WO2004083922A1

WO2004083922A1 - Optical muliplexer based on arrayed waveguide grating

Info

Publication number: WO2004083922A1
Application number: PCT/DK2004/000184
Authority: WO
Inventors: Lei Wang
Original assignee: Ignis Technologies As
Priority date: 2003-03-21
Filing date: 2004-03-18
Publication date: 2004-09-30

Abstract

The object of the present invention is to provide an optical component suitable for multiplexing or de-multiplexing optical signals within a relatively broad spectral range of wavelengths and with relatively wide channel spacing. The physical layout of the waveguides comprises two parts, layout I and layout II joined at an interface, each layout comprising N optical waveguides, i the optical path lengths increasing from waveguide 1 to waveguide N and the differences in optical path lenghts Δ(nL)I and Δ(nL)II between neighbouring waveguides being essentially constant for each layout I and II, respectively, and layout I is geometrically different from layout II, e.g. in the sense that the parts of the waveguides that are located the farthest from said interface contribute to the difference in optical path length Δ(nL) between neighbouring waveguides WGi, WGi+1. Small path length differences may be provided between neighbouring waveguides.

Description

OPTICAL MULTIPLEXER BASED ON ARRAYED WAVEGUIDE GRA_TING

Technical Field of the Invention:

This invention relates in general to the field of optical communications, in particular to broadband multiplexing or de-multiplexing of optical signals, e.g. in wavelength division multiplexing (WDM, CWDM as well as DWDM) systems.

The invention relates specifically to: An optical component suitable for use in separating or combining optical signals, the component comprising a number N of optical waveguides formed on a substrate, the waveguides WG_l5 WG₂, ...., WG_N having first and second ends connectable to first and second optical couplers, the waveguides having physical path lengths L_l5 L₂, ..., ^ between the first and second ends, the path lengths of the waveguides being defined by the centrelines of the waveguides, the optical path lengths increasing from waveguide WGj to waveguide WG_N, the difference in optical path length Δ(nL) between neighbouring waveguides WG_i5 WG_i+1 being essentially constant.

The invention may e.g. be useful in applications such as optical communication systems employing WDM.

Description of Related Art:

The need for bandwidth in long distance communications systems has led to the development of WDM-systems that are able to transmit a multitude of optical signals comprising independent information channels on a single optical waveguide by transmitting the signals at different wavelengths. This technology has a basic need for components that are able to multiplex and de-multiplex the optical signal(s) at each end of a waveguide. Various examples of such components have been implemented, typically either comprising resonator-based filters or interference- based filters. One example of the interference-based filter type is a so-called arrayed waveguide grating (AWG) device consisting of an input coupler (a splitter) and an output coupler (a combiner) and a phased array of multiple waveguides with constant path length difference between adjacent waveguides (thereby inducing a corresponding phase difference between the optical signals propagated in the waveguides). The input and output couplers are adapted to be coupled to input and output waveguides (e.g. optical fibres), respectively. Such a device is typically symmetric in the sense that it may work as a de-multiplexer (one input to a number P of outputs) or when the transmission direction is 'reversed' as a multiplexer (P inputs to one output). A schematic prior art arrayed waveguide grating device with the input and output couplers in the form of two focusing slab regions is shown in Figure 1 (cf. e.g. US-5,706₃377).

The theories of AWG are well documented (see e.g. M. Smit, Electron. Lett., p. 385,

1988, C. Dragone, Electron. Lett., p. 942, 1988 or H. Takahashi, et. al., Electron.

Lett., p. 87, 1990). For a centre wavelength λ₀, the order of the grating m is determined by

where c denotes the speed of light in vacuum, and FSR denotes the free spectral range, which is the spectral distance between the m±l order of interference, i.e. the frequency spacing of two resonance peaks (cf. e.g. K. Okamoto, Fundamentals of Optical Waveguides, Academic Press, 2000, p. 164, referred to as OKAMOTO in the following). The path length difference is given by: (2) n_cM = mλ₀ where n_c is the effective refractive index of the phased array waveguides. The effective refractive index is defined as the propagation constant β divided by wave number k (cf. e.g. OKAMOTO, p. 18). One can see from equation (1) that when FSR increases, m decreases. And from equation (2), when m decreases, ΔL decreases. For example, for an AWG designed for λ₀=1550 mn, FSR=10 nm, and n_c=1.5, one can determine from the above equations that m=155, ΔL=160.17 μm. Similarly, for an AWG with λ₀=1550 nm, FSR=500 nm, and n_c=1.5, we have m=3, ΔL=3.1 μm.

A typical layout of a prior art optical phased array is shown in Fig. 1. The waveguides of the phased array consist of straight waveguides and arc waveguides. When the path length difference decreases, it is more difficult to design a sufficient number of phased array waveguides that satisfies the constant ΔL requirements. Intuitively, one can see that if ΔL-0, two waveguides will have to be parallel which makes it impossible for them to join together at both the input slab centre and the output slab centre. As a result, for an AWG with a very broad spectral range, it is difficult if not impossible to design a large number of phased array waveguides to meet the requirements on crosstalk and insertion loss.

