Stresa, Italy, 25-27 April 2007
SELECTION OF HIGH STRENGTH ENCAPSULANT FOR MEMS DEVICES
UNDERGOING HIGH PRESSURE PACKAGING
Azrul Azlan Hamzah, Yusnira Husaini, Burhanuddin Yeop Majlis, Ibrahim Ahmad
Institute of Microengineering and Nanoelectronics (IMEN)
Universiti Kebangsaan Malaysia
43600 Bangi, MALAYSIA
e-mail: azlan@vlsi.eng.ukm.my, yusni458@tganu.uitm.edu.my, burhan@vlsi.eng.ukm.my,
ibrahim@vlsi.eng.ukm.my
ABSTRACT
Deflection behavior of several encapsulant materials
under uniform pressure was studied to determine the best
encapsulant for MEMS device. Encapsulation is needed
to protect movable parts of MEMS devices during high
pressure transfer molded packaging process. The selected
encapsulant material has to have surface deflection of less
than 5 ?m under 100 atm vertical loading. Deflection
was simulated using CoventorWare ver.2005 software
and verified with calculation results obtained using shell
bending theory. Screening design was used to construct a
systematic approach for selecting the best encapsulant
material and thickness under uniform pressure up to 100
atm. Materials considered for this study were polyimide,
parylene C and carbon based epoxy resin. It was observed
that carbon based epoxy resin has deflection of less than 5
?m for all thickness and pressure variations. Parylene C is
acceptable and polyimide is unsuitable as high strength
encapsulant. Carbon based epoxy resin is considered the
best encapsulation material for MEMS under high
pressure packaging process due to its high strength.
1. INTRODUCTION
A major ongoing challenge in MEMS is to develop a
more standardized packaging like those in IC while
maintaining the integrity and functionality of the device.
MEMS packaging varies extensively depending on device
function, thus driving packaging cost high [1]. In an
effort to develop a standardized low cost packaging for
MEMS, device capping followed by glob top
encapsulation technique is increasingly gaining popularity
[2].
Many types of MEMS devices, especially those
that do not require interaction with outside ambient, could
function well in standardized encapsulation and glob top
packaging. In this process, movable parts of a MEMS
device are covered by a metal or silicon cap, leaving a
gap between the top surface of the movable parts and the
bottom surface of the cap. As a result, the cap allows
movable elements to move freely while protecting them.
The cap is usually fabricated by deposition technique,
thus it is not robust enough to endure subsequent high
pressure transfer molding process, as the molding
pressure could be as high as 100 atm [3]. Therefore, a
strong glob top encapsulation is needed to protect the cap,
so that the device could be packaged using standardized
IC transfer molding process. This paper investigates
several materials to select for the best encapsulant under
given pressure loading and package thickness constraints.
2. ENCAPSULATION FABRICATION PROCESS
A typical MEMS device consists of sensor or actuator
elements fabricated on silicon substrate. These elements
are usually movable and are very sensitive to damage by
chemical contamination, presence of micro dusts, as well
as physical touch. MEMS devices such as RF switches,
inductors, filters, and accelerometers do not require
interaction with outside ambient to function. Hence, a
complete isolation of the sensor or actuator elements
would increase device performance as well as its lifetime.
For this reason, encapsulation is a favorable method for
packaging these types of MEMS devices.
Using encapsulation technique, a cap, usually of
metal or silicon is deposited on top of MEMS movable
elements via deposition processes such as CVD or
sputtering [4]. A sacrificial layer, usually of silicon oxide
or photoresist material, is pre-deposited on the movable
elements to create a gap between the elements and the cap
structure. After cap deposition, the sacrificial layer is
removed, consequently releasing the movable elements
within the enclosed cap. The cap is then sealed by another
step of deposition, usually of oxide or metal, thus
completely isolating the movable elements from outside
ambient.
