GB2365987A

GB2365987A - An electro-optic device

Info

Publication number: GB2365987A
Application number: GB0019772A
Authority: GB
Inventors: Adrian Petru Vonsovici
Original assignee: Bookham Technology PLC
Current assignee: Lumentum Technology UK Ltd
Priority date: 2000-08-11
Filing date: 2000-08-11
Publication date: 2002-02-27
Anticipated expiration: 2020-08-11
Also published as: AU2001276553A1; GB0019772D0; GB2365987B; WO2002014936A1

Abstract

An electro-optic device for controlling the phase shift between first and second waveguide paths (16A,16B) with pairs of doped regions (24, 25; 26, 27; 24, 28; 25, 29) of opposite polarity being provided on opposite sides of the respective waveguide paths (16A, 16B) for altering the refractive index of the waveguide paths by altering the density of charge carriers within the waveguide paths when a voltage is applied. Each waveguide path (16A, 16B) is provided with two pairs of doped regions, the polarity of the first pair (24, 26; 24, 28) being opposite to the polarity of the second pair (25, 27; 25, 29) and the first and second pairs being electrically connected in series. Each of the doped areas (24, 25, 26, 27, 28, 29) is subject to substantially the same misalignment with the respective waveguide path (16A, 16B), whereby a path imbalance induced between the first and second waveguide paths (16A,16B) due to the first pair of doped regions (24, 26; 24, 28) of each path is counteracted by a path imbalance in the opposite direction due to the second pair (25, 27; 25, 29) of doped regions of each path.

Description

2365987 AN ELECTRO-OPTIC DEVICE This invention relates to an electro-optic

device for controlling the phase shift between first and second waveguide channels, eg in a Mach- Zehnder (MZ) type modulator using a lateral injection p-i-n structure in both arms of the modulator, and to a method of driving the device.

A known waveguide Mach-Zehnder (MZ) modulator is shown schematically in Figure 1. The modulator has two inputs 1 and 2 and two outputs 3 and 4. A splitter 5 divides signals received on inputs 1 and 2 between two waveguide paths (PI and (p2 and a combiner 6 combines the outputs of the two paths and transmits them to outputs 3 and 4. The splitter 5 and combiner 6 may be 3-port devices, eg Yjunctions, or 4-port devices, eg directional or multimode interference couplers.

Carrier injection phase modulators, using diodes and transistors are known in Mach-Zehnder type devices. One known arrangement involves injection of charge carriers into the core of a rib or strip-loaded waveguide from a lateral p-i-n diode. The central intrinsic region of the diode coincides with the waveguide core and the two doped regions (p+, n+) are at either side of the central waveguide.

Figure 2 shows a rib waveguide structure formed on a silicon-on-insulator chip used for Mach-Zehnder intensity modulators. This structure comprises a rib 7A formed in a silicon layer 7 which is separated from a silicon substrate 8 by a layer of silicon dioxide 9. An n-doped region 10 is provided in the silicon layer 7 on one side of the rib 7A and a p-doped region 11 in the silicon layer 7 on the other side of the rib 7A. Metal layers 12 are provided to provide electrical connection to the doped regions 10 and 11 and a layer 13 of silicon dioxide is formed over the device. Further details of such a device are given in W095/08787 and an alternative form of such a device is described in WOOO/10039.

The doped regions 10 and 11 are typically phosphorous (n') and boron (p"), though any p-type or n-type dopant can be used at a concentration sufficient to form an ohmic contact and provide good injection efficiency. The doped regions 10 and 11 are placed laterally at a sufficient distance from the waveguide ridge 7 to avoid significant optical loss due interaction of the optical mode with the doped regions. This distance varies with the c,-.---entration of dopant and also determines the injection efficiency and therefore the modulation efficiency of the diodes.

The doped regions 10 and 11 and the rib waveguides 7A are defined at different steps of the fabrication process, which leads to potential alignment errors of the doped regions 10 and 11 with respect to the waveguide. The position of the waveguide is defined using a first lithographic mask (that will be named waveguide mask) followed by subsequent etching. The positions of the two doped regions are defined at another stage of fabrication by forming two doping holes separated by a distance D using another mask (that will be named doping mask). In the case of a perfect alignment, a line passing at mid-distance between the doped regions 10 and 11 should coincide with the waveguide centreline W. However, due to misalignment between the two masks during the fabrication process, there is an offset between the waveguide centreline W and a line R passing mid-distance between the two doped regions 10 and 11. The centre of the waveguide 7A will therefore be situated at distances d, and dp relative to the doped regions 10 and I I and d, D / 2 and dp # D / 2. In the case of a perfect alignment between masks, d,, = dp D / 2.