US-5,212,758 deals with an integrated optic device comprising an array of plural waveguides extending between a pair of couplers. The waveguides are laid out in an S -configuration providing equal path lengths for use of the component as an optical lens. A low order array multiplexer or de-multiplexer component may be implemented by insertion of an intermediate region between the two parts of the S- layout for providing small path length increments between neighbouring waveguides. A drawback of this solution is that it is aimed at providing relatively small path length differences in an otherwise equal-path length S-type layout by inserting paths consisting of concentric arcs of constant pitch between two identical (but mutually 180° rotated) halves of a C-configuration layout, thereby being relatively restricted in form and size. First of all, the symmetry of the two 'layout-halves' connecting to input and output slabs means the input and output slabs must be identical. Secondly, the path length difference is entirely achieved by the arcs in 'concentric arc' -section of the layout. This type of design creates three arc segments (A) and 5 straight (S) segments for each grating. The transition is S-A-S-A-S-A-S, often resulting in higher transition loss and bending loss.

Summary:

We propose a different approach to design the phased array waveguides for an AWG applicable for implementing any size of optical and/or physical path length differences Δ(nL) and/or ΔL, respectively, between adjacent waveguides (including very small values, such as e.g. < 10 μm), (nL) being the product of effective refractive index n and physical path length L, ideally integrated over the extent of the waveguide in question.

The object of the present invention is to provide a flexible scheme for implementing an optical component suitable for multiplexing or de-multiplexing optical signals within a relatively broad spectral range of wavelengths and with a relatively wide channel spacing.

This and further objects of the invention are achieved by the invention described in the accompanying claims and as described in the following.

An optical component suitable for use in separating or combining optical signals is provided, the component comprising a number N of optical waveguides formed on a substrate, the waveguides WG_l5 WG₂, ...., WG_N having first and second ends connectable to first and second optical couplers, the waveguides having physical path lengths L_l3 L₂, ..., L^ between the first and second ends, the path lengths of the waveguides being defined by the centrelines of the waveguides, and optical path lengths (nL)_j, (nL)₂, ..., (nL)_N, the optical path lengths increasing from waveguide WG_j to waveguide WG_N, the difference in optical path length Δ(nL) between neighbouring waveguides WG_i5 WG_i+1 being essentially constant (i=l, 2, ..., N-l). The physical layout of the waveguides comprises two parts, layout I and layout II joined at an interface, each layout comprising N optical waveguides WG_U, WG_{I 2}, ...., WG_IN and WG_π>1, WG_{π 2}, ...., WG_πN having effective refractive indices n_{effI 1}, n_effT;2, ..., n_effJN and n_effAι, n_effA2, ..., n_effAN, physical path lengths L_I;1, L₁₂, ..., L_IN and L_{n ι}, L_{π 2}, ..., L_πN and optical path lengths (nL)_Tjl, (nL)_{I 2}, ..., (nL)_IN and (nL)_{π A}, (nL)_{π 2}, ..., (nL)_πN, respectively, the optical path lengths increasing from waveguide 1 to waveguide N and the differences in optical path lengths Δ(n_ )_I^=(nL-)_{I i+1}-(nJ_-)_{I i} and

between neighbouring waveguides are essentially constant for each layout I and II, respectively. Layout I is geometrically different from layout II. In an embodiment of the invention, layout I is geometrically different from layout II in the sense that the parts of the waveguides that are located the farthest from said interface contribute to the difference in optical path length Δ(nL)=n_mL_i+1-n^~L_i between neighbouring waveguides WG_i5 WG_i+1.

In an embodiment of the invention, layout I is geometrically different from layout II in the sense that the combined layout of the waveguides lacks symmetry. The term lacks symmetry is taken to mean that the layout does not possess a rotational or mirror symmetry (or a combination thereof).

An advantage of the invention is that it provides an improved degree of design freedom compared to the prior art. A simpler structure of the layout may be achieved because the layout consists of two distinctively different portions of array waveguides. It is hence possible to design for any small or large ΔL with any number of phased array waveguides, because ΔL_{, ΔL_n and Ah are independent of ΔL and can be of any numbers that provide us with a desired geometry (where Δh is the spacing between the grating waveguides), including any values of the order m of the partial AWGs of layouts I and II. The degree of freedom is further increased, if the refractive index of the individual waveguides is allowed to vary, e.g. from waveguide to waveguide and/or over the length of a waveguide.

In an embodiment of the invention, layout I is geometrically different from layout II in the sense that the orders m of the partial AWGs of layouts I and II m_x and m_π, respectively, are different.

In an embodiment of the invention, layout I is geometrically different from layout II in the sense that Δm=m_I-m_π (where m_x-(n_xΔxL_x)/λ₀, X=I, II) is an odd number (2n + l, n=l, 2, ...).

In an embodiment of the invention, layout I is geometrically different from layout II in the sense that Δm=m_I-mι_I is an even number (2n, n=l, 2, ...).

The term "physical path length" of an optical waveguide e.g. the length L_{X p} of the p'th waveguide in layout X is in the present context taken to mean to be represented by the length of a centreline in a direction of light guidance of the waveguide between its end faces (one or both of which may be integrated with another waveguide (e.g. at the interface between layout I an II) or component (e.g. a coupler at the first or second ends of a particular waveguide), in which case the end point is taken at the interface in question. Similarly, by the term 'first and second ends' is in the present context meant the 'end faces' of a waveguide each typically coupled to (or integrated with) an optical coupler, e.g. a slab waveguide or another waveguide.