?EDA Publishing/DTIP 2007
ISBN: 978-2-35500-000-3
A. Hamzah, Y. Husaini, B. Y. Majlis, and I. Ahmad
Selection of High Strength Encapsulant for MEMS Devices
In order to strengthen the cap structure for the
subsequent transfer molding process, an encapsulation
layer needs to be added on top of the cap structure as
depicted in figure 1. A thin glob is dispensed on capped
MEMS device using standard glob top process, yielding
encapsulated device with uniform glob top after curing,
as depicted in figure 2 below.
gap
glob top encapsulation
substrate
cap
structure
seal layer
etch vents
MEMS movable
elements
bondpad
Figure 1. Schematic diagram of a glob top encapsulated
MEMS device.
(a)
(b)
Figure 2. SEM micrograph showing (a) an array of epoxy
resin encapsulated accelerometer devices and (b) a single
capped accelerometer device encapsulated with epoxy
resin. Note that the traces and bondpads are exposed,
while capped accelerometer fingers are enveloped
underneath the encapsulation.
3. ANALYTICAL APPROACH
A systematic screening approach is applied in encapsulant
selection process. The generalized selection process flow
is outlined in figure 3 below.
Figure 3. Flow of the encapsulant material selection
process.
Initially, dimensional requirements of the
encapsulation were determined. For our case, the glob top
encapsulation has to have a spherical dome shape with
thickness of less than 250 ?m (denoted by t in figure 4),
as the chip would be integrated into thin SMT package.
Dome shaped shell was chosen due to its ability to endure
Determine encapsulation design and
dimensions
Approximate governing equations for
deflection
Select candidate materials and determine
pressure loading conditions
Select factors and levels for Taguchi
experiment
Run CoventorWare simulation and
perform deflection calculation for
selected factors and levels
Compare simulation and calculated
deflection results
Perform reliability and sensitivity tests
on simulation results
500 ?m0
device
width
epoxy
glob top
encapsulation
900 ?m0
2400 ?m
encapsulation
bondpad traces
?EDA Publishing/DTIP 2007
ISBN: 978-2-35500-000-3
A. Hamzah, Y. Husaini, B. Y. Majlis, and I. Ahmad
Selection of High Strength Encapsulant for MEMS Devices
high external pressure applied, owing to compressive
stress distribution in meridional direction. The lateral
length of the glob-top encapsulant, denoted by 2b in
figure 4, should be approximately 2400 ?m, limited by
device width as shown in figure 2(b). Using the above
parameters, the values for shell radius a and vertical to
base angle ? could be determined by solving
simultaneous equations. The values for a and ? were
determined to be 3010 ?m and 23.5? respectively.
Figure 4. Schematic diagram showing the cross section
and the parameters involved in encapsulation structure
governing equations.
It is derived from the above parameters that the
ratio of shell thickness t over curvature a is approximately
0.08. Since this ratio is small, it follows that the radial
deflection w of the encapsulation structure when loaded
with uniform force P could be approximated using shell
bending theory. This theory could be well applied to
approximate deflection of thin semi-spherical structure in
which the radial deflection is much smaller compared to
shell thickness [5]. By means of this theory, radial
deflection of a thin shell under uniform loading P, as
depicted in figure 4, could be formulated as follows [5]:
?
?
?
?
?
?
?
?
?
+
+
?= ?
?
?
?? cos
cos1
1
cot
2
Et
Pa
w
(1)
where E and ? are the Young?s modulus and Poisson ratio
of the encapsulant material, t is the thickness of the shell,
? is the angle from vertical at which the deflection is
considered, and ? is the meridional deflection, which for a
spherical shape can be represented as follows [5]:
()
?
?
?
??
?
? sin
cos1
cos1
ln
cos1
1
cos1
11
2
?
?
?
?
?
?
?
?
?
?
?
?
?
?
+
+
+
+
?
+
+
=
Et
Pa
(2)
The encapsulant material selected has to be able
to withstand up to 100 atm external pressure without
excessive deflection. For our capped accelerometer
device, the gap between the cap and the device is 5 ?m.
