GB2370884A - Optical waveguide - Google Patents

Abstract

A method of forming an optical waveguide which tapers to a point at each end comprises the steps of:<BR> ```defining a first shape (of a photoresist) (8, Fig. 4A) having first and second parallel sides;<BR> ```defining a second shape (of a photoresist) (7, Fig. 4D) having third and fourth parallel sides orientated such that the third and fourth sides each intersect the first and second sides of the first shape (8) at an acute angle; and<BR> ```using the shape of the area where the first and second shapes overlap to define a mask of SiO<SB>2</SB> (6B, Fig. 4F) which is used to form an optical waveguide (1A, Fig. 4H) with an elongate parallelogram shape. The tapered waveguide (1A) may be formed of active material over and optically coupled with a passive waveguide (2A).

Description

1 2370884

OPTICAL WAVEGUIDE

This invention relates to a method of forming an optical waveguide which tapers to a point at each end and to waveguides formed thereby, in particular waveguides forming part of a semiconductor optical amplifier.

Such components are typically used in optical communications networks operating around 1.3-1.6 microns and semiconductor lasers and amplifiers in III-V materials and may be used in hybrid integrated planar lightwave circuits for amplification and wavelength conversion of optical signals.

Semiconductor optical amplifiers (SOA) comprising an active waveguide tapered at each end are described in US 5844929 and US5278926. These each describe a semiconductor optical device, for example a laser, having a composite optical waveguide comprising a tapered active ridge waveguide in optical contact with a substantially planar, passive waveguide. The fundamental optical mode supported by the composite waveguide varies along the length of the composite waveguide so that, in a laser, the laser mode is enlarged at the output of the device and is a better match to a single mode optical fibre.

The width of the active waveguide decreases at each end thereof to couple a narrow optical mode which is amplified and guided by the active waveguide to a wide optical mode which is guided by the passive waveguide and which has a mode size greater than that of the narrow mode (see Figure 1A). Such devices find application in the fabrication of optical transmitters for optical fibre telecommunication networks.

The active ridge waveguide is typically defined by lithography and wet etching. The lithography has to be of high precision to provide a sharp end to the ridge. In addition, the ridge etching has to be of high precision to copy the mask into the semiconductor structure. Typically, contact optical lithography can routinely provide an end dimension WAR (see Figure 3) to the ridge mask of 0.7 microns. The required shape is formed in photoresist and then etched into a semiconductor layer in a single lithographic step. The width of the shape may then be narrowed by wet etching to under-cut either the resist or a mask. However this process is not easily reproducible

and neither the active ridge width WA (see Figure 3) nor the end width WAS can be easily controlled within a production process.

W097/36353 discloses a method of manufacturing a tapered waveguide region using two masks each of which is shaped like a rectangular trapezoid one side of which is chamfered or inclined with respect to the opposite side, each mask being a mirror image of the other. Each mask defines one side of the waveguide and one side of the taper and the two masks are used successively in overlaid positions to define a symmetrical taper at one end of the waveguide. The width of the waveguide and the lateral and longitudinal positions of the taper are thus subject to any inaccuracy in alignment of the two successive masks.

US5574742 discloses a tapered beam expander waveguide in which a symmetrical taper is formed at one end of the waveguide by two sequential masking and etching processes, each step again defining one side of the waveguide and one side of the taper so the width of the waveguide and the position of the taper are subject to errors arising from alignment of the two masks. The second masking step is also carried out over a physical step approximately one micron high created by the first etching step which leads to distortion in the shape created by the second etching step.

Some areas of the device are also subject to both etch steps which may give rise to different properties in these areas. The sidewalls are described as intersecting to define a pointed edge having a radius of curvature of less than 500 Angstroms (0.05 microns). The present invention provides a method of making a waveguide tapered at both ends with both tapers terminating at a sharp point.

According to a first aspect of the invention there is provided a method of forming an optical waveguide which tapers to a point at each end comprising the steps of: defining a first elongate shape having first and second parallel sides;

defining a second shape having third and fourth parallel sides orientated such that the third and fourth sides each intersect the first and second sides of the first shape at an acute angle; and using the shape of the area where the first and second shapes overlap to define an optical waveguide.

According to another aspect of the invention, there is provided an optical waveguide which tapers to a point at each end having an elongate parallelogram shape the adjacent sides of which are acutely angled to each other to form said tapered points.

Preferred and optional features of the invention will be apparent from the following description and from the subsidiary claims of the specification.

The term optically active material is used herein to refer to material which provides optical gain.

