GB2227854A - Integrated optics asymmetric y-coupler - Google Patents

Abstract

An integrated optics beam splitting/combining coupler in which a first integrated optics waveguide (5) is optically coupled with second and third waveguides (7, 6) takes the form of an asymmetric Y. The first waveguide forms the stem of the Y, the third waveguide is in line with the first and forms a through arm, and the second waveguide forms a angled arm which is optically coupled with the stem by means of a mirror (9). <IMAGE>

Description

Integrated Optics Asymmetric Y-Coupler This invention relates to integrated optics devices consisting of or including one or more asymmetric Y-couplers. A Y-Coupler is a three-port device in which light launched into the stem of the Y is apportioned between the two branches. Alternatively the device may be used for combining into a single output from the stem light launched into either of the branches.

Devices of this kind find application for instance as optical taps and in optical cross-point switching. If, in an integrated optics Y-coupler at the point where the stem of the Y is optically coupled with the two branch waveguides, all three waveguides extend in the same direction then, because the integrated optics waveguides are unable to negotiate tight bends without losing their waveguiding properties, it is clear that the length of the device tends to be not inconsiderable in order to accommodate the swan-neck bends required to achieve adequate separation of the distal ends of the two branches. The overall length of such devices can however be substantially reduced by incorporating discrete reflectors into the structure.

An example of this approach is given in the paper by A.

Ajisawa et al entitled 'Monolithically Integrated Optical Gate 2x2 Matrix Switch sing GaAs/AlGaAs Multiple Quantum Well Structure' appearing in Electronics Letters 8th October 1987 Vol. 23 No. 21 pp 1121-2, to which attention is directed. The 2x2 matrix switch described in that paper employs four integrated optics Y-couplers in which light directed into the stem 0 of the Y strikes two discrete mirrors arranged at 90 to each other. The reflected light then enters the two branch waveguides which extend in opposite directions at right angles to the stem in a Tee configuration.The four (Tee configuration) Y-couplers are linked by rectilinear waveguides and a further eight discrete mirrors to form the 2x2 matrix switch.

The present invention relates to an asymmetric form of integrated optics Y-coupler which uses only one discrete mirror instead of the two mirrors of the Ajisawa et al paper referenced above. Using this approach it is possible to construct a 2x2 matrix switch which requires no additional discrete mirrors to link the four Y-couplers. Using this configuration light is required to undergo a maximum of two reflections in its passage through the switch, whereas with the Ajisawa et al configuration the light has to undergo four reflections.

According to the present invention there is provided an integrated optics asymmetric Y-coupler in which first, second and third optical waveguides are arranged in relation to a mirror such that light propagating in the first waveguide toward the mirror is partially intercepted by the mirror such that a portion of the light is reflected by the mirror into the second waveguide while substantially all the rest of the light is not intercepted by the mirror but is launched from the first waveguide, without reflection, into the third waveguide. Preferably the mirror and the second waveguide are disposed on opposite sides of the third waveguide.

There follows a description of integrated optics matrix switches and optical taps incorporating asymmetric Y-couplers embodying the invention in preferred forms. The description refers to the accompanying drawings in which: Figure 1 depicts a 2x2 matrix switch incorporating four asymmetric Y-couplers Figures 2 to 5 depict successive stages in the manufacture of one of the asymmetric Y-couplers of the switch of Figure 1 Figure 6 depicts an alternative form of asymmetric Y-coupler, and Figure 7 is a schematic diagram of a tap for a bidirectional data highway.

Figure 1 depicts the configuration of four asymmetric Y-couplers 1, 2, 3 and 4 linked to form a 2x2 matrix switch. The waveguides of the device are ridge waveguides. The stem of each Y-coupler is defined by a ridge 5; one branch of the Y-coupler, the through arm, by a ridge 6; and the other branch, the angled arm, by a ridge 7. The ridges 5, 6 and 7 are delineated in a planar surface by the etching of channels 8. The stem waveguide of each Y-coupler is optically coupled with its angled arm waveguide via a plane mirror provided by one perpendicular side wall 9 of an etched well 10. The angled arm waveguides of one pair of Y-couplers intersect those of the other pair.In order to minimise 0 cross-talk problems each mirror is inclined at 22.5 to the axis of its respective through arm waveguide so that the pairs of angled arm waveguides intersect at right angles.

