Effects of Sustained Exposure to Temperature and Humidity on the Reliability and Performance of MEMS Microphone

Abstract

MEMS microphones are extensively used in many applications that require reliability, small size, and high sound quality. For harsh environment reliability data MEMS microphones need to be monitored under conditions mimicking their areas of applications. MEMS microphones have an opening/sound port in order to interact with the environment, therefore cannot be sealed completely since the sensing mechanism requires interaction between sound waves and the sensing element. Little to no information exists on reliability data for MEMS microphones under low/high temperature operating life and temperature humidity bias condition. Our work is primarily focused on providing harsh environmental reliability data which can be useful to MEMS designers and engineers. In this paper the test vehicles with MEMS Microphones have been tested under three different harsh environmental conditions: high temperature operating life (HTOL) at 125°C at 3.3V, low temperature storage (LTS) at −35°C and temperature humidity 85°C/85%RH at 3.3V. The main motive of this study is to document the incremental shift and degradation in output parameters namely distortion, frequency response, power supply rejection capability of IC, frequency vs pressure characteristics and analog output voltage of the MEMS microphone. The survivability of MEMS microphone, ADMP401, has been demonstrated as a function of change in the output parameters. Failure analysis has been conducted on the microphone samples to study failure modes and sites using analytical methods such as SEM, EDS and X-ray.

Full text

EFFECTS OF SUSTAINED EXPOSURE TO TEMPERATURE AND HUMIDITY ON THE
RELIABILITY AND PERFORMANCE OF MEMS MICROPHONE
Pradeep Lall
Auburn University
Department of Mechanical Engineering and
NSF Center for Advanced Vehicle and
Extreme Environmental Electronics (CAVE3)
Auburn, AL 36849
Tele: (334) 844-3424
E-mail: lall@auburn.edu
Amrit Abrol
Auburn University
Department of Mechanical Engineering and
NSF Center for Advanced Vehicle and
Extreme Environmental Electronics (CAVE3)
Auburn, AL 36849
David Locker
US AMRDEC
Huntsville, AL
ABSTRACT
MEMS microphones are extensively used in many applications
that require reliability, small size, and high sound quality. For
harsh environment reliability data MEMS microphones need to
be monitored under conditions mimicking their areas of
applications. MEMS microphones have an opening/sound port
in order to interact with the environment, therefore cannot be
sealed completely since the sensing mechanism requires
interaction between sound waves and the sensing element.
Little to no information exists on reliability data for MEMS
microphones under low/high temperature operating life and
temperature humidity bias condition. Our work is primarily
focused on providing harsh environmental reliability data
which can be useful to MEMS designers and engineers. In this
paper the test vehicles with MEMS Microphones have been
tested under three different harsh environmental conditions:
high temperature operating life (HTOL) at 125oC at 3.3V, low
temperature storage (LTS) at -35oC and temperature humidity
85oC/85%RH at 3.3V. The main motive of this study is to
document the incremental shift and degradation in output
parameters namely distortion, frequency response, power
supply rejection capability of IC, frequency vs pressure
characteristics and analog output voltage of the MEMS
microphone. The survivability of MEMS microphone,
ADMP401, has been demonstrated as a function of change in
the output parameters. Failure analysis has been conducted on
the microphone samples to study failure modes and sites using
analytical methods such as SEM, EDS and X-ray.
KEYWORDS: Micro Electro Mechanical Systems (MEMS),
Sound Pressure Level (SPL), Distortion, Frequency response,
Welch power spectral density, Power supply rejection ratio,
Reverberation chamber, Energy dispersive X-ray spectroscopy
NOMENCLATURE
k spring constant, N/m
m mass, proof mass, kg
PSRR Power Supply Rejection Ratio
PPA Parallel Plate Actuator
INTRODUCTION
In past only few studies have been conducted regarding
reliability of MEMS microphones, of which some demonstrate
the effect of high impact loading [Fang 2013], authors studied
the mechanical reliability of MEMS microphones subjected to
very high g shocks and demonstrated the characteristics of
diaphragm under shock loadings of up to 30000G. Finite
element simulations have also been used to study the shock
impact loading on a MEMS microphone and study the potential
failure sites [Li 2013]. The effect of airborne impurities and
humidity on MEMS microphone has been also been studied
[Broas 2015], who observed galvanic corrosion in thin film
Silicon MEMS. Overall Failure mechanisms and reliability of
different classes of MEMS devices are well documented
Proceedings of the ASME 2017 International Technical Conference and Exhibition on Packaging and
Integration of Electronic and Photonic Microsystems
InterPACK2017
August 29-September 1, 2017, San Francisco, California, USA
IPACK2017-74252
1
Copyright © 2017 ASME
This work was authored in part by a U.S. Government employee in the scope of his/her employment.
ASME disclaims all interest in the U.S. Government’s contribution.

