The present invention relates to optical telecommunications, and more specifically to polarisation beam splitters.
M. Okuno et al., Birefringence Control of Silica Waveguides on Si and Its Application to a Polarization-Beam Splitter/Switch, Journal of Lightwave Technology, vol. 12, no. 4, pp. 625-633 discloses a polarisation beam splitter, built using integrated Mach-Zehnder interferometer and stress applying amorphous silicon layers. One drawback of this splitter is the additional manufacturing step for forming the α-Si layer.
Y. Hashizume et al, Integrated polarisation beam splitter using waveguide birefringence dependence on waveguide core width, Electronics Letters vol. 37, no. 25, December 2001, pp. 1517-1518, discloses a polarisation beam splitter, built using integrated Mach-Zehnder interferometer with one branch having an increased width.
One could also contemplate a bulk optics isolator, including two collimating lenses, two polarisation splitters, a Faraday rotator and a wave-plate. Rigorous alignment of these components would then be necessary for ensuring proper operation. Collimating lenses should notably be placed with a high precision. This would make it difficult to insert the isolator within a complex planar-based optical component.
Accordingly, there remains a need for an isolator or circulator, which is easy to manufacture and which may be inserted in a planar lightwave circuit.
The invention therefore provides an optical component, comprising
One or both polarisation beam splitter may comprise a Mach-Zehnder interferometer, having at least one branch with an increased width section; one may also provide at least one branch with a reduced width section. The Mach-Zehnder interferometer(s) may further comprise a heater adapted to heat at least one branch of the interferometer. A cooler is another possible option. One may also provide that the maximum isolation wavelength of the component is tuneable.
In one example, the wave plate is a quarter wavelength wave plate.
The assembly may be connected to the splitter via at least one segmented waveguide or via at least one tapered waveguide.
In another example, the Faraday rotator comprises at least two sections separated by waveguides. One may also provide that the wave plate and Faraday rotator are separated by waveguides.
The Faraday rotator may comprise a magnet on top of the planar lightwave circuit, an inverted U-shaped magnet sitting on top of the rotator or a C-shaped magnet sandwiching the edge of the planar lightwave circuit.
A circulator embodying the invention will now be described, by way of non-limiting example, and in reference to the accompanying drawings, where:
The invention is directed to a circulator embodied in a planar lightwave circuit (PLC). It preferably uses PLC polarisation beam splitters, which are discussed in reference to
Planar lightwave circuits made of Silica on Silicon show a small birefringence. This is advantageous in that it permits manufacturing of polarisation independent integrated optical components. However, the small birefringence of the silica glass waveguide layers is a drawback for manufacturing integrated polarisation splitters—since polarisation splitters clearly require the ability to separate light according to its polarisation. In the rest of this description, parallel polarisation is used for referencing light having a polarisation parallel to the substrate of the planar lightwave component: perpendicular polarisation is used for referencing light having a polarisation perpendicular to the substrate of the planar lightwave component.
The polarisation beam splitter of
The splitter of
Coupler 12 feeds symmetrically light received on input ports 4 and 6 into the two arms 14 and 16 of a Mach-Zehnder interferometer. Thus, a first end of the arms 14 and 16 is connected to the first coupler 12.
The other ends of the arms 14 and 16 of the Mach-Zehnder interferometer are connected to a second coupler 18. Coupler 18 is another 3 dB coupler, which may like first coupler 12 be a MMI coupler. Light received by the coupler from both arms of the Mach-Zehnder interferometer is symmetrically provided to output ports 8 and 10.
As explained above, the relative phase shift between the two arms 14, 16 of the Mach-Zehnder interferometer is obtained by varying the width of the waveguide forming at least one of both arms. In the example of
For realising polarisation independent waveguides in SiO2Si, one may balance the stress for TE and TM polarisation by using a higher doped cladding material. This balancing is related to square waveguides. If the waveguide width is changed the balancing is no longer right and birefringence occurs. Calculations showed that the birefringence increases linearly with the waveguide width.
Thus, the width of the waveguide forming one of the branches of a Mach-Zehnder interferometer may be adapted, so as to obtain a relative phase shift between the two branches of the interferometer, which varies according to the polarisation of the light. The example discussed above is one of an increased width section; one could also use a reduced width section, or use at the same time an increased width section in one arm and a reduced width section in the other arm.
On both sides of section 20 with an increased width, there are provided tapers 22, 24. Each taper connects the waveguide of the second arm 16 with the increased width section 20. Due to its width, section 20 may no longer be single mode. Tapers 22, 24 provide an adiabatic transition from the single mode waveguide to the increased width section 20; they avoid exciting higher order modes in the increased width section 20.