A waveguide is in the present context taken to be an optical waveguide comprising a longitudinally extending core region having a refractive index n-_ore(r,φ) in plane polar coordinates in a cross section perpendicular to the longitudinal direction, the core region being surrounded by a cladding region (possibly comprising several distinct regions) having a refractive index n_olad(r,φ), where n_core and n_clad and possible microstructures in the core and/or cladding region are so arranged and dimensioned relative to each other that the waveguide is able to guide light in its longitudinal direction.

The term "essentially constant" in connection with differences in optical and/or physical path length between neighbouring optical waveguides is in the present context taken to mean that the path length in question is constant within processing tolerances of the manufacturing process or processes used for implementing the optical component.

The term "layout I and layout II is joined at the interface" is in the present context taken to mean that the corresponding waveguides join at the interface, i.e. that end faces of layouts I and II are joined in a common plane, 'the interface'. In an embodiment of the invention, 'the interface' between layout I and layout II (each layout fulfilling the condition that the difference in optical path length between adjacent waveguides is constant) is a plane that is perpendicular to the centrelines of the waveguides of the resulting combined layout. In an embodiment of the invention the difference in optical path length between adjacent waveguides of layouts I and II is different from layout I to layout II.

In an embodiment of the invention, at least one of the waveguides has a linear section around the interface between layout I and II (i.e. it is linear when crossing and over a certain distance on each side of the interface). In embodiments of the invention, a majority or all of the waveguides have a linear section around the interface between layout I and II. In an embodiment of the invention a linear section is larger than twice the width of the waveguide in question, such as larger than 5 times the width, such as larger than 10 times the width, the width of a waveguide being taken as the width of the core region of the waveguide taken in a direction perpendicular to the direction of propagation of the waveguide and perpendicular to the growth direction of the layers constituting the waveguides on the substrate.

In an embodiment of the invention, the centrelines of corresponding waveguides from layout I and II are aligned. In this case, the centreline of the resulting waveguide constitutes a continuous curve over the interface. The centrelines do not, however, have to be aligned. In certain configurations it is an advantage to have the centrelines offset. In the transition from bend waveguides to straight waveguides e.g. - because the peak of the mode will shift towards the outer rim of the bend waveguides - it is advantageous to offset the interface so that the peak of the mode profile remains at the centre of the straight waveguides. In an embodiment of the invention, the 1. order derivatives of the parametric curve representation of the centrelines are continuous (i.e. the end faces are parallel at the interface).

The continuous curved waveguide does not only solve the problem of small loss but also reduces the noise floor since any change in the optical phase (that can be introduced by a non-continuous transition) will introduce noise in the transmission spectrum.

The term "geometrically different" is in the present context taken to mean that different geometrical templates for the construction of layout I and II may be used. For example, we can use straight - arc construction of Region I (comprising layout I) and 10th order polynomial for Region II (comprising layout II). Or any other templates we see fit. We are not limited by the fact that we have to create an AWG with equal length arrays first. Specifically, the resulting combined layout does NOT consist of two identical part layouts that are rotated 180 relative to each other and wherein the difference in path length is achieved by a third layout inserted between the two identical parts.

In an embodiment of the invention, the parametric descriptions of the curve family constituted by the centrelines of the waveguides of layout I and layout II, respectively, are different. In other words, it is not possible to generate a substantial section of the curve family of layout I from a corresponding section of layout II simply by a symmetry operation (rotation or mirroring). A substantial section is in the present context taken to mean more than 50% such as more than 90% of the physical length of a representative waveguide.

In an embodiment of the invention, the difference in length L_{j i}(s)-L_πN._(i.₁₎(s) of two parts of the waveguides WGy and Y^G^^^ of layout I and II, respectively, that are joined at an interface to constitute waveguide i (WG_;) - the partial lengths starting from the non-interface ends of the waveguides in question (respective first and second ends) - is different from 0 for s larger than a certain minimum value s₀ (i.e. for s > so), where s₀ « 1 (i.e. L_rιi(s₀)/L « 1 and

« !)> ^{where L}χ,_P(^s) is a parametric representation of the length of waveguide p of layout X (WG_{X p}), X=I, II and p=l, 2, ..., N, and where s=0 corresponds to the waveguide end farthest away from the interface connecting the two layouts and s=l corresponds to the interface- end. In other words L_{X p}(s^:=0)=0 and I__Xp(s=l)=L_XjP. The expression s₀ « 1 is taken to mean s₀ < 0.5, such as s₀ < 0.2, such as s₀ < 0.05.

In an embodiment of the invention the layout of the waveguides consists of two parts, layout I and layout II.

In an embodiment of the invention, the waveguides of layout I having the relatively shorter optical path lengths are connected at the interface to the waveguides having the relatively longer optical path lengths of layout II according to the following scheme: WG_U to WG_π>N, WG_{I 2} to WG_W, ..., WG_IN to WG^. In this embodiment the difference in optical path length Δ(nL) between adjacent waveguides is equal to Δ(nL)_I - Δ(nL)_π, i.e. to the difference in optical path length between adjacent waveguides of layout I minus the difference in optical path length between adjacent waveguides of layout II.