[Fonseca 2011] [Huang 2012] but all together not much effort
has been made in assessing the effect of extreme operating
environmental stresses on characteristics of MEMS
microphones. Previous studies and research areas do not study
and observe the incremental evolution of damage progression
in MEMS microphones. Also, reliability data for analyzing and
understanding the characteristics of MEMS microphones under
low temperature storage (LTS), high temperature operating life
(HTOL) and temperature humidity bias (THB) conditions is not
available. Importance of reliability data for MEMS designers
and engineers is of the utmost priority since extreme
environmental operating conditions affect the grade, reliability
and performance of a MEMS device. MEMS microphones have
a similar detection scheme as MEMS pressure sensors which
comprises of capacitive sensing using the technique of PPA.
Depending upon the application of the sensor it would either
have a diaphragm or a membrane. Generally the diaphragm or
the membrane will have a bossed structure in order to generate
a linear capacitance change. MEMS microphone used in this
investigation was ADMP401, an omnidirectional microphone
with bottom port and analog output, ideal for smartphones,
digital video cameras, video phones and tablets. The sensor is a
simple silicon capacitor consisting of two silicon surfaces
where one surface is fixed and the other is movable. The fixed
surface/plate is conductive in nature and is comprised of
acoustic holes. The movable plate/surface has ventilation holes
which allow the membrane/diaphragm to move back and forth.
An application specific integrated circuit is connected via wire
bonds to the MEMS structure which converts the polarized
capacitance into analog voltage output. The IC is comprised of
an impedance converter and an output amplifier. The MEMS
structure and the IC are housed in a 4.72mm x 3.76mm x 1mm
surface mount package which encloses all the components. For
ADMP401, the sound port is constructed by drilling the
substrate as per the location of MEMS sensing structure. In this
architecture the front chamber is the cavity of the MEMS sensor
and the back chamber is created by the package. Resonant
chambers are distinguished depending on the location of
respective chambers in reference to the progressing sound
wave. The MEMS sensor is right under the sound port and the
two components, MEMS sensor and IC, are attached onto the
substrate. MEMS microphones such as ADMP401 and its
internal circuitry are unintentionally exposed to harsh
environmental conditions when operating in their fields of
applications. Extreme environments such as high-g shock,
vibration, thermal stresses, noise sources, humidity and
intermode coupling can lead up to a number of error sources in
a MEMS microphone resulting in a bad consumer product thus
the importance of reliability data. Presence of sophisticated and
intricate MEMS technology in microphones generates a
significant lack of reliability data that looks into the impact of
low temperature storage (LTS), high temperature operating life
(HTOL) and temperature humidity bias (THB) on
characteristics of MEMS microphones. The fundamental
objective behind this research work is to observe and anatomize
the output parameters of a commercial MEMS microphone and
assess the accrued damage. The reliability data shown in this
paper represents damage progression in MEMS microphones
till failure. Three different sets of experiments where conducted
in this piece of study where in the first set the MEMS test board
assemblies, MEMS microphone mounted on a FR4 layer PCB,
were subjected to high temperature operating life 125oC at
3.3V. The MEMS test boards were put in a thermal aging oven
set at the above mentioned parameter and a total of 10 test board
assemblies were subjected to harsh environment operating
condition. For the second test the MEMS test board assemblies
were subjected to temperature-humidity bias environment,
85oC/85%RH at 3.3V. A total of 10 test board assemblies were
subjected to THB operating environment. Lastly, for the third
test the MEMS test board assemblies were subjected to low
temperature storage, -35oC. A total of 5 test board assemblies
were subjected to LTS operating environment. MEMS test
board assemblies from all the three sets where taken out from
their respective chambers after an interval of every 100 hours
and put to test in a reverberation chamber. A sine sweep sound
ranging from 100Hz to 15KHz is played as the test sound since
this range falls in the full scale band of the MEMS microphone.
Room equalizer wizard (REW), an open source acoustics
software has been used to capture the frequency response of the
MEMS microphone samples with respect to sound pressure
level. Further this frequency response from the microphone
samples has been used to compute the distortion over the entire
audio band. Another parameter which was monitored for
damage progression is Power supply rejection ratio in order to
analyze the degradation in the capability of the ASIC to reject
noise added to the supply voltage. The raw output voltage from
ADMP401 has also been recorded for all the samples on which
Fast Fourier analysis has been performed in order to see damage
progression in the amplitude of the voltage output. The
objective of this experiment was to compute and compare the
frequency response, distortion, PSRR and voltage outputs for
pristine and HTOL, LTS and THB samples thereby assessing
the accrued damage in the microphone. Degradation of MEMS
microphone parameters has been observed and documented.
TEST VEHICLE
Free-PCB, open source software was used to design the MEMS
test board. The dimensions of test board are 132mm x 77mm. It
is a JED EC standard double s ided FR4 material PCB. PCB pads
are solder mask defined with immersion finish. The test boards
were fabricated and then assembled using the surface mount
technology line present at CAVE3 electronics research center.
Figure 1:- Test Board Design
The printed circuit board used in this investigation has been
fabricated for the testing of Accelerometers, Oscillators,
Gy rosc opes and Pre ssur e senso rs. Howev er thi s parti cular study
only focusses on ADMP401, MEMS bottom ported analog
microphone.
2
Copyright © 2017 ASME

Figure 2:- ADMP401 block diagram and Pin Layout
Figure 3:- ADMP401 mounted on MEMS test board
ADMP401 has an internal output amplifier, an impedance
converter and requires a single 4.7μF ceramic capacitor serving
as the decoupling capacitor. Externally the ADMP 401 sensor
is connected to an operational non inverting amplifier,
OPA344. Figure shows the external circuitry which was
connected between the ADMP401 and the op amp gain stage.
Figure 4:- ADMP401 connections to the op amp gain stage
The MEMS structure and the IC are housed in a 4.72mm x
3.76mm x 1mm surface mount package which encloses all the
components. In this architecture the front chamber is the cavity
of the MEMS sensor and the back chamber is created by the
package. Internal structure properties of the MEMS
microphone are unknown since it is proprietary to Analog
devices. The MEMS sensor is right on top of the sound port and
the IC is located in the back chamber as discussed before. The
X-ray images was taken at CAVE3 center at Auburn
University.
Figure 5:- X ray images of the internal structure
Figure 6:- ADMP 401 with the metallic casing and decapped
ADMP401 samples were decapped in order to visualize the
internal structure of the Microphone. The samples were
decapped by placing them on a hot plate set at 100
o
C and then
applying a small amount of shearing force with a sharp metallic
tool so as to remove the lid/casing. Scanning electron
microscopy was performed on the decapped MEMS samples in
order to closely observe the internal structure. Again, since this
device is proprietary to Analog devices no information was
available regarding the interior edifice and its properties. Below
is a set of SEM images which clearly demonstrate the
diaphragm shape, IC and diaphragm wire bond sites, acoustic
holes, folded spring elements and the fixed plate which is on the
opposite side to the diaphragm.
Figure 7:- Top view of MEMS sensor and IC
3
Copyright © 2017 ASME

