1 Introduction

Radio frequency MEMS switches are very much used instead of the traditional gallium arsenide switches, field effect transistors and photo-injection-diode switches in RF and microwave systems, because of very low power dissipation and consumption, high isolation, low insertion and return losses, less cost and low weight (Rebeiz 2003). Surface micro-machining processes is used to fabricate the MEMS switches which consists of a very thin metallic membrane called as bridge (two end fixed beam) which regulates the transmitting of the electronic signal by mechanical deflection of the bridge over the electrode. The different actuation techniques such as electrostatic (Muldavin and Rebeiz 2000a, b; Pacheco et al. 2000), electromagnetic (Yao et al. 1999), piezoelectric (Ruan et al. 2001) and thermal (Lee et al. 2009) actuation are used to actuate the MEMS switch. Electrostatic actuation is widely used because of negligible power consumption and linearity of the device. Switching operation is achieved by generating an electrostatic force between switch membrane and lower electrode. MEMS switches suffers low switching speed, high actuation voltages in RF circuits. (Daneshmand et al. 2009). A compact structure of Rotary electrostatic comb-drive actuators as a drive mechanism of the switch which operates in three state offers high isolation of − 60 dB at 40 GHz. (Ilkhechi et al. 2017) A high-force and high isolation metal-contact RF MEMS switch has been proposed to increase the isolation of the switch and got − 36 dB at 10 GHz. (Deng et al. 2017) A novel laterally and micro-electro-thermally actuated RF MEMS switch is presented to have the isolation of − 20 dB at 100 GHz (Zolfaghari et al. 2018).

This paper presents an approach to increase the isolation by considering series–shunt configuration RF MEMS switch arranged over the substrate made up of quartz, and with dielectric medium TiO2 (K = 100). The devices results in excellent RF performance with high isolation and low insertion loss characteristics. The series and shunt membranes are designed at same actuation voltage for electrostatic operation. The Electromechanical, pull-in voltage, spring constant and stress distribution and RF performance analysis is carried out using finite element modeler (FEM) tool.

The rest of the paper is organized as follows, in Sect. 2, the design of the proposed RF MEMS switch is discussed with the device structure of both capacitive and shunt membrane, working principle and its specifications and material selection for the switch. In the Sect. 3, various results such as stress distribution analysis, spring constant, pull-in voltage, Cd and Cu ratio analysis and switching response of the switch has been discussed. In Sect. 4, the RF performance characteristics such as insertion and return losses in ON state and isolation losses in OFF state have been discussed.

2 Theory

2.1 Device structure and working principle

The proposed MEMS switch consists of a Coplanar Wave guide which transmits the RF signal, dielectric layer made up of TiO2 (K = 100) over the Quartz substrate. The actuation voltage is required to ON the series membrane and OFF the shunt membrane of the switch simultaneously. The coplanar wave guide consists of signal line with 31.5 µm width and having a gap of 126 µm (Fig. 1; Table 1).

Fig. 1
figure 1

Series–shunt configuration RF MEMS switch

Table 1 Device specification

The performance of the switch is very much depends on both series and shunt membranes. The membranes are designed to have same pull-in voltage to establish the optimized relation between pull-in voltage and electrostatic recovery forces. The lower spring constant is achieved by considering meandering technique for series and crab-leg structure for shunt membrane, these membranes are designed inside the cavity made up on quartz substrate. This cavity helps to restore the membranes.

When the RF input signal is applied at input port, then the series membrane has to be actuate to complete the circuit connection and shunt membrane remains in its original position offers a very low capacitance path for the transmission of signals resulting low insertion loss. Whereas in OFF state of the device, the shunt membrane is actuated and deflects the bridge vertically downwards which offers high capacitive path and at the same time series membrane retains its original position, i.e., opening the circuit to block the RF signal resulting high isolation. The gaps between electrodes are maintained at 3 and 2.5 µm for series and shunt membranes to achieve same pull-in voltage.

2.2 Capacitive/shunt switch membrane

The shunt membrane is designed with crab-leg structured meanders which are having lower spring constant and is 3.02 N/m. The circular perforations are taken in order to decrease the air squeeze film damping and the shunt membrane is shown in the Fig. 2 with dimensions.

