WO2001020382A1

WO2001020382A1 - Integrated optical crosspoint switch array based on hybrid interference and digital mode operation

Info

Publication number: WO2001020382A1
Application number: PCT/US2000/025424
Authority: WO
Inventors: Chi Wu; Siamak Forouhar
Original assignee: California Institute Of Technology
Priority date: 1999-09-16
Filing date: 2000-09-15
Publication date: 2001-03-22
Also published as: AU7383000A

Abstract

An integrated optical cross point switch and switch element that is based on interference and digital modes. In the through state, multimode interference is formed to maintain the optical length. In the switching state, single mode evolution is used. The device uses a silicon (120) on insulator optical waveguide (84) formed with its lower and upper cladding layers (82, 86) being silicon dioxided air and the single optical mode on a ridged waveguide. Lateral current flow is used to keep the device form having current mode that spread.

Description

INTEGRATED OPTICAL CROSSPOINT SWITCH ARRAY BASED ON HYBRID INTERFERENCE AND DIGITAL MODE OPERATION

This application claims the benefit of U.S. Provisional Application No. 60/154,210, filed on September 16, 1999.

STATEMENT AS TO FEDERALLY-SPONSORED RESEARCH The invention described herein was made in the performance of work under a NASA contract, and is subject to the provisions of Public Law 96-517 (U.S.C. 202) in which the contractor has elected to retain title.

BACKGROUND

Optical communication has been used for many purposes. Optical processing can be used to manipulate data, graphics and images. Optical processing has been used to process data from photonic sensors, optical communication elements, microphotonics instruments, fiber optic data buses and photon propulsion systems. Optical switches have also been used in telecommunication, datacommunications, spacecraft, advanced radar and RF photonics . Fiber optics are used to carry information traffic in the form of light. Optical switches can be used in these elements. Optical switch arrays have also been used for telephone protection switching, e.g, in a cross connect network, restoration wavelength routing, or in an ATM optical switch. There are various technologies for optical switch arrays: opto-mechanical liquid crystal, optical amplifier gated-array, integrated optics based switch which are on silica, LiNb03, polymer, GaAs, InP, and silicon materials. Mechanical optical switch array have low optical loss and cross-talk and is considered the most matured technology. However, its long-term reliability (due to moving parts) , low speed and large size may limit these switches in many applications. A semiconductor optical amplifier array with optical gain has been developed for a gated array switch. However, its noise, polarization sensitivity, power consumption and limited optical bandwidth may limit its practical applications. A silica/silicon based optical switch based on the interference concept is known. Due to material and fabrication variations across the wafer, the performance of each switch element may not be uniform. Special trimming techniques may be used for each switch element that forms the array, after the fabrication. In addition, the power consumption may be a concern, since such a switch array uses the thermal-optic effect and needs NxN electrodes to be turned "on" at all the time. A demonstrated 16x16 switch array, which is the largest waveguide switch array ever fabricated, needs a 6-inch silicon wafer. The stress issue may form a limiting factor for wafer size beyond 6" due to the nature of its fabrication process. Liquid crystal based optical switches have the advantage of low loss. But its polarization and temperature sensitive performance are still concerns.

LiNb0₃ based optical switch arrays have been demonstrated based on both two-mode interference concept and one-mode evolution concept. The former has a sinusoidal-like switching characteristic. This makes the switch sensitive to the variations of polarization, wavelength, operating voltage and temperature, as well as fabrication imperfections. It is difficult to make a high performance large array based on this concept. A digital optical switch, based on one-mode evolution, and with a step-like switching characteristic, is in general a superior technology and may offer many advantages to overcome the problems of the interference type switch. However, the LiNb0₃ based digital optical switch has also drawbacks of long device length and high operating voltage (20-60V) . It is difficult to make a high performance and low-cost large array. In addition, due to the limit wafer size of LiNb03 material (less than 3 inches), it is difficult to make a fully non-blocking array with size larger than 16x16.

InP and GaAs based optical switches have been demonstrated based on either interference type or digital type concept. Symmetric Y-branches have been used for a digital optical switch. The principle has originally been demonstrated on LiNb0₃, then on InP. For the InP digital optical switch, two electrodes have been placed on each of the two arms of the Y. If no current is turned on, the switch behaves like a -3dB power splitter. When electrical current is injected into one of the waveguide arms, say waveguide A, the index of refraction of the waveguide A will be decreased. This means that no more can be guided in waveguide A. The light will be switched to another waveguide, say waveguide B.

Similarly, if current is injected into waveguide B, the light will be mode evolution whenever the waveguide A or B is turned on. The light transmission versus applied current shows a step function characteristic. Therefore, this type of switch is often called a digital optical switch. An optical loss may occur when current is turned on. This may be due to (1) tilted optical reflection wall formed by the injection current spreading effect, (2) radiation and scattering loss due to the imperfection of the Y-tip.