Figure 8:- Zoomed in view of the Diaphragm
ADMP401 uses the capacitive sensing MEMS technology. The
diaphragm and the fixed plate comprise to make the variable
capacitor, where the diaphragm is movable as shown in Figure
1. The fixed plate, shown in Figure 11, has acoustic holes and
is conductive which allows sound waves to propagate through.
The movable diaphragm is connected to silicon spring element,
Figure 10, which helps in deflection. The movable surface is
full of ventilation holes, which allow the air compressed in the
back chamber to disperse out, support in the back, and forth
motion of the diaphragm. Internal structure properties of the
MEMS microphone such as spring constant, k, and mass, m are
unknown for analysis since it is proprietary to
STMicroelectronics.
Figure 9:- Ventilation holes for compressed air
ADMP401 operates anywhere between 1.5-3.3V, for this study
it was connected to a 3.3V power supply and 6 other passive
components listed in Table 1 referred from Figure 4. It is also
connected to a standard commercially available low power
single supply CMOS operational amplifier, OPA344. The op
amp is unity gain stable. The MEMS microphone sensor has a
flat frequency response between 100 Hz-15 kHz and a low
current consumption value of 250μA.
Figure 10:- Spring connected to the diaphragm
Figure 11:- Fixed plate on backside of ADMP401
Figure 12:- Zoomed in view of acoustic holes on fixed plate
It is an Omni-directional bottom ported analog sensor with a
typical sensitivity of -42dBV and a maximum acoustic input of
120dB SPL (sound pressure level). Table 2 provides a list of all
4
Copyright © 2017 ASME

the pin connections and pin layout for ADMP401 microphone
sensor, refer to Figure 2.
Table 1:- Passive Components and their values
Type Value Pin No.
Ceramic
Capacitor
4.7μF (C3) 1 , ADMP401
Ceramic
Capacitor
1μF (C2) 2, OP-Amp
Ceramic
Capacitor
100pF (C4) 4, OP-Amp
S. Mount Resistor 1.5kΩ (R1) 1, ADMP401
S. Mount Resistor 10kΩ (R2) GND
S. Mount Resistor 10kΩ (R3) GND
S. Mount Resistor 100kΩ (R4) 4, OP-Amp, Out
Table 2:- LPY510AL Pin Configuration
Pin No. Mnemonic Description, Pin No.
1 OUTPUT Analog output, 1
2 GND Ground, 2
3 GND Ground, 3
4 GND Ground, 4
5 Vdd Power Supply, 5
6 GND Ground, 6
RELIABLITY TEST MATRIX
A total of 15 ADMP401 microphone sensors were put to test in
this study which is further divided into three sub-studies each
comprising of 5 sensors each. The three different harsh
environmental conditions are high temperature operating life
(HTOL) 125
o
C at 3.3V, temperature humidity bias (THB)
85
o
C/85%RH at 3.3V and low temperature storage (LTS) at -
35
o
C. The main objective of this research paper is to put
forward the progressive evolution of damage in MEMS
microphone sensor while monitoring the distortion, frequency
response, power supply rejection of IC and frequency vs
pressure characteristics. In each sub study the ADMP401
samples were taken out from their respective chambers,
operating at different harsh conditions mentioned above, at a
regular interval of 100 hours and put to test in a reverberation
chamber. A sine sweep sound ranging from 0Hz-15kHz was
used to excite the MEMS acoustic sensor. Table 3 shows the
different test parameters used for the harsh environmental study
on ADMP401 MEMS microphones.
Table 3:- Reliability test matrix
Test
Condition
Test
Parameters
Data
Acquisition
Test
Samples
HTOL 3.3V 125
o
C 100hrs. 5
THB 3.3V 85
o
C/85%RH 100hrs. 5
LTS -35
o
C 100hrs. 5
Pristine ------------ ----------- 5
Figure 13:- ADMP401 test flowchart
Figure 13 shows the test flowchart for the harsh environmental
study on ADMP401 sensors. The source speaker is connected
to a power amplifier and the amplifier is connected to a
computer with an operational soundcard. Room equalizer
wizard is an open source software, which has been used in this
study, the software is used to generate a sine sweep sound
ranging from 0Hz-15kHz played via the soundcard on the
connected computer. The output of the microphone is fed into
an audio interface, Behringer UM2, via a XLR cable, which is
further connected to the soundcard on the operating computer.
Microphone analog output voltage has been monitored and
recorded using a high-speed data acquisition system,
Waverunner Xi series. Figure 14 shows the data acquisition
system used for the experiment.
Figure 14:- Waverunner Xi series oscilloscope
EXPERIMENTAL TEST SETUP
ADMP401 MEMS microphone sensor-test board assemblies
were assembled at the Surface Mount Technology line present
on CAVE3 facility. The Surface Mount technology (SMT) line
consists of three stages, the Stencil printer, Pick-and-Place
machine and reflow oven as shown in Figure 15 . SAC305
solder alloy was used for this study and was evenly distributed
on MEMS test board assemblies through the Stencil printer.
ADMP401 and other 6 passive components were placed on the
MEMS test board using the Pick and Place machine. Finally,
the test boards were baked in the reflow oven. Figure 16 shows
the three different chambers used for this this study.
5
Copyright © 2017 ASME

