Polarization independent broad wavelength band optical switches/modulators

[00002] This application claims priority to U.S. Provisional Application Ser. Nos. 60/204,774 and 60/204,775, both filed May 17, 2000, whose entire disclosure is incorporated herein by reference.

[00003] 1. Field of the Invention

[00004] The present invention relates to electro-optic devices, and in particular, to polarization independent broadband optical modulators and switches for wideband fiberoptic networks.

[00005] 2. Background of the Related Art

[00006] A balanced bridge optical switch has two input and two output waveguides. Two ½l_c (half a coupling length) 3-dB (i.e., 50:50 power splitter) directional couplers and a phase modulated interferometer waveguide pair are used. Note that “l_c” is the characteristic coupling length of a directional coupler, which is the length of the directional coupler necessary to transfer substantially all the power from a first waveguide to a second waveguide. The input 3-dB coupler is used to equally divide an input signal received by an upper input waveguide into both the upper and lower waveguides prior to entering an interferometer section of the balanced bridge optical switch. An output ½l_c directional coupler rejoins the electro-optical phase modulated signal from the interferometer section back into the upper output waveguide for a “straight-thru” path (i.e., “on” or (=) bar state) or the lower output waveguide in the “cross-over” path (i.e., “off” or switched (×) state). If the optical path lengths of the two waveguides are identical in the middle interferometer section, the two optical waves arrive at the output (½l_c) 3-dB coupler section and recombined coherently producing an optical signal that transfers all into the lower waveguide or crossover path. However, if there is a 180° phase difference between the two optical path lengths in the interferometer section, the two optical waves will recombine and transfer back to upper waveguides or the straight through path. Thus, electro-optical phase modulation of the two optical waves in the interferometer section results in an amplitude modulation at each of the output waveguides and the device can be used as a 1×2 or a 2×2 optical switch.

[00007] FIG. 1 illustrates a related art polarization-independent 2×2 electro-optic switch 100 based on a balanced-bridge interferometric waveguide structure. As shown in FIG. 1, the related art switch 100 includes upper and lower waveguide patterns 120, 130 formed in a X-cut Z-propagation lithium niobate (LiNbO₃) electro-optic substrate 110. The switch 100 includes a 3 dB-directional coupler section 142 (50:50 power splitter) having an interaction length of L_B=½L_dc, an interferometer section of length L_e, and an output 3 dB-directional coupler section 146 having an interaction length of L_B=½L_dc, where L_B=½L_dc, L_dc equals combined length of the two couplers, and l_c equals characteristic coupling length of the directional coupler. Note that “interaction length” and “effective coupling length” are used interchangeably to describe the actual length of the waveguides in a directional coupler over which a signal may couple from a first waveguide to a second waveguide. Also note that the effective coupling length may be described in terms or units of characteristic coupling length l_c. Thus a range of effective coupling lengths or interaction lengths for a directional coupler can be indicated as 0.75l_c to 1.1l_c. The input and output 3 dB-directional coupler sections operate as 50:50 power splitters and are preferably identical with a combined length where l_c is the characteristic coupling length of the switch 100. As shown in FIG. 1, the upper waveguide 120 receives an input optical signal I_in. Under straight-thru switch operations, the input signal I_in received by the upper waveguide 120 exits from the waveguide 120 as an output signal I_upper (=), and in cross-over switch operations, the input signal I_in enters the upper waveguide 120 but exits through the lower waveguide 130 as I_lower (×).

[00008] For the related art switch 100, the upper and lower waveguides 120, 130 are single mode for each of two polarizations (TE, TM), and therefore support one TE mode and one TM mode each, where the “TE mode” is the transverse electric field mode, and the “TM mode” is the transverse magnetic field mode. A normalized applied voltage V applied in the interferometer section 144 is applied with an electric field in the Y-direction (E_y) that induces a differential propagation constant Δβ (“Δβ”) between the two interferometric sections of the upper and lower waveguides 120, 130 of length L_e via the linear electro-optic effect. The length L_e is the length of the electrodes 150, 152,154. Electrodes 150, 152, 154 are arranged in a push-pull configuration to maximize the electro-optically induced Δβ between the upper and lower waveguides 120, 130 in the interferometer section 144. Thus, the electrode 152 receives the normalized applied voltage V and the electrodes 150 and 154 receive a ground potential. As shown in FIG. 1, the placement of the electrodes 150, 152, 154 maximizes the E-field along the Y-axis inside the waveguides.

[00009] For the X-cut substrate 110, when the waveguide propagation direction in the waveguides 120, 130 is along the Z-axis (optic axis), both the TE and TM modes see the same ordinary index (n_o). Thus, both the TE and TM polarization modes are nearly degenerated and will behave in approximately the same way. The electro-optic (“EO”) interaction for the TE and TM modes with the E_y field in the interferometer section are via EO coefficients that are equal but opposite in sign (i.e., r₂₂, −r₂₂) in the lithium niobate substrate 110. Therefore, the magnitudes of the Δβ_i for both the TM and TE modes are the same in the interferometer section 144 where Δβ_i is the difference in the propagation constants between the two waveguide pair in the middle “interference” section, and Δβ_dc is the difference in the propagation constants between the waveguide pair in the “input” and “output” directional coupler sections. In other words, the EO interactions in the interferometer section 144 in the TE mode is proportional to +r₂₂ (E_y) and the TM mode is proportional to −r₂₂ (E_y) as the corresponding change in the propagation constants is via the electric field in the Y-axis direction.

[00010] In FIG. 2A, I_cross-over and I_{straight-thru} conditions are illustrated for the related art switch 100, where optical power is a vertical axis and the ratio of normalized applied voltage to V_π, which is the voltage required to cause a 180° phase shift between the two arms of the interferometer V/V_π, is a horizontal axis. As shown in FIG. 2A, the input signal I_in is output as I_cross when V/V_π is equal to −4, −2, 0, 2, 4 and I_in is output as I_upper (I _{straight-thru}) when V/V_π is equal to −3, −1, 1, 3.

[00011] In FIG. 2A, the optical power is illustrated from 0 to 1 corresponding to an on (=) or off (×) state of the switch 100. For a high performance, low-crosstalk switching device, the input and output directional couplers 142, 146 of the switch 100 must behave as 3 dB-couplers for both the TE and TM modes simultaneously. In this case of zero (0) voltage, both the TE and TM modes entering an input port of the upper waveguide 120 will exit a lower output “cross-over” port (×) of the lower waveguide 130. When voltage is applied to the interferometer section 144 of the switch 100 with an electric field in the Y-axis direction, the EO induced change in the waveguide indices for the TE and TM modes are exactly equal, but with opposite sign because of the r₂₂, −r₂₂ linear EO coefficients. Because of the symmetric nature of the switching characteristics of the balanced bridge interferometer with respect to voltage, both the TE and TM modes would be switched to the upper output port as shown in FIG. 2A when a normalized applied voltage V=V_π (or V =−V_π) is applied. Thus, the related art optical switch 100 provides polarization independent switching. However, as shown in FIG. 2B, for effective operation of the related art switch 100, the length of each of the directional coupler sections 142, 144 L_B(=½L_dc) must be very precisely manufactured to be equal to ½l_c, which is the coupling length of the switch 100. When L_dc(=2 L_B) does not equal l_c, light entering the upper input channel cannot be switched completely from the output port (=) of the waveguide 120 to the output port (×) of the other waveguide 130. Note that L_dc equals 2×(L_B), and L_B=length of each 3 dB coupler (input and output). Thus, L_B must precisely equal ½l_c for a low crosstalk switch. When L_dc does not equal l_c, the crosstalk of the switch increases rapidly as L_dc deviates from l_c. This is shown in FIG. 2B, for L_dc=0.6l_c, 0.8l_c, 1.0l_c and 1.4l_c. Accordingly, when L_dc does not equal 1.0l_c precisely, the high crosstalk can make the switch 100 a non-working switch.