The invention will now be further described, merely, by way of example, with reference to the accompanying drawings, in which: Figure 1A is a schematic plan view of a known form of optical amplifier comprising an active waveguide tapered at both ends over a passive waveguide. Figures 1B and 1C are cross-sections of Figure 1A on lines B-B and C-C and Figure 1D shows a cross-section on line C-C of the final device structure; Figures 2A - 2D correspond to Figures 1A - 1D for a modified version of the device shown in Figure 1; Figure 3 is an enlarged view of part of Figure 2A showing dimensions thereof; Figures 4A - 4H are cross-sectional views of a device fabricated by a preferred form of a method according to the first aspect of the invention illustrating different stages in the method;

Figures 5A - 5C are plan views of the device illustrating stages of the method; Figure 6 is a schematic plan view of a first embodiment of an optical waveguide according to the second aspect of the invention; and Figure 7 is a schematic plan view of a second embodiment of an optical waveguide according to the second aspect of the invention.

A method of forming an optical waveguide which tapers at both ends will be described in relation to mode expanded SOAs in which an active waveguide 1 forming a confinement ridge is provided over a passive waveguide 2 supported on a substrate 3 as shown on Figures 1 and 2. As shown in these Figures, the passive waveguide 2 comprises a ridge of passive material provided on the substrate 3 and the active waveguide 1 comprises a narrower ridge of active material formed on the passive waveguide 2. Once these ridges have been formed, further semiconductor layers 4 and 5 are grown on the ridge structure to provide lower refractive index optical cladding layers as shown in Figure 1 D. The ends of the active waveguide 1 stop short of the ends of the passive waveguide 2. At the tapered ends of the active waveguide 1, the optical mode couples into the underlying passive waveguide 2. The optical mode associated with the passive guide 2 is considerably expanded compared to the mode in the active waveguide 1 for ease of coupling into waveguides formed in lower index materials or optical fibres.

However, coupling losses occur at the ends of the active waveguide 1 due to the shape of the tapered ends of the waveguide 1 which, as discussed above, are insufficiently sharp due to limitations in known production processes using a single mask to define the active waveguide rather than terminating at a sharp point.

Figure 2 shows a similar device in which the optical axes of the active and passive waveguides are inclined to the longitudinal axis of the substrate 3 so the end faces of the passive waveguide are not perpendicular to the optic axis thereof. This helps suppress problems caused by back reflections at these interfaces.

s A preferred form of a method according to a first aspect of the invention is illustrated in Figures 4 and 5. In this method, a mask of the required shape is formed by first forming a silicon dioxide mask 6 with an elongate shape and parallel sides and then modifying this by forming a parallel-sided photoresist mask 7 across the parallel-

sided oxide mask 6 at an acute angle to the length of the oxide mask 6. Parts of the oxide mask 6 not covered by the photoresist mask 7 are then removed leaving an elongate, parallelogram-shaped mask of silicon dioxide. This will be further explained with reference to Figures 4 and 5.

Figures 4A - 4H are cross-sections taken on line C-C of Figure 5 Figures 4A - 4C shows a cross-section of a semiconductor chip comprising a substrate 3, eg of InP, a passive waveguide layer 2, eg of InGaAsP Q1.1, um, on the substrate 3 and an active waveguide layer 1, eg of InGaAsP Q1. 55pm, over the passive waveguide layer 2. The active and passive waveguide layers 1 and 2 each having cladding layers (not shown) of lower refractive index, such an InP.

A layer of SiO2 6 having a thickness in the range 30 to 300 nanometers is formed over the active waveguide layer 1 and a first photoresist mask 8 having parallel sides is formed thereon as shown in Fig 4A. Areas of the oxide layer 6 not covered by the mask 8 are then removed as shown in Fig 4B. The first photoresist mask 8 is then removed leaving an oxide layer 6A having an elongate shape of width WA (see Fig 3) with parallel sides on the active waveguide layer 1, as shown in Figure 4C. The width WA IS preferably in the range 5 microns > WA 0.5 microns. Figure 5A shows a plan view of the device at this stage. As shown in Figure 5A, the longitudinal axis of the oxide layer 6A is inclined to the longitudinal axis of the substrate 3.

A second photoresist mask 7 with a width wider than WA having parallel sides is then formed over the oxide layer 6A as shown in Figures 4D and Figure 5B. As shown in Figure 5B, the parallel sides of the mask 7 intersect the parallel sides of the oxide layer 6A at an acute angle. Areas of the oxide layer 6A not covered by the mask 7 are then removed as shown in Figure 4E. The photoresist mask 7 is then removed to leave an oxide layer 6B having an elongate parallelogram shape on the active waveguide layer 1, as shown in Figures 4F and 5C.

