CHAPTER 2 RF MEMS BASICS This chapter provides the basic introduction to RF MEMS switches. RF MEMS have in general seen a remarkable growth in the past two decades due to the immense potentials in defense and commercial applications. The major part of this chapter is committed to the comparison of the RF MEMS switches with state of the art solid state switches. RF MEMS switches reported over last decade, applications and challenges related to the reliability of the devices are also discussed. 2.1 Switches for Microwave Applications Currently in microwave industry, mechanical and semiconductor switches are being used. The main use of RF or microwave switches is in signal routing and impedance matching [1]. Telecommunication applications cover a broad range of frequencies, ranging from MHz to radio frequencies (GHz). This broad spectrum requires variety of RF switches for different frequency bands. Also, personalized use of hand held electronic devices emphasize more on the scalability or downsizing of devices with high performance. The selection of the switch is largely governed by frequency and speed of the application under consideration, e.g. in some applications high power handling is required, thus a mechanical switch can be used for this type of requirement; in some areas very high speed is required with moderate isolation and low insertion loss and power handling is not an issue, thus PIN diodes or FETs can be used for this purpose; in some applications very high isolation with very low insertion loss is required with moderate speed and hence MEMS switches are best candidates to fulfill this criteria. Different types of microwave switches exist in the market, everyone with its own set of pros and cons. In the case of silicon FETs, they can handle high power at low frequencies, but the performance drops significantly with increase in frequency. Whereas in the case of GaAs MESFETs and PIN diodes, the high-frequency operation is quite well with small signal amplitudes or briefly it can be summed up that, at high frequencies, the solid state switches have high insertion loss and poor isolation. Thus, mechanical coaxial and waveguide switches offer the advantage of low insertion loss, high 11
isolation, large power handling and high linearity, but are heavy and slow. On the other hand semiconductor switches provide faster switching speed and are smaller in size, but have inferior performance than mechanical coaxial switches. For above problem, the best remedy can be the use of MEMS technology for RF applications. MEMS technology has its own set of advantages to be used for RF applications. MEMS switches provide the advantages of both mechanical and semiconductor switches. They provide high isolation and low insertion loss with almost zero DC power consumption with a small size and low weight as discussed in further sections of this chapter. 2.2 RF MEMS Micro-electro-mechanical systems (MEMS) have been developed since 1970s for different applications, e.g. pressure sensor, accelerometers, temperature sensor & other sensor devices. In view of the fact demonstrated by Peterson in 1979 [2], that a bulk micromachined cantilever can be used as a switching element, the standard device of the RF MEMS the switch, is the first and the most studied in this field. MEMS switches for low frequency applications have been demonstrated in the early 1980s but remained a laboratory curiosity for a long time. The first MEMS switch designed for RF applications was reported in 1990 by Larson [3], and its results were so outstanding that afterwards several groups like Texas Instruments, Rockwell Science Centre, Raytheon, LG, etc. start research in this field [4]. RF MEMS devices which work as basic building blocks for any RF system are: RF MEMS switches, high Q inductors, filters & resonators and tunable capacitors or varactors. 2.2.1 RF MEMS Switch In a variety of applications, high frequency switches are essential components, e.g. mobile phones, wireless local networks, radars and satellites etc. The thrust for RF MEMS switch applications in communication has been mainly due to the highly linear characteristics of the switches over a wide range of frequencies. The MEMS devices offer better isolation (>30 db) and low insertion loss (<0.15 db) compared to the contemporary solid state devices. With high levels of integration, negligible current, low power consumption and improved overall performance, RF switches are preferred for space, air borne and hand held communication applications. Phase shifters, switch matrices, receivers 12