Figure 15:- SMT Line at CAVE3; Stencil Printer (Top), Pick
and Place (Middle), Reflow Oven (Bottom)
Figure 16:- HTOL chamber (TOP), THB chamber (Middle)
and LTS chamber (Bottom)
The 5 samples which were pristine did not undergo any aging
or high/low temperature or humidity exposure. Our approach
for this research study supports for a constructive comparison
among the four sets of samples. Microphone gauge parameters
such as distortion, frequency response, power supply rejection
of IC and frequency vs pressure characteristics can demonstrate
on how sensitive the ADMP401 samples are to the four above
mentioned harsh environmental. Figure 18 shows the inside
view of the reverberation room where the samples were put to
test.
Figure 17:- YXLON CT and X-ray machine
6
Copyright © 2017 ASME

One of the main objectives behind this investigation is to
quantify whether or not the microphone sensor parameters are
drifting significantly with respect to the pristine benchmark
values and by how much; when exposed to harsh environmental
conditions. The X-ray images shown in Figure 5 were taken at
Auburn University, CAVE3 facility, using the state-of-the-art
YXLON computed tomography and X-ray machine in Figure
17.
Reverberation Chamber Setup
Figure 18:- Inside View of Reverberation room
Figure 19:- Inside View of Reverberation room
Figure 20:- Source speaker on top of MEMS test board
assembly
The reverberation chamber/room helps is achieving an even
state of acoustic energy. Acoustic impedances offered by the
surface of this chamber to the propagating sound waves are very
large thereby reflecting most of the acoustic energy back within
the room. The chamber comprises of a two layer hardwood and
timber setup in order to maximize the reverberation time, also
no cavities are present within the walls or the roof which help
in avoiding unwanted resonances. The source speaker was
carefully mounted on a wooden rack as shown in Figure 20 and
the MEMS test board assemblies were kept on a small circular
table top which was at a fixed distance of 20cm from the
speaker, see Figure 19 and Figure 20 . Again as mentioned
above the MEMS microphone test board assemblies were taken
out after every 50hrs from their respective harsh environmental
chambers and then put to test in the reverberation room.
Figure 21:- XLR cable and TRSS connector
Since the ADMP401 is an analog sensor, the output signal
cannot be directly fed into the audio interface, Behringer UM2.
A commercially available of the shelf TRSS connector,
manufactured by Spark fun, was used which takes in the analog
7
Copyright © 2017 ASME

output signal and transports it to the audio interface via a XLR
cable with an audio jack end as shown in Figure 21.
Figure 22:- Power amplifier and audio interface.
The other end of the XLR cable is connected to the audio
interface powered through a standard USB cable which is
connected to the computer with soundcard. Figure 22
demonstrates the XLR cable output into the mic/line of the
audio interface and the output of the audio interface into the
source speaker.
Figure 23:- Behringer UM2 2 x 2 audio interface
The amount of time is takes for a sound wave to fade away is
an enclosed space or area is referred to as the reverberation
time. Surfaces in the room, reflective in nature, will help in
repeated bouncing of sound waves and when these reflections
fuse together reverberation is produced. Highly bibulous
surfaces reduce the reverberation when hit by reflections. The
acoustic status of a room is mostly characterized by the
reverberation. RT60 (60dB decay) reverberation time
measurement is an ISO 3382-1 standard for performance spaces
and is measured as soon as the source signal is abruptly ended
[NTI Audio 2017]. In experimental phase we generally measure
20dB or the 30dB decay and since the decay is linear, one can
extrapolate for 60dB decay.
Figure 24:- RT60 decay for reverberation chamber
RT60 plot for the reverberation chamber, shown in Figure 24,
was generated before every test so in order to maintain the
continuity of the experiment.
EXPERIMENTAL RESULTS
The fundamental objective behind this research work is to
observe and anatomize the output parameters of a commercial
MEMS microphone and assess the accrued damage. The
reliability data shown in this paper represents damage
progression in MEMS microphones till failure. In this study the
term failure does not actually mean that the device is not
working at all but is a coined term used to describe the progress
of damage i.e. until the Microphone output parameters go
permanently out of the stable operating range or drift
significantly with respect to the ideal output(s). In order to
normalize the survivability of ADMP401, MEMS analog
microphone, for smartphones, digital video cameras, video
phones and tablets, it has been subjected to the aforementioned
harsh environments. ADMP401 microphone sensors have been
gauged on standard microphone metrics such as distortion,
frequency response, power supply rejection capability of IC,
frequency vs pressure characteristics and output voltage.
Figure 25:- Voltage response of ADMP401 for a sine sweep
sound from 0 Hz-15 KHz
Room equalizer wizard is an open source software which has
been used in this study, the software is used to generate a sine
sweep sound ranging from 0 Hz-15 kHz played via the
soundcard on the connected computer. Figure 25 shows a
typical output voltage response signal for a pristine microphone
sample.
-0.8 -0.6 -0. 4 -0.2 00.2 0.4 0.6 0.8
0.5
1
1.5
2
2.5
3
Voltage Response
Time (s)
Voltage (V)
8
Copyright © 2017 ASME

