WO2008032023A1

WO2008032023A1 - Optoelectronic device

Info

Publication number: WO2008032023A1
Application number: PCT/GB2007/003376
Authority: WO
Inventors: Andrew Knight; Graham Reed; David John Thomson
Original assignee: University Of Surrey; Mcmaster University
Priority date: 2006-09-11
Filing date: 2007-09-10
Publication date: 2008-03-20

Abstract

An optical switch (50) has a Silicon On Insulator (SOI) body (51) incorporating two optical waveguides (52, 53). The waveguides (52, 53) are arranged so that they cross one another to form a crossover (55). A carrier restrictive region (56) is provided in the Si layer of the SOI body (51), passing through the crossover (55). The carrier restrictive region (56) comprises a region of the Si layer in which Si ions have been implanted at sufficient energy and dose to cause at least partial amorphisation. The carrier restrictive region (56) is also doped with an n-type dopant, whereas the background doping of the SOI body (51) is p-type. Movement of free carriers between the SOI body (51) and the carrier restrictive region (56) is therefore restricted and part of one side (57) of the carrier restrictive region (56) defines a boundary in the crossover (55). Electrodes (59, 60, 64) are provided on the upper surface (61) of the SOI body (51), in contact with n-type doped regions (62, 63) and a p-type doped region (65) forming a p-i-n diode through the crossover (55) on one side of the boundary. This allows free carriers to be injected into the crossover (55) to alter the refractive index of the crossover (55) on that side of the boundary. Light propagating along an input path of the first waveguide (52) can be controllably reflected into the second waveguide (53) at the boundary in the crossover (55) when the refractive index of the crossover (55) differs on either side of the boundary.

Description

OPTOELECTRONIC DEVICE

Field of the Invention

This invention relates to an optoelectronic device and to methods of use and manufacture of the same. A particular application of the invention is an optical switch. Other applications of the invention include optical modulators, filters and attenuators.

Background to the Invention

Many existing optoelectronic devices are fabricated from semiconductors that exhibit a strong linear electro-optic effect, known as the Pockels effect, by which the refractive index of a semiconductor changes linearly with the strength of an electric field applied across the semiconductor. Semiconductors exhibiting the Pockels effect all have crystal structures that lack something known as "inversion symmetry". An example is gallium arsenide (GaAs). However, many semiconductors, importantly including Silicon (Si), do not exhibit the Pockels effect. This is unfortunate, as it is otherwise desirable to fabricate optoelectronic devices from such semiconductors. For example, Si is the semiconductor of choice for many existing semiconductor devices and the ability to fabricate optoelectronic devices from Si allows them to be integrated with these existing semiconductor devices. Similarly, semiconductor manufacturing techniques are generally better developed and more readily available for Si than many semiconductors that exhibit the Pockels effect.

Fortunately, some semiconductors that do not exhibit the Pockels effect, including Si, exhibit something known as the "Plasma Dispersion" effect, by which the refractive index of a semiconductor changes with the density of free carriers present in the semiconductor. This opens the possibility of fabricating optoelectronic devices from these semiconductors and, in particular, integrating optoelectronic devices with a variety of conventional semiconductor devices. Much current research is therefore directed at developing optoelectronic devices that use the Plasma Dispersion effect.

For example, referring to figures 1 and 2, an optical switch 10 of the prior art has a Si body 11 incorporating a light guiding layer that provides two optical waveguides 12, 13. The waveguides 12, 13 are arranged to cross one another to form a crossover 14. On an input side of the crossover 14, a first of the waveguides 12, extending from left to right into the page as shown in figure 1 , provides a light input path. The corresponding portion of the second of the waveguides 13, on the input side of the crossover 14, is not used in this example, although it also effectively provides a light input path, which can be used if desired. On an output side of the crossover 14, the first and second waveguides 12, 13 each provide a light output path. So, the switch 10 has the configuration of a 2x2 optical crossover switch, although only one input is normally used.

As can be seen more clearly in figure 2, which is a cross-sectional schematic view of the optical switch 10 along the line A-A shown in figure 1 , a first electrode 15 is provided on top of the crossover 14 in contact with a p- type doped region 16 in a top portion of the crossover 14. The first electrode 15 and p-type doped region 16 are elongate and extend approximately along the middle of the crossover 14 basically parallel to a direction from the input side to the output side of the switch 10. A second electrode 17 is provided on a top surface 18 of the Si body 11 to a side of the crossover 14 between the output path of the first waveguide 12 and the (unused) input path of the second waveguide 13 (the right hand side as shown in figure 1 ) in contact with an n-type doped region 19 in a top portion of the Si body 11. The doped regions 16, 19 are close to, but not in contact with, one another and so form a p-i-n diode through the crossover 14. When an electric potential is applied between the electrodes 15, 17, there is a change in the density of free carriers (e.g. electrons and holes) in a volume of the crossover 14 between the doped regions 16, 19. This causes the refractive index of that volume of the crossover 14 to change, as a result of the Plasma Dispersion effect. This is described in detail in the paper "Electro-optical Effects In Silicon", Soref et al, IEEE Journal of Quantum Electronics, volume 23, pages 123-129, 1987. For example, as can be seen from figure 3, the change in refractive index of Si for a given change in the density of free carriers for light at a wavelength of 1.55μm varies approximately linearly from around -0.002 to around -0.02, as the change in density of free carriers increases from around 1x10¹⁸ cm^'1 to around 1x10¹⁹ cm^*1.