In an embodiment of the invention, the waveguides of layout I and II having the relatively shorter optical path lengths are connected at the interface and the waveguides of layout I and II having the relatively longer optical path lengths are connected at the interface according to the following scheme: WG to WG_{π ι}, WG_{I 2} to WG_{π 2}, ..., WG_IN to WGn_>N. In this embodiment the difference in optical path length Δ(nL) between adjacent waveguides is equal to ΔriiL)_! + Δ(nL)_π, i.e. to the sum of the difference in optical path length between adjacent waveguides of layout I and layout II, respectively. In an embodiment of the invention, the difference in optical path length Δ(nL) between adjacent waveguides is the same for layout I and II. In an embodiment of the invention, the effective refractive index is identical for all waveguides. In an embodiment of the invention, the difference in optical path length Δ(nL) between adjacent waveguides is different for layouts I and II. In an embodiment of the invention, the difference in optical path length Δ(nL) between adjacent waveguides between layouts I and II is larger than 10%o, such as larger than 25%o, such as larger than 50%o, the relative difference being defined as (Δ(nL - Δ(nL_π))/(Δ(nL)_I + Δ(nL_π)).

An advantage of embodiments of the invention is that the (optical) path length difference of the final AWG is the summation of (optical) path lengths of layout I and II. The waveguides may in an embodiment comprise two arcs and three straight segments, which minimizes transition loss between straight and arc segments. In addition, the path length difference is imbedded in the gratings similar to the regular AWGs (see e.g. M. Smit, Electron. Lett., p. 385, 1988), rather than having one section dedicated to it, as disclosed in the prior art.

In an embodiment, an asymmetrical C-shaped AWG may be thus constructed. Being able to design an asymmetrical AWG for high order brings many advantages; one of them is the potential spacing-saving. In the case of a multiplexer/demultiplexer chip, the slab itself is has a significant impact on the necessary chip size (e.g. for a standard AWG, a slab may have a dimension of 17.2 mm), having two of them will make the chip rather large. However, since the input slab is only distributing light, a C-shaped construction enables the input slab to be only half of the length of the output slab, thereby reducing the chip size, and increasing the productivity (#of chips on a wafer), and thus to reduce cost per chip. The possibility of make asymmetric AWGs thus has the potential of optimizing the total layout of the optical component (e.g. optical MUX/DEMUX).

The freedom to design different slabs for input and output also brings freedom in designing waveguide pitch. In typical applications the frequency spacing is 100 GHz. The frequencies could e.g. be 193.1 THz, 193.2 THz, 193.3 THz, etc. However for certain applications the frequencies in between are used (i.e. 193.15 THz, 193.25 THz, 193.35 THz, etc.). This can normally only be achieved by a new design. However, designing the input slab to be longer than the output slab can enable the interleaving of more input waveguides. As an example, a typical 32 channel 100 GHz AWG has an input waveguide pitch of 20 μm and typical waveguide width of 12 μm. In such a configuration it is not possible to place the input waveguides with less than 100 GHz spacing. By increasing the length of the input slab with a factor of 1.5 the corresponding waveguide pitch for 100 GHz spacing will be 30 μm. It is now possible to position input waveguide in between with a 50 GHz offset. The same component will therefore be able to service both frequency plans.

An advantage of the above embodiments is that any difference in optical and/or physical path length between adjacent waveguides (including arbitrarily small and large differences) may be implemented in a variety of ways. Further, the layout may be adapted to the available space and locations relative to possible other components on a substrate of an integrated optical chip. In an embodiment of the invention, the pitch between adjacent waveguides at the interface is constant (i.e. identical for all waveguides of the AWG). This has the advantage of making it easier to design and link up the two regions comprising layout I and II. Alternatively the distance between L_; and L_i+1 perpendicular to the centrelines of the waveguides may vary according to some predetermined algorithm or be non-systematic, such as random.

In an embodiment of the invention, the distance between adjacent waveguides is larger than the largest width of the waveguides in question, such as larger than twice the width, such as larger than 4 times the width of the largest width of the waveguides in question.

In an embodiment of the invention, the centrelines of the waveguides WG_l5 WG₂, ...., WG_N are perpendicular to the interface. This has the advantage of enabling a modular design of layout I and II, because with a well defined interface, the design of each region does not need to be dependent on the other. Instead, as long as each region satisfies the boundary condition at interface, one can be sure that the two regions can be joined seamlessly. Alternatively the centrelines of the waveguides might cross the interface at an angle different from 90.

In an embodiment of the invention, the effective refractive indices n_effu, n_{effI 2}, ..., n_effIN of the waveguides of layout I are identical.

In an embodiment of the invention, the effective refractive indices n_effA1, n_effA2, ..., n_effAN of the waveguides of layout II are identical.

In an embodiment of the invention, the effective refractive indices of all waveguides are identical, so that the optical path length equals the physical path length for any given waveguide.

In an embodiment of the invention, the centrelines of at least one (and preferably all) of the waveguides WG_l3 WG₂, ...., WG_N has a continuously varying tangent. The term "a continuously varying tangent" is in the present context taken to mean that the centrelines are represented by curves that are 1. order continuous. In other words, at least one and preferably all of the waveguides are smooth, which has the advantage of reducing loss of optical power. They might, alternatively, be piecewise linear, e.g. with a small distance between each point of discontinuity.