Figure 26:- Frequency response vs Sound Pressure Level for
ADMP401
I. Effect of High temperature operating life (HTOL)
As mentioned before five MEMS test board assemblies were
subjected to 125oC operating at 3.3V. High temperature
operating life provides a harsh environment for the microphone
sensor where the stresses act dynamically and approach the max
absolute ratings in terms of junction temperature, load current
and internal power dissipation [ST microelectronics AN4428].
The main idea behind including HTOL test in our test matrix is
that this test simulates the worst case application harsh
conditions. Since the device operates in a biased condition
typical failure modes such as wire-bond aging, ASIC oxide
faults and metal degradation have been investigated.
A) Distortion
The ratio of output pressure to the input sound pressure when
sweeping from 0Hz - 15 kHz, where the reference sound
pressure in air is 20μPa, is referred as distortion.
Figure 27:- Distortion after 300 hours, 400 hours, 500 hours
and 600 hours of HTOL respectively
Figure 27 shows the distortion in five HTOL samples after 300,
400, 500 and 600 hours of 125oC at 3.3V. It can be clearly seen
that while most of the distortion between 2000Hz and 8000Hz
is not significant, the major frequencies contributing to
distortion are the higher and the lower end of the frequency
sweep band and this behavior is portrayed by all the HTOL
samples. Also the distortion peaks are as high as 10dB at 0Hz
which is due to the fact that ADMP401 has a flat frequency
response from 100Hz – 15kHz and the sine sweep frequency
band lies between 0Hz – 15kHz, therefore these peaks are
neglected and do not contribute to the distortion analysis.
Progression of HTOL with time yields higher distortion in all
the samples especially at lower (100Hz-200Hz) and higher
frequencies (10 kHz- 15 kHz). Also the distortion at the
intermediate frequencies (2000Hz – 8000Hz) does not tend to
remain flat.
B) Power Supply Rejection-Ratio
The ability of ASIC to reject the noise or a spurious signal
added to the supply voltage [ST microelectronics AN4426]. A
sine tone at 100mV pk-pk at 217Hz (GSM switching frequency
in phone applications) is added to the power supply and the
amplitude of output signal is measured. PSR is typically
expressed in decibels (dB). It is a gauge factor of how much
noise on the power supply will squeeze through to the output
signal. For complete power supply rejection the microphone
would have PSR equal to the A-weighted noise floor.
  20  

@217
0
20
40
60
80
100
120
0 2000 4000 6000 8000 10000 12000 14000 16000
SPL(dB)
Frequency(Hz)
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8 000 1000 0 1 2000 14000
Distortion(dB)
Frequency(Hz)
HTOL1
HTOL2
HTOL3
HTOL4
HTOL5
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8000 10000 12000 14000
Distortion(dB)
Frequency(Hz)
HTOL1
HTOL2
HTOL3
HTOL4
HTOL5
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8 000 1000 0 1 2000 14000
Distortion(dB)
Frequency(Hz)
HTOL1
HTOL2
HTOL3
HTOL4
HTOL5
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8000 10000 12000 14000
Distortion(dB)
Frequency(Hz)
HTOL1
HTOL2
HTOL3
HTOL4
HTOL5
9
Copyright © 2017 ASME

Figure 28:- PSR for HTOL samples
Five pristine samples were evaluated and the computed PSR
values were around 70.584dB for all of them. Figure 28 shows
the degradation in the PSR computed values for HTOL samples
as the HTOL operating time progresses. The deterioration of
ASIC over time reduces its capability to reject the noise added
to the supply voltage and also loses the IC parametric stability.
Average PSR decay is 0.63dB.
Power supply rejection ratio (PSRR) is the residual noise
amplitude at microphone output to the added spurious signal on
the supply voltage [10]. PSRR is typically expressed in decibels
(dB) as  20  


@217)
Figure 29:- PSRR for HTOL samples
Figure 29 shows the deterioration in the PSRR computed
values for HTOL samples till 600 hours. As mentioned before
both the PSR and PSRR parameters were evaluated using a
100mV pk-pk sine tone at 217Hz (GSM switching frequency in
phone applications) which is added to the power supply and the
amplitude of output signal is measured.
C) Frequency vs Pressure
The diaphragm of the ADMP401 sensor is disturbed by the
sound pressure deviations which is a sound field quantity. This
analysis can help us determine how the sound pressure is
detected by the diaphragm for a pristine case and a harsh stress
environment case. Since the same sine-sweep signal is used for
all the test cases and the distance of the MEMS test board
assemblies from the source speaker is kept same so ideally the
pressure detected by the diaphragm should be similar
irrespective of the test conditions.
With the increase in exposure time to HTOL test condition the
diaphragm expands thereby changing the dimensions of the
ventilation holes, on the micro scale, which in turn results in
change in the pressure detection by the diaphragm. It can be
observed from Figure 30 that the pressure detected by the
diaphragm for pristine ADMP401 sample is higher than the
HTOL samples especially at higher frequencies.
Figure 30:- Pressure detected by diaphragm vs Frequency for
Pristine (Top), HTOL test samples
68.6
68.8
69
69.2
69.4
69.6
69.8
70
70.2
70.4
70.6
70.8
0 100 200 300 400 500 600
PSR(dB)
HTOL Hours(125
o
C @ 3.3V)
Pristine
HTOL1
HTOL2
HTOL3
HTOL4
HTOL5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
0 100 200 300 400 500 600
PSRR (dB)
HTOL Hours(125
o
C @ 3.3V)
Pristine
HTOL1
HTOL2
HTOL3
HTOL4
HTOL5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
0 2000 4000 6000 8000 10000 12000 14000 16000
Pressure(Pa)
Frequency(Hz)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
0 2000 4000 6000 8000 10000 12000 14000 16000
Pressure(Pa)
Frequency(Hz)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
0 2000 4000 6000 8000 10000 12000 14000 16000
Pressure(Pa)
Frequency(Hz)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
0 2000 4000 6000 8000 10000 12000 14000 16000
Pressure(Pa)
Frequency(Hz)
10
Copyright © 2017 ASME