In the absence of an electric potential applied between the electrodes 15, 17, the refractive index of the crossover 14 is uniform. This means that light propagating along the input path of the first waveguide 12 can normally pass straight through the crossover 14 and continue along the output path of that waveguide 12. However, when the refractive index of the volume of the crossover 14 is changed by application of an electric potential between the electrodes 15, 17, the refractive index of the crossover 14 varies across the width of the crossover or, more specifically, is different on either side of a boundary of the volume. The electrodes 15, 17 and p-type and n-type doped regions 16, 19 are arranged such that this boundary is roughly at a plane extending downwardly into the crossover 14 from an edge of the p- type doped region 16 on top of the crossover 14 furthest from the second electrode 17 and n-type doped region 19. Ideally, the boundary is a plane of symmetry of the X-shape formed by the waveguides 12, 13, which plane is defined by a first line basically parallel to a direction from the input side to the output side of the switch 10 and a second line perpendicular to the directions in which light propagates in each of the waveguides 12, 13. This plane is illustrated in cross section by the line B-B in figure 2. So, light propagating along the input path of the first waveguide 12, can be reflected at the boundary in the crossover 14 such that it is diverted down the output path of the second waveguide 13. By selectively applying the electric potential between the electrodes 15, 17, light can be selectively reflected and hence switched between the output paths of the waveguides 12, 13.

A significant problem with this optical switch 10 is that the boundary of the volume of the crossover 14 across which the refractive index changes and at which the reflection takes place is not reliably well defined. Although the free carriers move- largely between the electrodes 15, 17, they also have

( a tendency to spread out in other directions to some extent. This means that the free carrier density, and hence the refractive index of the crossover 14, varies gradually across the boundary. More specifically, the refractive index of the crossover 14 tends to decrease gradually across the boundary when the electric potential is applied between the electrodes, as shown in figure 4, rather than with a sharply defined step. So, the reflection at the boundary tends to be inefficient.

This problem is recognised in the paper "Silicon-On-lnsulator Asymmetric Optical Switch Based On Total Internal Reflection", Zhao et al, IEEE Photonics Technology News, Volume 9, No. 8, August 1997. In this paper, it is suggested that careful selection of the positioning and dimensions of the electrodes and doped regions can cause the electric field lines between the two doped regions to be squeezed at the boundary and hence sharpen the change in free carrier density at the boundary. However, the carriers are still able to diffuse laterally in the crossover and any step in the refractive index within the crossover is therefore very likely to be gradual, meaning that efficient reflection is unlikely to be achieved.

The problem is also recognised in the paper "InGaAsP/lnP Optical Switches Using Carrier Induced Refractive Index Change", lshida et al, Applied Physics Letters, 50(3), pages 141 to 142, 1987. In this paper, it is proposed to diffuse Zinc (Zn) into regions above and below a crossover to form electrodes. Two Zn electrodes below the crossover are immediately offset from a central Zn electrode above the waveguide. This apparently achieves a carrier restricting structure and a corresponding abrupt change in the density of free carriers across a boundary. However, the carriers are again still able to diffuse laterally in the crossover, despite the so-called carrier restrictive structure described in this paper. The change in refractive index across the boundary at which the reflection takes place is still therefore very likely to be gradual and efficient reflection is unlikely to be achieved.

The present invention seeks to overcome this problem.

Summary of the Invention

According to a first aspect of the present invention, there is provided an optoelectronic device comprising an optical waveguide disposed in a semiconductor, wherein the refractive index of a volume of the waveguide can be altered under the influence of electric potential to cause light propagating along the waveguide to be selectively reflected at a boundary of the volume to generate an output, the boundary being defined by a variation in the density of defects and/or a dopant in the semiconductor. The refractive index of the volume of the semiconductor can be altered under the influence of electric potential by the Plasma Dispersion effect. In other words, the refractive index can change with changes in the density of free carriers (e.g. electrons and holes) in the volume of the waveguide. By defining the boundary of the volume with a variation in density of defects and/or a dopant in the semiconductor, diffusion of the free carriers can be restricted at the boundary. In particular, the dopant can allow the creation of a depletion region that restricts the diffusion of carriers at the boundary. Similarly, the presence of defects alone restricts carrier diffusion due to reduced lifetimes where the defects are present. So, the resulting step in free carrier density at the boundary can provide a sharp change in refractive index across the boundary. This can in turn provide efficient reflection at the boundary.

Furthermore, as diffusion of the free carriers is restricted at the boundary, the carriers can be retained in the volume and a smaller overall number of carriers may need to be injected into the volume to attain a given change in refractive index. This can improve the speed with which the refractive index can be changed and reduce the required current. In other words, faster switching speeds and lower switching currents can be attained. The boundary usually coincides with the variation, e.g. is at the variation. However, it is possible that the boundary is slightly offset from the variation, e.g. when a depletion region causes the restriction of free carriers at the boundary. In most applications of the invention, the variation is arranged to define the boundary across substantially a whole transverse section of the waveguide. This allows the boundary to reflect light propagating in substantially any transverse portion of the waveguide.

