Scholarly article on topic 'Simulating Nonlinearity in MEMS Resonators by a Charge Controlled Capacitor'

Simulating Nonlinearity in MEMS Resonators by a Charge Controlled Capacitor Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia Engineering
OECD Field of science
Keywords
{MEMS / nonlinearity / "equivalent circuit" / resonator}

Abstract of research paper on Materials engineering, author of scientific article — Haoshen Zhu, Joshua E.-Y. Lee

Abstract In this paper, a novel circuit formulation allowing co-simulation of nonlinearity in MEMS resonators and interface circuits in a generic circuit simulation platform is reported for the first time. A charge controlled capacitor is used to model the nonlinear restoring force in the mechanical resonator. The concept has been applied to the case of a longitudinal beam resonator, illustrating the effect of feedthrough in the limit of nonlinear behavior characterized by features such as spring softening and hysteresis.

Academic research paper on topic "Simulating Nonlinearity in MEMS Resonators by a Charge Controlled Capacitor"

Available online at www.sciencedirect.com

Procedía Engineering

ELSEVIER

ProcediaEngineering 25 (2011) 403 - 406

www.elsevier.com/locate/procedia

Proc. Eurosacsors XXV, September 4-7, 2(11, Athacs, Graaca

Simulating Noclicaprity ic MEMS Rasocators by p Charga

Coctrollad Capacitor

Haoshac ZhuP, Joshua E.-Y. LaaP*

In this paper, a novel circuit formulation allowing co-simulation of nonlinearity in MEMS resonators and interface circuits in a generic circuit simulation platform is reported for the first time. A charge controlled capacitor is used to model the nonlinear restoring force in the mechanical resonator. The concept has been applied to the case of a longitudinal beam resonator, illustrating the effect of feedthrough in the limit of nonlinear behavior characterized by features such as spring softening and hysteresis.

© 2011 Published by Elsevier Ltd.

Keywords: MEMS; nonlinearity; equivalent circuit; resonator

1. Introduction

MEMS resonators have been exploited in both sensory and timing applications. Depending on the geometry, material and method of transduction, these vibratory devices exhibit nonlinear behavior in the limit of large displacement, thereby placing a limit on performance. As such, nonlinear analysis is of great importance particularly for high-Q MEMS resonators, where the displacement is largely constrained by nonlinearity. From this viewpoint, a simulation platform capable of modeling the electromechanical characteristics of a resonator under nonlinear conditions will be much welcomed. While elementary circuit blocks which map displacement to voltage have been proposed [1], this requires a custom software since the form of the circuit defers the conventional RLC used to represent a mass-damper-spring system. In this paper, we present an alternative convention where displacement is represented by the charge on a capacitor, which is further modified in a circuit simulation suite to describe nonlinear behavior. While the

* Corraspocdicg author. Tal.: + 852- 2788- 9897; fax: +852- 2788-7791. E-mail address: joshup.laa@cityu.adu.hk.

department of Electronic Engineering, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong

Abstract

1877-7058 © 2011 Published by Elsevier Ltd. doi:10.1016/j.proeng.2011.12.100

work here has been achieved in ADS, the concepts apply to all commonly used simulation platforms. The results here therefore demonstrate the possibility of performing a full system level simulation that includes both MEMS and interface circuits where nonlinearities can be accounted for.

2. Formulation for mechanical to electrical equivalence

The motion of a generic MEMS resonator can be modeled as a linear mass-damper-spring system:

mx + ]d + f (x) = F (t) (1)

where m is the mass, y is the damping coefficient, f(x) and F(t) represent the restoring force of the spring and a sinusoidal driving force respectively. In the case of a linear system, f(x) = kx. While it has been shown that (1) can be modeled by a series RLC using the capacitor voltage to denote displacement [2], this formulation does not yield a unified agreement between all electrical and mechanical variables. In contrast, using charge (Q) to represent displacement provides such a unified agreement as summarized in the Table 1:

Table 1. Formulation for mechanical to electrical equivalence

Constants Variables

Mechanical Mass (d) Damping coefficient (7) Spring constant (k) Displacement (x) Excitation force (F)

Electrical Inductance (L) Resistance (R) Elastance (1/C) Charge (Q) Source voltage (U)

In the limit of nonlinearity, the Helmholtz-Duffing equation considers the higher order spring constants:

f (x) = k1x + k2 x2 + k3 x3 (2)

where k2 and k3 are quadratic and cubic spring constants. Based on the formulation using Q to denote x, we can then construct a charge-controlled capacitor [3] to describe the polynomial function given by:

U = QCi + Q2/C2 + Q3/Q (3)

where C2 and C3 are second-order and third-order capacitance coefficients respectively. Since (3) has a similar form to (2), we may use the total charge Q on the capacitor to indicate the displacement of the resonator. Therefore, simulating the proposed RLC circuit will provide the exact value of displacement which is equivalent to the total charge stored on the capacitor.