Fig. 2
figure 2

Shunt membrane of proposed switch

2.3 Ohmic/series switch membrane

It is essential to design the series membrane with same pull-in voltage with the shunt membrane. Two different types of meanders are designed to reduce the spring constant which is 2.17 N/m. The inner meanders are connected with the membrane helps to maintain it uniformly without bending during actuation. The width of the membrane is exactly equal to the spliced width of the signal line such that completes the signal path in down state. The membrane and its dimensions are shown in Fig. 3.

Fig. 3
figure 3

Series membrane of proposed switch

2.4 Material selection

The material properties play a key role to increase the performance of the RF MEMS switch. Quartz is taken as the substrate material because, crystals of quartz have high resistivity of 105–108 Ωm and acts as excellent insulators at elevated temperatures. Ashby’s approach is used for the selection of materials for the dielectric layer and membranes, in this method, at first, the materials suitable for the component are noted down and are then screened to suit the application. Then, the properties of the screened materials are studied, and, based on the parameters to be optimized, material selection charts are plotted, which gives us an unbiased picture of how the performance of the device is affected by selecting a particular material, and how the compromising effect of the parameters is reflected in the choice of materials. The significance of Ashby’s material selection approach is charts are plotted, which gives us an unbiased picture of how the performance of the device is affected by selecting a particular material, and how the compromising effect of the parameters is reflected in the choice of materials. Dielectric strength and resistivity of the dielectric layer are major parameters to enhance the performance in terms of isolation of the RF signal and charge distribution (Paldas and Gupta 2013) (Fig. 4).

Fig. 4
figure 4

Logarithmic value of resistivity vs. reciprocal of relative permittivity to select the material for dielectric layer of the switch

According to the Ashby for low pull-in voltage the value of 1/εr and log (ρ) should be minimum. Hence the extreme left bottom material is the best material for the dielectric (Sharma and Gupta 2012). The TiO2 is having the dielectric constant (K = 100) helps to increase the isolation in OFF state. For choosing the material for membrane young’s modulus and mass density is considered for better electromechanical performance. The operational frequency is very much dependent on the spring constant. The smaller young’s modulus of the material results in low pull-in voltage and density of the material should also be low for fast switching of the beam (Fig. 5).

Fig. 5
figure 5

Material selection for membranes of the switch

\( \sqrt E \) value must be very small for low pull-in voltage, \( \sqrt {E/\rho } \) value must be large for fast switching. Thus, Al, SiO2 and Au are good material choices for the membrane for having low pull-in voltage. Amongst gold is preferred in this proposed structure because of high electrical conductivity.

3 Fabrication flow

The fabrication process flow is similar to the standard fabrication steps where the surface micromachining technique is used to fabricate this structure.

The Fig. 6a shows the polished upper surface of the silicon substrate. A cavity is formed by etching the portion of the substrate. A mask is used to form patterns over the substrate under UV exposure and the pattern is etched 3 µm depth by the suitable etchant. The TiO2 is deposited with 0.5 µm thickness over the substrate which acts a substrate dielectric which is shown in Fig. 6c. The Fig. 6d shows the deposition of CPW over substrate dielectric. Figure 6e shows the addition of biasing pads and electrodes. Figure 6f reports the addition of Polyimide sacrificial layer to add dielectric layer over the signal line for capacitive switch membrane followed by Fig. 6g that highlights the additional dielectric layer over the signal line. Figure 6h shows the sacrificial layer deposition to fabricate a membrane. The cut-outs are because of the exact mask used. Figure 6i shows the deposited gold material over the Polyimide sacrificial layer. At last, the sacrificial layer is removed by plasma etching or DRIE method. The suspended membrane can be seen in Fig. 6j.

Fig. 6
figure 6

Fabrication flow of proposed MEMS Switch

4 Results and discussions

4.1 Stress distribution analysis

Finite element modeler is used to analyze various parameters of the switch such as spring constant (k), pull-in voltage etc. The analytical calculation of the spring constant is not possible for these type of complicated meander structure because till now analytical formulas are defined for simple cantilever beam, fixed—fixed beam and guided end structures [Nguyen 1998]. Hence there is a need of FEM tool in order to get the required parameters. So, to calculate spring constant the stress distribution need to be analyze.

The gold membrane can withstand the stress up to 100 MPa (Singh 2015). By using solid mechanics physics in COMSOL Multiphysics tool it is analyzed that maximum forced needed to produce the maximum stress of about 100 MPa, beyond this force the membrane cannot withstand and breakdown occurs (Figs. 7, 8).

