Optical switching system

Abstract

An optical switching device switches optical signals completely within optical media, from an arbitrary number of N input optical fibers to a set of M output optical fibers. The invention uses a thermally-activated bimorph optical switch to redirect optical radiation emitted by a laser light source, or from the end of an optical fiber, to an input end of another optical fiber. The modulated optical radiation from the input fiber optic bundle is collimated into parallel beams and projected in free space across the tops of an array of microcantilever bimorph optical switches. When selected, a particular switch is activated, thereby causing a first bimorph switch element to pop up, and intercept and reflect the optical radiation at an angle approximately 90° to the original optical path direction. This reflected radiation is intercepted by another essentially identical bimorph switch element, located above or below the first switch element, that directs the radiation in a plane essentially parallel to that of the original beam plane, but above or below the original plane, and at approximately right angles to the original beam direction. The optical radiation is directed by the second switch element to output optical fiber collimating optics. This radiation is then coupled into the selected output optical fiber.

Description

TECHNICAL FIELD

[0002] The present invention relates generally to a thermally-activated bimorph optical switching element. More particularly, the invention relates to an optical switching system for fiber optic cross connect and add-drop multiplexers in wavelength division multiplexing systems.

BACKGROUND OF THE INVENTION

[0003] The growth of the Internet and the World Wide Web has created a demand for large data bandwidths and high-speed channel switching within the telecommunications industry. Traditionally, telecommunications and other data transfer signals have been exchanged electrically via electrically-conductive wires and cables. Increased bandwidth requirements have led to the adoption of optical signal transfer via fiber optic cables. Optical fiber communications were originally developed for large bandwidth applications, such as long distance trunk cables connecting local metropolitan telephone exchanges. As the cost of these fiber optic networks has decreased and the demand for signal bandwidth increased, fiber optic communications networks are being installed in local metropolitan area networks, and as parts of local area networks (LANS) for data exchange between computers in offices and other commercial and government environments.

[0004] Crucial elements in these fiber optic communications networks are the switches and routers that direct the optical signals from various sources to any one of a multitude of destinations. Such switches can be used in a variety of applications, including N×N cross-connect switches for switching signals between arrays of input and output optical fibers, add-drop multiplexers in wavelength-division-multiplexing (WDM) and the more recent dense-wavelength-division-multiplexing (DWDM) systems, reconfigurable networks, and hot backups for vulnerable components and systems. These and other similar applications require switches with moderate switching speeds of one millisecond or less, low insertion loss, low noise, low dispersion, and low cross talk. Additional needs are for low cost, small form factor, and easily manufacturable components to facilitate the rapid adoption of these technologies in wider arrays of optical switching applications.

[0005] Many switching techniques have been developed to facilitate the generation and transfer of these optical signals in fiber optic networks. Initially, to switch signals between various arrays of optical fibers, the optical signals had to be converted to electrical signals, conditioned, amplified, and routed to laser light sources and modulators before being retransmitted to the output optical fiber. Electrical switching schemes are undesirable in these applications due to the reduced bandwidth, increased noise, and added cost and complexity of the resulting system designs. As a result, several all-optical switching schemes have been developed, or are under development, for use in telecommunications and data transfer networks, but all of these designs have limitations in these applications.

[0006] What is needed, therefore, is a reliable, low-cost, high-bandwidth, low-loss optical switching system.

SUMMARY OF THE INVENTION

[0007] The foregoing and other needs are met by a device for switching optical signals completely within optical media, from an arbitrary number of N input optical fibers to a different set of M output optical fibers. The invention uses a thermally-activated bimorph optical switch to redirect optical radiation emitted by a laser light source, or from the end of an optical fiber, to an input end of another optical fiber. The modulated optical radiation from the input fiber optic bundle is collimated into parallel beams and projected in free space across the tops of an array of microcantilever bimorph optical switches. When selected, a particular switch is activated, thereby causing a first bimorph switch element to pop up, and intercept and reflect the optical radiation at an angle approximately 90° to the original optical path direction. This reflected radiation is intercepted by another essentially identical bimorph switch element, located above or below the first switch element, that directs the radiation in a plane essentially parallel to that of the original beam plane, but above or below the original plane, and at approximately right angles to the original beam direction. The optical radiation is directed by the second switch element to output optical fiber collimating optics. This radiation is then coupled into the selected output optical fiber.

[0008] Both the input and output optical fibers and the bimorph switching elements can be operated in an interchangeable fashion. Consequently, the optical radiation can travel in either direction once a connection between two optical fibers has been established. Thus, the invention is a bi-directional optical switching system with complete symmetry between the inputs and outputs of the switch. Since the input and output optical radiation traverses two different and parallel optical planes, the input and output optical beams do not cross or interact when traversing the switch.

[0009] The microcantilever (or bimorph) optical switch elements are composed of two or more thin film layers that possess large differences in their thermal expansion coefficients. When heated, the difference in thermal expansion of the layers induces a surface stress on the cantilever structure, causing it to bend to relieve the resultant stress. The bending of the cantilever structures causes the free end of the cantilevers to move to intercept the optical radiation emitted by the input optical fiber.

[0010] The bottom layer of the bimorph switch elements is fabricated from a metal that is highly reflective at the wavelength of the optical radiation, and that possesses a large thermal expansion coefficient. A thin electrically-insulating middle layer is preferably included in the structure to electrically isolate the bottom layer from the upper layer. The upper layer is fabricated from a low thermal expansion, doped semiconductor material, which is patterned and doped to create a resistive heater for uniformly heating the bimorph structure.

[0011] When a switch is selected, a pulsed electrical current passes through the resistive heaters of the switch elements, thereby heating the bimorph structures. When the bimorph structures are heated, the free end of the cantilevers of both structures rise and intercept the optical beam. The bending angle of the bimorph sections of each of the switch paddles is preferably about 45°. The optical radiation is reflected by the upper metal surface of the lower bimorph switch paddle, or the lower metal surface of the upper switch paddle, in an intermediate direction at an angle approximately 90° to the original path of the optical beam. The optical radiation is then intercepted by the lower surface of the upper switch element, or the upper surface of the lower switch element, and is reflected at an angle that is approximately 90° to both the original beam direction and the intermediate beam direction, and in a beam plane that is parallel to, but not coplanar with, the original beam direction. The reflected optical radiation is directed to the collimating optics of an adjacent output optical fiber and is focused onto the core region of the output fiber.

[0012] The optical switching device of the present invention offers the following benefits: (1) high speed switching with thermal time constants of less than one millisecond, (2) N×N switching operation, (3) scalability to large switch arrays where N may be several hundred to several thousand switch array dimensions, (4) one-dimensional switch paddle controls with simple feedback control to precisely align the switch paddles to the input and output optical fibers, (5) in-plane switch construction with resultant ease of fabrication and alignment, (6) low insertion losses and cross talk, (7) low switch activation power on the order of milliwatts per switch element, (8) small physical size, and (9) low cost fabrication due to complete silicon integrated circuit compatibility.

[0013] In one aspect, the invention provides an optical switch for directing at least one optical beam from a first propagation plane to a second propagation plane that is parallel with the first propagation plane. The switch includes a first switch portion having a substantially-planar first reflective surface portion. The first reflective surface portion is operable to move in a first pitch plane between a first deflection angle and a second deflection angle. When at the first deflection angle, the first reflective surface portion intercepts the optical beam propagating along a first propagation path in the first propagation plane and reflects the optical beam at a first reflection angle into an intermediate propagation path. When at the second deflection angle, the first reflective surface portion does not intercept the optical beam. The switch also includes a second switch portion having a substantially-planar second reflective surface portion. The second reflective surface portion is operable to move in a second pitch plane between at least the first deflection angle and the second deflection angle, where the second pitch plane is substantially perpendicular to the first pitch plane. When at the first deflection angle, the second reflective surface portion intercepts the optical beam propagating along the intermediate propagation path, and reflects the optical beam at a second reflection angle into a second propagation path in the second propagation plane. When at the second deflection angle, the second reflective surface portion does not intercept the optical beam.

