure according to claim 8, wherein the channel waveguides include a core layer formed of a first material, wherein the photonic crystal section is formed from an array of rods or slots of a second material in the waveguide core, the second material having a different refractive index to the first material.
10. An arrayed waveguide grating structure according to claim 8, wherein each channel waveguide has a core layer and a cladding layer and wherein the photonic crystal section is formed in the cladding layer.

2 0
11. An arrayed waveguide grating structure according to claim 8, wherein each multiple channel waveguide includes two core layers and an intermediate cladding layer, and wherein the photonic crystal section is formed in the intermediate cladding layer.

2 5
12. An arrayed waveguide grating structure according to any one of claims 8 to 11, wherein the photonic crystal section includes defects.
13. An arrayed waveguide grating structure according to any one of claims 8 to 12, wherein the photonic crystal section is graded.
14. An arrayed waveguide grating structure according to any preceding claim, wherein the region of second effective refractive index includes an electro-optic or magneto-optic material and further includes means to tune the refractive index of the electro-optic or magneto-optic material.
15. An arrayed waveguide grating according to any preceding claim, wherein each of the channel waveguides includes an optically active material.

s
16. An arrayed waveguide grating according to any preceding claim, wherein each waveguide includes a Mach-Zehnder interferometer.
17. A method of fabricating an Arrayed Waveguide Grating comprising the steps of: forming an input waveguide; forming an input focussing slab coupled to the input waveguide; forming an output focussing slab; forming an array of at least three channel waveguides between the input focussing slab and the output focussing slab; and forming a plurality of output waveguides coupled to the output focussing slab, wherein each of the channel waveguides includes a region having a first effective refractive index and region having a second effective refractive index, and the optical path length of the region having the second effective refractive index is different in each of the channel waveguides.