Fig. 7
figure 7

Capacitive/shunt switch membrane for max. stress of 100 MPa showing maximum force of 11.6 µN

Fig. 8
figure 8

Ohmic/series switch membrane for max. stress of 99.4 MPa showing maximum force of 6.2 µN

It is observed that at 6.2 µN of force the series membrane is producing a stress of 99.4 MPa and at 11.6 µN the shunt membrane produces 100 MPa. Hence the maximum force required to displace the series membrane about 3 µm should be less than 6 µN and for shunt membrane about 2.5 µm should be less than 11.6 µN (Figs. 9, 10).

Fig. 9
figure 9

Capacitive/shunt switch membrane for max. displacement of 2.5 µm showing maximum force of 7.2 µN

Fig. 10
figure 10

Ohmic/series switch membrane for max. displacement of 3 µm showing maximum force of 5.9 µN

For series membrane and shunt membrane is simulated using FEM tool and maximum displacement of 3 µm is noted at 5.9 µN for series and 2.5 µm is noted at 7.2 µN for shunt which are less than the maximum forces that the membranes can withstand (Figs. 11, 12).

Fig. 11
figure 11

Stress distribution of capacitive/shunt switch membrane at 7.2 µN of force which displaces the membrane at a displacement of 2.5 µm

Fig. 12
figure 12

Stress distribution of ohmic/series switch membrane at 5.9 µN of force which displaces the membrane at a displacement of 3 µm

The maximum and minimum stress distribution on the two membranes are observed by applying maximum force at which the membranes are displaced through the gap is observed as 7.34E-15 MPa and 94.6 MPa for series membrane and 62.1 and 0.21 MPa for shunt membrane respectively. The series membrane completes the connection when it displaced at 3 µm and shunt membrane breaks the connection when it displaced 2.5 µm comes in contact with the dielectric layer.

4.2 Electromechanical analysis

4.2.1 Spring constant

The spring constant of the two switches is difficult to calculate with the analytical method because of the complicated structure of non-uniform meanders (Ravirala et al. 2018; Singh 2015). Hence the force is applied and vertical deflection is noted to determine the spring constant of both the membranes.

According to the Newton’s Second law mass-spring-Damper system (Marcelli et al. 2010), the mechanical suspension of the spring is directly proportional to the force applied.

$$ F\infty \delta $$
(1)

where F is applied force and is the deflection of the proof mass.

$$ F = k\delta $$
(2)
$$ K = F/\delta $$
(3)

The two membranes are simulated by applying maximum holding force of the membranes and corresponding vertical deflection is noted and ratio of those values gives the spring constant of the membranes as mentioned in equation (Figs. 13, 14).

Fig. 13
figure 13

Force vs. displacement showing the spring constant of series switch membrane

Fig. 14
figure 14

Force vs. displacement showing the spring constant of shunt switch membrane

It is observed that the simulated spring constant for the series membrane is 2.17 N/m and for the shunt membrane is 3.02 N/m with the force–displacement ratio. The spring constant for the two membrane plays a key role for low pull-in voltage.

4.2.2 Pull-in voltage

The pull-in voltage of the two membranes are calculated analytically with the Eq. 4 (Rao 2017)

$$ V_{p} = \sqrt {\frac{{8kg_{0}^{3} }}{{27\varepsilon_{0} A}}} $$
(4)

where k is the spring constant (k = 2.17 for series and k = 3.02 for shunt), \( \varepsilon_{0} \) = 8.85 × 10−12 F/m is the permittivity of the free space and A is the overlapping area of electrodes.

It is very much important to maintain the membranes at same pull-in voltage for the operation of switch. To maintain same pull-in voltage the overlapping area of series is taken as 4000 µm2 and shunt is 3000 µm2. The two membranes are electrostatically actuated by applying the voltage and the electrostatic force generated is responsible for the vertical deflection of the membrane (Figs. 15, 16).

Fig. 15
figure 15

Ohmic/series switch membrane at pull-in voltage of 23.2 V showing a displacement of 2.91 µm

Fig. 16
figure 16

Capacitive/shunt switch membrane at pull-in voltage of 23.5 V showing a displacement of 2.39 µm

Analytically the calculated pull-in voltage for series and shunt is 22.5 V and the simulated results for series is 23.2 and for shunt is 23.5 which are close to the analytical values (Fig. 17).