The operation of the polarisation beam splitter of
The parallel polarisation component of the light travelling through the second branch 16 of the Mach-Zehnder interferometer LP2 undergoes zero phase shift, relative to the parallel polarisation component LP1 of the light travelling through the first branch 14. In the second coupler 18, lights LP1 and LP2 interfere destructively on port 8 and constructively on port 10, due to the additional π/2 phase shift caused by the second coupler 18.
Similarly, the orthogonal polarisation component of the light travelling through the second branch 16 of the Mach-Zehnder interferometer LT2 undergoes a phase shift π, relative to the orthogonal polarisation component LT1 of the light travelling through the first branch 14. In the second coupler 18, lights LT1 and LT2 interfere constructively on port 8 and destructively on port 10.
Thus, the polarisation beam splitter 2 provides:
An example of possible dimensions of the planar lightwave component of
The operation of the polarisation beam splitter is clearly symmetric. Thus, the splitter may be used for splitting the light received on port 6; it then provides the parallel polarisation component of this light on port 8 and the orthogonal polarisation component of this light on port 10. The operation of the splitter for light received on ports 8 and 10 is similar.
The splitter of
Silica has a positive thermo-optic coefficient, so that its refractive index increases when it is heated by the resistor 26. Assuming the waveguide of the first branch is formed of silica, the index of the waveguide is increased, which increases the phase shift in this branch.
The heater may first be used for optimising the splitting. The “fine tuning” is helpful to get the right phases or working point for the polarisation beam splitter.
In addition, the heater may be used for thermoregulating the splitter. This may not be necessary. Ideally the phase shift of the two waveguide arms is very small: 0 for one polarisation π for the other polarisation. The temperature dependence of the total device is therefore very small. This might be different if one decides to use an asymmetric Mach-Zehnder with some picoseconds time delay.
The thin film heater can be used for a polarisation beam splitter switch. As the phase difference of the two arms can be changed. It is possible to launch the perpendicular polarisation to the upper arm 12 or to the lower arm 44—vice versa for the parallel polarisation. This has been demonstrated by NTT in the above-mentioned reference Okuno et al. The thermooptical phase shifter induces also a small birefringence. However, this can be suppressed if trenches are etched besides both waveguides. In the example of
The splitter discussed above was shown to have an extinction ratio of more than 20 dB. It operates in the C-band, over a range of more than 40 nm. Especially when the relative time delay/phase shift of the two Mach-Zehnder branches is matched—apart form the π phase shift for one polarisation— very broadband device covering C— and L-Band is obtained.
In the example of
In the example of
The presence of the heater has the further advantage of improving the symmetry of the splitter. It makes it possible to tune the wavelength where the maximum isolation is obtained. This makes it possible to adapt the splitter to the wavelength band where it is being used.
The polarisation beam splitter of
The assembly of the wave plate and Faraday rotator is asymmetric. The principal states of polarisation—parallel and orthogonal—are conserved from when light passes from left to right, and are exchanged when light passes from right to left. The term right and left refer to the drawing; the polarisation states are conserved when light enters the assembly through the λ/4 wave plate and are exchanged when light enters the assembly through the Faraday rotator.
The operation of the circulator of
Lights LP1 and LT1 pass through the wave plate 46 and Faraday rotator 48; as discussed above, this does not change their polarisation state.
Lights LP1 and LT1 then enter the output ports 50 and 52 of the second splitter 54. They interfere constructively on the second port 34 of the circulator, and destructively on the fourth port 38 of the circulator. Thus, light entering the circulator on the first port 32 is output on the second port 34.
Similarly, light entering the circulator on the third port 36 is output on the fourth port 38.
In the other direction of propagation of light, light L2 entering the circulator on the second port 34 is separated into
Lights LP2 and LT2 pass through the Faraday rotator 48 and wave plate 46; as discussed above, due to the fact that the rotator is asymmetrical, the orthogonal and parallel states of polarisation are exchanged. Thus, light entering the first output port 42 of the first splitter 40 is polarised parallel to the substrate, while light entering the second output port 44 of the second splitter 40 is polarised orthogonal to the substrate. As a consequence, these lights undergo destructive interference toward the first port 32 of the circulator and constructive interference toward the third port of the circulator.
Thus, light entering the circulator on the second port 34 is output on the third port 36. Similarly, light entering the circulator on the fourth port 38 is output on the first port 32.
The device of
As compared to the prior art device, the splitter does not use bulk polarisation splitters, but planar splitters. This is per se an important cost reduction; in addition, due to the use of planar splitters, there is no need to provide collimating lenses, which also makes the circulator less costly.
The assembly of the circulator is also simplified: the splitters may simply be integrated in the PLC. The wave plate and Faraday rotator are mounted into slots etched or diced in the PLC; the waveguides are segmented or tapered to the waveplate or rotator; in any case, there is no need for collimating lenses and optical alignment of the splitters with the wave plate and Faraday rotator.