In an embodiment of the invention, the centrelines of at least one of the waveguides WG_l5 WG₂, ..... WG_N joined from corresponding waveguides of layout I and II are aligned at the interface. In an embodiment of the invention, the centrelines of all the waveguides are aligned at the interface. This has the advantage of minimizing losses at the interface.

In an embodiment of the invention, widths w_y and w_{π i} of the waveguides WG_{r i}, and WG_{π i}, respectively, (i=l, 2, ..., N) of layouts I and II are equal at the interface where the waveguides are joined. In an embodiment of the invention, the widths of all waveguides of layout I and II are identical.

In an embodiment of the invention, the curvatures κ_{τ i} of the centrelines of the waveguides of layout I have the same sign. In an embodiment of the invention, the curvatures κ_{π i} of the centrelines of the waveguides of layout II have the same sign.

The term "the curvature of the centreline of a waveguide" is in the present context taken to mean the length of the vector of curvature calculated with sign, where the vector of curvature κ_n of curve r(s) at a point P (given as rj(s)) is defined as d²r_x/ds² = Kn, being the curvature of the curve at the point P and n a unit vector.

In an embodiment of the invention, curvature κ_υ and κ_{π i} of the centrelines of the waveguides of layout I and II, respectively, have opposite signs. This has the advantage of providing a compact S-shape configuration.

In an embodiment of the invention, curvature κ_τ>i and κ_{π i} of the centrelines of the waveguides of layout I and II, respectively, have the same sign.

In an embodiment of the invention, said first and second ends of said N optical waveguides are coupled to first and second optical couplers to form an optical multiplexer or de-multiplexer. In an embodiment of the invention, at least one of the couplers is a slab waveguide in the form of a planar area, which is large compared to the area of an individual waveguide and designed to support lightwave transmission between input and output waveguides. In an embodiment of the invention the couplers are implemented as star couplers. In an embodiment of the invention the couplers are implemented as P to P couplers, e.g. in the form of multimode interference couplers. The latter have advantages such as low loss, large bandwidth and simple design procedures.

The idea of having two different AWG's adjoined can have the advantage that since the purpose of the of slabs and the grating is to act as a lens between the input and output waveguides, combining two different gratings will have the effect of combining two different "lenses". This is not specifically related to low-order devices, but can also be used in DWDM applications (where the order is normally high). The different gratings can influence the transmission spectra in ways that cannot be achieved otherwise.

An advantage of using two different AWG-partial designs is that the input and output slabs of an optical MUX/DEMUX component need not necessarily be identical. The flexibility can be used to introduce different properties in the different parts of the waveguide gratings, e.g. one design with a particular loss, another design having a particular transmission curve, etc. Such dedicated 'sub-functions' may then be composed (as layout I or II) to a resulting waveguide grating having particular properties.

In an embodiment of the invention, the array waveguides and the optical couplers are implemented as an integrated optical circuit on the same substrate. This has the advantage of providing a small size component that is readily suitable for integration with other optical components such as lasers, switches, variable optical attenuators, splitters, etc.

In an embodiment of the invention, an optical component according to the invention is implemented in silica-on-silicon planar technology. Alternatively, waveguides may be formed on a silicon substrate with Si₃N₄ cores and silica cladding or in any other relevant technology for forming optical waveguides on a substrate.

In an embodiment of the invention, the optical component is optimized to wavelengths in the range 1200 to 1700 nm. In an embodiment of the invention, the optical component is optimized to wavelengths in the range 190 nm to 11 μm, such as in the range of 250 nm to 3.6 μm, such as in the range of 850 nm to 1800 nm, such as in the range of 1300 nm to 1600 nm.

In embodiments of the invention, the number of WDM-channels (as exemplified by the ITU channels around 1 550 nm) that are separable by the optical component is larger than 2, such as larger than 15, such as larger than 30, such as larger than 100.

The number N of waveguides in an AWG is typically larger than the number of channels separable by the optical component, e.g. four times larger. In embodiments of the invention, the number of waveguides is larger than 15, such as larger than 30, such as larger than 100, e.g. in the range 50 to 100.

In an embodiment of the invention, the channel spacing between adjacent channels is less than 100 GHz, such as less than 50 GHz such as less than 25 GHz.

It should be emphasized that the term "comprises/comprising" when used in this specification is taken to specify the presence of stated features, integers, steps or components but does not preclude the presence or addition of one or more other stated features, integers, steps, components or groups thereof.

Brief Description of the Drawings:

The invention will be explained more fully below in connection with a preferred embodiment and with reference to the drawings in which:

fig. 1 shows a prior art optical AWG multiplexer,

fig. 2 shows an S-form optical AWG multiplexer or de-multiplexer according to the invention,

fig. 3 shows a C-form optical AWG multiplexer or de-multiplexer according to the invention, and fig. 4 shows a cross section of an optical component according to the invention at the interface between layout I and II (line AA on fig. 2).

The figures are schematic and simplified for clarity, and they just show details which are essential to the understanding of the invention, while other details are left out. Throughout, the same reference numerals are used for identical or corresponding parts.