Fig. 17
figure 17

Deflection analysis of both series and shunt membranes with an applied voltage

The pull-in voltage is analyzed by plotting the curve between actuation voltage and vertical deflection. The above graph illustrates the series membrane displaces the gap height 3 µm and shunt membrane displace the gap height 2.5 µm at the same voltage (Vp = 23 V).

4.2.3 Up state (Cu) and down state (Cd) capacitance

Initially the device is in OFF state means the shunt membrane is actuated with voltage and is in down state and series membrane is in Up state. When the actuation voltage is applied to series membrane the membrane gets deflected to down state and at the same time the shunt membrane is not supplied with actuation voltage hence it is in upstate allowing the RF signal to pass.

The up state capacitance of the two membranes is calculated by the Eq. 5 (Lakshmi Narayana et al. 2017a, b)

$$ C_{up} = \frac{{\left( {\varepsilon_{0} \, A} \right)}}{{\left( {g_{0} + \frac{{t_{d} }}{{\varepsilon_{r} }}} \right)}} $$
(5)

where, \( \varepsilon_{0} \) is the permittivity of the free space, \( \varepsilon_{r} \) is the relative permittivity of the electrode material, A is the overlapping area between electrodes, g0 is the gap between the electrodes and td is the dielectric layer thickness electrodes.

The upstate capacitance for the shunt and series membrane is calculated as 0.24 fF. The downstate capacitance can be calculated by the Eq. 6 (Fig. 18).

Fig. 18
figure 18

Capacitance analysis of both series and shunt membranes with an applied voltage showing ON and OFF states of the switch

$$ C_{d} = \frac{{\varepsilon_{0} \varepsilon_{r} A}}{{t_{d} }} $$
(6)

The downstate capacitance of both series and shunt membrane is 14.2 fF. Hence the Cd/Cu ratio gives the capacitance sensitivity of the device and is 55.92 therefore the simulation results of capacitance of both series and shunt membrane need to be in the range of Cu–Cd (Fig. 18).

The two membranes are simulated by applying electro-mechanics physics using comsol tool and observed the upstate capacitance of 0.24 fF and downstate capacitance of 14.2 fF and Cd/Cu ratio is 59.16.

4.3 Modal frequency analysis

Eigen frequency analysis is carried out using FEM tool conveys the information of the first six vibrating modes of the membranes are presented in the Fig. 19.

Fig. 19
figure 19

Vibrating modes of both shunt and series membrane of the switch

The modal frequencies is obtained by solving the Lagrangian of the system and then applying the canonical equation. The motion of the membranes under harmonic force is considered for obtaining the equation, by modelling the second order mass-spring damper system the equation of the harmonic motion of the membrane (Singh 2015) is

$$ M_{ef} \frac{{d^{2} x}}{{dt^{2} }} + \gamma_{ef} \frac{dx}{dt} + kx = F_{E} , $$
(7)

where Mef is the effective mass of the membrane, γef is the effective damping coefficient of the dielectric material, FE is the electrostatic force and k is the overall spring constant (Table 2).

Table 2 Modal frequencies of the switch

The Eigen frequency analysis is carried out using COMSOL multiphysics tool to obtain the resonant frequencies of the switches. Amongst the six vibrating modes the vertical deflection mode (mode #1) is suitable for the proposed device. It is observed that the resonant frequency of the series and shunt membrane is 21.615 and 25.490 kHz, respectively. The other five vibrating modes gives uneven rise and fall of the capacitance. Hence the fluctuations in the capacitance will gives rise to the degradation of transmitting signal with the effect of noise.

4.4 Resonant frequency analysis

The resonant frequency of the switch is depends on the spring constant of the meanders and proofmass of the actuating Beam. Chakraborty et al. (2014) presented a formula to calculate the mechanical vibration of the beam has been written in Eq. (8).

$$ f_{0} = \frac{1}{2\pi }\sqrt {\frac{k}{m}} $$
(8)

where m is modal mass of the beam = 0.35 × (L × w × t) × d. The d is density of gold = 3.55 × 10−13 kg/m3, m = 118 × 10−6 kg.

Here the proofmass and the spring constant is fixed. Hence any variation in the structure of the meanders will vary the spring constant which results in variation of frequency. The calculation of values of the frequency of shunt and series switch on the fundamental mode obtained are 21.6 and 25.49 kHz.