[00012] As described above, the related art polarization independent optical switches have various disadvantages. Crosstalk of the switch depends on fabricating optimal 3 dB couplers. L_dc precision is limited by fabrication tolerances, and a precise length for the 3 dB couplers is hard to achieve. For example, a 10% variation in coupler length can render an optical switch defective. Thus, a low crosstalk (<−25dB) switch is difficult to achieve. Further, an optimal fabrication parameter to achieve a 3 dB-coupler for the TE mode is often somewhat different than the optimal fabrication parameter required for a 3 dB-coupler for the TM mode. Thus, it is difficult to achieve an exact 3 dB coupling for both TE and TM modes simultaneously. This will result in an undesirable high crosstalk for either TE or TM modes or both.

[00013] The above references are incorporated by reference herein where appropriate for appropriate teachings of additional or alternative details, features and/or technical background.

[00014] An object of the invention is to solve at least the above problems and/or disadvantages and to provide at least the advantages described hereinafter.

[00015] Another object of the present invention is to provide an optical switch and method for operating same that substantially obviates one or more problems caused by disadvantages and limitations of the related art.

[00016] Another object of the present invention is to provide a polarization independent balanced-bridge interferometric waveguide switch and method for operating same that is a low crosstalk switch (less than −20 dB).

[00017] Another object of the present invention is to provide a polarization independent 1×2 or 2×2 electro-optic switch and method for operating same that provides polarization-independent switch operation over a broad wavelength.

[00018] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that provides a reduced polarization mode dispersion.

[00019] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that provides a reduced polarization dependent loss switch.

[00020] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that uses differential propagation constant Δβ_dc coupling in the input and output couplers.

[00021] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that uses tunable Δβ_dc directional couplers in the input and output power splitters.

[00022] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that uses voltage induced linear electro-optic effect to induce Δβ_dc in the input and output directional couplers.

[00023] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that uses asymmetric-waveguide widths to achieve the proper Δβ_dc.

[00024] Another object of the present invention is to provide a polarization independent 2×2 electro-optic switch and method for operating same that uses directional couplers having different waveguide indices between the upper and lower waveguides that result in a Δβ_dc between the upper and lower waveguides.

[00025] In order to achieve at least the above-described objects of the present invention in whole or in part, there is provided a device including an optical input, a first coupler optically coupled to the optical input having a first +Δβ_dc mismatch, an optical interferometer optically coupled to the first coupler and having a first optical path and a second optical path, the interferometer having an input that receives a signal voltage, wherein an optical path length difference between the first and second optical paths is induced in response to the signal voltage, and a second coupler optically coupled to the optical interferometer and capable of having a second −Δβ_dc mismatch.

[00026] To further achieve the above-described objects of the present invention in whole or in parts, there is provided a device including an optical input a first coupler optically coupled to the optical input having a first +Δβ_dc mismatch, a first optical waveguide optically coupled to the first coupler and having a first path length, a second optical waveguide optically coupled to the first coupler and having a second path length, a control signal electrically coupled to at least one of the first and second optical waveguides, whereby an optical path length difference between the first and second optical paths is controlled in response to the control signal, and a second coupler optically coupled to the first and second optical waveguides and capable of having a second −Δβ_dc mismatch.

[00027] To further achieve the above-described objects of the present invention in whole or in parts, there is provided a device including at least one optical input, a first coupler optically coupled to the optical input having a first optical propagation constant Δβ mismatch, an optical interferometer optically coupled to the first coupler and having a first optical path and a second optical path, the interferometer having an input that receives a signal voltage, wherein an optical path length difference between the first and second optical paths is induced in response to the signal voltage, second coupler optically coupled to the optical interferometer and capable of having a second optical propagation constant Δβ mismatch, and at least one optical output optically coupled to the second coupler.

[00028] To further achieve the above-described objects of the present invention in whole or in parts, there is provided a device including at least one optical input, an input directional coupler optically coupled to the at least one optical input having a first optical propagation constant Δβ mismatch, a first optical waveguide optically coupled to the input directional coupler and having a first optical path length, a second optical waveguide optically coupled to the input directional coupler and having a second optical path length, a control signal electrically coupled to at least one of the first and second optical waveguides, whereby an optical path length difference between the first and second optical paths is controllable variable in response to the control signal, and an output directional coupler optically coupled to the first and second optical waveguides and capable of having a second optical propagation constant Δβ mismatch.

[00029] To further achieve the above-described objects of the present invention in whole or in parts, there is provided a balanced bridge optical switch including at least one input port, an input directional coupler having a first optical propagation constant Δβ mismatch coupled to the at least one input port, an interferometer optically coupled to the input directional coupler, an output directional coupler having a second optical propagation constant Δβ mismatch optically coupled to the interferometer, at least one output port optically coupled to the output directional coupler, whereby a first optical propagation constant Δβ mismatch is adjusted to provide an approximate 50% power split at the input directional coupler for a range of effective coupling lengths from approximately 0.75l_c to approximately 1.1l_c, and a second optical propagation constant Δβ mismatch is adjusted to provide an approximate 50% power split at the output directional coupler for a range of effective coupling lengths from approximately 0.75l_c to approximately 1.1l_c.

[00030] To further achieve the above-described objects of the present invention in whole or in parts, there is provided a balanced bridge optical switch including at least one input port, an input directional coupler having a first optical propagation constant Δβ mismatch coupled to the at least one input port, an interferometer optically coupled to the input directional coupler, an output directional coupler having a second optical propagation constant Δβ mismatch optically coupled to the interferometer, at least two output ports optically coupled to the output directional coupler, whereby the effective coupling length of the input directional coupler combined with the effective coupling length of the output directional coupler has a range from approximately 1.5l_c to approximately 2.2l_c, and a first optical propagation constant Δβ mismatch is determined and a second optical propagation constant Δβ mismatch is determined such that crosstalk between the at least two output ports is below a desired amount.

[00031] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objects and advantages of the invention may be realized and attained as particularly pointed out in the appended claims.

[00049] FIG. 3A is a diagram that illustrates a first preferred embodiment of a broad wavelength band polarization independent optical switch, in accordance with the present invention. “Broad wavelength band” includes the range of wavelengths from at least approximately 1530 nm (nanometer) to at least approximately 1610 nm. The switch's functioning wavelength range includes the C and L bands, where the C band spans approximately 1530 nm to 1565 nm, and the L band spans approximately 1565 nm to 1610 nm. However, the switch's functioning is not limited to the wavelength range encompassing the L and C band, and the switch will function substantially below and substantially above both those bands, inclusively. “Polarization independent” means that the switch may effectively switch any polarization or all components of an optical signal. The first preferred embodiment of the optical switch is a polarization-independent, broad-wavelength band 2×2 switch using a balanced-bridge interferometer with asymmetrically-biased (±Δβ_dc) directional couplers.