Detailed Description of Embodiments :

Fig. 1 shows a prior art optical AWG multiplexer as discussed in the section 'Description of related art' above.

Fig. 2 shows an S-form optical AWG multiplexer or de-multiplexer according to the invention.

In the embodiment of fig. 2, the optical device 1 is divided into two sections termed Region I and Region II. Each region consists of a slab 21, 22 (either input or output) and a group of array waveguides WG_U, ...., WG_IN and WG_{π ι}, ...., WG_{π N} of regions I and II, respectively, where N is larger than 2 and equal to 10 in the illustrations of figs. 2 and 4, but typically larger than 10. The waveguides of layouts I and II typically satisfy the following requirements, where it is assumed that all waveguides have identical effective refractive indices. 1. The physical path length differences between adjacent waveguides are constant ΔLj and ΔL_π (not ΔL) for Regions I and II (the physical layout of which are termed layout I and II in the following), respectively. 2. ΔLj and ΔL_U are chosen such that ΔL = ΔL_j - ΔL_U have a desired size.. When the above two conditions (1 and 2) are met, i.e. assuming that L_{τ i} = h_lΛ + (i- l)ΔLι and L_ιy = L_{π ι} + (i-l)ΔL_TI (i=l, 2, ..., N) and that the waveguides of region I having the relatively shorter physical path lengths are connected to the waveguides having the relatively longer physical path lengths of region II according to the following scheme: WG_U to WG_{π N}, WG_L2 to WG_π,N.„ ..., WG_{r j} to WG_{π N.(M)}, ..., WG_IN to WG_{π I}, the length of waveguide i (WG_;) of the combined layouts of region I and II is the sum of the length of waveguide i of region I (WG_{j ;}) and the length of waveguide N-(i-l) of region II (WG_ry^-_!)) equals L, = L_u + (i-l)ΔL, + L_I + (N-(i-l)-l)ΔL_π

L; = L_u + L_nΛ + (N-l)ΔL_π + (i-l)(ΔLj - ΔL_π) = constant + (i-l)ΔL, because L_u + L_π , + (N-l)ΔL_π is constant for a given layout.

Thus, the path length difference L_i+J - -_.,-, between adjacent waveguides of the resulting layout combining region I and II as indicated above is AL (^~ΔL_I - ALjJ.

Combined waveguides of layout I and II, i.e. WG_l5 WG₂, ..., WG_N have first and second ends 71 and 72 that are coupled to first and second couplers 21, 22 in the form of slab waveguides for respectively splitting the light from an incoming signal at its input face 211, the incoming signal comprising a multitude of wavelengths λ_l5 λ_j, ..., λ_p, and combining the waveguide outputs by constructive interference to present signals each comprising one of the wavelengths λ_1; λ₂, ..., λ_p at a predeteraiined location at the output face 212 of the coupler 22 (assuming a demultiplexing function of the component).

The waveguides of layout I and II meet at the interface 3, each waveguide being perpendicular to the interface. A schematic cross section of the layout perpendicular to the direction of light propagation along the interface (line AA) is shown in f g. 4.

In an embodiment of the invention, all waveguides are perpendicular to a line defining the interface (as in fig. 2 and 3) and the distance between adjacent intersection points of adjacent waveguides (as defined by the intersection of their centrelines with the interface line) is constant Ah. This has the advantage of making it easier to design and link up two regions, but it is not absolutely necessary. However, as long as the two regions can be joined at the interface to allow light to propagate from one region to another with minimum loss, these latter two conditions do not need to be observed.

Fig. 3 shows a C-form optical AWG multiplexer or de-multiplexer according to the invention. If, alternatively, as shown in fig. 3, the waveguides WG , ..., WG_IN and WG_π ι, ..., WG_πN of regions I and II having the relatively shorter physical path lengths are mutually connected and the waveguides of regions I and II having the relatively longer physical path lengths are mutually connected, i.e. according to the following scheme: WG_U to WG_W, WG_I>2 to WG_D>2, ..., WG_y to WG_IU, ..., WG_lN to WG_π>N, a constant length difference between adjacent waveguides of the combined layout ΔL = ΔL + ΔL_U is implemented (the length differences zL--_/5 ΔL_U between adjacent waveguides of, respectively, layouts I and II being constant).

The waveguides of the embodiment in fig. 3 each have a. linear part around the interface 3. For waveguide WG_N, being constituted by WG_IN of layout I and WG_πN of layout II, joined at the interface 3, the linear part is indicated by segments 81, 82 of waveguides WG_IN and WG_πN, respectively. In the embodiment shown in fig. 3, the length of the linear segments of the waveguides decrease from WG_N (WG_IN + WG_π;N) to WGi (WG_{T j} + WG_jy). The linear parts are perpendicular to the interface 3. The linear parts extend a distance of at least 5 times the width of the core region of the waveguide in question on both sides of the interface.

The distance Δh_t between adjacent intersection points of adjacent waveguides WG_i5 WG_i+ι (as defined by the intersection of their centrelines with the interface line) in the embodiment of fig. 3 is larger than twice the width of the waveguides WG_i5 WG_i+1 having the largest width. In the present embodiment, the corresponding distance between adjacent waveguides at the interface 71, 72 to the couplers 21, 22 is likewise larger than twice the width of the waveguides WG_i5 WG_i+] having the largest width.