Thus, it will be appreciated that the shape of the oxide layer 6B is defined by the shape of the area where the oxide layer 6A shown in Fig 5A overlaps with the shape of the photoresist mask 7 shown in Figure 5B.

The oxide layer 6B is then used as a mask to accurately define a corresponding shape in the active waveguide layer 1. Parts of the active waveguide layer 1 not covered by the oxide layer 6B are removed by a dry etching process as shown in Figure 4G and the oxide layer 6B is then removed to leave an active waveguide 1A having the shape of the parallelogram shown in Figure 5C, as shown in Figure 4H.

Figure 6 shows a plan view of the active waveguide 1A. The waveguide 1A thus comprises an elongate ridge waveguide which tapers to a point at each end formed over the passive waveguide layer 2.

The oxide layer 6B used as a mask preferably has a thickness of < 200nm and most preferably < 100nm. This compares to a typical photoresist thickness of >400nm and to a typical ridge height of 1000nm. The photoresist mask 7 applied over the oxide layer 6A is thus applied over only a small step between the oxide layer 6A and the active waveguide layer 1 which is advantageous in maintaining accurate definition of the shape of the photoresist mask. This step height is significantly smaller than the typical ridge height of an active waveguide over which a mask has to be formed if a two-stage masking and two-stage etching process is used.

A dry etch, eg using reactive ion etching (RIE), electron cyclotron reactive ion etching (ECR), inductively coupled plasma reactive ion etching (ICP) or an Ar sputter etch, will accurately copy the mask formed by the oxide layer 6B into the semiconductor layer 1 below. The ICP and ECR methods are also advantageous owing to reduced damage to the semiconductor surfaces.

Due to its elongate from, the parallelogram shape layer of active material forms an elongate ridge waveguide 1A the ends of which taper to points, although the tapers are not symmetrical about the optical axis of the waveguide. As the tapers are formed by the fabrication, at two separate stages, of straight edges which intersect

with each other, fabrication of the points at the ends thereof are not subject to the limitations inherent in some of the prior art processes.

A value of WAE smaller that 0.2 microns is desirable and this process enables this to be achieved without the use of under-cut etching, E-beam or X-ray lithography.

Furthermore, the above process simultaneously forms tapers at each end of the waveguide 1A as the masking process described above defines both ends of the waveguide 1A in the same operation. Separate definition of each end is not therefore required.

The process also forms an active waveguide of accurately known width WA as this is defined by the width of the first photoresist mask 8. The length of the waveguide between the tapered ends is also accurately known as this is determined by the acute angle at which the parallel sides of the mask 7 intersect the parallel sides of the oxide layer 6A and the lateral position of the tapered points are known as these aligned with the respective sides of the parallelogram shape formed. Thus, the only parameter not accurately defined by the process is the longitudinal position of the active waveguide 1A relative to the passive waveguide 2 but, as described below, this does not give rise to problems. i A ridge waveguide 2A is defined in the passive waveguide layer 2 as shown in Figure 6. As shown in this Figure, the optical axis of the passive ridge waveguide 2A is inclined to the longitudinal axis of the substrate 3. This helps reduce problems due to back reflections at the ends of the passive waveguide 2A. The definition of the passive ridge waveguide 2A may be done before or after the definition of the active ridge waveguide 1A.

Following formation of the active and passive ridge waveguides, further semiconductor layers are formed over the ridge structure so it is buried in semiconductor material of similar refractive index in a similar manner to that shown in Figure 1 D.

It may be thought that the non-symmetrical shape of a waveguide 1A such as that shown in Figure 6 may present difficulties in the design and fabrication of an ideal SOA structure due to the tapers being at an angle to the parallel section of the active waveguide 1A and because the taper bisector is not parallel to the optical axis of the passive guide 2A. Furthermore, whilst the length of the active ridge 1A is well defined by the angle between the second mask 7 and the ridge waveguide 1A, the positions of the ends of the tapers relative to the passive waveguide 2 are not well defined as this is subject to errors in registering the second mask 7 to the first mask 8. However, these difficulties can be minimised or tolerated as explained below.

Firstly, the parallelogram shaped area described above can be fabricated so that a line passing through the points at each end thereof lies substantially parallel to the axis of the passive waveguide 2A as shown in Figure 6. This requires the parallel sided portion of the active waveguide 1A from which the parallelogram shape is formed to lie at a small angle to the optical axis of the passive waveguide 2A.