D) Interval Plot for HTOL
Interval plot is a graphical representation for a sample
distribution which focuses on the sample’s tendency and
variability. Figure 31 compares the means of pristine and
HTOL test cases, sample 3 and sample 5 do not show any
significant difference as their respective interval bars easily
overlap with the pristine interval bar. On the other hand samples
1, 2 and 4 show different means and are also significantly
different from the pristine test case.
Figure 31:- Interval Plot for HTOL samples after 600
hours
E) FFT analysis of output voltage
Fast Fourier transform allows us to analyze the voltage signal,
which is in time domain, in frequency domain and study the
overall signal content in terms of energy and spectral density.
FFT techniques have been used in the past [Lall 2015] to
analyze the incremental deterioration in energy of the MEMS
output voltage signal.
FFT analysis shows a significant reduction in the amplitude of
the output voltage signals in HTOL test samples. Samples 1, 2
and 4 show significant reduction in peak amplitude and overall
signal energy of the output signal. A drift in the peak frequency
has also been observed for the harsh environment test cases
when compared to pristine.
Figure 32:- FFT overlays for Pristine, HTOL1, HTOL2,
HTOL3, HTOL4 and HTOL5 after 600 hours
HTOL5HTOL4HTOL3HTOL2HTOL1Pristi ne
250
225
200
175
150
Me an
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequ ency (Hz)
Amplitude
Pristine
HTOL 1
Vari abl e
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
HTOL 2
Varia ble
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
HTOL 3
Vari abl e
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
HTOL 4
Vari abl e
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
HTOL 5
Varia ble
11
Copyright © 2017 ASME

II. Effect of Low Temperature Storage (LTS)
As mentioned before five MEMS test board assemblies were
subjected to -35oC. Low temperature storage provides a harsh
environment for the microphone sensor where the device is
stored in an unbiased condition at the minimum temperature
allowed by the package materials, this harsh environment is
useful for investigating the failure mechanisms activated by
extremely cold environments for prolonged time [ST
microelectronics AN4428].
A) Power Supply Rejection
The ability of ASIC to reject the noise or a spurious signal
added to the supply voltage [ST microelectronics AN4426]. A
sine tone at 100mV pk-pk at 217Hz (GSM switching frequency
in phone applications) is added to the power supply and the
amplitude of output signal is measured. PSR is typically
expressed in decibels (dB). For complete power supply
rejection the microphone would have PSR equal to the A-
weighted noise floor.
  20  

@217
Five pristine samples were evaluated and the computed PSR
values were around 70.584dB for all of them. Figure 333 and
Figure 344 show the deterioration in the PSRR and PSR
computed values for LTS samples till 300 hours. It can be
clearly seen that in the LTS test case the drift in the PSR and
PSRR values is almost negligible (about 0.2 dB) as compared
to HTOL test case up until the 300th hour mark.
Figure 33:- PSR for LTS samples
Figure 34:- PSRR for LTS samples
B) Interval Plot for LTS
Interval plot is a graphical representation for a sample
distribution which focuses on the sample’s tendency and
variability. Figure 355 compares the means of pristine and LTS
test cases, it can be clearly seen that since the interval bars for
all of the samples easily overlap with the pristine test case
therefore there is no significant difference or drift between the
pristine and LTS samples.
Figure 35:- Interval Plot for LTS samples after 300 hours
C) FFT analysis of output voltage
Fast Fourier transform allows us to analyze the voltage signal,
which is in time domain, in frequency domain and study the
overall signal content in terms of energy and spectral density.
FFT techniques have been used in the past [Lall 2015] to
analyze the incremental deterioration in energy of the MEMS
output voltage signal.
FFT analysis shows that there is no significant reduction in the
amplitude of the output voltage signals in LTS test samples.
Samples 1, 2, 3, 4 and 5 show insignificant reduction in peak
amplitude and overall signal energy of the output signal. A drift
in the peak frequency has not been observed for the harsh
environment test cases when compared to pristine. Figure 36
shows the FFT overlays for pristine and LTS test case after 300
hours.
68.6
68.8
69
69.2
69.4
69.6
69.8
70
70.2
70.4
70.6
70.8
0 100 200 300 400 500
PSR(dB)
LTOL Hours(-35
o
C@ 3.3V)
Pristine
LTOL1
LTOL2
LTOL3
LTOL4
LTOL5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
0 100 200 300 400 500
PSRR (dB)
LTOL Hours(-35
o
C @ 3.3V)
Pristine
LTOL1
LTOL2
LTOL3
LTOL4
LTOL5
LTOL5LTOL4LTOL3LTOL2LTOL1Pristine
240
235
230
225
220
Me an
14900134101192010430894074505960447 0298014901
2000
1500
1000
500
0
Fre quen cy ( Hz)
Amplitude
Pristine
LTOL1
Varia ble
12
Copyright © 2017 ASME