RF Switch Series Shunt Capacitive type Ohmic type Capacitive type Ohmic type Lateral Movement Vertical Movement Lateral Movement Vertical Movement Lateral Movement Vertical Movement Lateral Movement Vertical Movement Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation Electrostatic Actuation Magnetostaic Actuation Piezoelectric Actuation Thermal actuation 2x2x2x4=32 Fig. 2.1: Configuration tree for RF MEMS switches. RF MEMS switch can be implemented in 32 different configurations. and transmitter sections are some of the applications being developed using MEMS switches. Like any other switch, a MEMS switch has two stable states, ON and OFF. Switching between these two states can be achieved through the movement of a free moving armature, moved using different types of actuation mechanisms; e.g, electrostatic, piezoelectric, thermal or magnetic actuation. Electrostatic actuation is the most popular because of its low power consumption, small electrode area and relatively short switching time [1]. The other advantages of using electrostatic actuation are low fabrication complexity, possibility of biasing the switch using high resistance bias lines and easy integration with existing fabrication technology, with coplanar waveguide (CPW) & microstrip lines. In MEMS switches, there is a mechanical armature, whose movement decides the working of the switch. This movement can be vertical or lateral. In a 13
transmission line a MEMS switch can be placed in series or in shunt configuration. Also, MEMS switches can be classified according to the contact type between the armature and the transmission line. The contact can be a metal to metal type (Ohmic switches) or capacitive type (Capacitive switches). Ohmic switches are good at low frequencies (<10 GHz), whereas at high frequencies capacitive switches show better performance. Overall there are 32 different configurations of RF MEMS switches depending upon the actuation mechanism, contact type, armature movement and circuit implementations. Fig. 2.1 shows the configuration tree for RF MEMS switches. The main performance characteristics of RF MEMS switches are high isolation in off-state, low insertion loss in on-state, return loss in Semiconductor Switch G MEMS Switch S D IN OUT S ON-state S OFF-state S OUT ON-state OUT OFF-state OUT R ON 0.5 Ω C OFF 0.25ff G 18 nm R ON 5 Ω C OFF 45ff Physical Gap of 3 µm D D D IN IN IN I sd I con Linearity V sd Linearity V con Fig. 2.2: Comparison between semiconductor and MEMS switches. 14
both states, power handling capability, low power consumption and linearity. Fig. 2.2 shows the comparison between semiconductor switches and MEMS switching devices. In on-state, semiconductor switches have significant resistance (5 Ω) between source and drain, leading to high insertion loss; whereas, in MEMS switches there is a direct metal to metal contact resulting in minimum resistance (<0.5 Ω), leading to very low insertion loss [4]. The low insertion loss in MEMS switches thus eliminates the requirement for amplifiers, etc. to compensate for signal loss due to switching elements. In solid-state switches, off state is poorer either because of the leakage current or the large parasitic capacitance between the source and drain. In the case of MEMS switches, higher isolation can be achieved because of the large physical gap between the bridge and the transmission line. As compared to the state of the art semiconductor switches and PIN diodes or FETs, MEMS switches have many advantages, such as [4]: 1. Nearly zero power consumption: Electrostatic actuation does not consume any current leading to very low power dissipation (10-100nj per switching cycle). 2. Very low insertion loss: RF MEMS switches have insertion loss of -0.1dB up-to 40 GHz. 3. Very high isolation: RF MEMS switches are fabricated with air gaps, and hence they Parameter MEMS Switches PIN FET Voltage (V) 10-80 3-5 3 5 Current (ma) 0 3-20 0 Power Consumption (mw) 0.05 0.1 5-100 0.05 0.1 Switching Time 1 300 µs 1 100 ns 1 100 ns Isolation (1-10 GHz) Very High High Medium Isolation (10 60 GHz) Very High Medium Low Isolation (60 100 GHz) High Medium None Insertion Loss (1 100 GHz) (db) 0.05 0.2 0.3 1.2 0.4 2.5 Power Handling (W) < 1 < 10 < 10 Cost (US$) 8-20 0.9-8 0.45-5 Life Time (cycles) > 10 9 > 10 9 > 10 9 Table 2.1: Comparison of MEMS switches with solid state devices. 15
have very low off state capacitance providing excellent isolation (capacitive switches). 4. Very low cost: RF MEMS switches are fabricated using surface micromachining techniques and batch fabrication techniques and can be built on LTCC, Si, GaAs or quartz substrates. 5. Low intermodulation products: MEMS switches are very linear devices, and therefore results in very low intermodulation distortion and no measurable harmonics. Table 2.1 summarizes the performance comparison of the MEMS switches with PIN diodes and FETs. Semiconductor switches provide the desired performance in terms of switching speed and have low cost, but present power constraints and significant loss at high frequencies makes them less preferable than MEMS switches. 