In order to provide a switching facility the through arm waveguides and angled arm waveguides need to incorporate some form of structure by means of which the attenuation of those waveguides can be electronically controlled. A multiple quantum well structure may be used for this purpose to control the attenuation by an electro-absorption effect. This is the approach employed in the matrix switch described by A. Ajisawa et al to which previous reference has been made. We generally prefer to use an actie strcct::r ;' capable of providing optical amplification.The use of amplification can make good the intrinsic loss occasioned by beam splitting, waveguide absorption and scattering loss, and hence makes it possible to provide an acceptable optical path through a cascade of several couplers. The peak gain of such an amplifier, and hence the maximum modulation it is capable of providing, will generally be at a wavelength close to that which is convenient for either an on-chip or an external semiconductor laser source. Additionally the spectral range of operation is liable to be broader than that of typical loss modulators using either the quantum confined Stark effect or the Franz Keldysh effect.

To provide active amplification regions for operation in the wavelength region of about 1.5 pm the matrix switch is manufactured from a conventional structure of semiconductor wafer designed for 1.5 pm emission ridge waveguide semiconductor lasers. Such a wafer comprises an n-type InP substrate on which are grown in succession a set of five epitaxial layers comprising an n-type InP layer, an undoped active layer of i = 1.5 pm quarternary material, a waveguide layer of A =1.18 material, a p-type cladding layer of InP, and a more heavily doped p-type capping layer of ternary or quaternary material.

The processing of the epitaxial layers to provide the ridge waveguides, the discrete mirrors and the top contacts of the matrix switch will now be described with particular reference to the sequence of Figures 2, 3, 4 and 5 which depict a perspective view of a portion of the matrix switch incorporating one of its component asymmetric Y-couplers. As depicted in Figure 2, first of all standard photolithographic techniques are used to delineate the ridge guides 5, 6 and 7 by etching channels 8 through the capping layer and the cladding layer. Channels with vertical side walls are required since parts of the side walls function as + upper parts of the mirrors 9, and so reactive ion etching is the preferred etching method.Reactive ion etching with a methane hydrogen mixture will not automatically halt at the interface between the cladding layer and the underlying waveguide layer, and so reactive ion etching is halted just short of this interface, and then a selective wet etch is used to expose the waveguide layer.

A second stage of photolithography is then used to delineate wells 10 which extend through the waveguide layer, the active layer, and about 1 ,um into the underlying InP. Reactive ion etching is used to etch this well and so provide the lower parts of the mirrors 9 as depicted in Figure 3. Good alignment of the lower parts of the mirrors 9 with their upper parts is clearly desirable. This is conveniently achieved by 'self-alignment' processing in which the second stage of photolithographic etching is performed without previously having removed the residual etch-resistant oxide masking material of the first stage of photolithography.

Oxide is then deposited on the surface to form an electrically insulating layer through which windows 40 (Figure 4) are opened to expose the underlying capping layer along all the ridges 5, 6 and 7 except for the immediate vicinity of the region where they join at the four Y-couplers. Next the surface is metallised, and then further photolithography and ion beam etching is employed to divide the metallisation into two contact regions 41 (Figures 1 and 4) covering pairs of windows on the ridges 5 defining the stems of the Y-couplers, two contact regions 42 covering windows on the ridges 6 defining the through arms of the couplers, and a further contact region 43 covering a common window on the ridges 7 defining all four angled arms.It will be observed that the ion beam etching required to separate the metallisation into discrete contact regions LXt.' 's ver ridges and channels. This can make for difficulty in achieving adequate electrical isolation of each contact from its neighbours. This problem can be eased by modifying the design of the ridge waveguides to include webs 50, as depicted in Figure 5, which are positioned to provide a substantially planar surface on which to form the entire perimeter of each of the contact regions 41, 42 and 43.