Figure 36:- FFT overlays for LTS test case after 300 hours
III. Effect of Temperature Humidity Bias
As mentioned before five MEMS test board assemblies were
subjected to 85oC/85%RH operating at 3.3V. Temperature
humidity bias test aims at examining failure mechanisms
especially in the die-package surroundings caused by harsh wet
conditions and electric potential .Since the device operates in a
biased condition typical failure modes such as electro-chemical
corrosion have been investigated. EDS analysis shown in
section III below demonstrate high oxygen content found in
THB samples especially in the back chamber which constitutes
the ASIC.
A) Distortion
The ratio of output pressure to the input sound pressure when
sweeping from 0Hz - 15 kHz, where the reference sound
pressure in air is 20μPa, is referred as distortion.
149001341011920104308940745059604470298014901
2000
1 500
1 000
500
0
Frequ ency (Hz)
Amplitude
Pristine
LTOL2
Variab le
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
LTOL3
Variab le
14900134101192010430894074505960447 0298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
LTOL4
Variab le
149001341011920104308940745059604470298014901
2000
1500
1000
500
0
Frequency (Hz)
Amplitude
Pristine
LTOL5
Variab le
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8000 10000 12000 14000
Distortion(dB)
Frequency(Hz)
THB1
THB2
THB3
THB5
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8000 10000 1 2000 14000
Distortion(dB)
Frequency(Hz)
THB1
THB2
THB3
THB5
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 2000 4000 6000 8000 10000 12000 14000
Distortion(dB)
Frequency(Hz)
THB1
THB2
THB3
THB5
13
Copyright © 2017 ASME

Figure 37:- Distortion after 100 hours, 200 hours, 300 hours
and 400 hours of THB respectively
Figure 37 shows the distortion in five THB samples after 100,
200, 300 and 400 hou rs of 85
o
C/85%RH at 3.3V. Unlike HTOL
test case for THB the distortion between 2000Hz and 8000Hz
is significant and more than what is observed for HTOL.
Similar to high temperature operating life the major frequencies
contributing to distortion are the higher and the lower end of the
frequency sweep band and this behavior is portrayed by all the
THB samples. Also the distortion peaks are as high as 10dB at
0Hz which is due to the fact that ADMP401 has a flat frequency
response from 100Hz – 15kHz and the sine sweep frequency
band lies between 0Hz – 15kHz, therefore these peaks are
neglected and do not contribute to the distortion analysis.
Progression of THB with time yields higher distortion in all the
samples especially at lower (100Hz-200Hz) and higher
frequencies (10 kHz- 15 kHz) and is as high as -8dB. One of
five tested THB samples did not show any output signal even
before the completion of 100 hours of THB test, which mainly
can be because of an electrical overstress subjected to the ASIC.
Electrical overstress which is an unexpected phenomenon
causes irreversible damage to ASIC such as burned areas or
voids in Silicon [ST microelectronics AN4428]
B) Power Supply Rejection
The ability of ASIC to reject the noise or a spurious signal
added to the supply voltage [ST microelectronics AN4426]. A
sine tone at 100mV pk-pk at 217Hz (GSM switching frequency
in phone applications) is added to the power supply and the
amplitude of output signal is measured. PSR is typically
expressed in decibels (dB). It is a gauge factor of how much
noise on the power supply will squeeze through to the output
signal. For complete power supply rejection the microphone
would have PSR equal to the A-weighted noise floor.
  20  




@217
Five pristine samples were evaluated and the computed PSR
values were around 70.584dB for all of them. Figure 38 and
Figure 39 show the degradation in the PSR and PSRR computed
values for Temperature-humidity biased samples as the THB
operating time progresses. The deterioration of ASIC over time
in this case is more than as in HTOL test case. Average PSR
decay was 1.8dB.
Figure 38:- PSR for THB samples
Figure 39:- PSRR for THB samples
III. Energy Dispersive X-ray Spectroscopy Analysis (EDS)
Temperature Humidity Bias
Energy dispersive X-ray spectroscopy analysis shown in Figure
40 demonstrate high oxygen content found in THB samples
especially in the back chamber which constitutes the ASIC.
Oxygen rich content as high as 25-40% has been discovered
which makes oxidation a big possibility and metal degradation
and oxide faults at the ASIC site are very much likely.
Sample I
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 200 0 4000 6000 8000 1000 0 12000 1 4000
Distortion(dB)
Frequency(Hz)
THB1
THB2
THB3
THB5
68.6
68.8
69
69.2
69.4
69.6
69.8
70
70.2
70.4
70.6
70.8
0 100 200 300 400 500
PSR(dB)
THB Hours(85
o
C/85% RH @ 3.3V)
Pristine
THB1
THB2
THB3
THB4
THB5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
0 100 200 300 400 500
PSRR (dB)
THB Hours(85
o
C /85% RH @ 3.3V)
Pristine
THB1
THB2
THB3
THB4
THB5
14
Copyright © 2017 ASME

Sample II
Sample III
Sample IV
Sample V
Figure 40:- EDS scan results for THB samples I, II, III, IV
and V respectively after 400hours of 85
o
C/85%RH at 3.3V
Figure 41:- ADMP401 THB sample after 400hrs (Left) and
Pristine sample (Right)
Pristine
Figure 42:- EDS scan results for pristine samples
For the same ASIC site Pristine samples do not show any traces
of Oxygen thereby indicating to the fact that ADMP401
samples subjected to THB test condition have Oxygen presence
within the MEMS package which causes the deterioration of the
MEMS assembly.
SUMMARY AND CONCLUSIONS
This research study presents a reliability and defect detection
technique to analyze the behavior and characteristics of a
commercially available MEMS microphone when exposed to
high temperature operating life and biased temperature-
humidity test conditions. This approach allows MEMS
designers and engineers to investigate and document the
evolution of damage in ADMP401 MEMS bottom ported
microphone. Through reliability and defect detection analysis
this methodology was successful in characterizing the overall
behavior of the MEMS device. The incremental damage
progression was reported in terms of distortion, frequency
response, power supply rejection capability of IC, frequency vs
pressure characteristics and output voltage. On comparing the
pristine and harsh environment test case samples, FFT analysis
show a significant reduction in the amplitude of the output
voltage signals in THB and HTOL test samples especially
samples from THB test condition. Power supply rejection
values were computed for all the test cases; a decreasing trend
was observed in the PSR values of HTOL and THB samples
with the increase in exposure time. The computed PSR values
show significant reduction, especially for biased temperature-
humidity samples where reduction is almost 2dB, when
exposed to harsh environmental operating conditions in
comparison to the pristine sample. Results put forward for
power supply rejection clearly established that the IC connected
to the MEMS device is very sensitive to high humidity and high
temperature exposure when in a biased condition. This is also
supported by the EDS scan results which show significant
15
Copyright © 2017 ASME

