WO2002031591A1

WO2002031591A1 - Poled optical modulator

Info

Publication number: WO2002031591A1
Application number: PCT/GB2001/004451
Authority: WO
Inventors: Peter George Robin Smith
Original assignee: University Of Southampton
Priority date: 2000-10-10
Filing date: 2001-10-08
Publication date: 2002-04-18
Also published as: GB2368402A; AU2001292107A1; GB0024809D0

Abstract

A temperature stabilised Mach-Zehnder optical modulator device (14-20) fabricated in a multi-domain substrate (12) having inverted and non-inverted domains (34, 36) separated by domain boundaries (35), wherein the electrodes (30, 31) are arranged so as to invert the biasing direction on crossing the domain boundaries which extend generally transverse to the central waveguide section (16, 18) of the optical modulator. Excess charge is thus cancelled by the multiple domains without the need to provide the usual conductive paths which are provided in the prior art to channel excess pyroelectric change away from the electrodes. This design is especially attractive for z-cut lithium niobate devices where pyroelectric effects are known to be problematic. The invention includes several other domain geometries and is applicable to a wide variety of optical and electrical devices other than modulators.

Description

POLED OPTICAL MODULATOR

BACKGROUND OF THE INVENTION

The invention relates to stabilising devices fabricated from polar or ferroelectric materials against effects resulting from the build-up of charge induced by the pyroelectric and other effects.

Polar and ferroelectric materials are widely used in electronic and optical devices because of their ferroelectric, pyroelectric, piezoelectric, converse piezoelectric, electro-optic and nonlinear optical properties [Lines and Glass]. These properties have in common that they all require a non-centrosymmetric material to be used. For example, quartz, which is non-centrosymmetric, may be used to make a piezoelectric oscillator. Another commercially important device is the optical modulator utilising the electro-optic effect. Optical modulators are often made of ferroelectric materials, most notably lithium niobate (LiNbO₃). Other materials such as lithium tantalate (LiTaO₃) are also used. Optical modulators utilising the electro- optic effect are in widespread use in optical telecommunications. Examples of devices that use optical modulators are optical intensity modulators, switches, phase shifters, frequency shifters, polarisation transformers and wavelength filters.

A well known shortcoming of optical modulators made of such ferroelectric materials is their temperature dependence. For some types of device and applications, temperature dependence may be critical. For example, the operation of interferometric devices is critically dependent on the optical path length difference between two or more waveguides.

One commercially important example are devices based on a Mach-Zehnder (MZ) interferometer arrangement. The thermal drift problem is especially severe in devices with asymmetric path lengths, for example in devices such as MZ optical modulators designed to give controllable pulse chirp. Changes in temperature may be caused by a change in the ambient environment or a change in the temperature of the active region of the modulator caused by device operation, e.g. by electrical dissipation of radio frequency (RF) drive power applied to an optical modulator's electrodes. Similar problems occur in arrayed waveguide grating (AWG) devices which also depend on an interferometric effect and are often made in polar materials such as GaAs and InP.

One main cause of the temperature dependence in lithium niobate and other ferroelectric and polar materials is the pyroelectric effect which is induced by changes in temperature, i.e. according to the derivative of temperature with respect to time. Temperature changes cause a change in the dipole moment in a ferroelectric material resulting in either: (a) the release of free charge; or (b) a deficiency of cancelling free charge on the surface of the material. Free charge appears in order to compensate for the bound charge at the surface of the material. Free!:charge may be in the form of ionised impurities, free electrons, ions from the atmosphere, or other mobile charge carriers. In the case of LiNbO₃ modulators, an insulating layer is used between the waveguide core and the electrodes to reduce optical losses. 'This insulating layer traps charges which allows the build-up of fields which, in turn, results in drift in the operating point of the modulator. The pyroelectric effect is known to cause particular problems for .z-cut

LiNbO₃, since the electric fields induced by the pyroelectric effect are then aligned perpendicular to the plane of the substrate, so that bound charge appears on the top and bottom surfaces of the substrate. The bound charge on the top surface then interferes with the electrodes. This problem is simple to avoid by using x-cut LiNbO₃ in which case the electric fields induced by the pyroelectric effect are aligned parallel to the plane of the substrate so that bound charge appears on the sides of the substrate well away the electrodes. However, z-cut LiNbO₃ is more desirable in some cases, e.g. for high frequency devices.

Various schemes have been proposed to overcome this problem, all based on the idea of using conductive layers to conduct away surface charge induced by the pyroelectric effect. For example, in [Heisman] the substrate is provided with a conductive layer on its top surface which is electrically isolated from the drive electrodes but electrically coupled to a further conductive layer on the bottom surface. The electric field in the crystalline lattice of the substrate resulting from the pyroelectric effect is thus equalised. This approach has been commercially exploited to manufacture z-cut Ti:LiNbO₃ optical modulators with low thermal drift of the bias point. Provision of such conductive layers does however complicate the fabrication process. In particular, there are conflicting requirements in that, on the one hand, the top conductive layer must not electrically short with the drive electrodes, but, on the other hand, the top conductive layer needs to be arranged as close as possible to the drive electrodes for efficient conduction of unwanted pyroelectric-induced surface charge away from the critical area of the drive electrodes.