A silicon based optical switch has been reported, which employs undoped silicon as waveguide layer and heavily doped n+ and p+ silicon as the upper and lower cladding layers to confine the light. Due to the high free carrier absorption loss of the n+ and p+ silicon material, the waveguide loss is high.

Micro-electro-mechanic optical switch, which is based on silicon MEMS technology, has shown low optical loss and low cross-talk performance. However, its long- term reliability is still unknown. In addition, its free space alignment and packaging technology need to be solved.

Another switch, made by the inventor of the present application, is described in U.S. Patent No. 5,581,643. The present application operates differently than that previous structure. SUMMARY The present application teaches a new kind of optical switch array that is based on a hybrid between interference and a digital mode operation. The present application teaches a Y type optical switch which has certain structural differences from prior systems, including a increased size portion; a lateral current injection scheme which allows formation of a lateral index of refraction wall, and different modes of operation.

BRIEF DESCRIPTION OF THE DRAWINGS These and other aspects will now be described in detail with reference to the accompanying drawings, wherein:

Figure 1 shows an embodiment of a cross point switch having a number of elements;

Figure 2A and 2B show the principle of operation of the present y switch; Figure 3 shows a cross section across the line A-A in Figure 2A, showing the lateral current wall;

Figures 4A and 4B shows the variable optical attenuator, integrated into the switch. DETAILED DESCRIPTION An optical switch array is an array of switch elements; each of which switch element operates to switch optical signals from an input to an output. An optical cross-point switch has many such inputs and outputs. The present application describes an optical switch and array of switches, with specific differences and advantages over the prior art.

The basic structure can be formed as a silicon-on- insulator based photonic element. This may have a number of advantages including improved fabrication capability. The present system defines a cross point switch in which only N switches need to be "turned on" out of an NxN array. This compares with a cross bar switch where all of the NxN switches may need to be turned on.

The basic device structure is shown in Figure 1. The structure includes a silicon-on-insulator based optical waveguide switch array with a waveguide layer that is formed of silicon. Upper and lower cladding areas are formed of silicon dioxide and air. The light is confined by a ridge waveguide structure in the substrate.

Using the SOI based structure as disclosed herein, very low optical propagation loss can be obtained. A silicon waveguiding layer can be used which has a low doping concentration, in order to provide even lower optical absorption loss.

The present application describes a Y junction based switch that uses different techniques than the prior art. Y junction based switches, based on interference of two modes have been reported. Digital switches based on adiabatic, one-mode evolution at switching states have also been reported. A total internal reflection type switch is also known; however, this requires a relatively high index change at the reflection point in order to meet the total internal reflection condition.

The present application uses multi-mode interference in the through state, and single mode evolution at the switching state.

This operation is inherent from the structures that are used. The light launced into the wider "straight- through" waveguide 90 will propagate through the multi- mode section. The light in the bypass waveguide, in contrast, is more constrained, since the waveguide is thinner. This light can only exist in a single mode. A step function switching characteristic can exist to the switching state. The multimode operation can significantly reduce the optical loss and cross talk in the through state.

One way in which this system is made more efficient is by adding an expanded portion 4 to the waveguide in the vicinity of the branch 5 between the straight through waveguide 90, and the branch waveguide 92. Light in a waveguide tries to stay in the central portion of the waveguide. By widening the waveguide in the vicinity of the branch 5, the preferred location of the light is moved. The waveguide is widened in the direction that is away from the direction of the branch. This, therefore, moves the light away from the branch.

In the straight through mode, this means that less light will enter the branch, and/or interfere with the branch portion. Therefore, the cross-talk to the branch waveguide is reduced.

A curved structure 6 may be used as the wall of the throughput waveguide farthest from the branch, in order to minimize throughput optical loss and cross talk. In U.S. Patent No. 5,581,643, electrodes for the p and n junction were placed on top of the ridge and on the bottom of the substrate. This caused electric current to flow from the top of the ridged waveguide to the bottom substrate. When the current flows, the index of refraction of the underlying semiconductor material is changed. This forms a refraction index discontinuity; effectively a reflection wall. By forming the reflection wall in the straight-through waveguide, light can be forced into the branch waveguide 92.

In the '643 patent, the current spreading effect caused the current to spread into both the throughput waveguide and into the branching waveguide. Current into the branching waveguide can affect the optical throughput and can result in certain losses during that switching state .

The present application, in contrast, uses a laterally arranged PIN structure. The current flows laterally, that is, in the same direction as the surface of the substrate and is directed away from the branching waveguide. This thereby suppresses current spread into the branch waveguide, and hence can minimize the optical loss in the switching state.