The variation is usually arranged to define the boundary obliquely to the direction in which light propagates in the waveguide. This allows the boundary to reflect light in a different direction to the direction in which light propagates in the waveguide. For example, the boundary might reflect light into another waveguide arranged at a different angle to the waveguide in which the light is propagating.

Many useful applications of the invention involve the use of total internal reflection. This can only occur when light is incident on the boundary (with respect to a line normal to the boundary) at greater than a so-called critical angle. In other words, in order for light propagating in the waveguide to be totally internally reflected at the boundary, the boundary must be oriented to the direction in which the light propagates in the waveguide such that the light is incident on the boundary at greater than the critical angle. The value of the critical angle depends on the difference in refractive index across the boundary, which in turn usually depends on a change in the density of free carriers brought about by the applied electric potential. However, for any given semiconductor material the refractive index can only practically vary over a limited range of values by the Plasma Dispersion effect. Only a limited range of critical angles are therefore available, and hence only a limited range of orientations of the boundary may be suitable. So, in a preferred example of the invention, the variation is arranged to define the boundary oriented to the direction in which light propagates in the waveguide such that the light can be substantially totally internally reflected at the boundary. The variation can be arranged in a variety of ways. It is typically arranged to occur substantially in a plane, at least at the boundary, as a planar boundary tends to reflect the light most usefully for most applications. It is usually at the edge of a carrier restrictive region. There may be more than one carrier restrictive region. This can be useful to restrict the free carriers at more than one boundary. For example, the variation may confine the extent of the volume along a length of the waveguide. This might be both at the boundary discussed above and at another boundary at a different location along a length of the waveguide for example. In another example, the carrier restrictive region defines a perimeter in a plane defined by the length and width of the waveguide. In other words, the carrier restrictive region may define be a loop or perimeter. It can therefore be appreciated that whilst the variation defines the boundary, the carrier restrictive region, and indeed the variation, may extend beyond the boundary. This can allow it to confine the free carriers in other parts of the waveguide or, indeed, other parts of the semiconductor in which the waveguide is disposed.

The invention also provides a method of operating the optoelectronic device described above, the method comprising applying an electric potential to alter the refractive index of the volume.

In one example, the volume may have a refractive index the same as that of the waveguide outside the volume at the boundary. In other words, the refractive index of the waveguide and the carrier restrictive region may be inherently substantially the same. This means that light propagating in the waveguide is admitted at the boundary when the refractive index of the volume is unaltered by the electric potential. In other words, light propagating in the waveguide may be admitted at the boundary in the absence of applied electric potential. The method may then comprise applying an electric potential to alter the refractive index of the volume such that (more of) the light is reflected at the boundary. The reflection may be substantially total internal reflection or otherwise. In another example, the volume may have a refractive index different to that of the waveguide outside the volume at the boundary, such that light propagating in the waveguide is reflected at the boundary when the refractive index of the volume is unaltered by the electric potential. In other words, light propagating in the waveguide may be reflected at the boundary in the absence of applied electric potential. In particular, the light propagating in the waveguide may be substantially totally internally reflected at the boundary when the refractive index of the volume of the waveguide is unaltered by the electric potential. This can be achieved by the volume being doped such that it has a refractive index different to that of the waveguide outside the volume at the boundary. It can also be achieved by the density of defects outside of the volume, e.g. in the carrier restrictive region, causing an increase in refractive index relative to the volume. The method may then comprise applying an electric potential to alter the refractive index of the volume such that (more of) the light is admitted at the boundary.

In either of these examples, the reflection at the boundary can be such that one polarisation mode of the light propagating in the waveguide is reflected and another polarisation mode is admitted. This can be useful for separating the polarisation modes of the light propagating in the waveguide. Again, the reflection may be substantially total internal reflection or otherwise.

As alluded to above, the variation may define the boundary at a crossover between the waveguide and another waveguide. This can be useful when the optoelectronic device is an optical switch or such like. Light admitted at the boundary may continue along the waveguide and light reflected at the boundary may be diverted along an/the other waveguide.

According to a second aspect of the present invention, there is provided a method of manufacturing an optoelectronic device, the method comprising: forming a waveguide from a semiconductor; and providing defects and/or a dopant in portion of the semiconductor such that there is a variation in the density of defects and/or the dopant defining a boundary to a volume of the semiconductor, the refractive index of which volume can be changed under influence of electric potential to cause light propagating along the waveguide to be selectively reflected at the boundary to generate an output.

The optoelectronic device may be similar to that described above. Typically, the semiconductor from which the waveguide is formed is silicon. The variation might be step-like. In particular, the variation might be a bulk change from one density of defects/dopant to another density of defects/dopant, e.g. throughout the entire volume mentioned above. Alternatively, the variation might be an interim change in the density of defects/dopant (either an increase or a decrease) providing a wall in the semiconductor having a defect/dopant density different to that either side of the variation.