It would therefore be desirable to address the problems associated with charge fluctuations caused by the pyroelectric and other effects by means other than simply providing conductive paths to conduct away charge from regions of the device where it causes problems.

SUMMARY OF THE INVENTION

According to the invention domain inversion is used to provide stabilisation against thermal drift in devices made in polar or ferroelectric materials. The inversion can be provided by ferroelectric domains or polar reversal domains, e.g. by multi- domain ferroelectric or polar twinned structures, respectively. The invention thus adopts a completely different approach to the approaches of the above-mentioned prior art, which are all based on the idea of providing conductive paths to transport surface charge away from the active parts of the device. In a first aspect of the invention there is provided a device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; and an electrode structure for biasing the device, the electrode structure covering a path in the substrate that extends across at least one of the domain boundaries, and being arranged so as to provide an inversion of the biasing on crossing the at least one of the domain boundaries.

The domains may be elongate, extending generally transverse to the path. The elongate domains may be arranged to extend perpendicular to the path or to extend at an oblique angle to the path in order to suppress back-reflections. Alternatively, the domains may be dispersed in a two-dimensional arrangement. The path may be an optical or electrical path. In the case of an optical path, this may be in a bulk material or defined by a waveguide. In the case of an electrical path, this may be a ballistic electron path in a quantum well, wire or dot. In addition, the domains may be arranged periodically or aperiodically along the path both in the one-dimensional (i.e. elongate strip) and two-dimensional (i.e. granular) domain arrangements. In a second aspect of the invention there is provided a device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; and an electrode structure for biasing the device, the electrode structure covering a path in the substrate that extends generally parallel to at least one of the domain boundaries, and being arranged so as to provide a unidirectional biasing for the path. The path may be an optical or electrical path. In the case of an optical path, this may be in a bulk material or defined by a waveguide. The domains may be arranged periodically or aperiodically along the path.

The second aspect thus contrasts with the first aspect in that the domain boundaries or electrode structure are aligned with the path, rather than across it. The same underlying principle is however exploited, namely the use of multiple domains to cancel out the surface charge locally and globally over the top surface of the device

(and also the bottom surface).

In a third aspect of the invention there is provided a device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; and a path including a waveguide that extends generally parallel to at least one of the domain boundaries.

The domains are preferably elongate, extending generally parallel to the path.

According to aspects and embodiments of the invention it is thus possible to counteract the build up of pyroelectric charge on the faces of the device, and thus improve differential temperature insensitivity, by domain inverting the material over a suitable length scale and alignment having regard to device operation.

An example of a device that suffers from pyroelectric induced thermal instability is a z-cut titanium diffused lithium niobate modulator in which pyroelectric charge induced by temperature changes causes a drift in the bias point. As described herein, ferroelectric domain inversion is used in several embodiments of the invention to allow cancellation of the pyroelectric charge in such devices. Accordingly, it is possible to make devices that are more inherently thermally stable than devices fabricated in single domain substrates. In general the approach of reversing the ferroelectric or polar domain direction can be applied to any device fabricated from a ferroelectric or polar substrate, examples include, surface acoustic wave filters, lithium niobate modulators in other geometries than z-cut, where a pyroelectric contribution results in unwanted charge build-up. The approach may also be used in AWG's, e.g. AWG's made in GaAs, InP, poled silica or poled polymer. In embodiments of the invention, one or more of three effects may be exploited to reduce thermal drift by the provision of multiple domains, namely:

1. Overall reduction in the field across the whole sample (this means that the surface area of up and down domains should be approximately equal for the whole device).

2. Ensuring that each optical (or electrical) path has equal amounts of up and down domains. This is particularly important in those parts of a structure in which there are split paths (e.g. in the arms of an interferometric device). This ensures that any net DC or low frequency field across the whole sample cancels in the different regions.

3. Local fine scale variation in the domain orientation, as will occur for fine period structures, for which small scale local charge migration and cancellation will cause a considerable reduction in local drift resulting in improved global stability.

In one example, the substrate is a z-cut of 3m point group crystal, e.g. lithium niobate or related compounds.

Devices embodying the invention may utilise one or more of the following effects: pyroelectric, electrooptic, piezoelectric, inverse piezoelectric, photorefractive and surface acoustic waves.

Various devices can be made according to the aspects of the invention such as arrayed waveguide gratings, optical modulators (with or without waveguides, or bulk modulators). An important example of an optical modulator whose performance can be improved by any of the above-recited aspects of the invention is a Mach-Zehnder interferometric modulator.

Specifically, according to the first aspect of the invention, there may be provided a Mach-Zehnder optical modulator device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; a waveguide arrangement comprising an input section that divides into first and second arm sections that combine into an output section, wherein the first and second arm sections extend across at least one of the domain boundaries; and an electrode structure for biasing the device, the electrode structure covering at least one of the first and second arm sections, and being arranged so as to provide an inversion of the biasing on crossing the at least one of the domain boundaries.

At this point it is mentioned that multidomain substrates have been used previously in modulators, with the domains aligned transverse to the waveguide direction [Schaff er]. However, the prior art device exploited the multiple domains for increasing RF bandwidth and, as a result, incorporated a conventional electrode structure with uniform biasing.