The PLC polarisation splitters were discussed in reference to
In both examples, the Faraday rotator uses a magnet (not represented on the figures). The magnet creates a magnetic field, roughly oriented in the direction of the beam and exceeding the saturation field of the rotating material. The field type extends over a zone less than 500 μm×100 μm×30 μm. Due to the low magneto-optic coefficient of the silica used for the PLC waveguides, the presence of the magnetic field has little if no polarisation rotation effect on the PLC components—notably on the splitters. Thus, a magnet of 1 mm3 size is sufficient for the Faraday rotator. The shape of the magnet is of little relevance. The magnet could for instance be a plate sitting on the top of the PLC or have an inverted-U shape sitting on top of the waveguide forming the Faraday rotator. Both solutions are compatible with having the circulator function anywhere within a complex PLC. If the circulator is close to the edge of the PLC, one may also use a C-shaped magnet sandwiching the edge of the PLC over the waveguide of the rotator. This simplifies assembly of the magnet and increasing the magnetic field within the rotator.
The insertion losses of the circulator of
The table below gives the minimum mode radius expansion for ensuring that the insertion loss of the circulator remains below 1 dB. The external coupling loss is not taken into account, this corresponding to the case where the circulator is integrated into a more complex PLC-based component or subsystem. The table considers the examples of rotators having a length of 400 μm or a length of 290 μm.
Losses would be even lower if the length of the rotator was further reduced, with a similar mode radius expansion; alternatively, mode radius expansion could be further reduced if the length of the rotator was further reduced. Losses in the wave plate are an order of magnitude lower, due to the reduced thickness of the wave plate, and their contribution may be disregarded. The table considers two values of the step index for the PLC silica waveguides; the value of 0.015 corresponds to bend radii of 3 mm and a −0.7 dB coupling loss to a G.652 step-index fibre; the value of 0.008 increases the bend radii to 12 mm, with however a −0.2 dB coupling loss.
The table shows that in the worst case of the 400 μm long rotator, insertion losses of 1 dB may be obtained, with a mode expansion of ×3.3. This value is still realistic. Otherwise, a mode expansion of no more than ×2 with a split rotator also provides losses of less than 1 dB, with a step index of 0.008. The table shows that low insertion losses may be obtained for the circulator.
As a comparison, the insertion loss of a standard bulk isolator coupled to a PLC by fibre pigtails would be 0.9 to 1.1 dB (single stage or double stage), to which the connection losses to the PLC should be added. This shows that insertion losses for the circulator of
With a splitter having extinction ratio of more than 20 dB, the extinction ratio of the circulator—that is the power ratio of e.g. port 3 or 4 to port 2, for light applied to port 1—is more than 20 dB. It is also possible to cascade PBSs to increase the extinction ratio to more than 40 dB. For such a configuration on each output arms 8 and 110 of the PBS another PBSs is cascaded working as a TE mode filter and a TM mode filter. The wavelength range over which the circulator operates may be higher than 100 nm. Indeed, the PLC polarisation beam splitter was shown to operate in C or L-bands, over a range of more than 100 nm. Commercially available wave plate and Faraday rotator may also operate over such a range. The circulator thus behaves identically or almost identically over a broad range of wavelength.
The examples described above discuss the use of the component of the invention as a circulator and as an isolator. The component is thus an isolator and circulator in one.
In the example where the polarisation beam splitter is switchable thanks to the thermooptical phase shifters, the circulator or isolator offers switching ability, by reversing the relative phase shift between parallel and orthogonal polarisation components of the light. Thermo-regulation also makes it possible to revert the cycle of the circulator; in the example, the circulator is a 1234 cycle circulator, where light entering port n is output at port n+1 (modulo 4). The switching ability of the polarisation beam splitter makes it possible to change the circulator to a 4321 circulator, where light entering port n is output at port n−1 (modulo 4).
Last, thanks again to thermooptic phase shifting, the maximum isolation is tuneable. For instance, the centre wavelength could be tuned for optimal operation in the C-band, or for optimal operation in the L-band, or for optimal operation in both bands. This makes it possible to use the same component for covering C band as well as L band. This reduces diversity.
In the examples of
Although this is not represented on the figures, the wave plate and Faraday rotator would usually be at an angle with respect to the waveguides; one may also coat their facets with anti-reflection coating; this reduces insertion losses and return losses.
In the examples of
Last, the Faraday rotator may comprise or not a waveguide. As discussed above, the circulator may be manufactured, with reduced insertion losses, even where the rotator does not have a waveguide. The Faraday rotator may also include a waveguide; this would help in controlling mode mismatch and would further reduce insertion losses.
Number | Date | Country | Kind |
---|---|---|---|
02292036.7 | Aug 2002 | EP | regional |