Preferred embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings.

Brief Description of the Drawings

Figure 1 is a schematic view of an optical switch according to the prior art;

Figure 2 is a cross-sectional schematic view of the optical switch of the prior art illustrated in figure 1 along the line A-A; Figure 3 is a graphical representation of change in refractive index of Si with change in free carrier density, as known in the prior art;

Figure 4 is a graphical representation of the refractive index of a crossover of the optical switch of the prior art illustrated in figures 1 and 2 along the line A-A in the presence of applied electric potential;

Figure 5 is a schematic view of an optical switch according to a first preferred embodiment of the invention;

Figure 6 is a cross-sectional schematic view of the optical switch illustrated in figure 5 along the line C-C; Figure 7 is a graphical representation of the refractive index of a crossover of the optical switch illustrated in figures 5 and 6 along the line C- C in the presence of applied electric potential;

Figure 8 is a schematic view of an optical switch of according to a third preferred embodiment of the invention; Figure 9 is a cross-sectional schematic view of the optical switch illustrated in figure 8 along the line E-E;

Figure 10 is a schematic view of an optical switch of according to a third preferred embodiment of the invention;

Figure 11 is a cross-sectional schematic view of the optical switch illustrated in figure 10 along the line G-G.

Detailed Description of the Preferred Embodiments

Referring to figures 5 and 6, an optical switch 50 according to a first preferred embodiment of the invention has a Silicon On Insulator (SOI) body 51. A silicon (Si) layer of the SOI body 51 incorporates a light guiding layer that provides two optical waveguides 52, 53. A silicon oxide (SiO₂) layer 54 of the SOI body 51 is disposed below the Si layer to provide part of a cladding for the waveguides 52, 53 and an electrically insulating base for the switch 50.

In this embodiment, the waveguides 52, 53 are rib waveguides. They are also monomode and are typically adapted to transmit light either around 1.3μm or around 1.55μm, which are the wavelengths most commonly used for optical communication. The waveguides 52, 53 are arranged to cross • one another to form a crossover 55. On an input side of the crossover 55, a first of the waveguides 52, extending from left to right into the page as shown in figure 5, provides a light input path. The corresponding portion of the second of the waveguides 53, on the input side of the crossover 55, is not used in this embodiment, although it also effectively provides a light input path, which can be used if desired. On an output side of the crossover 55, the first and second waveguides 52, 53 each provide a light output path. So, the switch 50 has the configuration of a 2x2 optical crossover switch, although only one input is normally used.

A carrier restrictive region 56 is provided in the Si layer of the SOI body 51. The carrier restrictive region 56 comprises a region of the Si body 51 in which Si ions have been implanted at sufficient energy and dose to cause at least partial amorphisation and Phosphorus (P) ions have been provided as a dopant. So, the carrier restrictive region 56 is an n-type doped region. This contrasts with the background doping of the Si layer of the SOI body 51 , which is p-type in this embodiment. The carrier restrictive region 56 comprises a perimeter in a plane defined by the lengths and widths of the waveguides 52, 53. The region 56 also extends (vertically) from the upper surface 61 of the SOI body to the SiO₂ base layer 54, so that it encloses a portion of the Si layer of the- SOI body 51 (although in other embodiments the region 56 can be shallower and not meet the SiO₂ base layer 54 if desired). Importantly, one side 57 of the region 56 runs along a plane that divides the crossover 55 in half. More specifically, it runs substantially along a plane of symmetry of an X-shape formed by the waveguides 52, 53, which plane is defined by a first line basically parallel to a direction from the input side to the output side of the switch 50 and a second line perpendicular to the directions in which light propagates in each of the waveguides 52, 53. This part of the one side 57 of the carrier restrictive region 56 defines a boundary in the crossover 55 at which the movement of free carriers is restricted. This boundary is illustrated by line D-D in Figure 6. The other side 58 of the region 56 also runs through the crossover 55 parallel to the plane, but offset toward the input path of the first waveguide 52 and the output path of the second waveguide 53, or to the left as shown in figure 5. So, one half of the crossover 55 is outside the region 56 and the other half of the crossover 55 includes the region 56. Where the region 56 crosses the waveguides 52, 53 away from the crossover 55, its sides 57, 58 are oriented perpendicularly to the direction of propagation of light in the respective waveguide 52, 53. This can help to minimise any unwanted affect of the region 56 on the propagation of light in the waveguides 52, 53. A pair of electrodes 59, 60 is provided on the upper surface 61 of the SOI body 51 , one electrode 59 of the pair on the input side of the crossover 55 and the other electrode 60 of the pair on output side of the crossover 55. Beneath each of the electrodes 59, 60 in the Si layer of the SOI body 51 are respective n-type doped regions 62, 63. On another side of the crossover 54, between the output path of the first waveguide 52 and the (unused) input path of the second waveguide 53 (e.g. on the right hand side of the switch. 50 as shown in figure 5), another electrode 64 is provided on the upper surface 61 of the SOI body 51. Beneath this other electrode 64 is a p-type doped region 65. The doped regions 62, 63, 65 are fairly close to one another, but not in contact with one another, and so form a p-i-n diode through the crossover 55. All of the electrodes 59, 60, 64 and doped regions 62, 63, 65 are within the perimeter formed by the carrier restrictive region 56. In addition, the electrode 64 in contact with the p-type doped region is connected to yet another electrode (not shown) in contact with the carrier restrictive region 56. This means that, when the p-i-n diode formed through the crossover 55 is forward biased, a reverse bias is applied between the carrier restrictive region 56 and the volume of the crossover 55 in which the p-i-n diode is formed.