Further, according to the second aspect of the invention, there may be provided a Mach-Zehnder optical modulator device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; a waveguide arrangement comprising an input section that divides into first ancl second arm sections that combine into an output section, wherein the first and second arm sections extend generally parallel to at least one of the domain boundaries; and an electrode structure arranged to provide unidirectional biasing of at least one of Jhe first and second arm sections. Specific examples of optical modulator devices that may benefit .from the present invention are a titanium diffused telecommunications modulator (either phase or amplitude modulated), a linearised modulator for CATV (cable access television), and a titanium diffused lithium niobate telecommunications switch (2 by 2 or higher, in general n by m). Particular and preferred aspects of the invention are set out in the accompanying independent and dependent claims. Features of the dependent claims may be combined with those of the independent claims as appropriate and in combinations other than those explicitly set out in the claims. BRIEF DESCRIPTION OF THE DRAWINGS

For a better understanding of the invention and to show how the same may be carried into effect reference is now made by way of example to the accompanying drawings in which:

Figure 1 shows a standard prior art optical modulator;

Figure 2 is a perspective view of an optical modulator according to a first embodiment of the invention having a multi-domain substrate and structured electrode;

Figure 3 is a section through the multi-domain substrate of Figure 2;

Figure 4 is a plan view of the modulator of Figure 2 showing the electrode structure in more detail;

Figure 5 is a plan view of the electrode structure of an optical modulator according to a second embodiment of the invention;

Figure 6 is a section X through Figure 5;

Figure 7 is a section Y through Figure 5;

Figure 8 shows in plan view an alternative domain alignment applicable to various embodiments of the invention,

Figure 9A is a plan view of an optical modulator according to a third embodiment of the invention with the electrodes omitted for clarity of representation of the relative alignment between the modulator waveguides and the ferroelectric domains;

Figure 9B corresponds to Figure 9 A and shows the electrodes;

Figure 10 is a plan view of an optical modulator according to a fourth embodiment of the invention;

Figure 11 is a plan view of an active arrayed waveguide grating (AWG) according to a fifth embodiment of the invention; and

Figure 12 shows a part of the AWG of Figure 11 in more detail.

DETAILED DESCRIPTION

Figure 1 shows a standard prior art optical modulator, generally as described in [Davis]. The basic features of this device are first described to allow a better appreciation of the following description of specific embodiments of the invention.

The modulator 10 is fabricated in a z-cut lithium niobate substrate 12 having a top surface 11 and a bottom surface 13. The modulator 10 is based on an integrated optic MZ interferometer and a pair of RF electrodes comprising a ground electrode 30 and a signal electrode 32 for inducing phase modulation in the MZ interferometer. The MZ interferometer includes first and second optical waveguides 16 and 18 that constitute first and second interferometer arms respectively. The waveguides 16 and 18 are formed in the substrate 12 and extend generally parallel to each other over their central portions. The MZ interferometer further includes an input waveguide 14 that leads to an input Y-j unction 15 which serves to divide an optical input signal into the first and second interferometer arms 16 and 18. The interferometer arms 16 and 18 rejoin at an output Y-junction 19 into an output waveguide 20. The input and output waveguides 14 and 20 terminate at the chip edge and can be connected by suitable pigtailing to input and output optical fibres 13 and 21, as illustrated. In the figure, the hatching of the electrodes 30 and 32 indicates the areas of the electrodes that perform a modulating function.

Figure 2 is a perspective view of an optical modulator 10 according to a first embodiment of the invention. The modulator 10 is fabricated in a lithium niobate substrate 12 having a top surface 11 and a bottom surface 13. The lithium niobate substrate is z-cut. An initially single domain sample is used that is oriented with the negative z-face as the top surface 11. The negative z-face is defined as the face that becomes negatively charged on cooling, whereas the positive z-face is defined as the face that becomes positively charged on cooling, as a result of the pyroelectric effect. The modulator 10 is based on an integrated optic MZ interferometer in combination with a pair of RF electrodes comprising a ground electrode 30 and a signal electrode 32 for applying an electric field to induce phase modulation. The MZ interferometer includes first and second optical waveguides 16 and 18 that constitute first and second interferometer arms 16 and 18 respectively. The waveguides 16 and 18. are formed in the substrate 12 and extend generally parallel to each other over their central portions. The MZ interferometer further includes an input waveguide 14 that leads to an input Y-j unction 15 which serves to divide an optical input signal into the first and second interferometer arms 16 and 18. The first and second interferometer arms 16 and 18 rejoin at a further, output Y-junction 19 into an output waveguide 20. The input and output waveguides 14 and 20 terminate at the chip edge and can be connected by suitable pigtailing to input and output optical fibres (not shown). In alternative embodiments, the substrate may be reversed, i.e. an initially single domain sample is used that is .oriented with the positive z-face as the top surface 11. In that case, slight modifications to the fabrication process described below may be necessary.