amount of Oxygen presence, 25-40% by weight, in and around
the IC site within the MEMS package. Oxide formation and
metal degradation are very likely to have caused the overall IC
parametric instability. The diaphragm of the ADMP401 sensor
is disturbed by the sound pressure deviations which is a sound
field quantity. In this research paper pressure detection by the
diaphragm for a pristine case and a harsh stress environment
case vs frequency was also analyzed. Since the same sine-
sweep signal was used for all the test cases and the distance of
the MEMS test board assemblies from the source speaker is
kept same so ideally the pressure detected by the diaphragm
should be similar irrespective of the test conditions. It was
observed that the pressure detected by the diaphragm for
pristine ADMP401 sample is higher than the HTOL samples
especially at higher frequencies. This can be due to the fact that
the compressed air in the back chamber is not being properly
vented out through the ventilation holes present on the
diaphragm for its proper back and forth motion. The Silicon
diaphragm within the MEMS device operating at 125oC at 3.3V
undergoes thermal expansion thereby changing the dimensions
of the ventilation holes present on the diaphragm. Distortion or
the ratio of output pressure to the input pressure was also
observed and the results have been documented. The results
shown above in the study demonstrate that for HTOL samples
most of the distortion between 2000Hz and 8000Hz was not
significant, the major frequencies contributing to distortion
were the higher and the lower end of the frequency sweep band
and this behavior was portrayed by all the HTOL samples.
Progression of HTOL test condition with time yields higher
distortion in all the samples especially at lower (100Hz-200Hz)
and higher frequencies (10 kHz- 15 kHz). Also the distortion
at the intermediate frequencies (2000Hz – 8000Hz) does not
tend to remain flat. For the THB test samples unlike HTOL test
case the distortion between 2000Hz and 8000Hz was significant
and more than what was observed for HTOL cases. Similar to
high temperature operating life the major frequencies
contributing to distortion were the higher and the lower end of
the frequency sweep band and this behavior was demonstrated
by all the THB test samples. Progression of THB with time
yields higher distortion in all the samples especially at lower
(100Hz-200Hz) and higher frequencies (10 kHz- 15 kHz) and
was as high as -8dB. Also the distortion peaks in both HTOL
and THB test conditions were as high as 10dB at 0Hz which
was due to the fact that ADMP401 has a flat frequency response
from 100Hz – 15 kHz and the sine sweep frequency band lies
between 0Hz – 15 kHz, therefore these peaks were neglected
and did not contribute to the distortion analysis.
ACKOWLEDGMENTS
The research results presented in this paper are based on
projects supported by industrial members of NSF-CAVE3
Electronics Research Center at Auburn University.
REFERENCES
[1] M. Broas et al., "Galvanic corrosion of silicon-based thin
films: A case study of a MEMS microphone," 2015 IEEE
65th Electronic Components and Technology Conference
(ECTC), 2015, pp. 453-459.
[2] J. Li et al., "Reliability assessment of a MEMS microphone
under shock impact loading," 2013 14th International
Conference on Thermal, Mechanical and Multi-Physics
Simulation and Experiments in Microelectronics and
Microsystems (EuroSimE), Wroclaw, 2013, pp. 1-6.
[3] W. Fang and Q. Huang, "A study of the mechanical
reliability of a MEMS microphone," Proceedings of the
20th IEEE International Symposium on the Physical and
Failure Analysis of Integrated Circuits (IPFA), Suzhou,
2013, pp. 716-719.
[4] Fonseca, D.; Sequera, M., "On MEMS reliability and failure
mechanisms” 2011 International Journal of Quality,
Statistics and Reliability, 2011
[5] P. Lall, A. S. Abrol, L. Simpson and J. Glover, “Effect of
Simultaneous High temperature and vibration on MEMS
based vibratory gyroscope" 2017, InterPACK
[6] NTI-Audio, Reverberation Time RT60, 2017
[7] Lall, P.; Abrol, A.S.; Simpson, L.; Glover, J.,
Survivability of MEMS accelerometer under sequential
thermal and high-g mechanical shock environments,
InterPACKICNMM 2015
[8] P. Lall, A. S. Abrol, L. Simpson and J. Glover, "A study on
damage progression in MEMS based Silicon oscillators
subjected to high-g harsh environments," 2016 15th IEEE
Intersociety Conference on Thermal and
Thermomechanical Phenomena in Electronic Systems
(ITherm), Las Vegas, NV, 2016, pp. 546-559.
[9] ST microelectronics, Best practices in manufacturing
process of MEMS microphones, Application Note,
AN4428
[10] ST microelectronics, Tutorial for MEMS microphones
Application Note, AN4426
[11] Analog Devices, ADMP401, Technical data sheet, Rev E
[12] Lall, P.; Abrol, A.S.; Simpson, L.; Glover, J., Reliability
of MEMS devices under multiple environments, Thermal
and Thermomechanical Phenomena in Electronic Systems
(ITHERM), pp.1313-1321,2014
[13] Brown, T.G.; Harsh military environments and
Microelectromechanical systems, Sensors, Proceedings of
IEEE, Vol.2, pp, 753-760, 2003
16
Copyright © 2017 ASME