Light propagating along the input path of the first waveguide 52 can be reflected in the crossover 55 at the boundary D-D when the refractive index of the crossover 55 differs on either side of the boundary. The switch 50 is ideally arranged such that substantially all the light propagating along the input path of the first waveguide 52 can be reflected down the output path of the second waveguide 53 by total internal reflection. This requires an angle of incidence of light propagating along the first waveguide 52 at the boundary (with respect to a line normal to boundary) to be greater than a so- called "critical angle" θ_c . Fundamentally, the value of the critical angle θ_c is dependent on the refractive index of the crossover 55 on each side of the boundary. Indeed,

sin 6>_c = -^ (1 )

where «, is the refractive index of the crossover 55 on the same side of the boundary as the input path of the first waveguide 52 and the carrier restrictive region 56, and n₂ is the refractive index of the crossover 55 on the other side of the boundary.

The refractive index n_x of the crossover 55 on the same side of the boundary as the input path of the first waveguide 52 is determined by the material from which the carrier restrictive region 56 is made. So, it is the natural refractive index of Si layer of the SOI body 51 taking into account the effects of the implanted Si ions and dopant P ions. As the amorphisation caused by the Si ions increases the refractive index of the Si layer and the presence of the dopant P ions decreases the refractive index of the Si layer, the effects can be balanced and the refractive index of the carrier restrictive region 56 is substantially the same as that of the Si layer of the SOI body 51. The refractive index n₂ of the volume of the crossover 55 on the other side of the boundary can be altered from the natural refractive index of Si layer by application of an electric potential between the electrodes 59, 60, 64 on either side of the p-i-n diode to cause a change in free carrier density. So, the critical angle θ_c can be varied within practical limits by applying electric potential between the electrodes 59, 60, 64. Generally, the smaller the critical angle θ_c , the shorter the length of the crossover 55 and hence the overall length of the switch 50 can be. As the length of the crossover 55 decreases, so generally does the distance between the electrodes 59, 60 in contact with the n-type doped regions 62, 63 and the electrode 64 in contact with the p-type doped region 65. This means that free carriers have less distance to travel between the electrodes 59, 60, 64 and the p-i-n diode can be switched more quickly. However, to attain a smaller critical angle θ_c , there must be a greater difference in refractive index across the boundary; and, in turn, a greater change in free carrier density. This tends to increase the time it takes to switch the p-i-n diode. Inevitably, there is therefore a trade-off between these conflicting criteria that leads to an optimum selection of critical angle θ_c and orientation of the waveguides 52, 53 to the boundary.

In operation, when an electric potential is applied between the electrodes 59, 60 in contact with the n-type regions 62, 63 and the electrode 64 in contact with the p-type region 65, there is a change in the density of free carriers (e.g. electrons and holes) in a volume of the crossover 55 within the perimeter formed by the carrier restrictive region 56. This causes the refractive index of that volume of the crossover 55 to change, as a result of the Plasma Dispersion effect. Light propagating along the input path of the first waveguide 52 can normally pass straight through the crossover 55 and continue along the output path of that waveguide 52. However, when the refractive index of the volume of the crossover 55 is changed by application of an electric potential between the electrodes 59, 60, 64, light can be reflected in the crossover 55 such that it is diverted down the output path of the second waveguide 53 by total internal reflection.

As mentioned above, when the p-i-n diode is forward biased, a reverse bias applied between the carrier restrictive region 56 and the volume of the crossover 55 on the other side of the boundary. This creates a depletion region between the carrier restrictive region 56 and the volume of the Si layer of the SOI body with the perimeter formed by the carrier restrictive region 56, in particular the rest of the crossover 55. So, most free carriers that are generated in the p-i-n diode stay within the perimeter defined by the carrier restrictive region 56. In particular, free carriers are confined at the boundary in the crossover 55. So, the boundary provides a clear limit to the diffusion of free carriers across the width of the crossover 55. Referring to figure 7, this means that when an electric potential is applied between the electrodes 59, 60, 64 to decrease the refractive index of the volume of the crossover 55 to which the free carriers are confined, there is a sharp change in refractive index across the boundary. In other words, there is a clear step in the refractive index at the boundary, as illustrated by the line D-D in figures 6 and 7 with distance along the line C-C shown in figure 5. This contrasts with the gradual change in refractive index of the prior art, as illustrated in figure 4. So, the reflection achieved at the boundary much more efficient than the prior art.