Unlike the conventional optical jmodulator of Figure 1, the substrate 12 of the optical modulator of the first embodiment comprises multiple ferroelectric domains in combination with a specially shaped ^'electrode structure to provide correction of thermal drift induced by the pyroelectric effect and other drift mechanisms such as stray electric fields or ion migration. , '

As illustrated, the substrate 12 has alternating inverted and non-inverted ferroelectric domains 36 and 34 respectively, separated by domain boundaries 35. In this embodiment, the ferroelectric domains are elongate and extend transverse to the principal optical path direction of the MZ interferometer arrangement in a striped fashion. More specifically, the domain boundaries extend perpendicular to the direction of extent of the interferometer arms 16 and 18. More generally, the domains should be sized and aligned so that the interferometer arms cross at least one, preferably several, domain boundaries, the significance of which will be appreciated by the discussions further below. The width of the domains may be chosen within a wide range. The domain width may be from about half the relevant path length of the device (e.g. up to 4 cm) down to one micron. In general, a smaller domain width will provide better cancellation of thermal drift. The primary limitations to using small domain sizes will be imposed by either the poling process used to create the inverted domains or the practicalities of fabricating a suitable matched electrode structure. The domain width shown in the figure is purely schematic, being selected for convenience of representation. It is also noted that the domains may be periodically spaced (e.g. generated by periodic poling) or non-periodically spaced. Indeed, aperiodic spacing of the domains may be beneficial to avoid coherent backscattering of light from the domain boundaries (or electrons in an electronic device, e.g. in a quantum electron device where the domain spacing is less than the electron mean free path).

Figure 3 is a schematic cross-section through the multi-domain substrate of Figure 2 with the positive and negative surface charge induced by the pyroelectric effect being shown with plus and minus symbols respectively.

As illustrated in the figure, the multi-domain nature of the ferroelectric substrate results in local cancellation of pyroelectric charge. The net pyroelectric charge on the top surface 11 of the substrate, where the electrodes are positioned, can thus be made zero by providing equal areas of inverted (up arrows) and non-inverted domains (down arrows) on a z-cut substrate. Similarly, the bottom surface 13 also benefits from the same charge cancellation effect. For example, if equal areas of the substrate are domain inverted and an electric field is applied in the z-direction, the electro-optic effect in the two orientations (up and down) will precisely cancel. The material will also develop a much smaller net pyroelectric charge on the whole surface because the different domain orientations will contribute charges of opposing signs resulting in both local and global charge cancellation.

It will be appreciated that the electro-optic effect will cancel out in any waveguide sections under uniform bias (provided that both inverted and non-inverted domains are crossed), since any induced refractive index changes or the like will be of opposite sign in the inverted and non-inverted domains. Undesired stray effects are thus also cancelled.

The substrate is made multi-domain from its original single domain state by selective domain inversion of parts of the substrate. Domain inversion can be performed in ferroelectric materials by a number of methods, which are commonly referred to as poling methods [Houe]. In the present embodiment, the domain inversion is generated by an applied electric field generally according to the method described in references [Amin, Hoffmann]. Reference is also made to [Yamaha, Myers, Webjorn]. A specific example relating to fabrication of a titanium diffused lithium niobate structure is now given:

Generally, fabrication of a titanium diffused lithium niobate waveguide according to this example is conventional, and as described for example in [Davis].

The fabrication process differs from the conventional process in that there is an additional poling step that is carried out after the titanium diffusion. This poling step is in itself conventional, and is also described in the literature [Amin, Hofmann].

More specifically, the process may proceed in the following stages:

1. Cleaning using standard acids and solvents (e.g. nitric acid, Ecoclear, acetone, isopropyl alcohol, water with detergent, water) then drying. Care must be taken to avoid thermal shock to the material otherwise spontaneous poling may occur.

2. Application 'of photoresist by spinning, followed by baking (examples are Shippley 1813 photoresist and baking for 30 minutes at 90 degrees centigrade, again care to avoid thermal shock is required).

3. Patterning and developing of the photoresist to define openings through the photoresist, these openings define the geometry of the waveguide - typically in a Mach-Zehnder structure openings of between 3 and 10 microns would be used. In some cases it is desirable to use a treatment to alter the surface layer of the photoresist, for example by soaking in nitro-benzene.

4. Deposition using sputtering or evaporation of a film of titanium under vacuum. Layer thicknesses of between 50 nm and 500 nm are normally suitable.

5. Lift-off of the photoresist and unwanted metal using a solvent such as acetone. This removes the resist leaving metallic tracks of titanium on the surface. 6. Diffusion of the titanium into the sample by placing the sample in a furnace. Examples of temperature cycles may be found in the literature, in particular at temperatures of around 1050 degrees centigrade for 10 hours in an inert atmosphere. In some processes, the final part of the diffusion step involves adding oxygen gas to the atmosphere to re-oxidise the material.

7. Repetition of steps 1 to 3, but with a different mask design placed in registry with the original design. This new mask has openings that correspond to the areas that will be periodically poled. The mask openings may be made slightly smaller than the desired poled dimensions to allow for domain spreading (typically 1 to 3 microns smaller at each boundary).

8. Application of an electrode material over the photoresist, for example, aluminium metal (on the positive z-face), or lithium chloride or an ECG (electrocardiogram) gel conductor such as used in medical practice (on the negative z- face). A planar electrode is applied on the opposing un-patterned z-face.