All the ridge waveguides are relatively highly lossy in the absence of any current injection through the contact regions. Injection of current through the contact regions 41 is employed either to remove substantially all loss from the ridge waveguides 5, or to provide these ridge waveguides with a small amount of gain. The size and extent of the mirrors 9 are chosen to provide a substantially equal division of power between the through arm and the angled arm. If straight through coupling is required, the contact region 43 is not energised, and hence the angled arms 7 remain highly attenuating and provide minimal cross-coupling. At the same time the two contact regions 42 are both energised sufficiently to provide substantial unity gain between the direct-coupled input and output ports.Conversely, if cross-coupling of the input and output ports is required, energisation of both contact regions 42 is maintained, contact region 43 is also energised, but contact regions 42 are left unenergised.

A performance analysis has been computed, using the overlap integral approach, to determine the proportion of the zero and first order mode power directed from the stem of one of the component Y-couplers towards the through and angled arms that is launched into those arms respectively in zero and first order guided modes in these arms. This analysis indicates that in the case of a ridge-guided InGaAsP structure with an effective dielectric constant step of 02, and 4pm stripe width for tlie ridges 5, and o:r stripe widths for the ridges 6 and 7, unity gain through the device can be achieved with current drives of a few tens of mA's.This computation was on the basis of a device, whose overall length is 400 um, and whose through arm length from Y-coupling region 1 to Y-coupling region 2 is 200 Fm, and assumes a reflector loss of 2.5dB per reflector. At the lower end of this range of drive level it is possible to arrange for the device to operate as a predominantly zero order mode device. In other circumstances operation will normally be confined to the zero and first order modes only.

A drawback that is possessed by the Tee-configuration couplers of A. Ajisawa et al, to which previous reference has been made, is that when light is directed into the stem of the coupler, diffraction occurs at the edge that is formed by the line joining the two mirrors that divide the power into its two components, and this edge registers with the region of highest field intensity for the zero order mode.The same drawback is also present in the asymmetric Y-couplers described above with reference to Figures 1 to 5, where the corresponding diffracting edge is the side edge of the mirror nearest the through and angled arms of the coupler, but the angles of the two facets that define this edge in the case of the asymmetric Y-coupler are more favourably oriented with respect to the incident light, presenting an obtuse angle to the light incident, and hence the diffraction loss is predicted to be smaller. In the case of a coupler operating in zero order mode this source of diffraction loss can be eliminated by ensuring that the optical field intensity is zero along its length. For this purpose the arrangement of the ridge waveguides is modified in the manner now to be described with reference to Figure 6.

In the asymmetric Y-coupler of Figure 6 the stem 60 waveguide, the through arm 61 waveajui;3e, ani the angled arm waveguide 62, all have the same width, but the stem waveguide is laterally coupled with an auxiliary waveguide 63. The length of the coupling region is chosen, in relation to the strength of coupling, so that it functions as a 3dB splitter in order that, at the coupling region, half the power launched into the stem 60 waveguide remains in that waveguide while the other half has been transferred to the auxiliary waveguide. This auxiliary waveguide 63 is optically coupled with the angled arm waveguide 62 by way of a mirror 64 formed by one vertical side wall of an etched well 65.Not only does this use of an auxiliary waveguide serve to avoid having optical power at the diffracting edge, it also provides a closer match between the field distributions directed at the ends of the through and angled arm. On the other hand the performance of this coupler of Figure 6 is not so good for propagation of light in the reverse direction since light launched into the coupler by way of the through arm waveguide 61 is coupled virtually exclusively into the stem waveguide 60, but then substantially half the power is lost by being coupled into the auxiliary waveguide 63. Similarly light launched into the coupler by way of the angled arm waveguide 62 is relatively efficiently coupled into the auxiliary waveguide 63, but only half that power is then coupled into the stem waveguide 60. Thus the asymmetric Y-couplers of Figure 6 are best suited to single mode applications in which the light propagation is unidirectional rather than bidirectional. In these circumstances such couplers are used for beam splitting, and if beam recombination is also required, then in those places asymmetric Y-couplers that have the wider stem waveguides and no auxiliary waveguide are employed.