[0014] In a preferred embodiment, the first switch portion includes a first paddle disposed adjacent the first propagation path. The first paddle has a fixed portion and a free end, with the first reflective surface portion adjacent the free end. The first paddle is operable to bend upon a change in temperature, thereby moving the first reflective surface portion between the first and second deflection angles. The second switch portion includes a second paddle oriented substantially perpendicular to the first paddle. The second paddle, which is disposed adjacent the second propagation path, has a fixed portion and a free end, with the second reflective surface portion adjacent the free end. The second paddle is operable to bend upon a change in temperature, thereby moving the second reflective surface portion between the first and second deflection angles.

[0015] Preferably, the first and second paddles each include an upper layer having a first coefficient of thermal expansion, and a lower layer mechanically coupled to the upper layer. The lower layer has a second coefficient of thermal expansion that is different from the first coefficient of thermal expansion. The change in temperature causes the upper and lower layers to change in size at different rates due to the differences in coefficients of thermal expansion, thereby creating stress in the paddle and causing the paddle to bend.

[0016] In another aspect, the invention provides an optical switching device for selectively transferring at least one optical beam between at least one first optical channel and M number of second optical channels. The device comprises a first optical structure for propagating the optical beam along a first propagation path in a first propagation plane. The device further comprises a 1×M-dimensional array of optical switches that includes M number of first switch portions adjacent to and aligned in parallel with the first propagation path. The first switch portions have substantially-planar first reflective surface portions that are operable to move in a first pitch plane between a first deflection angle and a second deflection angle. When at the first deflection angle, one of the first reflective surface portions intercepts the optical beam propagating along the first propagation path in the first propagation plane and reflects the optical beam at a first reflection angle into an intermediate propagation path. When at the second deflection angle, the first reflective surface portions do not intercept the optical beam. The array of optical switches further includes M number of second switch portions adjacent the corresponding first switch portions. The second switch portions have substantially-planar second reflective surface portions that are operable to move in second pitch planes between the first deflection angle and the second deflection angle. The second pitch planes are substantially perpendicular to the first pitch plane. When at the first deflection angle, one of the second reflective surface portions is operable to intercept the optical beam propagating along the intermediate propagation path and reflect the optical beam at a second reflection angle into one of M number of substantially parallel second propagation paths in a second propagation plane. The second propagation paths are substantially perpendicular to the first propagation path, and the second propagation plane is substantially parallel with the first propagation plane. When at the second deflection angle, the second reflective surface portions do not intercept the optical beam. The device also includes M number of second optical structures disposed in the second propagation plane. Each of the second optical structures is optically aligned with a corresponding one of the M number of second switch portions in a corresponding one of the M number of second propagation paths. Each of the second optical structures is operable to receive the optical beam from the corresponding one of the second reflective surface portions and direct the optical beam to a corresponding one of the M number of second optical channels.

BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Further advantages of the invention will become apparent by reference to the detailed description of preferred embodiments when considered in conjunction with the drawings, which are not to scale, wherein like reference characters designate like or similar elements throughout the several drawings as follows:

[0018] FIG. 1 depicts an optical switching device according to a preferred embodiment of the invention;

[0019] FIG. 2 is an isometric view of a portion of the optical switching device containing four switches according to a preferred embodiment of the invention;

[0020] FIG. 3A is an isometric view of one non-activated bimorph switch element according to a preferred embodiment of the invention;

[0021] FIG. 3B is an isometric view of one activated bimorph switch element according to a preferred embodiment of the invention;

[0022] FIG. 4A is an elevation view of one non-activated bimorph switch element according to a preferred embodiment of the invention;

[0023] FIG. 4B is an elevation view of one activated bimorph switch element according to a preferred embodiment of the invention;

[0024] FIG. 5A is a top view of one non-activated bimorph switch element according to a preferred embodiment of the invention;

[0025] FIG. 5B is a top view of one activated bimorph switch element according to a preferred embodiment of the invention;

[0026] FIG. 6 is schematic diagram of a circuit for selectively activating individual switches of the optical switching device according to a preferred embodiment of the invention;

[0027] FIG. 7 is an isometric view of an active feedback circuit used for alignment of the reflecting surfaces of the bimorph switch to optimize signal transmission through the switch according to a preferred embodiment of the invention;

[0028] FIGS. 8A-E depict expected locations of the optical beam on a four-quadrant optical detector under various alignment conditions according to a preferred embodiment of the invention;

[0029] FIG. 9 is a schematic diagram of an equivalent thermal circuit of a bimorph optical switch according to a preferred embodiment of the invention;

[0030] FIG. 10 is a graph depicting modeled response time of a bimorph switch element as a function of the length of a thermal isolation gap for several switch paddle lengths;

[0031] FIG. 11 is a graph depicting modeled switch positional responsivity in microns/C as a function of the thickness of the polysilicon resistive film layer for several different bi-material film thicknesses;

[0032] FIG. 12 is a graph depicting the maximum temperature required to fully activate a bimorph switch as a function of the polysilicon film thickness for several bi-material film thicknesses; and

[0033] FIGS. 13A-C show three graphs depicting modeled switch paddle tip position as a function of time after application of a switch heating pulse.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

[0034] FIG. 1 depicts a cross-connect optical switching device 10 for switching optical signals between a first optical structure, comprising N number of first optical fibers F1N, to a second optical structure, comprising M number of second optical fibers F2M. The optical signals, such as telephone or data signals in a telecommunication network, generally consist of modulated optical radiation. In the example depicted in FIG. 1, there are five first fibers F11-F15 (N=5) and five second fibers F21-F25 (M=5). However, it should be appreciated that the invention is not limited to any particular number of first or second channels.

[0035] In the preferred embodiment of the invention, the switching device 10 is a completely reciprocal device, such that optical fibers F1N and F2M can act as input fibers or output fibers, depending on the direction of propagation of the optical signals. Thus, it should be appreciated that the operation of the invention is not limited to the direction of propagation of the optical signals. Further, at any particular instant in time, any one or more of the fibers F1N could be input fibers, while other of the fibers F1N are output fibers. Similarly, at any particular instant in time, any one or more of the fibers F2M could be input fibers, while other of the fibers F2M are output fibers.

[0036] When the first fibers F1N act as input fibers, the optical signals radiate from the ends of the first fibers F1N and are captured by beam-collimating optics, such as lenses L1N. The lenses L1N collimate the optical signals into N number of parallel beams of light, each of which is projected in free space over M number of microcantilever optical switches SNM. When the second fibers F2M act as input fibers, the optical signals radiate from the ends of the second fibers F2M and are captured by beam-collimating optics, such as lenses L2M. The lenses L2M collimate the optical signals into M number of parallel beams of light, each of which is projected in free space over N number of the microcantilever optical switches SNM.

[0037] In an alternative embodiment, the first or second optical structures comprise modulated light sources, such laser sources or light-emitting diode (LED) sources, for generating modulated optical signals. In this embodiment, the first optical fibers F1N or the second optical fibers F2M are replaced by the modulated light sources.