Therefore, in order to avoid cap material touching the
movable parts, maximum allowable deflection on the
inner surface of the glob top encapsulant is limited to 5
?m. Materials selected for this study were PMDA
polymide, high strength parylene C, and carbon based
epoxy resin. These materials were selected for their
outstanding mechanical, thermal and chemical resistant
properties. Carbon based epoxy resin has the highest
Young?s modulus, followed by parylene C and polymide.
However, parylene C is hermetic, and polymide is more
delamination resistant [6]. Young?s modulus and poisson
ratio values for the selected materials are summarized in
table 1 below.
Material Young?s modulus
(GPa)
Poisson ratio
Polymide 7.5 0.35
Parylene C 27.59 0.4
Carbon epoxy resin 70 0.4
Table 1. Young?s modulus and poisson ratio values for
encapsulant canditate materials [7].
It could be seen from equations (1) and (2) that
given the dimensional parameters, Young?s modulus,
shell thickness, and external pressure applied are the
determinant factors for deflection. Poisson ratio on the
other hand, is somewhat consistent across the selected
candidate materials as shown in table 1. Hence, the
material selection criteria have to involve these factors. In
order to avoid simulating through the whole spectrum of
thickness, pressure, and material selection, Taguchi
method was used to construct a systematic approach to
screen for the best encapsulant material and thickness
under external pressure up to 100 atm. Screening design
was used to determine the optimal material and glob top
thickness required. The factors selected were Young?s
modulus, shell thickness, and applied external pressure,
and three levels were selected for each factor. Thickness
variations considered were 150, 200 and 250 ?m and
pressure variations were 80, 90 and 100 atm. These
values were inserted into L9 orthogonal matrix as shown
in table 2, and analyzed by Taguchi approach using JMP
software.
?
?
a
P
w
2b
t
?
?EDA Publishing/DTIP 2007
ISBN: 978-2-35500-000-3
A. Hamzah, Y. Husaini, B. Y. Majlis, and I. Ahmad
Selection of High Strength Encapsulant for MEMS Devices
Experiment Combination Material Thick
ness
(?m)
P
(atm)
1 - + + Polyimide 250 100
2 + - + C.E.Resin 150 100
3 - 0 0 Polyimide 200 90
4 0 + - Parylene C 250 80
5 - - - Polyimide 150 80
6 0 - 0 Parylene C 150 90
7 + 0 - C.E.Resin 200 80
8 0 0 + Parylene C 200 100
9 + + 0 C.E.Resin 250 90
Table 2. L9 orthogonal matrix showing factor and level
combinations for Taguchi experiment.
Simulation study was conducted using
CoventorWare ver.2005 software. The glob top
encapsulation was approximated as a shelled dome as
depicted in figure 4. The simulation model consists of
three main parts: silicon substrate, capped MEMS device,
and encapsulation shell. A cross-section of modeled
encapsulation structure used in simulation is shown in
figure 5. Uniform pressure was applied to the dome as to
imitate the molding pressure on the encapsulant material.
Deflection results obtained from CoventorWare
simulation were then compared with deflection results
obtained using shell bending equations. Finally, analysis
of variance (ANOVA) and sensitivity analysis were
conducted on the simulation results to verify the
reliability of the statistical model.
Figure 5. CoventoreWare screen capture of a cross-
section view of the encapsulation model used in
deflection simulation.
4. RESULTS AND DISCUSSION
First and foremost, CoventorWare simulation conducted
revealed that maximum deflection occurs at the center of
the encapsulation structure. Figure 6 shows a deflected
encapsulation structure under uniform external pressure,
with major deflection concentrated at the center of the
structure. This observation was adapted in deflection
calculation using shell bending equations, in which ?
value is set to approximately zero (? ? 0). This condition
was implemented to imitate maximum deflection at the
top of the encapsulation structure observed in
CoventorWare simulation.
Surface deflection values obtained from
simulation performed with the aforementioned materials,
pressure, and thickness variations are tabulated in table 3
below. Alongside the simulation results, deflection values
calculated using shell bending equations (1) and (2)
above are also tabulated for comparison. It could be seen
that the calculated and simulated results are close most of
the times, but deviate profusely at some instances.
However, one could observe that the deviation between
the two sets of results is generally less than 38%.