9. Application of a controlled high voltage pulse across the sample to cause domain inversion. The pulse duration is controlled to pass a specific charge. The charge flowing is given by twice the spontaneous polarisation multiplied by the area to be inverted, (71 micro-coulombs per square centimetre being the established value for lithium niobate). In some circumstances it may be necessary to adjust the amount of charge by as much as plus or minus 50% of the theoretical value, with such variation depending on the material source. In some situations it is desirable to use a current-controlled high voltage supply in which the poling current is matched to a specific profile by using current derived feedback.

10. Removal and cleaning of the sample followed by inspection of the domain quality under crossed polarisers.

The choice of crystal face for the process depends upon the electrode material used and on whether the lithium out-diffused layer on the positive z-face is removed from the sample by polishing after the diffusion process. In a particular implementation, the titanium diffusion is carried out on the negative z-face, and the periodic electrode using conducting gel as conductor is also placed on the negative z- face. The orientation of the poled regions should be matched to the preferred domain directions in the material. In particular, for lithium niobate, poled domains are preferentially formed along the y-direction and at 120 degrees to the y-direction.

After the poling step, the sample is processed in the conventional way for making modulators, namely by:

1. Deposition of a buffer layer such as silica;

2. Photolithography and deposition of electrodes;

3. Deposition of anti-reflection coatings to reduce coupling losses from fibre to waveguide;

4. Electrical connectorisation by wire bonding; 5. Attachment of optical fibres to couple light into and out of the waveguide (so-called pigtailing); and 6. Testing.

Having described fabrication according to a specific example, the novel electrode structure of the first embodiment is now described.

Figure 4 is a plan view of the modulator of the first embodiment showing more clearly the electrode structure arranged on the top surface 11 of the substrate. The electrodes 30 and 32 are arranged to cover the central section of one of the interferometer arms 16, and extend across several domain boundaries 35. As illustrated, the ground electrode 30 overlies the waveguide 16 for the inverted domains 36, whereas the signal electrode 32 overlies the waveguide 16 for the non- inverted domains 34, this being achieved by translating the electrodes transverse to the waveguide direction on crossing each domain boundary 35 in a kind of square- wave or zig-zag shape. The purpose of this translation is to invert the biasing on crossing each domain boundary. It will be understood that many other electrode geometries could be designed to provide the same functionality, i.e. to switch the biasing direction in the inverted and non-inverted domains. It will also be understood that the number of domain boundaries covered by the electrode structure may be varied widely, but at least one domain boundary should be crossed by the optical path underlying the electrode to provide the desired net charge cancellation.

In the present device that utilises the electro-optic effect, the electrode structure thus ensures that the electro-optic induced path length change in the inverted domains sums with the electro-optic induced path length change in the non-inverted domains. (If a conventional electrode structure was used, such as shown in Figure 1, the path length changes in the inverted and non-inverted domains would cancel, in which there would be no net electro-optic effect and the modulator would not function).

Figure 5 is a plan view of the electrode structure of an optical ;modulator according to a second embodiment of the invention. The optical modulator of the second embodiment is the same as that of the first embodiment, except for its electrode structure. By contrast to the first embodiment, where the electrode structure only covers one of the interferometer arms, the electrode structure of the second embodiment extends over the top surface of the substrate to cover both of the interferometer arms. More specifically, in the second embodiment, the electrode structure comprises an RF signal electrode 32 arranged between first and second ground electrodes 30 and 31. The electrodes 30, 31 and 32 are arranged to cover the central sections of both of the interferometer arms 16 and 18, and extend across several domain boundaries 35. As illustrated, in the inverted domains 36, the first ground electrode 30 overlies the first interferometer arm 16 and the RF signal electrode 32 overlies the second interferometer arm 18, with the second ground electrode 31 lying on the other side of the RF signal electrode 32 than the first ground electrode 30. For the non-inverted domains 34, the RF signal electrode 32 overlies the first interferometer arm 16 and the second ground electrode 31 overlies the second interferometer arm 18, with the first ground electrode 30 lying on the other side of the RF signal electrode 32 than the second ground electrode 32. As in the first embodiment, the electrodes are shifted transverse to the waveguide direction on crossing each domain boundary 35, thereby to invert the biasing on crossing each domain boundary 35. The electrode structure of the second embodiment constitutes a push-pull configuration which may be useful for reducing the drive voltages required. In such a push-pull configuration, the electric field in the waveguide core of one of the interferometer arms 16 or 18 causes an increase in refractive index and path length, whereas, in the waveguide core of the other interferometer arm 18 or 16, the electric field causes a decrease in refractive index and path length. There thus results a greater relative phase change between the two paths through the interferometer for a given drive voltage.

Here it is noted that a push-pull electrode configuration has been used in the past to increase RF bandwidth of modulators fabricated on single domain substrates [Alferness].

Figures 6 and 7 show the local operation of this push-pull electrode configuration in more detail. Figures 6 and 7 are sections X and Y through Figure 5. Figure 6 is thus a section through one of the inverted domains 36 and Figure 7 is a section through one of the non-inverted domains 34. In addition to the features familiar from the above-description, these figures also show an insulating layer 33 disposed between the top surface 11 of the substrate 12 and the electrodes 30, 31 and 32. The bottom surface 13 of the substrate 12 is also indicated. The dashed lines in the figures show the electric field lines at the same point in time. The solid arrow shows the ferroelectric domain orientation which is anti-parallel to the z-axis in the inverted domains and parallel to the z-axis in the non-inverted domains.