[0038] As shown in FIG. 1, a preferred embodiment of the device 10 includes optical switches S11- S15 aligned in a column substantially parallel to the beam emitted from the lens L11, the switches S21- S25 aligned in a column substantially parallel to the beam emitted from the lens L12, and so forth. As described in more detail hereinafter, when a particular one of the switches, such as S25, is selected, a first portion S25a of the switch S25 pops up, intercepts the corresponding optical beam, and reflects the optical beam toward a second portion S25b of the switch S25. The radiation is reflected from the first portion S25a in an intermediate direction at an angle of about 90°, travels a short distance in the intermediate direction, and is reflected from the second portion S25b that popped up when the switch S25 was selected. Based on the angle of reflection from the second portion S25b, which is also preferably about 90°, the radiation is directed into a second lens L25. The second lens L25 directs the optical radiation into the corresponding second fiber F25. Preferably, there are M number of second lenses L2M corresponding to the M number of second fibers F2M.

[0039] In the example of FIG. 1, switches S25, S31, S43, and S52 are selected to route optical signals between the first fiber F12 and second fiber F25, between first fiber F13 and second fiber F21, between first fiber F14 and second fiber F23, and between first fiber F15 and second fiber F22, respectively.

[0040] Thus, each column of switches, such as S11-S15, forms a 1×M-dimensional array which is capable of routing optical signals between any one of the N number of first optical fibers F1N and any one of the M number of second optical fibers F2M, and between any one of the M number of second optical fibers F2M and any one of the N number of first optical fibers F1N, completely in optical media. Since there is no need to convert the input signals from an optical format to an electrical format to accomplish the switching, there is little or no loss in bandwidth. Further, the switching system 10 introduces much less insertion loss than would be introduced by a system that requires conversion from optical to electrical and back to optical formats. Other advantages of the invention will be obvious as further details of the system 10 are described below.

[0041] With continued reference to FIG. 1, the preferred embodiment of the invention includes N number of output lenses Lo1N disposed on the opposite side of the array of switches from the first lenses L1N, and M number of output lenses Lo2M disposed on the opposite side of the array of switches from the second lenses L2M. If no switch SNM is selected in a given column of the array, then the optical beam from a first lens L1N corresponding to that column is incident upon the corresponding output lens Lo1N. For example, as shown in FIG. 1, the optical beam from the first fiber F11 and lens L11 is not intercepted by any of the switches S11-15, and is therefore incident upon the output lens Lo11. Also, the optical beam from the second fiber F24 and lens L24 is not intercepted by any of the switches S14-S54, and is therefore incident upon the output lens Lo24. The lenses Lo11 and Lo24 focus the beams onto corresponding optical light detectors D11 and D24 which convert the optical signals into electrical signals. The electrical signals from the detectors D11 and D24 are provided to a switch control logic unit 19.

[0042] Generally, the modulated optical signals carried by the optical beams include header information indicating the source and the intended final destination of the information carried by the optical signals. Preferably, before any switch in a column of an array is activated, the switch control logic unit 19 receives and decodes the header information in the optical beam, and determines which switch in the column should be activated to send the signal to the appropriate output fiber to carry the signal to its final destination. Based on the header information, the switch control logic unit 19 generates a switching signal that is used in selecting the appropriate switch in the array. Preferably, after a particular packet of information has been routed through to the appropriate output fiber by selection of one of the switches SNM, the selected switch is deactivated, and the optical beam is again provided to the corresponding one of the detectors D1N or D2M so that the header information in the next information packet may be processed.

[0043] As shown in FIG. 1, the preferred embodiment of the switching device 10 includes a number of spare switches SS11-SS25 that may be used in the event that any of the switches S11-S55 fail. For example, if any of the switches S11-S15 were to fail, an input signal on fiber F11 could be rerouted to the spare fiber Fs1, and the spare switches S11-SS15 would be used in place of the switches S11-S15.

[0044] In the preferred embodiment, the system 10 of FIG. 1 is operable in a number of switching modes. The optical beams can be redirected from N number of first optical fibers F1N to an equivalent number of second optical fibers F2N(M=N) in an arbitrary fashion to form an N×N switch. Alternatively, the signals from N number of first optical fibers F1N can all be directed into a single one of the second fibers F2M(M=1) to form an N×1 switch. In yet another configuration, the radiation from two or more first optical fibers F1N can be combined in an arbitrary fashion to produce an output array with up to N operational second optical fibers F2N(an N×M switch). Thus, N and M may be any number, and are completely independent.

[0045] FIGS. 2, 3A-B, 4A-B, and 5A-B depict portions of the switching system 10 in more detail. FIG. 2 is an isometric view of four of the switches, such as S11, S12, S21, and S22. Switch S21, in a non-activated position, is depicted in even greater detail in FIGS. 3A, 4A, and 5A. Switch S22, in an activated position, is depicted in FIGS. 3B, 4B, and 5B. Preferably, each switch SNM includes a cantilevered paddle 20 having a fixed end attached to a paddle support post 35 and a free end extending over a cavity region 22. The paddle section 20 is preferably 50-500 microns in length, 20-200 microns in width, and composed of three primary layers: a lower layer 24, a middle layer 26, and an upper layer 28. The lower paddle layer 24 is preferably metal that is highly reflective at the wavelength of the optical radiation, and which possesses a large coefficient of thermal expansion, such as Al, Au, Pb, or Zn. The middle paddle layer 26 is a thin electrically-insulating dielectric layer for electrically isolating the lower layer 24 from the upper paddle layer 28. Preferably, the middle layer 26 is composed of SiO2, SiC, Si3N4, SixOyNz or HxSiyCz. In the preferred embodiment, a very thin bonding layer 30, such as Cr, Ti, TiW, TiN, or nichrome, is used to bond the lower metal layer 24 to the middle insulating layer 26.

[0046] The paddle support post 35, which mechanically connects the fixed end of the paddle 20 to the substrate 21, is preferably constructed from silicon carbide, silicon nitride, or hydrogenated/oxygenated versions of these materials. As well as supporting the fixed end of the paddle 20, the paddle support post 35 provides electrical isolation between the vias 34 and 36. The substrate 21 is preferably formed from a silicon semiconductor wafer.

[0047] The upper paddle layer 28 is preferably fabricated from a semiconductor material having a low coefficient of thermal expansion, such as doped polysilicon. The upper paddle layer 28 is patterned, preferably in a serpentine shape, and doped to somewhat increase its electrical conductivity, thereby forming serpentine resistive heaters 32a and 32b. Preferably, the heaters 32a and 32b have a resistivity of a few hundred to a few thousand ohms. As described in more detail below, when an electric current flows through the heater 32a or 32b, it uniformly heats the paddle section 20 which causes activation of the switch. In the preferred embodiment of the invention, the resistive heaters 32a and 32b are connected to underlying control and voltage supply lines by via structures 34 and 36, which are preferably formed from aluminum.

[0048] In the preferred embodiment, the switch paddles 20 are surrounded by an inert gas filled environment. The gas is preferably contained within a switch enclosure that covers the switch arrays. As discussed in more detail hereinafter, among other things, the gas surrounding the paddles 20 serves to dampen oscillatory vibrations of the paddles 20 when the switches are activated.