Fig. 4 shows a cross section of an optical component according to the invention at the interface between layout I and II (line AA on fig. 2).

Fig. 4 shows a substrate 5 with bottom 61 and top 62 cladding layers in which core regions 4 are embedded, thereby defining optical waveguides WG_l5 ...., WG_N. The waveguide core of WG_j have thickness t_; 402 and width w_f 401. The mutual distance between adjacent waveguides WG_j and WG_i+I is d_j 403. In an embodiment of the invention, the thickness and width of the core regions in a cross section perpendicular to the direction of light propagation of the waveguides, respectively, are identical for all waveguides. In an embodiment of the invention, the distance between adjacent waveguides is the same for all waveguides at the interface between layout I and II.

An example of a centreline 405 of a waveguide is indicated by the point of intersection of the diagonals of the rectangular waveguide in fig. 4 (as seen in a cross section of the waveguide at the interface between layout I and II).

The design of the actual layout of a particular AWG is made with a view to several parameters including the following:

• Bend loss (the maximum acceptable bend loss (contributing to total insertion loss) influences the maximum curvature of the individual waveguides and thereby the branch geometry and hence the compactness of the design)

• Cross talk between neighbouring waveguides (the maximum acceptable cross talk (or minimum acceptable channel isolation) limits the minimum distance between neighbouring waveguides (and thus the compactness of the design) and is further influenced by the process tolerances governing waveguide imperfections (e.g. variations in cross sectional waveguide dimensions and refractive indices))

The actual simulation of a given layout optimized for a given technology (refractive indices and distribution), waveguide dimensions (width, thickness, mutual centre-to- centre distance), central wavelength, number of channels, free spectral range, etc. may be made by several methods, e.g. the Beam Propagation Method (BPM), e.g. using the OlympIOs design, simulation and mask layout software package from C2N, Enschede, The Netherlands.

Example of a fabrication technology for an optical component according to the invention:

An optical component according to the present invention can be fabricated in a number of different planar technologies such as in polymers, in Silicon-on-insulator (SOI), Litliiumniobate (LiNbO3), III-N-semiconductors (incl. GaAs- and InP -based systems), as well as in silica-on-Silicon and others. In a preferred embodiment of the present invention the silica-on-silicon planar technology is used. This technology presently produces the most advanced and technically developed planar waveguide components. Silica waveguides possess a number of highly attractive properties such as material compatibility (optical fibres are made from the same material, silica), optimum coupling between fibre and waveguide component (refractive indices and index differences are comparable), low absorption- and propagation losses, low birefringence, high stability and low cost. Furthermore, the technology used to fabricate these silica waveguides is identical to the technology used in fabricating integrated electrical circuits such as CPU's (Central Processing Units in computers) and e.g. RAM (Random Access Memory), thus this technology has matured during the last more than thirty years and is known to be capable of mass production.

The various steps in the fabrication of an optical component according to the present invention in a silica-on-Silicon technology are described in the following.

A clean and bare Silicon wafer (used as substrate, 5 in fig. 4) is firstly oxidized to provide an optical isolation layer of silica sufficiently thick that the magnitude of the evanescent field tail of the field pertaining to the waveguide cores is sufficiently low to ensure negligible propagation loss. This first layer of silica is referred to as the buffer layer (61 in fig. 4). On top of the buffer layer a layer of doped-silica is deposited, containing one or more dopants that effectively act to increase the refractive index of said layer. This doped layer of silica glass is referred to as the core layer (4 in fig. 4). Depending upon the method used to deposit the core layer a high temperature treatment (known as an anneal step) may be advantageous in order to stabilize the optical and/or mechanical properties of said layer. The optical waveguide circuitry is defined through standard optical lithography where a UN- transparent plate containing typically a chromium pattern replica of the waveguide design pattern is pressed against a layer of UN-sensitive polymer which has been spin coated onto the surface of the- core silica layer, subsequently the UN-sensitive polymer is exposed through the mask and the pattern is developed. Following the exposure and development of the waveguide pattern into the polymer layer, the polymer pattern is used as masking material for dry etching (e.g. RIE - Reactive Ion Etching, ICP - Inductively Coupled Plasma) into the core silica layer. Alternatively a second masking material is sandwiched between the silica core layer and the UN- sensitive polymer layer, which is used to enhance selectivity and waveguide core profile. In this way the design waveguide pattern is transferred into the core silica layer having predetermined cross-sectional properties as well as refractive index. In order to protect the recently defined waveguide core, and in order to enhance symmetry in the structure transverse to the direction of propagation a layer of silica (62 in fig. 4) with optical properties as close to those of the buffer layer as the chosen fabrication technology permits is deposited on top of the core structure.

The refractive index of core and/or cladding regions of a waveguide may e.g. be locally varied by UN induced index changes, local heating or electro optic effects. Thermal induced index changes may e.g. be implemented using a local heater element on top of the waveguide. Another way of changing the effective refractive index is by changing the width of the waveguide. A third way of changing the effective refractive index is by UN-induced index changes where a focused UN-laser beam is scanned across the waveguide causing an index change in a waveguide with an UV-sensitive dopant. Other means, such as electro-optical and magneto-optical induced index changes can also be applied.