As will be explained further below, this misalignment gives rise to a path imbalance between the two paths of the MZ interferometer. The present invention seeks to overcome, or at least substantially reduce, this problem.

According to a first aspect of the invention, there is provided an electro-optic device for controlling the phase shift between first and second waveguide paths, pairs of doped regions of opposite polarity being provided on opposite sides of the respective waveguide paths for altering the refractive index of the waveguide paths by altering the density of charge carriers within the waveguide paths when a voltage is applied between the doped areas, a path imbalance between the two waveguide paths arising due to misalignment of the doped regions with the waveguide paths, wherein each waveguide path is provided with two pairs of doped regions, the polarity of the first pair being opposite to the polarity of the second pair and the first and second pairs being electrically connected in series, each of the doped areas being subject to substantially the same misalignment with the respective waveguide path, whereby the path imbalance induced between the first and second waveguide paths due to the first pair of doped regions of each path is counteracted by a path imbalance in the opposite direction induced between the first and second waveguide paths due to the second pair of doped regions of each path.

According to another aspect of the invention there is provided a particularly advantageous method of driving a preferred embodiment of the electro-optic device.

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

The invention will now be further described, merely by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic view of a known 4-port MZ interferometer structure; Figure 2 is a schematic cross-sectional view of a known form of lateral p- i-n diode used in a MZ interferometer; Figure 3 and 4 are plan views of known forms of MZ interferometer using p- i-n diodes of the type shown in Figure 2; Figure 5 is a plan view of a MZ interferometer using p-i-n diodes according to one embodiment of the present invention; and Figure 6 is a further plan view of the device shown in Figure 5 illustrating the potential misalignment of the doped regions and waveguides thereof.

Figures 1 and 2 have been described above so will not be described further.

Figure 3 shows a plan view of a Mach-Zehnder interferometer using p-i-n diode phase modulators. The figure shows an input waveguide 14 a splitter in the form of a Y-junction 15, two substantially parallel waveguide paths 16A and 16B and a combiner in the form of a further Y-junction 17 leading to an output waveguide 18. A central n+ doped region 19 is provided between the arms 16A and 16B of the Mach-Zehnder and two p+ doped regions 20 and 21 on the opposite sides of the arms 16A and 16B. A first phase modulator is thus formed across arm 16A and a secono phase modulator is fc.7-med across the arm 16B.

As for any interferometric device, the relative phase and imbalance in power between the two arms is crucial for a good performance of the device. In practice, a phase error (or static phase shift) between the arms is introduced by imperfections in the fabrication process of lateral p-i-n diode structures as the alignment of the doped regions with the central rib waveguide has a certain tolerance. This results in differences between the effective indices of the two waveguides 16A and 16B due to the difference in positioning of the p+ and n+ doped regions 19, 20 and 21 with respect to the central rib waveguide. This is because the doped regions have different refractive indices from intrinsic silicon and their positioning relative to the waveguide core influences the mode field and the effective index of the guided optical mode.

This can be an important cause of path imbalance for the Mach-Zehnder interferometer. The path-imbalance (PI) is defined as:

pj p(2)) Lph = ko - (') - n (2)) Lph -27 - ( (1) - n (2)) L eff eff k eff eff ph 0 where ko is the wavelength, P('), n") and n(') are the propagation constants eff eff and effective indices of the first and second phase modulator respectively, and Lph is the length of the two phase modulator sections (as shown in Figure 3).

One way to compensate for the path-imbalance due to the misalignment of the doped regions is to replace the central n+ region 19 with two separate n+ and p+ contacts 22 and 23 as shown in the device illustrated in Figure 4. However, a disadvantage of this is that the distance between the arms 16A and 16B of the MZ interferometer is doubled. The length of the device is also increased compared to the device shown in Figure 3, as the curved sections of the splitters 15 and combiner 17 must be at least twice as long in order to maintain the same propagation loss.

The dotted lines in Figures 3 and 4 show the positions of metal contacts which provide electrical connections to the doped regions. The equivalent circuit diagram for the p-i-n diodes is also shown in a box next to each of the Figures.