[00050] As shown in FIG. 3A, an optical switch 300 includes upper and lower waveguide patterns 320,330 preferably formed in an X-axis cut, Z-axis propagation electro-optic substrate 310. The electro-optic substrate is preferably lithium niobate (LiNbO₃). The switch 300 includes a first directional coupler section 342, an interferometer section 344 and a second directional coupler section 346. The first directional coupler 342 has an electrode length of L_B, the interferometer section 344 has an electrode length L_e and the second directional coupler section 346 has an electrode length of L_B. The first and second directional coupler sections 342, 346 preferably operate the same with opposite directivities and respectively couple an input and output of the interferometer section 344. The first and second directional coupler sections 342, 346 have equal magnitude but opposite in sign differential propagation constants Δβ_dc, i.e., the input coupler is +Δβ_dc biased and the output coupler is −Δβ_dc biased and vice versa. While it is perferable that the signs of the differential propagation Δβ_dc mismatches for the directional couplers are opposite, such a sign difference is not necessary. Additionally, when there is a sign difference between the differential propagation Δβ_dc mismatches for the directional couplers, either order will work. That is, the first directional coupler's differential propagation Δβ_dc mismatch may have a positive sign and the second directional coupler's differential propagation Δβ_dc mismatch may have a negative sign, or the first directional coupler's differential propagation Δβ_dc mismatch may have a negative sign and the second directional coupler's differential propagation Δβ_dc mismatch may have a positive sign. The first and second waveguide pair sections 342, 346 experience evanescent coupling while the interferometer section 344 does not.

[00051] Both the first and second directional couplers may have a differential propagation Δβ mismatch, and the interferometer may have Δβ mismatch. The differential propagation Δβ mismatch for the couplers may be referred to as either Δβ or Δβ_dc, and the Δβ mismatch of the interferometer may be referred to as Δβ or Δβ_E. When the mismatch is simply referred to as Δβ, whether it is the Δβ mismatch of a coupler or Δβ mismatch of an interferometer is determined by the context of the reference. “Coupler” refers to either directional coupler of the switch. “First coupler,” “first directional coupler,” “input coupler,” and “input directional coupler” and the like are synonymous. “Second coupler,” “second directional coupler,” “output coupler,” “output directional coupler,” and the like are also synonymous. Similarly, a “first optical propagation constant Δβ mismatch” refers to the Δβ in the first or input coupler, and “second optical propagation constant Δβ mismatch” refers to the Δβ mismatch in the second or output coupler.

[00052] The optical propagation Δβ_dc mismatch in a coupler may be achieved by having the propagation constant in the first waveguide of the coupler be a first value β₁, and the optical propagation constant in the second waveguide of the coupler be a second value β₂. Thus, β₁−β₂=+Δβ_dc, and β₂−β₁=−Δβ_dc. A coupler with an optical propagation Δβ_dc mismatch has a value of β₁ for one of its waveguides and a value of β₂ for another of its waveguides.

[00053] As shown in FIG. 3A, the upper waveguide 320 of the switch 300 receives an input optical signal I_in that exits entirely from the waveguide 320 as an output signal I_upper in a “straight-thru” path or “bar (=)” state, and the input signal I_in enters the upper waveguide 320 but entirely exits through the lower waveguide 330 as I_lower in a “cross-over” path or “(×)” state. For the optical switch 300, the upper and lower waveguides 320, 330 are single moded for both TE and TM polarizations, and therefore support one TE mode and one TM mode each.

[00054] A signal voltage V_S applied as a switching voltage in the interferometer section 344 is applied with an electric field in the Y-direction (E_y) that induces Δβ_e between the two interferometer arms of the upper and lower waveguides 320,330 via the linear electro-optic effect. The electrode length of the two interferometer arms is L_e. The signal voltage can have various waveforms such as sinusoidal, stepped, chirped, sawtooth, DC, off/on, etc. Electrodes generating the electric fields are preferably arranged in a push-pull configuration to increase the electro-optically induced phase difference between the upper and lower waveguides 320, 330 in the interferometer section 344. Thus, in the X-axis cut, Z-axis propagation substrate 310, an electrode 352 receives the signal voltage V_S and electrodes 350 and 354 receives a ground potential.

[00055] As shown in FIG. 3A, electrodes are placed to maximize an electric field in the Y-axis direction of the lithium niobate crystal 310. In the X-axis cut substrate shown in FIG. 3A, the electrodes are placed along side the waveguide so as to increase or maximize horizontal electric field (E_y) inside the optical waveguide. Alternatively, tor Y-axis cut Z-axis propagation substrates, the electrodes are preferably placed on top of the waveguides so as to increase or maximize the vertical electric field (E_y) inside the optical waveguide (as shown in FIG. 10).

[00056] To be polarization independent, the optical switch 300 orients the propagation direction of the waveguides 320, 330 along the Z-axis (optic axis) so that both the TE and TM modes see the same ordinary index (n_o). The EO interaction for the TE and TM modes with the E_y field are via electro-optic coefficients that are equal but opposite in sign (i.e., r₂₂, −r₂₂ in a lithium niobate substrate). Thus, the EO interactions in the interferometer section 344 in the TE mode is +r₂₂ (E_y) and the TM mode is −r₂₂ (E_y) as the corresponding change in the propagation constants is via the electric field in the Y-axis direction. Thus, magnitudes of Δβ_e for both the TM and TE modes are the same, but opposite in sign in the interferometer section 344. As shown in FIGS. 5A-5C, optical power for the switch 300 is illustrated as a vertical axis and the ratio of voltages V/V_π is a horizontal axis. The optical power is measured by a ratio of an output optical power over an input optical power (e.g., I_upper/I_input) Accordingly, the input signal I_in launched via the upper input channel is transmitted as I_lower (I_cross-over) via the waveguide 330 when V/V_π is equal to −4, −2, 0, 2, 4 and I_in, launched via the upper input channel, is transmitted as I_upper (I_{straight-thru}) when the V/V_π is equal to −3, −1, 1, 3.

[00057] FIG. 3B shows the same embodiment of the switch as shown in FIG. 3A, with like numbers referring to like elements. The graph shows the output of the straight-thru port 320, as a function of applied switching voltage. The maximums of the output labeled “=” correspond to when substantially all the signal exits the interferometer switch 300, through the straight-thru port 320, and the minimum labeled “×” corresponds to when substantially all the signal exits the interferometer through the cross-over port 330.

[00058] FIG. 4 is a vertical two-dimensional slice of a three-dimensional plot of the power output of the upper waveguide 320 as a function of normalized applied voltage V/V_π. When light is launched into the upper input channel of the switch 300, where the X-axis is the ratio between L_dc/l_c and the Y-axis is the ratio of the normalized applied voltage V/V_π in the interferometer section 344. As shown in FIG. 4, crosstalk contour A, crosstalk contour B, and crosstalk contour C represent −10 dB, −20 dB and −30 dB crosstalk contour lines respectively. FIGS. 5A-5C show a particular case where electrodes 360, 362, 364 on the input and output couplers 342, 346 receive the normalized applied voltage bias of V_dc=Δβ_dcL_dc/π of approximately ±1.6 (where Δβ_dc=difference in the propagation constant of the two waveguides in the directional coupler). The characteristic coupling length of the switch 300 is represented as l_c, and the combined length of the two directional coupler is L_dc.