The optical switch 50 has been described above with the electrodes 59, 60, 64 and n-type and p-type doped regions 62, 63, 65 forming a p-i-n diode that can be used to inject free carriers into the volume of the crossover 55 on one side of the boundary to decrease the refractive index of that volume of the crossover 55. To achieve this, an electric potential is applied between the electrodes 59, 60, 64 to forward bias the p-i-n diode. There is, however, no reason why the n-type and p-type doped regions 62, 63, 65 cannot be interchanged, such that the electric potential is applied between the electrodes 59, 60, 64 with opposite polarity to forward bias the p-i-n diode and inject the free carriers into the volume. In this case, the electrodes 62, 63 on the input and output sides of the crossover 55 would be in contact with p-type regions and hence it is these electrodes 62, 63 that would also be connected to the electrode (not shown) in contact with the carrier restrictive region 56 to create the reverse bias across the boundary.

Furthermore, in another modification, the volume of the crossover 55 that has its refractive index changed by the applied electric potential can be doped such that it contains a given density of free carriers when no electric potential is applied between the electrodes 59, 61 , 64. In other words, the volume may have a different refractive index from the rest of the crossover 55 in the absence of an applied electric potential, e.g. such that total internal reflection occurs at the boundary. The electric potential may then be applied between the electrodes 59, 61 , 64 to reverse bias the p-i-n diode and reduce the density of free carriers in the volume, e.g. such that the refractive index of the volume is increased to the same refractive index as the rest of the crossover 55 and no reflection occurs at the boundary. In this case, the electrode in contact with the carrier restrictive region 56 is connected to the one or both of the electrodes 59, 61 in contact with the n-type regions 62, 63 of the p-i-n diode. This means that reverse biasing the diode also generates a reverse bias between the carrier restrictive region 56 and the volume of the crossover 55 on the other side of the boundary. This may enhance the depletion of the volume and improve device performance.

In another embodiment, the carrier restrictive region 56 has a different refractive index to the rest of the Si layer at the crossover 55. Indeed, it is possible that the carrier restrictive layer contains defects, e.g. is amorphised or comprises poly silicon, but is only lightly doped or not doped at all. This means that it has a higher refractive index than the rest of the Si layer at the crossover 55. The difference in refractive-index, caused by this alone can be sufficient that total internal reflection occurs at the boundary in the absence of an applied electric potential. In operation, an electric potential can then be selectively applied between the electrodes 59, 61 , 64 just to reverse bias the p-i-n diode and reduce the density of free carriers in the volume. This increases the refractive index of the volume to the same as that of the carrier restrictive region 56 and light is allowed to pass straight through the crossover 55 from the input to the output of the first waveguide 52. In this modification, the carrier restrictive region 56 can again also be doped and reverse biased during switching to enhance the depletion of the volume and improve device performance.

In yet another modification, the n-type doped regions 62, 63 and p- type doped region 65 may be in contact with one another to form a p-n diode.

The optical switch 50 has been described above in relation to total internal reflection of all of the light propagating along the input path of the , first waveguide 52. However, different polarisation states of light propagating in the waveguide 52 actually have slightly different reflection characteristics. More specifically, the reflection coefficient r_TE of the quasi- Transverse Electric (TE) polarised component of the light propagating in the waveguide 52 is different to the reflection coefficient r_m of the quasi- Transverse Magnetic (TM) polarised component of the light, e.g.

and

H₁ cos#, - n. Jn.,² - n,² sin² θ,

™ ^~ n₂ cos^, + H₁ TJ In-, 2 - «, 2 sin T⁼ O₁ ^ '

where θ_x is the angle of incidence of light at the boundary (with respect to a line normal to the boundary). So, in another embodiment the optical switch 50 may be modified to separate light having these different polarisation states by reflecting light having just one of the polarisation states.

Referring to figures 8 and 9, an optical switch 50 according to a second preferred embodiment of the invention is similar to the optical switch 50 of the first preferred embodiment of the invention illustrated in figures 5 and 6 and the same reference numerals are used for like components. However, in this embodiment, the optical switch 50 has two separate carrier restrictive regions 68, 69. A first of the regions 68 again has a side 70 that extends through the crossover 55 along a plane of symmetry of the X-shape formed by the waveguides 52, 53, which plane is defined by a first line basically parallel to a direction from the input side to the output side of the switch 50 and a second line perpendicular to the directions in which light propagates in each of the waveguides 52, 53. However, the region 68 is elongate and terminates on the input and output sides of the crossover 55. It does not form a perimeter as in the first preferred embodiment of the invention described above. A second carrier restrictive region 69 is located in the Si layer of the SOI body 51 , but does not extend into the crossover 55 or either of the waveguides 52, 53. Rather, it is on the other side of the electrodes 59, 60 and 64 and n-type and p-type doped regions 62, 63 and 65 to the first region 68.

So, the two carrier restrictive regions 68, 69 of the second preferred embodiment of the invention again confine any free carriers generated between the electrodes 59, 60, 64 and n-type and p-type doped regions 62, 63, 65 to one half of the crossover 55. They also confine the free carriers to a width of the Si layer of the SOI body 51 between the two regions 68, 69. At the same time, the carrier restrictive regions 68, 69 do not extend through the waveguides 52, 53, except at the crossover 55, which improves propagation of light through the waveguides 52, 53.