As shown, the structure in Figure 1 defines an optical system in which the input optical signals 100,

102 can be input at one branch and output through any of another branches 110, 112. Other branches, such as 104, 106 can also be input. The structure as defined includes a silicon substrate 120 which is formed of a n type silicon layer 80, "intrinsic" Sι0₂ layer 82, covered by an p type silicon layer 84. An N type silicon ridge 88 forms the waveguide thereby forming a PIN structure. Each input waveguide 80 has a y branch coupler associated therewith m the vicinity of its intersection with an output waveguide. The directional coupler has a PIN structure formed by electrodes 150, 152 is an electrical bias selectively applied to the waveguide. The y-branch may be tuned by the electro-optic effect (reverse bias on p-n junction) or by free carrier injection (forward bias on p-n junction) to selectively couple an optical signal from the straight through waveguide 90 to a branching waveguide lying in close proximity as described herein. A reflecting device 92 such as an etched mirror is positioned near the pass-through waveguιde9 so as to redirect the optical signal into each Y-branch guide 94. Controllable optical attenuators such as 18 can be formed on the respective output lines. These attenuators can suppress some of the optical energy m the line. These attenuators can be controlled to equalize the levels of the optical signals. For example, an optical signal that has passed through two switches may be more attenuated than a signal that has only passed through one switch. The attenuators can produce an attenuation level to equalize the different outputs.

One of the advantages of optical transmission system involving waveguides is that signals can cross waveguides without being affected.

Figure 1 shows the substrate 80, lower cladding layers 82, waveguide layer 84, upper cladding layer 86, as well as the waveguide matrix layer 88 are also shown. The hatched regions 90 represent electrodes on top of the Y-branch coupler guides. The angled portion 92 at the end of each guide is a mirror having an angle of approximately 45° to reflect the optical signal into the Y-branch guide. The mirrors are etched deeply to the lower cladding layer. A plurality of these y switches 140 and deeply etched mirrors are formed on the silicon substrate .

A cross section of one of the switching elements 140 is shown in Figure 2A, 2B. A cross-sectional view along the line A-A, which is effectively a cross-section through the substrate, is shown in Figure 3. The only contacts are formed in locations causing lateral spread, as described. The lateral spread will push the optical mode out of the branch waveguide. This could increase the switching loss. The integral attenuators are shown m further detail in Figures 4A and 4B. Each output line like 400 includes an optical attenuator 402 in series with the output. Each optical attenuator is individually controllable, so that the lines can be equalized.

Each switch includes two branches -- the straight- through waveguide 200, and the branching waveguide 205. With the voltage equal to zero, the switch is in a straight through state. The light travels in the center of the waveguide, and therefore mostly misses the branch waveguide 92. Therefore, with no applied voltage, the optical energy continues straight through, down the center of the waveguide. When a switching voltage is applied between the electrodes, it changes the index of refraction of the materials, raising a vertical reflection wall. This forces the optical beam into the bypass waveguide, the "switching state."

The current is applied between the electrodes 150 and 152 to make the maximum effect of the applied current. As shown, one electrode is located physically on the junction itself and forms a geometric right angle, with edges of the electrode being curved. The other electrode, located on the substrate, is also geometrically m the shape of an right angled portion, thereby forming a similar shape to the shape of the first electrode. The lateral current spread is minimized.

Although only a few embodiments have been disclosed m detail above, other modifications are possible. For example, while the present application describes using silicon, other materials may be possible. All such modifications are intended to be encompassed with the following claims m which:

Claims

What is claimed is:

1. An optical switch, comprising: a semiconductor substrate, formed of a first doping type of silicon, an insulating layer over said first doping type of silicon and a second doping type of silicon over said insulating layer, said second doping type of silicon forming a semiconductor surface; a first waveguide, formed of an optical layer, said light input portion including an ohmic contact which is coupled thereto, said ohmic contact being in a specified geometric shape, said first waveguide being formed of said first doping type of silicon; a second ohmic contact, formed of a similar geometric shape to said first ohmic contact, and formed on said semiconductor surface, adjacent said first ohmic contact in a location where current can flow therebetween; and a second waveguide, coupled optically in parallel with said light input structure; and said first and second ohmic contacts forming a switching structure, coupled between said first and second waveguides, wherein a specified current between said ohmic contacts causes a switching state which causes light to pass through said one of said waveguides, and lack of current therebetween allows the light to pass through the other of said waveguides.

2. A switch as in claim 1, wherein said second doping type is p type, and said first doping type is n type.

3. A switch as in claim 1, wherein said specified shapes cause current flow in a lateral direction.

4. A switch as in claim 1, wherein said geometric shapes cause current flow in a direction away from the second waveguide.