[00059] FIGS. 5A-5C are two-dimensional plots that are horizontal slices through the three-dimensional plot of FIG. 4. FIGS. 5A-5C each represent optical power output by the upper waveguide 320 while the ratio V/V_π varies for a fixed ratio of L_dc/l_c. FIG. 5A illustrates where L_dc is 2.2l_c, FIG. 5B illustrates the case where L_dc is 1.8l_c, and FIG. 5C illustrates the case where L_dc is approximately 1.6l_c. (L_dc is the total length of the two input and output directional couplers.)

[00060] In FIGS. 5A-5C, the optical power is illustrated from 1 to 0, which correspond to an “ON” or “OFF” state of the switch 300, respectively. The Δβ_dc (propagation difference between the directional waveguide pair) of the first and second directional couplers 342, 346 must be equal but opposite in sign. This means that the Δβ_dc of the second directional coupler 346 must be in the opposite direction of the Δβ_dc of the first input coupler 342 but of the same magnitude. This asymmetric Δβ_dc is achieved in the first preferred embodiment of the switch 300 by having an equal but opposite bias to the central electrodes 362 of the first and second directional couplers 342, 346 while the outer electrodes 360, 364 are coupled to the ground voltage. Further, the central electrodes 362 preferably have a length L_E. The outer electrodes 360, 364 preferably have an equal length being L_B.

[00061] When the signal voltage V_S is applied to the interferometer section 344 with an electric field in the Y-axis direction, the EO induced change in the waveguide indices for the TE and TM modes are exactly equal, but with opposite sign via the r₂₂, −r₂₂ linear EO coefficients. Thus, because of the symmetric nature of the switching characteristics of the balanced bridge interferometer with respect to voltage, both the TE and TM modes of the input optical signal I_in would be switched to the upper output port as shown when a V_S=V_π (or V_S=−V_π) is applied. The dark band in the center of each contour on FIG. 4 graphically shows an expanded “cross-over” switch region with very low (−30 dB) crosstalk for the switch when the input and output directional couplers are properly asymmetrically biased V_dc=Δβ_dcL_dc/π of approximately ±1.6. FIG. 4 also shows that an expanded −30 dB crosstalk contour (e.g., vertical slot) is achieved for an expanded range of L_dc/l_c value from 1.5 to 2.2. This particular family of crosstalk contours illustrated by FIG. 4 is achieved by applying a normalized bias voltage V_dc=Δβ_dcL_dc/π of approximately ±1.6 to the first and second (input and output) directional couplers 342, 346.

[00062] FIG. 5A is a two-dimensional plot showing the output of the upper waveguide 320 when L_dc=2.2l_c. Again, the X-axis illustrates power of the signal coming out of the upper waveguide 320, and the Y-axis shows the ratio of the switching voltage V/V_π. FIG. 4B shows that even though L_dc is not equal to l_c and in fact is 2.2l_c, the output of the upper waveguide 320 ranges from 0 to 1 thereby exhibiting very low crosstalk being <−30 dB between the ON (upper) and the OFF (lower) outputs of the switch 300. FIG. 5B is a two-dimensional graph for the special case of L_dc=1.8l_c.

[00063] FIG. 5B shows that when L_dc=1.8l_c the output from the upper waveguide 320 ranges from 0 to 1 through the range of normalized applied voltages even though L_dc does not equal l_c and in fact FIG. 4C shows that the switch 300 functions as well when L_dc=1.8l_c as it does when L_dc=2.2l_c. Thus, FIG. 5B shows the switch 300 exhibits crosstalk being <−30 dB for the ratio L_dc=2.2l_c.

[00064] FIG. 5C is another two-dimensional plot showing the output of the upper waveguide 320 for the values of L_dc≈1.6l_c for a range of switching voltages. Once again FIG. 5C shows that the output of the upper waveguide ranges 320 from 0 to 1 for the range of normalized applied voltages V when L_dc is not equal to l_c but L_dc≈1.6l_c.

[00065] Taken together, FIGS. 5A-5C graphically illustrate that the directional couplers 342, 346 can have a range of L_dc (or a range of ratios 1.5<L_dc/l_c<2.2) and the switch 300 still operates as a low −30 dB crosstalk switch. This is what is meant by the switch having expanded cross-over region and a −30 dB crosstalk contour (e.g., slot) for both the TE and TM modes. The switch 300 differs from the related art because the related art does not exhibit an expanded −30 dB crosstalk contour but merely has a −30 dB point when L_dc=l_c as shown in FIG. 2C.

[00066] According to the first preferred embodiment of the switch 300, l_dc is the total length of the combined directional couplers, l_c is the characteristic coupling length for the directional coupler sections 342 and 346. L_B is the electrode length of each directional coupler that is preferably equal to {fraction (1/2 )}L_dc. The signal voltage V_S is preferably a normalized applied voltage, and is equal to Δβ_eL_e/π where L_e is the electrode length for the interferometer section 344, and Δβ_e is the electron optically induced propagation difference between the two interferometric waveguide pair. Thus, normalized directional coupler bias voltage V_dc=Δβ_dcL_dc/π where L_B (=½L_dc) is the electrode length for each of the input and output directional couplers, and Δβ_dc is the electro-optically induced Δβ_dc between the waveguide pair in the directional coupler region.

[00067] Thus, the first preferred embodiment of the optical switch 300 uses a balanced bridge interferometer waveguide structure with asymmetrically-bias Δβ_dc-directional coupler structures 342 and 346, with an equal magnitude, but an opposite sign for the directivity. The directional couplers 342 and 346, are preferably designed such that the effective Δβ_dcL_dc/π is approximately equal to 1.6, which provides the favorable extinction ratios with an extended −30 dB crosstalk contour for a given coupling length ratio L_dc/l_c. Accordingly, to maintain a −30 dB optical switch, the total (including the input and output) directional coupler effective length (L_dc) can be in the range of ˜1.5 to 2.2 times the characteristic coupling length l_c of the directional coupler. Evanescent coupling occurs in the input and output directional couplers but not in the interferometer section. Light for both TE and TM polarizations is switched between the two outputs (e.g., I_upper, I_lower) by applying a signal voltage V_S to the interferometer section equivalent to V_π, which is the voltage required to electro-optically induce a phase mismatch of π (180°) between the two waveguides.

[00068] The first preferred embodiment generates the asymmetrically-biased Δβ_dc input and output directional couplers using an applied voltage to induce the Δβ_dc change using the linear electro-optic effect. However, the present invention is not intended to be so limited.

