MEMS Electrostatic Acoustic Pixel

Arpys Arevalo*1 , David Conchouso1 , David Castro1 , and Ian G. Foulds1,3

1 Computer, Electrical and Mathematical Sciences and Engineering (CEMSE), King Abdullah University of
Science and Technology (KAUST), 2 The University of British Columbia (UBC), School of Engineering,
Okanagan Campus.
*Corresponding author: 4700 KAUST, Kingdom of Saudi Arabia, [email protected]

Abstract: This paper reports the simulation of an element) where they demonstrated the sound re-
hexagonal membrane structure using COMSOL construction. Due to their fabrication methods
Multiphysics 5.0. The structure is a 5 µm thick each transducer in the array has a fixed electrode
polyimide layer with an integrated metal layer with a dome shape membrane. Therefore, there is
on top, to apply a bias voltage. The hexagonal an asymmetry in their system while actuating the
membrane is separated by a 3 µm air gap and membranes.
5 µm thick polyimide structural layer from the In previous work, we have tackled this problem
bottom electrode and a 3 µm and 5 µm thick by using a piezoelectric layer [7–9]. Using the lat-
polyimide structural layer from the top electrode. ter method for the driving mechanism help us to
The AC/DC Module was used to extract the avoid the asymmetry in the system, but adds some
capacitance and pull-in voltage needed to displace complexity to the fabrication process. Our work
the membrane toward the active electrode. A differs from other reports [10–14], specifically in
modal analysis was performed using the Struc- the dimensions of the actuator, the combination of
tural Mechanics Module to extract the structure’s materials used, the micro fabrications process and
resonance frequency and frequency modes. the method of sound reconstruction.
In this work we propose an electrostatic ap-
Keywords: MEMS, electrostatic speaker, proach using Polyimide as the structural layer.
digital sound reconstruction, Acoustic Pixel, Polyimide is a very attractive polymer for MEMS
polyimide. fabrication due to its low coefficient of thermal ex-
pansion, low film stress, lower cost than metals
and semiconductors and high temperature stability
1. Introduction compared to other polymers [15–17]. Polyimide
has been previously used in the microelectronics
The growth of the electronics industry demands industry for module packaging, flexible circuits
better components for improved electronic sys- and as a dielectric for multi-level interconnection
tems [1, 2]. Such components need to be more technology [18, 19]. Polyimide has demonstrated
advanced in order to keep up with the evolution great performance in Micro Electro Mechanical
of the digital era. The loudspeaker design mech- Systems (MEMS) devices [15, 20–34] and in the
anism has not been changed for almost a century microelectronics industry. The polymer can han-
[3–5]. In an era were almost every component is dle temperatures of up to 350◦ C and can be easily
digital, the loudspeaker driver design still is the processed and integrated with metal layers to de-
last analog component that needs to evolve to its velop devices with low complexity.
digital form, to achieve a true digital audio repro-
duction cycle.
Efforts to develop a direct digital method of 2. Computational Methods
sound reconstruction has been reported elsewhere
[6, 7]. In [4], authors reported a micro-speaker in COMSOL Multiphysics provides the Electro-
a single chip using CMOS-MEMS membrane ar- static Interface, which is available for 3D, 2D in-
rays and described their method behind the dig- plane and 2D axisymmetric components. In our
ital sound reconstruction concept. They fabri- particular application we have a capacitor which
cated a 3-bit electrostatic array (seven transducer will use relatively high voltage (up to 150 Volts).

