MEMS -> Elastomer adds muscle to optical MEMS

However, because MEMS evolved from integrated-circuit silicon processing technology, its developers adopted the materials traditionally used in IC fabrication, such as silicon oxide, silicon nitride, aluminum and even nickel. Unfortunately, all these are very rigid, not very compliant materials.

The limited range of mechanical properties severely narrows the design space for silicon-based MEMS devices. In addition, the extreme rigidity of silicon MEMS requires higher voltage to provide the force for mechanical deflection.

Spurred by the limitations o f silicon-based MEMS in optical networking applications,Solus Micro Technologies has developed a new class of MEMS devices that uses highly compliant polymeric materials as a principal design element. Compliant MEMS (CMEMS) technology employs a soft, rubber-like material called an elastomer (from the words elastic and polymer), which can be stretched as much as 300 percent--as opposed to less than 1 percent for silicon.

Accordingly, CMEMS devices require much lower voltages to achieve a given mechanical deflection, and their mechanical range of motion is much larger than with silicon MEMS for equivalent voltages.

Another advantage of CMEMS is in the deposition method. Silicon MEMS uses chemical vapor deposition (CVD) techniques, in which "sacrificial" layers are deposited on top of a silicon wafer to create a structure. CVD requires a large, complex, expensive machine.

CMEMS technology, on the other hand, uses spin-on deposition to pattern a structure, in much the same way that photor esist is applied in IC fabrication. With this technique, a silicon wafer is placed on a chuck that spins at a high speed, but instead of photoresist, elastomer is deposited, leveling out to form a thin smooth layer on top of the wafer. CMEMS technology employs the same spin-on equipment.

All of these factors combine to lower production costs and improve yields for CMEMS components. This is especially important in the optical networking industry, which is deploying denser and higher bandwidths to boost network capacities. This heightens the need for costeffective, higher-performance tunable photonic components that increase network efficiency and reliability.

The first components to use the new CMEMS technology will be tunable optical filters, often referred to as interferometers. Used for a variety of optical networking applications--such as optical performance monitors-- tunable filters transmit light of a predetermined wavelength and reflect the non-transmitted light back toward the source .

Tunable filters may be achieved using a number of methods, but the most commonly used is the Fabry-Perot-based interferometer, which consists of two parallel partial reflectors (plates) facing each other, with a small gap in between. When the optical path length between the plates is an integer number of half waves, the structure becomes optically resonant, with zero electric field intensity at the boundaries and energy coupled through the filter.

To make the interferometer tunable, one of the plates is fixed and the other movable, with the distance between them controlled--typically by piezo forces --to tune the wavelength that will pass through the filter.

Although the movable plate is intended to move away in a completely parallel fashion, in actuality it moves with a minuscule tilt error. Controlling this tilt error allows high finesse, which measures how narrow a spectral line can be filtered.

In dense wavelength division multiplexing (DWDM) systems, which transmit nu merous wavelengths of light simultaneously over a single optical fiber, high finesse is critical because all these channels are spaced extremely close together.

The objective is for the filter to transmit one of these channels but none of the adjacent ones. Accordingly, a very sharp, very narrow filter is required, and the higher the finesse, the sharper the filter.

See related chart

The first CMEMS tunable filters comprise three layers of micro-machined silicon wafers. The first layer is a silicon substrate with a high-reflectivity dielectric mirror on its interior surface and a high-efficiency anti-reflection coating on its outside surface. The second layer contains the high-reflectivity movable mirror that is the tuning element of the Fabry-Perot. It has a high-reflectivity dielectric mirror depo sited on a silicon island that is suspended from a perimeter frame using an elastomer.

Three layers
Three segmented electrodes on the backside of the elastomer layer match up with the electrostatic drive contained on the third and final layer. Voltages applied between the second and third layers control the gap between the first two layers, while the segmentation of the electrodes enables precise adjustment of the tilt.

The CMEMS Fabry-Perot tunable filters will include a 45-nanometer version designed to measure wavelength, power and optical signal-to-noise ratio. It is targeted for monitoring either the C or L bands in the telecommunications spectrum.