[00069] As shown in FIGS. 6A-6G, the crosstalk contour plots of −10 dB, −20 dB and −30 dB for a vertical axis in units of L_dc/l_c and a horizontal axis of a normalized applied voltage V/V_π are provided for a progressive directional coupler bias voltage V_dc applied to the switch 300. A center of a bullseye is a low point in the optical power output of the waveguide, and represents low (−30 dB) crosstalk, and is shown being a cross-over (×) state 502 for the upper waveguide 320 (in FIG. 3A) in FIG. 5A. For example, a cut on this contour plot along the line of L_dc/l_c=1 would be a horizontal cut through the center of the bullseye on V/V_π=0. The bullseye is a low point and one can move horizontally to the right to where V/V_π=1, which is a high point shown being the bar (=) state 504, for the upper waveguide 320 (in FIG. 3A) in FIG. 5A occurs at V/V_π=1. Moving horizontally, the next bullseye at V/V_π=2 is a low point. Accordingly, a horizontal line cut along the L_dc/l_c=1 horizontal line would produce a sinusoidal varying linear switch line. Similar sinusoids are shown in FIGS. 5A-5C. As another example, if a horizontal cut was taken along the L_dc/l_c=0 line along a bottom edge of the crosstalk contour plot in FIG. 4, there is no variation in the optical power as V/V_π goes from 0 to 2. Thus, there is no coupling to the lower waveguide and all the optical power remains in the upper waveguide output 320 of switch 300. As V/V_π is varied there is no switching effect because there is no coupling in the interferometer section 344. Similarly, at the L_dc/l_c=2 horizontal line in FIG. 6A, each one of the directional couplers is 100% coupling in the bar (=) state. Thus, the first directional coupler 320 couples a portion of the input signal I_in to the bottom waveguide 320 and the first directional coupler 320 is exactly one coupling length l_c so all the input signal I_in power in the upper waveguide 320 will couple down to the lower waveguide 330 at V/V_π=−2, 0, 2, etc. In the interferometer section 344, a phase shift would be applied, but since there is no signal in the upper waveguide 320 all power is phase shifted to the bottom waveguide 330. Upon passage through the output directional coupler 346, which is again one coupling length l_c, the input signal I_in is coupled back to the upper waveguide 320 and output as I_upper. Thus, the result is similar. All the input power signal is transmitted out the upper waveguide output 320 independent of V/V_π. As shown in the description of the related art, the bar state can generally be achieved by optical switches, but a −30 dB crosstalk cross-over (×) state is generally difficult to achieve.

[00070] When the cross-over (×) state (e.g., as indicated by a bullseye, peanut shape or slot in FIGS. 6A-6G) in the crosstalk contour plots cannot be reached, complete switching from 0 to 1 cannot be achieved by the optical switch and the switch would exhibit high crosstalk (e.g., greater than −20 dB, −30 dB outside the corresponding contours). As shown in FIGS. 6A-6G, a −10 dB cross-over (×) state 610 is larger than a −20 dB cross-over (×) state 620, which is larger than a −30 dB cross-over (×) state 630. In fact, as shown in FIG. 6A, where V_dc=0, a −30 dB state can rarely be achieved since L_dc must be precisely 1l_c and such manufacturing tolerances are rarely achieved. However, as shown in FIG. 6F, where V_dc≈±1.6, a −30 dB crosstalk contour extends from approximately from L_dc/l_c=1.5 to 2.2. Thus, a horizontal line across the crosstalk contour plot shown in FIG. 6F indicates that a −30 dB switch can achieve 100% switching from 0 to 1 for a large tolerance of L_dc/l_c ratios and accordingly a very large tolerance of L_dc lengths.

[00071] The contour plots of FIGS. 6A-6G can be achieved by applying different bias voltages ±V_dc at the directional couplers 342, 346 in the switch 300 illustrated in FIG. 3A. FIG. 6A is a diagram that shows a crosstalk contour plot with the bias voltage at 0, FIG. 6B is a diagram that shows the crosstalk contour plot with the bias voltage at 0.5, and FIGS. 6C-6G are diagrams that show crosstalk contour plots when ±V_dc is at 1.0, 1.4, 1.5, 1.6 and 1.7, note that V_dc is the normalized applied voltage in the unit of Δβ_dcL_dc/π where L_dc approximately equals the total length of the combined directional couplers (input and output), Δβ_dc equals the difference in propagation constant induced between the two directional coupler waveguide pairs. FIG. 6F is a crosstalk contour plot similar to that shown in FIG. 4 for a bias voltage of approximately 1.6, where V_dc is a normalized voltage in the unit of Δβ_dcL_dc/π.

[00072] The advantage of looking at different bias voltages V_dc for the switch 300 illustrated in FIG. 3A, is that an empirical understanding of the advantages according to preferred embodiments can be seen through the series of crosstalk contour plots. FIG. 6A is a crosstalk contour plot of an equivalent related art type switch where there is no asymmetric Δβ_dc coupling between the first and second directional couplers. As FIG. 6A illustrates, the cross-over state (×) occupies a very narrow region on the L_dc/l_c axis. This indicates that the ratio of L_dc to l_c must be tightly controlled in order for a low crosstalk switch to be achieved. FIG. 6B is a crosstalk contour plot showing the output when the normalized bias voltage in the unit of Δβ_dcL_dc/π, V_dc at the directional couplers 342, 346 is 0.5. The crosstalk contours shown in FIG. 6B are very similar to the contours shown in FIG. 6A. The primary difference being that the −10 dB regions 620 in FIG. 6B have moved closer to each other along the axes L_dc/l_c. Turning to FIG. 6C where the bias voltage V_dc at the directional couplers 342, 346 is 1, the plot illustrates how the −30 dB contours 610, 620, 630 are closer yet along the axis represented by L_dc/l_c. FIG. 6D shows the crosstalk contour plot where the bias voltages V_dc is 1.4. This contour plot shows yet again how the contours 610, 620, 630 have moved closer to each other due to the bias voltage V_dc and that there is some overlap in the −10 dB region 610 of the crosstalk contour plot.

[00073] FIG. 6E shows the crosstalk contour plot for a bias voltage V_dc of 1.5 where the contours 610,620,630 have moved even closer and there is almost overlap in the −20 dB region 620. FIG. 6F shows the crosstalk contour plots for the bias voltage 1.6 where the 30 dB contours 630 have merged along the Y-axis representing the L_dc/l_c ratio. This elongated crosstalk contour along the Y-axis indicates that for a wide range of ratios of L_dc/l_c, a cross-over state at −30 dB can be achieved. This indicates that the switch can be manufactured with a wide range of ratios of L_dc/l_c and the switch will still function as a low crosstalk switch. FIG. 6G shows contours for a bias voltage V_dc of 1.7 at the directional couplers 342, 346. FIG. 6G shows how as the bias voltage V_dc is increased above 1.6, the −30 dB crosstalk contours 630 come even closer to one another so that there is less of a range for the ratio of L_dc to l_c where the switch 300 will operate with about −30 dB of crosstalk. FIG. 6G is an example whereby the bias voltage V_dc has been increased beyond an optimal amount. FIGS. 6A-6G show that as the bias voltage V_dc at the directional couplers 342 and 346, is increased, the switch 300 becomes more and more tolerant of a range of ratios of L_dc to l_c with an optimal range being reached at about V_dc being equal to approximately 1.6 yet still exhibit low crosstalk.

[00074] Looking at the difference in the cross-over contours between FIG. 6A which illustrates equivalent operations of a related art switch and FIG. 6F which illustrates operations of a preferred embodiment of the present invention, it is clear that the switch in FIG. 6F allows a much wider range of L_dc over l_c, which translates into at least a larger tolerance for error in the manufacturing process for the length L_dc.