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble

The electrostatic equations are not to be taken The equation describes the electrostatic field in
literally as ”statics”, but as the observation or dielectric materials, the physical constant ε0 is the
time scale at which the applied excitation changes permittivity of vacuum with units [F/m], P is the
are in comparison to the charge relaxation time, electric polarization vector in [C/m2 ], and ρ is the
and that the electromagnetic wavelength and skin space charge density given in [C/m3 ].
depth are very large compared to the size of the For models in 2D, the interface assumes a sym-
domain of interest [35]. metry where the electric potential varies only in
For our device, we need to use the quasi-static the x and y directions and is constant in the z di-
electric fields and currents that are included in rection. Which implies that the electric field E is
the MEMS module, together with the AC/DC tangential to the xy-plane. The same equation is
Module, which do not include the wave propa- solved in the case of a 3D model. The interface
gation effects. The physics interfaces takes only sovles the following equation where d is the thic-
the scalar electric potential, which can be inter- ness int eh z direction:
preted in terms of the charge relaxation process.
The three equations used for this physic are: The −∇ · (ε0 ∇V − P) = ρ (6)
Ohm’s Law, the equation of continuity and the The axisymmetric version of the physics inter-
Gauss’ law. COMSOL combines this equation face considers the situation where the fields and
and uses the following differential equation for the geometry are axially symmetric. For this case the
space charge density in a homogeneous medium: electric potential is constant in the φ direction, im-
plying that the electric field is tangential to the rz-
δρ σ
+ ρ=0 (1) plane [35].
δt ε
with solution:
3. Design and Simulation Setup
−t
ρ(t) = ρ0 e τ (2) The main membrane of our device can be di-
where vided in three sections: outer hexagonal ring, teth-
ers and hexagonal membrane (see Fig. 1).
ε
τ= (3)
σ
which is the charge relaxation time. When us-
ing a good conductor material such as gold, τ is of
the order of 10− 19s whereas for a good insulator
like silicon oxide, its of the order of 103 s. It is the
relation between the external time scale and the
charge relaxation time that determines the physics
interface and study that we will use.

2.1 Electrostatic Equations

Under static condition the potential, V, is de-
fined as the following relationship:

E = −∇V (4)
Figure 1: Top view of the simulated membrane,
When combined with the constitutive relation- showing its three different sections.
ship D = εE + P bettwen the electric displacement
D and the electric field E, the Gauss’ law is repre- The device was evaluated with several differ-
sented as: ent tether designs and the present work is done us-
ing the final chosen design for fabrication. The de-
−∇ · (ε0 ∇V − P) = ρ (5) sign shows 5 tethers in each side of the hexagonal

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble

membrane. The standard structure has the follow-
Table 1: Materials Properties
ing dimensions: 250 µm membrane hexagon side,
Property Polyimide Gold
the hexagon was inscribed in a 500 µm diameter
Relative permittiv- 2.9 6.9
circumference and tether have a width of 8 µm for ity (εr )
each of them. Young’s modulus 3.1e9 [Pa] 70e9 [Pa]
The structure can be fabricated using two (E)
structural layers and two sacrificial layers. The Poisson’s ratio (ν) 0.34 0.44
structural layers are made of polyimide and have Density (ρ) 1300 [kg/m3 ] 19300[kg/m3 ]
a thickness of 5 µm. To be able to attract and re-
pel the membrane we need a set of electrodes. In
our simulations we use the bottom electrode made The Electromechanics physics module was
of gold, because of its good conductivity, which is setup with the following constraints: Fixed con-
located right on the silicon substrate. Also a mid- straint for all the six outer sides (faces bound-
dle electrode which is on top of the membrane and aries) of the full structure, Bottom Electrode as
a top electrode that is all the way to the top of the the ground and Middle Electrode as a Terminal.
structure. Fig. 2 shows a conceptual view of an The setup will allow the interaction between the
individual membrane. electrodes, and the capacitance will be calculated
by the software.
An interesting feature of our design is that there
will not be an electric short when pull-in occurs,
because all the electrodes are completely isolated
from each other with structural layer.
To see the behavior of the membrane we used a
Stationary Study with an auxiliary sweep to apply
voltages between a pair of electrodes ranging from
10V - 150V in steps of 10V. The boundary that
was set to be a terminal was given the declared
parameter ”Vin”.