2.2.2 Figure of Merit In semiconductor switches, with direct current, off-state to on-state resistance ratio can be used to characterize a switch. At microwave frequencies the off-state is determined by the capacitance and ratio of impedance is given be [5] The product of C and R is termed as FOM. A small FOM is better as it means a small on state impedance as compared to the off-state. In capacitive type of switches, performance characterization can be done on the basis of off-on capacitance ratio. For capacitive MEMS switch, FOM is given by where, ε r is the dielectric constant of the dielectric material, g o is the air gap and t diel is the dielectric layer thickness between the capacitors. For capacitive type of switches a high FOM is better as it means down-state capacitance should be as high as possible and up-state capacitance should be as small as possible. FOM for capacitive type of switches can be improved by choosing high-k dielectric material and scaling down the thickness of the dielectric layer as discussed in Chapter 4. 16
Anchor Switch contacts RF-in RF-out Side view Front view Pull down electrode Fig. 2.3: Electrostatically actuated metal to metal contact type MEMS switch. 2.2.3 RF MEMS Switch Contact Configurations As shown in Fig. 2.1, RF MEMS switch can be broadly classified on the basis of contact mechanism (e.g. ohmic and capacitive contact), armature movement (e.g. vertical or lateral), placement of armature with respect to transmission line (e.g. series or shunt) and actuation mechanisms (e.g. electrostatic, piezoelectric, magnetic or thermal) [6, 7]. In ohmic contact type switches, also known as metal to metal contact switches, there is a direct contact (resistive contact) between the movable membrane and the transmission line (Fig. 2.3). Such type of switches are useful from DC to 8 GHz. The reliability of ohmic contact type switches is largely determined by the metal to metal contact. Fig. 2.4 illustrates a capacitive type of RF MEMS switch. The performance of a capacitive contact type switch depends on the thickness and roughness of the dielectric layer and the gap between the membrane and the transmission line. The capacitance change between the up-state and down-state decides the transmission of the signal. The capacitance ratio (C down /C up ) is the key parameter as discussed. A high capacitance ratio is always desirable. Because of the coupling nature, MEMS capacitive switches are not suitable for low frequency applications. Metal Bridge Anchor Gap Anchor Metal Bridge CPW Ground CPW Ground CPW Ground CPW Ground Signal line Up-state Dielectric Signal Line Down-state Dielectric Fig. 2.4: RF MEMS capacitive contact type switch in up-state and down-state. In up-state signal passes from input port to output port, whereas in down state, signal couples to the ground through the capacitive contact between 17 the transmission line and the metal bridge.
Compared to the ohmic type switches, lifetime is not an issue due to capacitive contact, however, the reliability is undermined due to the dielectric charging. 2.2.3.1 Series and Shunt implementation From the application perspective, the MEMS switches are further classified as series or shunt switches. The series and shunt configuration is determined by the position of the metal armature with respect to the transmission line. Both the ohmic contact type and capacitive contact type switches can be used in series or shunt configurations. But, generally ohmic switches are used in series and capacitive switches are used in shunt configurations. Switches can further be classified as in-line and broadside switches. In an inline switch, the armature is an integral part of the transmission line (Fig. 2.5 (a)), whereas in broadside switches the armature is placed perpendicular to the transmission line and is connected to the CPW ground planes (Fig. 2.5 (b)). A basic capacitive shunt switch consists of a movable metal bridge anchored to the ground plane of the CPW as shown in Fig. 2.5 (b). When zero bias voltage is applied at the actuation electrodes, the membrane is at a gap from the transmission line and the RF signal transmits through the signal line. In this stage, the small overlap capacitance (femto-farads) results in signal loss termed as transmission loss and is given by S 21. When actuation voltage is applied at the inner electrodes, due to electrostatic force developed between the membrane and the actuation electrode, the movable bridge moves down and make a capacitive contact with the dielectric over the transmission line. CPW GND Signal Line CPW GND Membrane Membrane ground ground Actuation Electrode Signal Line Actuation Electrode Dielectric (a) Fig. 2.5: (a) Top view of in-line RF MEMS switch, (b) Cross-sectional view of broadside RF MEMS switch. (b) 18