[0049] An example of the operation of the bimorph switches S11-S55 is depicted in FIGS. 4A and 4B. As shown in FIG. 4A, which is an elevation view of the switch S21 in the direction I (FIG. 3A), optical radiation, substantially in the form of an input beam 60a, is emitted from the lens L12 (FIG. 1). Since switch S21 is not selected, the input beam 60a continues uninterrupted past the switch S21. Also, optical radiation from the lens L21 (FIG. 1), in the form of an input beam 62a, continues uninterrupted past the switch S21. As shown in FIG. 4B, which is an elevation view taken in the direction I (FIG. 3B), when the switch S21 is selected, an electrical current flows through the resistive heaters 32a and 32b, and the bimorph structures of the paddles 20 are heated to preferably less than 100° C. When the paddles 20 are so heated, the various layers of the paddles 20 begin to expand. However, since the lower layers 24 possesses a higher coefficient of thermal expansion than do the upper layers 28, the lower layers 24 expand at a greater rate than do the upper layers 28, thereby creating differential stress across the bimorph structures. This causes the paddles 20 to bend upward, thereby raising the free ends of the paddles by 10-20 microns or more. A reflective surface portion, also referred to herein as a mirror section 25, of the lower metal layer 24 of the switch portion S22a intercepts the input beam 60a, and reflects the input beam 60a in the intermediate direction to form an intermediate beam 60b propagating toward the paddle 20 of the switch portion S22b. The mirror section 25 of the switch portion S22b intercepts the intermediate beam 60b, and reflects the intermediate beam 60b to form the output beam 60c propagating toward the lens L22. As indicated in FIG. 4B, a selected switch, such as S22 completely intercepts the output beam 60c in this embodiment of the invention. Thus, any one input optical beam, such as emitted from any one of the first fibers F11-F15, is preferably directed to only a single one of the second fibers F21-F25 at any given time.

[0050] FIG. 6 depicts a preferred selection circuit 52 for selectively driving the heaters 32a and 32b to activate the switches S11-S55. The circuit 52 includes a second select logic circuit 54 and multiple first select logic circuits 56. Generally, the second select logic circuit 54 selects one or more rows of the switches S11-S55associated with one or more of the second fibers F2M, and the first select logic circuits 56 select which of the switches S11-S55 within the selected row or rows are to be activated. The first select logic circuits 56 are connected to the second select logic circuit 54 by second select lines LS21-LS25, and the first select logic circuits 56 are connected to semiconductor switching devices D11a-D55a and D11b-D55b by first select lines LS11a-LS55a and LS11b-LS55b. In the preferred embodiment, the semiconductor switching devices D11a-D55a and D11b-D55b are CMOS transistor devices. So as not to unnecessarily complicate FIG. 6, only two input select logic circuits 56a and 56b and two sets of associated switches S11-S51 and S12-S52 are depicted. It should be appreciated that the other input select logic circuits 56 and associated switches S13-S53, S14-S54, and S15-S55 are configured according to the pattern depicted in FIG. 6.

[0051] As shown in FIG. 6, one side of each of the semiconductor switching devices D11a-D55a and D11b-D55b is preferably connected to the low side of an associated one of the resistive heaters 32a and 32b, respectively. The other side of each of the semiconductor switching devices D11a-D55a and D11b-D55b is preferably connected to a common return bus 57. The high side of each of the resistive heaters 32 is preferably connected to a voltage supply 58 for providing the voltage Vs through the supply bus 55. In the preferred embodiment, the voltage supply 58 provides Vs in pulses of 3-5 volts in amplitude and having pulse widths of 1-10 μs. It is anticipated that during normal switch activation, the pulse duty cycle is approximately 50%. To facilitate rapid initial heating of the switch elements to decrease the switch activation time, the duty cycle may be increased to as much as 100%, if needed.

[0052] As an example of the operation of the selection circuit 52, consider a situation in which it is desired to transfer an optical signal from the first fiber F13 to the second fiber F21 (FIG. 1). The output select logic circuit 54 generates an output select signal on the line LS21 having a 3-bit binary value, such as 011. The input select logic circuit 56A decodes the value 011 and sets the lines LS31a and L31b high. When the lines LS31a and LS31b go high, the switching devices D31a and D31b are turned on, thereby connecting the low side of the associated heaters 32a and 32b to the ground bus 57. With the circuit completed to ground, current flows through the heaters 32a and 32b associated with the switch S31, thereby generating heat that activates the associated paddles 20.

[0053] As shown in FIGS. 3A-B, 4A-B, and 5A-B, in the preferred embodiment of the invention, the mirror section 25 that intercepts the optical beam extends beyond the ends of the middle layer 26 and upper layer 28. As the paddle 20 is heated, only the bi-material section of the paddle 20, where the upper layer 28 and the lower layer 24 overlap, bends due to the difference in thermal expansion rates. Thus, the mirror section 25 remains relatively planar when the paddle 20 is heated. As FIG. 4B indicates, since the mirror section 25 of the paddle 20 remains substantially planar, the upper edge of the optical beam 60a is incident upon the mirror section 25 at about the same angle as the lower edge of the beam 60a, such that the beam 60a does not substantially diverge when reflected from the mirror section 25.

[0054] One of the advantages of the invention is a somewhat relaxed requirement on the positioning of the free end of the paddle 20. When a switch is not activated, the free end of the paddle 20 merely needs to be positioned out of the beam path. When activated, the free end of the paddle 20 should get at least high enough to preferably intercept all of the optical radiation. Thus, the paddle 20 can be designed such that variations of several microns in the on or off position of the free end of the paddle 20 will not affect the operation of the switch.

[0055] As one skilled in the art will appreciate, if the mirror sections 25 of the two portions SNMa and SNMb of a switch SNM do not deflect to the correct angle θ upon activation, the reflected optical beam will not be correctly aligned with the lenses L1N and L2M. Preferably, as shown in FIGS. 3B and 4B, the deflection angles of the reflecting surfaces are about 45°. If the deflection angles θ are not correct, thereby causing beam misalignment, significant loss of signal power may occur.

[0056] One of the advantages of the present invention is that the orthogonal arrangement of the paddles 20 of the two switch portions SNMa and SNMb allows for simple and accurate alignment of the reflecting surfaces with respect to the first and second lenses L1N and L2M and optical fibers F1N and F2M. In a preferred embodiment, as depicted in FIG. 7, beam splitters 70 are interposed between the second lenses L2M and the second fibers F2M. Each beam splitter 70 extracts a small fraction of the optical signal incident thereon, and directs it onto an optical detector 72, which is preferably a four-quadrant photodetector. The position of the beam on the detector 72 is determined by the deflection angles θ of the two orthogonal mirror sections 25 of the paddles 20. If the beam is not exactly centered on the detector 72, with each quadrant of the detector 72 receiving exactly the same amount of signal, then the resultant offset signal is used as a feedback signal to control the voltage supply 58 to apply more or less power to one or both of the resistors 32a and 32b. By adjusting the voltage Vs applied to the resistors 32a and 32b, the heat generated by the resistors 32a and 32b is controlled, thereby controlling the angles of deflection θ of the mirror sections 25. In this manner, the beam reflected from the mirror section 25 may be properly centered on the lens L2M and in the fiber F2M.

[0057] Specifically, when the angle θ of the mirror section 25 of the switch portion SNMa is too large or too small, a vertical offset signal results in the detector 72. If the angle θ of the mirror section 25 of the switch portion SNMb is too large or too small, a horizontal offset signal results. Consequently, two distinct signals can be generated for independently controlling the angles θ of the mirror sections 25. In contrast to known multi-mirror switching approaches, each mirror section 25 of the device 10 moves in a one-dimensional space, thereby facilitating simple feedback and control of the alignment of the reflecting surfaces.

[0058] One skilled in the art will appreciate that the beam splitter 70 and the detector 72 could be positioned in the beam path between the first lenses L1N and the first fibers F1N, or between the second lenses L2M and the second fibers F2M, depending on the direction of propagation of the optical beam used for alignment.

[0059] In an alternate alignment approach, alignment radiation from an external light source, operating at a different wavelength from that of the optical signal-carrying beam, can be introduced, such as through a beam combiner, into the same optical path as that followed by the signal-carrying beam. The alignment beam is then intercepted and extracted from the signal-carrying beam, such as by using a dichroic mirror that highly reflects the alignment beam, but which is highly transmissive of the signal-carrying beam. The alignment beam is then directed onto a four quadrant detector as in the above-described method. In this manner, the alignment beam traverses exactly the same optical path through the optical switch SNM as does the signal-carrying beam. The advantage of this approach is that none of the signal-carrying beam light is used for the alignment detection function, so that the signal-to-noise ratio of the signal-carrying beam is not thereby reduced.