However, it should be noted that whilst the arrangement shown in Figure 4 compensates for systematic misalignments of the mask for the p/n layers, it does not compensate for random variations of their positioning due to resist shrinkage variations.

Figure 5 shows a device according to one embodiment of the invention in which a two-section diode is used in each of the arms 16A and 16B of the MZ interferometer. As will be described below, this arrangement combines the advantage of a reduced separation of the waveguides 16A and 16B by using a common central electrode and provides automatic compensation of the PI due to the misalignment of the doped regions relative to the rib waveguide centre.

The arrangement shown in Figure 5 is similar to that of Figure 4 except that two pairs of doped regions of opposite polarity are provided for each of the paths 16A and 16B, ie each of the doped regions shown in Figure 4 are replaced by two doped regions of opposite polarity. Thus, as shown in Figure 5, a pair of diodes D1 and D3 of opposite polarity are formed across the first waveguide path 16A by a central n-doped region 24 and a central p-doped region 25 with a p-cloped region 26 and an n-cloped region 27 provided on the opposite side of the waveguide 16A and a pair of diodes D2 and D4 of opposite polarity are formed across the second waveguide 16B by the central n-doped region 24 and the central p-doped 25 region with a p-doped region 28 and an n-doped region 29 on the opposite side of the second waveguide 16B. The central n-doped region 24 and the central pdoped region 25 are thus each common to the one of the diodes of the two waveguide paths, ie n-doped region 24 is common to diode D1 and D2 and pdoped region 25 being common to diodes D3 and D4.

The diodes D1 and D3 are connected in series by an electrical connection between the n-doped region 24 and the p-doped region 25. The diodes D2 and D4 are similarly connected in series by the same electrical connection. The dotted lines again show the positions of metal contacts which provide electrical connections to and between the doped regions and the equivalent circuit diagram for the diodes is shown in a box next to Figure 5. As this diagram shows, the first pair of diodes D1 and D3 is connected in series between an anode 1 and a cathode 3, the second pair of diodes D2 and D4 is connected in series between anode 2 and cathode 4 and the common electrical connection between the two pairs of diodes is labelled as electrode 5.

The diodes D1-D4 each have the same length, this being defined by the doping mask.

As indicated above, the diodes DI and D3 are in series and when a positive voltage bias is applied between the electrodes 1 and 3 they inject current into the first arm 16A of the Mach-Zehnder. Similarly, the diodes D2 and D4 inject current into the second arm 16B when a positive voltage bias is applied between electrodes 2 and 4. Each arm of the MZ interferometer is thus in effect provided with a two-stage p-in diode.

The separation between the two arms 16A and 16B may be made equal to that for the device shown in Figure 3.

Figure 5 shows the splitter and combiner as Y-junctions but other types of splitter or combiner (eg directional or multimode interference couplers) may be used instead.

Figure 6 illustrates the potential misalignment (shown exaggerated) between the waveguide mask and the doping mask. The dotted lines indicate the position of the doped regions if there was no misalignment. For clarity the electrical contacts are omitted from this Figure.

If 8n"ffP" 5n("lare the effective index changes for a given offset of the p+ and n+ eff regions, respectively, the induced PI between the arms 16A and 16B is:

PI = ko - [(n(P), 6,(n)) Lph n(P) + 8n(")) Lph 0 (2) eff eff 2 eff eff 2 Thus, with this structure, the PI induced by the misalignment of the doped regions relative to the waveguide is automatically compensated. As a comparison, the path imbalance for a device such as that shown in Figure 3 may typically be about 700 at 1.55pm for a 2gm misalignment between masks.

Further details of a device comprising pin diodes in which the doped regions on each side of the waveguide comprise a plurality of doped areas are given in co pending application No relating to an electro-optic device for altering the density of charge carriers within an integrated optical waveguide filed on 11 August 2000.

Another advantage for the dual-stage diode stems from the analysis of the modulation efficiency of the lateral diode as a function of the offset between the doped regions and the centre of the waveguide. As the injected carrier profile is not uniform across the waveguide (it typically has an exponential form) there will be a variation of the injection efficiency as a function of this offset. However, for a twostage diode with a misalignment error each section will have opposite efficiency changes. These tend to compensate for each other so there will be less dispersion of the injection efficiency compared to the arrangements shown in Figure 3 and 4. This improves the uniformity on a wafer scale of the power efficiency of the devices.