Z 0 Z 0 C u Z 0 Z 0 L C d R s (a) Fig. 2.6:Equivalent circuit of RF MEMS capacitive switch in (a) on-state and (b) off-state. (b) The high down state capacitance couples the RF signal to the ground and thus isolating the output port from the input port. In on or off state the switch can be modeled as shown in the Fig. 2.6 (a) and Fig. 2.6 (b) respectively. 2.2.3.2 Ohmic Series switch As shown in Fig. 2.3 front view, in ohmic series switch, the signal is interrupted by a break in the t-line. The movable armature could be a metallic membrane or a dielectric membrane with metallic contacts. Under no bias condition, the armature is in up position and there is a break in the transmission line; the switch is in off-state. When actuation voltage is applied at the actuation electrode the armature snaps down due to electrostatic force and the signal line gets connected through armature. This is the on-state of the switch. To keep the insertion loss as low as possible the on-state resistance of contacts should be as small as possible. 2.2.4 Other RF MEMS Components 2.2.4.1 MEMS Capacitors There are many broadband applications with specific design requirements in which the capacitor controls the critical electrical parameters. They include low-noise amplifiers, harmonic frequency generators and frequency controllers [8, 9]. Many of the modern wireless system constraints requirement of high quality, low phase noise, stable operation with wide tuning range voltage controlled oscillators (VCOs). The tuning range of these VCOs must be large enough to cover the entire frequency band of interest. These tunable capacitors or varactors are electronically controlled for the desired operation. Semiconductor 19
(b) (a) (c) (d) Fig. 2.7: (a) Top view of variable capacitor, suspended at four anchor points. (b) & (c) electrical equivalent of single electrode capacitor and variable capacitor with electrodes on top and bottom of the suspended membrane, (d) schematic of MEMS inductor coil suspended at a height of 5 µm from ground. on-chip varactors or MOS capacitors suffers from excessive series resistance and nonlinearity [10]. RF MEMS varactors on the other hand use highly conducting thick metal layers, with air as a dielectric, thus offering substantial improvement over conventional onchip varactor diodes in terms of power loss. In addition, the RF MEMS capacitors have excellent linearity, wide tuning range and ability to separate the control circuitry from the signal circuit, which greatly simplifies the overall design. The tuning of the capacitance can be achieved by three different ways; (a) tuning the dielectric constant (b) by tuning the air gap and (c) by tuning the overlap area. The first method can be implemented by considering the dielectric material, whose dielectric constant changes with the change in boundary conditions e.g. dielectric constant of BST changes with change in temperature. In the later two cases the air gap between the two plates or overlap area can be changed by electrostatic forces and capacitance can be linearly changed with the application of applied voltage. The principle of gap tuning of the variable capacitors and varactors is similar to the RF MEMS capacitive switches as shown in Fig.2.5 (b). The plate or membrane is suspended with the anchors to the CPW ground plane. When actuation voltage is applied at the actuation electrodes, the membrane starts moving down due to electrostatic force developed between the two plates. This operation can be linearly controlled by changing the dimensions of the membrane. The down state capacitance is determined by the dielectric layer over the transmission line and the overlap area. As the substrate is high resistivity 20
substrate and metal lines are fabricated over that dielectric substrate, the effect of parasitic capacitances is almost negligible. Fig. 2.7 (a) shows the top view of variable capacitor suspended over the actuation electrode. Fig. 2.7 (b) and (c) shows the electrical equivalent of a variable capacitor with single side actuation electrode and wide-range variable capacitor with actuation electrodes on both bottom and top side of the movable membrane. 2.2.4.2 MEMS Inductors RF inductors are needed in any wireless front-end circuitry; the performance of both transceivers and receivers depend heavily on this component. The key parameters that characterize the performance of inductors are the quality factor Q, inductance L and self resonance frequency. The Q-factor is an important characteristic for inductors and determines the energy dissipation in the inductors; high Q implies low energy dissipation. The quality factor for planar spiral inductors and junction diode capacitors are only of the order of low 10s at higher frequencies and hence alternative off-chip technologies including inductors and tunable capacitors are often used for high Q applications. High-Q inductors reduce the phase noise and the power consumption