MEMS Optical Devices — MEMS Optical Switch

Optical switches are widely employed in optical fiber communication systems. The approaches for an optical switch are variable, including mechanical optical switch, thermos-optic switch, acousto-optic switch, electro-optic switch, magneto-optic switch, liquid crystal optical switch, MEMS optical switch, et. al. MEMS optical switches are characterized by compact size, low power consumption and good scalability, which enable their wide applications.

Optical switch is a multiport device. The port configurations include 2×2, 1×N, N×N. Optical switch with N×N ports is usually called OXC (optical cross connect). According to the difference in port configurations, different MEMS chips are employed for realization of the devices. We will discuss optical switches with 2×2 and 1×N ports in this paper.

2×2 Optical Switch

2×2 optical switches are employed for protection of fiber ring networks. The structure of AON (Agile Optical Network) based on ROADM (Reconfigurable Optical Add/Drop Multiplexer) technology is shown in Fig.1. The optical network is a three-layer structure including long-haul network, metro-network and access network [1-3].


Fig.1 Structure of the AON based on ROADM technology [3]

The metro network is a dual fiber ring with ROADM nodes. Fig.2 shows the structure of a ROADM node. Two ROADM modules and OAs (Optical Amplifier) are connected by two 2×2 optical switches. The switches provide protection for the fiber ring. When breakdown happens at the fiber link or in the ROADM node, the breakdown point can be bypassed through bar→cross switching of the 2×2 optical switches [4].


Fig.2 Structure of a ROADM node [4]

With the rapid development of Internet applications, the demand for bandwidth increases rapidly, which boosts the construction of ROADM-based metro network. Thus the demand for 2×2 optical switches increases rapidly.

The principle of the 2×2 optical switch is shown in Fig.3. Four waveguides are aligned in four directions and a vertical MEMS mirror is aligned in 45° direction. When the mirror is not inserted into the optical path, optical beams from waveguides 1 and 2 are coupled to waveguides 3 and 4, respectively. The port connection is 1→3 & 2→4, which is the bar state. When the mirror is inserted into the optical path, optical beams from waveguides 1 and 2 are reflected by the mirror and then coupled to waveguides 4 and 3, respectively. The port connection is switched to 1→4 & 2→3, which is the cross state [5].


Fig.3 Principle of the 2×2 MEMS optical switch, left: bar state, right: cross state [5]

The stroke of the MEMS mirror is limited (usually tens of microns). It can’t switch the optical beams from optical fiber collimators (the beam diameter is usually hundreds microns), which are widely used in optical passive components. The MEMS mirror is designed to switch optical beams from optical fibers, which emit optical beams with a diameter of about 10 microns. The optical fiber must be well aligned for coupling of optical signals. Fig.4 shows the insertion loss introduced by lateral offset between two optical fibers. The tolerance of lateral misalignment is <1μm.


Fig.4 IL vs the lateral offset between two optical fibers

What’s more, the optical beams to be switched are not collimated. The longitudinal gap between optical fibers will introduce significant power loss due to divergence of the optical beams, as shown in Fig.5. Thus the gap between fiber need to be limited as <20μm.


Fig.5 IL vs the longitudinal gap between two optical fibers

Based on above considerations, the MEMS mirror is fabricated together with four alignment grooves for optical fibers, as shown in Fig.6. For well alignment of the optical fibers, a spring structure is fabricated in each groove, as we can see in the red circle.


Fig.6 SEM photograph of the chip including MEMS mirror and alignment grooves [1]

Scientists from University of Neuchael designed and fabricated the 2×2 MEMS optical switch. The switch has fast response of <1ms, while the loss is relatively high, especially for cross state. In the cross state, the optical beams are reflected by the MEMS mirror. The vertical mirror is fabricated by ion-beam etching process. The roughness of the mirror surface is not good as a polished surface.

1×N Optical Switch

As we know, the rapid development of Internet applications promotes the construction of ROADM-based AON. The new generation of CDC (Colorless, Directionless and Contentionless) ROADM can be constructed with 1×N WSS + N×M WSS (Wavelength Selective Switch) or 1×N WSS + MCS (Multi-Cast Switch), as shown in Fig.7 [6]. Based on cost consideration, the latter is more favored by the telecom operators and manufacturers. Thus the demand for MCS increases rapidly with the development of ROADM-based AON, especially when ROADM technologies sink to metro network from backbone network.


Fig.7 CDC ROADM based on 1×N WSS + N×M WSS or 1×N WSS + MCS [6]

The structure of an 8×16 MCS is shown in Fig.8, consisting of eight 1×16 power splitters (PS) and sixteen 8×1 optical switches (SW). The power splitter is usually fabricated by PLC (Planar Lightwave Circuit) technology, while the 1×N optical switch is usually based on MEMS technology. The most frequently used optical switches have 1×8 and 1×16 ports.