[00075] In the Z-propagation waveguide switches, even if the TE and TM modes have slightly different characteristics in their respective coupling behavior, or the waveguides using the +r₂₂, −r₂₂ EO coefficients, that slight difference would merely be represented as a small vertical offset between close horizontal slices in the crosstalk contour plot of FIG. 6F. Thus, the TE and TM modes can be designed to be completed completely within the large available −30 dB crosstalk contour 630 and thus, an effective polarization independent switch is achieved.

[00076] In addition, as coupling length is a function of wavelength (λ) of the input signal I_in, a broad band light source can be used while the optical switch still performs at a −30 dB crosstalk level. Directional coupling length (l_c) is a function of the operating wavelength. Therefore, a change in wavelength λ simply means a change in the effect directional coupling length (L_dc/l_c). Since the switch can be operated with very low crosstalk for an expanded range of (L_dc/l_c) when V_dc˜1.6, a switch can be operated with very low crosstalk for an expanded range of wavelength as compared to conventional switch.

[00077] For example, a broad band light source having wavelengths between 1.4 and 1.7 microns can be permitted for operating an input signal to the optical switch 300.

[00078] FIG. 7A is a diagram that illustrates a second preferred embodiment of the optical switch in accordance with the present invention. The second preferred embodiment is a polarization independent broad wavelength 2×2 switch relying on a balanced bridge Mach-Zender interferometer constructed on an X-axis cut Z-propagation lithium niobate waveguide (LiNbO₃) structure with asymmetrically biased directional couplers. In the second preferred embodiment, the asymmetric waveguide function is achieved by designing the waveguides to have different widths in the directional coupler section that result in a selected Δβ_dc between the waveguides.

[00079] As shown in FIG. 7A, an optical switch 700 is formed in an X-axis cut Z-axis propagation electro-optic substrate 710. The EO substrate is preferably lithium niobate LiNbO₃, however, it may be made of other suitable electro-optic substrates. The switch 700 includes a lower waveguide input 712 and an upper waveguide input 714, and the switch 700 also includes a first directional coupler 720, an interferometer section 722 and a second directional coupler 724. The output end of the switch 700 has a lower output waveguide 732 and an upper output waveguide 734. The first and second directional couplers 720 and 724 are close enough so that the two waveguides experience evanescent coupling. The interferometer section 722 of the switch 700 does not experience evanescent coupling between the waveguides. In the interferometer section 722 of the switch 700 are three electrodes 746, 744 and 742. The electrodes 746, 744, 742 parallel the waveguides in the interferometer section 722 with the electrode 746 being along one side of the waveguides, the electrode 742 along the opposite side of the waveguides and the electrode 744 located between the waveguides. Electrodes 742 and 746 are preferably electrically connected to ground potential and the center electrode 744 is connected to the signal voltage V_S. Referring to the first and second directional couplers 720 and 724, the total length of the two directional couplers 720,724 is L_dc, and L_dc can be anywhere in the range of 1.5 to 2.2 times the characteristic coupling length l_c. The directional coupler effective length L_dc is preferably divided between each directional coupler 720 and 724 as L_B and L_B. The asymmetric (Δβ_dc) nature of the couplers is achieved by the un-equal width of the waveguides in the coupler regions.

[00080] The asymmetrically biased nature of the directional couplers 720, 724 is achieved in the second embodiment by using asymmetric waveguide widths for the two waveguides in each directional coupler section 720 and 724, which effectively results in a Δβ_dc between the waveguides. The two waveguides with different widths are designed so that Δβ_dc is equal but opposite in sign, where the propagation difference between them is Δβ_dcL_dc/π˜1.6 for the output directional coupler. For example, that portion of the waveguide 714 in the first directional coupler 720, is wider, w₁, than the corresponding narrower, w₂, portion of the waveguide 712 in the first directional coupler 720, and that portion of the waveguide 714 in the second directional coupler 724, is narrower, w₂, than the corresponding wider, w₁, portion of the waveguide 712 in the second directional coupler 724. Δβ_dcL_dc/π≈1.6 is achieved by fabricating the appropriate Δw, where Δw=(w₁−w₂).

[00081] All the switch operations are similar to the first case where Δβ_dcL_dc/π˜1.6 using bias voltage. In operation, the switch 700 receives an optical signal into the input waveguide 714, the first directional coupler 720 divides the optical signal between the two arms of the interferometer section 722, the signal voltage V_S then controls whether the divided optical signal couples into the upper output 734 or the lower output 732 at the second directional coupler 724. The interferometer section 722 controls which waveguide the optical signal couples into using the signal voltage V_S by relying on the EO effect of the lithium niobate substrate. The propagation constant of the lithium niobate substrate is altered based on the direction and the magnitude of the electric field applied by the signal voltage V_S. The electrodes 746,744 and 742 are arranged to maximize the difference in Δβ_i when the signal voltage is applied. For an X-cut lithium niobate substrate with waveguide propagation direction along the Z-axis, the electrodes 746,744 and 742 are placed alongside the waveguides in order to maximize the horizontal electric field, E_y, inside the two arms of the interferometer section 722.

[00082] When the switch 700 is in the bar state (=), the optical signal entering the input waveguide 714 exits the switch 700 at output waveguide 734. When the switch 700 is in the crossover state (×), an input signal at input waveguide 714 exits the switch at output waveguide 732. Because the waveguides are arranged on the lithium niobate substrate so that the propagation direction is along the Z-axis, both the TE and TM modes see the same ordinary index n₀. Similar to the first case, when Δβ_dcL_dc/π˜1.6 using asymmetric waveguide widths, rather than an applied voltage, switch operation with low crosstalk can be achieved for a very wide range of directional coupler parameters (L_dc/l_c˜1.5 to 2.2). Therefore a broad wavelength, polarization independent switch can be achieved.

[00083] As shown in FIG. 7B shows the same embodiment of the switch as shown in FIG. 7A, with like numbers referring to like elements. The graph of FIG. 7B shows the output of the straight-thru port 734, as a function of applied switching voltage. The maximums of the output labeled “=” correspond to when substantially all the signal exits the interferometer switch 700, through the straight-thru port 734 and the minimum labeled “×” corresponds to when substantially all the signal exits the interferometer switch through the cross-over port 732.

[00084] A third preferred embodiment of an optical switch according to the present invention will now be described. As shown in FIG. 8, a third preferred embodiment of a polarization independent broad wavelength band 2×2 switch 800 uses a Mach-Zender type balanced bridge interferometer waveguide structure with asymmetrically biased directional couplers. The asymmetrically biased nature of the directional couplers is achieved by manufacturing the waveguide with different propagation indices, n₁ and n₂, between the waveguides in the directional couplers. Different propagation indices for the waveguides can be achieved, for example, by changing the thickness of the initial metal to be diffused or by using material loading effects on the waveguides in the coupler region. Either one of these methods achieves a selected difference in Δβ_dc between the waveguides.

[00085] Turning to the switch 800 shown in FIG. 8, the waveguide substrate material 810 is made preferably out of lithium niobate. The substrate material is X-cut with a preferred propagation direction along the Z-axis. There are two waveguide inputs 812 and 814. There is a first directional coupler 820 followed by an interferometer section 822, followed by a second directional coupler 824, and on the output end of the switch 800 are two output waveguides 832 and 834. In the interferometer section 822 of the switch 800, are three electrodes 842, 844 and 846. For the optical switch 800, the upper and lower waveguides 814 and 812 are single moded for both the TE and TM modes. For the X-cut lithium niobate substrate, the electrodes 842, 846 are preferably arranged on the outside along opposite sides of the waveguides and the electrode 844 is positioned between the interferometer arms. The electrodes 846 and 842 are electrically connected to ground potential. A signal voltage V_S can be applied to the electrode 844 to alter the propagation constant in the same amount but in different directions for each arm of the interferometer section 822. The first and second directional couplers 820 and 824 are asymmetrically biased directional couplers.