Figure 2: Conceptual view of the simulated de- 4. Results

vice. The image shows an exploded view on the
right, and an assembled view on the left. The simulation results give us an insight of the de-
formation of the membrane. It is known that the
To create the 3D model in COMSOL, we first pull-in voltage when the system is unstable, which
exported the 2D layout from Tanner L-edit soft- happens at approximately 1/3 of the distance be-
ware, which is the tool we use to design our tween the electrodes. Therefore, the pull-in will
devices for micro-fabrication. The CAD import occur when the membrane moves approximately
module was used, and the correct scale was set to 2.6 µm towards the active electrode. In Fig. 3 a
import the DXF file into COMSOL environment. graph of the simulated displacement vs the applied
The import was done in two different work-planes voltage is shown. Fig. 5 shows the graph of the
to be able to extrude the needed features. The fi- capacitance between the electrodes vs the applied
nal component was set to form composite faces to voltage.
eliminate unnecessary features and a union opera- From these results we were able to deduct that
tion. the pull-in voltage is between 140V and 150V, ap-
The selected materials for the electrodes was plying more than this voltage wont let the simu-
gold, as depicted in Fig. 2 in yellow color. The lation to converge. Fig. shows the result of the
structural layer was set to be polyimide, shown in displacement in the 3D model.
red color in Fig. 2. Also, all the gaps were set to be The resonance frequency and mode frequencies
air. Table 1 contains the material properties used were calculated using the Solid Mechanics Mod-
in the simulation. ule to study the behavior of the structure. An

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble

Figure 3: Total displacement on the ”z-axis” vs Figure 4: Capacitance vs applied voltage.
the applied voltage.

Eigenfrequency study was setup to find the first

6 modes of the structure, shown in Fig. 6.
From this simulation we can see that the mode
of interest is the first one at 9.4175 kHz, as this will
displace the air in a uniform mode with only one
deformation node. Since the transducer will be ac- Figure 5: (Top Left) Isometric view of the sim-
tuated at an expected sample frequency of 40 kHz, ulation results for displacement, (Top Right) Top
the closest mode is the sixth at 3.9267 kHaz. view of deformed structure, (Bottom) Side view
Mode 6, has one radial node and one central node, of the deformed structure at 150V.
but it will not have an impact in the performance
of the membrane because it will be out of the
5. Conclusions
range of the frequency.
If the membrane will be actuated at 40 kHz, this The proposed membrane design was simulated
means that the input signal will behave as a pulse with the intended operation voltages for the real
with a width of 25 µm. Therefore, we performed device. The results shows that the membrane is
a new simulation with at Time Dependent Study suitable for the intended acoustic transducer ele-
from t = 0 to t = 625 µs in steps of 25 µs to observe ment for the final transducer array. The membrane
the response time of the structure to a 150V con- geometry can be adjusted to change the structure
stant electric potential applied to one of the elec- resonance frequency, so that the element has the
trode. Fig. 7, shows the response time of the mem- optimal acoustic response. The next simulation
brane. From this graph we can see that it takes the steps will be the acoustic response. Now that we
membrane approximately 125 µs to reach the max- have found the total displacement of the structure
imum displacement of about 1.5 µm. Also, it can at an applied voltage, we can simulate the dis-
be seen that it the membrane reaches a stable posi- placement and calculate the sound pressure gen-
tion in approximately 500 µs at 1 µm displacement erated by this change.
from its original position. Nevertheless, the pulses Full arrays have already been designed and fab-
will only be 25 µs long and this means that the ricated. The processed chips are diced from a four
structure will only displace approximately 0.5 µm. inch wafer using our in house dicing method [36].

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble

 Figure 7: Response time of the structure to an in-
put signal of 150V

5. Baron John William Strutt Rayleigh and

Robert Bruce Lindsay. The theory of sound
(1945).
Figure 6: Frequency modes of the simulated struc-
tures 6. B.M Diamond, J.J. Jr Neumann, and K.J
Gabriel. Digital sound reconstruction using
arrays of CMOS-MEMS microspeakers. In
The chips are currently being tested and the next Micro Electro Mechanical Systems, 2002. The
steps include the validation of the presented model Fifteenth IEEE International Conference on,
and the experimental results. pages 292–295, (2002).