5. A switch as in claim 1, wherein said first waveguide has a substantially Y portion.

6. The switch as in claim 1, wherein said switching element comprises, a PN junction, which when energized, causes interference at an area of the PN junction, thereby causing the optical energy to flow through an area which does not have said interference.

7. A switch as in claim 1, wherein said geometric shape of both said first and said second ohmic contacts is substantially in the shape of a right angled corner.

8. A switch as in claim 1, further comprising a deep etched mirror, formed adjacent said second waveguide and between a plurality of said devices.

9. A device as in claim 1, wherein said switching structure is arranged to form multimode interference at a through state through the first waveguide, and to form single mode evolution at a switching state.

10. A method of optical switching, comprising: forming an optical switch which has a first state based on multimode interference and a switching state based on single mode evolution; and applying a control signal which switches between said multimode interference and said single mode evolution to control the switching state.

11. A method as in claim 10, wherein said straight through mode is formed on a silicon waveguide which is founded at one side at Si0₂, and at the other side by air.

12. A method as in claim 11, wherein said silicon waveguide has a specified doping type, and further comprising providing a semiconductor substrate adjacent said silicon layer with an opposite doping type to said specified doping type.

13. A method as in claim 12, further comprising providing an ohmic contact on said silicon waveguide and another ohmic contact on said substrate in a location that, when energized, causes lateral current flow.

14. A method as in claim 13, wherein said ohmic contact on said substrate is substantially in a right angled shape.

15. A method as in claim 10, further comprising switching by forming lateral current flow in substantially the same direction as the substrate.

16. A method as in claim 15, wherein a switching branch is optically in parallel with the other branch, and the lateral current flow is in a direction away from said switching branch.

17. A method as m claim 9, further comprising placing a switching branch wave guide m parallel with a mam wave guide and using said interference to determine which of said switching wave guide or said branch wave guide receives said optical energy.

18. A method as in claim 17, wherein said interference is caused by an electric current.

19. A method as in claim 18, further comprising minimizing lateral current flow into the branch waveguide .

20. A method as in claim 10, wherein said control signal forms a lateral index of refraction wall.

21. A method as in claim 19, further comprising providing a plurality of additional optical switches the form of an optical matrix; and placing some of said optical switches into a switch state, and other optical switches into a non-switch state to thereby allow input light to be coupled to any location m the light.

21. A method as in claim 20, further comprising forming the optical mode in a ridge waveguide structure.

22. An optical switch, comprising: a semiconductor substrate, formed of a first doping type of silicon, an insulating layer over said first doping type of silicon and a second doping type of silicon over said insulating layer, said second doping type of silicon forming a semiconductor surface; a first waveguide and a second waveguide, which form a Y with one another, said first waveguide being a straight through waveguide, and said second waveguide being a branch waveguide, said first waveguide including an expanded part, having an expanded size, in an area of said Y; and a switching structure selectively forcing said optical energy into one of said first or second waveguide based on an applied signal.

23. A switch as in claim 22, wherein said expanded part is an expanded curved part.

24. A switch as in claim 23, wherein said switching structure includes an electrode on said first waveguide, and another electrode spaced from said first waveguide in a direction away from said second waveguide.

25. A switch as in claim 23, wherein said electrode and said another electrode are of similar geometric shapes .

26. A switch as in claim 23, wherein said electrodes are placed to form lateral current injection.

27. An optical switch, comprising: a semiconductor substrate; a first waveguide and a second waveguide, which form a Y with one another, said first waveguide being a straight through waveguide, and said second waveguide being a branch waveguide, said first waveguide including an expanded part, having an expanded size, in an area of said Y; and a switching structure selectively forcing said optical energy into one of said first or second waveguide based on multi-mode interference in the through state through the first waveguide, and single mode evolution at the switching state through the second waveguide.

28. An optical switch as in claim 27, wherein said first waveguide is enlarged and curved in an area of said

Y.

29. An optical switch as in claim 27, wherein a wall of said first waveguide which is distant from said second waveguide, is enlarged and curved in an area of said Y on said wall distant from said first waveguide.

30. A switch as in claim 27, wherein said switching structure causes lateral current injection.

31. An optical switch array, comprising: a semiconductor substrate; a plurality of lines, running on said substrate, a first waveguide and a second waveguide, which form a Y with one another, said first waveguide being a straight through waveguide, and said second waveguide being a branch waveguide, each said branch waveguide coupled to an output waveguide; and a plurality of controllable optical attenuators, coupled on said output waveguides.

32. An array as in claim 31, wherein each of said output waveguides has a separate attenuator.

33. An array as in claim 31, further comprising a switching structure selectively forcing said optical energy into one of said first or second waveguide based on multi-mode interference in the through state through the first waveguide, and single mode evolution at the switching state through the second waveguide.

34. An array as in claim 31, wherein each said first waveguide including an expanded part, having an expanded size, in an area of said Y