Therefore, the calculated results provide a good basis for
simulation verification. Based on both sets of results, it
could be concluded that the simulated deflection values
are in close proximity to the actual deflection values
given the parameters and loading conditions stated above.
Exp Material Thick
ness
(?m)
P
(atm)
w
s
sim
(?m)
w
c
calc
(?m)
Error
w
c
/w
s
(%)
1 Polyimide 250 100 12.59 15.58 23.79
2 C.E.Resin 150 100 4.95 3.11 -37.25
3 Polyimide 200 90 18.94 17.53 -7.42
4 Parylene C 250 80 2.60 3.12 20.07
5 Polyimide 150 80 33.18 20.78 -37.38
6 Parylene C 150 90 2.93 5.85 99.78
7 C.E.Resin 200 80 1.70 1.54 9.44
8 Parylene C 200 100 5.39 4.88 9.44
9 C.E.Resin 250 90 1.16 1.38 19.69
Table 3. Comparison of simulated and calculated
deflection values for all combinations tested.
silicon
substrate
capped
MEMS
device encapsulation
structure
?EDA Publishing/DTIP 2007
ISBN: 978-2-35500-000-3
A. Hamzah, Y. Husaini, B. Y. Majlis, and I. Ahmad
Selection of High Strength Encapsulant for MEMS Devices
Figure 6. A simulation capture showing 150 ?m thick
carbon based epoxy resin with deflection of 4.95 ?m
under 100 atm pressure.
The deflection values obtained from
CoventorWare simulation were analyzed using Taguchi
method. Figure 7 shows JMP results that relate
encapsulation deflection values with each factor
considered. Factor desirability is plotted underneath
deflection values of each factor at each level, where
desirability increases with decrease in deflection value.
The vertical lines mark level selected for each factor. The
top horizontal lines mark the corresponding deflection for
the selected factor combination. On the other hand, the
bottom horizontal lines mark the desirability of each
factor at the level selected. The intersection of the
horizontal line and the slanted curve in the utmost right
column marks the total desirability of the level
combination. The slanted dashed curves at the very top of
both second and third columns indicate the range of
continuous thickness and pressure variation. The JMP
curves are used to determine optimal encapsulation
thickness given the deflection limit. Optimized result in
figure 7(a) shows that 172.5 ?m thick carbon based epoxy
resin glob top would deflect 4.09 ?m under 100 atm
pressure. Similarly, parylene C glob top of thickness 205
?m would deflect 4.98 ?m under 100 atm pressure (figure
7(b)). Deflection curves for polymide are not shown as it
deflects more than 5 ?m for any encapsulation thickness
and pressure combination.
(a)
(b)
Figure 7. JMP results showing deflection values for (a)
172.5 ?m thick carbon based epoxy resin under 100atm
pressure. (b) 205 ?m thick parylene C under 100atm
pressure. Note that the deflection values are less than 5
?m for both cases.
ANOVA was used to evaluate the adequacy of
the Taguchi model developed. Table 4 summarizes
ANOVA results. The P-value indicates that the model is
more than 95% significant in the study of surface
deflection, which statistically proves the high precision of
the results.
Source df SS MS F P-value
Simulation 4 1443.2734 360.818 7.5094 0.0242
Errors 5 240.2443 48.049
Total 9 1683.5177
Table 4. ANOVA test results on the adequacy of the
Taguchi model developed. The parameter df is degree of
freedom, SS is sum of square, MS is mean of square, F
is SS divided by MS, and P-value is the smallest alpha.
0
10
20
30
40
0.00
0.25
0.50
0.75
1.00
7.5
27.59
70
125 150 175 200 225 250 275
75 80 85 90 95
100 105
.00 .25 .50 .75
1.00
0
10
20
30
40
0.
00
0.
25
0.
50
0.
75
1.
00
7.
5
27.
59 70
125 150 175 200 225 250 275
75 80 85 90 95
100 105
.0
0
.2
5
.5
0
.7
5
1.