[0060] Some of the important physical parameters affecting the operation of the thermally-activated bimorph switch of the present invention include (1) the thermal time constant, τth, of the bi-material switching element, which determines the fastest switching time for a given set of switch design parameters, and which is dependent on the material properties and dimensions of the thin film structures; (2) the switch stabilization time, ts, which is the time required for the switch paddle 20 to stabilize, and which is dependent on the mass and stiffness of the paddle, and viscous air damping among other parameters; (3) the temperature, Tmax, required to fully activate the switch to intercept the optical radiation, which depends on the positional responsivity of the bimorph structure along with the maximum displacement of the free end of the paddle 20 and the precise deflection angle required to intercept and redirect the optical radiation beam; (4) the maximum power, Pmax, required to heat the bimorph structure to cause the mirror section 25 of the paddle 20 to fully intercept the optical beam, which should be minimized to reduce switch thermal management problems; and (5) the supply voltage, Vs, and polysilicon conductivity, σpoly, required to provide the maximum power Pmax.

[0061] As discussed in more detail below, there is also a maximum temperature rise ΔTmax within the bimorph structure which constrains the design. The material properties of the thin films of which the switch is comprised limit the maximum temperature to which these materials can be exposed over extended periods of time and number of switching operations before significant performance degradation and switch failure can be expected.

[0062] The thermal response of the switch is analyzed using an equivalent thermal circuit, such as depicted in FIG. 9. The thermal time constant τth of the switch elements determine at least in part the speed at which the switch can both be activated and deactivated. Generally, the faster the rate of increase and decrease in the switch paddle temperature, the more quickly the paddle 20 moves to intercept or unblock the optical radiation. The thermal time constant τth is determined by the thermal capacitance Cth and the thermal impedance Rth of the switch paddle structure according to:

τth=Rth×Cth.   (1)

[0063] The thermal capacitance Cth of the switch is determined by the capacities of the individual thin film structures which make up the switch paddle 20 shown in FIGS. 4A-B and 5A-B. The thermal capacitance of a specific thin film layer Cth(i) is determined according to:

C th (i)=ρ(i)×λ(i)×t(i)×l(i)×w(i),   (2)

[0064] where ρ(i) is the density, λ(i) is the specific heat, t(i) the thickness, l(i) the length, and w(i) the width of the specific layer. The lumped thermal capacitance of the switch paddle 20 is the sum of all the individual layer capacitances, i.e.: 1$\begin{array}{cc}{C}_{\mathrm{th}}=\sum _i{C}_{\mathrm{th}}\left(i\right).& \left(3\right)\end{array}$

[0065] Likewise, the thermal impedance of the switch is controlled by the thermal conductivity of the paths to the substrate through the switch structure and the surrounding environment as shown by the thermal impedance pathways represented in FIG. 9. The radiation thermal resistance, Rthr, for the paddle structures of the preferred embodiment is found from the following expression:

R −1 thr =4×Ad×(εb+εt)×σ×Ts3,   (4)

[0066] where Ad is the area of the paddle 20, εb and εt are the emissivities of the top and bottom surfaces of the switch paddle respectively, σ is the Stefan-Boltzmann constant, and Ts is the temperature of the paddle. For the paddle areas and materials of the preferred embodiment of the invention, Rthr is approximately 107 to 108 ° K./watt, and contributes negligibly to the switch power dissipation. In a gas-filled switch enclosure, the conductive and convective thermal impedances, Rcon, (FIG. 9) are difficult to calculate analytically. An approximate estimate of the contribution to thermal impedance losses can be found in “Advances in Amorphous Silicon Uncooled IR Systems” by J. Brady et al., SPIE Conference on Infrared Technology and Applications XXV, Orlando, Fla., April 1999, SPIE Vol. 3698. The Brady reference documents the thermal impedance of similar structures measured as a function of the fill gas pressure. For the switch paddle areas of the preferred embodiment, Rcon is approximately 106 to 107° K./watt. This mechanism also contributes only a small thermal loss in comparison with the thermal impedance (i.e., Rth=104 to 105° K./watt) and hence much higher power losses due to thermal conduction through the paddle structure to the support posts and underlying substrate. Thus, the Rcon loss mechanisms are neglected in the following analysis.

[0067] The remaining impedance pathways shown in FIG. 9 are conduction pathways through the various layers making up the bimorph switch structure. The thermal impedance for a path (i) is determined by: 2$\begin{array}{cc}{R}_{\mathrm{th}}\left(i\right)=\frac{l\left(i\right)}{2\times \kappa \left(i\right)\times t\left(i\right)\times w\left(i\right)},& \left(5\right)\end{array}$

[0068] where κ(i) is the thermal conductivity of the material of which path (i) is composed. The total conductive thermal impedance Rth to the interconnects and metal vias, and thus to the silicon substrate, is given by the following equation for the structure shown in FIGS. 4A-B and 5A-B: 3$\begin{array}{cc}{R}_{\mathrm{th}}={\left(\frac{1}{R_{\mathrm{poly}}}+\frac{1}{R_{\mathrm{insul}}}\right)}^{-1}+{\left(\frac{1}{R_{\mathrm{poly}}}+\frac{1}{R_{\mathrm{insul}}}+\frac{1}{R_{\mathrm{bond}}}+\frac{1}{R_{\mathrm{metal}}}\right)}^{-1}& \left(6\right)\end{array}$

[0069] where, as shown in FIG. 9, the thermal impedance Rpoly of the polysilicon upper layer 28 is in parallel with the thermal impedance Rinsul of the insulating middle layer 26 for the section closest to the switch substrate, and is in series with the section containing the parallel thermal impedances Rpoly, Rinsul, the bonding metal layer Rbond, and the thermal impedance Rmetal of the metal lower layer 24. In this analysis, the thermal impedances of the lower layer 24 and bonding layer 30 of the paddle 20 are assumed to be negligible in comparison with the other layers, as layers 24 and 30 are composed of high thermal conductivity metals. The overlapping insulation layer 26 is physically small such that it also makes a negligible contribution to the switch thermal impedance. In practice, the thermal response time of the switch is preferably controlled primarily by the length of the thermal isolation gap, Lgap, (FIGS. 4A-B) as shown by the modeled results in FIG. 10. The switch can be designed to have a thermal time constant τth in the range from several tens of microseconds to seconds, depending on the switching application (e.g. switch size, required paddle movement, materials, design, and power dissipation) as shown in FIG. 10 and in Table II below.

[0070] The thermal properties of several materials that are preferred for use in the invention are shown in Table I. The material properties shown in Table I are not necessarily accurate values for the thin films used in the fabrication of the switch, as they are, in most cases, bulk material properties, which can be significantly different from the thin film values, depending the film thickness and fabrication technique. Nevertheless, they can be used in the present calculations to estimate the dependence of the switch parameters on these materials. The electrical impedances of the insulator materials are not included in the tables as they, being very large, have no bearing on the calculation results presented in Table II below.