Referring to figures 10 and 11 , an optical switch 50 according a third preferred embodiment of the invention is similar to the optical switches 50 according to the first and second embodiments of the invention illustrated in figures 5, 6, 8 and 9, and the same reference numerals are used for like components. Indeed, the optical switch 50 of the third embodiment of the invention is the same as the optical switch 50 of the second embodiment of the invention except that the second carrier restrictive region 69 is dispensed with. So, whilst the free carriers are not confined across the width of the Si layer of the SOI body 51 , the relative positioning of the first carrier restrictive region 68 and the electrodes 59, 60, 64 ensures that the free carriers are confined to one half of the crossover 55. For most applications, the third embodiment of the invention should not therefore have any significant disadvantages in comparison to the second embodiment of the invention and is simpler to manufacture. However, it does require slightly greater current to achieve the same change free carrier density in the crossover 55, as the free carriers have more freedom to diffuse to other regions of the Si body 51. This diffusion could also potentially interfere with other semiconductor devices situated close to the optical switch 50 on the SOI body 51. So, if it is desired to situate other semiconductor devices close to the optical switch 50 on the SOI body 51 , provision of the second carrier restrictive region 69, as in the second preferred embodiment of the invention, or a region 56 forming a perimeter, as in the first preferred embodiments of the invention, may advantageously limit diffusion of the free carriers towards the other semiconductor devices.

In order to manufacture the optical switch 50, the waveguides 52, 53 are first etched into the surface of the SOI body 51. Defects are then provided by implanting Si ions into Si layer of the SOI body 51 in the carrier restrictive region(s) 56, 68, 69. In the illustrated embodiments, the carrier restrictive region(s) 56, 68, 69 extends from the upper surface 61 of the SOI body 51 to the SiO₂ base layer 54 of the SOI body 51. Annealing is then carried out and, in the illustrated embodiments, the density of defects provided in the region(s) 56, 68, 69 is sufficient that polysilicon is formed in the region 56 during the anneal. A high temperature drive in process is then used to put phosphorus (P) in the region(s) 56, 68, 69. This dopes the region 56 and forms a depletion region between the carrier restrictive region(s) 56, 68, 69 and the rest of the Si layer of the SOI body 51. Conveniently, the high temperature drive in is much more effective in polysilicon than crystalline silicon, as the P ions diffuse more readily in polysilicon. This allows the P ions to be selectively placed in th*^» carrier restrictive region(s) 56, 68, 69.

The n-type and p-type doped regions 62, 63, 65 are also provided by ion implantation or diffusion of an appropriate dopant and the electrodes 59, 60, 64 are provided by deposition of a suitable conductor. The described embodiments of the invention are only examples of how the invention may be implemented. Modifications, variations and changes to the described embodiments will occur to those having the appropriate skills and knowledge. For example, the invention may be applied to optical filters or attenuators. The refractive index of the volume of the crossover 55 on one side of the boundary might also be itself varied by light of an appropriate wavelength, by the interaction of photons with the semiconductor from which the crossover 55 is formed. A variety of materials other than Si and SiO₂ might alternatively be used to implement the invention. Similarly, the location of the defects and/or dopant of the carrier restrictive region(s) 56, 68, 69 might be inversed, so that the defects and/or dopant are provided elsewhere the SOI body 51 , but not in the carrier restrictive regions 58, 67, 70. Alternatively, relative changes in the defect and/or dopant density can be used, rather than an absence of defects and/or dopant in any one region or volume. Furthermore, the boundary may not coincide exactly with the side 57, 69 of the carrier restrictive region 56, 68. For example, the depletion region formed between the carrier restrictive region 56, 68 and the rest of the crossover 55 may extend out of the carrier restrictive region. This means that the boundary would be spaced away from the side 57, 69 of the carrier restrictive region 56, 68 in an outward direction with respect to the carrier restrictive region 56, 68. These modifications, variations and changes may be made without departure from the spirit and scope of the invention defined in the claims and its equivalents.

Claims

1. An optoelectronic device comprising an optical waveguide disposed in a semiconductor, wherein the refractive index of a volume of the waveguide can be altered under the influence of electric potential to cause light propagating along the waveguide to be selectively reflected at a boundary of the volume to generate an output, the boundary being defined by a variation in the density of defects and/or a dopant in the semiconductor.

2. The optoelectronic device of claim 1 , wherein the variation is arranged to define the boundary across substantially a whole transverse section of the waveguide.

3. The optoelectronic device of claim 1 or claim 2, wherein the variation is arranged to define the boundary obliquely to the direction in which light propagates in the waveguide.

4. The optoelectronic device of any one of the preceding claims, wherein the variation is arranged to define the boundary oriented to the direction in which light propagates in the waveguide such that the light can be substantially totally internally reflected at the boundary.

5. The optoelectronic device of any one of the preceding claims, wherein the variation is at the edge of a carrier restrictive region.