of Voltage Controlled Oscillators (VCO's) and amplifiers and reduce the return loss of matching networks and filters. The quality factor of inductors can be increased by using a thick metal layer and by isolating the inductor from the substrate. To isolate the inductor, bulk micromachining or self-assembly can be used. Further, tunable inductors allow for performance optimization of RF front-end circuits. Most of the reported MEMS inductors are static fixed value inductors. Few published papers report use of MEMS switches as variable inductors [11]. However, it provides only discrete values of inductance depending on the ON/OFF configuration of the switches. A tunable inductor using self-assembly technique has been reported [12]. The main disadvantage of using these type of inductors is that, they are suspended on the substrate, thus becoming prone to the electromagnetic interference in the transceiver system. The demand for fully integrated planar inductors and capacitors for the realization of MEMS and monolithic microwave integrated circuits (MMICs) is growing steadily. Conventional inductive components are inherently three-dimensional (3D) in nature and the implementation of these components in planar shape is quite challenging. Small size and weight, low power consumption, mass production, reliability and reproducibility are some of 21
the numerous advantages of integration of MICs with MEMS [13]. MEMS technology improves the on-chip inductor performance by etching away the sacrificial layer (spacer or lossy substrate) underneath the spiral inductor, resulting in a hanging membrane or suspended inductor coil. Fig. 2.7 (d) shows the schematic of spiral MEMS inductor, suspended at a height of 5 µm from the ground plane. 2.2.5 General Fabrication Process for RF MEMS Capacitive Switches In general RF MEMS capacitive switches are fabricated using surface micromachining techniques. Though designing and fabricating a RF MEMS switch with microstrip configuration results in smaller size, the CPW configuration results in easy fabrication process. The choice of materials and fabrication process design depends on the specifications of the device. Most of the reported switches, are fabricated using a five mask level process, excluding packaging. During the material selection for the fabrication, the overall thermal budget and the material etching at different phases are the general Spacer > 5 kω-m Thermal Oxide (a) Photoresist mould (d) Electroplated Bridge Ground Signal Line Actuation Electrodes Ground (b) (e) Final released switch Dielectric (c) (f) Fig. 2.8: (a) - (f): General process steps for RF MEMS capacitive switch. 22
compatibility issues for process design. For ohmic contact type switches, the contact material and the thickness of the contacts are the key parameters which will decide the lifetime of the switch. Below are the summarized, general fabrication steps for RF MEMS capacitive switch fabrication (Fig. 2.8): a. For RF MEMS switches, a high resistivity substrate (> 5kΩ-m) is the starting material. A thermal oxide layer is generally preferred over the Si substrate [3, 14, 15]. b. The actuation electrodes, transmission line and ground planes of CPW are fabricated in metal layers and patterned accordingly. c. A thin dielectric layer is deposited and patterned. This layer acts as a dielectric layer for the capacitive type of switch. d. A sacrificial layer or spacer layer, which could be a metal layer or a photo-resist is deposited and patterned to obtain the gap between the transmission line and the suspended membrane. e. After spacer patterning, a thin seed layer is generally deposited followed by a photoresist mould formation using thick photoresist lithography for suspended membrane formation. f. Suspended membrane is fabricated by electroplating process, followed by a release process. The release process can be dry plasma ashing or wet etching process, depending upon the spacer or sacrificial layer. 2.3 Existing RF MEMS Switches (MEMS Switch Library) Following section includes some RF MEMS switches, which were developed by industry, university and government laboratories. As discussed in former section, almost all of the MEMS switches are fabricated using surface micromachining techniques on a high resistivity substrate, which can be Si, GaAs or quartz [5]. 2.3.1. Raytheon Capacitive MEMS Shunt Switch The Raytheon shunt switch also known as Texas Instruments Switch, was developed by Chuck Goldsmith and his co-workers in 1995-2000 [14, 16]. The device is a capacitive type of MEMS switch with 1000Å of Si 3 N 4 as a dielectric layer. The bridge membrane is composed of 0.5 µm of aluminum that is suspended 3-5µm above the transmission line. 23