Secondly, to minimise losses in coupling light from the active waveguide 1A to the passive waveguide 2A, the bisectors of the tapers at each end of the active waveguide 1A should lie substantially on the optical axis of the passive waveguide 2A. In order to satisfy this requirement at each end of the active waveguide 1A, a small offset is provided in the passive waveguide 2A part-way along its length as shown in Figure 7. This Figure shows an abrupt offset but a gradual offset may be used instead, eg of sigmoid shape. The optical axis at one end of the passive waveguide 2A IS thus offset with respect to the optical axis at the other end of the passive waveguide 2A.

A typical mode expanded or large spot LS-SOA may has the following dimensions (with reference to Figure 3): WA IS in the range 0.5-5.0 microns, eg 1.4 microns WAE IS 0.2 microns or less Wp is in the range 310 microns, eg 5 microns LT IS in the range 100-800 microns, eg 500 microns LA-2LT is in the range 300-2000 microns, eg 600 microns

Lp is typically around 2000 microns so Lp-LA-2LT is typically in the range 50 500 microns Op is in the range 3-15 degrees, eg 7 degrees where WA IS the width of the active waveguide, WAE IS the width of the point of the tapered portions of the active waveguide, We is the width of the passive waveguide, LT IS the length of the tapered portion at each end of the active waveguide, LA-2LT is the length of the parallel sided portion of the active waveguide, Lo is the length of the passive waveguide, Lp-LA2LT is the difference between the lengths of the active and passive waveguides and Hp is the angle between the optical axes of the active and passive waveguides and the longitudinal axis of the substrate.

From these parameters it is clear that the angles of the taper and the angle of the passive guide 2A relative to the active waveguide 1A are small, eg c 2.8 degrees and preferably c 0.6 degrees.

The offset along the length of the passive waveguide 2A is given by: Offset WA(LA/2LT - 1) which, using the dimensions given above, is 2.1 microns for a 200 micron taper and 0.84 microns for a 500 micron taper, i. e. 42% and 17% of a 5 micron passive guide width, respectively. This has little effect on the operation of the device as, in the central section, the optical overlap between the active and passive waveguides 1A and 2A is at a minimum.

The effect of alignment is determined by the geometry described. Typical figures for alignment accuracy are: X = +/- 0.5 microns Y = +/- 0.5 microns 0= +/-1 e-5 Rad, 1.1 E-3 degrees

This rotational accuracy represents 0.5 microns across a 2 inch (5 cm) wafer and therefore 20 rim along the length of a device and can be ignored when compared to the Y accuracy. The error in the position of the end of the taper is: Xerror=Y/WAxLT which is +/-71.4 microns and +/-178 microns for LT = 200 microns and 500 microns, respectively. The passive waveguide 2A preferably extends beyond the ends of the active waveguide 1 by a distance of at least 50-500 microns, and preferably 200-300 microns, to ensure that the optical mode is stabilised before being transmitted to a further waveguide or optical fibre.

Variation in the length of the passive waveguide 2A section is not of great significance to the operation of the device. This is likely to result in a change in insertion loss which is very small compared to the 0.5 - 1.0 dB insertion loss caused by WEA=0.2 microns in the prior art

Furthermore, it is possible to cleave the device such that the passive waveguide length is known by using optical inspection. This is also desirable to avoid the device being longer than necessary and so help minimise the size of the device. A typical cleaving tolerance to a marker on a substrate is +/-5 microns. The passive waveguide loss is c 0.5 dB/cm.

The position of the offset in the passive waveguide 2A relative to the length of the active waveguide 1A also has an identical error. This is not significant as this section of the passive waveguide 2A has very little influence on the propagation as the optical mode is in the active waveguide 1A at this position. As the length of the active waveguide 1A is long compared to the tapers, the error in position is <10% from the mid position.

The method described above thus modifies the shape of the ends of the active waveguide 1A to reduce the coupling losses with the passive waveguide 2A as a reduction in the dimension WAE reduces problems due to reflection and scattering of light at the point. The method is simple but provides a significant improvement in the sharpness of the ends of the tapers formed. As discussed above, the method enables the width of the active waveguide to be accurately defined whilst using a two step process to define sharp tapers at each end thereof without compromising the performance of the device. The method is particularly suited to the formation of an active waveguide on a passive waveguide where the passive waveguide extends beyond each end of the active waveguide and accurate positioning of the active waveguide along the length of the optical axis of the passive waveguide is not required, such as in an SOA.

The example described above uses a silicon dioxide mask to form the tapered waveguide. However, other types of mask may be used, eg silicon dioxide, silicon nitride, aluminium oxide and tungsten suicide. Other masking arrangements in which two shapes each having parallel sides are overlaid to define an elongate parallelogram can also be envisaged.