Fig.8 Structure of an 8×16 MCS (PS: Power Splitter, SW: Switch)

The structure of a MEMS-based 1×N optical switch is shown in Fig.9, which consists of a MEMS torsion mirror, a collimating lens and a multi-fiber pigtail. The MEMS mirror is usually assembled on a TO base, then the collimating lens is joint to the sub-assembly through the TO cap. Finally, the multi-fiber pigtail is actively aligned to the sub-assembly.


Fig.9 Structure of the 1×N MEMS optical switch

The device structure in Fig.9 is very simple. However, it is not easy to fabricate a 1×N MEMS optical switch with large ports N and low loss. The maximum loss is measured at the outer ports, which have the maximum deviation Δmax from the optical axis and thus are most affected by optical aberration. Optical aberration deteriorates with the increment of relative aperture Δmax/f (f is the focal length of the collimating lens) of the optical system. Adding the focal length f will help to reduce the optical aberration. However, a larger f results in a larger collimated beam spot on the MEMS mirror according to Eq. (1).

image020.gif (1)

where ω0 is the radius of the beam spot from the optical fiber and ωc is the radius of beam spot on the mirror.

The size of the collimated beam spot is confined by the diameter Ф of the MEMS mirror. Ф>3ωc is required to cover 99% optical power of the collimated beam. However, the diameter Ф and maximum tilting angle θmax of the MEMS mirror is contradictory due to the limitation of MEMS technology. The specifications of a typical MEMS mirror include mirror diameter Ф=1mm and θmax=&plusmn;4&deg;. A larger mirror diameter Ф will result in a smaller θmax=Δmax/f, which limits the number of ports on the contrary. Thus we know, adding to the focal length f of the collimating lens can’t improve the port number N.

Considering above dilemma, there are three ways to improve the port number. The first is to change the alignment optical fibers in the multi-fiber pigtail, as shown in Fig.10. The left one require a 1-axis MEMS mirror, while the port number is smaller. The right one can obtain a larger port number, while it requires a 2-axis MEMS mirror. A 2-axis MEMS mirror is much expensive than a 1-axis MEMS mirror.


Fig.10 Alignment of optical fibers in the multi-fiber pigtail

The second way to improve the port number is to reduce the fiber diameter. As we know, the cladding diameter of a typical single mode fiber is 125μm. Chemical etching is commonly used to reduce the diameter of the optical fiber. The fiber diameter after etching is usually 60~80μm, which is still not small enough. Thus the port number is usually limited to be N≤16. Moreover, the process controlling is not easy, which adds to the cost of the multi-fiber pigtail.

The third way to improve the port number is select a collimating lens with smaller optical aberration. An aspherical lens or a Grin-lens has better performance than a C-Lens (a planar-convex lens with rod shape similar to Grin-lens).

[1]style="margin-bottom: 8px;" L. Eldada, Advances in ROADM technologies and subsystems, Proc of SPIE, Toronto, Canada, 2005: 597022-1.
[2]style="margin-bottom: 8px;" L. Eldada, 40-channel ultra-low-power compact PLC-based ROADM subsystem, Optical Fiber Communication Conference/National Fiber Optic Engineers Conference, Anaheim, USA, 2006: NThC4.
[3]style="margin-bottom: 8px;" Zhujun Wan, Research on Interleaving Filters and Thermo-Optic Switches Based on Planar Lightwave Circuit, Doctoral Dissertation, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, 2007
[4]style="margin-bottom: 8px;" J. J. O. Pires, Constraints on the design of 2-fiber bi-directional WDM rings with optical multiplexer section protection, Advanced Semiconductor Lasers and Applications/Ultraviolet and Blue Lasers and Their Applications/Ultralong Haul DWDM Transmission and Networking/WDM Components, the LEOS Summer Topical Meetings, San Diego, USA, 2001: MC3.2.
[5]style="margin-bottom: 8px;" Cornel Marxer and Nicolaas F. de Rooij, Micro-Opto-Mechanical 2×2 Switch for Single-Mode Fibers Based on Plasma-Etched Silicon Mirror, Journal of Lightwave Technology, 17(1): 2-6, 1999
[6]style="margin-bottom: 8px;" Manish Sharma, Per Hansen, Bimal Nayar, and Peter Wigley, Next-Generation ROADM Technologies and Architecture, Proc. of SPIE Vol.8283: 828309, 2012


Recruitmentcontact us

Copyright © 2018 All rights reserved.粤ICP备05129863号-1