The device length and width are 270-350 µm and 50-200 µm respectively. Thick polyimide used as sacrificial layer is released using a plasma etching technique. A capacitance ratio of 80-120 has been achieved for actuation voltages of 30-50 volts. The switching time is 3µs/5µs (Down/Up) providing an isolation and insertion loss of -35 db (at 30 GHz) and -0.07 db (10-40 GHz) respectively. 2.3.2. University of Michigan Capacitive MEMS Shunt Switches The university of Michigan has developed a novel low-voltage MEMS capacitive shunt switch with Nickel suspended membrane [17]. The membrane is suspended using meander type support structures, which results in a low spring constant and hence low actuation voltage for the switch. For the length of 500-700 µm, width of 200-250 µm and a thickness of 2-2.5 µm, actuation voltage between 6-20 Volts can be achieved for a gap height of 4-5 µm. The dielectric layer used is Si 3 N 4 having a thickness of 1000-1500Å. The down state switching time is between 20-40 µs with a capacitance ration of 30-50. The switch show an isolation of -25 db (at 30 GHz)in off-state and an insertion loss of -0.1 db (1-40 GHz). Due to the low spring constant, the switch is vulnerable to external mechanical forces such as acceleration and vibrations. This problem can be solved by incorporating a second electrode (pull-up) above the suspended membrane. This pull-up electrode will hold the membrane in up-state when the switch is not actuated, thus reducing the switch sensitivity to mechanical shocks and vibrations. Also, the up-state time of low spring constant switches can be improved by incorporating the electrode on top of the suspended membrane. 2.3.3. University of Michigan Capacitive MEMS Shunt Switches As discussed in the above example, to decrease the spring constant, spring type anchors can be used, and to reduce the devices vulnerability to shocks and vibration another electrode can be fabricated on the top of the suspended membrane. But, fabricating another electrode on top of the suspended membrane is a complex fabrication process, which requires need of special and costly releasing equipments and techniques. In view of this the University of Michigan also developed, a low height switch with Ti/Au membrane, having high spring constant [18, 19, 20, 21]. The switch is based on a 0.8-1.0 µm thick Ti/Au membrane, suspended at height of 1.5-2.2 µm above the transmission line. A low-gap 24
height results in a low pull-down voltage of 12-24 Volts while still maintaining a high spring constant for the membrane. The low-height switch therefore has a relatively high mechanical resonant frequency and a fast switching time. Also it is not sensitive to vibrations, with a compromise in capacitance ratio (20-40). An isolation of -30 db (at 30 GHz) with an insertion loss of -0.03 db (at 10 GHz) and -0.05 db (at 30 GHz) has been achieved with this switch. 2.3.4. The LG-Korea Capacitive Shunt Switch As discussed above, with small gaps, one has to compromise with the capacitance ratio. Park and team presented very high capacitance ratio MEMS capacitive switches by using high-k dielectric materials [22, 23]. The switch design is based on the fixed-fixed beam capacitive shunt switches with strontium-titanate-oxide (SrTiO 3 ) as a dielectric layer. The relative dielectric constant of SrTiO 3 is 30-120, depending on the deposition temperature with very low leakage current. The reported capacitance ratio of fabricated devices is 600 with a down state capacitance of 60 pf. The membrane is suspended through the low spring constant springs, resulting in actuation voltages of 8-15 Volts. The switch isolation is better than -40 db (at 3-5 GHz) and -30 db (at 10 GHz) with an insertion loss of -0.1 db (at 1-10 GHz). 2.3.5. DTIP low actuation voltage switch This switch is quite long with four actuation electrodes and a low actuation voltage of 7.5 Volts [24]. The structure consists of a large membrane supported over three pillars. At rest state, the membrane is at a nominal height over the transmission line, which is the offstate of the switch. Odd-states can be achieved by applying the actuation voltage at any of the odd or even electrodes. When actuation voltage is applied at the outer electrodes, a large up -state deflection can be obtained. The switch shows an isolation of -30 db (at 24 GHz) and an insertion loss of 0.65 db (at 24 GHz). As the membrane is only supported over the pillars and is not anchored to any plane, the switch is highly vulnerable to mechanical shocks and vibrations. 25