Various relevant aspects of the silica-on-silicon technology are e.g. discussed in M. Kawachi, "Silica waveguide on silicon an their application to integrated-optic components", Opt. Quant. Electr. 22 (1990) 391-416. Various relevant aspects of low loss plasma enhanced chemical vapour deposited planar waveguides are e.g. discussed in Christian Laurent-Lund, "PECVD grown Multiple Core Planar Waveguides with Extremely Low Interface Reflections and Losses", Photon. Technol. Lett. 10 (1998) 1431-1433.

Some preferred embodiments have been shown in the foregoing, but it should be stressed that the invention is not limited to these, but may be embodied in other ways within the subject-matter defined in the following claims.

Claims

Claims:

1. An optical component suitable for use in separating or combining optical signals, the component comprising a number N of optical waveguides formed on a substrate, the waveguides WG_l9 WG₂, ...., WG_N having first and second ends connectable to first and second optical couplers, the waveguides having physical path lengths L„ L₂, ..., L^ between the first and second ends, the path lengths of the waveguides being defined by the centrelines of the waveguides, the optical path lengths increasing from waveguide WG_j to waveguide WG_N, the difference in optical path length Δ(nL) between neighbouring waveguides WG_i5 WG being essentially constant wherein the physical layout of the waveguides comprises two parts, layout I and layout II joined at an interface, each layout comprising N optical waveguides WG_y and WG_{π i} having effective refractive indices n_effAi, and n_{effπ i}, physical path lengths L_y and L_{π i}, and optical path lengths (nL)_I;i and (nL)_{π i}, respectively, i=l , 2, ..., N, the optical path lengths increasing from waveguide 1 to waveguide N and the differences in optical path lengths Δ(nL)_I and Δ(nL)_π between neighbouring waveguides being essentially constant for each layout I and II, respectively, and layout I is geometrically different from layout II.

2. An optical component according to claim 1 wherein layout I is geometrically different from layout II in the sense that the parts of the waveguides that are located the farthest from said interface contribute to the difference in optical path length Δ(nL) between neighbouring waveguides WG_S, WG_i+]j.

3. An optical component according to claim 1 or 2 wherein layout I is geometrically different from layout II in the sense that the combined layout of the waveguides lacks symmetry.

4. An optical component according to any one of claims 1-3 wherein layout I is geometrically different from layout II in the sense that the orders m of the partial

AWGs of layouts I and II m_: and m_u, respectively, are different.

5. An optical component according to any one of claims 1-4 wherein the waveguides of layout I having the relatively shorter optical path lengths are connected at said interface to the waveguides having the relatively longer optical path lengths of layout II according to the following scheme: WG to WG_πN, WG_{I 2} to WG_πN._l5 ..., WG_IN to G_π,_.

6. An optical component according to any one of claims 1-4 wherein the waveguides of layout I and II having the relatively shorter optical path lengths are connected at the interface and the waveguides of layout I and II having the relatively longer optical path lengths are connected at the interface according to the following scheme:

WG_U to WG_IU, WG_U to WG_It2, ..., WG_IιN to WG_π,N.

7. An optical component according to any one of the preceding claims wherein the centrelines of the waveguides WG_l5 WG₂, ...., WG_N are perpendicular to the interface.

8. An optical component according to any one of the preceding claims wherein the effective refractive indices n_eff , n_{effI 2}, ..., n_effIN of the waveguides of layout I are identical.

9. An optical component according to any one of the preceding claims wherein the effective refractive indices n_effA1, n_effA2, ..., n_effAN of the waveguides of layout II are identical.

10. An optical component according to any one of the preceding claims wherein the centrelines of at least one of the waveguides WG_j, WG₂, ...., WG_N has a continuously varying tangent.

11. An optical component according to any of the preceding claims wherein the centrelines of at least one of the waveguides WG_l5 WG₂, ...., WG_N joined from corresponding waveguides of layout I and II are aligned at the interface.

12. An optical component according to any of the preceding claims wherein the curvatures κ of the centrelines of the waveguides of layout I have the same sign and the curvatures κ_{π i} of the centrelines of the waveguides of layout II have the same sign.

13. An optical component according to claim 12 wherein curvatures K_y and κ_{a i} of the centrelines of the waveguides of layout I and II, respectively, have opposite signs.

14. An optical component according to claim 12 wherein curvature K_y and κ_{π i} of the centrelines of the waveguides of layout I and II, respectively, have the same sign.

15. An optical component according to any of the preceding claims wherein said first and second ends of said N optical waveguides are coupled to first and second optical couplers to form an optical multiplexer or de-multiplexer.

16. An optical component according to claim 15 wherein at said optical couplers are implemented as slab waveguides.

17. An optical component according to claim 16 wherein the combination of the design of said first slab and layout I of said optical waveguides and the design of said second slab and layout II of said optical waveguides are individually optimized.

18. An optical component according to any of the preceding claims wherein the array waveguides and the optical couplers are implemented as an integrated optical circuit on the same substrate.

19. An optical component according to any of the preceding claims wherein the waveguides are implemented in silica-on-silicon or silicon-oxy-nitride planar technology.