In the inverted domains (Figure 6), the electric field induced by the electrodes at an arbitrary instant in time (i.e. with arbitrary polarity) is oriented anti-parallel to the z-direction in waveguide core 16 and parallel to the z-direction in waveguide core 18. In the non-inverted domains (Figure 7), the electric field at the same arbitrary instant in time is oriented parallel to the z-direction in waveguide core 16 and anti- parallel to the z-direction in waveguide core 18.

The reversal of the domain direction with the corresponding lateral shift in the electrode position thus results in the electric field being applied to the waveguide cores 16 and 18 with the same relative polarity in both the inverted and non-inverted domains. In other words, the electric field applied along the optical path concerned (path of waveguide 16 or 18) is applied in the same sense relative to the domain alignment direction. This is achieved by configuring the electrodes so that the applied electric field direction is reversed at each domain boundary to follow the reversal in domain alignment direction.

Figure 8 shows in plan view down onto the top surface 11 of the substrate an alternative arrangement of the ferroelectric domains applicable to various embodiments of the invention, including the above-described first and second embodiments. In this alternative arrangement, the domain boundaries 35 occurring between the non-inverted and inverted regions 34 and 36 are not set normal to the optical path, i.e. the direction of light propagation, but rather are set at an oblique angle α thereto. The oblique angle is chosen so that the back-reflections (arrows in the figure) occurring at the domain boundaries are not coupled into the waveguide core 116. This is achieved when the oblique angle exceeds the numerical aperture of the waveguide.

Figures 9A and 9B are corresponding plan views down on to the top surface 11 of the substrate of an optical modulator 10 according to a third embodiment of the invention. In Figure 9A, the electrodes are not shown to allow clearer appreciation of the geometric arrangement of the domains relative to the waveguide sections of the modulator. In Figure 9B, the electrodes are shown. The modulator 10 of the third embodiment is fabricated in a z-cut lithium niobate substrate 12 and includes a MZ interferometer of conventional waveguide architecture. Namely, the interferometer comprises first and second optical waveguides 16 and 18 that constitute first and second interferometer arms respectively. The waveguides 16 and 18 are formed in the substrate 12 and extend generally parallel to each other over their central portions. The interferometer further includes an input waveguide 14 that leads to an input Y- junction 15 which serves to divide an optical input signal into the first and second interferometer arms 16 and 18. The interferometer arms 16 and 18 rejoin at an output Y-junction 19 into an output waveguide 20. The input and output waveguides 14 and 20 terminate at the chip edge (not shown) and can be connected by suitable pigtailing to input and output optical fibres.

As in the above-described embodiments, the substrate has a plurality of alternating non-inverted 34 and inverted 36 ferroelectric domains separated by domain boundaries 35. The inverted domains are shown with cross-hatching in the figure and are created by poling, as described above. However, the third embodiment differs from the first and second embodiments in that the ferroelectric domains are aligned generally with the optical path of the interferometer, rather than across it.

This has the advantage of allowing a conventional electrode structure to be used for biasing the device, as shown in Figure 9B. The electrode structure comprises an RF signal electrode 32 arranged between first and second ground electrodes 30 and 31. The electrodes 30, 31 and 32 are generally straight over the active portion of the device, extending in the same direction as the central waveguide sections of the interferometer arms 16 and 18. With such conventional straight electrodes in combination with the co-extending ferroelectric domain boundaries, the pyroelectric charge cancellation occurs laterally, as will be appreciated with reference to Figure 3. The electrode structure thus covers a path in the substrate, namely the optical path of the interferometer arm 16, that extends generally parallel to the domain boundaries 35, rather than across them. In this way, the electrode structure is arranged so as to provide a unidirectional biasing for the path, as in a conventional electrode arrangement.

This embodiment may thus be thought of in terms of the relative alignment of the domain boundaries with either the electrode arrangement or the waveguide(s) that define the relevant optical path. Figure 10 is a plan view down on to the top surface 11 of the substrate of an optical modulator 10 according to a fourth embodiment of the invention. The modulator 10 is fabricated in a z-cut lithium niobate substrate 12 and includes a MZ interferometer of conventional waveguide architecture, as previously described in relation to the preceding embodiments using reference numerals 14-20. The non- inverted domains 34 are shown with plus symbols and the inverted domains 36 with minus symbols, and are separated by domain boundaries 35.

In this embodiment, the ferroelectric domains dispersed in a two-dimensional arrangement in a granular fashion to allow charge cancellation. The local electrode structure (not shown) is designed to follow the reversals of domain polarity (as in the first and second embodiments). ' As in the second embodiment, a push-pull configuration may be used to reduce drive voltage.

In general, any regular or irreeular domain pattern designed so that in any local area an approximately equal quantity, Of up and down domains exists will be suitable for embodying the invention. A correctly designed electrode will also be needed for any device embodying the invention which involves a physical effect that needs electrical biasing for inducing it (e.g. electro-optic, piezo-electric).