00
maximum
deflection
z
Young?s modulus thickness total
desirability
r
e
liabilit
y
d
e
fle
c
tion
pressure
Young?s modulus thickness total
desirability
re
l
i
a
b
i
lity def
l
ection
pressure
?EDA Publishing/DTIP 2007
ISBN: 978-2-35500-000-3
A. Hamzah, Y. Husaini, B. Y. Majlis, and I. Ahmad
Selection of High Strength Encapsulant for MEMS Devices
Another important aspect to consider is the
sensitivity of each factor in the model. The main effect
test was used to check the sensitivity of each factor
towards surface deflection. In this test, P-value indicates
the influence of a particular factor on surface deflection.
A higher P-value means that a factor is less sensitive.
Table 5 summarizes the main effect test results. It is
observed that Young's modulus is the most sensitive
factor in the model, compared to encapsulation thickness
and applied pressure, indicated by its small P-value. Thus,
a small change in Young's modulus value would greatly
affect surface deflection. It is therefore most crucial to
choose a material with correct Young's modulus value in
order to obtain the desired deflection result.
Source Nparm df SST F P-value
Young?s
Modulus
2 2 694.1383 7.2233 0.0335
Thickness 1 1 28.7176 0.5977 0.4744
Pressure 1 1 123.2013 2.5641 0.1702
Table 5. Main effect test results. The parameter Nparm is
number of parameters, df is degree of freedom, SST is
total sum of square, F is F distribution, and P-value is the
smallest alpha.
From the simulation, calculation, and analyses
conducted, the best material for MEMS encapsulation and
the important factors involved have been successfully
analyzed. Carbon based epoxy resin was determined the
best material for encapsulation of MEMS devices
undergoing high pressure packaging.
5. CONCLUSIONS
CoventorWare is an excellent tool for simulating
deflection of encapsulation surface under external
pressure. The simulated deflection values are in close
proximity to deflection values obtained using shell
bending equations. Screening design has effectively
simplified the encapsulant selection process. Carbon
based epoxy resin and parylene C are acceptable glob top
materials since their deflection under 100 atm loading are
less than 5 ?m for thickness within 250 ?m limit.
Polyimide is deemed unsuitable since its deflections are
greater than 5 ?m for the entire thickness and pressure
variations. Carbon based epoxy resin was selected as the
best encapsulation material for MEMS devices
undergoing high pressure packaging process due to its
high strength.
6. REFERENCES
[1] C.T. Hsieh, J.M. Ting, C. Yang, and C.K Chung, The
Introduction of MEMS Packaging Technology In Procs.
4
th
International Symposium on Electronics Materials and
Packaging, 2002, IEEE, 2002, pp. 300-306.
[2] J. Wu and C.P. Wong, ?Development of new low
stress epoxies for MEMS device encapsulation,? IEEE
Transactions on Components and Packaging
Technologies, Vol. 25, Issue 2, June 2002, pp. 278-282.
[3] A. Partridge, ?Lateral Piezoresistive Accelerometer
with Epipoly Encapsulation,? PhD thesis, Stanford
University, 2003.
[4] K.S. Lebouitz, A. Mazaheri, R.T. Howe, and A.P.
Pisano, Vacuum Encapsulation of Resonant Devices
using Permeable Polysilicon In Procs, Twelfth IEEE
International Conference on Micro Electro Mechanical
Systems, 1999 (MEMS?99), IEEE, 1999, pp. 470-475.
[5] A.C. Ugural, ?Stress in plates and shells,? McGraw-
Hill, New York, 1981.
[6] J. Wu, R.T. Pike, and C.P. Wong, ?Novel bi-layer
conformal coating for reliability without hermeticity
MEMS encapsulation,? Electronics Packaging
Manufacturing, Vol. 22, Issue 3, July 1999, IEEE, pp.
195-201.
[7] J. Dolbow and M. Gosz, ?Effect of out-of-plane
properties of a polyimide film on the stress fields in
microelectronic structures,? Mechanics of
Materials, Vol. 23, Issue 4, August 1996, pp. 311-321.
?EDA Publishing/DTIP 2007
ISBN: 978-2-35500-000-3