7. A Arevalo, D Conchouso, D Castro, N Jaber,

6. References M I Younis, and I G Foulds. Towards a Digital
Sound Reconstruction MEMS Device: Char-
1. J P Rojas, A Arevalo, I G Foulds, and M M acterization of a Single PZT Based Piezo-
Hussain. Design and characterization of ultra- electric Actuator. In 10th IEEE Interna-
stretchable monolithic silicon fabric. Applied tional Conference on Nano/Micro Engineered
Physics Letters, 105(15):154101, (2014). and Molecular Systems NEMS2015, Xia’an
(2015).
2. J P Rojas, Hussain A M, A Arevalo, I G
8. A Arevalo and I G Foulds. MEMS Acous-
Foulds, Sevilla GAT, JM Nassar, and M M
tic Pixel. 2014 COMSOL Conference, Cam-
Hussain. Transformational electronics are
bridge, England, (2014).
now reconfiguring. SPIE Defense+ Security,
9467:946709–946709–7, (2015). 9. A Arevalo and I G Foulds. Parametric Study
of Polyimide-Lead Zirconate Titanate Thin
3. Carl B Boyer and Uta C Merzbach. A History Film Cantilevers for Transducer Applications.
of Mathematics. Wiley (2011). In 2013 COMSOL Conference, Rotterdam,
Holland, (2013).
4. Brett M Diamond, John J Neumann, Jr, and
Kaigham J Gabriel. Digital sound reconstruc- 10. Zhihong Wang, Jianmin Miao, and Chee Wee
tion using arrays of CMOS-MEMS micros- Tan. Acoustic transducers with a perforated
peakers. In Micro Electro Mechanical Sys- damping backplate based on PZT/silicon
tems, 2002. The Fifteenth IEEE International wafer bonding technique. Sensors and Actu-
Conference on, pages 292–295, (2002). ators A: Physical, 149(2):277–283 (2009).