[00086] In the third preferred embodiment shown in FIG. 8, the difference in propagation constant (Δβ_dc) of the directional couplers is achieved by how the waveguides are constructed. In general, a waveguide is constructed on a lithium niobate substrate or other dielectric material by diffusing other materials into the substrate during the manufacturing process to change the optical index locally so that light will then propagate preferentially along the path created by the diffused material. By varying the amount of material diffused into the substrate, the propagation constant of a particular waveguide can be altered. In the third preferred embodiment, a Δβ_dc between the first and second directional couplers 820, 824 for each waveguide is achieved by starting with a different amount of metal to be diffused into the substrate. Otherwise known as differential optical loading, the differential propagation constant can also be achieved by asymmetric layering of a dielectric or metal over the waveguide.

[00087] By diffusing into, or layering upon, different amounts of metal or dielectric, different indexes of refractions and the corresponding waveguide indexes, n₁ and n₂, are achieved for each waveguide section in the coupler region. ±Δn is the waveguide's difference in the index of refraction in the coupler region, where Δn=n₁−n₂, and the sign indicates the direction of the change. Δn is the same but opposite in sign between the input (first) 820, and output (second) 824, directional couplers. Δn is chosen so that Δβ_dcL_dc/π≈±1.6.

[00088] Alternatively, one can also change the optical waveguide index by using material such as a dielectric or metal loading effect to change the waveguide propagation index. By properly selecting a Δn designed to achieve Δβ_dcL_dc/π to be approximately ±1.6 (equal but opposite in sign between the input and output couplers), a switch with low crosstalk can be achieved for an extended range of directional coupler parameters L_dc/l_c˜1.5 to 2.2.

[00089] In operation, the switch 800 as shown in FIG. 8 performs similar to the other preferred embodiments. Accordingly, a detailed description is omitted. Further, the switch 800 receives an input signal into input waveguide 814. The input signal is divided between the two waveguides at the first directional coupler 820. The divided signal travels through the upper and lower arms of the interferometer section 822. The upper and lower arms of the interferometer section receive a signal voltage V_S at the center electrode 844 while the outer electrodes 842 and 846 are held at ground. The signal voltage causes an electric field between the electrodes which changes Δβ_i for the interferometer section. This causes a phase shift between the optical signals in the upper and lower arms so that when both signals arrive at the second directional coupler 824, the optical signal is either coupled entirely into output waveguide 834 or output waveguide 832. If the signal couples into waveguide 834, the switch is in a bar state which is a straight-thru configuration. If the signal is coupled into waveguide 832, the switch is in a cross-over state, when the Δβ_dcL_dc/π˜1.6, and the directional couplers 820 and 824 have a total coupling length of L_dc equal to about 1.5 to 2.2 times the characteristic coupling length l_c. Thus, the total directional coupler length L_dc can be anywhere in the range of about 1.5 to 2.2 times the characteristic coupling length l_c of the directional couplers 820, 824.

[00090] FIG. 9 shows the same embodiment of the switch as shown in FIG. 8, with like numbers referring to like elements. The graph of FIG. 9 shows the output of the straight-thru port 834, as a function of applied switching voltage. The maximums of the output labeled “=” correspond to when substantially all the signal exits the interferometer switch 800, through the straight-thru port 834 and the minimum labeled “×” corresponds to when substantially all the signal exits the interferometer switch through the cross-over port 832.

[00091] In the first, second and third embodiments of the optical switch, an X-cut, Z-propagation LiNbO₃ crystal was used as the substrate. In addition to not being limited to LiNbO₃ as the substrate material, the invention is not intended to be limited to an X-cut crystal orientation. It is contemplated that other orientations will work for the switch substrate. For example, FIG. 10 shows a Y-cut, Z-propagation optically active crystal as the substrate. In FIG. 10, the interferometer switch 1000 has an upper and lower input port, 1002 and 1004, respectively, and an input and an output directional coupler 1006 and 1010, respectively. The input directional coupler 1006, has an upper input electrode 1012 and a lower input electrode 1014. The second directional coupler 1010, has an upper output electrode 1016 and a lower output electrode 1018. In the interferometer section 1008, there is an upper switch electrode 1020, and a lower switch electrode 1022.

[00092] In FIG. 10, because the crystal's X and Y axes are rotated about the Z axis by 90° from the X-cut examples, the electrodes 1012, 1014, 1016, 1018, 1020, and 1022 must be placed in different positions relative to the waveguides in order to induce the change in propagation constant, Δβ, in the Y-direction. As FIG. 10 illustrates, the electrodes must be placed above the waveguides to induce a Δβ in the Y direction. Also contemplated for the Y-cut crystal is placing some or all of the electrodes under the waveguides.

[00093] After a switch made according to the preferred embodiments has been installed, the aforementioned tuning feature of the switch (e.g., the switch 300 shown in FIG. 3A) can be used to optimize the switching function if the switch is degraded by local conditions such as for example, temperature, pressure, mechanical stresses such as bending or twisting, and chemical environment. The switch's tuning characteristics can also be used to optimize the switching function if it degrades over time. For example, switching degradation can be caused by substances diffusing into the switch substrate and waveguides, or by the frequency of the optical signal changing, etc. Virtually any change in the switching characteristics based on altered directional coupler functionality can be compensated by tuning the directional couplers using their associated electrodes.

[00094] FIG. 11A shows related art balanced-bridge interferometer 1100, with a graph of the signal output of the straight-thru port. 1110 and 1120, are the top and bottom input ports, respectively. 1130 and 1140, are the first and second directional couplers. 1150 is the interferometer section, and includes an upper and lower electrode 1160 and 1180, respectively, which are held at ground potential, and a middle electrode 1190, which receives a controlling signal. 1195 and 1197 are the straight-thru and cross-over outputs, respectively. All the waveguides are single-moded.

[00095] For the related art balanced-bridge interferometer switch 1100, to function with low cross-talk, each directional coupler 1130 and 1140, should have an effective length of one-half the characteristic coupling length. In operation, a differential phase shift is induced within the interferometer section 1150, by application of a switching signal to the middle electrode 1190.

[00096] The output of the straight-thru port as a function of applied switching voltage for coupler lengths of one-half the characteristic coupling length is shown in FIG. 11B. The maximums of the output labeled “=” correspond to when substantially all the signal exits the interferometer 1100, through the straight-thru port 1195, and the minimum labeled “×” corresponds to when substantially all the signal exits the interferometer through the cross-over port 1197. When the lengths of the first and second directional coupler 1130 and 1140, respectively, are one-half the characteristic coupling length, the interferometer exhibits low cross-talk as indicated by the minima of FIG. 11B being substantially equal to 0.