Although the preferred method is as described above, the parallelogram shaped waveguide can also be formed in other ways, eg a first parallel sided mask may first be used to form a parallel sided ridge waveguide and a second parallel sided mask then formed at an angle across the ridge waveguide to define the parallelogram shape. In a further alternative, the parallelogram shape may be first defined in a layer of photoresist by a two exposure process, a first exposure defining a first parallel sided shape and a second exposure defining a second parallel sided shape lying across the first shape at an angle so that the area of photoresist subject to a double exposure defines the required parallelogram shape.

In the methods described, the masks can also be orientated such that the parallel sides thereof may be written as true, continuous straight lines rather than as a series of stepped lines as used in some of the prior art.

The method reduces the insertion loss of a semiconductor optical amplifier (SOA) and so improves the overall gain and more importantly the noise performance of the SOA giving a noise advantage in systems where multiple SOAs are cascaded.

Claims (27)

1. Method of forming an optical waveguide which tapers to a point at each end comprising the steps of: defining a first elongate shape having first and second parallel sides; defining a second shape having third and fourth parallel sides orientated such that the third and fourth sides each intersect the first and second sides of the first shape at an acute angle; and using the shape of the area where the first and second shapes overlap to define an optical waveguide.

2. Method as claimed in claim 1 in which the step of defining the first elongate shape comprises forming a mask of the first elongate shape.

3. A method as claimed in claim 2 in which the second shape is used to modify said mask by removing parts thereof where the first and second shapes do not overlap.

4. A method as claimed in claim 2 or 3 in which the mask is formed of silicon dioxide.

5. A method as claimed in claim 3 in which the second shape is formed from photo resist.

6. A method as claimed in claim 4 in which the mask has a thickness of 300 nm or less, and preferably 100 rim or less.

7. A method as claimed in claim 3 in which the modified mask is formed on a waveguide layer and parts of a waveguide layer not covered by the modified mask are removed.

8. A method as claimed in claim 7 in which said parts are removed by a dry etch process.

9. A method as claimed in any preceding claim in which the optical waveguide is formed of an optically active material so as to form an active waveguide which tapers to a point at each end thereof.

10. A method as claimed in claim 9 in which the active waveguide is a ridge waveguide, which is buried within further semiconductor layers.

11. A method as claimed in claim 5 in which the active waveguide is formed over a passive waveguide, the active and passive waveguides being optically coupled.

12. A method as claimed in claim 6 in which the passive waveguide is formed so as to extend beyond each end of the active waveguide.

13. A method as claimed in any preceding claim in which the acute angle is 2.8 degrees or less, or preferably 0.6 degrees or less.

14. A method of forming an optical waveguide substantially as hereinbefore described with reference to Figures 4 and 5 of the accompanying drawings.

15. An optical waveguide which tapers to a point at each end having an elongate parallelogram shape the adjacent sides of which are acutely angled to each other to form said tapered points.

16. An optical waveguide as claimed in claim 15 in which the acute angle is 2.8 degrees or less, or preferably 0.6 degrees or less.

17. An optical waveguide as claimed in claim 15 or 16 in which the point of each taper has a width of 0.2 microns or less.

18. An optical waveguide as claimed in claim 15,16 or 17 in which each taper has a length in the range 100 - 800 microns, and preferably in the range 200 - 500 microns.

19. An optical waveguide as claimed in any of claims 15 - 18 in which a parallel-

sided portion of the waveguide between the tapered ends has a width in the range 0.5 - 5.0 microns.

20. An optical waveguide as claimed in any of claims 15 - 19 in which the waveguide is formed over and optically coupled with a passive waveguide.

21. An optical waveguide as claimed in any of claims 15 - 20 in which a line through the points of each taper lies substantially parallel to the optical axis of the passive waveguide.

22. An optical waveguide as claimed in claim 21 in which the optical axis at one end of the passive waveguide is offset with respect to the optical axis thereof at the other end so that lines bisecting the tapered point at each end of the tapered waveguide lie substantially on the optical axis of the passive waveguide at each end thereof.

23. An optical waveguide as claimed in any of claims 20 - 22 in which the passive waveguide extends beyond each end of the tapered waveguide.

24. An optical waveguide as claimed in any of claims 20 - 23 in which the passive waveguide extends beyond each end of the tapered waveguide by a distance in the range 50 - 500 microns, and preferably in the range 200 300 microns.

25. An optical waveguide as claimed in any of claims 15 - 24 in which the tapered waveguide is formed of an optically active material.

26. An optical waveguide as claimed in claims 25 in which the tapered waveguide forms part of a semiconductor optical amplifier.

27. An optical waveguide substantially as hereinbefore described with reference to and/or as shown in Figures 6 and 7 of the accompanying drawings.