	TABLE I
	Thermal
	Expansion	Thermal	Specific	Young's		Intrinsic
	Coefficient	Conductivity	Heat	Modulus	Density	Resistivity
	(10−6 K−1)	(W/m-K)	(J/g-K)	(1011 N/m2)	(g/cm3)	(Ω-cm)
Semi-
conductor
Polysilicon	4.7	33.2	0.702	1.9	2.33	1 × 105
Insulators
SiO2	0.55	1.4	0.17	0.73	2.65	—
SiC	3.5	2.1	0.17	7.0	3.21	—
A-HxSiyCz	5	0.34	0.17	0.7	3.2	—
Si3N4	0.8	19	0.17	3.85	3.17	—
Bimetals
Al	25.0	237	0.897	0.70	2.70	2.65 × 10−8
Au	14.2	317	0.129	0.8	19.3	2.27 × 10−8
Zn	35.0	116	0.388	0.79	7.13	5.5 × 10−8
Pb	28.9	35	0.129	0.16	11.3	20 × 10−8
Bonding
Metals
Cr	4.9	93.7	0.449	1.52	7.15	12.7 × 10−8
Ti	8.6	21.9	0.523	1.2	4.51	39 × 10−8
W	4.5	174	0.116	4.1	19.3	5.4 × 10−8
TiW	6	14.4	0.21	3.5	16.3	10 × 10−8
Ni	13.4	90.7	0.444	2.1	8.9	6.9 × 10−8
nichrome	10	12	0.445	2.1	8.5	11 × 10−8

[0071]

	TABLE II
	Symbol	Design 1	Design 2	Design 3	Design 4	Units
Parameter
Tip excursion	Zmax	26	26	26	35	μm
Total Paddle length	Lpaddle	250	250	225	275	μm
Paddle width	Wpaddle	20	20	20	20	μm
Length polysilicon	Lpoly	150	150	125	175	μm
Length bi-material region	Lmetal	150	150	150	200	μm
Length thermal gap	Lgap	50	50	25	25	μm
Length of straight mirror	Lmirror	50	50	50	50	μm
Thickness of bi-material	tmetal	0.2	0.3	0.2	0.2	μm
layer
Thickness of Bonding	Tbond	0.01	0.01	0.01	0.01	μm
metal
Thickness of polysilicon	tpoly	0.1	0.15	0.1	0.1	μm
Thickness of insulator	tinsul	0.02	0.01	0.01	0.01	μm
Supply voltage	VR	5	5	5	5	V
Switch base temperature	T0	300	300	300	300	K
Calculated Results
Switch thermal impedance	Rth	8.3 × 104	5.6 × 104	4.2 × 104	4.2 × 104	K/watt
Switch thermal capacitance	Cth	2.8 × 10−9	4.1 × 10−9	2.7 × 10−9	3.4 × 10−9	J/K
Thermal time constant	τth	0.23	0.23	0.113	0.142	msec
Responsivity	RT	0.58	0.39	0.58	1.04	μm/K
Change in temperature	ΔT	44	66.5	44	33	K
Thermal power	Pth	0.53	1.2	1.07	0.8	Watt
Electrical Impedance	RSi	5.3 × 104	2.9 × 104	4.7 × 104	4.7 × 104	Ω
Polysilicon resistivity	σSi	113	121	99	128	(Ω · cm)2

[0072] The results of several example calculations of the thermal response time for a given switch size and paddle thickness are shown in Table II. It should be appreciated that there are many possible combinations of materials and switch physical dimensions, with the optimum combination dependent on the particular switching application.

[0073] In the preferred embodiment of the invention, the switch is fully activated when the two following conditions are satisfied. First, the vertical displacement Δz of the free end of the switch paddles 20, as shown in FIG. 4A, must be great enough that the paddles 20 fully intercept the full extent of the input optical beams 60a and 62a projected across the top of the switch structure. Secondly, the deflection angle θ of the paddles 20 is most preferably about 45° with respect to the direction of propagation of the beams 60a and 62a across the switch structure. This ensures that the intermediate beam 60b reflected from the mirror section 25 of the switch portion S22a will travel in essentially a perpendicular direction with respect to the direction of the input beam 60a across the switch structure, to be intercepted by the mirror section 25 of the switch portion S22b. The relationship between these two requirements is preferably expressed by:

Δz=Δzbi+Lmirror×sinθ,   (7)

[0074] where Lmirror is the length of the mirror section (FIGS. 4A-B), and Δzbi is the vertical displacement of the bi-material section of the switch paddle 20 where the upper layer 28 and the lower layer 24 overlap, as given by: 4$\begin{array}{cc}\Delta z_{\mathrm{bi}}=\frac{180}{\theta }\times \frac{L_{\mathrm{bi}}}{\pi }\times \frac{\sin \theta }{\tan \left(90-\theta /2\right)},& \left(8\right)\end{array}$

[0075] where Lbi is the length of the bi-material section, as shown in FIGS. 4A-B.

[0076] Generally, the vertical movement of the paddle 20 is dependent on the change in temperature ΔT of the bimorph structure, and the switch responsivity Rp in units of microns/°C. The responsivity Rp determines the sensitivity, and for the switch structure shown in FIGS. 4A-B, is expressed as: 5$\begin{array}{cc}{R}_p=\frac{\Delta z}{\Delta T}=\frac{3\times L_{\mathrm{bi}}^2\times \left({\alpha }_{\mathrm{bi}}-{\alpha }_{\mathrm{Si}}\right)\times \left(1+x\right)}{t_{\mathrm{bi}}\times K},& \left(9\right)\end{array}$

[0077] where tbi is the thickness of the metal layer 24, αb1 and αs1 are the thermal expansion coefficients of the metal layer 24 and the polysilicon layer 28, respectively, and x is the ratio of the thicknesses of the polysilicon layer 28 and the metal layer 24, i.e. 6$\begin{array}{cc}x=\frac{t_{\mathrm{Si}}}{t_{\mathrm{bi}}}.& \left(10\right)\end{array}$

[0078] The term K is a correction factor necessitated by the difference in Young's moduli of the two layers 24 and 28, i.e.: 7$\begin{array}{cc}K=4+6x+4x^2+n{x}^3+\frac{1}{nx}.& \left(11\right)\end{array}$

[0079] The term n in equation (11) is the ratio of the Young's moduli, i.e.: 8$\begin{array}{cc}n=\frac{E_{\mathrm{Si}}}{E_{\mathrm{bi}}},& \left(12\right)\end{array}$

[0080] where Esi and Ebi are the Young's moduli of the polysilicon layer 28 and the metal layer 24, respectively.

[0081] The responsivity of the switch paddles is modeled in FIG. 11 as a function of the thickness of the polysilicon thin film layer 28 for several thicknesses of the metal film layer 24. These results show that, in general, the thinner the metal film 24, the greater the responsivity. These results also indicate that for each given thickness of the metal film layer 24, there is an optimum thickness of the polysilicon film layer 28 which maximizes the responsivity. Films that are thicker or thinner than the optimum tend to lower the responsivity of the bimorph structure, and increase the temperature and hence the power required to fully activate the switch.

[0082] Once the maximum tip displacement zmax required to fully intercept the optical radiation is known, the maximum change in temperature ΔTmax of the bimorph structure required to achieve this displacement can be found according to: 9$\begin{array}{cc}\Delta T_{\max }=\frac{z_{\max }}{R_p}.& \left(13\right)\end{array}$

[0083] The maximum switch operating temperature is critically dependent on the dimensions of the paddle structures. The dependence of ΔTmax on the thicknesses of the polysilicon and metal film layers 28 and 24 is shown in FIG. 12, and has a complex functional dependence on these and the other material properties and film dimensions. For reliable long term switch operation, it is desired that ΔTmax be kept below approximately 100° C. Higher operating temperatures can lead to creep and stress-related fatigue in the thin film structures. The modeled results shown in FIG. 12 indicate that it is possible to keep ΔTmax in the 50 to 70° C. range over a wide range of switch paddle dimensions.