2.4 Challenges in RF Switches Most of the challenges in RF MEMS development are interrelated or a trade off with other parameters. As an example, many of the fabricated devices have actuation voltages above 20 Volts and hence need up-converters in order to be integrated with other semiconductor based systems with standard voltage sources. Such a system consumes more space and is expensive. Also, high actuation voltage switches have low reliability in terms of dielectric charging and stiction. In some applications, high switching speed is needed, which is a major problem area in MEMS switches. To decrease the switching time, the spring constant of the membrane should be increased, which in turn results in increase of actuation voltage, which is not required. Also, power handling is one of the major issues in MEMS switches. RF MEMS metal to metal contact switches with life time up-to a few billion cycles can handle only (0.5-5 mw) of RF power. For high RF power handling, the reliability of MEMS switches reduces drastically. The failure mechanisms depend on the RF power used and can be due to thermal stress, dielectric breakdown, self-actuation and current density issues. Capacitive switches with their large contact area can handle more RF power than metal-to-metal contact switches and therefore are preferred for applications requiring 30-300 mw of RF power. Another major issue in RF MEMS switches is the packaging of devices. The operation of the device is highly affected by the presence of water vapors, contaminations and the hydrocarbons present in the atmosphere. Packaging contributes to almost 80% of the total cost of the device manufacturing cost and its performance and reliability highly depends on the packaging. Thus, in order to meet the cost, performance and reliability MEMS packaging tends to be customized to specific application. To avoid failure of RF MEMS switches proper hermetic packaging is required. Hermetic packaging is a complex technology, which costs almost 10 15 times higher for the MEMS devices as compared to the semiconductor devices. But, hermetic bonding process requires very high temperatures for achieving a good seal contact. For released structures or suspended membranes high temperature processing can bow the membranes by several microns, thus damaging the switch or deteriorating the reliability of switch. Other packaging issues and different packaging technologies and packaging levels are discussed in Chapter 5 of this thesis. 26
2.5 Symmetric Toggle Switch As discussed in former sections, low actuation voltage switches with low switching time and high reliabilty against mechanical shocks and vibrations are required. Few of the existing capacitive switches have high capacitance ratio, but they have high switching time and actuation voltages. Few have low actuation voltages, but are highly prone to mechanical shocks and vibrations. In the meander based switch design the reliabilty against self biasing and external mechanical shocks can be improved by incorporating an additional electrode on the top of the suspended membrane. This increases the process complexity and also adds the parasitic capacitances. To obviate the above problems, this thesis focusses on a novel switch topology called Symmetric Toggle Switch (STS) [25]. STS is implemented using standard 50 Ω CPW configuration (Fig. 2.9). The device consists of a pair of micro-torsion actuators placed symmetrically around the transmission line. They are suspended at a gap of 3µm above the transmission line, and anchored to CPW ground planes using euler beams called as 'spring'. They are connected to each other with connecting levers and an overlap area. The transmission line is divided into three parts. The input and output ports are in thick gold; whereas, the area under the bridge capacitive area is fabricated in multi-metal layer (Ti/TiN/Al:Si/Ti/TiN) to provide the smooth capacitive contact. There are four actuation electrodes beneath the micro torsion actuators for pull-in and pull-out. spring contact area micro-torsion actuator anchor connecting lever underpass area pull-out electrode pull-in electrode transmission line Underpass Area with Floating metal layer Fig. 2.9: 3 -D model of Symmetric Toggle Switch. 27
Additional electrodes (on the same plane), to clamp the beam in up-state makes the device impervious to self biasing and vibrations. The use of micro-torsion springs also improves the travel range. The device can also be configured as MEMS varactor with a wide capacitance range for a given gap and voltage, not achievable with conventional MEMS varactor design. Another, outstanding feature of the device as a RF MEMS switch is its tunability over a wide frequency range. The present configuration reduces the in-built stress related deformation, though devices become longer as compared to other similar topologies. Due to the presence of four actuation electrodes, inner two for pull-in and outer two for achieving minimum insertion loss and getting rid of external shocks and vibrations in on state, the switch dimensions are very large. Also, long size result in low spring constant implying low resonance frequency and therefore increase in up-state time. Thus, the outer two electrodes can be used to reduce the up-state time. Chapter 3 explains the design & modeling, fabrication and characterization of STS. Dimensional optimization of actuator area and the capacitive area in view of the required mechanical and electrical performance has been done by incorporating high-k dielectric material (hafnium oxide) further explained in Chapter 4. 28