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble

11. Youngki Choe, S Chen, and Eun Sok Kim. 20. S.Y. Xiao, L.F. Che, X.X. Li, and Y.L. Wang.
High fidelity loud microspeaker based on A novel fabrication process of {MEMS} de-
PZT bimorph diaphragm. In NSTI-Nanotech vices on polyimide flexible substrates. Mi-
2010, pages 316–319, (2010). croelectronic Engineering, 85(2):452 – 457,
(2008).
12. Sang-Soo Je and Junseok Chae. A Compact,
Low-Power, and Electromagnetically Actu- 21. Steve Tung, Scott R Witherspoon, Larry A
ated Microspeaker for Hearing Aids. Electron Roe, Al Silano, David P Maynard, and
Device Letters, IEEE, 29(8):856–858, (2008). Ned Ferraro. A mems-based flexible sen-
sor and actuator system for space inflatable
13. P Valousek. A digital loudspeaker: ex- structures. Smart Materials and Structures,
perimental construction. Acta polytechnica, 10(6):1230, (2001).
46:40–42, (2006).
22. J. Courbat, M.D. Canonica, D. Briand, N.F.
14. Nicolas-Alexander Tatlas and John N Mour- de Rooij, Damien Teyssieux, Laurent Thiery,
jopoulos. Digital Loudspeaker Arrays driven and Bernard Cretin. Thermal simulation and
by 1-bit signals. In Audio Engineering Society characterization for the design of ultra-low
Convention 116, (2004). power micro-hotplates on flexible substrate.
In Sensors, 2008 IEEE, pages 74–77 (2008).
15. A Arevalo, E Byas, D Conchouso, D Cas-
tro, S Ilyas, and I G Foulds. A Ver- 23. A Arevalo, E Byas, and I G Foulds. µHeater
satile Multi-User Polyimide Surface Micro- on a Buckled Cantilever Plate for Gas Sensor
machining Process for MEMS Applications. Applications. 2012 COMSOL Conference,
In 10th IEEE International Conference on Milan, Italy, (2012).
Nano/Micro Engineered and Molecular Sys-
tems NEMS2015, Xia’an (2015). 24. A Arevalo, D Conchouso, and I G Fould.
Optimized Cantilever-to-Anchor Configura-
16. A A A Carreno, D Conchouso, A Zaher, tion of Buckled Cantilever Plate Structures
I Foulds, and J Kosel. Simulation of a for Transducer Applications. 2012 COMSOL
Low Frequency Z-Axis SU-8 Accelerome- Conference, Milan, Italy, (2012).
ter in CoventorWare and MEMS+. In Com-
puter Modelling and Simulation (UKSim), 25. A Arevalo, S Ilyas, D Conchouso, and I G
2013 UKSim 15th International Conference Foulds. Simulation of a Polyimide Based
on, pages 792–797, (2013). Micromirror. 2014 COMSOL Conference,
(2014).
17. D Conchouso, A Arevalo, D Castro,
E Rawashdeh, M Valencia, A Zaher, J Kosel, 26. A Arevalo and I G Foulds. Polyimide Ther-
and I Foulds. Simulation of SU-8 Frequency- mal Micro Actuator. 2014 COMSOL Confer-
Driven Scratch Drive Actuators. In Computer ence, Cambridge, England, (2014).
Modelling and Simulation (UKSim), 2013
UKSim 15th International Conference on, 27. S Ilyas, A Ramini, A Arevalo, and M I You-
pages 803–808, (2013). nis. An Experimental and Theoretical In-
vestigation of a Micromirror Under Mixed-
18. A Bruno Frazier. Recent applications of poly- Frequency Excitation. Journal of Microelec-
imide to micromachining technology. In- tromechanical Systems, (2015).
dustrial Electronics, IEEE Transactions on,
42(5):442–448, (1995). 28. D Conchouso, A Arevalo, D Castro, , and
I G Foulds. Out-of-plane Platforms with Bi-
19. A B Frazier. Uses of polyimide for micro- directional Thermal Bimorph Actuation for
machining applications. In Industrial Elec- Transducer Applications. In 10th IEEE In-
tronics, Control and Instrumentation, 1994. ternational Conference on Nano/Micro Engi-
IECON ’94., 20th International Conference neered and Molecular Systems NEMS2015,
on, pages 1483–1487, (1994). Xia’an (2015).

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble

29. Loı̈c Marnat, A Arevalo Carreno, D Conc-
houso, M Galicia Martinez, I Foulds, and Atif
Shamim. New Movable Plate for Efficient
Millimeter Wave Vertical on-Chip Antenna.
IEEE Transactions on Antennas and Propa-
gation, 61(4):1608–1615, (2013).
30. A Alfadhel and Arpys Arevalo. Three-Axis
Magnetic Field Induction Sensor Realized on
Buckled Cantilever Plate. . . . IEEE Transac-
tions on, (2013).

31. D Castro, A Arevalo, E Rawashdeh, and

N Dechev. Simulation of a Micro-Scale Out-
of-plane Compliant Mechanism. In 2014
COMSOL Cambridge, England, (2014).
32. A Arevalo, D Conchouso, D Castro, M Diaz,
and I G Foulds. Out-of-plane buckled can-
tilever microstructures with adjustable angu-
lar positions using thermal bimorph actuation
for transducer applications. Micro and Nano
Letters, (2015).
33. A Arevalo, D Conchouso, E Rawashdeh,
D Castro, and I G fould. Platform Isolation
Using Out-of-Plane Compliant Mechanisms.
In 2014 COMSOL Conference, Boston, USA
(2014).
34. N Jaber, A Ramini, A Carreno, and MI You-
nis. Higher Order Modes Excitation of Mi-
cro Clamped-Clamped Beams. NANOTECH
Dubai 2015, (2015).
35. COMOSL. Comsol multiphysics reference
manual. (2015).
36. Y Fan, A Arevalo, H Li, and I G Foulds. Low-
cost silicon wafer dicing using a craft cutter.
Microsystem Technologies, (2014).

Excerpt from the Proceedings of the 2015 COMSOL Conference in Grenoble