Figure 11 is a schematic plan view of an active arrayed waveguide grating (AWG) according to a fifth embodiment of the invention. The device has conventional architecture comprising an array of waveguides 100 interconnecting first and second free space propagation regions 102 and 104, successive waveguides of the array 100 having an incrementally increasing optical path length, as is known in the art. In input waveguide 110 couples into the first free space propagation region 102 and a plurality of output waveguides 108 couple out from the second free space propagation region 104. The waveguide array 100 has arranged thereon a trapezoidal electrode 106 biased in use by a voltage V for imposing a linear phase profile on the waveguide array 100, thereby to tune the device so that a given input wavelength from the input waveguide 110 can be coupled to different ones of the output waveguides 108. To reduce thermal drift in the AWG, specifically in the waveguide array 100, the AWG is provided with multiple domains, as in the previous embodiments. The AWG of this embodiment is made in GaAs via structured growth [Ebert] to selectively produce polar twinned regions at the electrode 106. Figure 12 shows in more detail the electrode region 112 ringed in Figure 11.

As illustrated by the hatching, the structured growth has been used to provide polar twinned or antiphase regions (inverted domains) 36 separated from normal phase regions (non-inverted domains) 34 by boundaries 35. The antiphase regions 36 are arranged coincident with the waveguide sections traversing the electrode 106. In terms of the relative arrangement of electrodes, domains and waveguides, this AWG embodiment may thus be compared with the modulator embodiment of Figure 9 A.

In further AWG embodiments, the domains may be arranged in elongate strips arranged to cross the waveguides (see Figure 2), or in a two-dimensional arrangement (see Figure 10). Further, as an alternative to structured growth, wafer bonding could be used

[Yoo]. Moreover, other alternative materials which could be considered for fabricating an AWG embodying the invention are poled silica or poled polymer.

The above described embodiments of the invention may be implemented in a variety of materials. As well as lithium niobate, further examples of suitable materials are lithium tantalate, sodium barium niobate, strontium barium niobate, potassium titanyl niobate (KTN), potassium titanyl phosphate, rubidium titanyl arsenate, isomorphs of KTP, RTA, barium titanate, potassium titanate, GaAs, InP, quartz, poled silica and poled polymer. Alloys based on these compounds may also be used. In addition, any of the preceding materials may include dopants such as titanium, zinc, erbium, neodymium, ytterbium, holmium, barium, cerium, rubidium, magnesium, magnesium oxide or iron.

It will also be understood that references to substrate in the present document follow usage in the art, but should not be construed to exclude epitaxial structures. For example, the multi-domain structure may be constructed in a ferroelectric or polar epitaxial layer on a non-ferroelectric or non-polar substrate. Further, the domain inversion may be imposed only in an upper part or layer of a thicker block of material which may be a buried layer or a surface layer, for example.

Moreover, it will be appreciated that for embodiments comprising waveguides, the waveguides may be formed with a variety of fabrication techniques including proton exchange, zinc-indiffusion, titanium diffusion, direct bonding, strip loading and physical machining, direct bonding. In the case of antiphase, i.e. twinned, polar structures, these may be fabricated by molecular beam epitaxy, optionally in combination with self-assembly or nano-manipulation with scanning probe microscopes. The invention may also be applied to free-space devices, and is expected to be particularly beneficial for miniaturised free-space devices.

Further, the inversion may be performed by various methods such as electrical poling, orientated;.;growth, direct bonding, poling by diffusion, thermal poling or optical poling. It will also 'be appreciated that the above-described techniques based on multi- domain substrates; are compatible with prior art conductive path techniques such as [Heisman]. It is therefore possible to combine both techniques in a single device, as

^' .1 ' may be advantageous.

REFERENCES

[Alferness] R.C. Alferness US patent 4,448,479 "Travelling wave, electrooptic devices with effective velocity matching" May 15, 1984

[Amin] Amin J, Pruneri, V, Webjorn J, Russell PS, Hanna DC, Wilkinson JS, "Blue light generation in a periodically poled Ti:LiNbO3 channel waveguide", OPTICS COMMUNICATIONS 135: (1-3) 41-44 FEB 1 1997

[Burfoot] J.C. Burfoot and G.W. Taylor, "Polar dielectrics and their applications", ISBN 0333179587, 1979

[Davis] Davis, Christopher C, "Lasers and electro-optics : fundamentals and engineering" ISBN: 0521308313, 1996

[Ebert] Ebert C B, Eyres L A, Fejer M M, Harris J S "MBE growth of antiphase GaAs films using GaAs/Ge/GaAs heteroepitaxy" Journal of Crystal Growth, vol. 202, pages 187-193, May 1999

[Heisman] Heismann, Korotky, Veselka, 1995 (AT&T Corporation), US .patent number 5,388,170 "Electrooptic device structure and method for reducing thermal effects in optical waveguide modulators"

[Hofmann] Hofmann D, Schreiber G, Haase C, Herrmann H, Grundkotter W, Ricken R, Sohler W, "Quasi-phase-matched difference-frequency generation in periodically poled Ti : LiNbO3 channel waveguides", OPTICS LETTERS 24: (13) 896-898 JULY 1, 1999 [Houe] "An introduction to methods of periodic poling for 2nd-harmonic generation", Houe M, Townsend PD, Journal of Physics D-Applied Physics, 28: (9) 1747-1763 SEP 14 1995