4.5 Switching time analysis

Switching speed of the device is considered as the utmost important factor for the performance of the device. The switching speed is also a major factor that influence the working of switch for transmit/receive applications. The switching time is calculated by the Eq. 9 (Lakshmi Narayana et al. 2017a, b)

$$ T_{s} = \frac{{3.67V_{p} }}{{\omega_{0} \, V_{s} }} $$
(9)

where Vp is the pull-in voltage, ω0 is the resonant frequency, Vs is the input source voltage.

The resonant frequency is determined by the spring constant and effective mass of the membrane (Singh 2015) (Fig. 20),

Fig. 20
figure 20

Source voltage vs. switching time

$$ \omega_{0} = \sqrt {\frac{k}{m}} . $$
(10)

The switching response of the series membrane is 24.8 µs and shunt membrane is 21.8 µs hence overall switching time of the device is observed as 23.8 ns which is average of both switching times.

4.6 RF performance analysis

The performance of the switch in GHz frequency range is to be analyzed to use in RF applications. The scattering losses is analyzed by using HFSS tool.

The device contains reflection loss or return loss and insertion loss in ON state. The device is on when the shunt membrane is in upstate and series membrane is in downstate and offers very low insertion and return losses during transmission of the RF signal. During ON condition of the device return loss (S11) is the measure of impedance matching of the device (Muldavin and Rebeiz 2000a, b) and given as

$$ S_{11} = \frac{{ - j\omega_{0} C_{u} Z_{0} }}{{2 + j\omega C_{u} Z_{0} }} $$
(11)

and for S11 ≤ − 10 dB.

$$ \left| {S_{11} } \right|^{2} \simeq \frac{{\omega^{2} C_{u}^{2} Z_{0}^{2} }}{4} $$
(12)

where ω is resonant frequency, Cu is the upstate capacitance and Z0 is the input impedance and is taken as 50 \( \varOmega \) (Fig. 21).

Fig. 21
figure 21

Reflection loss of proposed switch at 5–40 GHz range during ON state

The return loss of − 59.8 dB is observed in the frequency of 5 GHz and − 24.5 dB at 35 GHz.

Insertion loss is the loss of signal power resulting from the insertion of a device in a transmission line and is usually expressed in decibels (dB). Insertion loss is the loss occurs during the transmission of the RF signal. The insertion loss depend mainly on the dimensions and material of the substrate, dielectric layers and the bridge (Fig. 22).

Fig. 22
figure 22

Insertion loss of proposed switch at 5–40 GHz range during ON state

By taking TiO2 (K = 100) as a substrate dielectric the insertion loss reduced and is − 0.1 dB at 5–40 GHz frequency range.

The three types of switches which are single series, single shunt and series–shunt configuration switches are simulated using HFSS tool to obtain comparison of isolation loss of switches. The isolation loss of single series and shunt switches are observed as − 54 dB for single series switch at 21 GHz and − 61 dB for shunt switch at 25 GHz (Figs. 23, 24).

Fig. 23
figure 23

Isolation loss of series switch at 1–50 GHz range during OFF state

Fig. 24
figure 24

Isolation loss of shunt switch at 1–50 GHz range during OFF state

Isolation loss of the device occurs when both the switches are in OFF state i.e., the series membrane in upstate and shunt membrane is in downstate. Hence the Cu of the series membrane and Cd of the shunt membrane plays major role in attenuating the signal (Fig. 25).

Fig. 25
figure 25

Isolation loss of proposed series–shunt switch at 5–40 GHz range during OFF state

The isolation of 82.47 dB is observed at 26 GHz frequency. The switch exhibits high isolation of 82.47 dB around 26 GHz and having a bandwidth of 22 GHz typically in the range of Satellite communication applications (Table 3).

Table 3 Comparison of proposed switch results with existing switches

5 Conclusion

The RF MEMS switch is designed to achieve high isolation in frequency band used for satellite communication by incorporating the series and shunt membranes on the same coplanar waveguide. The two membranes are designed to work on same actuation voltage for better performance of the switch. Low pull-in voltage is achieved by considering the meandering structured springs for the series membrane and crab-leg structured springs for the shunt membrane such that both the membranes have acceptable stress levels. The Electromechanical parameters such as pull-in voltage, spring constant, Cd, Cu, modal frequencies and RF performance characteristics has been analyzed in this paper. The proposed RF MEMS switch can be used in antennas of high frequency satellite communication applications where the high isolation is a major concern. The proposed series–shunt switch shows excellent isolation compared to the switches having series and shunt membranes individually with a wider bandwidth for satellite communications applications.