[00097] FIGS. 12A through 12E show the output of the straight-thru port 1195, as a ratio between input signal and output through the straight-thru port, of the interferometer switch 1100, for a range of directional coupler 1130 and 1140, lengths from 0.6l_c to 1.4l_c. When the ratio is close to one, substantially all the signal exits through the straight-thru port, and when the ratio is close to 0, substantially all the signal exits through the cross-over port. In FIG. 12C, where the minimum represented by the reference letter “A” is close to 0, the interferometer switch 1100 is exhibiting low cross-talk, and has a minimum of cross-talk when L_dc=1.0l_c. The interferometer switch 1100 exhibits increasing cross-talk as indicated by the minimum at A increasing in value as L_dc deviates from 1.0l_c. FIGS. 12A, 12B, 12D, and 12E are for L_dc=0.6l_c, L_dc=0.8l_c, L_dc=1.2l_c, L_dc=1.4l_c

[00098] FIG. 13A is an example of a balanced bridge interferometer switch 1300, with an input directional coupler 1302, an interferometer section 1304, and an output directional coupler 1306. The input directional coupler 1302, has a pair of electrodes 1308 and 1310. The output directional coupler 1306, has a pair of electrodes 1312 and 1314. The interferometer section has a pair of electrodes 1316 and 1318. The interaction lengths L_B, of the input and output directional couplers, 1302 and 1306, respectively are each equal to one-half the characteristic coupling length l_c.

[00099] Via the electrodes 1308, 1310, 1312, and 1314, associated with the directional couplers 1302 and 1306, a difference in the propagation constant Δβ_dc, of the waveguides in the couplers can be induced. A normalized applied voltage of Δβ_dcL_B/π=±0.8 is the preferred voltage to be applied between the electrodes 1308 and 1310, as well as between 1312 and 1314. It is further preferred that the each directional coupler be biased the same amount but with an opposite sign, and this would be achieved by suitably controlling the polarity of the bias voltages. By applying a normalized voltage of Δβ_dcL_B/π=±0.8, the widest range of coupler interaction length L_B is allowed for a coupler to function as a 3 dB coupler.

[00100] FIG. 13B is a contour plot for a single directional coupler which could be used as either the input directional coupler 1302 or the output directional coupler 1306 of the balanced bridge interferometer switch 1300. The horizontal axis is the induced change in propagation constants between the waveguides of a directional coupler as expressed in units of Δβ_dcL_B/π, where Δβ_dc is the difference in the propagation constant between the waveguides and L_B is the interaction length of the coupler. The vertical axis is the normalized interaction length between the couplers as expressed in units of L_B/l_c, where L_B is the same as for the horizontal axis and l_c is the characteristic coupling length. Characteristic coupling length is the interaction length between the two waveguides sufficient to transfer all the power in one waveguide into the other. Reference letter “A” indicates the locus of the 3 dB half power points. Reference letter “D” is where the locus of the 3 dB half power point crosses the zero vertical axis. Reference letter “C” is that section of the 3 dB locus corresponding to where Δβ_dcL_B/π is approximately equal to 0.8. Reference letter “D” indicates the contour for when all of the output power exits from the straight through output with less than 30 dB power exiting through the crossover output. Reference letter “E” indicates the power contour region when all the output power exits through the cross-over output with less than 30 dB power exiting through the straight-thru output.

[00101] The line where the normalized applied voltage bias equals zero on the horizontal axis corresponds to the 0 bias state of related art directional couplers. Hence FIG. 13B indicates that, with a normalized applied voltage bias equal to zero, the line associated with the interaction length of a coupler bisects the locus of the 3 dB half power point near 0, and consequently intersects the locus in a region corresponding to a point.

[00102] The physical implication of FIG. 13B is that the interaction length of the coupler must be precisely controlled in order to achieve the 3 dB coupler. When the normalized applied voltage bias (Δβ_dcL_B/π) equals either negative 0.8 or positive 0.8, the edge of the locus of the 3 dB point is intersected. Consequently, the coupler interaction length can have a much greater variation, and 3 dB coupling would still being achieved. Corresponding to ±0.8 normalized voltage bias, the 3 dB coupling length of the coupler ranges from 0.75 to 1.1 in units of l_c. In other words, at the bias point of Δβ_dcL_B/π˜±0.8, 3 dB splitting can be achieved for an extended range of L_B/l_c˜0.75 to 1.1 (i.e. L_dc/l_c˜1.5 to 2.2), not a singular point as in the conventional design.

[00103] The 3 dB locus in the regions of Δβ_dcL_B/π=+0.8 and −0.8 normalized voltage bias has an opposite curvature. The bias of the input and output directional couplers for the balanced bridge interferometer switch are chosen to have the same magnitude, but opposite polarity to balance out this slight asymmetry.

[00104] The contour plot of FIG. 13B is derived using standard coupled wave formula. The derivation and starting equations can be found, for example, in the “Handbook of Microwave and Optical Components,” Vol. 4, Chapter 4, “Optical Modulation, Electro-Optic devices,” by Suwat Thaniyavarn, which is incorporated herein by reference.

[00105] FIGS. 14A through 14H show the output of the switch 300 shown in FIG. 3A for various coupling lengths ranging from 1.2l_c to 2.6l_c. When the ratio shown in FIGS. 14A through 14H is close to one, substantially all the signal exits through the straight-thru port, and when the ratio is close to 0, substantially all the signal exits through the cross-over port. In FIGS. 14B through 14G, which corresponds to L_dc=1.4l_c to 2.4l_c, reference letter “A” indicates that there is substantially 0 cross talk. This is a much wider operational range than the related art interferometer switch represented by FIGS. 12A through 12E.

[00106] FIG. 15 shows contour plots of straight-thru and cross-over switch states for a related art balanced bridge interferometer. The “×” corresponds to the cross-over switch state with three levels of cross-talk of <−30 dB, <−20 dB, and <−10 dB. The “=” corresponds to straight-thru switch states with three levels of cross-talk of <−30 dB, <−20 dB, and <−10 dB.

[00107] As described above, the present invention provides various advantages. According to the preferred embodiments, polarization independent switching over a broad wavelength can be realized because of relaxed coupler length tolerances. Further, the polarization independent switching has a very low polarization mode dispersion (PMD) and polarization dependent loss (PDL) through the use of Z-axis (i.e., optic-axis) propagation waveguide orientation on an electro-optic substrate where TE and TM see an identical ordinary index.

[00108] A low crosstalk switch (less than −25 dB) can be achieved for a very large range of L_dc values (first preferred embodiment 1.5l_c to 2.2l_c). In accordance with the large range of the directional coupler effective coupling length L_dc, a fabrication tolerance of L_dc equal to 1.5l_c to 2.2l_c can be achieved, which creates an increased production yield. Further, since the characteristic coupling length l_c of the directional coupler is a function of an optical wavelength and polarization (and to a small degree a function of environmental parameters such as waveguide loading effects from a dielectric layer, electrode layer, stress, temperature, etc.), the switch can be designed to operate in a very broad range of wavelengths for both TE and TM polarizations.

[00109] The foregoing embodiments and advantages are merely exemplary and are not to be construed as limiting the present invention. The present teaching can be readily applied to other types of apparatuses. The description of the present invention is intended to be illustrative, and not to limit the scope of the claims. Many alternatives, modifications, and variations will be apparent to those skilled in the art. In the claims, means-plus-function clauses are intended to cover the structures described herein as performing the recited function and not only structural equivalents but also equivalent structures.