The modulator described above may be driven so as to inject carriers in the first arm 16A of the modulator to induce a 7r-phase shift between arms (the OFF-state). This is achieved by driving D1 and D3 using the following combination of voltage biases (electrodes 2,4,5 are commoned and floating):

1. -ON-state (0-phase shift between arms): V1=0, V3=0(GND) 2. -OFF-state (7E-phase shift between arms): Vl=2Vf, V3=0(GND) Vf is the forward bias on one diode for which the injected carriers induce an effective index shift equal to:

AnM;o (3) eff - 2 - Lph Assuming that D1 and D3 are identical, the total phase shift induced in the arm 1 is:

A(p A(p + A(p "' = 2 - ko - AnM. Lph = 7T (4) 3 eff 2 However, the modulator may also be driven in a particularly advantageous manner with electrodes 1 and 4 commoned and electrodes 2 and 5 commoned using the following biases:

1. - ON-state (0-phase shift between arms): V4=Vl=O, V3=0(GND), V2=V5=Vf2. - OFF-state (Ti-phase shift between arms): V4=Vl=2Vf, V3=0(GND), V5=V2=VfAs before, when a forward bias equal to Vf is applied to one of the diodes 1-4, the induced effective index shift of the waveguide mode is given by equation (3).

It should also be noted that with this arrangement the device is driven using just one driving electrode (electrode 1 which is commoned with electrode 4).

This method of driving (referred to herein as quasi-push-pull) has the advantage of switching just half of the charge stored from one arm to the other in order to obtain the 7t-phase shift. This will improve significantly the modulation bandwidth of the modulator compared to the MZ with a central n+ region. This is due to the fact that the switching riseand fall-time for the phase change depends on the amount of maximum charge concentration to be injected or evacuated respectively. This driving method requires the injection of half of the charge needed for the same phase shift difference between the arms of the MZ.

This type of operation also provides correction of small path imbalances that may arise due to the random resist shrinkage. By slightly varying the bias on the electrode 5 (commoned with electrode 2), the operating point can be adjusted for a perfect balance between the arms.

Tables 1 a and 1 b shown below summarise the effective index variation for each diode and for the two arms in the case of OFF- and ON-states.

Table 1. Effective index variations for D1-D4 and the phase shift between the two arms in the OFF(I a) and ON(l b) states for quasi-push-pull driving.

OFF-state D1 D3 Total arm I Effective index 0 An() /2 AnM /2 change eff eff D2 D4 Total arm 2 Effective index 0 An(') /2 An(') / 2 change eff eff Phase difference AY (1) - A(p (2) 0 between arms 1 a) I ON-state D1 D3 Total arml Effective index An(')/2 AnM /2 AnM change eff eff eff D2 D4 Total arm2 Effective index 0 0 0 change Phase difference A(p A(p (2) 7Z between arms 1 b) A driving system (not shown) is thus preferably provided to drive the diodes in the manner described above. To achieve this, the anode 1 and cathode 4 are electrically connected to receive a signal and the anode 2 and the electrode 5 are electrically connected to receive a voltage bias. In a first state, ie the ON-state referred to above, a reference potential, or ground, is applied to anode 1, cathode 4 and cathode 3, and a forward bias Vf is applied to anode 2 and electrode 5. In a second state, ie the OFF-state referred to above, a forward bias of substantially 2Vf is applied to anode 1 and cathode 4, whilst the reference potential on anode 3 is maintained, and the forward bias of Vf is maintained on anode 4 and electrode 5. In practice, the forward bias applied to anode 1 and cathode 4 may not be exactly 2Vf due to the small differences between fabricated diodes.

Another important advantage of the quasi-push-pull driving method is thermal stability as the power driven through the circuit for the OFF and ON states is constant.

In the example described above, the doped regions are placed laterally at sufficient distance from the waveguide rib to avoid significant optical loss due interaction of the optical mode with the doped regions. However, these lateral doped regions could be part of a more complex structure with three or more doped regions placed in a variety of different geometries as known in the art. For example, one doped region could be on top of the waveguide rib in a three terminal device and the lateral regions positioned in the manner described above to compensate automatically for any path imbalance due to the misalignment of the lateral doped regions with the waveguide.

The arrangement described above may also be extended to provide an even number of pairs greater than two of doped regions for each waveguide path, eg four or six pairs for each path, to improve the linearity of the device and to achieve lower power consumption.