Asymmetric Y-couplers that have the wider stem waveguides and no auxiliary waveguide are also usefu] for tapping power fro a main optical highway.

plurality of such taps may be employed optically in cascade along such an optical highway. Figure 7 is a schematic diagram showing how power may be tapped from a bidirectional highway 70. A fraction of the optical power propagating along the highway in the direction indicated by arrow 71 is tapped off by an asymmetric Y-coupler 72 to emerge from output port 73. Similarly a fraction of the optical power propagating along the highway in the direction of arrow 74 is tapped off in a asymmetric Y-coupler 75 to emerge from output port 76.

In a power tapping application there is clearly no need for a Y-coupler to tap half the power from the main highway, or for the angled arm to be necessarily inclined at 45 to the through arm. Generally the mirror, and the through arm and angled arm waveguides, will be dimensioned so that the tapping fraction is much less than half.

The waveguides of all the specific embodiments of the invention described with reference to the accompanying drawings are ridge waveguides, but it is to be clearly understood that the invention cam be embodied using other types of waveguide structure such as, for instance, buried heterostructure waveguides.

Claims (16)

CLAIMS.

1. An integrated optics asymmetric Y-coupler in which first, second and third optical waveguides are arranged in relation to a mirror such that light propagating in the first waveguide toward the mirror is partially intercepted by the mirror such that a portion of the light is reflected by the mirror into the second waveguide while substantially all the rest of the light is not intercepted by the mirror but is launched from the first waveguide, without reflection, into the third waveguide.

2. An integrated optics asymmetric Y-coupler as claimed in claim 1, wherein the mirror and the second waveguide are disposed on opposite sides of the third waveguide.

3. An integrated optics asymmetric Y-coupler as claimed in claim 1 or 2, wherein the first and third waveguides are collinear, and the second waveguide is inclined at substantially 450 to the third waveguide.

4. An integrated optics asymmetric Y-coupler as claimed in claim 1, 2 or 3, wherein said portion of the light reflected by the mirror is substantially 508.

5. An integrated optics asymmetric Y-coupler as claimed in any preceding claim wherein said first, second and third optical waveguides are single mode waveguides.

6. An integrated optics asymmetric Y-coupler as claimed in any claim of claims 1 to 4, wherein said first, second and third optical waveguides are waveguides that guide only zero and first order modes.

7. An integrated optics asymmetric Y-coupler as claimed in any claim of claims 1 to 4, wherein said second and third optical waveguides are single mode waveguides, wherein said first waveguide consists of or includes a length of laterally coupled twin waveguide forming a four-port 3dB splitter, two of whose ports are coupled substantially exclusively with the second. a.'.

third waveguides respectively.

8. An integrated optics asymmetric Y-coupler as claimed in any preceding claim, wherein included in the second waveguide is a portion provided with means for rendering its optical attenuation electrically controllable.

9. An integrated optics asymmetric Y-coupler as claimed in any preceding claim, wherein included in the third waveguide is a portion provided with means for rendering its optical attenuation electrically controllable.

10. An integrated optics asymmetric Y-coupler as claimed in claim 8 or 9, wherein the or each said electrically controllable optical attenuation portion whose optical attenuation is electrically controllable is a portion that is capable of being rendered optically amplifying.

11. An integrated optics asymmetric Y-coupler substantially as hereinbefore described with reference to Figures 2, 3, 4 and 5 or 2, 3, 4, 5 and G of the accompanying drawings.

12. An integrated optics device including two or more asymmetric Y-couplers as claimed in any preceding claim connected optically in cascade.

13. An integrated optics device as claimed in claim 11, which device is an optical matrix switch.

14. An optical network including, connected optically in cascade, two or more integrated optics devices, each including one or more integrated optics asymmetric Y-couplers as claimed in any claim of claims 1 to 12.

15. A monolithically integrated optical waveguide 2x2 matrix switch having first and second input waveguide ports optically coupled with first and second output waveguide ports via gated rectilinear waveguide sections and only four reflectors.

16. A method of making an asymmetrice' Y-cop1err as claimed in any claim of claims 1 to 12, wherein the waveguides are ridge waveguides formed by photolithography, wherein a first photolithographic etching stage is employed to create the ridge waveguides and a first part of the or each mirror, and then a second photolithographic etching stage is employed to create the remainder of each mirror using residual mask material of the first photolithographic etching stage for self alignment of the remainder part of the or each mirror with its corresponding first part.