[Lines and Glass] "Principles and applications of ferroelectrics and related materials" by M.E. Lines and A.M. Glass, ISBN 0198512864., 1977

[Myers] Myers L.E., Eckardt, R.C., Fejer M.M., Byer, R.L., Bosenberg W.R., Pierce J.W. "Quasi phase matched optical parametric oscillators in bulk periodically poled PPLN", J. Opt. Soc. Am B, Vol 12, No. 11, November 1995

[Schaffner], US Patent 5,278,924 "Periodic domain reversal electro-optic modulator", ',,1994

I [Sturman] Sturman B, Aguilar M, AguUoLopez F, Pruneri V, Kazansky PG, ι"Photorefractive nonlinearity of periodically poled ferroelectrics" JOURNAL OF

JTHE OPTICAL SOCIETY OF AMERICA B-OPTICAL PHYSICS 14: (10) 2641-

1

-2649 OCT 1997

[Webjorn] "Quasi-phase-matched blue-light generation in bulk lithium-niobate, electrically poled via periodic liquid electrodes", WEBJORN J, PRUNERI V, RUSSELL PS, BARR JRM, HANNA DC, ELECTRONICS LETTERS 30: (11) 894- 895 May 26, 1994

[Yamada] Yamada M, Nada N, Saitoh M, Watanabe K "1st order quasi-phase matched LiNbO₃ waveguide periodically poled by applying an external field for efficient blue 2nd harmonic generation" Applied Physics Letters vol. 62, no. 5, pages 435-436 (1993) [Yoo] Yoo S J B, Bhat R, Caneau C, Koza M A "Quasi-phase matched 2nd harmonic generation in AlGaAs waveguides with periodic domain inversion achieved by wafer bonding" Applied Physics Letters, vol. 66, pages 3410-3412, June 1995

Claims

1. A device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; and an electrode structure for biasing the device, the electrode structure covering a path in the substrate that extends across at least one of the domain boundaries, and being arranged so as to provide an inversion of the biasing on crossing the at least one of the domain boundaries.

2. A device according to claim 1, wherein the domains are elongate, extending generally transverse to the path.

3. A device according to claim 2, wherein the domains are arranged so as to extend at an oblique angle to the path.

4. A device according to claim 1, wherein the domains are granular and dispersed in a two-dimensional arrangement.

5. A device according to any one of the preceding claims, wherein the path is an optical path.

6. A device according to claim 5, wherein the optical path is defined by a waveguide.

7. A device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; and an electrode structure for biasing the device, the electrode structure covering a path in the substrate that extends generally parallel to at least one of the domain boundaries, and being arranged so as to provide a unidirectional biasing for the path.

8. A device according to claim 7, wherein the path is an optical path.

9. A device according to claim 8, wherein the optical path is defined by a waveguide.

10. A device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; and a path including a waveguide that extends generally parallel to at least one of the domain boundaries.

11. A device according to any one of claims 7 to 10, wherein the domains are elongate extending generally parallel to the path.

12. A device according to any one of the preceding claims, wherein the device is an optical modulator.

13. A device according to claim 12, wherein the optical modulator is a bulk modulator.

14. A device according to claim 12, wherein the optical modulator is a waveguide modulator.

15. A device according to claim 14, wherein the optical modulator is a Mach- Zehnder interferometric modulator.

16. A device according to any one of claims 1 to 11, wherein the device is an arrayed waveguide grating.

17. A device according to any one of the preceding claims, wherein the device utilises at least one of the following effects: pyroelectric, electrooptic, piezoelectric, inverse piezoelectric, photorefractive and surface acoustic waves.

18. A Mach-Zehnder optical modulator device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; a waveguide arrangement comprising an input section that divides into first and second arm sections that combine into an output section, wherein the first and second arm sections extend across at least one of the domain boundaries; and an electrode structure for biasing the device, the electrode structure covering at least one of the first and second arm sections, and being arranged so as to provide an inversion of the biasing on crossing the at least one of the domain boundaries.

19. A device according to claim 18, wherein the domains are elongate, extending generally transverse to the first and second arm sections.

20. A device according to claim 19, wherein the domains are arranged so as to extend at an oblique angle to the first and second arm sections.

21. A device according to claim 18, wherein the domains are granular and dispersed in a two-dimensional arrangement over the first and second arm sections.

22. A Mach-Zehnder optical modulator device comprising: a substrate with a plurality of alternating inverted and non-inverted domains separated by domain boundaries; a waveguide arrangement comprising an input section that divides into first and second arm sections that combine into an output section, wherein the first and second arm sections extend generally parallel to at least one of the domain boundaries; and an electrode structure arranged to provide unidirectional biasing of at least one of the first and second arm sections.

23. A device according to claim 22, wherein the domains are elongate, extending generally parallel to the first and second arm sections.

24. A device according to any one of the preceding claims, wherein the substrate is a z-cut of a crystal having a 3m point group.

25. A device according to claim 24, wherein the crystal is lithium niobate.

26. A device according to any one of the preceding claims, wherein the domains are ferroelectric domains.

27. A device according to any one of claims 1 to 23, wherein the domains are non- ferroelectric polar domains.