6. The optoelectronic device of claim 5, having more than one carrier restrictive region.

7. The optoelectronic device of claim 5 or claim 6, wherein the carrier restrictive region(s) confine(s) the extent of the volume along the length of the waveguide.

8. The optoelectronic device of claim 5 or claim 6, wherein the carrier restrictive region(s) define(s) a perimeter in a plane defined by the length and width of the waveguide.

9. A method of operating the optoelectronic device of any one of the preceding claims, the method comprising applying an electric potential to alter the refractive index of the volume.

10. The optoelectronic device of any one of claims 1 to 8, wherein the volume has a refractive index substantially the same as that of the waveguide outside the volume at the boundary, such that light propagating in the waveguide is admitted at the boundary when the refractive index of the volume is unaltered by the electric potential.

11. The method of claim 9, wherein light propagating in the waveguide is admitted at the boundary in the absence of applied electric potential, the method comprising applying an electric potential to alter the refractive index of the volume such that the light is reflected at the boundary.

12. The optoelectronic device of any one of claims 1 to 8, wherein the volume has a refractive index different to that of the waveguide outside the volume at the boundary, such that light propagating in the waveguide is reflected at the boundary when the refractive index of the volume is unaltered by the electric potential.

13. The optoelectronic device of any one claims 1 to 8 and 12, wherein the volume is doped such that it has a refractive index different to that of the waveguide outside the volume at the boundary.

14. The optoelectronic device of any one of claims 1 to 8, 12 and 13, wherein the volume has a different defect density than, such that it has a different refractive index to, a/the carrier restrictive region outside the volume at the boundary.

15. The optoelectronic device of any one of claims 1 to 8 and 12 to 14, wherein light propagating in the waveguide is substantially totally internally reflected at the boundary when the refractive index of the volume of the waveguide is unaltered by the electric potential.

16. The method of claim 9, wherein light propagating in the waveguide is reflected at the boundary in the absence of applied electric potential, the method comprising applying an electric potential to alter the refractive index of the volume such that the light is admitted at the boundary.

17. The method of any one of claims 9, 11 or 16, wherein the reflection at the boundary is such that one polarisation mode of the light propagating in the waveguide is reflected and another polarisation mode is admitted.

18. The method of any one of claims 9, 11 , 16 or 17, wherein the reflection at the boundary is substantially total internal reflection.

19. The optoelectronic device of any one of the preceding claims, wherein the variation defines the boundary at a crossover between the waveguide and another waveguide.

20. The optoelectronic device of any one of the preceding claims, wherein light admitted at the boundary continues along the waveguide ar._ light reflected at the boundary is diverted along an/the other waveguide.

21. A method of manufacturing an optoelectronic device according to any one of the preceding claims. .

22. A method of manufacturing an optoelectronic device, the method comprising: forming a waveguide from a semiconductor; and providing defects and/or a dopant in portion of the semiconductor such that there is a variation in the density of defects and/or the dopant defining a boundary to a volume of the semiconductor, the refractive index of which volume can be changed under influence of electric potential to cause light propagating along the waveguide to be selectively reflected at the boundary to generate an output.

23. The method of claim 22, wherein the variation is arranged to define the boundary across substantially a whole transverse section of the waveguide.

24. The method of claim 22 or claim 23, wherein the variation is arranged to define the boundary obliquely to the direction in which light propagates in the waveguide.

25. The method of any one of claims 23 to 25, wherein the variation is arranged to define the boundary oriented to the direction in which light propagates in the waveguide such that the light can be substantially totally internally reflected at the boundary.

26. The method of any one of claims 22 to 25, wherein the variation is at the edge of a carrier restrictive region.

27. The method of claim 26, comprising providing more than one carrier restrictive region.

28. The method of claim 26 or claim 27, wherein the carrier restrictive region(s) confine(s) the extent of a volume along the length of the waveguide.

29. The method of claim 26 or claim 27, wherein the carrier restrictive region(s) define(s) a perimeter in a plane defined by the length and width of the waveguide.

30. The method of any one of claims 22 to 29, wherein the semiconductor from which the waveguide is formed is silicon.

31. The method of any one of claims 22 to 30, wherein the defects are provided by implanting ions in the semiconductor.

32. The method of claim 31 , wherein the defects are provided by implanting Si ions in the semiconductor.

33. The method of claim 31 or claim 32, comprising annealing the semiconductor following ion implantation.

34. The method of any one of claims 31 to 33, wherein the ion implantation generates a region of polysilicon in the semiconductor.

35. The method of any one of claims 22 to 34, wherein the dopant is Phosphorus.

36. The method of any one of claims 23 to 35, comprising doping a/the volume of the waveguide such that it has a different refractive index to that of the waveguide outside the volume.

37. The method of any one of claims 22 to 36, comprising providing the volume with a different defect density than, such that it has a different refractive index to, a/the carrier restrictive region outside the volume at the boundary.

38. An optoelectronic device, substantially as described with reference to any of the accompanying drawings.

39. A method of operating an optoelectronic device, substantially as described with reference to any of the accompanying drawings.

40. A method of manufacturing an optoelectronic device, substantially as described with reference to any of the accompanying drawings.