[0084] After the switch is activated by application of heat to the switch paddle 20, there is a time lag before the switch paddle 20 reaches its final optical beam intercept position, as shown in FIG. 4B. This is due to the finite time required to heat the structure to its final operating temperature, which is a function of the thermal time constant τth (Eq. 1) and the mechanical responsivity of the paddle structure. The time dependence of the temperature rise when the switch is activated, and the temperature fall after the switch has been turned off, T(t), is determined according to:

T (t)=Tmax×(1−e−t/τth).   (14)

[0085] The maximum operating switch temperature Tmax should be low enough such that the switch does not experience long term performance degradation due to, for example, heat induced stress gradients in the paddle structure, or fatigue and ion mobility issues in the thin film structures. These issues may become problems for standard CMOS IC devices at temperatures much in excess of 100° C. However, this is well above the expected operating temperature of the preferred embodiment of the present invention. (See Table II.)

[0086] An additional consideration in determining the time required to activate the switch is the fact that the switch paddles 20 are free standing mechanical structures, fixed at one end, but free to move at the other. These structures vibrate at given resonant frequencies when excited by the thermal activation pulse or other external mechanical forces. A further time constant, referred to herein as the settling time ts, can be defined which indicates the time required for the switching transients and induced vibrations to be dampened sufficiently for the switch to operate without excessive modulation of the transmitted optical radiation.

[0087] The switch can be modeled as a vibrating spring by the following expression: 10$\begin{array}{cc}{m}_{\mathrm{switch}}\times \frac{d^2}{d{t}^2}z\left(t\right)+b\times \frac{d}{dt}z\left(t\right)+k_{\mathrm{switch}}\times z\left(t\right)=F_{\mathrm{thermal}}\left(t\right)& \left(15\right)\end{array}$

[0088] where mswitch is the total mass of each of the switch paddles 20, b is a damping constant, which for a gas-filled switch is mainly due to the viscous drag of the air in the enclosure, kswitch is the spring constant of the paddle 20, and Fthermal(t) is the thermal force applied to the paddle 20 as the paddle 20 is heated. Without the damping force due to the gas within the switch enclosure (i.e. b≅0), the paddle 20 will oscillate indefinitely, unless the paddle resonant time period (τswitch=(mswitch/kswitch)1/2) is significantly shorter than the initial response time (τth) to the thermal heating force. Under these conditions, there is generally no phase lag between the applied force and the switch mechanical response, and thus little if any mechanical overshoot and vibration of the switch paddles 20. These conditions are shown graphically in FIGS. 13A and 13B. With the addition of air damping, this condition is relaxed, as long as there is sufficient gas pressure and the resulting viscous drag to dampen any induced mechanical motion in the paddle 20 within the required switching time. At atmospheric gas pressures, air damping occurs well within the one millisecond switch time required of the present switch. This condition is modeled in FIG. 13C for expected switch conditions.

[0089] Similar arguments can be made for externally induced mechanical vibration of the switch paddle 20. The fundamental switch paddle resonant frequency fres can be found from the following expression: 11$\begin{array}{cc}{f}_{\mathrm{res}}=\frac{1}{2\times \pi }\times {\left(\frac{m_{\mathrm{switch}}}{k_{\mathrm{switch}}}\right)}^{1/2}.& \left(16\right)\end{array}$

[0090] Typically, fres for the switch paddle 20 will be in the range of 5×103 to 5×104 Hz, depending on the switch dimensions and materials. For mechanical noise with a frequency well below the resonant frequency of the switch paddle 20, the paddle 20 will move in phase with the external applied force and will not vibrate when the force is removed. At higher frequencies, air viscous drag will tend to dampen any vibrations with periods well under one millisecond, as shown by the modeled results in FIG. 13C. Since the mass of the switch paddles 20 is extremely small (on the order of 10−11 kg), the resultant inertial and gravitational forces on the paddle structure are correspondingly minute.

[0091] As mentioned previously, an important design consideration is thermal power management, both from a power usage requirement, and switch temperature stability and heat dissipation requirements. The heating power Pmax required to raise the switch to its operating temperature Tmax is given by the following expression: 12$\begin{array}{cc}{P}_{\max }=\frac{T_{\max }}{R_{\mathrm{th}}},& \left(17\right)\end{array}$

[0092] where Rth is the thermal impedance of the structure of the switch paddle 20. Knowing the switch supply voltage Vs, it is possible to estimate the electrical impedance REpoly of the doped polysilicon resistive heating layer 32 required to achieve the switch operating temperature. Thus, REpoly may be determined according to: 13$\begin{array}{cc}{\mathrm{RE}}_{\mathrm{poly}}=\frac{V_S}{P_{\max }}& \left(18\right)\end{array}$

[0093] The required electrical conductivity σpoly of the polysilicon resistive heater 32 can be found according to: 14$\begin{array}{cc}{\sigma }_{\mathrm{poly}}=\frac{2\times L_{\mathrm{poly}}}{{\mathrm{RE}}_{\mathrm{poly}}\times t_{\mathrm{poly}}\times w_{\mathrm{poly}}}& \left(19\right)\end{array}$

[0094] Patterning the polysilicon thin film layer 28 as shown in FIGS. 5A-B, and doping the patterned layer 28 using ion implantation techniques, allows the resistivity of the polysilicon layer 28 to be tailored to the desired value.

[0095] Results of several calculations of the switch performance, as a function of the specific switch structures and the total length of the switch paddle 20, are shown in Table II. These calculations are designed to show the sensitivity of the switch parameters (i.e. switch response time, temperature change, and power requirements) to changes in the switch dimensions. Designs 1 and 2 show the effect of increasing the thickness of the metal and polysilicon layers 24 and 28 on the switch parameters. Thicker structures have a higher resonant frequency, and are potentially more robust, but require a higher thermal power and resultant higher switch activation temperature. This effect is also shown graphically in FIG. 12. Changes in film thicknesses typically have little effect on the thermal response time of the switch. Generally, the response time is highly dependent on the length of the thermal isolation gap Lgap, as shown by a comparison between Designs 1 and 3. This effect is also shown graphically in FIG. 10. Smaller values of Lgap tend to lead to smaller thermal time constants and faster switching. However, the power required to activate the switch is generally increased correspondingly. Thus, there appears to be a clear trade-off between speed of response and switch power consumption.

[0096] The switch Design 4 shown in Table II indicates the effect on the switch parameters of changes in the length of the bimaterial section Lbi of the switch paddle 20. In a comparison between Designs 3 and 4, increasing the length of the bimaterial section reduces switch activation temperature and thus the thermal power, but increases the response time and the size of the individual switch elements. The switching time and power requirements are also significantly dependent on the other input parameters listed in Table II, along with the thermal and mechanical properties (shown in Table I) of the of the materials of which the switch is composed.

[0097] The manufacturing techniques used to fabricate the optical switching device 10 includes MEMS surface micro-machining steps as well as standard silicon CMOS IC fabrication steps. Both the MEMS and CMOS fabrication processes have been designed to be compatible with, and transferable to, standard silicon foundries. The switch structures described herein may be manufactured in accordance with the fabrication processes described in copending U.S. patent application Ser. No. 09/628,536, filed on Jul. 31, 2000 by the present inventor, or with slight variations to those processes which are within the capabilities of one of ordinary skill in the art. The entire contents of application Ser. No. 09/628,536 are hereby incorporated by reference.

[0098] It is contemplated, and will be apparent to those skilled in the art from the preceding description and the accompanying drawings that modifications and/or changes may be made in the embodiments of the invention. Accordingly, it is expressly intended that the foregoing description and the accompanying drawings are illustrative of preferred embodiments only, not limiting thereto, and that the true spirit and scope of the present invention be determined by reference to the appended claims.