Ultimately, this motion is captured as a current due to the modulation of the capacitive gap when a DC bias voltage is applied across the nominal gap (d). This effect can be modeled by a variable capacitor:

u = Q/C(x) = jIm/(s0 A/(d - x)) (4)

where C(x) is the capacitance of the transducer under different displacements and Q is the total charge stored on the transducer. Furthermore, the proposed variable capacitor also allows for the presence of high-order harmonics stemming from the feedthrough to be captured, which would otherwise be seen as merely amplitude changes at the fundamental tone as the excitation voltage is increased.

3. Case Study

The reported parameters for a nonlinear electrostatic longitudinal beam resonator by [4] were adopted in our model for validation. The resonator's significant quadratic mechanical spring constant and strong

spring softening effect makes for an excellent case study to apply our circuit model, which is shown in Fig. 1. The lumped parameters of the longitudinal bulk mode beam resonator are listed in Table 2.

Fig. 1. Schematic of the equivalent circuit

Table 2. Lumped parameters of the longitudinal beam resonator of gap d = 1^m

Effective Mass Damping Coefficient Linear Spring Quadratic Spring Cubic Spring Transduction

(m) kg (y) №s/m Constant (kg) N/m Constant (fe) N/m2 Constant (fe) N/m3 Capacitor (Q fF

16.8 x 10-12 6.88 x 10-9 91.4 x 103 -1.764 x 109 -4.23 x 1013 0.708

The harmonic balance (HB) method was used to study the nonlinear features of the resonator in the frequency domain. Firstly, only electrostatic nonlinear effects were considered by setting the higher order spring constants to zero. Fig. 2(a) shows the simulated displacement for increasing AC input, illustrating skewing in the resonant peak towards lower frequencies due to spring softening. Inclusion of higher order spring constants is shown to intensify the spring softening effects when we compare Fig. 2(b) to Fig. 2(a); where the maximum frequency shift increases from 40Hz to 250Hz for the same AC excitation inputs. Moreover, hysteresis can be observed as the AC input excitation increases as highlighted by the red dotted lines. The resonator becomes more nonlinear and easier to reach the hysteresis regime by adding the mechanical nonlinearity.

(a) (b)

Fig. 2. Displacement with increasing AC input with higher-order spring constants (a) excluded; (b) included

Finally, we simulate the full electrical characteristics of the MEMS resonator where the output has both the motional as well as feedthrough currents as observed in actual measurements. Apart from the direct coupling through the capacitive transducer (i.e. variable capacitor), a parasitic coupling capacitor connecting the drive and sense ports was added to account for coupling through the substrate and bond

wires. Fig. 3(a) shows the magnitude of the full electrical transmission characteristic (S21) that has been embedded in feedthrough with the output connected to a 50Q termination load. The anti-resonance valley due to feedthrough can be clearly observed along with the increased skewing in the resonant peak towards lower frequencies as the AC excitation input is increased. Furthermore, the hysteresis becomes more clearly observable with the increase of the AC driving voltage. By re-routing the AC input to feed only into the mechanical resonator (variable capacitor thus only sees a DC bias), the S21 transmission curves containing only the motional current are obtained are shown in Fig. 3(b).

(a) (b)

Fig. 3. Magnitude of the S21 transmission with feedthrough (a) included; (b) excluded

4. Conclusion

We have used a charge controlled nonlinear capacitor model with higher order capacitance coefficients to capture the effects of both mechanical and electrostatic nonlinearity in MEMS resonators. A variable capacitor model was implemented to capture the modulating effect of the transducer gap. We show that the whole equivalent circuit can well predict the nonlinear effects for micro-resonators, demonstrating the possibility for system-level simulation of non-linear MEMS resonators together with interface circuits. In short, we demonstrate this capability for implementation in commonly used circuit simulation platforms.

Acknowledgement

Simulation support in ADS was provided through the State Key Laboratory of Millimeter Waves. This work was supported by a Strategic Research Grant from the City University of Hong Kong (project #7008088).

References

[1] Veijola T, Mattila T. Modeling of nonlinear micromechanical resonators and their simulation with the harmonic-balance method. Ietruentiaenl Jauuenl af RF nedMicuawnvr Cadputru-Aidrd Ledierruied 2001;11: 310-21.

[2] Zhu Y, Lee J, Seshia A. System-level simulation of a micromachined electrometer using a time-domain variable capacitor circuit model. Jauuenl af Micuadrchneics nedMicuaregierruied 2007;17: 1059-65.

[3] Oksasoglu A, Huelsman L. Interaction of Capacitive and Resistive Nonlinearities in Chua's Circuit. Aenlag Ietrguntrd Ciucuits ned Sigenl huacrssieg 1997;12: 239-53.

[4] Mattila T, Kiihamaki J, Lamminmaki T, Jaakkola O, Rantakari P, Oja A et al. A 12 MHz micromechanical bulk acoustic mode oscillator. Sresaus nedActuntaus A: hhysicnl 2002;101: 1-9.