References: [1]. H. A. C. Tilmans, MEMS components for wireless communications, in Eurosensors XVI, 16th European Conference on Solid State Transducers, Prague, CZ, September 15-18, 2002, pp. 1-34. [2]. K. E. Peterson, Micromachined Membrane Switches on Silicon, IBM J. Res. Dev., vol. 23, 1979, pp. 376-385. [3]. L. E. Larson, R. H. Hackett, M. A. Melendes and R. f. Loher, Micromachined Microwave actuator (MIMAC) technology a new tuning approach for microwave integrated Circuits, in Microwave and Millimeter wave Monolithic Circuits Symposium digest, Boston, MA, USA, June, 1991, pp. 27-30. [4]. Gabriel M. Rebeiz, RF MEMS Theory, Design and Technology, 1st ed. Wiley- Interscience, 2003. [5]. P. D. Grant, M. W. Denhoff, A comparison between RF MEMS switches and semiconductor switches, Proceedings of the 2003 International conference on MEMS, NANO and Smart Systems (ICMENS'04), 2004. [6]. Elliot R. Brown, RF-MEMS switches for reconfigurable integrated circuits, IEEE Transactions on Microwave theory and Techniques, vol. 46, no. 11, November 1998, pp 1868-1880. [7]. Gabriel M Rebeiz, Jeremy B Muldavin, RF MEMS switches and switch circuits, IEEE microwave magazine, vol. 59, December 2001. [8]. J Jason Yao, RF MEMS from a device perspective, J of Micromech. and Microeng., vol. 10, April 2000. [9]. S. Lucyszyn, Review of radio frequency micromechanical systems technology, IEE Proc. Sci Meas. Technol., vol. 151, no. 2, March 2004. [10]. M. Soyer, K. A. Jenkins, J. N. Burghartz and M. D. Hulvey, A 3V 4GHz NMOS voltage controlled Oscillator with integrated resonator, in Proc. IMAPS 2001, Baltimore, Maryland, USA, October 9-11 2001, pp. 471-476. [11]. W-H. Chu, M. Mehregany, and R. L. Mullen, Analysis of Tip Deflection and Force of a Bimetallic Cantilever Micro-actuator, J. Micromech. Microeng., vol. 3, 1993, pp. 4-7. 29
[12]. V. M. Lubecke, B. Barber, E. Chan, D. Lopez, M. E. Gross, and P. Gammel, Self- Assembling MEMS Variable and Fixed RF Inductors, IEEE Trans. Microwave Theory and Techniques, vol. 49, no. 11, November 2001, pp. 2093-2098. [13]. Vijay K. Vardan, K. J Vinoy, K. A. Jose, RF MEMS and Their Applications, 1st ed. Wiley-Interscience, 2003. [14]. C.L. Goldsmith, Z. Yao Z, S. Eshelman and D. Denniston, Performance of low loss RF MEMS Capacitive Switches, IEEE Microwave and Guided Wave Letters, vol. 8, no. 8, August 1998, pp. 269-271. [15]. C. Goldsmith,T. H. Lin, B. Powers, W-R. Wu and B. Norwell, Micromechanical Membrane switches for microwave application, in Tech. Digest, IEEE Microwave Theory and Technique Symp., 1995, pp. 91-94. [16]. Z. J. Yao, S. Chen, S. Eshelman, D. Denniston, and C. L. Goldsmith, Micromachined low-loss microwave switches, IEEE J. Microelectromechanical Systems, vol. 8, June 1999, pp. 129-134. [17]. D. Peroulis, S. Pacheco, and L.P.B Katehi, MEMS devices for high isolation switching and tunable filtering, in IEEE MTT-S International Microwave Symposium Digest, Boston, MA, June 2000, pp. 1217-1220. [18]. J. B. Muldavin and G. m. Rebeiz, High isolation MEMS shunt switches; Part 1: Modeling, IEEE Trans. Microwave Theory Tech., vol. 48, no. 6, June 2000, pp. 1053-1056. [19]. J. B. Muldavin and G. M. Rebeiz, High isolation MEMS shunt switches; Part 2: Design, IEEE Trans. Microwave Theory Tech., vol. 28, no. 6, June 2000, pp. 1053 1056. [20]. N. S. Barker and G. M. Rebeiz, Distributed MEMS true-time delay phase shifters and wideband switches, IEEE Trans. Microwave Theory Tech., vol. 46, no. 11, November 1998, pp. 18881-1890. [21]. J. B. Muldavin and G. M. Rebeiz, Inline capacitive and DC contact MEMS shunt switches, IEEE Microwave Wireless Comp. Letters, vol. 11, no. 8, August 2001, pp. 334-336. 30
[22]. J. Y. Park, G. H. kim, K. W. Chung, and J. U. Bong, Fully integrated micromachined capacitive switches for RF applications, in IEEE MTT-S International Microwave symposium Digest, Boston, MA, June 2000, pp. 283-286. [23]. J. Y. Park, G. H. kim, K. W. Chung, and J. U. Bong, Electroplated RF MEMS capacitive switches, in IEEE International conference Microelectromechanical Systems, January 2000, pp. 639-644. [24]. Renaud Robin, Salim Touati, Karim Segueni, Olivier Millet, Lionel Buchaillot, A new four states high deflection low actuation voltage electrostatic MEMS switch for RF applications, Proc. of MEMS and MOEMS, April 9 - April 11, 2008. [25]. K. J. Rangra, Electrostatic low actuation voltage RF MEMS switches for Telecommunication, Ph. D. Thesis, International Doctorate school in information and communication technologies, DIT, University of Trento, Italy, February 2005. 31
32