MEMS (Micro ElectroMechanical Systems) combine electrical and mechanical components with feature sizes of the order of several microns. MEMS technology has evolved to its present state thanks mainly to the fabrication techniques pioneered by the semiconductor industry. MEMS can be found in large volume applications such as airbag-control sensors for the automotive industry; however, there are many other MEMS applications in every day use. MEMS components in these applications may have moving mechanical parts together with fluidic, electrical, optical, chemical, and biomedical elements.

MEMS for RF Applications

Passive components such as inductors, capacitors, and resistors can take up valuable amounts of real estate on a printed-circuit board (PCB). It is therefore desirable to integrate as many of these components as possible into networks. Traditionally, these networks were fabricated using thin-film techniques. Not only does MEMS technology offer high packing densities, but the technology also has brought with it an additional advantage in that it is possible to make devices that are tunable. This article will focus on such a device: an LC filter.

Simulating RF Behavior
In order to predict the behavior of an RF component, we use computer simulation. By using a High Frequency Structure Simulator (HFSS), it is possible to perform a full-wave electromagnetic analysis of a 3D structure, using the finite-element method. The simulation solves Maxwell's equations, which describe electromagnetic phenomena completely, thus allowing us to accurately predict the performance of the component, taking into account electrical losses and electromagnetic-radiation effects. HFSS is commonly used to model components such as filters, connectors, IC packages, and antennas found in cellular telephones and broadband communications systems. However, because it analyzes the underlying physics of these components, you can apply HFSS with equal success to modeling the electrical performance of MEMS devices.

A major advantage of simulation is that you can predict and optimize the performance of a device without having to fabricate test structures. This capability is particularly important when the cost of building a physical prototype is high.

The MEMS LC Filter
In this example, the main inductor is fabricated in thick copper and isolated from the underlying silicon substrate by a twin dielectric layer (Figure 1a). Electrically, the structure is similar to an inductor in parallel with a small capacitor.

Figure 1:  Fabrication of a tunable LC resonator

A second conductive spiral is formed on a membrane suspended above the first (using MEMS fabrication techniques). This provides a large parallel capacitance, which can be varied by changing the gap between the two spirals using standard MEMS actuation methods based on electrostatic attraction or thermal expansion (Figure 1b). The structure thus behaves as a tunable resonator.

In modeling the electrical behavior, HFSS takes into account the losses in the underlying silicon as well as losses in the copper tracks themselves—this is important since these losses directly affect the quality factor (Q) of the inductor, which is an important parameter for RF applications.

Figure 2:  Resonator performance vs. frequency for different capacitor gaps

Figure 2 shows return loss of a MEMS resonator as a function of frequency for different capacitor gaps. The shape of these characteristic curves is determined by the dimensions of the spiral track and the number of turns in the spiral. Simulation lets the engineer explore this design space and produce a structure that achieves as high a Q factor as is possible over a required tuning range. The use of parameterization and optimization modules to control the simulation makes this a faster and more cost-effective route than the manufacture of physical prototypes.

Conclusion
MEMS technology can be applied in a very wide range of application areas. Particular benefits in the RF arena are the technology's ability to produce compact, low-loss, tunable devices. The characterization and design of such devices through physical prototyping can be both time-consuming, and expensive. Electromagnetic simulation offers a fast and accurate means to predict behavior and hence develop optimum designs.

About the Author