The fundamentals of Analog Devices’ revolutionary MEMS switch technology
05 November 2018
Over the last 30 years, MEMS (microelectromechanical systems) switches have been consistently touted as a superior replacement to limited performance electromechanical relays – and when compared to traditional electromechanical relays, Analog Devices’ (ADI) MEMS switch technology enables a huge leap forward in RF and DC switch performance, reliability, and in miniaturisation. This tutorial describes their breakthrough in MEMS switch technology.
For the digital issue of this piece, please visit this link – or click here to register for EPDT's magazine.
MEMS switches (see Figure 1 above) effectively revolutionise how electronic systems are realised, by providing an easy-to-use, small form factor switch that can route 0 Hz/DC to 100s of GHz signals – reliably and with minimal losses. This performance advantage impacts across a huge spectrum of equipment types and applications.
Electrical test and measurement systems, defence systems applications and healthcare equipment are just some areas that can reach previously unattainable levels of performance and form factor, all enabled by MEMS switch technology.
Contemporary switching technologies all have drawbacks, with no one technology providing an ideal solution. Relay drawbacks include narrow bandwidths, limited actuation lifetimes, limited number of channels and large package sizes.
MEMS technology has always had the potential to deliver world-class RF switch performance and orders magnitude improvements in reliability in a small form factor, compared to relays. The challenge, however, which has thwarted many companies who have tried to develop MEMS switch technology, has been in the delivery of reliable products in high-volume mass production.
ADI has a rich history with MEMS. The first MEMS accelerometer product successfully developed, manufactured and commercialised was ADI’s ADXL50 accelerometer, which was released in 1991. ADI also released the first integrated MEMS gyroscope, the ADXRS150, in 2002. From these beginnings, ADI has built a huge MEMS product business and a reputation for manufacturing reliable, high-performance MEMS products. ADI has shipped over one billion inertial sensors for automotive, industrial and consumer applications.
MEMS switch fundamentals
Central to ADI’s MEMS switch technology is the concept of an electrostatically actuated, micromachined cantilever beam switching element. In essence, it can be thought of as a micrometre-scale mechanical relay, with metal-to-metal contacts that are actuated via electrostatics.
The switch is connected in a three-terminal configuration. Functionally, the terminals can be thought of as a source, gate and drain. Figure 2 (see below, or click link to view online magazine version) shows a simplified graphic representation of the switch with Case A showing the switch in the ‘off’ position. When a DC voltage is applied to the gate, an electrostatic pull-down force is generated on the switch beam. This is the same electrostatic force as would be seen in a parallel-plate capacitor, having positive and negative charged plates that attract each other.
When the gate voltage ramps to a high enough value, moreover, it creates enough attraction force (red arrow) to overcome the resistive spring force of the switch beam, and the beam starts to move down until the contacts touch the drain.
This is shown in Case B in Figure 2. This completes the circuit between the source and the drain, meaning the switch is now on. The specific force it takes to pull the switch beam down is related to the spring constant of the cantilever beam and its resistance to movement.
Notice that even in the ‘on’ position, the switch beam still has a spring force that pulls the switch up (blue arrow); but as long as the down-pulling electrostatic force (red arrow) is larger, the switch will remain on.
Finally, when the gate voltage is removed (see Case C in Figure 2) – that is 0V on the gate electrode – then the electrostatic attraction force disappears, and the switch beam acts as a spring with sufficient restoring force (blue arrow) to open the connection between the source and the drain, and then, returns to the original ‘off’ position.
Figure 3 (see immediately above) shows the four main steps in fabricating a switch using MEMS technology. The switch is constructed on a high-resistivity silicon wafer (1), which has a thick dielectric layer deposited on top to provide superior electrical isolation from the substrate below. A standard back-end CMOS interconnect process is used to realise interconnections to the MEMS switch.
Low-resistivity metal and polysilicon are used to make an electrical connection to the MEMS switch, and are embedded into the dielectric layer (2). Metal vias marked in red (2) are used to provide a connection to the switch input, output, and the gate electrode to wire bond pads elsewhere on the die. The cantilever MEMS switch itself is surface-micromachined using a sacrificial layer to create the air gaps under the cantilever beam. The cantilever switch beam structure and bond pads (3) are formed using gold.
Switch contact and gate electrodes are formed using a low-resistance thin metal, deposited on the surface of the dielectric. Wire bond pads are also built using the above steps. Gold wire bonding is used to connect the MEMS die to a metal lead frame, encapsulated into a plastic QFN (quad-flat, no-lead) package for easy surface-mounting on PCBs. The die is not limited to any one type of packaging technology. This is due to the fact that a high-resistivity silicon cap (4) is bonded to the MEMS die to form a hermetic protective housing around the MEMS switch device.
Hermetically enclosing the switch in this way increases the environmental robustness and cycle lifetime of the switch, regardless of what external package technology is used.
Figure 4 (click link or see below) shows a zoomed-in graphic of four MEMS switches in a single-pole four-throw (ST4T) multiplexer configuration. Each switch beam has five ohmic contacts in parallel to reduce resistance and increase power handling when the switch is closed.
The MEMS switch requires a high DC drive voltage to electrostatically-actuate the switch. To make the part as easy to use as possible and further guarantee performance, a companion driver integrated circuit (IC) has been designed by – to generate high DC voltages – and co-packaged with the MEMS switch in a QFN form factor. In addition, the high-actuation voltage generated is applied to the gate electrode of the switch in a controlled manner. It is ramped up to a high voltage in microsecond timescales.
Such ramping helps to control how the switch beam is attracted and pulled down, and improves the actuation, reliability and cycle lifetime of the switch. Figure 5 shows the driver IC and MEMS die in situ in a QFN package. The driver IC only requires a low-voltage, low-current supply, and is compatible with standard CMOS logic drive voltages. This co-packaged driver makes the switch very easy to use – and it has very low-power requirements, in the region of 10 mW to 20 mW.
A key tenet to any new technology is, of course, how reliable it is, and the new MEMS technology manufacturing process was the base that enabled the development of mechanically-robust, high-performance switch designs. This, coupled with a hermetically-sealed, silicon capping process were both crucial to the delivery of truly reliable long-life MEMS switches.
To successfully bring the MEMS switch to commercialisation required extensive reliability testing specific to MEMS, such as switch cycling, lifetime testing and mechanical shock testing. In addition to this qualification, and to guarantee the highest level of quality possible, the part has been qualified using a whole range of standard IC reliability tests.
Long-switch actuation lifetimes are of utmost importance in RF instrumentation applications. The MEMS technology has been developed to bring an order of magnitude improvement in cycle lifetimes compared to electromechanical relays. The high temperature operation lifetime (HTOL) test at 85°C, and the early-life failure (ELF) qualification test, rigorously guarantee the cycle lifetime of the part.
Continuously on lifetime (COL) performance is another key parameter for MEMS switch technology. For example, RF instrumentation switch usage can be varied, and a switch can be left in its ‘on’ condition for extended periods of time. ADI has recognised this fact and has focused on achieving excellent COL lifetime performance for the MEMS switch technology to mitigate lifetime risks. From an initial COL performance level of seven years (mean time before failure) at 50°C, ADI has further developed the technology to deliver a class-leading 10 years of COL at 85°C.
The MEMS switch technology has undergone a comprehensive suite of mechanical robustness qualification tests. Table 1 lists a total of five tests that ensure the mechanical endurance of the MEMS switch. Due to the small size and low inertia of the MEMS switch element, it is significantly more robust than electromechanical relays.
Contact Details and Archive...