Claims

1. An optical switch for directing at least one optical beam from a first propagation plane to a second propagation plane that is parallel with the first propagation plane, comprising: a first switch portion having a substantially-planar first reflective surface portion, the first reflective surface portion operable to move in a first pitch plane between at least a first deflection angle and a second deflection angle, when at the first deflection angle, the first reflective surface portion for intercepting the optical beam propagating along a first propagation path in the first propagation plane and for reflecting the optical beam at a first reflection angle into an intermediate propagation path, and when at the second deflection angle, the first reflective surface portion for not intercepting the optical beam; and a second switch portion having a substantially-planar second reflective surface portion, the second reflective surface portion operable to move in a second pitch plane between at least the first deflection angle and the second deflection angle, the second pitch plane being substantially perpendicular to the first pitch plane, when at the first deflection angle, the second reflective surface portion for intercepting the optical beam propagating along the intermediate propagation path and for reflecting the optical beam at a second reflection angle into a second propagation path in the second propagation plane, and when at the second deflection angle, the second reflective surface portion for not intercepting the optical beam.

2. The optical switch of claim 1 wherein the first reflective surface portion reflects the optical beam into the intermediate propagation path such that the intermediate propagation path is substantially perpendicular to the first and second propagation planes.

3. The optical switch of claim 1 wherein the second reflective surface portion reflects the optical beam into the second propagation path such that the second propagation path is substantially perpendicular to the first propagation path.

4. The optical switch of claim 1 wherein the first and second deflection angles are approximately 45 degrees, and the first and second reflection angles are approximately 90 degrees.

5. The optical switch of claim 1 wherein the first and second reflective surface portions are substantially parallel and are disposed on either side of the second propagation plane when at the second deflection angle.

6. The optical switch of claim 1 wherein the first reflective surface portion intersects the first propagation plane at approximately 45 degrees when at the first deflection angle, and the second reflective surface portion intersects the second propagation plane at approximately 45 degrees when at the first deflection angle.

7. The optical switch of claim 1 wherein: the first switch portion further comprises a first paddle disposed adjacent the first propagation path, and having a fixed portion and a free end, with the first reflective surface portion adjacent the free end, the first paddle operable to bend upon a change in temperature, thereby moving the first reflective surface portion between the first and second deflection angles; and the second switch portion further comprises a second paddle oriented substantially perpendicular to the first paddle, the second paddle disposed adjacent the second propagation path, and having a fixed portion and a free end, with the second reflective surface portion adjacent the free end, the second paddle operable to bend upon a change in temperature, thereby moving the second reflective surface portion between the first and second deflection angles.

8. The optical switch of claim 7 wherein the first and second paddles each include: an upper layer having a first coefficient of thermal expansion; and a lower layer mechanically coupled to the upper layer, and having a second coefficient of thermal expansion that is different from the first coefficient of thermal expansion, where the change in temperature causes the upper and lower layers to change in size at different rates due to the differences in coefficients of thermal expansion, thereby creating stress in the paddle and causing the paddle to bend.

9. The optical switch of claim 8 wherein the upper layer includes a resistive heater portion that generates heat upon passage of an electric current there through.

10. The optical switch of claim 8 further comprising an electrically-insulating layer disposed between the upper and lower layers.

11. The optical switch of claim 10 wherein the electrically-insulative layer is a material selected from the group consisting of SiO2, SiC, Si3N4, SixOyNz, and HxSiyCz.

12. The optical switch of claim 10 further comprising a bonding layer disposed between the electrically-insulative layer and the lower layer.

13. The optical switch of claim 12 wherein the bonding layer is a material selected from the group consisting of Cr, Ti, TiW, TiN, nichrome, and alloys thereof.

14. The optical switch of claim 8 wherein the lower layer has a coefficient of thermal expansion greater than that of the upper layer.

15. The optical switch of claim 8 wherein the lower layer is an optically-reflective metal selected from the group consisting of Au, Al, Pb, and Zn.

16. The optical switch of claim 8 wherein the upper layer is doped polysilicon having a coefficient of thermal expansion less than that of the lower layer.

17. The optical switch of claim 8 wherein the upper layer is disposed only between the fixed portion and a point located between the fixed portion and the free end, whereby a portion of the lower layer extends beyond the upper layer at the free end of the first and second paddles.

18. The optical switch of claim 17 wherein the portion of the lower layer extending beyond the upper layer of the first and second paddle comprises the first and second reflective surface portions, respectively, which remain substantially flat when at the first deflection angle, such that the first and second reflective surface portions reflect the optical beam with minimal beam divergence.

19. The optical switch of claim 7 further comprising a thermal isolation gap portion adjacent the fixed portion of the first and second paddle, where in the thermal isolation gap portion, the lower layer does not overlap the upper layer.

20. The optical switch of claim 9 further comprising: a voltage supply for selectively providing electrical power to the resistive heating portions of the upper layers of the paddles; a beam splitter disposed in the second propagation path for redirecting at least a portion of the optical beam into a third propagation path; an optical detector disposed in the third propagation path for providing at least one beam position signal indicative of a position of the optical beam on the optical detector; the voltage supply operable to adjust a characteristic of the electrical power provided to the resistive heating portions based on the beam position signal, such that the resistive heating portions generate more or less heat as determined by the beam position signal, thereby causing the first and second paddles to provide larger or smaller first and second deflection angles to thereby center the optical beam on the detector.

21. An optical switching device for selectively transferring at least one optical beam between at least one first optical channel and M number of second optical channels, comprising: a first optical structure for propagating the at least one optical beam along a first propagation path in a first propagation plane; and a 1×M-dimensional array of optical switches, comprising: M number of first switch portions adjacent to and aligned in parallel with the first propagation path, the first switch portions having substantially-planar first reflective surface portions that are operable to move in a first pitch plane between at least a first deflection angle and a second deflection angle, when at the first deflection angle, one of the first reflective surface portions for intercepting the optical beam propagating along the first propagation path in the first propagation plane and for reflecting the optical beam at a first reflection angle into an intermediate propagation path, and when at the second deflection angle, the first reflective surface portions not intercepting the optical beam; and M number of second switch portions adjacent corresponding first switch portions, the second switch portions having substantially-planar second reflective surface portions, the second reflective surface portions operable to move in second pitch planes between at least the first deflection angle and the second deflection angle, the second pitch planes being substantially perpendicular to the first pitch plane, when at the first deflection angle, one of the second reflective surface portions operable to intercept the optical beam propagating along the intermediate propagation path and reflect the optical beam at a second reflection angle into one of M number of substantially parallel second propagation paths in a second propagation plane, the second propagation paths being substantially perpendicular to the first propagation path, and the second propagation plane being substantially parallel with the first propagation plane, and when at the second deflection angle, the second reflective surface portions not intercepting the optical beam; and M number of second optical structures disposed in the second propagation plane, each of the second optical structures optically aligned with a corresponding one of the M number of second switch portions in a corresponding one of the M number of second propagation paths, each of the second optical structures operable to receive the optical beam from the corresponding one of the second reflective surface portions and direct the optical beam to a corresponding one of the M number of second optical channels.

22. The optical switching device of claim 21 wherein the first and second optical structures further comprise optical fibers.

23. The optical switching device of claim 21 wherein the first optical structure further comprises a modulated laser light source.

24. The optical switching device of claim 21 wherein the first optical structure further comprises a modulated light-emitting diode light source.

25. The optical switching device of claim 21 further comprising: N number of the first optical structures for propagating N number of optical beams along N number of substantially parallel first propagation paths in the first propagation plane; and N number of the 1×M-dimensional arrays of optical switches, each array aligned with a corresponding one of the N number of first propagation paths.

26. The optical switching device of claim 25 wherein the first optical structure, the array of optical switches, and the second optical structures are reciprocal devices, such that the optical beam may travel from the first optical structure through one of the optical switches and into one of the second optical structures, or from one of the second optical structures through one of the optical switches and into the first optical structure.