Although it is preferred for common doped regions to be used between the two waveguide paths, in some arrangements separate doped areas may be provided, ie as shown in Figure 4 but with each doped region comprising two doped regions of opposite polarity. Such an arrangement would thus comprise four diodes as in Figure 5 with D1 and D3 connected in series and D2 and D4 connected in series but without the common connection between these two pairs of diodes.

As described above, the device preferably comprises silicon rib waveguides formed on a silicon-on-insulator chip. However, it is also applicable to other waveguide arrangements, eg strip-loaded waveguides.

A two channel MZ interferometer has been described above but it will be appreciated that the invention is applicable to an interferometer with any number of channels or to control the phase shift between two waveguide channels in other forms of device.

The arrangement described above thus uses two-stage lateral p-i-n diodes to compensate automatically for the path imbalance for a Mach-Zehnder modulator due to fabrication errors and reduces to half the separation between arms in comparison to the previous compensated MZ structures (eg as shown in Figure 4). The two-stage p-i-n diodes can also be operated in a quasi-push-pull regime with increased switching speed and thermal stability of the chip.

Claims

1 An electro-optic device for controlling the phase shift between first and second waveguide paths, pairs of doped regions of opposite polarity being provided on opposite sides of the respective waveguide paths for altering the refractive index of the waveguide paths by altering the density of charge carriers within the waveguide paths when a voltage is applied between the doped areas, a path imbalance between the two waveguide paths arising due to misalignment of the doped regions with the waveguide paths, wherein each waveguide path is provided with two pairs of doped regions, the polarity of the first pair being opposite to the polarity of the second pair and the first and second pairs being electrically connected in series, each of the doped areas being subject to substantially the same misalignment with the respective waveguide path, whereby the path imbalance induced between the first and second waveguide paths due to the first pair of doped regions of each path is counteracted by a path imbalance in the opposite direction induced between the first and second waveguide paths due to the second pair of doped regions of each path.

2. An electro-optic device as claimed in claim 1 in which the first and second waveguide paths form part of a Mach-Zehnder interferometer.

3. An electro-optic device as claimed in claim 1 or 2 in which the first and second waveguide paths are substantially parallel and a common doped region provided between the waveguide paths forms one part of one of the pairs of doped regions of each of the waveguide paths.

4. An electro-optic device as claimed in claim 1, 2 or 3 in which each pair of doped regions form a pin diode across a respective waveguide path.

5. An electro-optic device as claimed in any preceding claim in which the waveguide paths comprise rib waveguides.

6. An electro-optic device as claimed in any preceding claim formed on a silicon-on-insulator chip.

7. An e' tro-optic device as claimed in any preceding claim in which the doped regions form two pairs of diodes, a first pair formed across the first waveguide path and the second pair formed across the second waveguide path, the diodes of the first pair being connected in series by a first electrical connection and the diodes of the second pair being connected in series by a second electrical connection.

8. An electro-optic device as claimed in claim 7 in which the first and second electrical connections are electrically connected or formed by a common electrical connection.

9. An electro-optic device as claimed in claim 8 in which the first pair of diodes are connected in series between an anode 1 and a cathode 3, the second pair of diodes are connected in series between an anode 2 and a cathode 4 and the common connection is designated electrode 5, anode 1 and cathode 4 being electrically connected and anode 2 and electrode 5 being electrically connected.

10. An electro-optic device as claimed in claim 9 with drive means arranged to apply a reference potential, or ground, to anode 1, cathode 4 and cathode 3 and to apply a forward bias Vf to anode 2 and electrode 5 in a first state and to apply a forward bias of substantially 2Vf to anode I and cathode 4, whiist maintaining the reference potential on anode 3 and the forward bias of Vf on anode 4 and electrode 5 in a second state.

11. An electro-optic device substantially as hereinbefore described with reference to Figures 5 and 6 of the accompanying drawings.

12. A method of driving an electro-optic device as claimed in claim 9 in which cathode 3 is held at ground and, in a first state, anode 1 and cathode 4 are also connected to ground whilst a forward bias of Vf is applied to anode 2 and electrode 5, and, in a second state, anode 1 and cathode 4 are connected to a forward bias of substantially 2Vf whilst the forward bias of Vf is maintained on anode 2 and electrode 5.

13. A method of driving an electro-optic device substantially as hereinbefore described with reference to the accompanying description.